Article Text

Download PDFPDF

Ancillary molecular analysis in the diagnosis of soft tissue tumours: reassessment of its utility at a specialist centre
  1. Katherine Vroobel1,
  2. David Gonzalez2,
  3. Dorte Wren2,
  4. Lisa Thompson2,
  5. John Swansbury3,
  6. Cyril Fisher1,
  7. Khin Thway1
  1. 1Sarcoma Unit, Royal Marsden Hospital, London, UK
  2. 2Department of Molecular Diagnostics, Royal Marsden Hospital, Sutton, Surrey, UK
  3. 3Department of Clinical Cytogenetics, Royal Marsden Hospital, Sutton, Surrey, UK
  1. Correspondence to Dr Khin Thway, Sarcoma Unit, The Royal Marsden NHS Foundation Trust, 203 Fulham Road, London SW3 6JJ, UK; khin.thway{at}


Aims The histological diagnosis of soft tissue tumours (STTs) can be difficult, sometimes requiring a combination of morphology, immunophenotype and ancillary molecular tests. Many STTs are associated with characteristic genetic aberrations that can be assessed using fluorescence in situ hybridisation (FISH), reverse transcription-PCR (RT-PCR) or mutational analysis. We have previously assessed the practicality and sensitivity of using these modalities as part of the routine diagnosis of STT in paraffin-embedded material and now revisit the subject in light of further experience in this field.

Methods 200 consecutive cases from 2013 that had undergone FISH, RT-PCR or mutational analysis were assessed to evaluate their diagnostic utility compared with preliminary histological assessment.

Results 218 FISH, 91 RT-PCR and 43 mutational analysis tests were performed. Compared with the previous study, FISH for MDM2 amplification in possible well-differentiated/dedifferentiated liposarcomas, and mutational analysis for assessing KIT, PDGFR and BRAF mutations made up a large proportion of the workload (107 and 43 tests, respectively). As in the previous study, alveolar rhabdomyosarcoma showed the best FISH:RT-PCR concordance. Unlike previously, RT-PCR showed marginally higher sensitivity than FISH (78.9% and 76.9%), while continuing to demonstrate higher specificity (90.9% and 84.3%). RT-PCR again showed an increased failure rate (5.5%; 1% for FISH).

Conclusions We demonstrate the continuing utility of RT-PCR and FISH for STT diagnosis, and that each has advantages in specific contexts. These ancillary molecular tests are important tools in both defining and excluding diagnoses of STT, which is crucial in determining prognosis and guiding appropriate management.

  • FISH

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The diagnosis of soft tissue tumours is challenging as their morphology and immunophenotype often overlap with each other and with tumours of other lineages such as carcinoma and melanoma. Many soft tissue tumours are associated with characteristic, reproducible genetic aberrations (chromosomal translocations that generate novel chimeric fusion genes, gene amplifications and gene mutations), several of which have only been recently defined, such as EWSR1 rearrangement in soft tissue myoepithelial tumours, and these can be used for diagnostic purposes. The ability to achieve reliable molecular test results from formalin-fixed, paraffin-embedded (FFPE) tissue has been well documented; fresh frozen tissue remains optimal, but attaining such material in routine histopathological practice is generally impractical, particularly for referred cases. The molecular diagnostic service at the Royal Marsden Hospital (RMH) comprises two laboratories, performing fluorescence in situ hybridisation (FISH) and reverse transcription-PCR (RT-PCR), respectively. We previously evaluated the diagnostic usefulness of FISH and RT-PCR on soft tissue neoplasms in FFPE material at a relatively early time in the set up of molecular and molecular cytogenetic diagnosis as routine ancillary diagnostic testing.1 We now reassess this 5 years later, comparing to the previous study, to see whether there were discernible improvements in test efficacy. Since this last study, routine molecular testing for gastrointestinal stromal tumours (GISTs) to assess the mutation status of KIT and PDGFRA genes has been implemented, and FISH to assess MDM2 amplification in differentiated (WDL)/dedifferentiated liposarcomas (DDL) has also been introduced. The experience has also increased within the department as more molecular tests are routinely performed.

