|Home | About | Journals | Submit | Contact Us | Français|
Lung cancer is the leading cause of cancer-related deaths across the world. In this study, we present therapeutically relevant genetic alterations in lung adenocarcinoma of Indian origin.
Forty-five primary lung adenocarcinoma tumors were sequenced for 676 amplicons using RainDance cancer panel at an average coverage of 1500×(reads per million mapped reads). To validate the findings, 49 mutations across 23 genes were genotyped in an additional set of 363 primary lung adenocarcinoma tumors using mass spectrometry. NIH/3T3 cells over expressing mutant and wild-type FGFR3 constructs were characterized for anchorage independent growth, constitutive activation, tumor formation and sensitivity to FGFR inhibitors using in vitro and xenograft mouse models.
We present the first spectrum of actionable alterations in lung adenocarcinoma tumors of Indian origin, and shows that mutations of FGFR3 are present in 20 of 363 (5.5%) patients. These FGFR3 mutations are constitutively active and oncogenic when ectopically expressed in NIH/3T3 cells and using a xenograft model in NOD/SCID mice. Inhibition of FGFR3 kinase activity inhibits transformation of NIH/3T3 overexpressing FGFR3 constructs and growth of tumors driven by FGFR3 in the xenograft models. The reduction in tumor size in the mouse is paralleled by a reduction in the amounts of phospho-ERK, validating the in vitro findings. Interestingly, the FGFR3 mutations are significantly higher in a proportion of younger patients and show a trend toward better overall survival, compared with patients lacking actionable alterations or those harboring KRAS mutations.
We present the first actionable mutation spectrum in Indian lung cancer genome. These findings implicate FGFR3 as a novel therapeutic in lung adenocarcinoma.
Lung cancer is the leading cause of cancer-related deaths worldwide, accounting for over a million deaths annually . Molecularly targeted therapies improve outcome for lung adenocarcinoma patients whose tumors harbor mutant EGFR or translocated ALK, RET or ROS1, with an encouraging response for those with mutated BRAF, MET, NTRK-1 & 2 and ERBB2 [2–5]. Such oncogenic somatic alterations though vary across populations/ethnic groups, e.g. EGFR mutations are present in over 30% of East Asian lung adenocarcinoma patients, however, they are only found in ~23%–25% of Indian and 10% of Western lung adenocarcinoma patients [6–8]. Similarly, KRAS mutations are present at 60% lower frequency in Indian lung adenocarcinoma patients than compared with the Caucasian population [3, 9, 10]. The diversity in somatic alterations lends similarity to the known plurality in clinical response based on ethnicity and divergent genetic and environmental factors , Thus, besides the unmet need for additional therapeutic targets in lung adenocarcinoma patients, it is equally pertinent to profile known oncogenic somatic alterations across different populations to understand their landscape of variability.
Here, in an attempt to profile for activating alterations, we have generated a comprehensive mutational spectrum of activating alterations prevalent among lung adenocarcinoma patients of Indian origin, considered to be an admixture population of non-European Caucasian and Ancestral South Indians. We also report the first incidence of activating and drug sensitive FGFR3 mutations in lung adenocarcinoma. FGFR3 mutated samples, with ~5% population frequency, form a distinct subclass apart from EGFR, KRAS and EML4-ALK.
To profile for therapeutically relevant genome alterations in lung adenocarcinoma of Indian origin, FFPE blocks with known EGFR mutation status for 45 consecutive histologically confirmed lung adenocarcinoma patients tumor samples (stage IV, 49% males and 51% non-smokers) for sequencing and an additional set of 363 consecutive lung adenocarcinoma patients tumor samples (stage IV, 62% males and 73% non-smokers) for mass spectrometry were retrospectively collected from Tata Memorial Hospital (supplementary Table S1, available at Annals of Oncology online).
