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 84 (2014) 121–126

Contents lists available at ScienceDirect

Lung Cancer journal homepage: www.elsevier.com/locate/lungcan

ALK, ROS1 and RET fusions in 1139 lung adenocarcinomas: A comprehensive study of common and fusion pattern-specific clinicopathologic, histologic and cytologic features Yunjian Pan a,b,1 , Yang Zhang a,b,1 , Yuan Li b,c , Haichuan Hu a,b , Lei Wang a,b , Hang Li a,b , Rui Wang a,b , Ting Ye a,b , Xiaoyang Luo a,b , Yiliang Zhang a,b , Bin Li a,b , Deng Cai a,b , Lei Shen b,c , Yihua Sun a,b,∗∗ , Haiquan Chen a,b,∗ a

Department of Thoracic Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, China Department of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China c Department of Pathology, Fudan University Shanghai Cancer Center, Shanghai 200032, China b

a r t i c l e

i n f o

Article history: Received 2 December 2013 Received in revised form 30 January 2014 Accepted 11 February 2014 Keywords: Lung adenocarcinoma ALK fusions ROS1 fusions RET fusions Histology Cytomorphology

a b s t r a c t Background: To have a comprehensive investigation of the clinicopathologic, histologic and cytologic features of fusion-positive lung adenocarcinomas. Methods: Quantitative real-time reverse transcriptase PCR (qRT-PCR) and reverse transcriptase PCR (RTPCR) were simultaneously performed to screen ALK, ROS1 and RET fusions in resected tumor samples from 1139 Chinese lung adenocarcinoma patients, with validation of positive results using fluorescent in situ hybridization. Clinicopathologic characteristics, predominant histologic subtype and cytomorphology were assessed in fusion-positive lung adenocarcinomas and compared to those harboring EGFR, KRAS, HER2 or BRAF mutations. Results: There were 58 (5.1%) ALK fusions, 11 (1.0%) ROS1 fusions and 15 (1.3%) RET fusions. Tumors with ROS1 fusions had significantly larger diameter than ROS1 fusion-negative tumors (P = 0.007), whereas all the 15 tumors harboring RET fusions were ≤3 cm in diameter (P = 0.001). The three fusion genes were all more prevalent in solid-predominant adenocarcinoma. Compared to fusion-negative lung adenocarcinomas, tumors harboring a fusion gene had significantly higher prevalence of extracellular mucin (P < 0.001), cribriform pattern (P < 0.001), signet ring cells (P < 0.001) and hepatoid cytology (P < 0.001). No significant difference in relapse-free survival (P = 0.147) and overall survival (P = 0.444) was observed between fusion-positive and fusion-negative patients. Conclusions: This study showed fusion-positive lung adenocarcinomas had identifiable common and fusion-pattern specific clinicopathologic, histologic and cytologic features, offering implications for fusion genes screening. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

∗ Corresponding author at: Department of Thoracic Surgery, Fudan University Shanghai Cancer Center, 270 Dong-An Road, Shanghai 200032, China. Tel.: +86 21 64175590; fax: +86 21 62686511. ∗∗ Corresponding author at: Department of Thoracic Surgery, Fudan University Shanghai Cancer Center, 270 Dong-An Road, Shanghai 200032, China. Tel.: +86 21 64175590; fax: +86 21 62686511. E-mail addresses: Sun [email protected] (Y. Sun), [email protected] (H. Chen). 1 contributed equally to this work and should be considered co-first authors. http://dx.doi.org/10.1016/j.lungcan.2014.02.007 0169-5002/© 2014 Elsevier Ireland Ltd. All rights reserved.

