Minimal Residual Disease in Hematologic Disorders

Advances in the treatment of numerous malignant hematologic disorders over the past 3 decades have resulted in improved response rates and, in many cases, long-term disease-free survival. About 80% of children with acute lymphoblastic leukemia (ALL) can be considered cured today from a condition that not long ago was uniformly fatal.1 Median survival times in chronic myelogenous leukemia (CML) have doubled, with 50% to 60% of patients alive at 5 years and more than 30% at 10 years.2 With modern chemotherapy and radiation therapy programs, up to 70% of patients with Hodgkin disease are cured.3 Response rates in non-Hodgkin lymphomas are often more than 50%, with long-term survival in a substantial proportion of patients.3 Despite this success, in many individuals, disease eventually recurs because of the persistence of low numbers of malignant cells that have not been eradicated with induction therapies. Clinical relapse and death from complications thereof or additional treatments is the frequent consequence.

Despite having achieved a clinical complete remission or response (CR), a patient may still harbor up to 10¹⁰ leukemia cells that persist at levels undetectable by conventional cytomorphologic methods such as light microscopy.4 This level of disease is referred to as minimal residual disease (MRD). To what extent detection and efforts at eradication of residual disease are important and how far they influence prognosis is unclear and the object of numerous ongoing studies. Consequences of any approach are crucial. Patients are subjected to either undertreatment with risk of relapse or overtreatment with exposure to therapy-related morbidity and mortality. The last decade has therefore seen the emergence of various laboratory tools for the detection and assessment of residual disease, including flow cytometry, immunologic studies, fluorescence in situ hybridization (FISH), and cytogenetic assays. Polymerase chain reaction (PCR) techniques in particular have provided a means to analyze tumor-specific DNA sequences with custom-built probes and added a level of sensitivity of detection in the range of 1 malignant cell in 10⁴ to 10⁶ normal cells. Despite this progress, however, many incongruities exist among clinical studies that try to establish prognosis, outcome, and clinical decision making on the basis of these assays.

Various methods have been developed for the detection of MRD (Table 1). Techniques differ in the cellular structures identified and in their sensitivity. Many approaches are still investigational, and numerous clinical trials are ongoing to validate their applicability in clinical practice.

The value of morphologic detection of residual disease is limited by its low sensitivity. Generally, only 1 of 100 cells can be identified as malignant. Occasional inability to distinguish between immature cancer cells and early regenerating cells further restricts the usefulness of standard morphologic methods. However, sensitivity and specificity of morphology can be enhanced when combined with other tools such as immunophenotyping or FISH.5 

In vitro culture techniques have been developed in which bone marrow samples from patients with ALL are grown under conditions favorable for stimulation of leukemic cells.6 The advantage of blast colony assays is that they identify populations of occult malignant cells that can be expanded, hence allowing their biologic characteristics and growth requirements to be further studied. In addition, single colonies can be analyzed by immunologic, cytogenetic, or molecular techniques. Disadvantages of clonogenic assays include their dependence on growth rates of leukemic progenitor cells and the cumbersome nature of assays targeted at single colonies.

Use of monoclonal antibodies by means of flow cytometry or fluorescence microscopy to detect nuclear, cytoplasmic, and surface antigens that are expressed by malignant cells can be fast and reliable.7 Sensitivity of detection can range as high as 1 abnormal cell per 10⁴ to 10⁵ normal cells using double- or triple-color immunofluorescence techniques or fluorescence microscopy screening multiple slides per sample.8 Drawbacks of immunophenotype analysis, however, can reduce the sensitivity of detection to the level of light microscopy and morphology. These include (1) the lack of antigen specificity for malignant cells as these cells represent the counterparts of normal cells with, in many cases, identical or similar antigen profiles; (2) the existence of several subpopulations, some of them as minor clones, that are difficult to identify; and (3) the inability to identify phenotypic switch, a phenomenon that may occur at relapse, albeit at a low rate.9 

Cytogenetic analysis has become an important tool in risk stratification and as a prognosticator for both ALL and CML. Cytogenetic studies provide specificity because they can unambiguously identify malignancy-specific markers and detect cytogenetic signs of clonal evolution at relapse. The pitfalls of conventional cytogenetic analysis are its labor intensity (because it requires in vitro cultures) and the dependency of karyotype identification on dividing cells, thereby missing populations of residual cells with a low proliferative index. Despite its specificity, the sensitivity of the method is low and does not exceed that of morphologic assessment of marrow smears.

