The Pathologic Diagnosis of Pediatric Soft Tissue Tumors in the Era of Molecular Medicine: The Sarcoma Pediatric Pathology Research Interest Group Perspective

Depending on the diagnosis and tumor type, specimens obtained for tissue diagnosis must serve multiple objectives: (1) establish the diagnosis, (2) identify features that determine prognosis and/or influence treatment, (3) identify any features of underlying genetic syndromes, and (4) archive materials for future patient care needs, tumor banking, disease registries, research, or clinical trial enrollment. Establishing and confirming the diagnosis is the foremost priority because this determines the context for all other considerations. For malignant tumors in which the histologic grade may influence the treatment approach, evaluating and documenting the tumor's grade, including the system used, is also important.

Surgeons or interventional radiologists frequently ask how much tissue they should obtain, and the answer depends on multiple factors, including tissue viability, cellularity, extent of tumor content, and the anticipated amount needed for potential diagnostic and prognostic testing, tumor banking, research, or clinical trials. Tumor location, accessibility, and individual patient factors may also impact how much tissue can be safely retrieved during an operative procedure. Multiple studies have shown that core needle biopsy (CNB) in the pediatric oncology setting is associated with fewer complications and decreased recovery time compared with open biopsy (OB) procedures, and that the adequacy rate (89% CNB versus 95% OB); diagnostic rate (87% CNB versus 95% OB); and adequacy for ancillary testing needs (89% CNB versus 92% OB) are comparable between the procedure types.²  Furthermore, 1 core may be entirely diagnostic and adequate for ancillary needs in some cases, whereas multiple cores or OBs can still be inadequate in other cases. The pediatric studies described above demonstrate a range of core biopsy numbers, from 1 to 20, with the mean number of cores collected ranging from 4 to 6.²  Accordingly, we make the recommendation that 4 to 6 cores of viable tissue will be adequate for establishing a diagnosis supported by molecular data in most cases. For each additional tissue allocation (research, banking, etc), another 1 to 2 cores, comprising high tumor percentage, should be obtained for each request. In total, this sampling strategy is likely to satisfy most needs.

The amount of tissue needed for genetic testing depends on the laboratory and the platform(s) used for testing, especially when a tiered approach to diagnosis is used. Directed tests using polymerase chain reaction (PCR) assays may require as little as 50 ng, whereas some next-generation sequencing (NGS) assays could require micrograms of nucleic acid. Samples with a volume greater than 8 mm³ easily yield at least 1 μg of DNA, sufficient for most NGS assays, and half that volume frequently yielded up to 1 μg of DNA in one study of paraffin-embedded tissues of multiple types.³  Calculations from this study suggest that one 9-mm, 18-gauge core biopsy will produce 1 μg of DNA 90% of the time, and two 18-gauge cores will produce 1 μg of DNA 99% of the time.³  Jamshidi et al⁴  showed that one 18-gauge core biopsy of lung tumor tissue yielded about 5 times the amount of nucleic acid compared with one 20-gauge core biopsy of the same tissue (average, 1840 ± 405 ng per 18-gauge core versus 383 ± 68 ng per 20-gauge core).⁴  Twenty-gauge needle core biopsies are less successful than 10-, 14-, 16-, or 18-gauge needle cores in obtaining sufficient tissue.²  Smaller-gauge needles (larger bore size) yield more tissue than higher-gauge needles (smaller bore size) and result in higher nucleic acid recovery. End-cut biopsy needles retrieve more tissue than side-cut needles. Hoang et al⁵  showed that one 16-gauge needle core biopsy (end or side cut) obtains 1 μg of DNA 99% of the time, as does one 18-gauge end-cut biopsy, but more than one 18-gauge side-cut biopsy and as many as five 20-gauge side-cut biopsies may be required to obtain 1 μg of DNA.⁵ 

Research or clinical trial enrollment tissue requirements vary considerably, and some request multiple grams of tissue. Volumetrically, a 17-mm, 18-gauge needle core biopsy is about 13 mm³. Accordingly, 77 such cores would be needed to approach 1 g of tissue.⁵  For this reason, a clear understanding of study requirements and clinical feasibility is important before tissue is obtained.