Lung adenocarcinoma, which is the most common histologic subtype of lung cancer, is now a disease that could be largely classified into clinically relevant molecular subsets by oncogenic driver mutations, each with unique clinicopathologic features and potential opportunities for targeted therapies. Recently, gene fusions have been identified as recurrent oncogenic events in lung adenocarcinoma. The first reported fusion oncokinase is ALK rearrangement [1], which occurs in approximately 5% of lung adenocarcinomas [2,3]. In ALK-positive lung cancer patients, crizotinib has demonstrated its treatment efficacy in terms of a higher response rate and improved progression

122

Y. Pan et al. / Lung Cancer 84 (2014) 121–126

free survival compared with standard chemotherapy [4]. ROS1 is another identified receptor tyrosine kinase that forms fusions, and is reported to be present in about 1–2% of lung adenocarcinomas [2,5,6]. ROS1 fusions also serve as therapeutic targets for crizotinib [6]. RET fusions have been identified as the novel oncogenic drivers in about 1–2% of lung adenocarcinomas [2,7–10], and the potential therapeutic targets of multi-targeted kinase inhibitors, vandetanib, sunitinib and sorafenib [2,7,8]. As the prevalence of fusion genes in lung adenocarcinoma is comparatively low, identifying the enriched population has definite implications for efficient screening. However, despite their critical importance in targeted therapies, the three fusion genes have not been examined together along with other well-identified driver mutations in a large cohort of lung adenocarcinoma to comprehensively define their common and fusion-pattern specific clinicopathologic, histologic and cytologic features. In this study, we used a combination strategy of quantitative real-time reverse transcriptase PCR (qRT-PCR) and reverse transcriptase PCR (RT-PCR) for screening of ALK, ROS1 and RET fusion genes in over 1139 lung adenocarcinoma patients, with validation of positive results using fluorescence in situ hybridization (FISH). Mutational analyses of EGFR, KRAS, HER2 and BRAF were also carried out in the same cohort of patients. Clinicopathologic, histologic and cytologic features of fusion-positive lung adenocarcinomas were comprehensively analyzed and compared to those of other wellidentified molecular subsets of lung adenocarcinoma.

2. Materials and methods 2.1. Patients and samples From April 2007 to October 2012, we consecutively collected lung tumors resected at the Department of Thoracic Surgery, Fudan University Shanghai Cancer Center. Eligible subjects must have pathologically confirmed lung adenocarcinoma, a minimum of 50% of tumor cells and sufficient tissue for comprehensive mutational analyses. Patients who received neoajuvant chemotherapy were excluded. Total RNA were extracted as per standard protocols (RNeasy Mini Kitt; Qiagen, Hilden, Germany) after frozen tumor specimens were dissected into TRIzol (Invitrogen). Total RNA samples were then reverse transcribed into cDNA using RevertAid First Strand cDNA Synthesis Kit (Fermentas, St Leon-Rot, Germany). We prospectively collected the following clinicopathologic data: sex, age at diagnosis, smoking history, and pathologic TNM stage. Neversmokers were defined as patients who never smoked any cigarettes in their lifetimes [3]. Histologic subtypes of lung adenocarcinoma were classified according to the new International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS) multidisciplinary classification of lung adenocarcinoma [11]. This study was conducted in accordance with the Helsinki Declaration, and was approved by the Institutional Review Board of the Fudan University Shanghai Cancer Center. All patients gave written informed consent. To investigate the associations between fusion genes and cytologic features including signet ring cells, hepatoid cytology, extracellular mucin and cribriform pattern, pathologic slides of all the 84 fusion-positive lung adenocarcinomas and 260 consecutive fusion-negative lung adenocarcinomas resected in 2012 (January to September) were re-reviewed by two certified pathologists (Lei Shen and Yuan Li) who were blinded from clinical data and mutational status. Rare discrepancies were resolved by re-examination of the slides and discussion. Hepatoid cytology was defined as the solid pattern with hepatoid tumor appearance characterized by abundant eosinophilic cytoplasm, round and relatively monomorphic nuclei, and prominent nucleoli. A cut-off value of 10% was used

