ROS1 is a receptor tyrosine kinase that has recently been shown to undergo gene rearrangements in~1%–2% of non-small cell lung carcinoma (NSCLC) and in a variety of other tumours including cholangiocarcinoma, gastric carcinoma, colorectal carcinoma and in spitzoid neoplasms, glioblastoma and inflammatory myofibroblastic tumours. The ROS1 gene fusion undergoes constitutive activation, regulates cellular proliferation and is implicated in carcinogenesis. ROS1 fusions can be detected by fluorescence in situ hybridisation, real-time PCR, sequencing-based techniques and immunohistochemistry-based methods in clinical laboratories. The small molecule tyrosine kinase inhibitor, crizotinib has been shown to be an effective inhibitor of ROS1 and has received Food and Drug Administration approval for treatment of advanced NSCLC. The current review is an update on the clinical findings and detection methods of ROS1 in clinical laboratories in NSCLC and other tumours.
- lung cancer
- cancer genetics
Statistics from Altmetric.com
Human c-ros oncogene 1 (ROS1) was first discovered as proto-oncogene and transforming sequence for UR2 avian sarcoma virus.1 2 ROS1 is a receptor tyrosine kinase (RTK) of the insulin receptor family and encodes a transmembrane protein with intracellular C-terminal tyrosine kinase domain and shares marked sequence homology and structural similarities to anaplastic lymphoma kinase (ALK) oncogene.3 The first naturally occurring ROS1 fusion transcript was described by Charest et al; where interstitial deletion of chromosome 6 resulted in fusion of 3` sequences of ROS1 to 5` region of fused in glioblastoma (FIG) and gave rise to the fusion product FIG-ROS1 transcript in the human glioblastoma cell line U118MG.4 5 Subsequently, multiple ROS1 fusion partners have been since described in non-small cell lung carcinoma (NSCLC) and other epithelial and non-epithelial malignancies.
The widespread and renewed interest in ROS1 stems from the rapid clinical response demonstrated in advanced NSCLC to small molecule tyrosine kinase inhibitor (TKI), crizotinib.6
ROS1 is located in chromosome 6q22 and encodes a 2347 amino acid long protein.7 The gene product of ROS1 is comprised of an extracellular N-terminal domain, a hydrophobic stretch of transmembrane spanning region, an intracellular tyrosine kinase domain and a carboxy terminal tail (figure 1A)8 The transmembrane RTK shares significant structural similarities to the ALK gene product. ROS1 is a unique RTK, partly due to the relatively large size of its extracellular domain (~1800 amino acids) as well as due to the presence of multiple extracellular domain repeat motifs with homologous amino acid sequences to fibronectin type III-like repeats (FN-III-like-repeats), almost resembling the structure of cell adhesion molecule.8 Not surprisingly, there is high density of evolutionary conserved domains that stretch along its promoter sequences and coding exons.8
ROS1 protein expression is highest in the kidneys, followed by the cerebellum, peripheral neural tissue, stomach, small intestine and colon with lower expression in several other tissues.9 Interestingly, full-length ROS1 protein expression is reportedly absent in normal human lung tissue when evaluated by immunohistochemistry (IHC).9 However, consistent c-ROS expression in adult human lung by northern blot and expression array databases was described by Acquaviva et al.8 Studies of ROS1 induced signalling pathway and elucidation of its biological role have been hampered largely due to lack of understanding of as yet unknown ligand for c-ROS and in part due to the inability to overexpress full-length wild type c-ROS in cellular models.8 It has been hypothesised that c-ROS1 expression plays a developmental role in epithelial-mesenchymal interactions and in cellular differentiation cascade of epithelial tissues.8
ROS1 signalling pathway
