Initial Diagnostic Workup of Acute Leukemia: Guideline From the College of American Pathologists and the American Society of Hematology

This guideline was developed using an evidence-based methodology intended to meet recommendations for a report from the Institute of Medicine.⁹  This guideline is based on the results of a systematic review (SR) of available evidence. A detailed description of the methods and SR (including the quality assessment and complete analysis of the evidence) used to create this guideline can be found in the supplemental digital content, which also includes 24 tables and 2 figure files, at www.archivesofpathology.org in the October 2017 table of contents.

The CAP Pathology and Laboratory Quality Center (the Center) and the ASH members included 7 pathologists, one hematologist, one hematologist/oncologist, and one methodologist consultant. These panel members served as the expert panel (EP) for the systematic evidence review and development of the guideline statements. An advisory panel including one patient advocate, one cytogeneticist, 3 hematologists/oncologists (including one pediatric hematologist/oncologist), one medical oncologist, and 2 hematopathologists assisted the EP in determining the project scope and reviewing and providing guidance on the draft recommendations and manuscript development.

In accordance with the CAP conflict of interest policy (in effect April 2010), members of the expert panel disclosed all financial interests of possible relevance to the guideline, from 12 months before appointment through publication of the guideline. Individuals were instructed to disclose any relationship that could be interpreted as constituting an actual, potential, or apparent conflict. Disclosures were collected by the CAP staff before beginning the SR and were updated continuously throughout the project at each virtual and face-to-face meeting. A separate oversight group (consisting of staff and members of the CAP and ASH) reviewed the disclosures and agreed that most of the expert panel had no conflicts of interest. Complete disclosures of the expert panel members are listed in the Appendix. Disclosures of interest judged by the oversight group to be conflicts are as follows: D.A.A., consultancy and board/advisory board with Celgene Corporation (Summit, New Jersey), board/advisory board of DAVA Oncology (Dallas, Texas), Bristol-Myers Squibb (New York, New York), Novartis (Deerfield, Illinois), and Agios Pharmaceuticals (Cambridge, Massachusetts); M.J.B., grants received from Amgen Inc (Thousand Oaks, California), Beckman Coulter (Brea, California), Becton, Dickinson and Company (San Jose, California), Bristol-Myers Squibb (New York, New York), Genzyme Corporation (Cambridge, Massachusetts), MedImmune (Gaithersburg, Maryland), and Micromet (Rockville, Maryland); K.F., consultancy with Celgene Corporation (Summit, New Jersey); R.P.H., consultancies with Cancer and Leukemia Group B, Genzyme Corporation (Cambridge, Massachusetts), and Incyte Corporation (Wilmington, Delaware); S.A.W., consultancy with Genzyme Corporation (Cambridge, Massachusetts), board/advisory board with, and grants received from, Seattle Genetics, Inc (Bothell, Washington), and GlaxoSmithKline plc (Brentford, United Kingdom). Most of the EP (6 of 11 members) was assessed as having no relevant conflicts of interest. The CAP and ASH provided funding for the administration of the project; no industry funds were used in the development of the guideline. All panel members volunteered their time and were not compensated for their involvement, except for the contracted methodologist. Please see the SDC for full details on the conflict of interest policy.

The objective of the guideline is to recommend laboratory testing for the initial workup for proper diagnosis, determination of prognostic factors, and possible future monitoring of ALs, including AML, ALL and ALs of ambiguous lineage, in children and adults.

The key questions were as follows:

• 1.

  What clinical and laboratory information should be available during the initial diagnostic evaluation of a patient with AL?

• 2.

  What specimens and sample types should be evaluated during the initial workup of a patient with AL?

• 3.

  At the time of diagnosis, what tests are required for all patients for the initial evaluation of an AL?

• 4.

  Which tests should be performed on only a subset of patients, including in response to results from initial tests and morphology?

• 5.

  Where should laboratory testing be performed?

• 6.

  How should test results and the diagnosis be correlated and reported?

A systematic literature search was completed on October 4, 2011, for relevant evidence using OvidSP (Ovid Technologies, New York, New York), PubMed (US National Library of Medicine, Bethesda, Maryland), and Science Direct (Elsevier, Amsterdam, the Netherlands) to identify literature published from January 2005 through September 2011. A literature refresh was completed on April 24, 2013, and again on August 11, 2015, to identify recently published material. Database searches were supplemented with expert panel recommendations and the references from those supplemental articles were reviewed to ensure all relevant publications were included.

Selection at all 3 levels of the SR was based on predetermined inclusion/exclusion criteria for the outcomes of interest. Detailed information about the literature search and selection can be found in the supplemental data.

An assessment of the quality of the evidence was performed for all retained studies after application of the inclusion and exclusion criteria by the methodologist (see Supplemental Table 6). Using that method, studies deemed to be of low quality would not be excluded from the SR but would be retained and their methodological strengths and weaknesses discussed where relevant. Studies would be assessed by confirming the presence of items related to both internal and external validity, which are all associated with methodological rigor and a decrease in the risk of bias. The quality assessment of the studies was performed by determining the risk of bias by assessing key indicators based on study design against known criteria. Only studies obtained from our SR were assessed for quality by these methods and any additional articles brought in to support the background and to contextualize the findings were not. Each study was assessed individually (refer to the SDC for individual assessments and results by guideline statement) and then summarized by study type. A summary of the overall quality of the evidence was given considering the evidence in totality.

A rating for the strength of evidence is given for guideline statements for which quality was assessed (ie, only studies obtained from our SR). Ultimately, the designation (rating) of the strength of evidence is a judgment by the expert panel of their level of confidence that the evidence from the studies informing the recommendations reflects a true effect. Table 1 describes the grades for strength of evidence. (Refer to the SDC for a detailed discussion of the quality assessment.)

Development of recommendations required that the EP review the identified evidence and make a series of key judgments, including the balance of benefits and harms. Grades for strength of recommendations were developed by the CAP Pathology and Laboratory Quality Center and are described in Table 2.

This guideline will be reviewed every 4 years, or earlier in the event of the publication of substantive and high-quality evidence that could potentially alter the original guideline recommendations. If necessary, the entire EP will reconvene to discuss potential changes. When appropriate, the EP will recommend revision of the guideline to the CAP and ASH for review and approval.

Practice guidelines and consensus statements reflect the best available evidence and expert consensus supported in practice. They are intended to assist physicians and patients in clinical decision-making and to identify questions and settings for further research. With the rapid flow of scientific information, new evidence may emerge between the time a practice guideline or consensus statement is developed and when it is published or read. Guidelines and statements are not continually updated and may not reflect the most recent evidence. Guidelines and statements address only the topics specifically identified therein and are not applicable to other interventions, diseases, or stages of diseases. Furthermore, guidelines and statements cannot account for individual variation among patients and cannot be considered inclusive of all proper methods of care or exclusive of other treatments. It is the responsibility of the treating physician or other health care provider, relying on independent experience and knowledge, to determine the best course of treatment for the patient. Accordingly, adherence to any practice guideline or consensus statement is voluntary, with the ultimate determination regarding its application to be made by the physician in light of each patient's individual circumstances and preferences. The CAP and ASH makes no warranty, express or implied, regarding guidelines and statements and specifically excludes any warranties of merchantability and fitness for a particular use or purpose. The CAP and ASH assumes no responsibility for any injury or damage to persons or property arising out of, or related to any, use of this statement or for any errors or omissions.

Of the 4901 unique studies identified in the SR, 174 published, peer-reviewed articles were included, which underwent data extraction and qualitative analysis. Among the extracted documents, 55 articles/documents did not meet any inclusion criteria and were excluded from the SR but retained for discussion purposes.

The EP met 23 times through teleconference webinars from June 8, 2011, through August 16, 2016. Additional work was completed via email. The panel met in person July 19, 2013, to review evidence to date and draft recommendations.

A public comment period was held from August 10 through August 31, 2015, on the ASH Web site. Twenty-nine draft recommendations and 2 demographic questions were posted for peer review.

Agree and disagree responses were captured for every proposed recommendation. The Web site also received 789 written comments. Twenty-six draft recommendations achieved more than 90% agreement, 2 draft statements achieved more than 80% to 90% agreement, and 1 received more than 70% to 80% agreement. Each EP member was assigned 3 draft statements for which they had to review the public comments and present them to the entire panel for group discussion. After consideration of the comments, 2 draft recommendations were maintained with the original language, 25 were revised, and 2 draft recommendations were combined into other statements, which resulted in 27 final recommendations.

The panel convened again September 14, 2015, to review the comments received and revise the recommendations. Resolution of all changes was obtained by unanimous consensus of the panel members using a nominal group technique (rounds of subsequent teleconference webinars and email discussions). Final EP recommendations were approved by a formal vote. The panel considered laboratory efficiency and feasibility throughout the entire process, although neither cost nor cost-effectiveness analyses were performed.

An independent review panel, masked to the EP and vetted through the conflict of interest process, provided a review of the guideline and recommended the guideline for approval by the CAP Council on Scientific Affairs and the ASH Executive Committee. The final recommendations (guideline statements) are summarized in Table 3.

The treating clinician should provide relevant clinical data or ensure that they are readily accessible by the pathologist.

These data include, but are not limited to, the patient's age; sex; ethnicity; history of any hematologic disorder or known predisposing conditions or syndromes; any prior malignancy; exposure to cytotoxic therapy, immunotherapy, radiotherapy, or other possibly toxic substances; and any additional clinical findings of diagnostic or prognostic importance. The treating clinician should also include any history of possibly confounding factors, such as recent growth factor therapy, transfusions, or other medications that might obscure or mimic the features of AL. The treating clinician should also obtain and provide information regarding any family history of any hematologic disorder or other malignancies.

The strength of evidence was convincing to support this guideline statement.

Twenty-eight studies, comprising 2 nonrandomized clinical trials (NRCTs)^10,11  and 26 prospective cohort studies (PCSs)^12–37  support including data on age. Most of the PCSs had a risk of bias determination of low to moderate, except for 3 studies determined to be low^12,24,30  and 6 determined to be moderate.^{14,22,26,29,32,33}  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 7 in the SDC for the quality-assessment results for the studies included for statement 1. For the quality assessment and summaries of study data for family history, ethnicity, and performance status, for which there were fewer studies identified in our SR, refer to Supplemental Tables 1 and 2.

Although it may seem that inspection of blood and marrow samples is sufficient to make a diagnosis of AL, in fact, clinical information is often essential for the correct diagnosis, classification, and/or determination of prognosis. For example, a patient with a high white blood cell (WBC) count and 20% blasts in a peripheral blood (PB) leukocyte differential might be erroneously diagnosed as having AL if the pathologist is not aware that the blood was from a 1-week old baby with Down syndrome, in which case, the more likely diagnosis is transient abnormal myelopoiesis rather than AL.^38–40  Further, if the clinician and/or pathologist are not aware that a newly diagnosed patient with AML has a strong family history of leukemia or other hematologic abnormalities, appropriate genetic testing may not be performed to confirm a myeloid neoplasm with a germline predisposition, which, if present, is important not only for genetic counseling of the patient's family but also for the selection of family members as potential donors for hematopoietic stem cell therapy for the patient.^8,41  Although somewhat unusual, these examples illustrate the importance of a detailed clinical history, including information regarding possible predisposing factors, such as a family history of hematopoietic neoplasms or other hematologic abnormalities, exposure to cytotoxic therapies or other leukemogenic toxins, and exposure to any medications or known factors that might mimic the clinical and morphologic features of AL.

The most routine and basic clinical information—the patient's age and sex—are important because of their effect on prognosis in AL. In our SR of the literature, age emerged as a statistically significant prognostic factor in AL.^10,-37 In ALL children 1 to 9 years old generally had a more-favorable outcome than those younger than 1 or older than 10 years,^{17,22,27,34,36}  whereas, in AML, patients 60 years old or older have worse outcomes compared to younger patients.^{10,16,23,26,28,34,37} 

Published evidence revealed that sex was also prognostically important. In childhood ALL, males tended to have a worse overall prognosis than females did,^19,42  although that difference was not as clear in adult ALL.⁴³  In AML, males fared worse than females did.^44,45 

Although familial acute leukemia is generally regarded as rare, an inherited predisposition to hematopoietic neoplasms—including AL—is likely more common than appreciated and can only be recognized by detailed clinical information.⁴¹  Because of the importance of hematopoietic neoplasms with germline predisposition for genetic counseling and for the detection of family members as potential donors for hematopoietic stem cell therapy, verified cases of inherited-predisposition syndromes should be documented in the patient's medical record and in the pathology report. A number of predisposing syndromes with germline mutations have been described and included in the most recent revision of the WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues⁸; the most well-known of which include AML with germline CEBPA mutation,^46,47  myeloid or lymphoid neoplasms with germline RUNX1 mutation,^48,51  myeloid or lymphoid neoplasms with germline ANKRD26 mutation,^52,53  and myeloid neoplasms with GATA2 mutation,^54,58  among others. In addition, AL arising in patients with a background of inherited bone marrow (BM) failure syndromes, such as dyskeratosis congenita and other telomerase biology diseases, and Fanconi anemia, should be recognized in the medical record.^41,59,60  Family histories that include a first-degree or second-degree relative with AML, ALL, myelodysplastic syndrome (MDS), persistent thrombocytopenia, clinical bleeding propensity, immunodeficiency, or a hematologic malignancy at a young age are important clues for AL with germline predisposition.^41,61,62 

Therapy-related myeloid neoplasms, including therapy-related AML (t-AML), are late complications of cytotoxic chemotherapy and/or radiotherapy administered for a prior neoplastic or nonneoplastic disorder.^63,67  Currently, t-AML comprises nearly 15% to 20% of all cases of AML, but the incidence is rising as more patients survive treatment for their initial cancers.⁶⁸  Therapy-related lymphoblastic leukemia has been reported but is much less common than t-AML.^69,70 

In the WHO classification system, therapy-related myeloid neoplasms are recognized as a distinct category for patients who have a history of prior exposure to alkylating agents, topoisomerase II inhibitors, antimetabolites, antitubulin agents, and/or ionizing radiation.⁸ 

Information regarding the specific therapy implicated in the pathogenesis of therapy-related myeloid neoplasms is important because of the correlations between the clinical, morphological, genetic findings, and prognosis with the prior therapeutic regimen.^{65,67,68,71–73} 

Although most cases of t-AML are thought to be related to mutational events induced by prior cytotoxic therapy, the exact mechanisms and pathways involved are not clear. Most patients treated with cytotoxic therapies do not develop t-AML, suggesting there may be underlying predisposing genetic factors.^74,75  Thus, the history of previous cytotoxic therapy in a patient with t-AML or therapy-related lymphoblastic leukemia is important for diagnosis and classification but also, perhaps, for identification of an inherited predisposition to drug-induced cancer.⁷⁶ 

As noted in the preceding section, a history of radiation therapy—either alone or in combination with chemotherapy for prior neoplastic or nonneoplastic conditions—is recognized as being associated with an increased risk for AL, particularly AML. Radiation exposure for individuals near natural disasters or atomic bomb explosions is also associated with an increased risk of leukemia,^77,78  and reportedly, radiation exposure after diagnostic procedures, including computed tomography scans in children, increases the risk for leukemia.^79,80 

Patients receiving hematopoietic stem cell therapy are at increased risk for development of MDS and AML. Such patients usually receive a combination of chemotherapy and radiation therapy, and the rate of developing MDS or AML ranges from 2% to 7.6% in the studies reviewed.^81–83  The incidence appears to be increased in older patients and in patients who received total-body radiation.⁷⁸ 

Exposure to certain chemicals is associated with an increased risk of development of AL, particularly AML. Benzene exposure, especially at high levels, is associated with an increased risk of AML.^84,85  Other exposures are more controversial. Embalmers and funeral-home workers exposed to formaldehyde are reported to have an increased mortality rate from AML,⁸⁶  but a more recent meta-analysis found no such increase in risk of leukemia for workers exposed to formaldehyde when results were adjusted for smoking history.⁸⁷ 

The use of recombinant granulocytic growth factors, such as granulocyte colony-stimulating factor and granulocyte-macrophage colony-stimulating factor, may transiently increase blasts in the blood and/or BM, which, in some cases, may account for 20% or more of the cells and lead to an erroneous diagnosis of AML. The increase in blasts may persist up to 5 weeks after cessation of growth factor therapy.⁸⁸  The pathologist should be aware of any growth factor or other cytokine therapy and its time of administration relative to the BM or blood sample in question to ascertain whether an elevated blast count is due to AML or could be attributable to a transient growth factor effect.

