Article Tools
ORIGINAL REPORTS
Rapid Communications
Article Tools
Next-Generation Sequencing of Pulmonary Sarcomatoid Carcinoma Reveals High Frequency of Actionable MET Gene Mutations
B.H. and A.C.B. contributed equally to this work.
Abstract
To further understand the molecular pathogenesis of pulmonary sarcomatoid carcinoma (PSC) and develop new therapeutic strategies in this treatment-refractory disease.
Whole-exome sequencing in a discovery set (n = 10) as well as targeted MET mutation screening in an independent validation set (n = 26) of PSC were performed. Reverse transcriptase polymerase chain reaction and Western blotting were performed to validate MET exon 14 skipping. Functional studies for validation of the oncogenic roles of MET exon 14 skipping were conducted in lung adenosquamous cell line H596 (MET exon 14 skipped and PIK3CA mutated) and gastric adenocarcinoma cell line Hs746T (MET exon 14 skipped). Response to MET inhibitor therapy with crizotinib in a patient with advanced PSC and MET exon 14 skipping was evaluated to assess clinical translatability.
In addition to confirming mutations in known cancer-associated genes (TP53, KRAS, PIK3CA, MET, NOTCH, STK11, and RB1), several novel mutations in additional genes, including RASA1, CDH4, CDH7, LAMB4, SCAF1, and LMTK2, were identified and validated. MET mutations leading to exon 14 skipping were identified in eight (22%) of 36 patient cases; one of these tumors also harbored a concurrent PIK3CA mutation. Short interfering RNA silencing of MET and MET inhibition with crizotinib showed marked effects on cell viability and decrease in downstream AKT and mitogen-activated protein kinase activation in Hs746T and H596 cells. Concurrent PIK3CA mutation required addition of a second agent for successful pathway suppression and cell viability effect. Dramatic response to crizotinib was noted in a patient with advanced chemotherapy-refractory PSC carrying a MET exon 14 skipping mutation.
Pulmonary sarcomatoid carcinoma (PSC) is a recognized category of highly aggressive and poorly differentiated non–small-cell lung carcinoma (NSCLC), with five different subtypes: pleomorphic, spindle, giant cell, carcinosarcoma, and pulmonary blastoma. Although uncommon (0.1% to 0.4% of all pulmonary malignancies1,2), their clinical importance is underscored by poorer prognosis2-4 and higher rate of resistance to conventional chemotherapy5,6 than other NSCLCs. This leaves few treatment options for patients with advanced PSC.
Limited studies based on specimen numbers and limited scope of molecular analysis have so far failed to identify actionable gene alterations with any significant frequency in PSC.7-14 A larger-scale and comprehensive study of molecular alterations in this unique lung cancer subtype is needed to better understand biology and guide therapy. Therefore, we performed whole-exome sequencing (WES) on a large and well-characterized collection of PSCs, identified and validated known and novel somatic alterations, and, in doing so, uncovered an expanded spectrum of functionally significant MET mutations in PSC. Not only are these mutations potentially targetable in this treatment-resistant disease, but they also may explain the mechanisms of the unique dual-differentiation pattern in this tumor type and could serve as novel biomarkers to define subsets of sarcomatoid and other carcinomas for MET targeting.
This study was reviewed and approved by the ethics committee of Columbia University. WES was performed on DNA isolated from fresh frozen tumor and corresponding normal tissue in the discovery set. Subsequently, mutated genes that fulfilled appropriate selection criteria were chosen for ranking and further analysis. A parallel sequencing study using the TruSeq Amplicon–Cancer 48 gene panel (Illumina, San Diego, CA) was performed on DNA from formalin-fixed paraffin-embedded (FFPE) tissue of the 10 patient cases in the discovery set to confirm mutations in known oncogenes and tumor suppressor genes (TSGs). Seven of the top genes on the ranking list, as well as key cancer genes that had not been confirmed in the TruSeq Amplicon–Cancer 48 gene panel, were then chosen for Sanger sequencing confirmation. Five genes, including CDH4, CDH7, SCAF1, LMTK2, and MET, were further sequenced in the independent validation set of 26 PSC DNA samples. Lastly, validation of MET exon 14 skipping at the RNA and protein levels was performed in patient cases of PSC for which frozen tumor tissues were available; functional studies in two cell-line models with MET exon 14 skipping (H596 and Hs746T) were also performed.
From the pathology database spanning a period from 1997 to 2013, 41 patient cases of PSC (excluding carcinosarcoma and blastoma) were reviewed according to WHO criteria1 (A.C.B.) and staged according to American Joint Committee staging manual (seventh edition) criteria.15
Sanger sequencing for EGFR, KRAS, and BRAF was performed, as was anaplastic lymphoma kinase (ALK) immunohistochemistry (D5F3 clone) or ALK fluorescent in situ hybridization using the Vysis break-apart probe (Abbott Molecular, Abbott Park, IL). Five of 41 patient cases were excluded from genomic studies because of prior neoadjuvant therapy (n = 2) or lack of quality tumor or normal DNA sample (n = 3). The selected group of 36 PSCs was divided into two sets: 10 PSCs with both fresh frozen tumor and normal tissue and FFPE tissue were assigned to the discovery set, with the remaining 26 PSCs assigned to the validation set.
