Mini-symposium: molecular testing of solid tumours
Molecular testing of gastrointestinal tumours

https://doi.org/10.1016/j.mpdhp.2017.08.004Get rights and content

Abstract

Cancers of the gastrointestinal tract are a major cause of morbidity and mortality globally, and not surprisingly, there has been intense interest in identifying prognostic and predictive markers for new targeted therapies. The cancers of the gastrointestinal tract are unique in exemplifying in microcosm the axioms of molecular pathology: that amongst the various organs of the gastrointestinal tract, different molecular markers have different implications; that of the innumerable molecular alterations identified in these cancers, only an extremely small minority have proven clinical relevance; and perhaps most importantly, that molecular pathology practitioners must be prudent in maximising the use of small tissue samples to garner clinically important information. Here, we review the clinically-validated molecular targets in carcinomas of the oesophagus, stomach, colorectum and anal canal, and in gastrointestinal stromal tumours, and discuss a handful of markers which are likely to be of use in the future.

Introduction

There has been a recent shift in paradigms within histopathology: no longer is the classification of a tumour dependent purely on its appearance and expression of particular antigens. Rather, there is an increasing – and increasingly rapid – shift towards using molecular criteria for determination of a given tumour's prognosis and most appropriate management. This is exemplified in its most developed form by the recent WHO classification of tumours of the central nervous system, which for the first time, makes particular molecular features mandatory for rendering particular histopathological diagnoses.

In contrast, while cancers of the gastrointestinal tract are still diagnosed on the basis of their morphology, a few key molecular markers have been validated in classifying individual tumour types into molecular subgroups on the basis of their behaviour and responses to therapies. The fact that cancers of the gastrointestinal tract are so common renders these few key markers absolutely central in the management of human cancers.

Before any discussion of the specific details of the molecular pathology of any particular tumour, however, it is germane to review a handful of the challenges and caveats which are inherent to the practice of molecular pathology.

Above all else, it must always be borne in mind that in real-life practice, the vast majority of specimens received are small, must be reported within very short periods of time to benefit patients, and must be tested in the context of the robust quality controls to ensure good practice. This requires that molecular pathology practitioners are judicious in their use of tissue not only for molecular but also for immunohistochemical testing, and that well-organised logistics are in place.

Secondly, tumours are by definition genetically unstable, and contain enormous numbers of molecular alterations. In routine practice, however, only a limited number have been validated as having clinical significance. The difficulty, then, is to avoid being overwhelmed by the detail of the many thousands of recognised alterations, at the expense of the few which are known to be clinically valuable. In this vein, although the examination of large panels of genes or even sequencing of whole genomes certainly has a vital role in research and in recruitment into clinical trials, the assessment of extremely large banks of molecular alterations is currently of rather limited use in routine practice.

Thirdly, while molecular pathology certainly involves the interrogation of DNA sequences themselves, this is by no means the limit of its scope: abnormalities of chromosomal structure, of epigenetics and of protein expression have all been validated as predictive and prognostic markers in various tumours. It is incumbent upon molecular pathology services, therefore, to make use of multiple platforms for the full molecular profiling of tumours.

Furthermore, as the range of validated molecular alterations of all types in a given tumour increases – and especially given that many of them have complex interrelationships – their interpretation becomes exponentially more difficult, and the answer to the clinical question, ‘what is best for this patient?’, rapidly becomes unclear. This is made yet more complex by the fact that it is not molecular alterations alone which guide clinical interpretation: molecular alterations must be interpreted in the context of the tumour type and of the clinical scenario. The BRAF mutation (discussed in more detail later) perhaps exemplifies this best: the same mutation is predictive of response to targeted therapy in melanoma, but not in other tumour types (at least as a monotherapy, in colorectal cancers); it is, in contrast, a validated prognostic marker in microsatellite-stable (MSS) colorectal cancers and can be used as a diagnostic marker in equivocal thyroid lesions. This has an important bearing on the rationale for wider sequencing; algorithms for interpretation of any findings must be tumour-specific.