Traditionally, in RTK fusions (eg, ALK), the tyrosine kinase activation is induced by dimerisation domains that are typically recruited from the fusion gene partners.10 The seminal work in FIG-ROS1 fusion protein showed co-localisation to Golgi apparatus as the key step for inherent cell transforming ability, but, intriguingly, not the tyrosine kinase activity of the fusion protein.11 Much of the current understanding of ROS1 signalling is derived from FIG-ROS1, CD74-ROS1 or SDC4-ROS1 fusion product expression studies in fibroblasts or Ba/F3 cell lines. Autophosphorylation of ROS1 and phosphorylation of the SH-2 domain containing tyrosine phosphatase appears to be a key and possibly the rate-limiting step for downstream signal transduction.12 With acquisition of gene fusions, there is constitutive activation of RTK which subsequently impacts cell differentiation, proliferation, growth and survival.8 The downstream signalling cascade is achieved by navigating through phosphoinositide 3 kinase/mammalian target of rapamycin pathway, phosphorylation/activation of signal transducers and activators of transcription 3 (STAT3) and vav 3 guanine nucleotide exchange factor 1 eventually leading to regulation of anchorage-independent cellular growth and cell morphology.8 12–16 Interestingly, in a recent report, Jun et al demonstrated that the activation of downstream signalling pathways differ based on fusion partners of ROS1, as phosphorylation/activation of E-Syt1 (extended synaptotagmin 1) led to an invasive phenotype in CD74-ROS1 transduced cells, but not in FIG-ROS1 transduced cells.15 Figure 2 summarises14 the key downstream signalling pathways known to be involved in downstream cell proliferation, survival, migration and metastasis, transformation and invasiveness.8 14
ROS1 gene fusions in NSCLC
So far, over 20 ROS1 fusions have been described in solid tumours (Figures 1B, 3). Rikova et al reported the first ROS1 translocation in human NSCLC (CD74-ROS1 and solute carrier family 34 (sodium phosphate) A2 (SLC34A2)-ROS1) by global survey of phosphotyrosine kinase signalling in NSCLC cell lines and primary lung tumours.17 Subsequently, in a large-scale NSCLC screening study using high-density molecular data, Bergethon et al first described the prevalence of ROS1 fusion tumours and demonstrated excellent clinical response to crizotinib.18 Since then, a multitude of independent studies have corroborated and contributed to further data on the role of ROS1 in NSCLC. Figure 3 depicts the common gene fusions seen in NSCLC. CD74-ROS1 is by far the most commonly reported fusion (figure 3), with an estimated frequency of 32% in NSCLC. The other common fusion partners include SLC34A2-ROS1 (~17%), tropomyosin 3 (TPM3-ROS1;~15%), syndecan 4 (SDC4-ROS1;~11%), ezrin (EZR-ROS1;~6%) and FIG-ROS1 (~3%). Less common (~1% or less) fusions include leucine-rich repeats and immunoglobulin-like domains 3 (LRIG3-ROS1), endoplasmic reticulum protein retention receptor 2, coiled-coil domain containing 6 (CCDC6-ROS1) and others. Details of ROS1 fusions in NSCLC have been described elsewhere.14 16 19 In this review, we will discuss some of the salient features of ROS1 fusion tumours in NSCLC and in other tumours.
Epidemiology of ROS1 fusion in NSCLC
ROS1 fusion tumours are enriched in female patients without a smoking history and this observation has been confirmed by multiple independent studies and by a recent meta-analysis.18 20 Although seemingly rare, however, ROS1 fusion tumour is an exceedingly attractive target in lung cancers, due to the availability of targetable therapy with TKIs. Large-scale population data and distribution in various ethnicities is still forthcoming; preliminary observations show similar prevalences across Asia (2.1%), Europe (1.6%) and North America (1.8%) (table 1). Although the prevalence appears lower in Australia (0.9%), it should be noted that the observation is from a single study and thus is limited by sample size constraints and sampling bias. There are no reported cases or case series from African or South American continents to our knowledge. Overall, based on data from published literature, the worldwide prevalence of NSCLC ROS1 fusion tumour is 1.9% and in accordance to the estimated prevalence.