FISH has emerged in recent years as a promising new tool for the identification on a molecular level of both chromosomal aberrations and malignancy-specific DNA sequences.10 The main advantages of FISH, besides its specificity, are the larger number of cells that can be studied in a time-efficient manner and without the need for in vitro cultures, the quantifiability of results, and the applicability to archival material such as blood smears and histologic sections. In addition, FISH allows analysis of both metaphase and nondividing interphase cells.11 Interphase FISH can even be performed on peripheral blood samples, thus avoiding the need for marrow aspiration.

Molecular techniques allow the detection of leukemia-specific gene rearrangements by identifying either leukemia-specific translocations or clone-specific immunoglobulin heavy chain (IgH) gene and T-cell receptor (TCR) gene rearrangements. The technical aspects of PCR have been reviewed elsewhere.12–15 Whereas the sensitivity of Southern blotting is similar to that of morphologic assessment and therefore too low to be useful for the detection and follow-up of residual disease, PCR assays are characterized by unparalleled sensitivity. A single malignant cell can be identified among 10⁴ to 10⁶ normal cells.15 Both PCR and its variants such as reverse transcriptase (RT)-PCR and nested PCR have become the preferred tools in studies addressing the detection and role of MRD. The drawback of PCR lies in its extraordinary sensitivity and, hence, the facile manner in which sample contamination and false-positive results can occur. In addition, since RT-PCR measures transcript expression, it is conceivable that a malignant clone that transiently does not express its diagnostic transcript would evade detection by this technique.

Many structural chromosomal abnormalities and their molecular genetic consequences have been described and can be exploited for the identification of residual disease by PCR in various malignant hematologic disorders (Table 2). We will use 2 diseases, ALL and CML, to illustrate the advances and pitfalls in understanding MRD. In the area of residual disease in childhood ALL (Table 3),16 PCR assays have, at times, produced discrepant results. For example, using probes for either TCR or IgH gene rearrangements and PCR assays at various time points after induction and maintenance therapy has caused different investigators to arrive at different conclusions. Whereas some investigators claim that detection of residual disease immediately following induction or during the first 6 months predicts likelihood of relapse,19–21 others have found that, in many patients, MRD can persist for 24 months and even longer and that time points farther away from diagnosis and induction may be more useful as a prognosticator for disease recurrence.22 Other data indicate the importance of serial assessments and quantification of residual disease measurements, rather than focusing on timing in relation to induction or maintenance therapy.23–25 

The PCR data should be examined critically. A powerful technique, PCR also has some shortcomings that may cause both false-positive and false-negative results. Problems such as contamination of samples or degradation of target molecules need to be avoided with appropriate technical precautions. In addition, biologic pitfalls must be considered. Changes in disease markers as reflected by clonal evolution, oligoclonality, subclone formation, and incomplete target sequence rearrangements of the malignant cell population should be excluded before associating a negative PCR finding with absence of malignant cells. Furthermore, disease quantification has proved difficult with PCR assays. Initial assays did not allow measurements of disease burden, and no information could be obtained as to the kinetics of disease disappearance and recurrence in individual patients. Quantitative PCR assays have helped this dilemma and could demonstrate some correlation between changes in disease burden and relapse or remission. Notably, in many studies, PCR-negative patients relapsed, and conversely, PCR-positive patients stayed in remission.26,27 Therefore, for disease quantification to be useful, a threshold of disease load must be defined above which relapse is likely and below which remission can be sustained.

Further, molecular tools such as PCR or Southern blotting do not give any information about the biologic nature of the cells they characterize. Residual cells that carry phenotypic, cytogenetic, or molecular markers of a malignant clone such as IgH or TCR gene rearrangements may not be malignant in a clinical sense, especially if they did not induce the disease recurrence. They rather may represent mutant descendants of the original malignant clone that have either lost or gained genetic alterations rendering them less tumorigenic or turning them into an irrelevant subclone destined for apoptosis.28 

Targeting the BCR-ABL transcript, the hallmark of CML,29 using PCR, studies of MRD in CML have been undertaken. After interferon alfa treatment, initial studies demonstrated persistence of the malignant clone in virtually all patients.30,31 In a study on residual disease in CML, Hochhaus et al31 measured levels of BCR-ABL expression in CML patients in cytogenetic remission after treatment with interferon alfa. They described a median of 750 transcripts per microgram of RNA. However, this cutoff point does not determine a threshold for relapse vs remission, and no other studies to date have provided reliable quantitative levels either. Furthermore, the detection of MRD may be time dependent, as demonstrated by more recent studies of interferon alfa–treated patients.32 Patients who had remained in continuous complete cytogenetic remission for 42 months were often PCR negative, whereas those who had been followed up for a median of 21 months often remained positive.