A multidisciplinary approach to each case optimizes tissue retrieval procedures. Communication with requesting providers clarifies other aims aside from establishing or confirming the diagnosis, including needs for study enrollment, banking, or treatment-related testing. These conversations help determine how much tissue is needed, which must then be relayed to the physicians who procure the tissue, whether they are surgeons, interventional radiologists, or other types of physicians. Depending on the features of the tumor, such as size, location, and extent of intratumoral necrosis, the amount of tissue procured may be limited. Intraprocedural evaluation of the collected tissue for adequacy by the pathologist improves diagnostic tissue yield by ensuring tumor representation and viability, reducing inadequacy rates.^2,6  Rapid on-site evaluation, intraoperative touch preparations, or frozen section examination can meet this need, and both touch preparations (best if unstained) and frozen section remnants can subsequently be used for molecular testing. Also, if molecular characterization is a known need, coordinating collection of normal tissue for paired germline mutation analysis may be important. A summary of these points is provided in Table 1 for ease of reference.

Formalin-fixed, paraffin-embedded tissue (FFPE) is often the most important resource to meet a variety of tissue needs. Well-prepared hematoxylin-eosin–stained sections comprise the first step to generating a morphologic differential diagnosis and in turn direct other ancillary testing (Table 2). Immunohistochemical (IHC) stains are often the first ancillary tests to consider, as discussed below in our pattern-based approach. Depending on the gauge of the CNB, anywhere from 10 to 30 histologic sections can be successfully created. If all cores are equal in tumor content, division of multiple cores of tissue into multiple tissue blocks increases the number of possible tissue sections; optimally 1 to 3 cores per block depending on individual laboratory practices. Unstained levels of tissue cut at intervals can be quickly used for immunohistochemistry or other ancillary testing without requiring re-facing of the tissue block and unnecessary loss of tissue.

Use of FFPE for molecular testing is steadily improving, even for archival tissue. Although some laboratory-based details may vary, a few generalizations can be made to increase success in tumor sequencing. Decreasing time to fixation minimizes nucleic acid degradation due to cold ischemic time and results in better morphology after routine processing.⁷  However, formalin can impact nucleic acid quality (particularly RNA) if fixation time is prolonged, especially if more than 72 hours.⁷  Thicker scrolls, thicker sections, increased numbers of sections, or use of an entire block of very limited biopsies may be necessary for successful genetic testing. For larger portions of tissue, nucleic acid retrieval may be more successful from thinner sections of large tumor sections than from thick scrolls because increased cellularity and high paraffin content may compromise the purification process, and in one study they contributed to a higher rate of testing failure.⁸  Tumors with high fibrous or myxoid matrix content typically also have lower nucleic acid content and may require more material for testing than other tumor types.⁸ 

An important consideration concerns tissue that requires decalcification. Use of strong acids, such as hydrochloric or nitric acid, to decalcify tissue may result in denaturation of nucleic acids, precluding performance of molecular diagnostic techniques or eligibility for research studies. Ethylene diamine tetraacetic acid (EDTA)–based decalcification results in better preservation of the nucleic acids, although it requires increased time to adequately decalcify the tissue for sectioning. EDTA and formic acid combination solutions can decrease the time required for decalcification compared to EDTA alone, but formic acid may still adversely affect nucleic acid recovery if exposure is prolonged.⁹ 

Preparation for handling a resection should include review of any radiographic imaging and available operative notes to help with specimen orientation and to provide contextual information. Specimen photography should be used to document steps of handling, such as before and after sectioning and following application of surface ink for margin definition. These photographs can be useful in multidisciplinary tumor boards to inform margin discussions and to guide other aspects of surgical and medical care.