In addition, vitamin B12 or folate deficiency can markedly alter the BM appearance because of a proliferation of immature erythroblasts, potentially mimicking acute erythroid leukemia. It is critical to exclude vitamin B12 or folate deficiency before making a diagnosis of AL if the BM shows numerous blasts with erythroid features.^89,90 

There were 200 respondents to statement 1, of whom, 97% (n = 194) agreed, and 3% (n = 6) disagreed with the statement. The reason for the 3% disagreement was not clear. There were 42 written responses; most of which suggested that the statement include even more-specific and more-detailed information and that the data should reside in the patient's electronic medical record, where it would be accessible to the pathologists and treating physicians. The comments were considered in the final draft of statement 1 in this document.

There is strong evidence, based on our SR, as well as in literature gathered outside of our SR, to support statement 1. The clinical history is the starting point for the workup of AL and provides information that may be necessary for diagnosis, classification, and prognosis. In addition to the literature, the statement was enthusiastically supported—almost unanimously—by the respondents during the open comment period. Refer to Table 4 for study data on age.

The treating clinician should provide relevant physical examination and imaging findings or ensure that those results are readily accessible by the pathologist.

This includes, but is not limited to, neurologic exam findings and the presence of tumor masses (eg, mediastinal), other tissue lesions (eg, cutaneous), and/or organomegaly.

The strength of evidence was convincing to support this recommendation.

This recommendation was supported by 4 PCSs.^19,20,22,91  One study was deemed to have a low risk of bias,⁹¹  2 were deemed to have a low to moderate risk,^19,20  and one was deemed to have a moderate risk of bias.²²  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings.

This recommendation was based on evidence from our SR as well as from expert consensus opinion. Evidence was available from the SR only for the relevance of central nervous system (CNS) involvement—which typically manifests as cranial nerve abnormalities or meningeal symptoms—for the outcome of AL (refer to Supplemental Table 3); based on expert opinion, other studies, which did not meet the criteria for SR, informed the recommendation for tumor masses, cutaneous lesions, and organomegaly.

Our SR identified 4 PCSs relevant to CNS status at presentation and outcomes in lymphoblastic leukemia.^19,20,22,91  In a study of 4959 noninfant patients with B-precursor ALL treated between 1986 and 1999 by various protocols, patients having CNS disease at presentation were associated with poor outcomes in a multivariate analysis (hazard ratio [HR], 1.34; P = .04).¹⁹  A Nordic study of 2668 children with ALL, who were treated in 2 successive trials between 1992 and 2007, showed a significantly increased risk of treatment failure (P < .01 in both study cohorts) if CNS disease (defined as ≥5 × 10³/mm³ leukemic cells in the diagnostic spinal tap) was present at presentation.⁹¹  A study of 546 consecutive, pediatric patients from St Jude Children's Research Hospital (Memphis, Tennessee) treated from 1984 to 1991 with 2 protocols showed generally worse event-free survival (EFS) and overall survival (OS) when CNS disease was present at presentation (study 11: EFS, P < .001, OS, P = .01; study 12: EFS, P = .03, OS, P = .22).²⁰  In subsequent studies, improvements in identifying those at high risk of CNS relapse and with reinduction treatment as an integral component of overall therapy, differences in EFS and OS were lessened between those with and those without CNS involvement.²⁰  A Brazilian study of 229 consecutive children with ALL treated with a single protocol, for whom 220 were evaluable for CNS status at presentation, reported a statistically significant difference (P < .001) in 5-year EFS between children with (n = 9; 5-year EFS mean [SD], 79.4% ± 3.1) and without (n = 211; 5-year EFS mean [SD], 40% ± 17.4) CNS involvement.²²  Refer to Supplemental Table 8 for the quality-assessment results of included studies for statement 2.

Acute leukemia may involve extramedullary sites, such as mediastinum (thymus and lymph nodes) and skin at presentation; those sites may be involved before, or concurrent with, BM and PB cells and may be sites of disease relapse. In ALL, an anterior mediastinal mass is present in 8% to 10% of childhood cases and in 15% of adult cases.⁹²  Mediastinal enlargement visualized on imaging studies may point to T-lymphoblastic lymphoma/leukemia,⁹³  whereas mediastinal adenopathy was associated with inferior survival outcome in pediatric ALL (P = .01) in one PCS.⁹⁴  Myeloid sarcoma and leukemia cutis are 2 manifestations more commonly associated with AML and usually require a tissue biopsy for diagnosis⁹⁵ ; lymphoblastic leukemias can also involve the skin.^96–98  Enlargement of liver and spleen are most common sites of extramedullary involvement in ALL with marked organomegaly being more frequent in children and uncommon in adults.⁹²  In AML, palpable splenomegaly or hepatomegaly occurs in about one-quarter of patients.⁹⁹  Knowledge of involvement at those sites can assist the pathologist in making an accurate diagnosis, offering recommendations for sampling other sites, and performing ancillary testing.

The EP noted that the physical examination and imaging information was typically obtained in the course of routine workup and making such information available should pose no additional burden on the clinician submitting the samples.

There were 195 respondents, of whom, 97.44% (n = 190) agreed and 2.56% (n = 5) disagreed with statement 2. There were 21 written comments, including a number that suggested adding imaging studies and that comments regarding mediastinal disease and cutaneous manifestations be specifically mentioned. Others commented that the data should be available through the electronic medical record. These comments were taken into consideration in the final draft of statement 2 presented in this article.

The pathologist should review recent or concurrent complete blood cell (CBC) counts and leukocyte differentials and evaluate a PB smear.

The strength of evidence was convincing to support this guideline statement.

This statement is supported by 51 studies,^* comprising 2 randomized, controlled trials (RCTs)^113,117 ; 5 NRCTs^{11,106,111,112,131} ; and 44 PCSs.^† For the 2 RCTs, the trial by Lange et al¹¹³  was deemed to have a very low risk of bias, and the trial by Schneider et al¹¹⁷  was deemed to have a high risk of bias. For the 5 NRCTs, 4 reported a risk of bias of low to moderate,^{11,111,112,131}  and one reported a moderate risk of bias.¹⁰⁶  For the 44 PCSs, 7 of the studies were deemed to have a low risk of bias,^‡ 30 were deemed to have a low to moderate risk of bias,^§ and 7 were deemed to have a moderate risk of bias.^** None of these studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 9 for the quality-assessment results of studies included for statement 3.

The diagnosis of AL is usually first suspected when a patient presents with symptoms related to an abnormal CBC, such as fatigue or weakness because of low hemoglobin, bleeding, or bruisability from thrombocytopenia, or fever caused by an infection related to neutropenia. Thus, an abnormal CBC is frequently the starting point for the workup and evaluation for AL. The CBC may also identify hematologic abnormalities, such as a dangerously low hemoglobin level or platelet count, for which therapeutic measures are immediately indicated. Lastly, according to the guidelines for application of the WHO classification, inspection of the PB smear is critical because it may provide evidence for a diagnosis of AL as well as provide information on features that aid in its classification.¹³² 

The literature strongly supports the importance of some CBC parameters as independently significant prognostic indicators in AL. In AML, numerous studies show a significant effect of the WBC count on response to therapy, OS, and/or EFS (P ≤ .05) in all studies cited.^†† Similar results are reported in ALL.^‡‡ In some reports of AL, the platelet count,^§§ hemoglobin level,^30,116,117  and the percentage of blasts^14,103  were also independent, significant factors in OS and EFS.

There were 195 respondents, of whom, 96.41% (n = 188) agreed, and 3.59% (n = 7) disagreed with the statement. The reasons for the disagreement responses were not stated. The 31 written comments submitted generally offered strong endorsement for the statement. Some emphasized that the pathologist should review the specimen personally and not rely on reported data; a few indicated the specimen review should be performed by the treating clinician. These comments were taken into consideration for the final draft of statement 3 that is presented in this article.

The knowledge gained from evaluation of the CBC and review of the PB provides information that is important for diagnostic, prognostic, and classification purposes. The evidence obtained from our SR strongly supports this statement, and opinions gathered during the open comment period were also supportive.

Refer to Table 5 for study data on CBCs.

The treating clinician or pathologist should obtain a fresh BM aspirate for all patients suspected of AL, a portion of which should be used to make BM aspirate specimens for morphologic evaluation. If performed, the pathologist should evaluate an adequate BM trephine core biopsy, BM trephine touch preparations, and/or marrow clots, in conjunction with the BM aspirates.

If BM aspirate material is inadequate or if there is compelling clinical reason to avoid BM examination, PB may be used for diagnosis and ancillary studies if sufficient numbers of blasts are present. If a BM aspirate is unobtainable, touch-imprint preparations of a core biopsy should be prepared and evaluated, and an additional core biopsy may be submitted, unfixed, in tissue culture medium for disaggregation for flow and genetic studies. Optimally, the same physician should interpret the BM aspirate specimens and the core biopsy specimens, or the interpretations of these specimens should be correlated if performed by different physicians.

Numerous studies confirm the utility of BM aspirate specimens in the diagnosis of AL. The BM aspirate is the optimal specimen for both blast enumeration and dysplasia assessment for the myeloid and erythroid lineages.¹³²  The confirmation of a blast percentage of at least 20% in blood or BM is a WHO requirement for AL diagnosis.¹³²  A manual differential cell count performed on BM aspirate specimens is the standard procedure for blast enumeration.

An adequate BM trephine core biopsy provides essential diagnostic information in patients with AL, including overall cellularity, assessment of residual hematopoietic cells, extent of leukemia effacement of the BM as well as distinctive features that could affect treatment response, such as necrosis and fibrosis.^132–138  In addition, numerous special stains can be performed on a BM core biopsy section that can provide both diagnostic and prognostic information in patients with AL.^{132,135–138}  The breadth of potential stains includes numerous immunohistochemical stains and some in situ hybridization stains (ones not adversely affected by decalcification), and DNA-based molecular studies can also be performed on the BM core or on clot biopsy sections. Core biopsy sections assume a greater role in leukemia diagnosis when an adequate BM aspirate is not obtained.

Adequacy of the BM trephine core biopsy specimen has been assessed in several studies. In general, an intact core biopsy specimen that is 1 cm or larger is considered optimal for diagnosis.^{133–135,137,139}  This specimen cannot consist largely of cortical or subcortical regions but, instead, must contain intact hematopoietic regions of the BM.

Touch preparations of the BM core biopsy can potentially facilitate the AL diagnosis in a variety of ways.^{132–135,138}  Because the touch preparations can be readily prepared before the BM core biopsy specimen is placed in fixative, they should be made every time a BM core biopsy is obtained. The cells in those touch preparations can be evaluated by Wright-Giemsa stain; differential cell counts can be performed as well as dysplasia assessment.^{132–134,138}  In addition, these touch-preparation slides can be used for cytochemical stains, fluorescence in situ hybridization (FISH), and molecular studies. Touch-preparation slides are even more essential in cases in which the BM aspiration is unsuccessful. This is not an uncommon event in AL cases, and, in that situation, the touch preparation slides become the key BM specimen for the blast enumeration, which is essential for AL diagnosis. The utility of the touch preparation in BM diagnosis has been confirmed by multiple comparative studies.^132,135,138 

Once adequate numbers of BM aspirate specimen slides have been prepared, the remaining BM aspirate specimen clots into a semisolid tissue specimen, which can be wrapped in filter paper and submitted for routine tissue processing. Because there are no bony trabeculae in this coagulated aspirate specimen, decalcification is not necessary. Thus, the value of the clot sections is enhanced because it can be used for the full breadth of molecular studies as well as for all special stains, immunohistochemical stains, and for in situ hybridization.^{132–134,138} 

The BM clot section can be used for any diagnostic technique, including many molecular diagnostic techniques that have been validated for formalin-fixed, paraffin-embedded (FFPE) tissue.^132,133,138  The BM clot section assumes a greater role in AL diagnosis when an adequate BM core trephine biopsy specimen has not been obtained.

There were 186 respondents, 94.09% (n = 175) of whom agreed with this recommendation, and 5.91% (n = 11) who disagreed. The reasons for the disagreement were stated by only 7 respondents and were either (1) that, in some cases, the number of blasts in the PB were sufficient for diagnosis and for all required ancillary studies, and thus, BM studies were not required (5 respondents); or (2) disagreement as to whether the specimens should be reviewed by a hematologist or a pathologist (2 respondents). These comments were considered in the final draft of statement 4 presented in this article.

In addition to morphologic assessment (blood and BM), the pathologist or treating clinician should obtain sufficient samples and perform conventional cytogenetic analysis (ie, karyotype), appropriate molecular-genetic and/or FISH testing, and FCI. The flow cytometry panel should be sufficient to distinguish acute myeloid leukemia (including acute promyelocytic leukemia), T-ALL (including early T-cell precursor leukemias), B-cell precursor ALL (B-ALL), and AL of ambiguous lineage for all patients diagnosed with AL. Molecular genetic and/or FISH testing does not, however, replace conventional cytogenetic analysis.

If sufficient BM aspirate or PB material is not available for FCI, immunohistochemical studies may be used as an alternative method for performing limited immunophenotyping. In addition, a second BM core biopsy can be obtained and submitted, unfixed, in tissue culture media for disaggregation for genetic studies and flow cytometry.

No studies from our SR directly informed this statement.

Specialized testing is essential in the diagnosis of AL and provides necessary prognostic information and a “fingerprint” of the neoplastic clone that can be used for optimal minimal residual disease (MRD) monitoring.

Flow cytometry immunophenotyping has an essential role in the diagnosis and classification of AL. Together with cytomorphology and cytochemistry, FCI is crucial for the detection of blasts and lineage assignment of blast cells and to define AL of ambiguous lineage. In addition, specific immunophenotypic profiles have been associated with prognosis and/or unique cytogenetic and molecular abnormalities, such as AML with t(15;17)(q24.1;q21.2), t(8;21)(q22;q22.1).

An adequate FCI panel should be able to determine not only AML, B-ALL, and T-ALL but also AL of ambiguous lineages, including acute, undifferentiated leukemia and MPAL. Acute leukemia of ambiguous lineage has been confirmed to be a poor-risk disease.¹⁴⁰  Furthermore, within each subset of AL, FCI panel/markers should be able to effectively subcategorize the blasts. In AML, FCI may help to determine blasts with monocytic differentiation and myeloperoxidase (MPO) expression¹⁴¹  and to recognize acute promyelocytic leukemia (APL) or APL mimics^142–144  and blasts with erythroid¹⁴⁵  or megakaryoblastic differentiation.¹⁴⁶  In T-ALL, the FCI panel should include sufficient markers to identify early T-cell precursor lymphoblastic leukemia.¹⁴⁷  In addition, FCI may provide therapeutic marker measurement, such as CD20 in B-ALL, for a frontline rituximab-containing regimen.¹⁴⁸ 

Although no standard FCI panels are mandated for all laboratories, there are recommendations for instrumentation, preanalytic variables, panel design, data analysis, and validation by the EuroFlow Consortium (Leiden, the Netherlands),¹⁴⁹  the British Committee for Standards in Haematology (London, United Kingdom),¹⁵⁰  and the International Clinical Cytometry Society (Glenview, Illinois).^151,152 

The role of cytogenetics in diagnosis, classification, and prognosis in AL is well established and predates the period used for our SR.^153,154  During the past 30 years, cytogenetic studies have become an integral part of the diagnosis, prognosis, and treatment of AML and ALL in children and adults.^155–157  Studies that preceded the dates of our SR and expert opinion informed this recommendation.