A set of 45 lung adenocarcinoma samples were sequenced using pooled sequencing approach to capture low-frequency variants [12–14]. Briefly, 45 samples were divided into duplicate pools of different population sizes (supplementary Figure S1, available at Annals of Oncology online), i.e. 2 pools of 5 individuals (5XA and 5XB), 2 pools of 10 individuals (10XA and 10XB) and 1 pool of 15 individuals (15X) for next-generation sequencing (NGS) of 676 genomic regions of 158 genes as described earlier .
FASTQ files were analyzed using BWA-Picard-GATK/MuTect pipeline generating 3349 unique variants (supplementary Table S2, available at Annals of Oncology online). Polymorphisms overlapping with dbSNP database (v.142) and Indian specific SNP database TMC-SNPdb derived from whole exome sequencing of 62 normal samples  were filtered (supplementary Figures S2 and S3, available at Annals of Oncology online). Stringent mutation analysis was carried out as further detailed in supplementary methods, available at Annals of Oncology online to derive list of significant mutations for further validation (supplementary Tables S2 and S3, available at Annals of Oncology online).
Briefly, PCR and extension primers for 49 mutations in 23 genes were designed using single base extension based mass spectrometry assay design 3.1 software (supplementary Table S4, available at Annals of Oncology online). Mutation calls were analyzed using Typer 4 (Sequenom, Inc., USA) and were reviewed by manually observing mass spectra.
NIH/3T3 cells transduced with FGFR3 wild-type and mutant construct were used for induction and drug inhibition study as detailed in supplementary methods, available at Annals of Oncology online. Anchorage independent growth assay and immunoblotting were carried out as described earlier , and as detailed in the supplementary methods, available at Annals of Oncology online.
A cohort of eight NOD-SCID or nude mice per clone were subcutaneously injected with five million cells for tumor formation in 2–3 months. Inhibitor BGJ-398  was given at 15 and 30mg/kg along with vehicle control (10% Tween-80) independently to randomized xenograft groups after tumor size reaching ~150mm3. Tumor size was measured every alternate day using a Vernier caliper during the 14 day period of drug treatment. microPET-CT scan was carried out at the end of the drug treatment.
Overall survival (OS) of patients was assessed using Kaplan–Meier method using R and IBM SPSS software package, as detailed in supplementary methods, available at Annals of Oncology online. The end point was taken as date of death with censoring implied at the date of the last contact.
To identify low frequency and ethnic-specific therapeutically relevant genome alterations in lung adenocarcinoma of Indian origin, we sequenced 45 primary lung adenocarcinoma stage IV tumors (supplementary Table S1, available at Annals of Oncology online) for 676 amplicons at an average coverage of 1500× (reads per million mapped reads), as described in supplementary Figures S1–S5 and Tables S2, S3, available at Annals of Oncology online. To validate the findings, we selected 49 mutations occurring across 23 genes based on their recurrence and therapeutic significance (supplementary Table S4, available at Annals of Oncology online), for genotyping in an additional set of 363 primary lung adenocarcinoma stage IV tumors (Figure (Figure1A;1A; supplementary Table S5, available at Annals of Oncology online) using mass spectrometry.
Based on the mutation profiling of 363 lung adenocarcinoma patients, we present the first portrait of activating mutations present in the Indian lung cancer genome (Figure (Figure1B),1B), wherein 160 of 363 patients were found to harbor activating mutations across 8 genes at following frequency: EGFR (28.4%), KRAS (13%), ALK (3.8%), AKT1 (2.5%), PIK3CA (1.4%), FGFR4 (0.4%) and ERBB2 (0.3%) as shown in Figure Figure1A,1A, consistent with earlier reports [6, 8, 9]. In addition, 3 of 79 patients were found to harbor EML4-ALK translocation as determined by FISH. Among the other most significantly mutated genes, we found recurrent FGFR3 mutations in 20 of 363 tumors (5.5%), of which 7 co-occurred in samples harboring EGFR (n=5) and KRAS (n=2) mutations. In total, 16 patients harbored FGFR3 (S249C) mutation; and 4 patients harbored a novel FGFR3 (G691R) mutation (Figure (Figure1A1A and C, upper panel; supplementary Figure S6 and Table S6, available at Annals of Oncology online). Interestingly, FGFR3 (S249C) mutation has previously been described as activating and drug sensitive in lung squamous , while the novel FGFR3 (G691R) mutation was predicted to be deleterious based on using 7 of 10 functional prediction tools (supplementary Table S3, available at Annals of Oncology online).