Tumor mapping is an underused aspect of tumor resection assessment. Submitting an entire cross section of a tumor, similar to the standard handling process required for posttreatment osteosarcoma resections, allows validation of the radiographic and gross assessment of extent of tumor necrosis or treatment response. If a mapping strategy is not used, representative sections of the tumor should assess the integrity of all surgical margins and provide a thorough sampling of different regions of the tumor, especially when there is heterogeneity of the gross appearance. Gross estimates of percent necrosis (with histologic correlation) inform tumor grading. Thorough sampling can reveal a spectrum of grades or divergent or dedifferentiated components not sampled in initial biopsies that may alter therapeutic plans. Sampling areas of normal adjacent tissue can be useful for comparing germline and tumor genetic alterations, studying tumor microenvironment, and characterizing surgical margins. Liberal tumor banking should be considered whenever possible, for future study needs.

The pathology report for resection specimens should include the features important for treatment planning: maximal tumor diameter, grade and grading system used (if applicable for the tumor type), and distance of tumor from surgical resection margins. For cancer resection specimens obtained after neoadjuvant therapy, the extent of necrosis should also be reported as it may influence prognosis and treatment approach.¹ 

Assessment of the predominant histologic pattern is the first step in developing the differential diagnosis and guides the selection of ancillary tests. Common nonmesenchymal pediatric tumors, such as Wilms tumor, lymphoma, histiocytic neoplasms, epithelial neoplasms, melanoma, and germ cell tumors, should be ruled out before more detailed evaluation of soft tissue tumor diagnoses. Frozen section, touch preparation, or routine histology usually is sufficient to accomplish initial categorization.

Distinction of generalized patterns, including round cell, spindle cell, epithelioid, and pleomorphic patterns, is a basic but useful approach. Round and spindle cell patterns are the most common in children and are addressed below in detail, with their most salient features displayed in Tables 3 and 4, respectively. Tumors with epithelioid features that occur in children often have round cell features, and they generally fall within the differential diagnoses summarized in Table 3. Epithelioid tumors with evidence of neural differentiation (S100 or SOX10 IHC staining expression) can be differentiated using the spindle cell tumor approach in Table 4. Epithelioid vascular tumors, including epithelioid hemangioma, pseudomyogenic hemangioendothelioma, epithelioid hemangioendothelioma, and epithelioid angiosarcoma, are all rare lesions than can occur in young patients, typically display at least 1 or more of a variety of vascular markers (ERG, CD34, CD31, WT1), and have distinctive molecular alterations, reviewed elsewhere.¹⁰  Pleomorphic soft tissue tumors are much less common in children than in adults and are often characterized by TP53 loss. Copy number or chromosomal testing to confirm genetic complexity in these tumors may be useful for treatment stratification.

Because it is performed on the FFPE tissue block, IHC typically precedes other types of testing. IHC findings can be difficult to interpret if there is nonspecific or weak staining, and because availability and quality of reagents can change over time.

Round cell tumors typically generate a robust differential diagnosis. Ewing sarcoma is the prototypical small round blue cell tumor. Hematolymphoid malignancies, particularly lymphoblastic lymphoma, may have similar IHC and morphologic features but can often be recognized on touch preparations because of a lack of cellular cohesion. Other diagnostic considerations include rhabdomyosarcoma, neuroblastoma, synovial sarcoma, desmoplastic small round cell tumor, and malignant rhabdoid tumors/SMARCB1-deficient tumors, among others, depending on tumor location, patient age, and other morphologic features. An initial panel of immunohistochemistry for undifferentiated round cell tumors typically includes CD99 or Nkx2.2, desmin, myogenin or MyoD1, INI1, PHOX2B, a pancytokeratin, CD43, and/or TdT.^11–16  If those stains do not provide a diagnosis, less common diagnoses should be considered (like CIC- or BCOR-altered sarcomas). IHC stains for BCOR, ETV4, ERG, and DUX4 are available at specialized institutions and may be very useful in determining which, if any, molecular confirmatory testing is necessary.

Pediatric tumors with spindle cell morphology include embryonal rhabdomyosarcoma, spindle/sclerosing rhabdomyosarcoma, malignant peripheral nerve sheath tumor, synovial sarcoma, inflammatory myofibroblastic tumor, infantile fibrosarcoma, other NTRK-rearranged spindle cell tumors, and dermatofibrosarcoma protuberans, the primary features of which are reviewed below (Table 4).