The value of conventional cytogenetic studies as a critical prognostic indicator in AL has been proven in numerous clinical trials in AML^*** and ALL^††† and now provides a basis for classification¹⁷⁴  and choice of initial and postremission therapy.^175–178  With testing widely available in academic and reference laboratories, conventional cytogenetics reveals a clonal abnormality in 40% to 50% of patients with AML and in 60% to 85% of patients with ALL; the success rate at diagnosis is typically in excess of 84% for ALL and 90% for AML in experienced laboratories.^156,157,179  Importantly, new karyotypic abnormalities continue to be described that may not be apparent by other routine techniques.^179,180  Including cytogenetic analysis as part of the diagnostic workup of AML, ALL, and AL of ambiguous lineage is endorsed by the National Comprehensive Cancer Network (Fort Washington, Pennsylvania) clinical practice guidelines¹⁸¹  and in the (now archived) British Committee for Standards in Haematology Guidelines on the Management of Acute Myeloid Leukaemia in Adults.¹⁸² 

Molecular genetic and/or FISH testing should be considered complementary to an adequate conventional cytogenetic analysis. Unless the cytogenetic analysis is suboptimal because of poor chromosome morphology or insufficient cells for analysis or is completely unsuccessful because of no growth, FISH analysis may be an expensive, redundant technology, particularly in AML.^183–185  On the other hand, certain abnormalities encountered in ALL, such as t(12;21)(p13.2;q22.1) ETV6-RUNX1 fusion or intrachromosomal amplification of chromosome 21 (iAMP21), can be cytogenetically cryptic and are optimally detected by interphase or metaphase FISH.^183,186  In general, the utility of FISH should be considered in the context of each case, and to that end, algorithmic approaches for using conventional cytogenetics and FISH have been proposed by several groups.^{185,187–189} 

Because of the critical importance of FCI and various genetic studies in AL diagnosis, prognosis, and disease monitoring, it is essential that a battery of specialized tests be performed on all cases. Consequently, a concerted effort to obtain adequate specimens for specialized studies is mandatory.

There were 186 respondents, of whom, 94.09% (n = 175) agreed, and 5.91% (n = 11) disagreed with the statement. No clearly defined reasons for disagreeing were stated. There were 34 written comments, including one comment that only qualified cytogeneticists or pathologists subspecialized in cytogenetics, rather than “pathologists” should “perform” the cytogenetic studies, and another comment said that targeted FISH and molecular methodologies should replace conventional karyotyping for patients with ALL. The comments were considered in the final draft of statement 5 presented in this article.

For patients with suspected or confirmed AL, the pathologist may request and evaluate cytochemical studies to assist in the diagnosis and classification of AML.

The strength of evidence was insufficient for this statement.

No evidence from our SR informed this statement.

Cytochemical stains were historically the primary laboratory adjunct to routine morphology for classification of AL. Other techniques, particularly immunophenotyping, have largely supplanted them. Nonetheless, cytochemical studies still have some utility in some circumstances.

Evidence identified external to our SR indicated that 2 cytochemical stains have continued utility in AML diagnosis: MPO and nonspecific esterase stains. Both can be performed on air-dried, unfixed, unstained slides of blood, imprint specimens, and BM aspirate specimens.^133,136,190  Myeloperoxidase is uniquely valuable because it can be performed within 5 to 10 minutes and is available 24 hours a day, 7 days a week in many laboratories.¹⁹⁰  The rapid diagnosis of APL is considered to be a medical emergency because of the risk of major hemorrhage,¹⁹¹  and MPO staining can be particularly helpful in cases of the microgranular variant of APL. The detection of MPO positivity in 3% or more of the blasts is a criteria for myeloid lineage delineation in AML.¹⁹²  Similarly, nonspecific esterase positivity is one of the defining criteria for monoblast/promonocyte identification in AML.¹⁹²  Other historically used stains, including Sudan black B, periodic acid–Schiff, or acid phosphatase, were not considered to have sufficient specificity for routine use in the classification of AL.

There were 185 respondents, of whom, 92.97% (n = 172) agreed, and 7.03% (n = 13) disagreed. Although the reason for the disagreements was not clearly stated, there were 29 written comments. Three individuals explicitly commented that the recommendation should have been stronger, whereas another 3 commented that cytochemical stains were no longer necessary. These comments were considered in the final draft of statement 6 in this article.

In part because there are no studies showing the independent value of these stains, but recognizing that the speed and low cost of these techniques can have utility in some circumstances, expert consensus opinion supports the optional use of these stains.

The treating clinician or pathologist may use cryopreserved cells or nucleic acid, nondecalcified FFPE tissue, or unstained marrow aspirate or PB specimens obtained and prepared from PB, BM aspirate or other involved tissues for molecular or genetic studies in which the use of such material has been validated. Such specimens must be properly identified and stored under appropriate conditions in a laboratory that is in compliance with regulatory and/or accreditation requirements.

The strength of evidence was adequate to support this guideline statement.

This statement was supported by 4 PCSs^193–196  that met the inclusion criteria for our SR. Two of the studies^193,194  were deemed to have a low risk of bias, and the other 2^195,196 were deemed to have a low to moderate risk of bias. None of these studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 10 for the quality-assessment results of studies included studies in statement 7.

The purpose of this guideline statement is to encourage preservation of cells from blood, BM, or other tissues that can be used to identify molecular, genetic, and/or antigenic abnormalities of leukemia cells that may be of further diagnostic or prognostic importance or that may be a target for a specific therapeutic agent but that were not assessed during the initial evaluation. The availability of preserved leukemic cells could spare the patient an additional BM procedure if there are no blasts in the PB, or, if, after therapy, there are few or no neoplastic cells in the marrow. This recommendation applies to initial diagnostic specimens as well as subsequent specimens with evidence of residual or recurrent disease.

Our SR provided no data that specifically informs this recommendation. Preserved cells are, however, invaluable for clinical research and are often used in cooperative group studies for clinical trials in which genetic abnormalities are performed in a central laboratory. Nevertheless, only a few studies in the SR specifically mention or allude to the use of preserved cells as the specimen source for the studies performed.^193–196 

Historically, cryopreservation was the most commonly employed method of preserving cells.¹⁹⁷  Cryopreserved leukemia cells can be used for the extraction of DNA and RNA for molecular genetic studies.^198–200  Although karyotypic analyses of cryopreserved cells have been reported, the number of analyzable metaphases is fewer and their quality is inferior to those from fresh samples.^201,202  Thus, fresh cells are clearly preferable for karyotyping, and cryopreserved cells should be used only when no other cells are obtainable, and the results should be considered with caution. In contrast, cryopreserved cells can be successfully studied for specific chromosomal abnormalities by FISH,^203,204  which can also be applied to properly stored specimens and cytospin or touch preparations on glass slides. Immunophenotyping by flow cytometry may be performed on cryopreserved cells, although some antigens or cells expressing those antigens may deteriorate during the freeze-thaw cycle. These latter, detrimental effects appear to depend, in part, on the protocols used for cryopreservation and cell storage and, perhaps, on the cell lineage.^205–216 

Cryopreservation procedures require mononuclear cells be suspended in a solution containing a cryoprotective agent (usually dimethyl sulfoxide), cooled in a cooling device to the storage temperature, and then stored in liquid nitrogen (for use in studies in which viable cells are necessary) or a −80° freezer, which is satisfactory for most molecular studies. Cryopreservation does require specialized equipment and storage facilities, which may not be available in all laboratories.

Nucleic acid extraction from FFPE tissues can be used for molecular studies as well, particularly for NGS and microarray technologies. In general, the use of FFPE cells has been most successful for DNA-based analyses,^217–221  whereas RNA extracted from such specimens is fragmented by formalin fixation and is often of poor quality.^222–225  Nevertheless, the expression pattern of small RNAs, eg, miRNAs, extracted from FFPE is reportedly similar to that derived from cryopreserved cells.^226,227  DNA and RNA for molecular studies may also be obtained from archived cytology and specimen preparations on glass slides.^228–230 

There were 184 respondents to this statement, of whom, 97.28% (n = 179) agreed, and 2.72% (n = 5) disagreed. There was no specific issue identified by those who disagreed. However, there were 19 written comments, among which, were 4 that emphasized that the preserved specimens and cells should be held in Clinical Laboratory Improvement Amendments of 1988 (CLIA '88)–approved facilities under controlled conditions and should only be used for studies for which such specimens had been validated. These comments were considered in the final draft of statement 7 in this article.

For patients with ALL receiving intrathecal therapy, the treating clinician should obtain a cerebrospinal fluid (CSF) sample. The treating clinician or pathologist should ensure that a cell count is performed and that examination/enumeration of blasts on a cytocentrifuge preparation is performed and is reviewed by the pathologist.

Acute lymphoblastic leukemia may involve the CNS, both at diagnosis and at relapse, and patients with CNS involvement require specific therapy. For that reason, knowledge of CSF status at the time of diagnosis has long been known to be important for proper management of patients. Because alteration of CNS therapy based on CSF findings has been standard practice for so long, there are no large-scale clinical trials using modern therapy that demonstrate adverse prognosis of CNS leukemia in the absence of therapy, although several have investigated the outcome of patients based on CSF involvement in the context of CNS-directed therapy.

Although there was no evidence from our SR to inform this statement, evidence obtained external to our SR and relevant to this statement includes 2 practice guidelines^231,232  that speak to the need to perform CSF cell count and morphology on patients with ALL, resulting in the now standard classification of CNS involvement as CNS-1 (negative), CNS-2 (blasts with WBC counts <5/mm³), and CNS-3 (blasts with WBC counts ≥5/mm³).

Several retrospective reviews of RCTs of pediatric ALL^233–235  specifically investigated the effect of CNS involvement and concluded that long-term outcome was not affected, provided appropriate therapy was given, although CNS relapse rates among patients with CNS-2 or CNS-3 disease have sometimes been found to be different.²³³  One older study²³⁶  showed that patients with blasts had an adverse prognosis, independent of blast count, but therapy was different in that study. Limited data were identified for adult patients with ALL, although one small study²³⁷  suggested that CSF blasts were associated with adverse outcome.

There were 180 respondents, of whom, 92.22% (n = 166) agreed, and 7.78% (n = 14) disagreed. There were 20 written comments. Although a few respondents did not believe this to be required in all cases, others pointed out that there may be contraindications to obtaining a CSF in some patients, particularly when the peripheral blast count is high and that the initial CSF is typically obtained not at the time of diagnosis, but rather, after the diagnosis was made and intrathecal therapy planned, even if circumstances delayed the timing of the lumbar puncture. In the initial wording of this statement, there was also some confusion about who should obtain the CSF. The issue of obtaining flow cytometry on the CSF fluid was also raised. These important comments were considered in the final draft of statement 8 in this article. The wording was changed to reflect the clinical practice of obtaining CSF at the time of intrathecal therapy, and wording about the clinical contraindication was not included because that was presumed to be covered by the association with administration of intrathecal therapy. Also raised in commentary was the need for flow cytometry, but that is covered in statement 10.

Irrespective of conclusions regarding the independent prognostic significance of finding blasts, it is clear that classification of CNS status requires knowledge of both the cell count and the morphologic assessment for blasts and affects most ALL protocols; thus, we consider this a strong recommendation.

For patients with AL, other than those with ALL, receiving intrathecal therapy, the treating clinician may, under certain circumstances, obtain a CSF sample when there is no clinical contraindication. The treating clinician or pathologist should ensure that a cell count is performed and that examination/enumeration of blasts on a cytocentrifuge preparation is performed and is reviewed by the pathologist.

The strength of evidence was insufficient to support this guideline statement. No data from the SR informed this statement.

This statement is regarding CSF evaluation at the time of diagnosis for patients with AL, other than those who have ALL, undergoing intrathecal therapy/prophylaxis and covered in statement 8. This recommendation is based on expert consensus opinion arising from controversial issues regarding clinical significance, indications, and timing of CSF examination in AL.

The CSF evaluation may be indicated for patients with any CNS signs and symptoms, for those who are suspected of having ocular involvement, for patients with increased risk of CNS involvement or later CNS relapse, or per protocol requirement. Central nervous system involvement at the time AML is diagnosed is uncommon in adults, and the risk factors include younger age, high leukocyte count, high lactate dehydrogenase (LDH) level, African American ethnicity, and 11q23.3/KMT2A abnormalities.^238–241  Because of a low incidence of CNS involvement, routine evaluation is often not recommended for adult patients with asymptomatic AML. The reported incidence of CNS involvement in childhood AML ranges from 6% to 29%^123,242  and is higher in patients younger than 2 years old, in patients with AML and high WBC and peripheral blast counts, in patients with AML and monocytic differentiation, in patients with AML and inv(16)(p13.1q22) or t(16;16)(p13.1;q22), and in patients with AML and hyperdiploidy.^123,242,243  Rarely, CNS disease may present as extramedullary involvement that precedes clinically evident AML.^244–246  Clinical risks for lumbar puncture (LP) in patients diagnosed with AL include increased intracranial pressure and severe coagulopathy. The presence of high numbers of circulating blasts may introduce blasts in the CNS if a traumatic LP occurs.^247,248  An LP should be performed only when there is no clinical contraindication.

If an LP is performed, CSF should be sent for WBC count with differential, total RBC count, and a cytocentrifuge preparation, similar to any other routine CSF examination for suspected CNS tumor involvement. The pathologist should review the slides to determine the presence or absence of blasts via light microscopy. The characteristic features of leukemia cells, either lymphoblasts or myeloblasts, are best seen on Wright-Giemsa–stained preparations, although use of other stains, such as Papanicolaou and Diff-Quik, varies in clinical laboratories. In addition, the detection of blasts via cytomorphology in a low–cell-count, CSF specimen may depend on the sensitivity of the cytocentrifuge used for preparation of the specimen.²⁴⁹ 

Although CNS involvement in ALL is classified into 3 categories as CNS1, CNS2, and CNS3, according to total WBC and RBC counts and the presence or absence of blasts,^231,233  the threshold for a definitive, positive CNS involvement in nonlymphoblastic leukemia varies among different groups. The Italian Cooperative Study Group on Chronic Myeloid Leukemia (Bologna, Italy),²⁵⁰  the Pediatric Oncology Group (Monrovia, California),²⁵¹  and St Jude Children's Research Hospital²³⁸  define CNS involvement as the presence of any blasts, regardless of total cell count, whereas the Children's Cancer Group (Monrovia, California)²⁵²  and the International Berlin-Frankfurt-Münster Study Group (Kiel, Germany)²⁵³  use a WBC threshold of 5 and 10/mm³ with the presence of blasts, respectively. In children, CNS involvement at the time of AML diagnosis is often thought to confer a worse prognosis^252,254,255 ; however, recent studies showed that CNS involvement might not have significant effect on survival.^{123,242,256,257}  A large cohort study by the Children's Oncology Group (Monrovia, California)¹²³  showed that, although CNS involvement at AML diagnosis had no effect on OS, affected patients did have an increased risk of isolated CNS relapse and had an inferior leukemia EFS. On the other hand, the prognostic significance of CNS involvement at the time of AML diagnosis in adults is controversial.^240,241,258  The use of chemotherapeutic agents that offer greater penetration of the CNS, such as high-dose cytarabine and cladribine, may have increased the eradication of low-level CNS involvement without requiring additional CNS-directed therapy.^259,260  Of note, recent induction chemotherapy regimens designed primarily for elderly patients with AML do not include high-dose cytarabine, and whether those induction protocols would result in an increased risk for CNS relapse is unknown.²⁴¹ 

There were 184 respondents, 96.2% (n = 177) of whom agreed with the statement, and 3.8% (n = 7) disagreed. There were 31 written comments. The comments received during the open comment period were related to the indications, timing, and risks of LP. Specifically, a number of commenters suggested deferring LP until after the first cycle of chemotherapy to avoid a traumatic LP in patients with high circulating blasts. Some also advocated performing CSF evaluation in all patients with AL. These comments and concerns have been incorporated in the final draft of statement 9 and are addressed in the preceding text.