Finally, using RT-PCR, our group33 studied 7 patients with CML who achieved complete cytogenetic remission after treatment with interferon alfa. We found that individual myeloid and erythroid colonies from blood and bone marrow samples in 2 of these patients expressed BCR-ABL transcripts despite their peripheral blood and bone marrow samples being PCR negative. One of these patients remains in ongoing complete cytogenetic remission, off therapy, several years after demonstration of BCR-ABL–positive clonogenic cells. These results imply that (1) a BCR-ABL–positive clone with potential for expansion remains present even in some patients who are PCR negative when peripheral blood or bone marrow is tested and (2) the malignant clone may remain clinically dormant.

Long-lasting remissions in the presence of persistent residual disease have also been observed in a variety of other hematologic malignancies. Roberts et al26 measured residual disease in 24 children with pre-B ALL by combining a quantitative limiting dilution PCR assay for IgH gene rearrangements with the clonogenic blast colony assay, as described by Estrov et al.34 Of the 24 patients, 7 relapsed, and 17 remained in remission up to 35 months after completion of maintenance therapy. Among these 17 patients in remission, 15 had evidence of residual disease. The blast colony assay was performed on 12 of the 15 PCR-positive patients and was positive in 7 indicating self-renewal capability of the malignant cells grown in colonies. Davis et al35 analyzed 5 patients with non-Hodgkin lymphoma who achieved prolonged complete remissions for 4 to 10 years after treatment with anti-idiotype monoclonal antibodies. They used enzyme-linked immunosorbent assays, flow cytometry, and PCR for clonal gene rearrangements of t(14;18) translocations. In all 5 patients, low-level residual lymphoma was detected without clinical relapse. Miyamoto et al36 described the detection of AML1-ETO messenger RNA transcripts by RT-PCR in bone marrow and peripheral blood samples from 18 patients with t(8;21)–positive acute myelogenous leukemia (AML) who have been in complete remission for up to 150 months without evidence of relapse. Likewise, persistence of RAR-α/PML transcripts in patients with t(15;17)–bearing acute promyelocytic leukemias and of CBFβMYH11 transcripts in patients with inversion of chromosome 16 and AML have been found, in select cases, to be compatible with prolonged clinical remission.37,38 The identification of residual disease markers such as the BCR-ABL transcripts in healthy individuals, although rare and at low levels, provides further explanations for positive tests in leukemia patients in long-term remission and also underscores the observation that cure is not necessarily synonymous with absence of disease.39,40 This concept has important implications for our understanding of disease biology, tumor dormancy, and mechanisms of immune surveillance.

That immune surveillance or the lack thereof plays an important role in control of residual disease can be concluded from various clinical observations. Patients with underlying immunodeficiencies or dysregulations of the immune system are at increased risk to develop certain lymphomas. As an example, B-cell lymphomas occur up to 50 times more frequently in patients after organ transplantation and subsequent immunosuppression than would be otherwise expected.41 In CML, a role for immunosurveillance in the control of residual disease can be concluded from several findings: (1) some patients with CML who relapse after allogeneic stem cell transplantation regain cytogenetic remissions after donor lymphocyte infusions42; (2) a positive correlation is found between the absence of graft-vs-host disease and disease recurrence; and (3) a positive correlation may exist between cytogenetic responses and the presence of interferon alfa–associated autoimmune phenomena.43 

Some patients with hematologic malignancies can expect long-lasting, disease-free survival and eventually cure from their disease. However, a significant proportion of patients still relapse after initially successful therapy and will die from their disease or complications of subsequent therapy. Disease recurrence is believed to arise from surviving clones of malignant cells that have somehow evaded containment in the host organism. It is therefore imperative to be able to assess the amount and relevance of residual disease after a complete clinical remission with induction treatments. Crucial questions arise in this setting. Is further therapy warranted in the presence of residual disease in order to forestall relapse or should further therapy be withheld? Is it safe, in some circumstances, to observe a patient with measurable residual disease, thereby avoiding the morbidity and mortality of treatment?

The identification of disease beyond the threshold of clinical detection has been greatly aided by the development of powerful molecular techniques, most notably PCR and its variants. Numerous studies of residual disease have significantly enhanced our knowledge in this area. Despite its potency, PCR assays should be interpreted cautiously, and possible pitfalls leading to false-positive and false-negative results should be considered. At this stage it is too early to translate results of clinical trials into clinical decision making. However, it appears that (1) there is no 1-to-1 correlation between PCR positivity and recurrence and relapse vs cure, and (2) the results of serial PCR studies may be more valuable than those at single time points.

Finally, one of the most notable findings to emerge is the observation that patients with evidence of residual disease can, at times, remain in long-lasting remissions. The physiologic and immune conditions mediating tumor dormancy, therefore, merit further investigation.