Materials and methods

A retrospective reaudit was performed for 200 consecutive cases that had FISH, RT-PCR or mutational analysis performed, beginning in March 2013. Cases were retrieved from the molecular and molecular cytogenetics databases (DG and JS). Similar to those in the previous study, most cases for which molecular analysis was performed carried a degree of uncertainty as tests were carried out either for confirmation of a suspected entity or for exclusion of a possible, but less likely, tumour type. For FISH, 2–4-µm-thick FFPE sections were treated as previously described.1 However, since the previous report, an extra step in the processing of slides for FISH has become routine: 3.5 L of distilled water and 35 mL of antigen retrieval buffer are brought to a boil in a pressure cooker; the de-waxed slides are added to this solution and the pressure is raised. After 5 min, the cooker is cooled and the slides are washed twice in distilled water before proceeding to the hybridisation stage, performed overnight according to manufacturer's protocols. RT-PCR was performed to assess for fusion transcripts according to standard or previously described methods.1 ,2 For each patient, results of RT-PCR and FISH were compared, and these findings were compared with the histology.


Two hundred cases seen consecutively in the Sarcoma Unit that required molecular testing were identified. A total of 218 FISH tests, 91 RT-PCR tests and 43 tests for mutational analysis were requested on these cases. In total, 90 were internal cases processed locally, whereas 110 were received as blocks from elsewhere (for second opinion or diagnostic review). Tests using FISH were positive (identified a translocation or gene amplification) in 72 cases (33.3% of successful tests) and negative in 144 (66.7% of successful tests). Two tests were unsuccessful (1% of total), both of which were from referred blocks. Of the RT-PCR tests (corresponding to 41 cases), positive results (for which a fusion transcript was identified) were obtained for 17 tests (19.8% of successful tests) and negative results for the remaining 80.2% (69 tests). Five tests were unsuccessful, three of which were performed on internal cases and two on external (5.5% of total tests). This is markedly lower than in the previous study, for which 25.2% of RT-PCR tests and 5.9% of FISH tests were unsuccessful. In this new study, mutational analysis was included in the assessment; there was a total of 43 tests (corresponding to 22 cases) for PDGFRA, KIT and/or BRAF mutations. Of these cases, 16 had a mutation (72.7%), and 2 tests failed, one of which was an external case and one of which was internal. As described in previous studies, we demonstrated that the most common mutation in gastrointestinal stromal tumours was within exon 11 of KIT (60% in this study).

Where the diagnosis was relatively certain on morphological assessment, molecular tests (either FISH or RT-PCR) were positive in 100% of alveolar rhabdomyosarcoma, myxoid liposarcoma and synovial sarcoma. In these cases, both tests, when performed together, were concordant (table 1). FISH for MDM2 amplification for diagnosis or exclusion of WDL/DDL made up a large proportion of all FISH tests requested (107 requests, 49% of total). MDM2 amplification was found in 81.6% of all cases thought to represent WDL/DDL on morphological assessment, and in 14.5% of cases for which it was performed for diagnostic exclusion (eg, in cases of pleomorphic lipoma, large or deep lipoma; table 2). Of the other entities, the second most frequent request was for synovial sarcoma (26 cases), and this is similar to the findings of the previous study. The least frequent requests were for clear cell carcinoma, desmoplastic small round cell tumour and infantile fibrosarcoma (one case each).