For patients with suspected or confirmed AL, the pathologist may use flow cytometry in the evaluation of CSF.

As discussed in statements 8 and 9, examination of the CSF is indicated in cases of ALL and may be indicated in some cases of AML. Definitive determination of CSF involvement by AL is based on identification of blasts by visual inspection of a CSF cytocentrifuge preparation; however, flow cytometry may provide immunophenotypic information to confirm the morphologic impression of the presence of blasts.

No evidence from our SR informed this statement; however, evidence identified outside our SR indicated that flow cytometry can effectively detect disease in CSF samples from patients with ALL^261,262  and can detect subtle leukemic involvement in some cytologically negative CSF samples from both pediatric and adult patients with B-ALL and T-ALL.^263,264  Moreover, patients with ALL and CSF disease not detected by visual inspection of cytocentrifuge preparations but detectable by flow cytometry involvement have shorter OS times than do those with no involvement detected by flow cytometry (P = .01 on multivariate analysis).²⁶⁵  Based on this evidence, a recommendation is made that flow cytometry be performed on CSF samples taken to evaluate for leukemic involvement in patients diagnosed with ALL; although the expert opinion is that AML blasts can also be detected in the CSF by flow cytometry, we identified one study to support that practice for AML.²⁶⁶ 

There were 181 respondents, of whom, 92.82% (n = 168) agreed, and 7.18% (n = 13) who disagreed. There were 27 written comments. Of the 27 who wrote specific comments, 15 thought that the words “pathologists may use flow cytometry” should be changed to “should use flow cytometry.” The question in the survey was, however, slightly different and indicated that “the pathologists may use flow cytometry in the evaluation of CSF when sufficient cells are available.” These comments were taken into consideration in the final draft of statement 10 in this article.

For patients who present with extramedullary disease without BM or blood involvement, the pathologist should evaluate a tissue biopsy and process it for morphologic, immunophenotypic, cytogenetic, and molecular genetic studies, as recommended for the BM.

Additional biopsies may be indicated to obtain fresh material for ancillary testing.

No evidence from our SR informed this statement.

Patients with AL may present initially with extramedullary disease. Extramedullary AL (myeloid sarcoma) has been shown to present months, or even years, before AML becomes evident in the BM or PB in some patients.^267–271  Myeloid sarcoma also can occur in patients with MDS, myeloproliferative neoplasms, or myelodysplastic/myeloproliferative neoplasms whose BM or PB may never meet the diagnostic criteria for AML. The incidence of extramedullary leukemic involvement varies widely among reports in the literature.^{268,269,272–275}  Sites of isolated myeloid sarcoma can occur in the bone, periosteum, soft tissues, lymph nodes, CNS, orbit, intestine, mediastinum, epidural region, testis, uterus, or ovary. Similarly, patients with acute lymphoblastic leukemia/lymphoma (ALL/LBL) may present with extramedullary disease. Patients with T-cell ALL/LBL often present with a mediastinal mass, lymphadenopathy, or other extranodal tissue mass.^276,277  Most patients with B-ALL present with PB and BM leukemia, but about 10% of patients may have an isolated extramedullary presentation.^278,279  Skin, bone, and soft tissue are the most frequently reported sites of extramedullary involvement in B-cell ALL/LBL.²⁸⁰ 

Fresh specimens, either fine-needle aspirate, excisional biopsy, or body-effusion fluids, are adequate for leukemia workup.^{267,281–286}  Similar to the analysis for BM and PB samples, the workup includes morphologic examination, FCI, and cytogenetic and molecular studies. In patients with suspected CNS leukemia,²⁸⁷  a CSF sample may be obtained. Cytocentrifuge preparation with Wright-Giemsa staining should be performed for morphologic examination, and, if sufficient numbers of blasts are present, CSF can be used for ancillary studies. Fine-needle aspiration samples or body fluids with cell-block preparation and excisional or core needle biopsy with a portion of the tissue fixed and paraffin embedded can be used for morphologic examination, immunohistochemistry and targeted FISH, and some mutational studies. The targeted FISH panel and molecular studies for these tissue samples should be performed according to the recommendations for the BM and PB workup for different types of leukemia. For myeloid sarcoma, the frequent cytogenetic abnormalities found in adults are +8, KMT2A translocations, inv(16), +4, −16, del(16q)/loss of 16q, del(5q), del(20q), and +11.^255,286,288  Nucleophosmin (nucleolar phosphoprotein B23, numatrin) (NPM1) mutations have been reported in 15% of patients with extramedullary myeloid tumors, as determined by immunohistochemical studies,²⁸⁹  and fms-related tyrosine kinase 3 internal tandem duplication (FLT3-ITD) mutations have been reported in 15% of cases of myeloid sarcoma.²⁹⁰ 

There were 185 respondents, of whom, 97.84% (n = 181) who agreed, and 2.16% (n = 4) who disagreed. There were 17 written comments. Concerns raised in these comments revolved around several practical issues. First, what are the indications to prompt an AL workup on tissue biopsies or fluid specimens when BM and PB have no sign to suggest AL? Air-dried touch imprints of tissue biopsy specimens or smears/cytocentrifuge preparations of fluids may have great utility in that scenario. Wright-Giemsa–stained slides allow for the assessment of the cytologic features. Cytochemical studies for MPO or nonspecific esterase can also be performed to raise the suspicion, thereby facilitating triage of the specimen for further workup. Touch imprints can also be used for FISH studies when a diagnosis of AL is confirmed.²⁷¹  Of note, in reviews of myeloid sarcoma published in the literature,^272,275,291  nearly 50% of cases were initially misdiagnosed, most often as malignant lymphoma. With the advent of immunohistochemistry, most cases can be diagnosed correctly by applying proper immunohistochemical stains. A number of immunohistochemical markers have been shown to have diagnostic utility in extramedullary leukemia,^{267,281,292,293}  including immature hematopoietic markers, such as CD34, CD117, and TdT; lineage markers, such as CD3, PAX5, CD19, CD79a, MPO, and CD61; monocytic markers; and markers for blastic plasmacytoid dendritic cell neoplasms. The second concern raised was that the biopsy tissue sample may be small, especially in patients with cutaneous involvement, and there may not be sufficient tissue for a complete AL workup. Paraffin-embedded tissue samples should be the priority in such circumstances. Paraffin-embedded tissue is adequate for immunohistochemical analysis, FISH, and some molecular testing.^{267,281–284,286}  When a diagnosis of extramedullary leukemia is confirmed or highly suspected based on the initial biopsy, additional biopsies may be indicated to obtain fresh material for ancillary testing.

Some comments also suggested that BM or PB that contains a substantial proportion of blasts, even less than 20%, is a more desirable material than tissue/fluids for an AL workup. Although this is largely true for lymphoblastic lymphoma/leukemia, it is less clear in myeloid sarcoma. Pileri and colleagues²⁸⁶  compared the results of FISH performed on myeloid sarcoma tissues and conventional cytogenetic analysis performed on BM or PB. A full concordance between the FISH and conventional cytogenetic results was found in 71% of patients with available results. This finding suggests that conventional cytogenetic studies on BM or PB and targeted FISH analysis on myeloid sarcoma are complementary and may be pursued in the appropriate clinical setting. Some responders also suggested that extramedullary tumors need a complete workup, even in patients with 20% or more blasts in BM and PB. Although there is no clear evidence to support or reject that approach, the consensus is that, in a case with a full leukemia workup completed using BM and PB, the workup using tissue samples may primarily focus on confirming the diagnosis.

Based on those findings, the EP concluded that, for patients who present with extramedullary disease without BM or PB involvement, the pathologist should evaluate a tissue biopsy specimen and process it for morphologic, immunophenotypic, cytogenetic, and molecular genetic studies, as recommended for the PB and/or BM.

For patients with suspected or confirmed AL, the pathologist or treating clinician should ensure that flow cytometry analysis or molecular characterization is comprehensive enough to allow subsequent detection of MRD.

The strength of evidence was convincing to support this guideline statement.

This recommendation was supported by 15 studies that met the inclusion criteria for our SR,^{18,20,22,91,294–304}  comprising one RCT³⁰⁴  and 14 PCSs.^{18,20,22,91,294–303}  The single RCT, reported by Yin et al³⁰⁴  in 2012, was deemed to have a moderate risk of bias. For the 14 PCSs, 2 studies^91,299  were deemed to have a low risk of bias, 10 were deemed to have a low to moderate risk of bias,^‡‡‡ and 2 were deemed to have a moderate risk of bias.^22,294  Overall, none of these 15 studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 11 for the quality-assessment results of studies for statement 12.

Minimal residual disease is a powerful predictor of adverse outcome in patients with AL. Although measurement of MRD per se is outside the scope of this recommendation, because it is, by definition, not performed at the time of diagnosis, it is important that the initial diagnostic material be handled in such a way that MRD testing on subsequent samples is possible. Minimal residual disease may be measured by flow cytometry or by molecular testing; the latter encompasses a variety of techniques, including quantitative PCR detection of antigen-receptor rearrangements, of fusion transcripts of leukemic translocations, or of mutated genes. Next-generation sequencing has increasingly been used as a tool for MRD detection.

Nine studies demonstrated the importance of detecting MRD in ALL.^§§§ The studies differed in methodology (PCR versus flow cytometry), the timing of the sample, cutoff values, and the outcome variable measured (EFS, relapse rate, OS), but all showed a statistically significant effect on the outcome variable measured.

There were 4 studies that demonstrated the importance of detecting MRD in AML.^{294,298,301,304}  Two of those studies used reverse transcription-PCR (RT-PCR) and were limited to core-binding factor (CBF) AML.^298,304  The other 2 used flow cytometry.^294,301  All showed that MRD was an important factor in relapse-free survival (RFS) and, in some studies, OS.

One study demonstrated the prognostic significance of MRD detection in MPAL.²⁹⁶ 

Three studies addressed the question of comparing the value of MRD detection by flow cytometry compared with molecular studies in ALL.^22,297,300  In all 3, neither method was found to be superior. There were no studies identified that addressed that question in AML

Numerous publications were identified external to our SR, which offered evidence that information obtained from diagnostic material was important for the subsequent detection of MRD, and only a few are referenced here. Flow cytometric MRD detection in both ALL and AML often depends on the persistence of cells with a particular leukemia-associated immunophenotype identified at the time of diagnosis.^305–308  Even when flow cytometry methods that depend on recognizing differences between normal and abnormal cells are used,^309,310  it is advantageous to be able to compare initial and posttreatment phenotypes because those often change in predictable ways. Molecular methods of detecting MRD in both ALL and AML require that leukemic cells be characterized and sequenced at diagnosis, whether MRD detection is performed by conventional PCR-based techniques^311–313  or by NGS.^314,315 

There were 179 respondents, of whom, 94.41% (n = 169) agreed with the statement, and 5.59% (n = 10) who disagreed. There were 19 written comments. However, for the open comment period, the original draft statement was written to address only patients with suspected or confirmed ALL. Comments offered were largely directed at the conduct and/or timing of subsequent MRD studies and were considered in the final draft of statement 12 for this article.

This statement is designated as having a strong recommendation. Clinicians and pathologists should be mindful, at the time of initial workup of AL, of the requirements for subsequent MRD studies. Much of the molecular testing required can be performed on preserved material obtained as specified in statement 7. However, material is frequently not preserved in such a way that flow cytometry can be readily performed after the fact. Thus, it is imperative that, in settings in which flow cytometric MRD detection is contemplated, initial immunophenotyping be performed in such a manner so as to optimize that testing. Refer to Table 6 for study data on MRD.

For pediatric patients with suspected or confirmed B-ALL, the pathologist or treating clinician should ensure that testing for t(12;21)(p13.2;q22.1); ETV6-RUNX1, t(9;22)(q34.1;q11.2); BCR-ABL1, KMT2A (previously MLL) translocation; iAMP21; and trisomy 4 and 10 is performed.

The strength of evidence was considered adequate to support this guideline statement.

This statement is supported by 6 PCSs^{18–20,91,124,316}  that met the inclusion criteria for our SR. Risk of bias assessments ranged from low^91,316  to low to moderate.^18–20,124  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 12 for the quality-assessment results of studies included for statement 13.

Prognosis in pediatric patients with B-ALL varies with low-risk patient 5-year EFS approaching 90%, whereas high-risk patient 5-year EFS was less than 45%.¹⁷⁰  Risk stratification supports risk-directed therapy to optimize patient care while minimizing unnecessary risks associated with treatment. Factors used for risk assessment include age, WBC count, genetic abnormalities (as in this recommendation), early response to therapy, CNS involvement, and MRD level.

Systematic literature review shows several important markers for risk stratification in pediatric B-ALL, including t(12;21)(p13.2;q22.1); ETV6-RUNX1, t(9;22)(q34.1;q11.2); BCR-ABL1, KMT2A (previously MLL) translocation; iAMP21; and trisomy 4 and 10. The PCSs showed specific prognostic information allowing risk stratification associated with each of those markers. In the most recent clinical trial from each article, the presence of t(12;21)(p13.2;q22.1); ETV6-RUNX1^18,19,91,316 or trisomy of the 4 and 10 chromosomes¹⁹  conferred improved 5-year EFS (P < .001 for both) and OS (P < .001 for both) in cooperative group studies. Interestingly, the more-recent studies (St Jude studies 13A and 13B) in another series failed to show significant prognostic impact for t(12;21) on EFS or OS,²⁰  perhaps because of improved OS for patients with B-ALL resulting from therapeutic improvements and the introduction of MRD evaluation for risk stratification. The t(12;21) is the most common recurrent cytogenetic abnormality in pediatric B-ALL (approximately 25% of patients) and is cryptic by classic cytogenetic evaluation. An alternate method, such as FISH or multiplex RT-PCR, is required for detection of that abnormality and should be performed in all pediatric patients with B-ALL.

Other abnormalities conferring a poor prognosis include t(9;22); BCR-ABL1; hypodiploidy (or decreased DNA index); KMT2A (previously MLL) translocation with slow early treatment response; and iAMP21. However, recent evidence indicates that tyrosine kinase inhibitor (TKI) therapy, combined with intensive chemotherapy, leads to a good outcome in children and adolescents with B-ALL who have the BCR-ABL1 mutation as their sole abnormality.³¹⁷  Fortunately, BCR-ABL1⁺ B-ALL is significantly less common in children than it is in adults, accounting for only 2% to 4% in that patient population. Our SR included PCSs demonstrating decreased 5-year EFS (P < .01)^18–20,91  and OS (P < .01)^18–20  in patients with t(9;22); BCR-ABL1; and t(4;11) but not other 11q23 (KMT2A/MLL) abnormalities.^18–20,91  In addition, iAMP21 was associated with decreased EFS (P < .001)^124,318  and OS (P = .01)¹²⁴; (P = .02)³¹⁸  in pediatric patients with B-ALL, as well as separately in the subset of standard-risk but not high-risk patients.³¹⁸  In addition, intensifying treatment for patients with iAMP21 reduced the likelihood of relapse and improved survival.³¹⁹ 

Pediatric treatment algorithms rely on accurate risk stratification so that patients with higher-risk disease receive appropriately intensified therapy. Genetic abnormalities, as defined in this statement, are an important aspect of therapeutic decision making with t(12;21) and trisomies 4 and 10 conferring improved prognosis, whereas t(9;22)(q34.1;q11.2); BCR-ABL1; KMT2A (previously MLL) translocation; and iAMP21 conferring poor prognosis requiring intensification of therapy. Other genetic markers associated with adult and pediatric B-ALL are discussed in statement 15.