H596 and Hs746T cell lines were obtained from the American Type Culture Collection (Manassas, VA). The MET inhibitor crizotinib and the PIK3CA inhibitor GDC0941 were purchased from Selleck Chemicals (Houston, TX). Recombinant human hepatocyte growth factor was purchased from Life Technologies (Carlsbad, CA).
For all frozen tissue, cresyl violet (0.3% in 70% ethanol) –stained sections were needle microdissected with microscope guidance after review of hematoxylin and eosin–stained cryostat sections. DNA extraction was performed using the DNeasy blood and tissue kit (Qiagen, Valencia, CA). RNA extraction was performed using the RNeasy blood and tissue kit with on-column DNase digestion (Qiagen) and evaluation using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). For paraffin tissue, the QIAamp DNA FFPE tissue kit was used (Qiagen).
WES at an average of 120× coverage for tumor and 60× coverage for normal samples was performed at the Columbia Genome Center. Genomic DNA was fragmented and exomes captured using the Agilent SureSelect Human All Exon kit. The captured DNA was sequenced with 100-bp paired-end reads on an Illumina HiSeq 2000 system. Sequencing reads were aligned to the human assembly hg19 using BWA software (version 0.5.9; http://bio-bwa.sourceforge.net). GATK software (version 1.0; UnifiedGenotyper tool; https://www.broadinstitute.org/gatk) was used to refine local alignment of reads and recalibrate base quality score within targeted regions. Samtools software (version 0.1.17; http://www.htslib.org) was used to call somatic mutations from paired tumor–normal samples. In addition to using the default filters in Samtools (bcftools pair mode), variants were further filtered for genotype minimum quality of 30, minimum quality > depth of five, minimum strand bias of −0.10, and maximum fraction of reads with mapping quality of zero at 10%. Annotated variants were subsequently filtered to exclude variants ≥ 1% of minor allele frequency based on dbSNP135 and the 1000 Genomes Project. Variants predicted to be pathogenic by either the SIFT (http://sift.jcvi.org) or PolyPhen2 tool (http://genetics.bwh.harvard.edu/pph2) were further analyzed.
The strategy for the selection of significantly mutated genes is shown in Appendix Figure A1 (online only). Genes at the top of the ranking list were then chosen for validation and further study (Data Supplement). In a second analysis, all nonsynonymous somatic variants were analyzed using CRAVAT software (http://www.cravat.us) and its CHASM analysis16,17 and filtered for known oncogenes and TSGs. In addition, these CHASM-derived lists were processed using the PROVEAN algorithm,18 providing the PROVEAN and SIFT predictions shown in Appendix Figure A2 (online only).
Extracted DNA was used for the TruSeq Amplicon–Cancer 48 gene panel (Illumina) 212-amplicon multiplexed targeting resequencing assay using the MiSeq system. FASTQ files were imported, aligned, and analyzed for variants using NextGENE software (Softgenetics, State College, PA).
The primer sets for sequencing MET exon 14 and flanking introns were as follows: forward, 5′-TGTCGTCGATTCTTGTGTGC-3′; reverse, 5′-AGTTGGGCTTACACTTCGGG-3′; and forward, 5′-CGCTACGATGCAAGAGTACAC-3′; reverse, 5′-TCAAATACTTACTTGGCAGAGGT-3′. Primer sequences for validation and mutation screening of other genes are available on request. All polymerase chain reaction (PCR) products were submitted for bidirectional Sanger sequencing. Observed variations were then verified to be of somatic origin using paired normal tissue DNA.
Reverse transcriptase PCR (RT-PCR) was performed using a one-step RT-PCR system purchased from Life Technologies. The primers for RT-PCR in validation of MET exon 14 skipping were as follows: forward, 5′-TCAACCGTCCTTGGAAAAGT-3′; reverse, 5′-AGCACTGAGGTCAATGTGGA-3′. For patient cases suspected to have exon 14 skipping, relevant RT-PCR products were cut from the gel, extracted using a gel extraction kit (Qiagen), and then submitted for bidirectional Sanger sequencing.
Cell-viability assays were performed as described previously.19 The combined effect of drugs was evaluated by MTS assay at a one-to-one ratio of each drug. CIs were calculated using both the Bliss model20 and CalcuSyn software (Biosoft, Cambridge, United Kingdom). CI values < one, equal to one, and > one indicated synergism, additive effect, and antagonism, respectively. Similar CI trends were observed with both methods.
Immunoblotting was performed as described previously.19 Antibodies against total MET and total AKT were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies to total ERK1/2, phosphorylated MET, phosphorylated AKT, phosphorylated ERK1/2, and glyceraldehyde 3-phosphate dehydrogenase were obtained from Cell Signaling Technology (Danvers, MA).
Knockdown of MET was performed using Silencer Select (Life Technologies) validated small interfering RNA (siRNA) targeting MET. Introduction of siRNA was performed with the use of Dharmafect 1 transfection reagent (GE Dharmacon, Lafayette, CO).