Table 1

Cases tested for histological confirmation

Table 2

Cases tested for diagnostic exclusion

In determining the sensitivity and specificity of correct detection of individual soft tissue neoplasms, 167 cases were suitable for analysis for FISH and 41 for RT-PCR. For FISH testing, the sensitivity was 76.9% and specificity 84.3%. For RT-PCR, the sensitivity was 78.9% and specificity 90.9%. Two FISH tests failed (1%), five RT-PCR tests failed (5.5%) and two mutational analysis PCR tests failed (4.7%). Thirty cases had both FISH and RT-PCR performed. Of these, 13 were both positive (43.3%) and 14 were both negative (46.7%). Discordant results were obtained in two cases (6.7%) (table 3). One case demonstrated an EWSR1 gene rearrangement but without detection of the fusion transcript by RT-PCR; this was interpreted as the presence of an EWSR1 translocation with a rare fusion partner that could not be identified by the probes available to us, rather than a true discrepancy. The final diagnosis in this latter case was of myoepithelial carcinoma.

Table 3

Cases with discordant results


A significant proportion of soft tissue neoplasms is associated with characteristic, reproducible genetic abnormalities, and their detection is now an integral part of the diagnostic investigation of soft tissue neoplasms (figure 1A–D). This follows from the recommendations by the National Institute for Health and Care Excellence that patients with histological or radiological provisional diagnoses of bone or soft tissue sarcomas should be referred to specialist multidisciplinary units for diagnostic evaluation by specialists; therefore, all hospitals should have defined referral pathways for the diagnosis of putative soft tissue tumours, with all cases being amenable to facilities for ancillary molecular diagnosis.3–5 The positive impact of ancillary molecular analysis on soft tissue tumour diagnosis is undisputed; a recent analysis of its utility in assessment of >700 sarcomas found that molecular tests were useful in confirming a probable diagnosis or allowing a possible diagnosis in a significant proportion of soft tissue neoplasms including GISTs, suspected translocation sarcomas and atypical lipomatous tumour (ALT)/WDL and DDL, and recommended that ancillary molecular tests be performed on any tumour with a suspected specific genomic abnormality and for which the diagnosis was uncertain.6

Figure 1

(A) This example of dedifferentiated liposarcoma (DDL) shows markedly pleomorphic spindled cells, including tumour giant cells and has a pattern of undifferentiated pleomorphic sarcoma (UPS), not otherwise specified. As there is no adjacent well-differentiated liposarcomatous component and the patient did not have a previous history of well-differentiated liposarcoma, it is not possible to differentiate this neoplasm from UPS, which has prognostic implications, as DDL tends to behave less aggressively than other pleomorphic sarcomas. (B) Fluorescence in situ hybridisation (FISH) for assessing MDM2 gene amplification is the diagnostic gold standard for well-differentiated and dedifferentiated liposarcomas. This is particularly useful as immunohistochemistry for MDM2 is often technically challenging. Here, the CEP 12 signals (green) are sited on the chromosome 12 centromere, while the MDM2 signals (red) are located on the long arm of chromosome 12. (C) Ancillary molecular investigations are particularly useful in the diagnostic evaluation of undifferentiated small round cell and spindle cell neoplasms. These typically require extensive immunohistochemistry, which may still not lead to a conclusive diagnosis as there is immunohistochemical overlap with several tumours of different lineages. This example of poorly differentiated synovial sarcoma was difficult to distinguish morphologically and immunohistochemically from both Ewing's sarcoma and small cell carcinoma. However, molecular investigations with FISH and reverse transcription-PCR were diagnostic. Additionally, for many malignant undifferentiated neoplasms, molecular investigations are valuable in excluding specific diagnoses. (D) Molecular diagnostic investigations can also be helpful in diagnosing unusual variants of some neoplasms. Low-grade fibromyxoid sarcoma usually comprises patternless distributions of spindle cells that lack cytological atypia. However, this example showed marked nuclear pleomorphism. RT-PCR revealed FUS-CREB3L2 fusion transcripts and FUS rearrangement was seen with FISH, confirming the diagnosis.

