Clinicopathological significance and diagnostic approach of ROS1 rearrangement in non-small cell lung cancer: a meta-analysis: ROS1 in non-small cell lung cancer

Abstract

Purpose:

The aim of this study was to investigate the rate of ROS1 rearrangement and concordance between ROS1 immunohistochemistry (IHC) and molecular tests in non-small cell lung cancer (NSCLC).

Methods:

The study included 10,898 NSCLC cases from 21 eligible studies. ROS1 rearrangement rates were evaluated in NSCLC by a meta-analysis, including subgroup analyses. In addition, we performed a concordance analysis and a diagnostic test accuracy review of ROS1 IHC in NSCLC.

Results:

The estimated overall rate of ROS1 rearrangement and IHC positivity was 2.4% (95% confidence interval (CI) 1.5, 3.7). In the subgroup analysis, which was based on tumor subtype, the rate of ROS1 rearrangement and IHC positivity was 2.9% (95% CI 1.9, 4.5) and 0.6% (95% CI 0.3, 1.2) in adenocarcinoma and non-adenocarcinoma, respectively. The overall concordance rate between ROS1 IHC and molecular tests was 93.4% (95% CI 78.3, 98.2). In ROS1 IHC positive and negative cases, the concordance rates were 79.0% (95% CI 43.3, 94.9) and 97.0% (95% CI 83.3, 99.5), respectively. The pooled sensitivity and the specificity of ROS1 IHC were 0.90 (95% CI 0.70, 0.99) and 0.82 (95% CI 0.79, 0.84), respectively. The diagnostic odds ratio and the area under the curve of the summary receiver operating characteristic curve were 118.01 (95% CI 11.81, 1179.67) and 0.9417, respectively.

Conclusion:

The rates of ROS1 rearrangement differed by tumor histologic subtype in NSCLC. ROS1 IHC may be useful for the detection of ROS1 rearrangement in NSCLC. Detailed criteria for evaluating ROS1 IHC are needed before it can be applied in daily practice.

Keywords

Non-small cell lung cancer ROS1 rearrangement immunohistochemistry meta-analysis diagnostic test accuracy review

Introduction

Although lung cancer has long been a disease characterized by late-stage diagnosis and no progress in treatment options, the last decade has yielded substantial progress with systemic therapies for molecular subgroups of patients with non-small cell lung cancer (NSCLC).¹ NSCLCs are subdivided by histologic type, including adenocarcinoma, squamous cell carcinoma, and adenosquamous carcinoma. Known oncogenic drivers of NSCLC include EGFR, ALK, MET, KRAS, HER2, RET, and ROS1.^2–4 These genetic alterations may be important for the application and prediction of various targeted therapies in NSCLC patients.⁵

ROS1, a tyrosine kinase receptor, belongs to the insulin receptor family.^6,7 Constitutively activated ROS1 fusion proteins induce the activation of oncogenic pathways, such as the PI3K/AKT/mTOR and RAS/MAPK/ERK pathways, and are involved in cell proliferation, survival, and cell cycling.^8,9 ROS1 genetic alterations are found less frequently in cancer, including NSCLC, and the clinicopathological significance of ROS1 rearrangement is controversial.^10-12 For the detection of ROS1 rearrangement, molecular tests, such as in situ hybridization (ISH) or DNA sequencing, are required. Although molecular tests are confirmative, it is difficult to test all NSCLC patients in daily practice. In addition, the disadvantages of these molecular tests include high costs, multiple steps, requirement of advanced technical skills, and high turnaround time compared to immunohistochemistry (IHC). Recently, ROS1 IHC has been introduced and studied; however, its diagnostic accuracy has not been fully elucidated in NSCLC.

In order to evaluate its role, we performed a concordance analysis between ROS1 IHC and molecular tests, and a diagnostic test accuracy review of ROS1 IHC in NSCLC. In addition, ROS1 rearrangement was investigated by subdividing histologic types and detection methods.