Histopathological features of ROS1 fusion in NSCLC
Among NSCLC, ROS1 fusions have thus far been described in adenocarcinoma (ADC). No fusion events were noted in NSCLC with squamous histology.18 21 22 Likewise, ROS1 fusions are not described in neuroendocrine carcinomas including small cell carcinoma or large cell neuroendocrine carcinoma of the lung. Lung ADC with cribriform/acinar architecture, mucinous features, solid pattern histology, hepatoid morphology, signet ring cell morphology and tumours with psammomatous calcification are the most common histological patterns that have been described and associated in ROS1 fusion lung ADCs.23 24
It is widely assumed that ROS1 fusion is mutually exclusive with other known driver oncogene mutations and fusions. Interestingly however, a number of studies now have reported concurrent driver mutations in ROS1 fusion tumours.9 25 26 Rimkunas et al reported co-occurrence of epidermal growth factor receptor (EGFR) mutations in two cases where there was demonstrable ROS1 expression by IHC.9 Scheffler et al reported 19 fluorescence in situ hybridisation (FISH)-confirmed ROS1 fusion tumours, of which, four tumours showed coexisting mutations in EGFR (P848L mutation), MET (R988C) and BRAF (G469S and c1742-cG>T).25 The EUROS1 trial reported one patient with ROS1 fusion and concomitant KRAS mutation.27 Similarly, Shaw et al reported a patient with ROS1 translocation and MET amplification in the PROFILE 1001 trial.28 The biological role of these coexisting driver mutations in the presence of putative oncogenic gene fusion is as yet unclear, and this phenomenon appears to be a rare event. As more and more studies implement next-generation sequencing (NGS)-based approaches in research design, more detailed information regarding the prevalence and possible biological role will emerge.
ROS1 and other tumours
ROS1 and cholangiocarcinoma
Cholangiocarcinoma (CCA) is a highly aggressive tumour, with dismal prognosis. Most patients present at an advanced-stage with limited treatment options. To date, there are no therapeutic agents available to target the most common genetic alterations (TP53 and KRAS) in CCA. Surgical resection remains the most important treatment but the survival rates even for the localised disease are not impressive. Gu et al studied protein tyrosine kinase signalling events in 23 patients with CCA by using a phosphoproteomic technique and demonstrated aberrant ROS1 expression in 2 (8.7%) patients.13 FIG-ROS fusion was identified in the two ROS1-positive CCAs by real-time (RT)-PCR. In a mouse model of intrahepatic CCA, FIG-ROS1 fusion was recognised as an oncogene which, in the presence of mutant p53 and KRAS, significantly promotes tumorigenesis.29 A recent study described ROS1 protein expression by IHC in 72 of 194 (37%) intrahepatic CCAs; however, none of the cases demonstrated ROS1 rearrangement by FISH.30 Interestingly, it was observed that ROS1 protein expression was associated with well differentiated tumours, papillary or mucinous histology, tumours with periductal or intraductal growth and better survival.30 Lim et al studied 261 CCA cases and identified ROS1 gene rearrangement by FISH analysis in 3 cases (1.1%) of intrahepatic CCA, whereas ROS1 protein expression by IHC was reported in 38 cases (19%).31 Further data are needed to study the role of ROS1 inhibition in ROS1 rearranged CCA.
ROS1 and spitzoid neoplasms
Spitzoid neoplasms are uncommon melanocytic tumours and include benignSpitz nevi, malignant spitzoid melanomas and atypical spitzoid tumours (tumours with overlapping histomorphological features of unknown malignant potential). A recent genomic analysis highlights the role of kinase fusions in pathogenesis of spitzoid neoplasms.32 It was found that 72 of 140 (51.4%) spitzoid tumours harboured mutually exclusive RTK fusions (ROS1 (17%), NTRK1 (16%), ALK (10%), RET (3%), BRAF (5%)). ROS1 rearrangement was observed in 26% of Spitz nevi, 9% of spitzoid melanoma and 8% of atypical spitzoid tumours, with nine different fusion partners (HLA-A, MYO5A, PPFIBP1, ERC1, PWWP2A, CLIP1, TPM3, ZCCHC8, KIAA1598).32
ROS1 and breast cancer
ROS 1 expression has also been studied in breast cancers. Eom et al recently examined the expression of ROS1 protein and gene in invasive ductal carcinoma (IDC) of the breast.33 Tumour samples from 203 patients were included to investigate the expression of ROS1 protein by IHC and ROS1 gene by RT-PCR. In this study the association between ROS1 expression and well known prognostic factors of IDC was observed. It was found that in IDC low ROS1 expression correlates with higher histological grade, higher mitoses, low oestrogen receptor expression and higher Mib1 proliferative index. It has been suggested that higher expression of ROS1 in IDC may be associated with better prognosis.