Refer to Table 7 for study data for t(12;21)(p13.2;q22.1); ETV6-RUNX1, t(9;22)(q34.1;q11.2); BCR-ABL1, KMT2A (previously MLL) translocation; iAMP21; and trisomy 4 and 10.

There were 172 respondents, 94.19% (n = 162) of whom agreed with the statement, and 5.81% (n = 10) who disagreed. There were 39 written comments, most of which were supportive but which covered a range of issues including queries regarding specific methodology for detection of the genetic abnormalities, detection of Philadelphia chromosome (Ph)-like B-ALL, the necessity for obtaining BCR-ABL1 transcripts at diagnosis, and the costs of these tests. These comments were taken into consideration in the final draft statement for this article.

For adult patients with suspected or confirmed B-ALL, the pathologist or treating clinician should ensure that testing for t(9;22)(q34.1;q11.2); BCR-ABL1 is performed. In addition, testing for KMT2A (previously MLL) translocations may be performed.

The strength of evidence was adequate to support this guideline statement.

This statement was supported by 2 PCSs^17,171  that met the inclusion criteria for our SR. These studies were deemed to have a low to moderate risk of bias. None of these studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 13 for the quality-assessment results of studies included for statement 14.

The presence of the Ph chromosome or BCR-ABL1 fusion was the most-common, recurrent abnormality in adult B-ALL affecting approximately 25% of patients. This abnormality is an independent risk factor conferring poor prognosis, as demonstrated by several studies and was supported by the Moorman et al¹⁷¹  study that was part of our SR. The t(9;22)(q34.1;q11.2), associated with the BCR-ABL1 fusion, is detected by conventional cytogenetic studies in approximately 95% of patients; however, a molecular genetic method, such as RT-PCR or FISH analysis, is required for detection in the remaining, approximately 5%, of cases. In addition, with the incorporation of BCR-ABL1 TKIs as front-line therapy for Ph⁺ B-ALL, rapid detection of the abnormality is often required and may be best obtained using RT-PCR or FISH analysis, thus, allowing a more-rapid treatment decision in those patients. Demonstrating the BCR-ABL1 fusion by quantitative RT-PCR at the time of diagnosis is necessary if subsequent MRD monitoring by the same method will be used.

In the PCS by Moorman et al,¹⁷¹  1522 patients with ALL were studied, and BCR-ABL1 fusion detected was detected in 19% by conventional cytogenetics, RT-PCR, and/or FISH. Patients with Ph⁺ ALL had significantly inferior 5-year EFS (16% versus 36%) and OS (22% versus 41%) (both P < .001 adjusting for age, sex, and WBC count)¹⁷¹  in comparison to patients lacking the BCR-ABL1 fusion. Those patients were treated on protocols before the incorporation of imatinib for patients with Ph⁺ disease. Studies incorporating TKIs into therapeutic regimens show improved outcome in adults with Ph⁺ B-ALL in comparison to Ph⁺ patients not receiving TKIs^320,321  and suggest that improvement can be enhanced by the addition of hematopoietic cell transplantation.^322,323  Thus, the detection of BCR-ABL1 fusion in adults with B-ALL is essential to determine prognosis and to identify those patients who will benefit from a BCR-ABL1 TKI.

KMT2A (previously MLL) translocations, an abnormality present in approximately 10% of adult patients with ALL, are also considered a poor-risk abnormality in adult patients with B-ALL. In the study by Moorman et al,¹⁷¹  patients with a cytogenetic presence of t(4;11) had significantly inferior EFS (P < .001) and OS (P < .001) when compared with patients with Ph⁻ disease. Other KMT2A translocations did not show statistically significant differences.

The SR also included studies showing the possible prognostic effect of other markers, although those markers did not reach the level of evidence required for recommendation. Of note, one study¹⁷  investigating brain and acute leukemia cytoplasmic (BAALC) expression by RT-PCR showed elevated levels were associated with an immature phenotype and primary therapy resistance in adult patients with B-ALL (P = .01). In addition, patients with BCR-ABL1⁻ or KMT2A⁻ disease with higher BAALC expression had shorter OS rates (P = .03).¹⁷ 

Predicting the prognosis and determining the optimal therapy is important for all patients with AL. In adults with B-ALL, the most significant prognostic factor is the presence of the BCR-ABL1 fusion, a finding associated with a poor prognosis. Optimal therapy for that patient subset requires identification of BCR-ABL1 and initiation of an appropriate TKI therapy. Other genetic markers associated with adult and pediatric B-ALL are discussed in statement 15.

There were 180 respondents, of whom, 95% (n = 171) agreed, and 5% (n = 9) disagreed. There were 26 written comments that were, generally, supportive but were similar to those for statement 13 regarding clarification of specific methodology and the necessity of BCR-ABL1 transcripts at diagnosis for BCR-ABL1⁺ ALL. These comments were considered in the final draft of statement 14 for this article.

For patients with suspected or confirmed ALL, the pathologist or treating clinician may order appropriate mutational analyses for selected genes that influence diagnosis, prognosis, and/or therapeutic management, which includes, but is not limited to, PAX5, JAK1, JAK2, and/or IKZF1 for B-ALL and NOTCH1 and/or FBXW7 for T-ALL. Testing for overexpression of CRLF2 may also be performed for B-ALL.

The strength of evidence was adequate to support this guideline statement.

This recommendation was supported by 14 PCSs.^**** One of the studies was deemed to have a low risk of bias,³³⁴  10 were deemed to have a low to moderate risk of bias,^†††† and 3 were deemed to have a moderate risk of bias.^327,331,332  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 14 for the quality-assessment results of studies included for statement 15.

In addition to gene fusions from chromosomal translocations, such as BCR-ABL1, and numerical abnormalities, such as trisomy 4 and 10 in pediatric B-ALL, a number of gene alterations have also been shown to have independent prognostic and therapeutic effect in ALL. Literature review revealed several genes that may contribute to risk stratification, including PAX5, Janus kinase 1 (JAK1), Janus kinase 2 (JAK2), IKAROS family zinc finger 1 (IKZF1), cytokine receptor-like factor 2 (CRLF2), notch homolog 1 (NOTCH1), and F-box and WD repeat domain containing 7 (FBXW7). We also recognize that information regarding genetic information in ALL is rapidly expanding, and evidence supporting disease-related, relevant markers continues to evolve. Additional markers may also be useful.

PAX5, a transcription factor required for B-lymphoid development and located on chromosome arm 9p, may be important for leukemogenesis, although evidence for independent contribution of PAX5 alterations to patient prognosis has varied. PAX5 alterations occur in approximately 30% to 35% of B-ALL by deletion, fusion translocations, or point mutations. Deletions of PAX5 occur in the greatest proportion of patients when concurrent BCR-ABL1 or TCF3-PBX1 is present.

Our SR included 4 PCSs that addressed PAX5 mutation.^{36,302,325,327}  Only one study³⁶  on adolescent/adult BCR-ABL1⁻ B-ALL showed significantly improved prognosis in a multivariate analysis for patients with PAX5 alterations, including improved EFS (P = .02), RFS (P = .05), and OS (P = .03). Another study in adult B-ALL demonstrated improved complete response (CR) rate (P = .03), although PAX5 alterations had no significant effect on cumulative incidence of relapse and disease-free survival (DFS).³²⁵  The 2 pediatric studies showed no significant difference in prognosis in high-risk B-ALL³⁰²  and no difference in PAX5 alterations when comparing a small relapse-prone cohort with an unselected B-ALL cohort.³²⁷  Although PAX5 abnormalities are common in B-ALL, more data may be needed to determine whether these abnormalities assist with risk stratification. Of note, a small subset of patients with B-ALL (2%–3%) show structural rearrangements resulting in fusion proteins, with PAX5-JAK2 identified as a recurrent abnormality. The presence of PAX5-JAK2 fusion protein can result in the constitutively activated JAK-STAT pathway raising the possibility that this represents a TKI target for therapeutic intervention.

IKZF1 encodes for the IKAROS zinc finger binding protein and is associated with a poor prognosis in B-ALL. Alterations in IKZF1 occur in more than 80% of patients with BCR-ABL1⁺ ALL who may be associated with resistance to TKIs.³³⁵  Our SR included 3 PCSs (1 adolescent/adult³⁶  and 2 pediatric^302,327) and overall supported the poor prognosis of IKZF1 alterations in BCR-ABL1⁻ B-ALL. In adolescent/adult patients with Ph⁻ B-ALL, IKZF1 alterations were associated with inferior EFS (P = .01), DFS (P = .06), and OS (P = .10) in univariate analysis; although significance was not demonstrated in multivariate analysis.³⁶  In pediatric patients with high-risk B-ALL (very high risk and BCR-ABL1 excluded), IKZF1 alteration was associated with elevated MRD (P = .04, day 8; P = .001, day 29) and increased incidence of relapse (P < .001).³⁰²  That study noted similarity between the gene expression signature of high-risk BCR-ABL1⁻ ALL from the original cohort and the gene signature of BCR-ABL1⁺ ALL in the validation cohort. The poor prognosis of IKZF1 was independent of BCR-ABL1 status. Another pediatric study³²⁷  supported the poor prognosis of IKZF1 alterations with an increased proportion of patients with IKZF1 deletion in their small cohort of relapse-prone ALL.

Overexpression of the cytokine receptor CRLF2 is associated with a poor prognosis in B-ALL and often results from translocations of CRLF2 with partner genes, such as the immunoglobulin heavy-chain gene (IGH) or P2RY8. CRLF2 overexpression is seen in 5% to 16% of pediatric and adult B-ALL, more than 50% of Down syndrome ALL, and approximately 50% of BCR-ABL1–like B-ALL (see discussion of BCR-ABL1–like B-ALL below). CRLF2 alterations are also associated with concurrent IKZF1 deletion and/or mutation, JAK1/JAK2 mutations, and a poor prognosis in adult and pediatric B-ALL.^36,330,336  Our SR included 3 PCSs (1 adolescent/adult³⁶  and 2 pediatric^330,332). CRLF2 deregulation in adolescents/adults was associated with decreased 5-year RFS (P = .03) and OS (P = .04).³⁶  In pediatric patients, CRLF2 high-level expression was associated with worse EFS rates (P = .01) and greater cumulative risk of relapse (P = .01), mainly because of the high incidence of relapse in non–high-risk patients with P2RY8-CRLF2.³³⁰  Gene expression profiling in pediatric patients with high-risk B-ALL showed an expression cluster with CRLF2 rearrangements, JAK mutations, IKZF1 deletions, BCR-ABL1–like signature, and a very poor prognosis.³³²  Increasing interest in CRLF2 status has occurred, particularly in light of its high incidence in BCR-ABL1–like ALL. CRLF2 overexpression may be detected by flow cytometry or FISH assays.

Regarding the prognostic significance of JAK1 and JAK2 mutations alone in ALL (without CRLF2), no data were available from our SR.

like (or Ph-like) ALL has been recently recognized and is of particular prognostic importance. These leukemias lack the BCR-ABL1 fusion but have a gene expression profile similar to BCR-ABL1⁺ leukemia and are associated with a poor prognosis. One expression array study in pediatric B-ALL patients identified the BCR-ABL1–like phenotype and showed an increased relapse rate (P < .05) and decreased 5-year DFS (P < .03) in that subset of patients when compared with other forms of B-ALL (BCR-ABL1⁺ ALL excluded).³²⁸  A separate, large study³³⁷  of B-ALL reported a frequency of BCR-ABL1–like ALL ranging from 10% for standard-risk, pediatric patients with ALL, up to 27% among young adults with ALL. In most of those patients (91%) a kinase-activating alteration, such as ABL1, ABL2, CRLF2, JAK2, or platelet-derived growth factor receptor β (PDGFRB), was identified suggesting that at least some patients may benefit from TKI therapy. The most common gene expression alteration identified in BCR-ABL1–like ALL, as well as Down syndrome–associated ALL, was elevated CRLF2 expression, occurring in approximately 50% of patients. Concurrent JAK2 or JAK1, IL7R, FLT3, SH2B3, and NRAS mutations were also present in 30% to 55% of patients with CRLF2 overexpression. Deletions involving IKZF1, PAX5, and EBF1 were also detected in patients with Ph-like ALL.⁸ 

For T-ALL, our SR revealed data about alterations of several genes.

NOTCH1 and FBXW7 mutations frequently occur in T-ALL, and both result in decreased NOTCH1 activity. Some studies suggest an improved early response to therapy and prognosis in these patients, although others lack prognostic significance. Those discrepancies may be the result of different treatment regimens. Our SR included 5 PCSs (3 adult; 2 pediatric). Of those studies, results of studies of adults with T-ALL and NOTCH1 and/or FBXW7 mutations include one showing an improved median EFS (P = .02) and OS (P = .01) in multivariate analysis,³²⁶  whereas 2 showed no prognostic significance,^21,329  although a trend toward improved EFS was seen in one (P = .10).²¹  In pediatric T-ALL, one study identified an improved early response to therapy (P < .01), decreased early (FBXW7 and/or NOTCH1) and late MRD (NOTCH1, P < .01), and improved EFS (NOTCH1; P = .01). Patients with both NOTCH1 and FBXW7 mutations had similar outcomes to those with NOTCH1 mutations alone.³³⁴  The second pediatric study showed improved early response to therapy with NOTCH1 and/or FBXW7 mutation but no difference in EFS or OS.³²⁴ 

JAK1 mutations were associated with reduced DFS (P = .01) and OS (P < .01) in a small cohort of patients with T-ALL,³³¹  but no recommendation regarding JAK1 mutation testing in T-ALL was made because of a lack of multiple or larger studies confirming that report.

Refer to Table 8 for study data on PAX5 and other mutations in patients with B-ALL. Refer to Table 9 for study data on NOTCH1 and FBXW7 mutations in patients with T-ALL.

There were 174 respondents, 90.8% (n = 158) of whom agreed, and 9.2% (n = 16) who disagreed with the statement. There were 32 written comments, most of which were very supportive, but some of which expressed that the panel was too limited, others questioning its clinical utility, and others suggesting the inclusion of copy number aberrations and genetic abnormalities characterizing Ph-like ALL be added.

For pediatric and adult patients with suspected or confirmed AML of any type, the pathologist or treating clinician should ensure that testing for FLT3-ITD is performed. The pathologist or treating clinician may order mutational analysis that includes, but is not limited to, IDH1, IDH2, TET2, WT1, DNMT3A, and/or TP53 for prognostic and/or therapeutic purposes.

The strength of evidence was adequate to support this guideline statement.

The recommendation for FLT3-ITD testing was supported by 13 PCSs^‡‡‡‡ that met the inclusion criteria for our SR and 8 other studies^26,343–349  that were found external to our systematic search (or did not meet the inclusion criteria) but were retained for discussion. Of the 13 studies, one was deemed to have a low risk of bias,³⁴²  10 were deemed to have a low to moderate risk of bias,^§§§§ and 2 were deemed to have a moderate risk of bias.^14,294  None of these studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 15 for the quality-assessment results of studies included on FLT3-ITD testing.