Materials and methods

Published study search and selection criteria

Relevant articles were obtained by searching the PubMed and MEDLINE databases until 30 November 2017.These databases were searched using the following key words: “non-small cell lung cancer or NSCLC” and “ROS1.” The titles and the abstracts of all searched articles were screened for exclusion. Review articles were also screened to find additional eligible studies. Search results were then reviewed and included if (a) the study was performed on human NSCLCs; and (b) there was information about ROS1 genetic alteration and IHC expression. Articles were excluded if they were (a) case reports or non-original articles; or (b) non-English language publications.

Data extraction

Data from all eligible studies were extracted by two independent authors. The included data were extracted from each of the eligible studies:^2,13–32 the first author’s name, the year of publication, the study location, the detection method for ROS1 genetic alteration, and the number of patients analyzed. For the meta-analysis, we extracted all data associated with the results of ROS1 genetic alteration and IHC.

Statistical analyses

For the meta-analysis, all data were analyzed using the Comprehensive Meta-Analysis software package (Biostat, Englewood, NJ, USA). We investigated the rates of ROS1 genetic alteration and IHC expression in NSCLC. The concordance rates were measured by agreement rates between ROS1 IHC and molecular tests. Because the eligible studies used various methods for the detection of ROS1 genetic alteration in various populations, a random-effects model was more suitable than a fixed-effects model. Heterogeneity between the studies was checked using the Q and I² statistics, and presented using P-values. Additionally, a sensitivity analysis was conducted to assess the heterogeneity of eligible studies and the impact of each study on the combined effect. To assess publication bias, Begg’s funnel plot and Egger’s test were used. If a significant publication bias was found, the fail-safe N and trim-fill tests were performed to confirm the degree of publication bias. The results were considered statistically significant at P< 0.05.

The review of diagnostic test accuracy was performed using the Meta-Disc program (version 1.4, Unit of Clinical Biostatics, the Ramon y Cajal Hospital, Madrid, Spain).³³ In order to calculate the pooled sensitivity and specificity, data were collected from each eligible study and forest plots were obtained. The summary receiver operating characteristic (SROC) curve was initially constructed by plotting “sensitivity” and “1-specificity” of each study, and the curve fitting was performed through linear regression using the Littenberg and Moses linear model.³⁴ Because the data were heterogeneous owing to differences in the evaluation criteria, the accuracy data were pooled by fitting a SROC curve and measuring the value of the area under the curve (AUC).³³ An AUC close to 1 indicated a strong test and an AUC close to 0.5 was considered a poor test. In addition, the diagnostic odds ratio (OR) was calculated by the Meta-Disc program.

Results

Selection and characteristics

A total of 354 reports were identified in the database search for the meta-analysis. Among them, 156 were excluded because of insufficient information on ROS1 gene rearrangement as determined by molecular tests and/or ROS1 IHC expression. Also, 150 were excluded because they were non-English or non-original articles, and 23 were excluded because they reported the results of other diseases or used animal models or cell lines. Four duplicate articles were also excluded. In total, 21 studies were included in the meta-analysis (Figure 1 and Table 1), and the number of patients was 10,898. Table 1 shows the ROS1 rearrangement and IHC positivity data for the eligible studies. In all immunohistochemistry, the antibody clone was D4D6, which was manufactured by Cell Signaling Technology.

Figure 1.

Flow chart of study search and selection methods.

Table 1.

Main characteristics of the eligible studies.

Study	Country	Antibody clone	Detection system	N	Number of genetic alteration in molecular test						ROS1IHCpositive
Study	Country	Antibody clone	Detection system	N	Total	AdCa	SCC	ASC	PCa	LCC	ROS1IHCpositive
Bergethon et al. 2012¹³	USA			1073	18	18/694	0/200	0/10
Bu et al. 2017¹⁴	China			1875	11
Cai et al. 2013¹⁵	China			392	8	7/231	0/119	1/37
Clavé et al. 2016¹⁶	UK	D4D6	FlexPlus	283	5	4/260	0/11			0/6	4/282
Fu et al. 2015¹⁷	China			204	4	4/137	0/67
Go et al. 2013¹⁸	Korea			515	8	8/236	0/167
Ha et al. 2015¹⁹	Korea			198	2	2/198
Jurmeister et al. 2015²⁰	Germany			473	4	4/335	0/108
Kim et al. 2013²¹	Korea			208	7	7/208
Lee et al. 2013²²	Korea			150	33
Mescam-Mancini et al. 2014²³	France			107	9	9/107
Rogers et al. 2015²⁴	Australia	D4D6	OptiView universal	317	3						3/322
Sacher et al. 2016²⁵	USA			1368	29
Scheffler et al. 2015²⁶	Various			1035	19	19/1035
Selinger 2017²	Australia	D4D6	UltraView
Retrospective cohort				278	1	1/277		0/1			89/278
Prospective cohort				104	12	11/98		1/3		0/3	17/102*
Song et al. 2016²⁷	China			385	8	8/385
Viola et al. 2016²⁸	UK	D4D6	FlexPlus	85							2/85
Warth et al. 2014²⁹	Germany	D4D6	FlexPlus	1387							66/1387
Wu et al. 2015³⁰	China			127	5	5/127
Zhang et al. 2016³¹	China			120	2
Zhao et al. 2016³²	China			214	0		0/214