For clinicopathologic variables and MET mutations, 2 × 2 tables were analyzed using Fisher's exact test. MTS assay data are expressed as mean ± standard deviation from at least triplicates. Analysis of difference between treated and control groups was performed by paired-sample t test using SPSS software (version 13.0; SPSS, Chicago, IL). Differences were considered significant at P < .01.
We analyzed clinical parameters and tumoral features from 41 patients with PSC, summarized in Table 1 and detailed in the Data Supplement. A majority of patients with PSC were current or former smokers, with 50% considered heavy smokers. Average tumor size (4.65 cm) and frequent pleural invasion (72%) were reflected in the stage groupings; frequent vascular invasion was also a typical feature of PSC biologic behavior. KRAS mutation was identified in six tumors (15%; 21% of 29 adenocarcinomas), and one BRAF mutation was seen; no EGFR mutations or ALK translocations were identified. This is in keeping with the high smoking rate.
|
| Characteristic | N = 41 |
|---|---|
| Age, years | |
| Median | 69.3 |
| Range | 38-87 |
| Sex, No. | |
| Male | 20 |
| Female | 21 |
| Tumor size, cm (n = 37) | |
| Average | 4.65 |
| Range | 1.3-10.2 |
| Pleural invasion, % | 72 |
| Vascular invasion, % | 75 |
| Stage, % | |
| Ia | 14.5 |
| Ib | 14.5 |
| IIa | 20 |
| IIb | 22 |
| IIIa | 24 |
| IIIb | 0 |
| IV | 5 |
| Smoking status, % (n = 38) | |
| Current | 28 |
| Former | 64 |
| Never | 8 |
| Pack-years, % (n = 34) | |
| 0 | 8.8 |
| 0.1-20 (light) | 41.2 |
| > 20 | 50 |
| Mutation status, % | |
| KRAS (n = 40) | 15 |
| EGFR (n = 40) | 0 |
| ALK (n = 38) | 0 |
| BRAF (n = 31) | 3 |
The mean somatic mutation rate was 3.44 per megabase of DNA (range, 1.01 to 7.03). Overall, 1,461 somatic mutations, including 850 missense (58.2%), 76 nonsense (5.2%), four stop loss (0.3%), and 26 splice-site mutations (1.8%), 78 deletions (5.3%), 33 insertions (2.3%), and 394 silent mutations (27.0%) were identified (Appendix Fig A3, online only; Data Supplement).
After selection of mutated genes with a score > 3 (using ranking algorithm; Appendix Fig A1; Data Supplement), 37 mutated genes were analyzed using the CRAVAT algorithm and CHASM driver analysis. These genes were divided into known oncogenes and TSGs (n = 12) and other genes with somatic mutations (n = 25), as summarized in Appendix Figure A2.
PROVEAN analysis was performed, and alterations in eight known oncogenes or TSGs and 17 other mutated genes were assessed as detrimental or deleterious mutations. According to the ranking list and PROVEAN analysis results, nine known oncogenes or TSGs and six other mutated genes were included for validation (highlighted in yellow in Appendix Fig A2). The Illumina TruSeq panel was performed for the known nine oncogenes or TSGs; Sanger sequencing was performed for selected other mutated genes and oncogenes or TSGs that had not been validated by the TruSeq panel. TP53 (six of 10), PIK3CA (two of 10), MET (two of 10), KRAS (two of 10), STK11 (one of 10), SMARCA4 (one of 10), RB1 (one of 10), NOTCH1 (one of 10), and MLL2 (one of 10), as well as six novel mutated genes, including RASA1 (two of 10), CDH4 (two of 10), CDH7 (two of 10), LAMB4 (three of 10), SCAF1 (two of 10), and LMTK2 (two of 10), were validated by TruSeq panel and/or Sanger sequencing (Data Supplement).
Four novel mutated genes with a relatively short coding sequence, including CDH4, CDH7, LMTK2, and SCAF1, and a known oncogene (MET) were chosen for targeted mutation screening in an independent validation set of 26 PSC tumor DNA samples. One additional tumor with a somatic mutation of CDH4 was identified (Data Supplement); however, no additional LMTK2, CDH7, or SCAF1 somatic mutations were seen.
Because the two MET mutations identified by WES were located in the splice sites of MET exon 14, we bidirectionally sequenced MET exon 14 and its flanking introns in all 36 patient cases. Six additional MET exon 14 splice-site mutations were detected and confirmed as somatic. The frequency of MET exon 14 splice-site mutations was 22.2% (eight of 36). Of these eight mutations, three were 5′ splice-site deletions, five were 3′ donor splice-site point mutations (Table 2; Fig 1A; Appendix Fig A4, online only), and four were c.G3028C/p.D1010_splice mutations in the 3′ donor splice site (previously reported). Previously reported23,24 MET exon 14 splice-site mutations (intron 14 + 1G>T) in pulmonary adenosquamous carcinoma (H596) and gastric adenocarcinoma cell lines (Hs746T) were also confirmed (Appendix Fig A5, online only).