The discovery of gene mutations that affect prognosis in AML was a major advance of the past decade, and the more-recent use of NGS techniques has increased access to mutation panels in the diagnostic setting. Although the prognostic significance of gene mutations were first recognized in patients with normal karyotype AML (NK-AML), it is now recognized that some mutations may define specific disease-classification groups, such as AML with mutated NPM1, AML with biallelic mutations of CEBPA, and the provisional entity of AML with mutated RUNX1, whereas others, such as FLT3, may provide prognostic information across different classification groups.⁸  Mutations in FLT3 most commonly result in ITDs but may also be point mutations in the tyrosine kinase domain. Many gene mutations are now, however, reported in AML,^7,349  creating challenges in understanding which individual genes and/or gene combinations are significant in the disease and warrant testing. Although NGS panels may allow for routine study of multiple genes, the literature review tended to focus on the significance of individual genes. It is understood that, with more study, stronger recommendations for genetic testing in AML may be appropriate in the near future.

Mutations in FLT3-ITD are now recognized as predictors of a poor prognosis in AML, especially in NK-AML. Most patient cohort studies have found a worse DFS or OS in patients with this mutation, although differences in CR are not always present.^***** Similar findings are found in young adult patients with AML and cytogenetic abnormalities, including t(15;17)(q24.1;q21.2), t(8;21)(q22;q22.1), and t(6;9)(p23;q34.1), as well as mutations of NPM1 and CEBPA.^100,349  Fewer studies have failed to find mutations of FLT3-ITD to be associated with prognosis, and the significance may be less in pediatric AML.^{122,298,338–340}  The mutation level was also directly associated with worse survival,^26,342,343  including 2 patient cohort studies, and the level of mutation should be investigated in cases with a mutation detected.

Refer to Table 10 for study data on NPM1, FLT3-ITD, CEBPA, KIT, and RUNX1 testing.

The recommendation for testing of other mutations in AML is supported by 21 studies,^{†††††} comprising one SR based meta-analysis,³⁵³  3 NRCTs,^10,11,354  and 17 PCSs.^{‡‡‡‡‡} The meta-analysis, reported by Zhou et al³⁵³  was deemed to have a low risk of bias. The 3 NRCTs^10,11,354  were all deemed to have a low to moderate risk of bias. For the 17 PCSs, 3 were deemed to have a low risk of bias,^30,101,119  13 were deemed to have a low to moderate risk of bias,^§§§§§ and one was deemed to have a moderate risk of bias.²⁹  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 16 for the quality-assessment results for all other molecular tests, excluding FLT3-ITD testing.

Among studies of the effect of isocitrate dehydrogenase 1 (IDH1) mutation R132 in adult AML, one meta-analysis³⁵³  found the mutation associated with a worse EFS, but not associated with OS. One NRCT³⁵⁴  found the mutation to be associated with a worse OS and DFS, but that study combined the results of IDH1 and IDH2 mutations. Three PCSs^340,351,352  found IDH1 mutation to be associated with a worse prognosis in patients with AML, who have intermediate-risk cytogenetics or NK-AML, but one of those studies³⁵¹  also combined the results of both IDH1 and IDH2 mutations for analysis. Three PCSs found no prognostic significance to detecting this mutation in adult AML.^15,119,338  A single PCS of pediatric AML did not detect the R132 mutation in any cases.³³⁸  One PCS of the IDH1 single-nucleotide polymorphism (SNP) rs11554137 found the presence of that SNP to be associated with a worse OS.¹⁵ 

For mutations in IDH2 in adult AML, most studies combined the results of the R140 and R172 mutations, and some combined IDH2 results with those of IDH1. One meta-analysis³⁵³  found improved OS, but no effect on EFS in AML with mutated IDH2. Two NRCTs^10,354  and 2 PCSs^351,352  found a worse prognosis with that mutation, although the worse prognosis was only associated with the R172 mutation in one study.³⁵²  Two PCSs^119,340  found no prognostic significance to the presence of an IDH2 mutation in AML, and one³⁵²  found no prognostic significance when the R140 mutation was present in AML.

One NRCT and one PCS studied the prognostic significance of tet methylcytosine dioxygenase 2 (TET2) mutations in AML. The NRCT³⁵⁴  found no prognostic significance to the detection of that mutation, whereas the PCS¹⁰¹  found TET2 mutations to be associated with a worse outcome.

Mutations of Wilms tumor 1 (WT1), usually involving exons 7 and 9, were evaluated in 12 studies and found to be associated with a significantly worse prognosis in AML, usually in NK-AML, in 8 of the 12 studies—one NRCT¹⁰  and 7 PCSs,^******—and was not prognostically significant in the other 4 PCSs.^{29,109,119,120}  One positive PCS was in pediatric patients.¹⁹⁵ 

The prognostic significance of detecting the WT1 SNP rs16754 was evaluated in one NRCT¹¹  and 3 PCSs^15,29,119  and was found to be significantly associated with an improved prognosis in all studies.

One PCS in adult AML evaluated the prognostic significance of DNA methyltransferase 3 alpha (DNMT3A) mutations in AML and found mutations to be significantly associated with a worse OS and EFS.¹³⁰ 

No studies were identified with the search parameters of the SR evaluating additional sex combs like 1, transcriptional regulator (ASXL1), mutations in AML, although a recent study found a worse OS in patients with AML with myelodysplasia-related changes (AML-MRC), and ASXL1 mutations compared with patients with AML-MRC and no such mutation (P = .01).³⁵⁵  Future studies may clarify the significance of that gene mutation in AML.

Two PCSs evaluated the prognostic significance of KMT2A (MLL)-PTD (partial tandem duplication) mutations in AML. One found that mutation to be associated with a worse progression-free survival,¹⁰⁹  and one found no significance to that mutation on OS in NK-AML.¹¹⁹ 

One PCS evaluated the prognostic significance of neuroblastoma RAS viral oncogene homolog (NRAS) mutations in NK-AML and found no effect on OS.¹¹⁹ 

Additional studies on mutations in DNMT3A, usually R882, show variable results, with some also showing such mutations associated with shorter remission and survival,^356,357  whereas others showed no effect on remission or survival duration.^357,358  Mutations may be associated with a worse outcome in younger adults (younger than 60 years),^130,358  but one study suggests worse DFS and OS in older adults with the R882 mutation.³⁵⁶ 

Mutations in tumor protein p53 (TP53) were highly associated with complex karyotypes and other myelodysplasia-related cytogenetic abnormalities that involve chromosomes 5, 7, and 17.^359–361  Such cases were either therapy-related AML or AML-MRC, but the mutations may be germline. TP53 mutations are associated with worse RFS, EFS, and OS. However, the effect of that mutation may not be significant, with complex karyotypes defined as 5 or more abnormalities.³⁶⁰ 

Refer to Table 11 for study data for IDH1, IDH2, TET2, WT1, DNMT3A testing.

There were 172 respondents, 65.7% (n = 113) of whom agreed, 1.74% (n = 3) who disagreed, and 32.56% (n = 56) who did not commit but responded with written comments to the initial draft statement. The comments included suggestions to include more mutation analysis as well as suggestions that fewer mutation studies should be listed. Some confusion arose from the omission of mutation studies for NPM1 and CEBPA in this statement, but those studies were recommended in statement 19. Some respondents suggested restricting that statement only to NK-AML, and others wanted to expand testing to a large NGS panel for all AML cases. Several objected to the listing of specific protein expression assays that were included in the original statement because such tests were not widely available, and others suggested adding testing for other proteins. Those comments were considered in the final draft of statement 16 in this article.

For adult patients with confirmed CBF-AML—AML with t(8;21)(q22;q22.1); RUNX1-RUNX1T1 or inv(16)(p13.1q22)/t(16;16)(p13.1;q22); CBFB-MYH11—the pathologist or treating clinician should ensure that an appropriate mutational analysis for KIT is performed. For pediatric patients with confirmed CBF-AML—AML with t(8;21)(q22;q22.1); RUNX1-RUNX1T1 or inv(16)(p13.1q22)/t(16;16)(p13.1;q22); CBFB-MYH11—the pathologist or treating clinician may ensure that appropriate mutational analysis for KIT is performed.

The strength of evidence was adequate to support testing for KIT in adult patients with CBF-AML, but insufficient to support testing for KIT in pediatric patients with CBF-AML.

This recommendation was supported by 2 PCSs^298,339  obtained from our SR. The risk of bias assessment for both was low to moderate. Overall, none of the studies providing the evidence base for statement 17 were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 17 for the quality-assessment results for the studies included for statement 17.

This recommendation for adult patients is evidence based and is supported by multiple cohort studies outside of our SR. Core-binding factor AML is a distinct cytogenetic subtype of AML that is characterized by the presence of t(8;21)(q22;q22.1) RUNX1-RUNX1T1 or inv(16)(p13.1q22)/t(16;16)(p13.1;q22) CBFB-MYH11 and, in general, is considered a favorable risk.^156,162  However, KIT mutations were identified in approximately 20% to 30% of CBF-AML^298,362,363  and are associated with increased recurrence risk and decreased survival. In adults, available data supported an adverse prognostic effect from KIT mutations in CBF-AML.^362–364  In one study of 33 patients,³⁶⁴  the few patients (n = 8; 24.2%) with t(8;21) and a KIT mutation had a significantly lower EFS (244 days versus 744 days); that difference was not seen in patients with NK-AML. A similar result was seen in a larger study of 110 adult patients with CBF-AML enrolled in multiple RCTs.³⁶³  In AML with inversion 16, the KIT mutation was associated with a greater 5-year cumulative incidence of relapse (56% versus 29%; P = .05) and a worse OS when adjusted for sex (P = .01), compared with cases without the KIT mutation. In AML with t(8;21), the KIT mutation was associated with increased 5-year cumulative incidence of relapse (70% versus 36%; P = .02), but there was no statistically significant difference in OS. A third, larger study³⁶²  of 354 patients with CBF-AML also showed a higher incidence of relapse among the 99 patients with KIT mutations, but further showed that effect was only significant in multivariate analysis if it was limited to those patients with higher levels of the mutation.

The prognostic effect of KIT mutations in pediatric CBF-AML is more controversial. One study met inclusion criteria in our SR.³³⁹  In that retrospective study of 203 pediatric patients with CBF amassed from several different trials, 38 showed KIT mutations, but there was no difference in the 5-year EFS between patients with and those without KIT mutations, either when they were looked at as a whole or when they were divided separately into those with t(8;21) or inv(16) leukemia. In contrast, a separate pediatric study from the external review of t(8;21) AML showed 94.7% DFS for the 38 patients without KIT mutations compared with 37.5% for the 8 patients with the mutations (P ≤ .001).³⁶⁵  Significant differences in 4-year OS were also observed between patients with the KIT mutation (50.0%) and those without KIT mutation (97.4%; P = .001). The outcome of the KIT⁻ t(8;21) patients in this study was much better than the 59% 5-year EFS reported in the larger CBF study, suggesting that the prognostic significance may be different in the context of different therapies. KIT was not prognostic in a separate pediatric AML study reviewed in the external review, which did not look separately at patients with CBF leukemia.³⁶⁶ 

One study from our SR included both pediatric and adult CBF AML.²⁹⁸  For combined adult/pediatric patients with AML and t(8;21)(q22;q22.1); RUNX1-RUNXT1, this study showed no significant association of KIT mutation and outcome (P = not significant [NS] for relapse rate [P = .39] or OS [P = .58]), although with a trend for inferior OS in patients with the KIT mutation (P = .14). For combined adult/pediatric patients with AML and inv(16)(p13.1q22)/t(16;16)(p13.1;q22); CBFB-MYH11, this study showed no significant association of the KIT mutation with relapse rate (P = .41) or OS (P = .70).²⁹⁸ 

Based on those findings, the EP concluded that KIT mutation testing should be performed in cases of confirmed CBF leukemia for further prognostication of this AML category. The evidence is strongest in adult patients, but there are data to suggest a negative prognostic effect in the pediatric population.

There were 173 respondents, with 84.39% (n = 146) who agreed, 3.47% (n = 6) who disagreed, and 12.14% (n = 21) who wrote comments. There was strong support for the recommendation, especially for the adult population. The prognostic significance of the KIT mutation in the pediatric AML was felt to be more controversial, and after review of the comments, the recommendation to test in the pediatric population was changed from should to may. Some felt that this testing should be performed only in patients in whom it affects clinical management or those who are transplant candidates. The public comments were taken into considered in the final draft of statement 17 in this article.

For patients with suspected APL, the pathologist or treating physician should also ensure that rapid detection of PML-RARA is performed. The treating physician should also order appropriate coagulation studies to evaluate for disseminated intravascular coagulation (DIC).

APL is defined by the presence of PML-RARA rearrangement. Because APL is treated differently from other AML subtypes, rapid diagnosis of this type of AML is critical. Although no evidence from our SR informs this statement, evidence external to our SR indicates that conventional karyotyping should be performed in all patients with suspected APL, but the testing may miss rare, cryptic PML-RARA rearrangements.³⁶⁷  Reverse transcription PCR for PML-RARA can rapidly confirm a diagnosis of APL, even in patients who are leukopenic and in those with cytogenetically cryptic PML-RARA rearrangements.^367–369  Interphase FISH studies using dual-fusion probes for PML and RARA can also be used to confirm the rearrangement.³⁷⁰  Immunofluorescence staining methods for PML protein can rapidly confirm the presence of PML-RARA rearrangement because of the differential nuclear distribution of the PML protein in APL, but that staining is not widely available and may miss variant RARA translocations.^367,371,372  The body of evidence supports a strong recommendation for the use of a rapid-detection method to confirm PML-RARA rearrangement in APL; the determination of which of the several alternate methods to use should be made by each individual laboratory.

Patients with APL are at high risk for DIC, which can be evaluated by coagulation studies. No evidence from our SR informed this statement. Evidence outside our SR indicates that compared with other AML subtypes, APL is more often associated with DIC, has more fibrin degradation products, higher D-dimer levels, and lower fibrinogen levels.³⁷³  In patients with APL, a prolonged prothrombin time has been associated with greater risk of clinical bleeding³⁷⁴  and a high International Society of Thrombosis and Hemostasis (Carrboro, North Carolina) DIC score (based on platelet count, D-dimer level, prothrombin time, and fibrinogen level)³⁷⁵  has been associated with a greater risk of fatal bleeding events.³⁷⁶  Based on that evidence, a strong recommendation was made to perform coagulation testing in patients with suspected APL.

There were 172 respondents, of whom, 89% (n = 154) agreed, 1.74% (n = 3) disagreed, and 8.72% (n = 14) wrote comments, among which were a more-precise definition of rapid and the preferred methodology for detection of PML-RARA rearrangement. The comments were considered in the final draft of statement 18 in this article.

For patients other than those with confirmed CBF-AML, APL, or AML-MRC cytogenetic abnormalities, the pathologist or treating clinician should ensure that mutational analysis for NPM1, CEBPA, and RUNX1 is performed.

The strength of evidence was adequate to support this guideline statement.

This statement was supported by 2 PCSs^100,339  that met the inclusion criteria for our SR. Both of those studies were deemed to have a low to moderate risk of bias. Neither of the studies was found to have methodological flaws that would raise concerns about their findings. Refer to Supplemental Table 18 for the quality-assessment results for the studies included for statement 19.