AdCa: adenocarcinoma; ASC: adenosquamous carcinoma; FISH: fluorescence in situ hybridization; IHC: immunohistochemistry; LCC: large cell carcinoma; N: number of patients; PCa: pleomorphic carcinoma; SCC: squamous cell carcinoma.

Two cases were unable to be tested.

Meta-analysis for ROS1 genetic alteration in non-small cell lung cancer

The estimated overall rate of ROS1 rearrangement and IHC positivity was 2.4% (95% CI 1.5, 3.7; Table 2). There was no significant publication bias identified by Egger’s test (P = 0.072) or Begg’s funnel plot. Subgroup analysis based on histologic type was performed. The rate of ROS1 rearrangement and IHC positivity in adenocarcinoma and non-adenocarcinoma was 2.9% (95% CI 1.9, 4.5) and 0.6% (95% CI 0.3, 1.2), respectively. Among non-adenocarcinomas, the rate of ROS1 rearrangement and IHC positivity in adenosquamous carcinoma and pleomorphic carcinoma was 3.5% (95% CI 1.1, 10.3) and 3.8% (95% CI 0.6, 19.1), respectively. However, ROS1 rearrangement was not found in squamous cell carcinoma. The ROS1 rearrangement rates varied by the detection method. The range of ROS1 rearrangement rates was 0.6, 22.0% and 2.0, 2.1% by fluorescence in situ hybridization (FISH) and DNA sequencing, respectively. In a subgroup analysis based on the detection method, the estimated rate of ROS1 rearrangement was 2.5% (95% CI 1.4, 4.5), 2.1% (95% CI 1.3, 3.3), and 0.8% (95% CI 0.2, 2.6) by FISH, DNA sequencing, and the NanoString technique, respectively. The range of ROS1 IHC positivity was 0.9, 32.0% in four eligible studies. All studies used the clone D4D6 (Cell Signaling Technology, Danvers, MA, USA) at a dilution of 1:50 or 1:300. The positive rate of ROS1 IHC was 2.0% (95% CI 0.8, 5.3). In addition, for the subgroup analysis based on the evaluation criteria, the eligible studies were classified into two subgroups: (a) staining intensity alone; and (b) a combination of intensity and percentage of positive tumor cells. The ROS1 IHC positivity of the intensity-only subgroup was significantly higher than that of the combination subgroup (4.8%, 95% CI 3.8, 6.0 vs. 1.3%, 95% CI 0.7, 2.6).

Table 2.

Meta-analysis for ROS1 genetic alteration and IHC expression in non-small cell lung cancer.