to define the presence of the four aforementioned cytomorphologic features. 2.2. Mutational analyses of EGFR, KRAS, HER2 and BRAF Briefly, EGFR (exons 18-22), HER2 (exons 18-21), KRAS (exons 2-3) and BRAF (exons 11-15) were amplified by PCR using cDNA, as previously described [3]. Direct dideoxynucleotide sequencing was used to analyze the amplified products. 2.3. Detection of ALK, ROS1 and RET fusions Gain-of-function gene rearrangement events result in a higher transcription level of the kinase domain after the break point (ABP), compared to non-kinase region before the break point (BBP). Based on this unbalanced transcription level, we have developed quantitative real-time reverse transcriptase PCR (qRT-PCR) method to rapidly and accurately detect ALK fusions and RET fusions in NSCLC [9,12]. The detection methods for ALK and RET fusions (a combination of RT-PCR and qRT-PCR in all the samples) have been described previously [9,12]. To explore whether the qRT-PCR method is also applicable to detect ROS1 fusions, we carried out qRT-PCR and RT-PCR simultaneously in all the samples. Primers used for RT-PCR and qRT-PCR were listed in Supplementary Table S1 and S2, respectively. Exon junction expression values were expressed in ln2−CT [−CT = −(CT␤-actin − CTROS1 )]. Average expressions of BBP and ABP were calculated using (SUMBBP -MAXBBP -MINBBP )/3 and (SUMABP -MAXABP -MINABP )/4, respectively. Unbalanced index (AVERAGEBBP /AVERAGEABP ) was used to assess unbalanced level of mRNA expression. Samples with unbalanced index >1 and 36 additional “pan-negative” lung adenocarcinomas (samples without known mutations in EGFR, KRAS, HER2, BRAF, ALK, ROS1 or RET) with unbalanced index <1were subject to FISH break apart detection. 2.4. FISH FISH assays for confirmation of ALK and RET fusions have been described in our previous studies [9,12]. ROS1 FISH was performed in formalin-fixed, paraffin-embedded specimens using dualcolor breakapart probe obtained from Innovation Center China, Astrazeneca, with 5 ROS1 signal labeled with SpectrumRedTM (red) and 3 ROS1 signal labeled with SpectrumGreenTM (green). Hybridized slides were then stained with 4, 6-diamino-2phenylindole and examined with a BX51 fluorescence microscope (Olympus, Tokyo, Japan). Samples were considered to be positive if more than 15% of tumor cells showed split signals [6]. 2.5. Immunohistochemistry (IHC) Cases with ALK, ROS1 or RET fusions were subject to IHC analysis for TTF-1 and p63. Briefly, after deparaffinization and rehydration, slides were treated with EDTA and microwaving for antigen retrieval. Goat serum was used to block nonspecific immunoglobulin binding. Slides were then incubated with mouse monoclonal antihuman p63 (1:400, 4A4, DAKO) or TTF-1 (1:100, 8G7G3/1, DAKO). The slides were then washed, and incubated with secondary antibodies followed by incubation with DAB. Positive IHC staining was defined as more than 10% of tumor cells showed nuclear staining. 2.6. Statistical analysis We used the Pearson’s 2 test or Fisher’s exact test to assess the correlations between driver mutations and clinicopathologic,

Y. Pan et al. / Lung Cancer 84 (2014) 121–126

123

3.2. Clinicopathologic, histologic and cytologic characteristics of ALK, ROS1 or RET-positive lung adenocarcinomas

Fig. 1. Spectrum of well-identified oncogenic driver mutations in 1139 lung adenocarcinoma. “Panneg” refers to EGFR, KRAS, HER2, BRAF, ALK, ROS1 and RET negative.

histologic and cytologic variables. Multivariate analysis was done using penalized maximum likelihood logistic regression to overcome the problem of “separation” in conventional logistic regression [13]. All the 84 patients with fusion genes and patients with other gene status diagnosed from October 2007 to August 2011 (sufficient follow up period) were included for survival analysis. Survival curves were drawn by the Kaplan-Meier method. Relapse-free survival (RFS) and overall survival (OS) were compared using the log-rank test. The statistical analyses were done in SPSS 16.0 for Windows (Chicago, IL) and Stata 11.1 (StataCorp, College Station, Tex). All tests were two tailed, and a P value of <0.05 was set as statistically significant.