ROS1 and thyroid cancer
Thyroid cancer is one of the most common endocrine malignancies and the majority including papillary thyroid carcinoma are well differentiated and tend to have good prognosis.34 Recently, ROS1 fusion was documented in a 24-year-old patient who presented with a locally aggressive papillary thyroid carcinoma, solid variant (pT4aN1b). CCDC30 was identified as a fusion partner for ROS1 by sequencing.35 This is the first reported case of thyroid carcinoma exhibiting CCDC30-ROS1 gene fusion. This discovery may provide additional insight in the biological behaviour of these tumours and may have potential therapeutic implications.
ROS1 and gastric cancer
Gastric cancers are aggressive tumours with dismal prognosis especially in advanced-stage disease. Histologically, these tumours are classified into two main subtypes; intestinal-type and diffuse-type.36 The role of humanised monoclonal antibody (Herceptin) against HER2 in combination with chemotherapy in providing prolonged overall survival in patients with HER2-positive advanced gastric adenocarcinomas (GAs) has been well established.37 Recently, ROS1 rearrangement has been documented in GAs by Lee et al.38 Of 495 GA tumour samples included in the study 23 (4%) were positive for ROS1 by IHC. It was observed that ROS1 IHC-positive GAs were more likely to be of well to moderately differentiated intestinal-type and had decreased propensity of lymph node involvement and lymphatic invasion. Of these 23 cases, 3 (0.6%) were positive for ROS1 rearrangement by FISH, 2 of which showed SLC34A2-ROS1 fusion by RT-PCR. Both patients with SLC34A2-ROS1 fusion demonstrated poorly differentiated adenocarcinomas of diffuse-type involving the gastric antrum, developed recurrence and died within 2 years of surgery.38 GA and ROS1 rearrangement and respective TKI therapy should be studied more in detail as potential avenues of therapy.
ROS1 and colorectal adenocarcinoma
ROS1 rearrangement has been demonstrated in metastatic colorectal adenocarcinoma (CRC). Aisner et al reported 2 of 268 cases of metastatic CRC that were positive for ROS1 rearrangement by FISH.39 In one case the ascending colon was the primary site of the tumour with liver and lymph node metastasis whereas in another case the primary tumour site was rectosigmoid with metastasis to the lungs. One ROS1-positive case was shown to have an SLC34A2 fusion partner while the fusion partner for the other case was unknown.39 Larger-scale studies are needed to further evaluate the role of these genetic aberrations that may have the potential for significant impact in patient outcomes.
