Tumor Heterogeneity and Testing Discrepancy Confound ROS1 Detection in NSCLC

Abstract

For advanced NSCLC, the correct and timely identification of actionable mutations has become paramount in therapy decision making. In about 1% of NSCLC, malignant growth is driven by the rearrangement of ROS1, a tyrosine kinase receptor.1Herbst R.S. Morgensztern D. Boshoff C. The biology and management of non-small cell lung cancer.Nature. 2018; 553: 446Crossref PubMed Scopus (1933) Google Scholar ROS1 rearrangements are traditionally detected by using fluorescence in situ hybridization (FISH), though next-generation sequencing (NGS) is increasingly being used, as it can detect an entire panel of genetic alterations simultaneously.2Lin J.J. Shaw A.T. Recent advances in targeting ROS1 in lung cancer.J Thorac Oncol. 2017; 12: 1611-1625Abstract Full Text Full Text PDF PubMed Scopus (160) Google Scholar The accurate detection of ROS1 fusions directly affects the choice of therapy, as ROS1 inhibitors are preferred over chemotherapy if a ROS1 fusion is present.3Dagogo-Jack I. Shaw A.T. Expanding the roster of ROS1 inhibitors.J Clin Oncol. 2017; 35: 2595-2597Crossref PubMed Scopus (15) Google Scholar With more than one detection methodology now available, there will be an increasing number of cases in which they produce conflicting test results, posing a diagnostic and therefore therapeutic challenge. Additionally, recent studies with multiregion sequencing have shown significant tumor heterogeneity with subclone-specific mutations, although none of these studies investigated the clonal nature of ALK receptor tyrosine kinase gene (ALK) or ROS1.4Zhang J. Fujimoto J. Zhang J. et al.Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing.Science. 2019; 346: 256-259Google Scholar, 5de Bruin E.C. McGranahan N. Mitter R. et al.Spatial and temporal diversity in genomic instability processes defines lung cancer evolution.Science. 2019; 346: 251-256Google Scholar, 6Jamal-Hanjani M. Wilson G.A. McGranahan N. et al.Tracking the evolution of non-small-cell lung cancer.N Engl J Med. 2019; 376: 2109-2121Google Scholar Here we report two rare cases of patients with positive FISH and negative NGS results but with opposing responses when treated with a ROS1 inhibitor. The second patient later showed positive NGS results in other regions of the tumor, showing evidence of ROS1 tumor heterogeneity as the cause of the discrepant testing results. A 55-year-old female nonsmoker presented with stage IV adenocarcinoma of the lung with liver and brain metastases. On mediastinal lymph node biopsy, testing for ROS1 was positive by FISH in 40% of nuclei (Abbott probes [Abbott Molecular]) (Fig. 1A). Our institutional hybrid capture–based targeted NGS panel for actionable mutations7Yang S.R. Lin C.Y. Stehr H. et al.Comprehensive genomic profiling of malignant effusions in patients with metastatic lung adenocarcinoma.J Mol Diagn. 2018; 20: 184-194Abstract Full Text Full Text PDF PubMed Scopus (28) Google Scholar identified a hitherto unpublished ROS1–olfactomedin 3 gene (OLFM3) fusion of unknown significance, as it did not contain the ROS1 kinase domain. Given the FISH positivity, which was known before the NGS results were available, ceritinib therapy was started; however, 3-week surveillance imaging revealed significant disease progression. Treatment with carboplatin and pemetrexed was then started. A month later, because of worsening respiratory symptoms, bronchoscopy was done for tumor debulking and stent placement. Interestingly, the pathologic examination this time revealed invasive, well-differentiated squamous cell carcinoma instead of adenocarcinoma, with the exact same mutation profile as that of the adenocarcinoma biopsy specimen (variants in phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha gene [PIK3CA], tumor protein p53 gene [TP53], cyclin dependent kinase inhibitor 2A gene [CDKN2A], cyclin dependent kinase inhibitor IB gene [CDKN1B], neurofibromin 2 gene [NF2], and nuclear factor erythroid 2, like 2 gene [NFE2L2], as well as high programmed death ligand 1 expression). ROS1 was still identified on FISH with 8.5% of nuclei (Fig. 1B), which is below the usual 15% cutoff, and ROS1-OLFM3 was again detected on NGS. Because of likely squamous cell transformation, treatment with carboplatin, paclitaxel, and pembrolizumab was started; however, the patient continued to have disease progression and died 3 months later. A 52-year-old man with a minimal smoking history presented with stage IV adenocarcinoma of the lung metastatic to the brain, bone, and muscle in 2015. He initially received cisplatin, pemetrexed, and stereotactic radiation to four brain lesions, resulting in with many months of disease control. Upon disease progression, after FISH testing returned a positive result for ROS1 (87% of nuclei) (Cytocell probes [Cytocell Ltd.]), the patient began receiving crizotinib. However, 2 weeks later, use of our institutional NGS panel on the 1-cm3 tumor specimen (80% tumor) did not identify any ROS1 fusions but did find 15% KRAS and U2 small nuclear RNA auxiliary factor 1 gene (U2AF1) mutations, absent any TP53 mutations. The patient continued receiving crizotinib given the symptomatic improvement. A week after this, FoundationOne8Thomas R.K. Baker A.C. DeBiasi R.M. et al.High-throughput oncogene mutation profiling in human cancer.Nat Genet. 