3. Results Tumor samples from 1139 lung adenocarcinoma patients were included in the comprehensive mutational analysis. There were 58 (5.1%) ALK fusions, 11 (1.0%) ROS1 fusions and 15 (1.3%) RET fusions. Mutations in EGFR, KRAS, HER2 and BRAF were detected in 735 (64.5%), 86 (7.6%), 27 (2.4%) and 14 (1.2%) patients, respectively. The spectrum of well-identified oncogenic driver mutations in 1139 lung adenocarcinoma was shown in Fig. 1.

3.1. The application of quantitative real-time PCR in the detection of ROS1 fusions The high accuracy of qRT-PCR method to detect ALK fusions and RET fusions in NSCLC has been well demonstrated in our previous studies [9,12]. In the present study, the results were consistent using the methods (including primers and cutoff for unbalanced index) described previously [9,12], with validation using FISH. Here, all the ROS1 fusion-positive samples detected by RT-PCR and direct sequencing showed unbalanced index >1.2 in qRT-PCR tests; meanwhile, all the tumors with unbalanced index >1.2 in qRTPCR tests were confirmed positive in both RT-PCR and ROS1 FISH (Supplementary Figure S1). We also carried out ROS1 FISH analysis in samples with unbalanced index between 1.0 and 1.2 and 36 additional “pan-negative” samples with unbalanced index <1, and showed that all these tumors were ROS1 FISH-negative.

Individual patient data of the 84 fusion-positive lung adenocarcinoma patients were listed in Supplementary Table S3. The detailed genotypic information of EGFR, KRAS, HER2 and BRAF mutations were listed in Supplementary Table S4. Patients harboring ALK fusions were significantly younger than ALK wild-type patients (56.0 vs. 59.6 years, P = 0.011) (Table 1). RET fusion-positive patients also seemed to be younger at diagnosis, although without statistical significance (54.5 vs.59.5 years, P = 0.071). Tumors with ROS1 fusions had significantly larger diameter than ROS1 fusion-negative tumors (4.1 vs. 2.7 cm, P = 0.007), whereas all the 15 tumors harboring RET fusions were ≤3 cm in diameter (P = 0.001). No adenocarcinoma in situ (AIS) or minimally invasive adenocarcinoma (MIA) harbored any fusion genes. ALK fusions were also not found in lepidic predominant adenocarcinoma (P = 0.036). Acinar predominant adenocarcinoma were less likely to harbor ROS1 fusions (P = 0.037). Solid predominant adenocarcinoma harbored a significantly higher proportion of RET fusions (P = 0.005), and higher percentages of ALK (P = 0.074) and ROS1 (P = 0.080) fusions with borderline significance. A significantly higher prevalence of ALK fusions were found in invasive mucinous adenocarcinoma (P = 0.006). Invasive mucinous adenocarcinoma (P = 0.007) and solid predominant adenocarcinoma (P < 0.001) were positively associated with fusionpositive status (Table 1). In multivariate analysis incorporating clinicopathologic and histologic variables, younger age (≤60 years) at diagnosis (P = 0.026) and invasive mucinous adenocarcinoma subtype (P = 0.017) independently predicted the presence of ALK fusions. Larger tumor size (>3 cm in diameter) independently predicted the presence of ROS1 fusions (P = 0.043), while smaller tumor size (≤3 cm in diameter) was an independent predictor of RET fusions (P = 0.029). To have a more comprehensive understanding of the clinicopathologic and histologic features of lung adenocarcinoma harboring a fusion gene, we also compared them with those harboring mutations in EGFR, KRAS, HER2 or BRAF (Supplementary Table S5, S6, S7 and S8). The distributions of females and never-smokers were similar in patients harboring one of the three fusion genes. Percentages of females or never-smokers in lung adenocarcinoma harboring a fusion gene were significantly higher than patients with KRAS or BRAF mutations, comparable with patients harboring EGFR mutations, and lower than patients harboring HER2 mutations. When comparing the clinicopathologic features of one fusion gene to those of another, we found RET-positive lung adenocarcinomas were significantly more likely to be small-sized tumors than those harboring ALK (P = 0.008) or ROS1 fusions (P = 0.001). Compared to fusion-negative lung adenocarcinomas, tumors harboring a fusion gene had significantly higher prevalence of extracellular mucin (57.1% vs. 16.9%, P < 0.001), cribriform pattern (52.4% vs. 12.7%, P < 0.001), signet ring cells (32.1% vs. 2.3%, P < 0.001) and hepatoid cytology (22.6% vs. 7.7%, P < 0.001). Interestingly, the prevalence of these cytomorphologies was very similar among lung adenocarcinomas with ALK, ROS1 or RET fusions (Table 2). Representative images of cytomorphology of fusionpositive lung adenocarcinomas were shown in Fig. 2.