Acute myeloid leukemia with mutated NPM1 defines a specific and unique category of AML under the WHO classification.¹³²  Mutations in NPM1, a nucleocytoplasmic shuttling protein, are the most-common mutations in adult AML, occurring in 27% to 35% of cases.^377,378  Frameshift mutations in exon 12 (chromosome band 5q35) result in an elongated protein that is retained in the cytoplasm.³⁷⁷  Those mutations are most frequent in NK-AML (45%–60%),^377–379  and occur only rarely in association with the recurrent cytogenetic abnormalities that define CBF-AML—AML with t(8;21) or inv(16)/t(16;16)—or APL.^{377,378,380,381}  This was supported by 2 studies from our SR, for both adult (Gaidzik et al¹⁰⁰  reported P < .05 incidence of NPM1 mutations compared with other AML subtypes) and pediatric patients (Pollard et al³³⁹  reported 0% incidence of NPM1 and CEBPA mutations). Cytogenetic abnormalities associated with NPM1 mutations are most frequently single genetic abnormalities (ie, +8, +4, −Y, del(9q), +21)^378,382  and are only rarely associated with a complex karyotype (P < .001).³⁷⁸  These mutations are less frequent in childhood AML (6.5%–8%) but occur primarily in pediatric cases with NK-AML (22%–27.1%).^383–385 

The favorable prognostic effect of the NPM1 mutation has been shown in multiple cohort studies and is strongest when combined with a lack of the FLT3-ITD mutation. In NK-AML, the NPM1 mutation alone is associated, in some studies, with improved CR (P < .03) without significant effect on OS.^378,380  However, when evaluated in context of the FLT3 mutation, the NPM1 mutation in the absence of the FLT3 mutation is associated with significantly higher OS (P < .03),^378–380  DFS (P < .04),³⁷⁸  EFS (P = .01),³⁸⁰  and RFS (P < .001)^379,380  compared with all other NPM1/FLT3 groups. In addition, the availability of a human leukocyte antigen–matched family donor in the NPM1⁺FLT3⁻ patient group does not affect RFS (P = .57) but was shown to significantly affect RFS in all other groups (P = .001),³⁷⁹  suggesting those patients should be excluded from transplant as first-line therapy. Studies in the pediatric population are limited by small patient numbers. In one pediatric AML study, the NPM1 mutation in the absence of FLT3-ITD mutation (n = 13) was associated with a trend toward a favorable 5-year EFS (P = .51).³⁸³  Within the FLT3⁻ subset, the NPM1 mutation was shown to have similar outcome to t(8;21) and inversion 16 AML. In a second cohort study, NPM1 mutation was associated with favorable EFS (P = .02 overall; P = .01 in NK-AML).³⁸⁵  FLT3 did not appear to affect outcome, but analysis was limited by small numbers (n = 10 of 25 patients with the NPM1 mutation). The favorable prognostic effect of the NPM1 mutation was not altered by an aberrant, non-MDS karyotype³⁸²  or by multilineage dysplasia³⁸⁶  in de novo AML. The AML with mutated CEBPA also defined a specific and unique category of AML under the WHO classification, but classification under that category is now restricted to cases with biallelic mutation.^8,132  CEBPA mutations have also been associated with a favorable prognosis in AML. CEBPA belongs to the CCAAT/enhancer binding protein family of transcription factors, is expressed exclusively by myelomonocytic cells, and is upregulated in granulocyte differentiation.³⁸⁷ 

Mutations in CEBPA are reported in 10% to 15% of patients with AML, and the most frequent mutations include either N-terminal frameshift mutations or C-terminal in-frame insertions/deletions.^388,389 

Three mutational patterns have been identified: single-mutated (involving one allele), double-mutated (typically biallelic), and homozygous CEBPA mutation because of a loss of heterozygosity.^390,391  These mutations occur most frequently in NK-AML (70%) and less frequently in AML with intermediate risk (most frequently trisomy 8) or unfavorable cytogenetic abnormalities.³⁹²  These mutations do not occur with favorable, recurrent cytogenetic abnormalities—t(15;17);PML-RARA, t(16;16) or inv(16); CBFB-MYH11, or t(8;21); RUNX1-RUNX1T1.³⁸⁹  FLT3-ITD and NPM1 mutations rarely occur in combination with biallelic-mutated CEBPA.³⁹² 

The favorable prognosis associated with CEBPA mutations has been shown in multiple cohort studies and is confined to biallelic-mutated CEBPA. In a prospective, multicenter clinical trial of 135 patients with AML, the presence of a CEBPA mutation (single or biallelic mutated) compared with wild-type CEBPA was associated with longer OS (P = .04), EFS (P = .04), and DFS (P = .05).³⁹³  A second prospective trial of 224 patients with AML demonstrated that the favorable prognostic significance of CEBPA mutation was confined to cases that were biallelic mutations.¹¹⁵  Biallelic CEBPA mutations were associated with improved OS (P = .01) and DFS (P = .01) compared with a single CEBPA mutation. There was no difference in OS or DFS in AML with a single CEBPA mutation when compared with wild-type CEBPA. These findings were further supported by a large, multicenter study, which included 2296 patients with AML enrolled in 2 large, prospective clinical trials.³⁹²  Biallelic CEBPA mutations, but not single CEBPA mutations, were verified as an independent favorable prognostic factor. When compared with single-mutated and wild-type CEBPA, biallelic-mutated CEBPA was associated with improved OS (P < .001 and P < .002) and longer EFS (P = .008 and P = .012). There was no difference in outcome between NK-AML and AML with intermediate-risk cytogenetic abnormality for both biallelic CEBPA mutations and single CEBPA mutations.

One NRCT¹⁰  and one RCS¹⁰⁰  evaluated the prognostic significance of RUNX1 mutations in AML and both found worse OS and EFS in patients with that mutation. Unlike NPM1 and CEBPA mutations, mutations of RUNX1 were relatively commonly associated with MDS-related cytogenetic abnormalities or prior therapy, features that continue to take precedence over the RUNX1 mutation for disease classification. Based on those studies, the EP recommends testing for NPM1, CEBPA, and RUNX1 mutations in AML other than APL, CBF-AML, or AML with MRC cytogenetic abnormalities. Cases meeting criteria for AML-MRC based on multilineage dysplasia alone should also be tested. In addition to defining specific categories of AML under the WHO classification, mutations of NPM1 and biallelic mutations of CEBPA were associated with a favorable risk. The favorable prognosis for NPM1 mutation was confined to cases that lacked FLT3-ITD and, for CEBPA, to the group with biallelic mutations.

There were 172 respondents, of whom 75.58% (n = 130) agreed, and 4.07% (n = 7) disagreed. However, 20.35% (n = 35) of the respondents provided a range of comments. Most of the comments referred to the necessity of evaluating NPM1 and CEBPA mutations in the context of the FLT3 mutation, which was addressed in the discussion and in statement 16. Several respondents suggested restriction of those markers for NK-AML or FLT3-ITD⁻ AML. Those comments were addressed in the discussion and were considered in the final draft of statement 19 for this article.

For patients with confirmed AL, no recommendation is made for or against the use of global/gene-specific methylation, miRNA expression, or gene expression analysis for diagnosis or prognosis.

The strength of evidence was insufficient to support this guideline statement.

This statement was supported by 10 studies,^{††††††} comprising 2 NRCTs^11,131  and 8 PCSs.^{‡‡‡‡‡‡} Both of the nonrandomized studies were deemed to have a low to moderate risk of bias assessment. For the PCSs, the risk of bias assessments ranged from low^{24,30,101,193,395}  to low to moderate.^34,127,394  None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 19 for the quality-assessment results of studies included for statement 20.

The interplay between mutational analysis and gene expression profiling in predicting prognosis is an area of ongoing research, especially in patients with cytogenetically normal AML. Although protein and gene-expression profiling studies for ERG, BAALC, or MECOM/EVI1 have reported prognostic significance for outcome in some studies, overall, independent, prognostic significance may not be present when mutational analysis is integrated into multivariate analyses.¹¹  Often, the prognostically significant effect of deregulated expression of a specific gene was most apparent in highly selected patient cohorts, such as patients older than 70 years with cytogenetically normal AML.¹³¹  In AML with KMT2A (MLL)-rearrangement, overexpression of MECOM/EVI1 is associated with an inferior prognosis (P = .01 for OS).³⁹⁶  In one study, a 24-gene prognostic signature independently predicted OS and EFS in AML (P < .001).³⁹⁷ 

In studies outside our SR, low global DNA methylation (as assayed by a luminometric methylation assay) was associated with favorable outcome in non-APL de novo AML, independent of karyotype risk and NPM1, FLT3, and CEBPA mutation status.³⁹⁸  High methylation of polycomb target genes was associated with better progression-free survival and OS in cytogenetically normal AML.³⁹⁹  Within our SR, 2 studies were identified for gene-specific methylation, one finding no significant prognostic effect of BMP/retinoic acid inducible neural specific 1 (BRINP1/DBC1) methylation in AML³⁹⁴  and one of which found that methylation status of any of 9 specific genes adversely affected OS in ALL (P < .05).³⁴ 

In our SR, 5 studies found a significant effect of miRNA expression levels on outcome in adult AML (P < .05),^{24,30,127,193,395}  including one in which miRNA expression patterns were correlated with expression of other prognostic markers,¹²⁷  whereas one study found no significant effect on outcome.¹⁰¹  One study found significant association of miRNA with OS in adult and pediatric ALL (P < .05).³⁴ 

Overall, many studies identified both within, and outside of, our SR indicated that deregulated gene expression, miRNA expression, and global as well as gene-specific methylation may affect outcome in AL. However, much of the data are relatively recent; moreover, those studies are not currently standard clinical laboratory tests, even in reference laboratories. Thus, no recommendation was made for, or against, those specialized tests at the time of AL diagnosis. With technological advances, it is possible that these specialized studies will become more widespread in clinical practice, similar to the current standardized assessment for mutations in key leukemia-associated genes recommended in statements 15, 16, and 19.

Refer to Supplemental Table 4 for study data on global/gene-specific methylation, miRNA expression, and gene expression analysis.

There were 162 respondents, of whom, 58.64% (n = 95) agreed, and 11.11% (n = 18) who disagreed. However, 30.25% (n = 49) provided written comments, which indicated they thought the clinical utility of gene expression, miRNA expression, and global/gene-specific methylation studies currently have limited clinical utility and that the studies were not widely available for clinical use. Those comments were considered in the final draft of statement 20 in this article.

For patients with confirmed MPAL, the pathologist or treating clinician should ensure that testing for t(9;22)(q34.1;q11.2); BCR-ABL1, and KMT2A (MLL) translocations is performed.

Although one study from our SR informed this recommendation,¹⁴⁰  the guideline statement is evidence based and supported by 2 additional studies outside our SR.^400,401 

Mixed-phenotype acute leukemia is associated with a variety of cytogenetic abnormalities, among which, are included t(9;22)(q34.1;q11.2); BCR-ABL1 and KMT2A (MLL) translocations.^140,02  Those cytogenetic abnormalities are associated with distinctive features and define specific entities under the WHO classification: MPAL with t(9;22); BCR-ABL1 (Ph⁺) and MPAL with t(v;11q23.3); KMT2A (MLL) rearranged.¹³² 

In studies using 2008 WHO criteria for MPAL, t(9;22)(q34.1;q11.2) was identified in 15% to 20% of cases, and KMT2A rearrangement in 4% to 8% of cases.^140,403 

With the advent of TKIs, it has become important to recognize the BCR-ABL1 fusion in cases of AL. Three studies were identified that specifically addressed the question of the outcome of MPAL with t(9;22)(q34.1;q11.2) (Ph⁺). Overall, those patients have a poorer outcome and, in the preimatinib era, had the worst prognosis of any cytogenetic group in MPAL.¹⁴⁰  In one small (nonrandomized) series of 21 patients with adult MPAL, the poor general outcome was confirmed, and patients receiving imatinib fared better with a 1-year OS of 43% in the imatinib group compared with no survivors in the chemotherapy-alone group.⁴⁰⁰  In a different series, which compared the outcome of 42 patients with Ph⁺ AL treated with imatinib-containing regimens, the 5-year OS and DFS of the 13 patients with MPAL was no different from that of the patients with Ph⁺ ALL, suggesting that imatinib therapy improved the outcome of the patients with Ph⁺ MPAL and should be considered the standard of care.⁴⁰¹  Similarly, KMT2A (MLL) rearrangement was associated with reduced survival.⁴⁰⁴ 

Based on those data, the EP recommends testing for t(9;22)(q34.1;q11.2); BCR-ABL1 and KMT2A (MLL) translocations in MPAL.

There were 174 respondents, 89.66% (n = 156) of whom agreed, 1.15% (n = 2) who disagreed, and 9.2% (n = 16) who wrote comments. In response to the comments, language regarding the specific methodology to test for these rearrangements was removed from the final draft of statement 21 in this article.

All laboratory testing performed for the initial workup and diagnosis of a patient with AL must be performed in a laboratory that is in compliance with regulatory and/or accreditation requirements.

The strength of evidence was insufficient to support this guideline statement, but its justification seems intuitive because laboratory testing occurs in a highly regulated environment. No evidence-based data were available from our SR.

This guideline statement was based on expert consensus opinion and codifies the importance of good laboratory practices in patient care. In the United States, clinical laboratory testing is regulated under CLIA '88, as administered by the Centers for Medicare and Medicaid Services (Baltimore, Maryland); the amendments were updated in 2003.^405,406  Certain medical devices and laboratory tests used in an evaluation for AL have been approved and are regulated by the US Food and Drug Administration, whereas others have been developed by, and are performed within, accredited laboratories (laboratory-developed tests). It is important to ensure that laboratory-developed tests have been appropriately validated and performance characteristics established before being used in patient care. The penalties for regulatory noncompliance are significant and can include loss of laboratory director responsibilities for 2 years, monetary fines, closure of the laboratory, and inability to receive Medicare reimbursement. The EP included this guideline statement to ensure that tests performed in research laboratories (eg, in the United States, non-CLIA–approved laboratories) would not be used for patient care or be included in the medical record.

There were 172 respondents, 97.67% (n = 168) agreed, and 2.33% (n = 4) who disagreed. There were 5 comments that were generally supportive, although they raised the question of an emergency situation that might arise outside of the availability of an accredited laboratory. The final draft statement in this document was modified to apply to national and international situations by not specifying individual regulatory and/or accrediting agencies. It is the laboratory director's responsibility to be aware of applicable regulations.⁴⁰⁷ 

If, after examination of a PB specimen, it is determined that the patient will require immediate referral to another institution with expertise in the management of AL for treatment, the initial institution should, whenever possible, defer invasive procedures, including BM aspiration and biopsies, to the treatment center to avoid duplicate procedures, associated patient discomfort, and additional costs.

The strength of evidence was adequate to support this guideline statement.

This statement was supported by 28 studies,^§§§§§§ comprising 2 meta-analyses,^408,410  one RCT,⁴¹⁴  2 NRCTs,^106,112  and 23 PCSs.^******* For the 2 meta-analyses, both were deemed to have a low to moderate risk of bias. One trial, an RCT reported by Vance et al⁴¹⁴  was deemed to have a moderate to high risk of bias. For the 2 NRCTs,^106,112  the trial by Moorman et al¹⁰⁶  was deemed to have a moderate risk of bias. The trial reported by Aricò et al¹¹²  was deemed to have a low to moderate risk of bias. For the PCSs, 3 were deemed to have a low risk of bias,^101,102,194  14 were deemed to have a low to moderate risk of bias,^{†††††††} and 6 were deemed to have a moderate risk of bias.^{‡‡‡‡‡‡‡} None of those studies were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 20 for the quality-assessment results of studies included for statement 23.

The level of expertise for the diagnosis and treatment of AL varies, with some centers having experience in virtually all case types, and some never treating such patients. In centers that do not routinely treat patients with AL and do not offer most of the testing needed to make a comprehensive diagnosis of AL, it is recommended that the patient be transferred to a treating center before the complete diagnostic evaluation is performed. This helps to reduce duplication of testing and the associated expense and discomfort that occur with such testing. It is recognized that some centers may determine that transfer of care is appropriate only after a complete diagnostic workup is completed or will need to make a rapid diagnosis to initiate therapy before transfer, such as in the diagnosis of APL with PML-RARA, and this guideline statement should not interfere with testing that is considered emergent before transfer.