Stratified analysis	Numberof studies	Fixed effect(95% CI)	Heterogeneity(P-value)	Random effect(95% CI)	Egger’stest
Overall	22	3.5% (3.1, 3.9)	< 0.001	2.4% (1.5, 3.7)	0.072
Tumor type
Adenocarcinoma	14	4.2% (3.6, 4.9)	< 0.001	2.9% (1.9, 4.5)	0.018
Non-adenocarcinoma	9	0.6% (0.3, 1.2)	0.954	0.6% (0.3, 1.2)	0.556
Adenosquamous carcinoma	4	3.5% (1.1, 10.3)	0.639	3.5% (1.1, 10.3)	0.107
Pleomorphic carcinoma	2	3.6% (0.7, 16.2)	0.285	3.8% (0.6, 19.1)	–
Molecular tests	20	3.2% (2.7, 3.6)	< 0.001	2.2% (1.3, 3.8)	0.123
FISH	16	3.4% (2.9, 3.9)	< 0.001	2.5% (1.4, 4.5)	0.248
DNA sequencing	2	2.1% (1.3, 3.3)	0.971	2.1% (1.3, 3.3)	–
NanoString technique	2	0.8% (0.2, 2.6)	0.351	0.8% (0.2, 2.6)	–
IHC	4	4.1% (3.3, 5.2)	0.003	2.0% (0.8, 5.3)	0.076
Only intensity	1	4.8% (3.8, 6.0)	1.000	4.8% (3.8, 6.0)	–
Combination	3	1.3% (0.7, 2.6)	0.432	1.3% (0.7, 2.6)	0.919

CI: confidence interval; FISH: fluorescence in situ hybridization; IHC: immunohistochemistry.

Next, to assess the usefulness of ROS1 IHC, a concordance analysis between ROS1 IHC and molecular tests was performed in NSCLC cases (Table 3). The overall concordance rate was 93.4% (95% CI 78.3, 89.2). In the subgroup analysis based on ROS1 immunoreactivity, the estimated concordance rate was 79.0% (95% CI 43.3, 94.9) and 97.0% (95% CI 83.3, 99.5) in the ROS1 IHC positive and negative subgroups, respectively. In the subgroup analysis based on evaluation criteria, the concordance rate of the intensity-only subgroup was significantly higher than that of the combination subgroup (98.6%, 95% CI 56.1, 100.0 vs. 90.7%, 95% CI 69.7, 97.6). In ROS1 IHC positive cases, the concordance rate was 90.0% (95% CI 32.6, 99.4) and 74.2% (95% CI 26.3, 95.9) in the intensity-only and the combination subgroup, respectively. In ROS1 IHC negative cases, there was no difference in the concordance rate between the intensity-only and the combination subgroup (99.8%, 95% CI 97.2, 100.0 vs. 94.9%, 95% CI 73.7, 99.2).

Table 3.

The concordance analysis of ROS1 immunohistochemistry and genetic alteration in non-small cell lung cancers.

Stratified analysis	Numberof studies	Fixed effect(95% CI)	Heterogeneity(P-value)	Random effect(95% CI)	Egger’stest
Overall	9	75.3% (70.7, 79.4)	< 0.001	93.4% (78.3, 98.2)	0.063
Only intensity	2	98.7% (91.3, 99.8)	0.045	98.6% (56.1, 100.0)	–
Combination	7	74.5% (69.8, 78.6)	< 0.001	90.7% (69.7, 97.6)	0.166
IHC positive	4	78.8% (47.6, 93.8)	0.281	79.0% (43.3, 94.9)	0.328
Only intensity	1	90.0% (32.6, 99.4)	1.000	90.0% (32.6, 99.4)	–
Combination	3	74.0% (36.2, 93.4)	0.186	74.2% (26.3, 95.9)	–
IHC negative	5	75.2% (70.5, 79.3)	< 0.001	97.0% (83.3, 99.5)	0.023
Only intensity	1	99.8% (97.2, 100.0)	1.000	99.8% (97.2, 100.0)	–
Combination	4	74.5% (69.7, 78.7)	< 0.001	94.9% (73.7, 99.2)	0.087

CI: confidence interval; IHC: immunohistochemistry.

Diagnostic test accuracy review

To confirm the diagnostic accuracy of ROS1 IHC for predicting ROS1 rearrangement, we performed a diagnostic test accuracy review. The pooled sensitivity and specificity were 0.90 (95% CI 0.70, 0.99) and 0.82 (95% CI 0.79, 0.84), respectively (Figure 2). The sensitivity of the eligible studies ranged from 0.33 to 1.00, and the specificity ranged from 0.50 to 1.00. The AUC of the SROC curve was 0.9417. The diagnostic OR was 118.01(95% CI 11.81, 1179.67).

Figure 2.