At this point in the development of molecular pathology, therefore, there is a pressing need to perform assessment of molecular alterations in an algorithmic fashion, taking account of all molecular, morphological and immunophenotypic characteristics of a tumour, and the clinical circumstances of the patient. Even with only a few parameters, this form of algorithmic interpretation is quickly beyond the ability of an unaided human, and so the essential role of computation in cancer management is obvious.

In this review, we describe the clinically relevant molecular alterations of tumours of the gastrointestinal tract. By necessity – because different alterations are of different significance in different tissues – each organ of the gastrointestinal tract is examined in turn.

Section snippets

Oesophageal carcinomas

At present, in routine practice, there are no molecular alterations of proven clinical use in either oesophageal squamous cell carcinomas or adenocarcinomas.1 Nonetheless, in both, there is early evidence that certain molecular markers may be of use in clinical management in the future.

For example, though controversial at present, there is evidence that human papillomavirus (HPV) is implicated in a subset of oesophageal squamous cell carcinomas, and this may – in future – be of interest in the

HER2 amplification

Trastuzumab has been licensed for a number of years for use in advanced gastro-oesophageal adenocarcinomas in combination with chemotherapy, following the promising the results of the ToGA trial,2 in which HER2 amplification was found in 22.1% of almost four thousand tumours. Prescription of this drug requires assessment for HER2 amplification; the method employed in the ToGA trial was the classical two-step process with initial immunohistochemistry and confirmatory FISH for equivocal cases

Colorectal adenocarcinoma

In the gastrointestinal tract, it is carcinomas of the colorectum which have provided the greatest number of clinically-relevant molecular alterations both for prognosis and for prediction of response to therapy. From a practical perspective, newly-diagnosed colorectal cancer – generally diagnosed by colonoscopic biopsy of a tumour – are increasingly assessed for DNA mismatch repair (MMR) status, KRAS and NRAS mutations, and BRAF mutations.

Anal squamous cell carcinoma

Squamous cell carcinomas of the anal canal are relatively rare and, at present, there is no molecular alteration in routine use in clinical practice.19 Nonetheless, there is early evidence that markers may come into clinical use in the future.

Given that most of these tumours are driven by HPV infection, there is good reason to believe that such cancers may in the future be good candidates for immunotherapy. Results of two phase II trials have recently been published, demonstrating good

Gastrointestinal stromal tumour (GIST)

It is now well-established in many centres that GISTs should be routinely tested for mutations in the KIT and PDGFRA genes, mainly for their significance in predicting response to tyrosine kinase inhibitor therapy (e.g. imatinib, sunitinib).21 Increasingly often, such testing is performed using small biopsy and FNA specimens. Overall, approximately 85–90% of GISTs harbour a mutation in either of these genes, with 70–80% of these being in KIT, and 10–20% in PDGFRA. By far the most common

Discussion

The gastrointestinal tract illustrates in microcosm the principle that personalised medicine requires a panel of markers related to multiple targets. We feel that this review is exhaustive in terms of those markers which are of use in routine clinical practice. The main challenges are more logistic than technical, and particularly demand the integration of morphological and molecular features of tumours. This requires, if not complete embedding of molecular pathology within histopathology, at

Conclusion

Molecular profiling of carcinomas and sarcomas of the gastrointestinal tract is technically and logistically achievable in routine practice using real-life samples, assuming that the correct expertise and resources are available. It remains uncertain in most of the public health systems when and how this service delivery should be located and how it should be funded. The journey from a putative predictive or prognostic alteration to a fully-fledged clinically-significant molecular marker is by

References (25)

  • L. Zhu et al.

    Microsatellite instability and survival in gastric cancer: a systematic review and meta-analysis

    Mol Clin Oncol

    (2015)
  • C. Ma et al.

    Programmed death-ligand 1 expression is common in gastric cancer associated with Epstein-Barr virus or microsatellite instability

    Am J Surg Pathol

    (2016)
  • Cited by (1)

    View full text