Definition of ROS1
The ROS1 (c-ros oncogene 1) gene is located at 6q22 on the long arm of chromosome 6. ROS1 has been described as an ‘orphan’ receptor tyrosine kinase and has no known ligand. The protein product of the gene is a member of the insulin receptor family and is related to ALK. ROS1 activation signals downstream in cells, activating the STAT3, PI3kinase, as well as the RAS/RAF/MAPK pathway by phosphorylation of RAS.
The physiological functions of ROS1 are not clear but in oncogenic terms, it seems likely that the downstream effects are similar to those exerted by oncogenic activation of ALK or EGFR.
ROS1 Abnormalities in NSCLC
The first lung tumour described with ROS1 rearrangement was, in fact, the HCC78 non-small cell lung cancer (NSCLC) cell line, but shortly after this, numerous reports in actual lung tumour samples emerged. During rearrangement, the break point in the ROS1 gene usually occurs in exons 32, 34 or 35. The ROS1 3’ fragment forming the fusion gene contains the tyrosine kinase domain and there are numerous partner genes described in lung cancer ROS1 fusions, including CD74, SLC34A2, TMP3, SDC4, EZR, LRIG3, FIG, KDELR2, and CCDC6. The fusion gene leads to constitutive activation of the tyrosine kinase, driving the oncogenic effect. Most of these partners involve rearrangements with other chromosomes, but occasional partners involve inversions within chromosome 6. This may be part of the reason why the fluorescence in situ hybridisation (FISH) findings in ROS1 rearranged tumours are highly variable, including widely separated 3’ and 5’ signals and, more often, several single signals. Findings are often heterogeneous within tumours. Oncogenicity of fusions has been demonstrated in animal models.
The epidemiology of ROS1 rearranged tumours is remarkably similar to that for ALK gene rearranged NSCLC. Prevalence has been reported at anything from 0.6% to over 3% in unselected populations but the data are somewhat confounded by variation in the cell type of the denominator; some reports refer to NSCLC without clarifying the proportion of the tested population that was adenocarcinoma, whilst in others, only adenocarcinomas were tested. It seems likely that prevalence will be somewhere between 1-2% of adenocarcinoma cases. Very occasional reports of ROS1 fusion in squamous cell carcinoma may be spurious or reflect unusual clinical circumstances (never smoking). ROS1 fusions are commoner in never smokers and tend to be found in younger patients. There may be a female predominance but there seems to be similar prevalence in East Asian and Caucasian populations.
As with ALK rearrangement and EGFR mutation, ROS1 fusion is an addictive oncogenic driver associated with adenocarcinogenesis in the lung peripheral epithelium (the terminal respiratory unit) unrelated to tobacco-driven carcinogenesis. The vast majority of reported cases are in adenocarcinomas. Histology of these cases has been quite variable. There does, however, seem to be an excess of solid or cribriform pattern tumours, signet ring cells are sometimes found, some studies report more poorly differentiated, pleomorphic tumours or mucinous tumours but there is no histological pattern that could be relied upon for preferential selection of patients for testing.
ROS1 as a Predictive Biomarker
So far, there are no established data to suggest there is any difference, clinically or in therapeutic response to crizotinib, between the different fusion proteins reported. Limited clinical data suggest that ROS1 rearranged adenocarcinoma is very responsive to crizotinib with response rates of over 70% reported. Crizotinib is now approved by the US Food and Drug Administration (FDA) for treatment of ROS1 rearranged NSCLC. Other agents which are active against ROS1-rearranged tumours, e.g. entrectinib, PF-06463922 (lorlatinib) are in early stages of development.
The core methodology generally used for ROS1 testing is the break apart FISH assay, although, for all the reasons already mentioned, testing is not yet widespread and mainstream, at least within Europe. Fusion gene mRNA gene product may be identified by a reverse transcriptase polymerase chain reaction (RT-PCR) assay and various approaches to massive parallel sequencing (Next Generation Sequencing – NGS) can identify fusion genes at the DNA level. The ROS1 fusion gene is associated with elevated expression of the protein product of the fusion gene which exerts oncogenic activity. This protein may be detected by immunohistochemistry (IHC).
FISH testing using dual colour break apart probes is probably the most accessible method for detecting ROS1 rearranged NSCLC. Until crizotinib is licensed by the European Medicines Agency, testing requirements for drug use in EU will remain unknown. As a standard approach, a FISH test generally requires a minimum read on at least 50 assessable cells, ideally by two different readers. As described above, the ROS1 fusion gene normally involves inter-chromosomal rearrangement, as opposed to intra-chromosomal inversion. As a result split signals are usually fairly clear; much easier to read in comparison to ALK break apart assays. Consequently, most laboratories would accept a minimum threshold of 10% abnormal cells for a ‘positive’ FISH test. Abnormal cells in the same tumour may show a range of abnormal signal patterns. There is also some variation in the coverage of fusions that may be detected by different commercially available probe sets and this should be borne in mind when an assay is being considered.