To test whether the recurrent FGFR3 mutations found in this study are activating we transduced NIH/3T3 fibroblast cells with retroviruses encoding the FGFR3 G691R mutation along with WT FGFR3 and the previously characterized FGFR3 (R248C) and (S249C) mutations . Similar to FGFR3 R248C and S249C, the ectopic expression of the novel G691R mutant clone in pooled NIH/3T3 cells conferred anchorage-independent growth, forming threefold more colonies in soft agar than cells expressing WT FGFR3 (Figure (Figure1C,1C, left panel), despite higher expression levels of WT FGFR3 (Figure (Figure1C,1C, right panel). The transformation was accompanied by elevated phosphorylation of the downstream ERK1/2 and AKT1 in a constitutive manner (Figure (Figure1D).1D). Furthermore, consistent with the in vitro data, NIH/3T3 cells expressing transforming FGFR3 mutations or WT when injected subcutaneously into NOD/SCID mice formed tumors within 2 months post injection of cells. While 3 of 11 mice injected with cells expressing FGFR3 WT formed tumors, 12 of 12 mice injected with cells expressing FGFR3 S249C; and 6 of 12 mice injected with cells expressing FGFR3 G691R formed tumors (Figure (Figure1E).1E). The tumor size doubling time was ~7 days for cells expressing FGFR3 (G691R), ~5 days for cells expressing FGFR3 (S249C); the FGFR3-WT tumors doubled in size in ~9–10 days.
We further show that inhibition of FGFR3 kinase activity using pan FGFR inhibitor PD173074 block the constitutive phosphorylation of ERK1/2 (Figure (Figure2A).2A). Similarly, the treatment of cells harboring activating FGFR3 mutations with PD173074 result in a marked decrease in colony formation in soft agar and cell survival in liquid culture (Figure (Figure2B).2B). Extending the effect in vivo studies, when tumors reached ~100–200mm3 in all mice injected with NIH/3T3 cells began treatment with 15 or 30mg/kg BGJ398—a selective FGFR inhibitor currently under clinical trials for various cancer types (as PD173074 is incompatible with in vivo studies ), or vehicle for 14 days. Tumors treated with BGJ398 slowed or reversed their growth compared with vehicle (Figure (Figure2C,2C, upper panel), so that by the end of the study, the effect on tumor burden in vehicle-treated versus BGJ398-treated mice were 3.3-folds in FGFR3 (S249C), three-fold in FGFR3 (G691R) and 2.25-fold in FGFR3-WT xenografts (Figure (Figure2D).2D). This reduction in tumor size was paralleled by a reduction in the amounts of phospho-ERK1/2 in immuno-histochemical analyses (Figure (Figure2C,2C, lower panel) of explanted tumors, validating our in vitro findings that MAPK signaling is the key pathway engaged by mutated FGFR3.
Clinically, lung adenocarcinoma patients with FGFR3 mutation positive tumors expressing higher activated MAPK levels (supplementary Figure S7, available at Annals of Oncology online) show a better trend in OS with 17 months (n=8; 95% CI: 6.4–27.5; HR: 0.6) compared with 14 months (n=197; 95% CI: 8.7–13.2) in patients with wild-type FGFR3 (Figure (Figure2E).2E). The OS trend in lung adenocarcinoma patients though is similar to bladder urothelial carcinomas and skin cutaneous melanoma patients, but not to head and neck cancer and lung squamous carcinoma patients, based on our analysis using cBioPortal for survival of patients harboring activating FGFR3 mutations in different cancers (supplementary Figure S8, available at Annals of Oncology online). Furthermore, the FGFR3 mutations were observed to be significantly higher in patients<45 years (9 of 95) than in patients>45 years (11 of 269) (P=0.048) but not with their gender and smoking status (supplementary Table S7, available at Annals of Oncology online). The sample size in this study, however, is underpowered to reach statistical significance for survival data.