3.3. Survival analysis A total of 573 patients were included for survival analysis, including 58 ALK fusions, 11 ROS1 fusions, 15 RET fusions, 346 EGFR mutations, 41 KRAS mutations, 15 HER2 mutations and 8 BRAF mutations. Patients harboring ALK fusions had significantly longer relapse-free survival than those with KRAS mutations (P = 0.044) or “pan-negative” status (P = 0.042). No statistical significance was

124

Y. Pan et al. / Lung Cancer 84 (2014) 121–126

Table 1 Clinicopathologic and histologic characteristics of 1139 lung adenocarcinomas harboring ALK, ROS1 or RET fusions. ALK (n = 58)

ROS1 (n = 11)

+

−

+

−

Age (years) Mean SD ≤60 >60

56.0 9.6 41 17

59.6 10.8 572 509

0.008

61.2 10.9 5 6

59.4 10.8 608 520

Sex Female Male

35 23

592 489

0.405

7 4

Smoking history 44 Never Ever 14

727 354

0.172

Tumor size (cm) Mean SD ≤3 >3

2.8 1.7 39 19

2.8 1.6 789 292

Stage I II–IV

25 33

Histologic subtypes 0 AIS 0 MIA 0 Lepidic 28 Acinar 7 Papillary Micropap 1 14 Solid IMA 8

RET (n = 15)

Fusions (n = 84)

+

−

0.576

54.5 11.5 11 4

59.5 10.7 602 522

620 508

0.763

10 5

7 4

764 364

1.000

0.339

4.1 2.5 4 7

2.7 1.6 824 304

596 485

0.073

5 6

18 41 87 535 165 16 166 49

0.620 0.165 0.036 0.857 0.508 1.000 0.074 0.006

0 0 1 2 3 0 4 1

P

0.011

0.886

+

−

0.127

56.4 10.2 57 27

59.7 10.8 556 499

617 507

0.363

52 32

575 480

0.189

12 3

759 365

0.410

63 21

708 347

0.137

0.012

2.1 0.6 15 0

2.8 1.6 813 311

0.016

2.8 1.7 58 26

2.8 1.6 770 285

616 512

0.544

7 8

614 510

0.539

37 47

584 471

0.045

18 41 86 561 169 17 176 56

1.000 1.000 1.000 0.037 0.387 1.000 0.080 1.000

0 0 1 5 1 0 7 1

18 41 86 558 171 17 173 56

1.000 0.674 1.000 0.209 0.492 1.000 0.005 1.000

0 0 2 35 11 1 25 10

18 41 85 528 161 16 155 47

0.391 0.067 0.059 0.139 0.594 1.000 <0.001 0.007

P

0.594

0.007

P

0.071

0.001

P

0.006 0.007

0.657 0.436

Bold represents statistically significant, P < 0.05. Abbreviations: SD, standard deviation; AIS, adenocarcinoma in situ; MIA, minimally invasive adenocarcinoma; Micropap, micropapillary; IMA, invasive mucinous adenocarcinoma.

achieved in other comparisons between a fusion gene and another driver mutation with regard to RFS or OS (Fig. 3). Compared to fusion-negative counterparts, the 84 patients with one of the fusion genes showed no significant difference in relapse-free survival (P = 0.147) and overall survival (P = 0.444) (Supplementary Figure S2). There were no significant correlations between molecular status and RFS or OS when the survival analysis was limited to a specific histologic or cytologic subtype. 3.4. IHC analysis All the 84 fusion-positive lung adenocarcinomas showed positive expression of TTF-1. A small proportion (7 out of 58, 12.1%) of ALK-positive lung adenocarcinomas also had positive staining of p63. Positive p63 staining was not found in any of the ROS1-positive or RET-positive cases. 4. Discussion Our previous studies have investigated ALK and RET fusions in non-small cell lung cancer in relatively smaller samples [9,12].