The sensitivity (a) and specificity (b) of ROS1 immunohistochemistry in non-small cell lung cancer (NSCLC).

Discussion

Various biomarkers, such as BRAF, ERBB2, MET, ROS1, and RET, have been evaluated for targeted therapies in daily practice. In addition, their genetic status can correlate with prognosis. Accurate tests for various biomarkers are important for predicting therapeutic effects in NSCLC. In previous studies, we confirmed the diagnostic role and accuracy of ALK and c-MET IHC in NSCLC.^35,36 Taken together, an IHC panel including ALK, c-MET, and ROS1 can be useful for screening genetic alterations in NSCLC.^35–40 However, the clinical usefulness and diagnostic accuracy of ROS1 IHC is not fully understood. The present study is the first meta-analysis to assess the ROS1 rearrangement rate and the diagnostic accuracy of ROS1 IHC in NSCLC.

Although treatments using tyrosine kinase inhibitors (TKIs) have been developed and applied in NSCLC, therapeutic effects can differ according to tumor nature, such as genetic alteration and histologic type.⁵ Patients with ROS1 rearrangement had poorer outcomes after treatment with EGFR-TKIs.²¹ Crizotinib, an ALK inhibitor, shows therapeutic efficacy in NSCLC patients with ROS1 rearrangement.^41,42 Of 50 patients, 72% had objective response to crizotinib with a median progression-free survival of 19.2 months. Tumor responses were observed regardless of the specific ROS1 fusion partner.³⁷ A European retrospective study of 32 patients reported showed similar results.⁴² Other ROS1 inhibitors are currently being evaluated, including ceritinib, cabozantinib and entrectinib.¹ To obtain optimal therapeutic effect, accurate molecular tests are warranted in daily practice. Studies for ROS1 genetic alteration in NSCLC have been reported since 2007. ROS1 rearrangement status correlates with patient status, including female gender, history of non-smoking, and adenocarcinoma histology.⁴³ However, in some studies, the correlation between ROS1 rearrangement and tumor stage is controversial.²⁹ In addition, the correlation between ROS1 rearrangement and patient survival is unclear. Some studies have reported that ROS1 rearrangement significantly correlated with poor survival in patients with NSCLC.^15,44,45 In other studies, there was no significant correlation between ROS1 rearrangement and survival.^13,29,46-48 To obtain the conclusive information regarding ROS1 genetic alteration, large-scale prospective studies and/or meta-analyses are required.

Adenocarcinoma and squamous cell carcinoma are the two most common histological types of NSCLC, accounting for 50% and 30% cases, respectively.⁴ The incidence of ROS1 rearrangement was 2.4% of patients with adenocarcinoma.⁴³ Concordant with our results, the rate of ROS1 rearrangement in adenocarcinoma was higher than that in non-adenocarcinoma in a previous meta-analysis.⁴³ However, there was no ROS1 rearrangement in squamous cell carcinoma found in the eligible studies. The rates of ROS1 rearrangement in other histologic types, such as adenosquamous carcinoma and pleomorphic carcinoma, were higher than that in adenocarcinoma (Table 2). Therefore, molecular tests for ROS1 can be required for all NSCLCs, except squamous cell carcinomas.⁴³ In addition, ROS1 rearrangement was frequently found in female patients, non-smokers, and patients with advanced stage cancer. In selected patients with triple wild type EGFR/KRAS/ALK genotype and non-smoking history, ROS1 rearrangement accounted for 7.4–8.3%.^23,30,44 Patient selection according to clinicopathological characteristics would not be effective for detecting ROS1 genetic alterations. Since the incidence rate of ROS1 genetic alterations is very low in NSCLCs, molecular tests cannot be applied to all patients, and an effective screening method is needed for daily practice.