We present the first portrait of clinically actionable alterations in lung adenocarcinoma of Indian origin that includes EGFR, KRAS, EML4-ALK, AKT1, PIK3CA, FGFR4 and ERBB2, similar to that identified in other ethnic groups [5, 22, 23], and an additional subset of patients with FGFR3 mutations. Ethnic-specific variations have been well known in lung cancer [24, 25] across different populations. We observed 28.4% EGFR mutations and 13% KRAS mutations in lung adenocarcinoma patients, consistent with our previous report [6, 9]. Similarly, variation in frequency of other molecular alterations is also observed such as 3% EML4-ALK alteration in our study compared with 8% in Caucasian population  and in 5% Chinese population . ERBB2 mutation found at<1% frequency in our cohort exists at ~2%–3% among the Caucasian  and Chinese populations . Similarly, AKT1 mutations were found at higher than the reported<1% in both Caucasian  and Chinese populations  indicating the higher therapeutic relevance of AKT1 targeted compounds in Indian population.
We have also identified frequent and recurrent drug sensitive FGFR3 mutations in lung adenocarcinoma patients. Among the Caucasians, activating mutations in FGFR3 have been earlier reported in bladder carcinoma , lung squamous cell carcinomas  and cervical cancer , but were found to be largely absent in lung adenocarcinomas [23, 28, 29], except for Imielinski et al. who reported non-recurrent somatic FGFR3 mutations of unknown functional significance in 3 of 183 lung adenocarcinoma patients . On the other hand, the presence of frequent FGFR3 mutations (with unknown driving potential) is tangentially referred to in the literature among Korean lung adenocarcinomas patients . Along with these reports, our finding of activating FGFR3 mutations in lung adenocarcinoma patients provides an interesting convergence with mouse genetic experiments wherein activated FGF9-FGFR3 signal acts as the primary oncogenic pathway involved in initiation of lung adenocarcinoma [31, 32].
Analyzing the potential effect of FGFR3 driver mutations on survival of lung cancer patients, we observed a trend toward better survival for FGFR3 mutations in lung adenocarcinoma, compared with lung adenocarcinoma patients with wild-type FGFR3 and those harboring KRAS mutation, similar to as reported in the bladder and skin cancer . Thus, FGFR3 mutation represents an opportunity for targeted therapy in lung adenocarcinoma. FGFR inhibitors, which are currently in clinical testing in tumor types bearing genetic alterations in FGFR genes [34, 35], may be extended to evaluate in patients with FGFR3-mutated lung adenocarcinoma. Finally, with a broader emerging role across different cancers [20, 36–39], this study further underscores that FGFR family may potentially join the EGFR family as a widespread target for therapeutic intervention in several human cancers.
All members of the Dutt laboratory for critically reviewing the manuscript. RainDance technologies, Inc. for providing NGS library preparation services. Sandor LifeSciences Pvt. Ltd for providing NGS services. AD is supported by an Intermediate Fellowship from the Wellcome Trust/DBT India Alliance (IA/I/11/2500278), by a grant from DBT (BT/PR2372/AGR/36/696/2011), from Terry Fox Foundation through TMC-Research Administrative Council (TRAC; project 108); and intramural grants (IRB project 92 and 55). PC is supported by a senior research fellowship from ACTREC. PU is supported by a senior research fellowship from CSIR. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
This work was supported by Wellcome Trust/Department of Biotechnology India Alliance [IA/I/11/2500278], Department of Biotechnology, Government of India [BT/PR2372/AGR/36/696/2011], and intramural grants [IRB project 92, 55 and 108].
The authors have declared no conflicts of interest.