The overlapping histology of pediatric spindle cell tumors presents diagnostic challenges in distinguishing these lesions. A useful initial panel of IHC stains includes SMA, desmin, myogenin, MyoD1, S100 or SOX10, β-catenin, pancytokeratin/EMA, and CD34.^11–16 

Genetic testing can inform diagnosis but can also identify prognostic biomarkers, therapeutic targets, or germ-line mutations that impact future health and cancer screening.^16–18  For diagnostic genetic testing, chromosomal and molecular testing are most commonly used.

Chromosomal testing (conventional cytogenetic analysis or karyotyping) examines entire chromosomes for large structural changes, such as large deletions, translocations, inversions, and duplications; smaller alterations typically are not detected. Karyotyping does not require a suspected diagnosis, must be performed on fresh cultured tumor tissue, and can take multiple weeks to complete analysis, depending on how well the tissue culture grows. Chromosomal microarray detects large or small structural abnormalities, as small as 100 kb in size, but not point mutations.

Molecular testing interrogates specific gene targets, individually or together as part of a panel. Molecular testing for diagnosis can roughly be divided into targeted/limited assays and broad/panel-based assays, with strategies that may focus on either DNA or RNA analysis. It is important that panel testing is enriched for known pediatric tumor gene targets because many widely available panels are designed with adult solid tumors in mind. The best approach to molecular testing varies according to the morphology-guided differential diagnosis, testing availability/access, tissue availability, time required for testing, and sometimes cost. Just as IHC approaches vary by pattern, so does the recommended molecular testing approach. Round cell tumors often involve an initial targeted approach because several entities in the differential diagnosis are far more common than the rest and targeted assays may have faster turnaround time than broad panel or chromosomal testing. Spindle cell tumor types can have significant morphologic overlap, making a broad-scope, panel testing approach highly appropriate, if IHC does not reveal a specific diagnosis. The Figure summarizes an approach the authors find useful and efficient.

There are a range of targeted assays that can be used to support a diagnosis. The primary benefits of targeted testing are that it is readily available, relatively rapid, and covers the most common, canonical molecular alterations (often fusion transcripts). These are particularly valuable when the differential diagnosis is limited and the assay is being used to rule in or rule out a particular entity. Targeted assays interrogate a specific point mutation, a single gene, or a fusion gene. Single genetic “lesion” assays, such as point mutation detection (eg, MYOD1 L122R, BRAF V600E) or internal tandem duplication detection (eg, BCOR ITD) may be performed using a variety of methods including standard PCR, Sanger sequencing, and digital droplet PCR. These alterations can also be detected by NGS methods, but in that case are typically included as part of a larger panel (see next section).

Recurrent gene rearrangements are readily targeted by fluorescence in situ hybridization (FISH) assays using break-apart probes for the gene of interest (eg, EWSR1, FUS, ALK). FISH has the advantage of rapid turnaround time compared with other molecular modalities, and it does not require a known fusion partner. The major limitation of FISH testing is the inability to identify fusion partners, and false-negative results may occur. For example, EWSR1 rearrangements can be seen in Ewing sarcoma, desmoplastic small round cell tumor, angiomatoid fibrous histiocytoma, myoepithelial tumors, clear cell sarcoma, and other entities. All of these tumors could have a positive result for the EWSR1 break-apart probe. Therefore, depending on the clinicopathologic, morphologic, and immunophenotypic features, nucleic acid–based testing may be necessary to distinguish tumors associated with EWSR1 fusions. Reverse-transcription PCR (RT-PCR) is another technique with relatively quick turnaround time that can be used to confirm the presence of a suspected fusion. The downside of this technique is that it can only be applied to fusions with reproducible and well-described breakpoints and partner genes. Common fusions like those seen in Ewing sarcoma, alveolar rhabdomyosarcoma, desmoplastic small round cell tumor, and synovial sarcoma can be easily detected using single or focused panel RT-PCR testing. If multiple targeted assays are required, material available for testing is limited, or if a targeted test is negative, then broad multigene testing may be a more cost-effective and efficient initial testing choice. As a side note, if a targeted assay is performed and is negative, reimbursement can be difficult when subsequent multiplex testing includes the same target as the initial targeted test.