RT-PCR is a very specific technique, but it lacks somewhat in sensitivity and reliability. Rare fusion genes may be missed if the primer set for the multiplex PCR reaction does not cover the fusion gene in question, and quality mRNA may not be available from formalin fixed paraffin embedded (FFPE) tissue, the usual source of lung cancer diagnostic material. This technology is not widely available and requires special expertise.
There is an emerging literature on the use of immunohistochemistry for the detection of ROS1 protein in NSCLC samples. This has largely been pursued in the hope that a screening approach, like that used in many centres for ALK testing, could be taken as an efficient method of detecting this uncommon molecular abnormality. FISH testing is perceived as being relatively expensive and time consuming as a screening tool. Initial reports described a close correlation between IHC and FISH positivity. However, a mixture of subsequent studies and anecdotal experience has suggested that the close correlation between ALK IHC and FISH may not be present to the same degree for ROS1 IHC and FISH, using currently available reagents. The issue seems to be a higher prevalence of IHC positivity which is not confirmed by FISH, confounded by several aspects of artefactual staining and positivity in non-tumour cell populations such as macrophages. However, with a knowledge of these pitfalls, a low threshold for ordering confirmatory FISH testing would seem to be an adequate approach to testing.
Detection of fusion genes using NGS approaches is feasible and although not widely practiced in the routine clinical setting, use is certainly increasing. Undoubtedly, as technology and experience improve, this approach will be widely adopted. ROS1 is usually amongst the fusions genes covered in commercially available panels of tests used on NGS platforms. Whether this test will be adopted as the primary diagnostic or as a screening tool, requiring confirmation by FISH, RT-PCR or IHC, remains to be seen.
ROS1 Testing in NSCLC
Most existing guidelines do not make any strong recommendation about testing for ROS1 rearrangements. This is because crizotinib, the drug which has, to date, mostly been used in ROS1 rearranged cases, has not received widespread approval and reimbursement for use. If there is access to the drug, perhaps in a clinical trial or through some special access scheme, testing may be considered. If this is the case, then clinical parameters – smoking habit, gender, age, ethnicity etc should, in general, not be used to select patients for testing. Excluding smokers with adenocarcinoma would potentially risk missing patients with a ROS1 fusion.
Currently, crizotinib is used in second or greater line. This, and the rarity of the fusion, plus the issues relating to drug access mean that routine, reflex testing for ROS1 fusion genes is unusual. With drug approval and increased access, this could change. A common testing scenario currently practiced, is to consider ROS1 testing only after an adenocarcinomas has been shown to be negative for the more common driver molecular alterations, such as EGFR or KRAS mutation, perhaps BRAF mutation, and ALK gene rearrangement. Screening such a population for ROS1 rearrangements has been reported to increase prevalence to 7-12% or more. This screening may be limited to such triple or quadruple –negative tumours in selected patients who are never or long-time ex-smokers.
The introduction of next generation sequencing approaches which will detect ROS1 fusion genes will change this testing approach. In this scenario, ROS1 fusion will be part of a multiplex panel of mutations and fusion genes covered by the NGS screen. ROS1 fusion data will be provided as part of the NGS data output, whether or not it is specifically requested by an oncologist.
Samples for Testing
Any pathological samples containing tumours cells may be used for ROS1 testing: biopsy samples, cytology material, surgically resected tumour. Whilst samples for FISH require a minimum number of assessable tumour cells, and both RT-PCR and NGS techniques depend to mRNA or DNA quality and quantity, ROS1 IHC may be read on a very small number of cells. However, data on IHC testing are limited and some of the known false positive staining reactions may render a positive call on a very small number of tumour cells unwise. Depending on the IHC technology used, pathologists must be aware of the pitfalls of testing, as they must when reading FISH tests, in order to avoid false positive or false negative results. With the current state of knowledge, it is necessary to confirm any positive IHC test with a FISH test or other reliable technology, if available.
Ensuring Timely and Quality ROS1 Testing
There are many steps and many individuals involved in the work chain from the decision to test, through sample handling, submission to the molecular lab, assimilation of the molecular and pathological data and reporting back to the treating physician. IHC can be carried out and read in a number of hours – a same day service may be feasible, a next day service quite routine. FISH testing usually takes longer. Currently NGS test results can take many days to be verified. Efficient communication and working systems are vital for timely diagnosis. Currently, these may not be pertinent issues whilst crizotinib has limited availability and approval. When ROS1 testing becomes routine, all the usual factors will become more important.
All labs issuing clinical reports should participate in, and perform adequately in external quality assurance programmes. Regular in-house monitoring of performance, and awareness of potential false positive and false negative tests is important.
Reporting ROS1 Test Results
ROS1 test results should be reported in the context of the sample which was diagnosed with NSCLC and assimilated into the overall pathology report. Context is vital, to avoid misinterpretation of results. Expert pathologist input is required for FISH test assessment, since identification of tumour cells is challenging.
The report to the treating physician should provide detail of the ROS1 test methodology used and, in the case of the FISH test, the number of cells assessed and the % showing abnormalities. Generic comments about the implications of any findings for therapy are acceptable, but specific treatment recommendations should be avoided.