|
| Patient | Mutation Type | Somatic or Germline | Sanger Sequencing | WES | TruSeq | Previous Reports |
|---|---|---|---|---|---|---|
| 2 | Intron 13 −(1-15) deletion and c.A2888del | Somatic | Yes | Yes | No | No |
| 5 | Intron 13 −(1-28) deletion | Somatic | Yes | No | No | No |
| 5 | c.T2998A, p.S1000T | Somatic | Yes | Yes | Yes | No |
| 6 | c.G3028C, p.D1010_splice | Somatic | Yes | NA | NA | Yes21 |
| 16 | c.G3028C, p.D1010_splice | Somatic | Yes | Yes | Yes | Yes |
| 22 | c.T2889C, synonymous | Somatic | Yes | NA | NA | No |
| 31 | c.G3028C, p.D1010_splice | Somatic | Yes | NA | NA | Yes |
| 32 | c.G3028C, p.D1010_splice | Somatic | Yes | NA | NA | Yes |
| 38 | Intron 13 −(12-22) deletion | Somatic | Yes | NA | NA | No |
| 39 | Intron 14 + 3 A>G | Somatic | Yes | NA | NA | No |
| 4 | c.C2975T, p.T992I | Germline | Yes | No | NA | rs56391007 |
| 19 | c.C2975T, p.T992I | Germline | Yes | NA | NA | rs56391007 |
| 25 | Intron 13 −(52-53) insertion CT | Germline | Yes | No | No | Yes22 |
Abbreviations: NA, not applicable; PSC, pulmonary sarcomatoid carcinoma; WES, whole-exome sequencing.
*Isoform b: NM_000245.2; NP_000236.2; CCDS: CCDS43636.1.
Fig 1. Identification of tumor-specific mutations in MET leading to exon 14 skipping. (A) Schematic diagram showing position of nucleic acid deletions (red dashes) and point mutations (red arrowheads) in splice-site junctions of MET exon 14 (based on RefSeq NM_000245.2). Three patients with exon 14 5′ splice-site deletions and five patients with 3′ splice-site point mutations were identified. (B) Real-time polymerase chain reaction (RT-PCR) using paired frozen tumor–normal tissue RNA proved presence of 141-bp MET exon 14 skipped RNA variant in five tumors (2T, 5T, 6T, 16T, 32T) with splice-site mutations, as well as two cell lines—H596 and Hs746T—with previously reported MET skipping. (C) Western blots show smaller MET precursor and β subunit proteins in four tumors (2T, 5T, 16T, 32T) and two cell lines (H596, Hs746T) with splice-site mutations. Protein lysates from tumors (3T, 4T) and cell lines (HCC827, H23, Calu-3, A549) without MET exon 14 splice-site mutations, as well as normal tissues (4N), were used as negative controls. (D) Representative sequencing histograms of patient 16 showing c.G3028C/p. D1010_MET point mutation in genomic DNA and MET exon 14 skipping in complementary DNA amplified by RT-PCR in tumor tissue, but not in adjacent normal tissue.
RT-PCR on paired tumor–normal total RNA isolated from fresh frozen tissue showed an additional smaller RT-PCR (141-bp smaller) product in PSC tumors and two cell lines with known MET exon 14 splice-site mutations (Fig 1B). Western blotting of paired tumor–normal protein extracts from PSC showed smaller MET protein in tumors and cell lines with MET exon 14 splice-site mutations but not in tumors without MET mutations or in normal controls (Fig 1C). MET exon 14 skipping by bidirectional Sanger sequencing was confirmed in complementary DNA synthesized by RT-PCR (Fig 1D).
The relationship between MET mutations and clinical and pathologic parameters is shown in Appendix Figure A6 (online only). MET exon 14 skipping mutations were found to be mutually exclusive with mutations in KRAS, EGFR, BRAF, and ALK. One tumor (patient case 2) harbored concurrent MET exon 14 skipping and PIK3CA somatic mutations. MET exon 14 skipping mutations were only found in PSCs with an epithelial component of adenocarcinoma (seven [87.5%] of eight) or PSCs without an epithelial component (one [12.5%] of eight). Incidence of MET exon 14 skipping in PSCs with acinar adenocarcinoma was significantly higher than in PSCs without acinar adenocarcinoma (P = .005).
We used the H596 and Hs746T cell lines as powerful in vitro tools for studies of the functional role of MET exon 14 skipping. Two additional cell lines—Calu-3 and HCC827—with high expression of wild-type MET but no exon 14 skipping events (Fig 1C; Appendix Fig A5) were used as negative controls. Treatment of cells with the MET/ALK/ROS inhibitor crizotinib decreased cell proliferation and inhibited downstream AKT as well as mitogen-activated protein kinase activation in Hs746T cells, whereas effects of MET inhibition on cell proliferation and downstream protein activation were modest in H596 cells and negligible in Calu-3 and HCC827 cells (Figs 2A and 2C). Ablation of MET signaling with MET siRNA showed effects on cell proliferation and downstream protein activation similar to those of crizotinib (Figs 2B and 2D), demonstrating specificity of these findings.