This study demonstrates that FFPE tissue continues to represent a reliable and practical resource for molecular tests. In an experienced laboratory, the overall failure rate has been reduced to <5%, which represents a vast improvement compared with the previous study (RT-PCR failure rate reported previously at 25.2% of all cases). The changes in diagnostic pattern and workload are also reflected in this re-evaluation, with two main increases in the tests performed. The number of samples requiring mutational analysis has risen significantly since the previous study, as this is now mandatory for all GISTs, to enable the assessment of treatment response and for prognostication. Also, FISH for the assessment of MDM2 amplification was introduced and has been an invaluable diagnostic adjunct (figure 1A, B), particularly in distinguishing ALT/WDL from lipomas. There is a proportion of differentiated adipocytic neoplasms whose morphology confounds even soft tissue pathologists, and MDM2 FISH is considered the definitive method for their diagnosis.7

There are a number of reasons for the improvement of both RT-PCR and FISH technical success rates in the five years since the previous study. Previously, the failure rate for FISH was 10/176 (5.7%) compared with 2/218 now (0.92%). The improvement may be due to the introduction of the extra step described in the methods, which has been shown to be particularly beneficial for material that is suboptimal in some way. The marked improvement in RT-PCR success rates (including with material from other institutions) is likely to be attributable to the continued experience of the staff of both histopathology and molecular laboratories, in the optimal processing, preparation and assessment of specimens for molecular analysis (figure 2). Second, in the genomic era, is the recognition among all histopathology departments of the importance of molecular diagnostics as a crucial part of the patient pathway for the assessment of genetic aberrations for solid tumours with available targeted therapies (eg, lung, breast and colorectal cancers). This has led to marked general improvements from all hospitals in the fixation and processing of specimens to generate optimal quality material for molecular assessment, in contrast to the specimens assessed in the previous study. In our molecular diagnostics laboratory, the only protocol change from previously is the use of 3–5 µm tissue sections rather than 20 µm scrolls for RT-PCR and mutational analysis. Possible reasons for the increased technical success might be because deparaffinisation is now performed on slides rather than in microcentrifuge tubes, and also that the previous use of a larger amount of (non-microdissected) material may have caused saturation of PCR columns with paraffin and/or undigested tissue, or carryover of PCR inhibitors.

Figure 2

Example of samples positive for the SS18-SSX2 fusion of synovial sarcoma, but with different RNA qualities. RNA quality is assessed for each sample by analysing amplification of the house-keeping gene B2M: amplification with a cycle threshold (CT) <30 indicates acceptable RNA quality. Sample A shows good RNA quality (B2M CT of 25 (pale blue line)), whereas the B2M CT of sample B is 33.5 (purple line). The fusion transcript is clearly amplified in sample A (dark blue line, CT of 31) whereas in sample B amplification for SS18-SSX2 (light purple line) is only visible at cycle 38 out of 40 PCR cycles.

As before, it is optimal that FISH and RT-PCR are used in conjunction to maximise tumour detection rates. Studies comparing the utility of FISH and RT-PCR in diagnosing individual soft tissue neoplasms have highlighted that these modalities should be used complementarily; while it is generally recognised that FISH remains more sensitive overall in detecting a given genetic abnormality, FISH or RT-PCR may each identify abnormalities that the other fails to detect.8 For FFPE material, FISH has been considered the more technically robust method, less subject to the variability of different specimen fixation and processing methods. However, gene rearrangements may fail to be detected by FISH in samples containing low proportions of lesional cells, in contrast to RT-PCR, which usually benefits from the overexpressed fusion mRNA present. FISH may be unable to detect gene rearrangements in specimens for which fusion transcripts are detected by RT-PCR due to tumours harbouring variant gene fusions with small substitutions or insertions of DNA that are not detected by the commercial break-apart probes used.1