Although ROS1 IHC in NSCLC has been introduced and studied, its diagnostic accuracy has not been fully elucidated. Whether ROS1 IHC in NSCLC can replace molecular tests, such as FISH or direct sequencing, is not yet known. In many laboratories, IHC is used more commonly than molecular tests. In addition, molecular tests remain expensive and labor-intensive. In the present meta-analysis, the overall concordance rate analysis between ROS1 IHC and molecular tests was 93.4% (95% CI 78.3, 98.2). In ROS1 IHC negative cases, the concordance rate was very close to 1 (97.0%, 95% CI 83.3, 99.5). According to our results, the ROS1 IHC negative cases could be considered as negative for rearrangement without additional molecular testing, such as for c-MET and ALK IHC in NSCLC.^35,36 Therefore, ROS1 IHC can be useful for screening ROS1 rearrangement in NSCLC. However, the concordance rate between ROS1 IHC and molecular tests in ROS1 IHC positive cases was lower than that in ROS1 IHC negative cases. The ROS1 IHC positive rates were variable, ranging from 0.6% to 4.8%.^16,24,28,29 The variability among the reports may be caused by different evaluation criteria and patient populations. In addition, ROS1 immunostaining can be heterogeneous, such as in cases of breast and gastric cancers with 2+ HER2 scores, and the possibility of false positivity of ROS1 IHC should be considered. In the current meta-analysis, subgroup analysis based on the evaluation criteria of ROS1 IHC was also performed. The concordance rate in the intensity-only subgroup was significantly higher than that in the combination subgroup (4.8%, 95% CI 3.8, 6.0 vs. 1.3%, 95% CI 0.7, 2.6, P < 0.001). Therefore, cumulative studies are needed to obtain the definitive evaluation criteria of ROS1 IHC.

There are some limitations in the current meta-analysis. First, the ranges of the sensitivities and the specificities of the eligible studies were 0.33–1.00 and 0.50–1.00, respectively. Only four eligible studies showed the correlation between ROS1 IHC and molecular tests. Therefore, to obtain more conclusive information for the sensitivity and the specificity of ROS1 IHC, further cumulative studies are required. Second, concordance analysis based on subgroups by molecular tests could not be performed due to insufficient information from eligible studies. Third, in ROS1-positive cases, the concordance rate was as low as 79.0% (95% CI 43.3, 94.9), compared to ROS1 negative cases. However, in the current meta-analysis, the cause of the low concordance rate in ROS1-positive cases could not be found because there were too few eligible studies analyzed.

In conclusion, the rates of ROS1 rearrangement differed according to histologic types of NSCLC and detection methods. In addition, ROS1 IHC showed good concordance with ROS1 molecular tests in NSCLC. Further studies of detailed evaluation criteria and diagnostic flow for the detection of ROS1 rearrangement are required before application in daily practice.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the 2017 Inje University research grant.

References

Hirsch

Scagliotti

Mulshine

et al . Lung cancer: current therapies and new targeted treatments. Lancet 2017; 389 (10066): 299–311.

Selinger

Pavlakis

et al . Screening for ROS1 gene rearrangements in non-small-cell lung cancers using immunohistochemistry with FISH confirmation is an effective method to identify this rare target. Histopathology 2017; 70 (3): 402–411.

Travis

Brambilla

Burke

et al . WHO classification of tumours of the lung, pleura, thymus and heart. 4th ed. Lyon, France: International Agency for Research on Cancer, 2015.

Perez-Moreno

Brambilla

Thomas

et al . Squamous cell carcinoma of the lung: molecular subtypes and therapeutic opportunities. Clin Cancer Res 2012; 18 (9): 2443–2451.

Rothschild

. Targeted therapies in non-small cell lung cancer-beyond EGFR and ALK. Cancers (Basel) 2015; 7 (2): 930–949.

Rikova

Guo

Zeng

et al . Global survey of phosphotyrosine signaling identifies oncogenic kinases in lung cancer. Cell 2007; 131 (6): 1190–1203.

Soda

Choi

Enomoto

et al . Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature 2007; 448 (13): 561–566.

Chin

Soo

Soong

et al . Targeting ROS1 with anaplastic lymphoma kinase inhibitors: a promising therapeutic strategy for a newly defined molecular subset of non-small-cell lung cancer. J Thorac Oncol 2012; 7 (11): 1625–1630.

Davies

Doebele

. Molecular pathways: ROS1 fusion proteins in cancer. Clin Cancer Res 2013; 19 (15): 4040–4045.

10.