Fig 2. Abrogation of MET signaling decreased cell proliferation and inhibited downstream protein activation in MET exon 14 skipped cells. (A, B) Cell viability assay results showing that (A) MET inhibitor crizotinib and (B) MET small interfering RNA (siRNA) treatment decreased cell proliferation in MET exon 14 skipped Hs746T cells, but not in either H596 cells harboring concurrent MET exon 14 skipping and PIK3CA mutations or Calu-3 and HCC827 cells with high expression of wild-type MET but no exon 14 skipping. Data are from three independent experiments and expressed as percent of viability of vehicle-treated cells. Results shown as mean ± SEM. (*) Indicates significant differences were observed as compared with control groups treated with nontarget siRNA (P < .01). (C, D) Western blots show ablation of MET signaling by (C) crizotinib and (D) MET siRNA blocked phosphorylation of downstream proteins AKT and ERK1/2 in Hs746T cells, whereas effects were modest in H596 cells and negligible in Calu-3 and HCC827 cells. Data are representative of three independent experiments.
Given the limited effect of MET ablation on cell proliferation and downstream protein activation and the previous report of a concurrent oncogenic PIK3CA mutation (p.E545K) in H596 cells,25 we hypothesized that combined ablation of MET and PIK3CA signaling might be needed to block downstream signaling and produce growth inhibition. This might have clinical relevance, because PIK3CA mutations frequently co-occur with other oncogenic driver alterations (eg, EGFR mutations driving resistance of upstream signaling inhibition),26,27 and patient case 2 similarly had concurrent MET and PIK3CA mutations. GDC0941 is a potent inhibitor of PI3Kα/δ with an IC50 of 3 nmol/L. We treated H596 cells using GDC0941 alone and along with crizotinib or MET siRNA. As shown in Figure 3, diminished MET signaling with crizotinib or MET siRNA and concomitant inhibition of PI3K activation with GDC0941 showed synergy in preventing H596 cellular proliferation and downstream AKT phosphorylation.
Fig 3. Concomitant inhibition of MET and PIK3CA signaling pathways is necessary for cells with concurrent MET and PIK3CA mutations. (A) MET inhibitor crizotinib and PI3K kinase inhibitor GDC0941 synergistically inhibited cell proliferation in H596 cells with concurrent MET exon 14 skipping and PIK3CA mutations, as measured by MTS assay. H596 cells were treated with crizotinib, GDC0941, or combination of crizotinib with GDC0941 at one-to-one ratio for 72 hours. CI values were used to measure combined effects. (*) Indicates CI < 1, consistent with synergistic effect. (B) Western blots show crizotinib and GDC0941 synergistically blocked downstream AKT phosphorylation in H596 cells. H596 cells were serum starved overnight and then treated with indicated doses of GDC0941 in presence or absence of crizotinib 1 μmol. Data are representative of three independent experiments. (C) Cell viability assay results showing H596 cells were more sensitive to PI3K kinase inhibitor GDC0941 in presence of MET small interfering RNA (siRNA). H596 cells were treated with indicated dose of GDC0941 in presence of nontarget siRNA 10 nmol (Ctl siRNA) or MET siRNA 10 nmol for 5 days. Results are shown as mean ± SEM. (*) Indicates significant differences were observed as compared with control groups treated with nontarget siRNA (P < .01). (D) GDC0941and MET siRNA synergistically blocked downstream AKT phosphorylation in H596 cells. H596 cells were transfected with nontarget siRNA 10 nmol or MET siRNA 72 hours before GDC0941 treatment. Cells were then starved overnight and treated with indicated dose of GDC0941 for 3 hours. Data are representative of three independent experiments.
A 74-year-old woman (patient case 39) with past medical history of smoking and asbestos exposure and stage II PSC showed radiographic progression during neoadjuvant platinum/taxane chemotherapy, in keeping with aggressive histology and treatment resistance. Widespread recurrent disease with lung, liver, and bulky mesenteric disease was noted 3 months after resection. Further chemotherapy was not advised because of resistance and poor overall patient condition. Expanded molecular testing revealed MET amplification (nine copies), with a concurrent splice-site mutation consistent with exon 14 skipping (84% of sequences were mutated, suggesting preferential amplification). On the basis of the presence of MET amplification, crizotinib 250 mg orally twice daily was started, with rapid dramatic clinical improvement, later confirmed as an excellent radiographic partial response by computed tomography scanning (Fig 4).
Fig 4. Representative pre- (upper row) and postcrizotinib (lower row) treatment computed tomography (CT) images in 74-year-old woman with pulmonary sarcomatoid carcinoma harboring MET exon 14 skipping. After 2 months of crizotinib treatment, extensive metastatic lesions involving right lower lung (white arrow), liver (white arrowheads), and mesentery in lower abdomen (long white arrows) became markedly reduced compared with those in precrizotinib treatment CT images (note lack of oral contrast in postcrizotinib images).