2007; 39: 347Crossref PubMed Scopus (874) Google Scholar testing (performed by Foundation Medicine) on a different region of the same specimen identified the common CD74-ROS1 fusion, U2AF1 mutation, and TP53 mutation. Given the conflicting NGS results, we performed additional testing, again using our institutional NGS panel but on a different section of the specimen, and confirmed the CD74-ROS1 fusion along with U2AF1, 2% KRAS, and 22% TP53 mutations (Fig. 2). The patient achieved a partial response after 1 month of treatment with crizotinib. However, 3 months later he had already experienced development of progressive disease with new intracranial metastases and is currently taking another tyrosine kinase inhibitor. In the rare cases in which FISH and NGS testing produce opposing results regarding presence or significance of a ROS1 fusion, a difficult decision must be made regarding whether to initiate ROS1 inhibitor therapy over chemotherapy. Both patients had a positive FISH result for ROS1 fusion but NGS results that were not confirmatory, either showing a fusion of unknown significance or yielding an initially negative results. The lack of response to an inhibitor in patient 1 is likely due to a nonfunctioning fusion, as the protein did not contain the kinase domain, which is retained in all previously published ROS1 fusion genes (Fig. 1C).9Davies K.D. Doebele R.C. Molecular pathways - ROS1 fusion proteins in cancer.Clin Cancer Res. 2013; 19: 4040-4045Crossref PubMed Scopus (248) Google Scholar Although the reciprocal fusion involving the kinase domain appears to be detected by FISH (see Fig. 1A and B), we were not able to detect this fusion by NGS, perhaps because of the breakpoint being in intron 31, a region that is more difficult to capture and sequence.10Davies K.D. Le A.T. Sheren J. et al.Comparison of molecular testing modalities for detection of ROS1 rearrangements in a cohort of positive patient samples.J Thorac Oncol. 2018; 13: 147-1482Google Scholar In any event, this reciprocal fusion is unlikely to be functional given that the partner is still in an intergenic region. For patient 2, three different NGS tests were ultimately performed on different regions of the 1-cm3 tumor specimen. The results showed at least two subclonal populations based on mutation profile: high-KRAS and low-TP53 versus low-KRAS and high-TP53 populations. Notably, the former subclone lacked the ROS1 fusion (see Fig. 2). The patient did have a partial response to crizotinib at 1 month but experienced development of progressive disease by as early as 3 months (compared with the shortest progression-free survival in the PROFILE 1001 trial, which was 14.4 months with a median of 19.2 months).11Shaw A.T. Ou S.H.I. Bang Y.J. et al.Crizotinib in ROS1-rearranged non–small-cell lung cancer.N Engl J Med. 2014; 371: 1963-1971Crossref PubMed Scopus (1409) Google Scholar Very little is known about the clonal nature of ROS1, as prior multiregion sequencing studies have not included patients with this mutation.4Zhang J. Fujimoto J. Zhang J. et al.Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing.Science. 2019; 346: 256-259Google Scholar, 5de Bruin E.C. McGranahan N. Mitter R. et al.Spatial and temporal diversity in genomic instability processes defines lung cancer evolution.Science. 2019; 346: 251-256Google Scholar, 6Jamal-Hanjani M. Wilson G.A. McGranahan N. et al.Tracking the evolution of non-small-cell lung cancer.N Engl J Med. 2019; 376: 2109-2121Google Scholar Our case raises the possibility that ROS1 may be present as a subclonal mutation instead of a truncal mutation, thereby limiting the efficacy of a ROS1 inhibitor. This case also cautions against excluding patients with positive FISH but negative NGS results in future ROS1 clinical trials. Given our aforementioned experience with conflicting ROS1 diagnostic results, a trial of ROS1 inhibitor may be prudent if at least one of the tests returns positive results, followed by a short interval surveillance.