Pan

Zhang

et al . ALK, ROS1 and RET fusions in 1139 lung adenocarcinomas: a comprehensive study of common and fusion pattern-specific clinicopathologic, histologic and cytologic features. Lung Cancer 2014; 84 (2): 121–126.

11.

Zhang

et al . ALK and ROS1 as targeted therapy paradigms and clinical implications to overcome crizotinib resistance. Oncotarget 2016; 7 (11): 12289–12304.

12.

Tan

Yen

et al . ROS1 as a “druggable” receptor tyrosine kinase: lessons learned from inhibiting the ALK pathway. Expet Rev Anticancer Ther 2012; 12 (4): 447–456.

13.

Bergethon

Shaw

et al . ROS1 rearrangements define a unique molecular class of lung cancers. J Clin Oncol 2012; 30 (8): 863–870.

14.

Wang

Pan

et al . Clinicopathologic characteristics of patients with HER2 insertions in non-small cell lung cancer. Ann Surg Oncol 2017; 24 (1): 291–297.

15.

Cai

et al . ROS1 fusions in Chinese patients with non-small-cell lung cancer. Ann Oncol 2013; 24 (7): 1822–1827.

16.

Clavé

Gimeno

Muñoz-Mármol

et al . ROS1 copy number alterations are frequent in non-small cell lung cancer. Oncotarget 2016; 7 (7): 8019–8028.

17.

Liang

Lin

et al . The frequency and clinical implication of ROS1 and RET rearrangements in resected stage IIIA-N2 non-small cell lung cancer patients. PLoS One 2015; 10 (4): e0124354.

18.

Kim

et al . Clinicopathologic analysis of ROS1-rearranged non-small-cell lung cancer and proposal of a diagnostic algorithm. J Thorac Oncol 2013; 8 (11): 1445–1450.

19.

Choi

Cho

et al . Lung cancer in never-smoker Asian females is driven by oncogenic mutations, most often involving EGFR. Oncotarget 2015; 6 (7): 5465–5474.

20.

Jurmeister

Lenze

Berg

et al . Parallel screening for ALK, MET and ROS1 alterations in non-small cell lung cancer with implications for daily routine testing. Lung Cancer 2015; 87 (2): 122–129.

21.

Kim

Lim

Kim

et al . The frequency and impact of ROS1 rearrangement on clinical outcomes in never smokers with lung adenocarcinoma. Ann Oncol 2013; 24 (9): 2364–2370.

22.

Lee

Seol

Kim

et al . ROS1 receptor tyrosine kinase, a druggable target, is frequently overexpressed in non-small cell lung carcinomas via genetic and epigenetic mechanisms. Ann Surg Oncol 2013; 20 (1): 200–208.

23.

Mescam-Mancini

Lantuéjoul

Moro-Sibilot

et al . On the relevance of a testing algorithm for the detection of ROS1-rearranged lung adenocarcinomas. Lung Cancer 2014; 83 (2): 168–173.

24.

Rogers

Russell

Wright

et al . Comparison of methods in the detection of ALK and ROS1 rearrangements in lung cancer. J Thorac Oncol 2015; 10 (4): 611–618.

25.

Sacher

Dahlberg

Heng

et al . Association between younger age and targetable genomic alterations and prognosis in non-small-cell lung cancer. JAMA Oncol 2016; 2 (3): 313–320.

26.

Scheffler

Schultheis

Teixido

et al . ROS1 rearrangements in lung adenocarcinoma: prognostic impact, therapeutic options and genetic variability. Oncotarget 2015; 6 (12): 10577–10585.

27.

Song

Cheng

et al . Programmed death-ligand 1 expression associated with molecular characteristics in surgically resected lung adenocarcinoma. J Transl Med 2016; 14 (1): 188.

28.

Viola

Maurya

Croud

et al . A validation study for the use of ROS1 immunohistochemical staining in screening for ROS1 translocations in lung cancer. J Thorac Oncol 2016; 11 (7): 1029–1039.

29.

Warth

Muley

Dienemann

et al . ROS1 expression and translocations in non-small-cell lung cancer: clinicopathological analysis of 1478 cases. Histopathology 2014; 65 (2): 187–194.

30.