Using WES followed by a careful ranking and validation strategy, we define a broad mutation spectrum for PSC, including several known oncogenes and TSGs (TP53, PIK3CA, MET, KRAS, STK11, SMARCA4, RB1, NOTCH1, and MLL2), as well as six novel recurrent mutated genes, including RASA1, CDH4, CDH7, LAMB4, SCAF1, and LMTK2. Subsequent targeted mutation screening demonstrated a strikingly high frequency of splice-site somatic mutations leading to MET exon 14 skipping, mutually exclusive with known driver mutations. Additional functional studies indicated that inhibition of MET-driven oncogenic pathways shows promise as a biomarker-driven targeted approach for PSC therapy. Dramatic response to the MET inhibitor crizotinib in a woman with chemotherapy-refractory and rapidly progressive advanced PSC with concurrent MET splice-site mutation and amplification of the mutated allele provides clinical validation and rapid translatability of our findings.
Marked advances have been made in the development of targeted agents for the treatment of molecularly defined subsets of NSCLC, with dramatic efficacy of epidermal growth factor receptor (EGFR) and ALK tyrosine kinase inhibitors in EGFR- and ALK-mutated lung adenocarcinomas.28 However, the frequency of EGFR mutations in PSC is controversial. Italiano et al7 and Pelosi et al9 found no EGFR mutations in two separate cohorts, including 22 and 20 European patients with PSC, respectively. On the contrary, Leone et al8 reported that two of 22 patients with PSC harbored EGFR exon 19 deletions, and Asian series have reported up to 20% incidence of EGFR mutations.11,12 Despite these rates, however, the presence of EGFR-activating mutations in PSC in the Asian series did not predict response to gefitinib.14,29 In accordance with the results in European patients, no EGFR mutations were found in our series. Therefore, whether because of low incidence of mutation or lack of clinical response, the clinical benefit from anti-EGFR drugs in PSC is expected to be limited. ALK inhibition in PSC also seems to be an ineffective strategy, with no ALK mutations or rearrangements reported. The lack of EGFR and ALK mutations in our series is in line with the high frequency of smoking in patients with PSC. In our study, KRAS mutations were found in 15% (six of 40) of PSCs, and 21% of those with an adenocarcinoma component, a rate that is similar to that in adenocarcinomas overall.30,31 Our study strongly supports KRAS mutation testing in patients with PSC to guide eligibility for therapeutic trials.
MET splice-site mutations have been previously shown to lead to exon 14 skipping,23,24 and loss of the juxtamembrane c-Cbl binding site was hypothesized as oncogenic through enhanced activation of MET phosphorylation by decrease of MET protein ubiquitination and degradation.24,32 Several large series defining the genomic landscape of lung adenocarcinoma identified MET exon 14 skipping mutations in 3% to 4% of tumors.21,33-36 One series of 77 sarcomatoid carcinomas identified MET exon 14 mutations in only 3% of cases,10 whereas our series confirmed MET exon 14 somatic mutations in PSC at a much higher rate (22%). The reason for this difference in rate mainly reflects the limited methodologic sensitivity of their assay; Vieira et al10 focused only on deletions in the 3′ splice site of MET exon 14 rather than on all variations, including deletions and single-nucleotide variants in both 3′ and 5′ splice sites. A lower proportion of sarcomatoid carcinomas with an adenocarcinoma component (30%) in their series (as compared with 68% in our series) may also contribute. The high incidence of oncogenic MET mutations in our series of PSC as compared with typical adenocarcinomas indicates that MET-dependent signaling pathways contribute to highly aggressive and treatment-recalcitrant PSCs, and this will need to be further confirmed through expression studies of such MET variants. The mutual exclusivity of MET exon 14 skipping with known driver mutations also strengthens the confidence of our hypothesis that MET exon 14 skipping is a driver oncogenic mutation in PSC. In addition, the role of the hepatocyte growth factor/MET pathway in mesenchymal growth as well as epithelial–mesenchymal transition37,38 makes its frequent mutational activation in PSC particularly intriguing. If MET exon 14 skipping mutations are critical to the pathogenesis of tumors with combined carcinomatous and sarcomatous components, it will also be imminently relevant to address the frequency of analogous MET alterations in sarcomatoid carcinomas affecting a wide variety of organs.
Two cell lines, an NSCLC cell line (H596)24 and a gastric adenocarcinoma cell line (Hs746T),23 were previously reported to exhibit MET exon 14 skipping. Both genetic ablation and pharmacologic inhibition significantly decreased cell proliferation and inhibited downstream signaling activation in MET-mutated Hs746T cells, but not in cells overexpressing wild-type MET. This further indicates that MET exon 14 skipping is a driver oncogenic event and drug-sensitive mutation. Dramatic response to the MET inhibitor crizotinib in a patient with chemotherapy-refractory and widely metastatic PSC harboring MET amplification/exon 14 skipping provides powerful clinical correlation to our in vitro findings and suggests oncogene addiction in a treatment-refractory smoking-related tumor analogous to those responses noted in lung adenocarcinomas with EGFR mutation and ALK translocation. A recent case report of an excellent clinical response to the MET inhibitor crizotinib in a patient with lung adenocarcinoma harboring a MET exon 14 splice-site mutation further corroborates the clinical relevance of our findings.39
However, the effect of deactivation of MET signaling showed limited effects in H596 cells, which harbor a concurrent oncogenic PIK3CA E545K point mutation. Similarly, in our study cohort, we noted a PSC tumor harboring concurrent MET exon 14 skipping and PIK3CA somatic mutations. Therefore, we hypothesized that concurrent MET and PIK3CA mutations synergistically drive the oncogenic behavior of these tumors, and combined ablation of MET and PIK3CA signaling might be needed. In other settings, such as EGFR-mutated NSCLC, it is known that PI3K mutations can co-occur, and in such cases, EGFR inhibition is not sufficient.40 In line with our expectations, combined treatment with the MET inhibitor crizotinib or MET siRNA and PI3K inhibitor indeed synergistically decreased H596 cell proliferation and downstream signaling activation. These findings suggest that although MET exon 14 skipping is oncogenic and a potential target in PSC, concurrent PIK3CA mutation requires addition of a second agent for successful pathway suppression. This calls for concurrent testing of both biomarkers to maximize efficacy of further translational development.