There is increasing availability of unique assays for broad-based molecular interrogation of tumors using limited amounts of tissue. These range from readily available panels of 50 to hundreds of genes to whole exome sequencing (WES) and whole genome sequencing (WGS). Each panel will have its own set of detection capabilities depending on the assay design, validated sensitivity, gene list, and laboratory pipeline. When selecting a broad assay, the pathologist should consider all possible differential diagnoses to determine if the assay will adequately identify the potential alterations under consideration.

The most commonly used methodology for these broad assays is NGS, although others exist as well.¹⁸  NGS technologies, as opposed to traditional Sanger sequencing, share a common foundation of nucleic acid extraction, library preparation, and massive parallel sequencing of millions of small fragments of DNA (or RNA).^19,20  Data analysis is then performed and involves bioinformatics pipelines that align the multiple sequencing reads of DNA/RNA to the known reference structure of the human genome with identification of genomic alterations/variants. Various NGS assays may examine whole genome, whole exome, or a defined panel of genes of interest.^19,20  Several different NGS platforms are available for the detection of a variety of genomic alterations, including small base changes, insertions-deletions (indels), and/or larger genomic changes including deletions of whole exons, inversions, translocations, and copy number alterations, the differences of which are beyond the scope of this manuscript. Other broad-based technologies that have been used in clinical laboratories include chromosomal microarray for copy number detection and NanoString for fusion and RNA expression. One can subclassify broad assays in different ways, but a helpful distinction is to consider the type of nucleic acid tested.

DNA-based NGS assays may include WGS, WES, or selected gene panels.^19–21  Focused gene panels are the most widely and readily available assay. These panels sequence exonic, select intronic, and promoter regions of specific genes in order to detect certain gene alterations. Their detection limits are variable but are typically on the order of 2% to 10%, depending on variant type. WES involves sequencing exonic (protein encoding) regions of the genome, which is about 2% of the mammalian genome.^21,22  WGS is the broadest type of sequencing but is not widely available at this time for clinical diagnosis. WGS interrogates all regions of the DNA including introns and promoters and can allow for the detection of structural rearrangements (fusions) without the need for RNA analysis. This is advantageous because although both DNA and RNA degrade with age and in the process of formalin fixation and paraffin embedding, DNA is less susceptible to degradation than RNA.^19,20  Copy number detection using DNA input can be achieved by NGS methods or also using chromosomal microarrays.^19–21 

Methylation array/profiling methodology has gained momentum because of its clinical utility in the diagnosis and management of brain tumors, and recent publications span other tumor types, including sarcoma.²³  This technology looks at the tumor methylome and has found unique methylation profiles for various tumors. This technology can be helpful in tumors where no diagnostic genetic aberration is found via NGS or other molecular methodologies, in order to find a “best fit” methylation profile. Recent studies have demonstrated that within the group of round cell sarcomas, Ewing sarcomas (FET::ETS fusions) have methylation profiles distinct from other round cell sarcomas, including CIC-rearranged sarcomas, EWSR1::NFATc2 fused sarcomas, and BCOR-altered sarcomas, among others.²⁴  On the other hand, pleomorphic sarcomas, such as pleomorphic liposarcoma and undifferentiated pleomorphic sarcoma, have similar methylation profiles, making methylation studies less helpful in this context.²³  In general, methylation may be a useful tool in specific contexts, but in others it may not be definitive, at least until there are more data surrounding methylation testing in sarcomas.