In this study, we extended the fusion gene detection to more lung adenocarcinoma samples and also included ROS1 fusions. Some previous studies reported that lung cancers harboring ALK, ROS1 or RET fusions shared some clinical characteristics, such as younger age of onset and never/light smoking history [2,6,14]. In this study, we also found some clinical characteristics in common. Although the gender distributions and smoking history were not significantly different between fusion-positive and fusionnegative patients, the statistical insignificance can be explained by the high prevalence of EGFR and HER2 mutations in East Asian lung adenocarcinoma patients, as we found that proportions of females or never-smokers in patients with any fusion gene all tended to be higher than patients with KRAS or BRAF mutations, comparable with patients harboring EGFR mutations, and lower than patients harboring HER2 mutations. The clinical outcomes for patients harboring one of the fusion genes seemed to be similar. However, in comparison to Takeuchi and colleagues’ [2] findings that positive kinase-fusion status was an indicator of good prognosis, we found no significant difference in relapse-free and overall survival between fusion-positive and fusion-negative status in this series of Chinese lung adenocarcinoma patients.

Table 2 Cytomorphologic features of fusion-positive lung adenocarcinomas.

Extracellular mucin Cribriform pattern Signet ring cells Hepatoid cytology

ALK+ (n = 58)

ROS1+ (n = 11)

RET+ (n = 15)

Fusion+ (n = 84)

Fusion− (n = 260)

Pa

32 (55.2%) 30 (51.7%) 19 (32.8%) 13 (22.4%)

6 (54.5%) 5 (45.5%) 3 (27.3%) 2 (18.2%)

10 (66.7%) 9 (60.0%) 5 (33.3%) 4 (26.7%)

48 (57.1%) 44 (52.4%) 27 (32.1%) 19 (22.6%)

44 (16.9%) 33 (12.7%) 6 (2.3%) 20 (7.7%)

<0.001 <0.001 <0.001 <0.001

Bold represents statistically significant, P < 0.05. a Comparison between fusion+ and fusion− lung adenocarcinomas.

Y. Pan et al. / Lung Cancer 84 (2014) 121–126

125

Fig. 2. Representative images of cytomorphology of fusion-positve lung adenocarcinomas. (A) ALK-positive lung adenocarcinoma shows cribriform pattern and extracellular mucin (patient ALK#8). (B) ROS1-positive lung adenocarcinoma shows extracellular mucin (patient ROS1#11). (C) RET-positive lung adenocarcinoma shows signet ring cells (patient RET#3). (D) ALK-positive lung adenocarcinoma shows hepatoid cytology (patient ALK#7).

Several significant distinctions were also observed among the three fusion genes. Patients with ALK or RET fusions seemed to be younger at diagnosis, while the mean age of ROS1-rearranged patients was comparable with ROS1-negative counterparts. RET fusion-positive tumors were characterized by small size (≤3 cm in diameter), whereas tumors harboring ROS1 rearrangements were significantly larger. These findings suggested the three fusion genes should not be viewed as totally identical when selecting enriched populations for screening.