ROS1 and glioblastoma multiforme
As mentioned earlier, the initial reports of ROS1 gene rearrangement, specifically FIG-ROS1 translocation, was described in glioblastoma multiforme.5
ROS1 and inflammatory myofibroblastic tumour
Inflammatory myofibroblastic tumour (IMT) is a mesenchymal tumour of intermediate biological behaviour due to its low risk of metastasis and tendency of local recurrence. These tumours are typically seen in the mesentery, omentum, retroperitoneum and lungs but can affect various anatomical sites. Histologically, a tumour is composed of myofibroblastic spindled cells arranged in fascicles embedded in myxoid stroma and associated with dense lymphoplasmacytic inflammatory infiltrate. Approximately 50% of IMTs are positive for ALK rearrangement and have shown good clinical responses to crizotinib in patients with ALK-positive IMT.40–42 In a recently reported study, TFG-ROS1 fusion was identified by molecular profiling in a young patient with ALK-negative, treatment refractory IMT. The patient was subsequently treated with crizotinib and showed excellent clinical response.43 The TFG-ROS1 translocation in IMTs was later found in gastric IMTs by Fujita et al, and confirmed in two independent studies in larger sample size by Hornick et al and Antonescu et al, respectively.44–46
ROS1 in vascular and other tumours
Recently, ROS1 rearrangements were also identified in soft tissue tumours; angiosarcoma (AS) and epithelioid haemangioendothelioma (EHE). Of 33 AS and 20 EHE cases investigated, ROS1 rearrangement was documented in 1 of 33 AS (0.3%) and 1 of 20 (5%) EHE. Interestingly, a novel CEP85L-ROS1 rearrangement was identified in AS. These findings suggest a possible role of ROS1 in the pathogenesis of AS and endothelial-derived tumour such as EHE.47
Atypical meningioma has been reported to harbour oncogenic ROS1 fusion recently.48 In an ongoing trial (AcSe trial), two patients with atypical meningioma and IMT were treated with crizotinib and have shown partial response to crizotinib therapy.49 The reported ROS1 fusions in tumours other than NSCLC are shown in table 2.
Detection of ROS1 in clinical laboratories
There are four main approaches to detect ROS1 translocation: (1) by FISH; (2) RT-PCR (3) NGS approaches and (4) IHC.
ROS1 detection by FISH
Thus far, detection by FISH, using dual-colour break-apart hybridisation probes is considered the gold standard for ROS1 fusion detection. In the event of a gene rearrangement event, the probes located in ROS1 fuse with partner genes to create the oncogenic ROS1 fusion gene. The widely established criteria for detection by FISH includes a threshold of ≥15% positive cells out of a count of at least 100 cells. The specific guidelines for ROS1 detection by FISH are well documented and published.23 50 It should be noted however, that there are limited data with regard to technical failure rates of FISH for ROS1 detection. The FISH methodology and specifics for ROS1 detection are discussed in detail by a recent review article by Bubendorf et al.19
ROS1 detection by RT-PCR
The RT-PCR approach requires a priori knowledge of the candidate fusion genes, and is designed with multiple specific primer sets to detect and identify fusion variants, that can later be confirmed by subsequent sequencing.51 As noted earlier, the common break point regions in ROS1 are located at exons 32, 34, 35 and 36 and RT-PCR-based detection of ROS1 fusion genes (SLC34A2, CD74, TPM3, SDC4, EZR, LRIG3, FIG, KDELR2, CCDC6) have been successfully designed and used in studies with high level of sensitivity and varying range of specificity (85%–100%) with respect to FISH.27 52 The multiplexed assay with RT-PCR is a rapid detection system that is relatively easy to perform and requires a less cumbersome laboratory set-up. However, the drawbacks include extraction of good quality RNA from formalin-fixed paraffin embedded tissue commonly used in tissue fixation in pathology laboratories worldwide and the likelihood of missing rare ROS1 fusion variants due to the abundance of ever-growing new candidate fusion gene partners. The recently described nanostring technology is a multiplexed assay for detection of known fusion gene variants, has shown good concordance with FISH results and is a promising platform for ROS1 fusion detection.53
ROS1 detection by NGS
NGS-based techniques have attracted widespread popularity in recent times, due to their ability to concurrently provide clinically relevant molecular data that cater to both actionable mutations (ie, EGFR, KRAS, BRAF) as well as targetable structural variations (ie, ALK, ROS1, RET gene rearrangements, gene amplification, gene deletion, etc). A number of innovative and promising approaches have been described recently.54–57 However, NGS-based techniques remain expensive tools requiring complex laboratory set-up and in-depth bioinformatics analytical tools. The widespread implementation of NGS technologies in day-to-day clinics remains a work in progress.