RNA-based assays interrogate RNA transcripts present in tumor cells. Most commonly in clinical practice, RNA sequencing is used to examine mRNA (derived from exons) and to identify gene fusions. In using RNA for fusion identification, intronic regions are no longer a barrier for sequencing. Panel-based fusion assays are widely available and often use technology that is partner agnostic; therefore, these methods are most useful in gene fusions with multiple rare partners and for discovery of novel partners.^19,21,25  However, these panels have important limitations based on gene/exon selection and assay design, making them sometimes inadequate to detect all possible fusion genes. For example, a recent comparative study evaluating 7 RNA-based targeted sequencing assays demonstrated that although all assays listed coverage of all 3 NTRK genes, only 2 of the 7 assays were designed to cover all known NTRK gene fusions within those genes.²⁶  Other panel technologies for fusion detection, such as NanoString, do require knowledge of both fusion partners but have the advantage of no amplification requirements and fast turnaround time (∼48 hours), making them relatively robust for degraded nucleic acid and diagnostic purposes.²⁷  Sequencing the entire transcriptome to identify fusions (RNA-seq) has also been used to identify novel fusions and fusions not covered by panel assay.²⁸  However, rare tumors can have very cryptic genetic abnormalities, such as CIC-rearranged sarcomas, for which many different molecular modalities, including RNA-seq, can be ineffectual.²⁹  Alternative methods of successful confirmation can include immunohistochemistry and DNA sequencing. RNA expression can be measured by a variety of methods, including whole-transcriptome shotgun sequencing, NanoString, and expression arrays.^19,25,27,30 

Tumors without an identified genetic driver that present at high stage, that recur, or that progress despite therapy may need to be subjected to more advanced molecular testing, like WGS or whole-transcriptome shotgun sequencing to find obscure genetic alterations, secondary alterations, or therapeutic targets.

Reserving tissue for research and for biobanking is an investment in our future knowledge about the biology of tumors. This tissue can be used to identify new biomarkers, look for newly discovered biomarkers across different kinds of tumors, identify therapeutic targets, and apply new technologies that may not have been available at the time of diagnosis. Annotated tumor samples obtained from patients enrolled in well-constructed prospective clinical trials are often the optimal source for these kinds of investigations. Initial diagnostic tissue is often ideal for these studies, but as discussed above, increasing use of minimally invasive biopsy procedures often challenges the ability to bank tissue for research or archival purposes. Collection of tissue from posttreatment resection specimens, at the time of disease recurrence, progression, or metastasis, also provides value as we explore biologic differences between different specimen types for a single patient or between different patients. Finally, tumor harvesting at the time of autopsy can provide ample tissue from multiple anatomic sites for research purposes.^31,32 

The collection, characterization, and storage of biospecimens from tumor specimens require coordinated collaboration across disciplines. Institutional biorepositories require consistent, standardized procedures for sample acquisition, processing, annotation, and storage. The College of American Pathologists provides guidelines for biorepository processes, and College of American Pathologists accreditation ensures the quality of the biorepository. Tissue that is archived for research purposes should be divided into multiple small portions (at least 25 mg each), collected in cryosafe tubes for snap-freezing, and stored in a −80°C freezer. The number of tubes stored depends on the study protocol and the amount of tissue received. Ideally, at least 1 vial of tumor and 1 vial of normal tissue are saved for potential future studies. Other important considerations include verification of approval by the appropriate Institutional Review Board for studies using collected tissue, at least before tissue is released for research studies. Creating guidelines for the prioritization of tissue distribution by a multidisciplinary group may also help optimize biorepository utilization.

Pediatric soft tissue tumors can be difficult to diagnose given overlapping morphologic features, increasing knowledge of their biology, and evolving classification systems. The foundational principles we have outlined here are intended to provide an approach to tackling these tough cases. Optimal handling of limited specimens can lead to faster diagnosis and may obviate a need for additional biopsy procedures after depletion of the initial sample. The pathologist is the expert custodian of this tissue and needs to make careful decisions about how these precious specimens should best be used and archived. Although a fully characterized diagnosis is always the first priority, sensitivity toward tissue handling positions the pathologist centrally in a collaborative culture of scientific advancement. Such foresight creates exponential opportunities for improving how we diagnose, manage, and treat this diverse group of diseases.

The Sarcoma Pediatric Pathology Interest Group (SPPRITes) includes pediatric pathologists from multiple organizations in the United States who focus on the study of pediatric bone and soft tissue neoplasia and endorse collaboration and fellowship, professional development, research, education, and clinical advancement.