We previously have showed the frequency of driver mutations in lung adenocarcinoma varies with histologic subtypes of lung adenocarcinoma according to the new IASLC/ATS/ERS classification system [15,16]. Here, we showed this adenocarcinoma classification system could provide some guidance for fusion genes detection. First, the three fusion genes were not detected in any of the 58 AIS or MIA cases. Although we could not totally rule out the presence of fusion genes in AIS or MIA, they were implied to be present in a very small proportion, if not totally absent. Second, none of

Fig. 3. (A) Relapse-free and (B) overall survival of lung adenocarcinoma patients harboring ALK, ROS1 or RET fusions as well as other molecularly defined cohorts. “Panneg” refers to EGFR, KRAS, HER2, BRAF, ALK, ROS1 and RET negative. Patients harboring ALK fusions have significantly longer relapse-free survival than those with KRAS mutations (P = 0.044) or “pan-negative” status (P = 0.042).

126

Y. Pan et al. / Lung Cancer 84 (2014) 121–126

the lepidic predominant adenocarcinoma was found to harbor ALK fusions. Nishino and colleagues [17] also found a significantly lower prevalence of ALK fusions in lepidic predominant adenocarcinoma than non-lepidic predominant counterparts, although they did find a small proportion of lepidic predominant adenocarcinoma harbored ALK fusions. Our study suggested that invasive mucinous adenocarcinoma might be an enriched population for ALK fusions screening, although a significantly positive correlation between ALK fusions and invasive mucinous adenocarcinoma has not been reported previously to our knowledge. Finally, solid predominant adenocarcinoma might be an enriched population for the screening of all the three fusion genes, especially RET fusions. Previous cytomorphologic analysis suggested that extracellular mucin [18,19], cribriform structure [18,19], signet ring cells [17,18] and hepatoid cytology [17] were positively associated with the prevalence of ALK rearrangements. Here, we also applied these cytologic analyses to ROS1 and RET fusions. Interestingly, we found that the prevalence of these morphologies in lung adenocarcinomas with ROS1 or RET fusions were very similar to those in ALK positive tumors, and all the four cytologic variables were significantly associated with the prevalence of a fusion gene, indicating fusion genes in lung adenocarcinomas had identical and identifiable cytomorphologic features. Break-apart FISH is currently considered the “Golden Standard” in the detection of the three fusion genes. However, its high expense and dependence on specialized equipments and expertise limited the application in the screening of large samples. RT-PCR can identify the specific fusion partner and breakpoint variant, but may miss unknown fusion partners or fusion variants. We previously found that qRT-PCR method, which is simple and high-throughput, was able to detect ALK and RET fusions with high accuracy [9,12]. Here, we showed that qRT-PCR could also be applied in the screening of ROS1 fusions. Although no new ROS1 fusion partners or variants were detected in this series of patients, through the application of 5 RACE in samples with unbalanced transcription level demonstrated by qRT-PCR, we have identified novel ALK and RET fusions which were missed in RT-PCR tests in our previous studies [9,12]. We might be able to find novel ROS1 fusions in the future screening using qRT-PCR methods. Collectively, we have developed an effective diagnostic method for the detection of all the three fusion genes. 5. Conclusion In conclusion, through comprehensive analysis of ALK, ROS1 and RET fusions along with other well-identified driver mutations in 1139 lung adenocarcinomas, we have comprehensively defined the common and fusion-pattern specific clinicopathologic, histologic and cytologic features of fusion-positive lung adenocarcinomas. Our results have implications for clinical trials seeking to enrich for patients with fusion genes as well as therapeutic strategies. Conflict of interest The authors declare no conflicts of interest. Funding source This work was supported by the funds from Key Construction Program of the National “985” Project (985III-YFX0102), the National Natural Science Foundation of China (81172218,