There remain specific advantages with each modality. It is increasingly apparent that gene fusions may be characteristic but not specific for given soft tissue neoplasms; EWSR1-CREB1 and EWSR1-ATF1 fusions are associated with several tumours of both mesenchymal and non-mesenchymal differentiation, which differ markedly clinically and histopathologically,9–12 and the EWSR1 gene is recognised as a partner in a diverse spectrum of neoplasms of different lineages.13 A large number of partner genes has been characterised for EWSR1 in Ewing sarcoma alone,14 but commercial probes are only available for the common fusions. The wealth of molecular advances in recent years has led to consistent genetic abnormalities being defined in a large array of soft tissue tumours; however, for economic reasons or lack of commercial availability, even the largest molecular diagnostics laboratories are unable to carry primers or probes for some rarer variant translocations or fusion transcripts. Since our last study, specific entities such as soft tissue myoepithelial tumours have been much more comprehensively characterised genetically. Up to about half of soft tissue myoepithelial neoplasms harbour EWSR1 gene rearrangements, with partner genes identified including POU5F1, PBX1, ZNF444 and ATF1;15–22 and FUS rearrangements are also described.23 ,24 As it is unfeasible for molecular diagnostics laboratories to carry the full range of primers or probes to assess for myoepithelial neoplasms, for putative myoepithelial tumours it would be more practical to use FISH to assess for EWSR1 and FUS gene rearrangements, whereby the presence of a rearrangement in the context of appropriate morphology, immunoprofile and clinical features (and the exclusion of other EWSR1-rearranged or FUS-rearranged neoplasms) would strongly support the diagnosis.

It is important to emphasise the need to interpret molecular findings in light of tumour morphology, immunoprofile and clinical features, and the now widespread use of ancillary molecular analysis has shown that molecular diagnosis does not always equate to the ‘gold standard’. While some genetic abnormalities remain specific for tumour types, such as the SS18-SSX fusions of synovial sarcoma (figure 2), and PAX3/7-FOXO1 of alveolar rhabdomyosarcoma, for many other soft tissue neoplasms, such as angiomatoid fibrous histiocytoma and clear cell sarcoma of soft parts, which have defining translocations that are shared with other tumour types, molecular diagnosis remains as a useful, highly supportive ancillary diagnostic method in the appropriate histopathological and clinical context.

In summary, the findings from this re-evaluation are that FISH and RT-PCR are useful complementary platforms in the pathological assessment of soft tissue neoplasms, with certain tests influenced by clinical need or pathological value. In contrast to the earlier study, RT-PCR and FISH were seen to show similar sensitivities. RT-PCR showed a higher specificity, and while it again showed a higher failure rate than FISH, this was markedly reduced in comparison to that seen five years ago. Due to the relative non-specificity of FISH, in many cases it may be worthwhile to perform RT-PCR first if suitable primers are available, and then FISH as a second-line test if RT-PCR is non-informative. The better overall results of this current study reflect the increasing technical expertise that has arisen with the widespread use of molecular and molecular cytogenetic methods as part of the histopathological workup of soft tissue neoplasms.

Take home messages

  • Fluorescence in situ hybridisation and reverse transcription-PCR (RT-PCR) remain useful complementary platforms in the pathological assessment of soft tissue tumours, with certain tests influenced by clinical need or pathological value.

  • Formalin-fixed paraffin-embedded tissue continues to represent a reliable and practical resource for molecular tests. In an experienced laboratory, the overall failure rate has been reduced to <5%, which represents a vast improvement compared with the previous study (RT-PCR failure rate reported previously at 25.2% of all cases).

  • There has been a marked improvement in RT-PCR success rates since the advent of ancillary molecular diagnostic testing. This is likely to be attributable to the increased experience of the staff of both histopathology and molecular laboratories, in the optimal processing, preparation and assessment of specimens for molecular analysis, as well as the recognition among all histopathology departments of the importance of molecular diagnostics as a crucial part of the patient pathway for the assessment of genetic aberrations for solid tumours with available targeted therapies.


The authors are grateful to the many laboratory colleagues who have performed the routine practical work and analysis.



  • Handling editor Cheok Soon Lee

  • Funding The authors acknowledge support from the NIHR Royal Marsden/ICR Biomedical Research Centre.

  • Contributors KV and KT wrote the manuscript. KV, DG, JS, CF, LT and KT analysed the data. DW and KT prepared images.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.