Wang

Zhou

et al . Clinicopathological characteristics and outcomes of ROS1-rearranged patients with lung adenocarcinoma without EGFR, KRAS mutations and ALK rearrangements. Thorac Cancer 2015; 6 (4): 413–420.

31.

Zhang

Sun

Kang

et al . Combined analysis of rearrangement of ALK, ROS1, somatic mutation of EGFR, KRAS, BRAF, PIK3CA, and mRNA expression of ERCC1, TYMS, RRM1, TUBB3, EGFR in patients with non-small cell lung cancer and their clinical significance. Cancer Chemother Pharmacol 2016; 77 (3): 583–593.

32.

Zhao

Choi

Song

et al . ALK, ROS1 and RET rearrangements in lung squamous cell carcinoma are very rare. Lung Cancer 2016; 94: 22–27.

33.

Zamora

Abraira

Muriel

et al . Meta-DiSc: a software for meta-analysis of test accuracy data. BMC Med Res Methodol 2006;6:31.

34.

Moses

Shapiro

Littenberg

. Combining independent studies of a diagnostic test into a summary ROC curve: data-analytic approaches and some additional considerations. Stat Med 1993; 12 (14): 1293–1316.

35.

Pyo

Kang

Cho

et al . Clinicopathological significance and concordance analysis of c-MET immunohistochemistry in non-small cell lung cancers: a meta-analysis. Pathol Res Pract 2016; 212 (8): 710–716.

36.

Pyo

Kang

Sohn

. ALK immunohistochemistry for ALK gene rearrangement screening in non-small cell lung cancer: a systematic review and meta-analysis. Int J Biol Markers 2016; 31 (4): e413–e421.

37.

Facchinetti

Rossi

Bria

et al . Oncogene addiction in non-small cell lung cancer: Focus on ROS1 inhibition. Cancer Treat Rev 2017; 55: 83–95.

38.

Ilie

Bence

Hofman

et al . Discrepancies between FISH and immunohistochemistry for assessment of the ALK status are associated with ALK ‘borderline’-positive rearrangements or a high copy number: a potential major issue for anti-ALK therapeutic strategies. Ann Oncol 2015; 26 (1): 238–244.

39.

Rossi

Jocollé

Conti

et al . Detection of ROS1 rearrangement in non–small cell lung cancer: current and future perspectives. Lung Cancer (Auckl) 2017; 8: 45–55.

40.

Rossi

Ragazzi

Tamagnini

et al . Does immunohistochemistry represent a robust alternative technique in determining drugable predictive gene alterations in non-small cell lung cancer? Curr Drug Targets 2017; 18 (1): 13–26.

41.

Shaw

Bang

et al . Crizotinib in ROS1-rearranged non-small-cell lung cancer. N Engl J Med 2014; 371 (21): 1963–1971.

42.

Kwak

Bang

Camidge

et al . Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. N Engl J Med 2010; 363 (18): 1693–1703.

43.

Zhu

Zhan

Zhang

et al . Clinicopathologic characteristics of patients with ROS1 fusion gene in non-small cell lung cancer: a meta-analysis. Transl Lung Canc Res 2015; 4 (3): 300–309.

44.

Kim

Shim

Kang

et al . Clinical and prognostic implications of ALK and ROS1 rearrangements in never-smokers with surgically resected lung adenocarcinoma. Lung Cancer 2014; 83 (3): 389–395.

45.

Jin

Sun

Kim

et al . ROS1 gene rearrangement and copy number gain in non-small cell lung cancer. Virchows Arch 2015; 466 (1): 45–52.

46.

Mazières

Zalcman

Crinò

et al . Crizotinib therapy for advanced lung adenocarcinoma and a ROS1 rearrangement: results from the EUROS1 cohort. J Clin Oncol 2015; 33 (9): 992–999.

47.

Yoshida

Kohno

Tsuta

et al . ROS1-rearranged lung cancer: a clinicopathologic and molecular study of 15 surgical cases. Am J Surg Pathol 2013; 37 (4): 554–562.

48.

Chen

Hsieh

et al . Clinical and the prognostic characteristics of lung adenocarcinoma patients with ROS1 fusion in comparison with other driver mutations in East Asian populations. J Thorac Oncol 2014; 9 (8): 1171–1179.