Another intriguing observation is the preponderance of MET exon 14 skipping transcript and protein in our mutated tumor samples, with relative paucity of wild-type message and protein. Relative absence of wild-type MET exon 14 message was also seen in the RNASeq data reported by Seo et al34 and Dhanasekaran et al.33 Although amplification of MET could explain this finding, a majority of reported cases of MET exon 14 skipping do not harbor MET amplification.33 In addition, allele frequency in our WES patient cases (12% and 31%) did not support amplification of mutated or loss of wild-type allele. The current hypothesis regarding loss of MET protein degradation is also insufficient to explain these findings. Whether preponderance of mutated protein is necessary for crizotinib treatment response is unknown, as is the mechanism of wild-type MET transcript and protein loss, both of which are topics for future study.
PSC is a rare disease, and as a result, small sample size is a limitation of our study. A lack of formal confirmation of the oncogenic role of MET exon 14 skipping using appropriate expression constructs in noncancer cell lines is another limitation. Given that MET is a known oncogene, and MET exon 14 skipping has previously been reported as an oncogenic event in lung cancer, such confirmation seems likely. In the index patient case, crizotinib was chosen, because extended mutation testing demonstrated MET amplification in addition to the exon skipping mutation. Preliminary results for crizotinib in MET-amplified adenocarcinoma41 showed four partial responses among 12 treated patients. Although it is not possible to separate the role of MET amplification versus exon 14 skipping in our patient, the preferential amplification of the mutated allele, along with the rapid and dramatic response, certainly argues for the utility of MET inhibition in this setting. In summary, our study has identified that mutational events of MET leading to exon 14 skipping are frequent and potentially readily targetable events in PSC.
See accompanying article on page 879
Processed as a Rapid Communication manuscript.
Supported by Uniting Against Lung Cancer and Lungevity (B.H.).
Authors' disclosures of potential conflicts of interest are found in the article online at www.jco.org. Author contributions are found at the end of this article.
Conception and design: Yuxia Jia, Balazs Halmos, Alain C. Borczuk
Financial support: Balazs Halmos, Alain C. Borczuk
Administrative support: Balazs Halmos, Alain C. Borczuk
Provision of study materials or patients: Mark B. Stoopler, Balazs Halmos, Alain C. Borczuk
Collection and assembly of data: Xuewen Liu, Yuxia Jia, Haiying Cheng, Jinli Chen, Sanjay Koul, Balazs Halmos, Alain C. Borczuk
Data analysis and interpretation: Xuewen Liu, Yuxia Jia, Mark B. Stoopler, Yufeng Shen, Haiying Cheng, Mahesh Mansukhani, Balazs Halmos, Alain C. Borczuk
Manuscript writing: All authors
Final approval of manuscript: All authors
| 1. | Pathology and Genetics of Tumours the Lung, Pleura, Thymus and Heart 341,2004 Lyon, France IARC Press Google Scholar |
| 2. | S Yendamuri, L Caty, M Pine, etal: Outcomes of sarcomatoid carcinoma of the lung: A Surveillance, Epidemiology, and End Results database analysis Surgery 152:397–402,2012 Crossref, Medline, Google Scholar |
| 3. | LW Martin, AM Correa, NG Ordonez, etal: Sarcomatoid carcinoma of the lung: A predictor of poor prognosis Ann Thorac Surg 84:973–980,2007 Crossref, Medline, Google Scholar |
| 4. | T Mochizuki, G Ishii, K Nagai, etal: Pleomorphic carcinoma of the lung: Clinicopathologic characteristics of 70 cases Am J Surg Pathol 32:1727–1735,2008 Crossref, Medline, Google Scholar |
| 5. | T Vieira, N Girard, M Ung, etal: Efficacy of first-line chemotherapy in patients with advanced lung sarcomatoid carcinoma J Thorac Oncol 8:1574–1577,2013 Crossref, Medline, Google Scholar |
| 6. | HM Bae, HS Min, SH Lee, etal: Palliative chemotherapy for pulmonary pleomorphic carcinoma Lung Cancer 58:112–115,2007 Crossref, Medline, Google Scholar |