Review of diagnostic material for the diagnosis of AL varies in the literature. At diagnosis, morphologic evaluation is often performed and interpreted at the local institution,^{§§§§§§§} often, with review of slides in a central laboratory or tertiary care center if part of a clinical trial,^25,411,415  or samples may be prepared and interpreted at a central laboratory or tertiary care center,^{112,297,340,412,413}  including any immunophenotypic or other studies required for diagnosis.⁴¹⁵  A discrepancy rate of 12% was reported between local and central review of AML diagnoses.⁴¹¹  At diagnosis, flow cytometry testing may be performed at the primary institution, with results reviewed by a central laboratory or tertiary care center if part of a clinical trial,^14,324,415  or flow cytometry may be performed and interpreted at a central laboratory.^297,413  At diagnosis, cytogenetic testing may be performed and karyotypes interpreted at the primary institution^********; cytogenetics may be performed, but karyotype images be reviewed at a central laboratory or tertiary care center,^{††††††††} or all cytogenetic testing and interpretation may be performed at a central laboratory or a tertiary care center.^{‡‡‡‡‡‡‡‡} In one study, 32% of AML and 38% of ALL karyotypes were revised or rejected as inadequate upon central review of the local karyotype images.¹⁷⁹  Fluorescence in situ hybridization studies as part of clinical trials were typically performed at a central laboratory or tertiary care center.^{106,112,413,414}  Molecular studies confirming mutations and gene rearrangements as part of clinical trials were typically performed at a central laboratory or tertiary care center^{§§§§§§§§} as were DNA methylation studies.¹⁹⁶ 

If testing involves making specimens for routine or cytochemical staining or performing flow cytometry on BM at a central laboratory, the sample should be shipped overnight and processed within 24 hours of being obtained.^112,297  For testing involving cytogenetics performed centrally, overnight shipping of a heparinized BM sample is recommended.²³  Molecular genetic testing may be performed on samples that are cryopreserved,^{194,324,411,412}  with some studies recommending the use of Trizol reagent (Thermo Fisher Scientific, Waltham, Massachusetts)¹⁹⁴  and storage of samples in liquid nitrogen,^324,412  and others recommending overnight shipping of sodium-citrate, anticoagulated samples for preparation of DNA (dry pellets stored at −80°C) or RNA (pellets in 4M guanidium isothiocyanate stored at −20°C).²³  For DNA methylation testing, BM samples should be shipped to the central laboratory in heparinized tubes for processing within 24 to 36 hours of being obtained, should be frozen immediately upon receipt (2–10 million cells), and should be stored at −70°C.¹⁹⁶ 

There were 171 respondents, of whom, 91.81% (n = 157) agreed, and 8.19% (n = 14) disagreed. Despite the support for the initial draft of this statement, there were 32 written comments that raised several patient concerns. Most of those concerns related to delays in transfer of patients or the inability to obtain acceptance for transfer without a complete diagnosis. In addition, the need for a rapid diagnosis of APL was raised by several commenters. Based on those comments, the final draft of statement 23 in this article was altered to clarify that it applies to patients needing immediate transfer.

Refer to Supplemental Table 5 for study data informing this statement.

If a patient is referred to another institution for treatment, the primary institution should provide the treatment center with all laboratory results, pathology slides, flow cytometry data, cytogenetic information, and a list of pending tests at the time of the referral. Pending test results should be forwarded when they become available.

The strength of evidence was insufficient to support this guideline statement.

This statement was supported by 2 PCSs^179,411  that met the inclusion criteria for our SR. The study by Barbaric et al⁴¹¹  was deemed to have a low to moderate risk of bias, and the study reported by Mrozek et al¹⁷⁹  was deemed to have moderate risk of bias. Neither of the studies was found to have methodological flaws that would raise concerns about their findings. Refer to Supplemental Table 21 for the quality-assessment results of studies included for statement 24.

This guideline statement was based on expert consensus opinion that knowledge of test results performed at the primary institution is optimal to rapidly confirm a diagnosis of AL and to allow more cost-effective management of the patient at the referral institution. Morphologic, flow cytometric, cytogenetic, and molecular genetic studies can pose significant cost to the health care system, especially if they are repeated without knowledge of the initial results. Certain tests may not need to be repeated at the referral institution if the information from the primary institution is available for review in a timely manner, and the findings are confirmed. We found no studies that compared interinstitutional results; however, comparisons between local institution and central review in the context of clinical trials were available for morphologic and cytogenetic reviews, and, as mentioned previously, indicated a discrepancy rate of 12% between local and central review of AML diagnoses,⁴¹¹  and in one study, 32% of AML and 38% of ALL karyotypes were revised or rejected as inadequate upon central review.¹⁷⁹  These findings support the need for confirming the results of diagnostic testing at the referral institution, with repeat testing employed judiciously.

There were 169 respondents, 98.22% (n = 166) of whom agreed, and 1.78% (n = 3) who disagreed. There were 9 written comments. Those comments generally dealt with logistical issues regarding communication between the primary institution and treatment center. However, as one respondent indicated, “seems simple, but in the real world it is often difficult.” In the era of shared medical records, these procedures may become easier. The comments received were considered and are reflected in the final draft of statement 24 in this article.

In the initial report, the pathologist should include laboratory, morphologic, immunophenotypic, and, if performed, cytochemical data on which the diagnosis was based, along with a list of any pending tests. The pathologist should issue addenda/amended reports when the results of additional tests become available.

Our SR provided no data to inform this statement. However, the panel believed that the benefits of implementing the recommendation vastly outweighed the harms and thus designated this guideline statement with a strong recommendation.

Both routine and more-specialized testing results must be incorporated into initial and subsequent, integrated reports. Because morphology, cytochemical stains, FCI, and immunohistochemical stains are typically available within a day or so after the BM has been obtained, the interpretation and integration of those results should be included in the initial reports.

Specialized testing is an integral component of AL diagnosis and is required in virtually all cases of AL to provide either diagnostic or prognostic information. In addition, specialized tests may provide evidence to validate the use of a specific type of targeted therapy for an individual patient. Based on an assessment of the morphologic and immunophenotypic features of the AL, genetic testing must be performed in a cost-effective manner using evidence based criteria. Conventional karyotyping is considered the standard of care for all cases of AL. In addition, some molecular genetic tests are also considered to be warranted in cases of AL meeting specific criteria as listed in this clinical practice guideline.

Consolidating all routine and specialized test results into an integrated consultation report is optimal for effective communication with treating physicians and patients and for optimal therapy.⁴¹⁶  Because of the time delays inherent in some molecular genetic testing, pathologists need to have mechanisms to track pending test results. Those results need to be integrated with morphologic and immunophenotypic data to enhance the original diagnosis, using current WHO classification criteria, as well as to provide prognostic information. Either an addendum format or, more optimally, an integrated, interpretive format is optimal.⁴¹⁶ 

There were 171 respondents, 94.15% (n = 161) of whom agreed, and 5.85% (n = 10) who disagreed. Despite the overwhelming support for this statement, several of the comments recognized the difficulty of getting all the data into the electronic medical record. Those comments were taken into account for the final statement in this article.

The pathologist and treating clinician should coordinate and ensure that all tests performed for classification, management, predicting prognosis, and disease monitoring are entered into the patient's medical records.

This information should include the sample source, adequacy, and collection information, as applicable.

Our SR provided no data related to this key question or statement. The EP, however, strongly recommended that all critical information related to the diagnosis and prognosis of a patient with AL be available in the medical record. Ideally, this would be summarized in a single consolidated report.⁴¹⁶  However, it is recognized that, in some settings, not all information needed for pathologists to generate such a report might be available because of some testing being sent directly to other laboratories by the treating physician. Therefore, all this information should be either interfaced with, or scanned into, the medical record, so that all the material needed for the determination of a comprehensive diagnosis is present in the patient's record. It is recognized that improved and more-integrated pathology information systems that directly interface with the electronic medical record are needed to allow for optimal patient care.

There were 169 respondents, 99.41% (n = 168) of whom agreed, and 0.59% (n = 1) of whom disagreed. There were 8 written comments, all of which endorsed this concept.

Treating physicians and pathologists should use the current WHO terminology for the final diagnosis and classification of AL.

The strength of evidence was convincing to support this guideline statement.

This guideline statement was supported by 40 PCSs obtained in our SR.^********* The risk of bias-assessment scores were low,^12,101  low to moderate,^{†††††††††} and moderate.^{‡‡‡‡‡‡‡‡‡} Overall, none of the studies providing the evidence base for statement 27 were found to have methodological flaws that would raise concerns about the studies' findings. Refer to Supplemental Table 22 for the quality-assessment results for the studies included for statement 27.

A review of the studies published in recent years shows that various classification systems have been used for the diagnosis and subclassification of AL. In AML, the classification systems used included the FAB system,^{§§§§§§§§§} the WHO classification (2001 version),^417,425  and the WHO classification (2008 version),^********** or more than one classification system because of lengthy study periods.^{13,71,159,341,423}  For studies that did not include a classification system,^{31,127,129,419,421}  cytogenetic-defining groups were variably used. For acute B-ALL, most studies^{††††††††††} used the 2008 WHO classification system or a cytogenetic classification system similar to the 2008 WHO system. The National Cancer Institute (Frederick, Maryland) risk classification¹⁶⁷  of ALL was used in conjunction with the cytogenetic risk classification in some studies.^{19,20,295,330}  The FAB system alone was mostly used to describe morphology.¹²⁴  A few studies used immunophenotypic classification only^17,296,426  or no classification system at all.^171,327  The T-ALL classification was mainly based on immunophenotyping alone, including a system proposed by the European Group for the Immunological Characterization of Leukemias (Nancy, France).^{‡‡‡‡‡‡‡‡‡‡}

A uniform categorization of AL is essential to facilitate understanding between health care workers and to provide a framework for clinical practice, data comparison, epidemiologic studies, and new genetic and molecular investigations. The WHO classification of neoplasms of the hematopoietic and lymphoid tissues, published in 2001,⁴²⁵  and updated in 2008,⁵  with a new revision recently summarized,⁸  represents a worldwide consensus on the diagnosis of hematopoietic tumors. That classification derives from numerous published clinical and scientific studies and is the result of collaboration and consensus among pathologists, cytogeneticists, and treating physicians. The WHO classification attempts to incorporate the disease characteristics that have been proven to have clinical and biologic relevance into a useful working nomenclature. A retrospective study of 5848 adult patients with AML showed that the FAB morphologic subclassification did not provide prognostic information if the specific genetic and morphologic WHO categories and WHO provisional entities, such as “AML with mutated NPM1” and “AML with mutated CEBPA,” were included.⁴²⁸  The applicability of the WHO classification in pediatric AML, on the other hand, may require additional molecular genetic data for further disease delineation.⁴²⁹  Nevertheless, since the first edition in 2001, the WHO classification system has been adopted for numerous studies, and its clinical practicality and reproducibility has been demonstrated in diverse international settings. Treating physicians and pathologists should use the most current WHO terminology for the diagnosis and classification of AL, including adoption of the current revision.

There were 167 respondents, 98.8% (n = 165) of whom agreed, and 1.2% (n = 2) who disagreed. There were 11 written comments. Some commented that the WHO classification of AL requires cytogenetic and molecular data, which are often not available at the time of diagnosis or before the initiation of treatment. The other concern was that the WHO classification system was not always up to date. Since the last updates in 2008, numerous molecular genetic discoveries have been published, some of which have been shown to have a significant effect on the treatment and prognosis of AL, and more discoveries will certainly be described after the 2016 classification. Therefore, the final report should incorporate those new data for the purposes of therapy and prognosis. New technologies in molecular genetic discovery are evolving quickly and provide new insights in disease biology, thereby refining disease classifications and guiding clinical practice. The panel acknowledged those comments, which are reflected in the final statement in this article.

Refer to Table 12 for study data that informed this guideline statement.

The 27 statements that comprise the ASH/CAP guideline for the initial diagnostic workup of AL address the 6 key questions initially proposed:

• 1.

  What clinical and laboratory information should be available? (Statements 1 and 2.)

• 2.

  What samples and specimen types should be evaluated? (Statements 3, 4, 7, 8, and 11.)

• 3.

  What tests are required for all patients during the initial evaluation? (Statements 3, 5, 6, 9, and 12.)

• 4.

  What tests are required for only a subset of patients? (Statements 10, 14, 13, 16, 17, 18, 19, 20, and 21.)

• 5.

  Where should laboratory testing be performed? (Statements 22, 23, and 24.)

• 6.

  How should the results be reported? (Statements 25, 26, and 27.)

As noted at the beginning of this article, the initial workup and evaluation of AL has become increasingly complex during the past decade, due, in part, to the availability of new laboratory techniques—particularly genetic studies—that have resulted in better characterization of AL and in classification schemes with improved clinical and scientific relevance. However, not only is the diagnosis and classification of AL important but also of importance is the identification of prognostic factors, antigens, or genetic abnormalities that may be targets for specific therapy, and markers that can be used to follow the response to therapy and monitor residual disease. In addition, the workup must be performed quickly, efficiently, and at a reasonable cost. When all these factors are considered, along with the realization that the recent revision of the 4th edition of the WHO classification of AL has nearly 50 distinct subtypes of AL,⁸  the value of the guidelines described in this article and their applicability for the workup of AL can be readily appreciated.

Although many pathologists and clinicians will consider the most crucial guidelines to be those indicated that relate to specific tests necessary to make an accurate diagnosis and to identify prognostic factors, the guidelines also highlight recommendations for general and logistic considerations that may be neglected in the urgency of the initial workup of AL. For example, statement 1 emphasizes the value of the patient's history, particularly of any predisposing conditions or syndromes, previous therapies, and any family history of leukemia or other neoplasms. This latter recommendation is important in view of the inclusion in the revised WHO classification of the provisional entity myeloid neoplasms with germline predisposition, which is likely more common than currently recognized. In addition, recommendations are made for preservation of cells and tissue from the initial diagnostic specimen for any future studies that may be relevant for prognosis or therapy (statement 7). For patients who are transferred from one institution to another, the guidelines recommend the avoidance of duplicate testing and invasive procedures whenever possible (statement 23), and the transfer of all test results and tests in progress, along with the diagnostic specimens, to the receiving institution (statement 24). The guidelines also indicate the central role of the pathologist in issuing the diagnostic reports, and for updating those reports as additional data are accumulated (statements 25 and 26). Thus, the recommendations are comprehensive in providing guidelines from the time a patient suspected of having AL is first encountered until the final diagnostic reports are generated.

The ASH and the CAP have cooperated in developing this guideline. That joint effort underscores the cooperation that is necessary between the treating clinician and the pathologist in the workup, diagnosis, and care of patients presenting with a suspected diagnosis of AL.

We thank advisory panel members Frederick R. Appelbaum, MD; Clara D. Bloomfield, MD; William L. Carroll, MD; Laura Housley, BS; Jerry Hussong, MD; Steven H. Kroft, MD; Michelle Le Beau, PhD; and Martin S. Tallman, MD. We thank ASH staff Kendall Alexander, MPH; Robert Kunkle, MA; Suzanne Leous, MPA; and Robert Plovnick, MD. We also thank CAP staff Sophia Dimoulis, BA; Carol Colasacco, MLIS, SCT(ASCP); Lisa Fatheree, BS, SCT(ASCP); Megan Wick, MT(ASCP); and consultant methodologist Christina Lacchetti, MHSc.