ROS1 detection by IHC
ROS1 fusion events are rare in NSCLC and other tumours. The best approach towards detecting rare events is by employing an easier, rapid and sensitive initial screening detection method, followed by subsequent confirmation by using more specific (and often, expensive) tests. The IHC-based detection approach for ROS1 is an automatic choice and would appropriately suit the purpose of detecting these rare, but therapeutically relevant fusion tumours. The seminal work in the IHC-based detection method was based on the work by Rimkunas et al who introduced the widely used D4D6 antibody clone (Cell Signalling Technology, Massachusetts, USA).9 Across multiple studies, the IHC-based method is an adequately sensitive marker for detecting ROS1 rearrangements; however, the problem remains with its specificity (figure 4).23 50 52 58–63 The drawbacks of ROS1 IHC detection mostly arises from experimental variability across various detection systems, antigen retrieval methodologies and staining conditions—which may explain the heterogeneity of results (especially specificity) that has been reported thus far.
Therapy of ROS1 fusion NSCLC
Crizotinib, a small molecule TKI was originally developed for MET inhibition; incidentally it was found to inhibit other kinases such as ALK and ROS1.6 18 There is significant amino acid sequence homology between ALK and ROS1 at their tyrosine kinase, ATP binding and crizotinib binding site domains.6 64 Thus far, data from four clinical trials have been published that corroborate significant clinical activity of crizotinib in ROS1 fusion tumours.26 28 65 66 In the first published trial (PROFILE 1001), Shaw et al reported a median progression-free survival (PFS) of 19.2 months and an objective response rate (ORR) of 72%.28 Mazieres et al reported similar ORR of 80%, however, the median PFS was lower (7.2 months) in the EUROS1 trial.26 The two ongoing phase II trials have reported similar ORRs (69% and 63%, respectively); other trial-specific survival statistics and PFS have not yet been published.65 66 Eventually, with long-standing therapy with TKIs, most of the tumours develop resistance mutations. The data on crizotinib TKI resistance mutations in ROS1 are still emerging; thus far, only a handful of studies have described crizotinib resistance mutations in CD74-ROS1 and EZR-ROS1 fusion tumours.67 68 Fortunately, the next-generation selective TKIs ioratinib, ceritinib, alectinib and entrectinib are active against most of these crizotinib-resistant tumours, have excellent activity to other RTKs including ALK and NTRK, display good clinical activity in the central nervous system, and have shown excellent clinical response overall.69
Future direction and conclusion
Given the availability of targetable molecules, encouraging clinical response and with recent Food and Drug Administration approval of crizotinib, reflex testing for ROS1 is expected to increase exponentially. In many centres, ROS1 detection is prompted often in conjunction with EGFR mutation and ALK rearrangement in advanced NSCLCs. Optimal utilisation of tumour tissue, especially from small biopsies or cytology specimens and optimisation of multiplexed approaches remains an area of active research and development. We discussed the prevalence of ROS1 fusions in NSCLC as well as in carcinomas from various organs and other non-epithelial malignancies. It remains to be seen whether the large-scale targeted clinical trials of ROS1 inhibitors will be expanded in the therapy of other malignancies.
Take home messages
ROS1 is a tyrosine kinase and ROS1 fusion transcript functions as an oncogene by constitutively activating many common downstream signalling pathways that regulate cellular proliferation and survival.
ROS1 fusions, although rare, are identified in non-small cell lung carcinoma and in a number of other epithelial and non-epithelial malignancies.
ROS1 tumours are amenable to targeted therapy by crizotinib and other next-generation tyrosine kinase inhibitors (TKIs).
There is widespread interest to screen and detect ROS1 fusions in clinical laboratories by various detection methods including fluorescence in situ hybridisation, next-generation sequencing, real-time PCR and immunohistochemistry.
Multiple clinical trials have shown dramatic clinical response with crizotinib and other TKIs in ROS1 fusion tumours.
Handling editor Runjan Chetty
Contributors PP performed the literature search, prepared the tables and figures, and drafted the manuscript. ZK assisted with the literature search, and drafted a portion of the manuscript. PP and ZK completed final revision.
Competing interests None declared.
Provenance and peer review Commissioned; internally peer reviewed.
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.