81101760, 81101761, and 81372525), the Science and Technology Commission of Shanghai Municipality (Program of Shanghai Subject Chief Scientist; Grant No. 12XD1402000), and the Shanghai Hospital Development Center (No. SHDC12012308). Acknowledgements We thank the AstraZeneca Innovation Center China for excellent technical support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.lungcan.2014.02.007. References [1] Soda M, Choi YL, Enomoto M, Takada S, Yamashita Y, Ishikawa S, et al. Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature 2007;448:561–6. [2] Takeuchi K, Soda M, Togashi Y, Suzuki R, Sakata S, Hatano S, et al. RET, ROS1 and ALK fusions in lung cancer. Nat Med 2012;18:378–81. [3] Sun Y, Ren Y, Fang Z, Li C, Fang R, Gao B, et al. Lung adenocarcinoma from East Asian never-smokers is a disease largely defined by targetable oncogenic mutant kinases. J Clin Oncol 2010;28:4616–20. [4] Shaw AT, Kim DW, Nakagawa K, Seto T, Crino L, Ahn MJ, et al. Crizotinib versus chemotherapy in advanced ALK-positive lung cancer. N Engl J Med 2013;368:2385–94. [5] Li C, Fang R, Sun Y, Han X, Li F, Gao B, et al. Spectrum of oncogenic driver mutations in lung adenocarcinomas from East Asian never smokers. PLoS One 2011;6:e28204. [6] Bergethon K, Shaw AT, Ou SH, Katayama R, Lovly CM, McDonald NT, et al. ROS1 rearrangements define a unique molecular class of lung cancers. J Clin Oncol 2012;30:863–70. [7] Kohno T, Ichikawa H, Totoki Y, Yasuda K, Hiramoto M, Nammo T, et al. KIF5BRET fusions in lung adenocarcinoma. Nat Med 2012;18:375–7. [8] Lipson D, Capelletti M, Yelensky R, Otto G, Parker A, Jarosz M, et al. Identification of new ALK and RET gene fusions from colorectal and lung cancer biopsies. Nat Med 2012;18:382–4. [9] Wang R, Hu H, Pan Y, Li Y, Ye T, Li C, et al. RET fusions define a unique molecular and clinicopathologic subtype of non-small-cell lung cancer. J Clin Oncol 2012;30:4352–9. [10] Ju YS, Lee WC, Shin JY, Lee S, Bleazard T, Won JK, et al. A transforming KIF5B and RET gene fusion in lung adenocarcinoma revealed from whole-genome and transcriptome sequencing. Genome Res 2012;22:436–45. [11] Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger KR, Yatabe Y, et al. International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 2011;6:244–85. [12] Wang R, Pan Y, Li C, Hu H, Zhang Y, Li H, et al. The use of quantitative real-time reverse transcriptase PCR for 5’ and 3’ portions of ALK transcripts to detect ALK rearrangements in lung cancers. Clin Cancer Res 2012;18:4725–32. [13] Heinze G, Schemper M. A solution to the problem of separation in logistic regression. Stat Med 2002;21:2409–19. [14] Yoshida A, Kohno T, Tsuta K, Wakai S, Arai Y, Shimada Y, et al. ROS1-rearranged lung cancer: a clinicopathologic and molecular study of 15 surgical cases. Am J Surg Pathol 2013;37:554–62. [15] Li H, Pan Y, Li Y, Li C, Wang R, Hu H, et al. Frequency of well-identified oncogenic driver mutations in lung adenocarcinoma of smokers varies with histological subtypes and graduated smoking dose. Lung Cancer 2013;79:8–13. [16] Zhang Y, Sun Y, Pan Y, Li C, Shen L, Li Y, et al. Frequency of driver mutations in lung adenocarcinoma from female never-smokers varies with histologic subtypes and age at diagnosis. Clin Cancer Res 2012;18:1947–53. [17] Nishino M, Klepeis VE, Yeap BY, Bergethon K, Morales-Oyarvide V, DiasSantagata D, et al. Histologic and cytomorphologic features of ALK-rearranged lung adenocarcinomas. Mod Pathol 2012;25:1462–72. [18] Yoshida A, Tsuta K, Nakamura H, Kohno T, Takahashi F, Asamura H, et al. Comprehensive histologic analysis of ALK-rearranged lung carcinomas. Am J Surg Pathol 2011;35:1226–34. [19] Jokoji R, Yamasaki T, Minami S, Komuta K, Sakamaki Y, Takeuchi K, et al. Combination of morphological feature analysis and immunohistochemistry is useful for screening of EML4-ALK-positive lung adenocarcinoma. J Clin Pathol 2010;63:1066–70.