| 7. | A Italiano, AB Cortot, M Ilie, etal: EGFR and KRAS status of primary sarcomatoid carcinomas of the lung: Implications for anti-EGFR treatment of a rare lung malignancy Int J Cancer 125:2479–2482,2009 Crossref, Medline, Google Scholar |
| 8. | A Leone, P Graziano, R Gasbarra, etal: Identification of EGFR mutations in lung sarcomatoid carcinoma Int J Cancer 128:732–735,2011 author reply 736 Crossref, Medline, Google Scholar |
| 9. | G Pelosi, P Gasparini, A Cavazza, etal: Multiparametric molecular characterization of pulmonary sarcomatoid carcinoma reveals a nonrandom amplification of anaplastic lymphoma kinase (ALK) gene Lung Cancer 77:507–514,2012 Crossref, Medline, Google Scholar |
| 10. | T Vieira, M Antoine, AM Ruppert, etal: Blood vessel invasion is a major feature and a factor of poor prognosis in sarcomatoid carcinoma of the lung Lung Cancer 85:276–281,2014 Crossref, Medline, Google Scholar |
| 11. | K Kaira, Y Horie, E Ayabe, etal: Pulmonary pleomorphic carcinoma: A clinicopathological study including EGFR mutation analysis J Thorac Oncol 5:460–465,2010 Crossref, Medline, Google Scholar |
| 12. | YL Chang, CT Wu, JY Shih, etal: EGFR and p53 status of pulmonary pleomorphic carcinoma: Implications for EGFR tyrosine kinase inhibitors therapy of an aggressive lung malignancy Ann Surg Oncol 18:2952–2960,2011 Crossref, Medline, Google Scholar |
| 13. | S Lee, Y Kim, JM Sun, etal: Molecular profiles of EGFR, K-ras, c-met, and FGFR in pulmonary pleomorphic carcinoma, a rare lung malignancy J Cancer Res Clin Oncol 137:1203–1211,2011 Crossref, Medline, Google Scholar |
| 14. | A Ushiki, T Koizumi, N Kobayashi, etal: Genetic heterogeneity of EGFR mutation in pleomorphic carcinoma of the lung: Response to gefitinib and clinical outcome Jpn J Clin Oncol 39:267–270,2009 Crossref, Medline, Google Scholar |
| 15. | P Goldstraw, J Crowley, K Chansky, etal: The IASLC Lung Cancer Staging Project: Proposals for the revision of the TNM stage groupings in the forthcoming (seventh) edition of the TNM classification of malignant tumours J Thorac Oncol 2:706–714,2007 Crossref, Medline, Google Scholar |
| 16. | C Douville, H Carter, R Kim, etal: CRAVAT: Cancer-related analysis of variants toolkit Bioinformatics 29:647–648,2013 Crossref, Medline, Google Scholar |
| 17. | WC Wong, D Kim, H Carter, etal: CHASM and SNVBox: Toolkit for detecting biologically important single nucleotide mutations in cancer Bioinformatics 27:2147–2148,2011 Crossref, Medline, Google Scholar |
| 18. | Y Choi, GE Sims, S Murphy, etal: Predicting the functional effect of amino acid substitutions and indels PLoS One 7:e46688,2012 Crossref, Medline, Google Scholar |
| 19. | Z Zhang, JC Lee, L Lin, etal: Activation of the AXL kinase causes resistance to EGFR-targeted therapy in lung cancer Nat Genet 44:852–860,2012 Crossref, Medline, Google Scholar |
| 20. | H Peng, J Wen, H Li, etal: Drug inhibition profile prediction for NFkappaB pathway in multiple myeloma PLoS One 6:e14750,2011 Crossref, Medline, Google Scholar |
| 21. | Comprehensive molecular profiling of lung adenocarcinoma Nature 511:543–550,2014 Cancer Genome Atlas Research Network Crossref, Medline, Google Scholar |
| 22. | PC Ma, T Kijima, G Maulik, etal: C-MET mutational analysis in small cell lung cancer: Novel juxtamembrane domain mutations regulating cytoskeletal functions Cancer Res 63:6272–6281,2003 Medline, Google Scholar |
| 23. | Y Asaoka, M Tada, T Ikenoue, etal: Gastric cancer cell line Hs746T harbors a splice site mutation of c-Met causing juxtamembrane domain deletion Biochem Biophys Res Commun 394:1042–1046,2010 Crossref, Medline, Google Scholar |
| 24. | M Kong-Beltran, S Seshagiri, J Zha, etal: Somatic mutations lead to an oncogenic deletion of met in lung cancer Cancer Res 66:283–289,2006 Crossref, Medline, Google Scholar |
| 25. | JM Spoerke, C O'Brien, L Huw, etal: Phosphoinositide 3-kinase (PI3K) pathway alterations are associated with histologic subtypes and are predictive of sensitivity to PI3K inhibitors in lung cancer preclinical models Clin Cancer Res 18:6771–6783,2012 Crossref, Medline, Google Scholar |
| 26. | JE Chaft, ME Arcila, PK Paik, etal: Coexistence of PIK3CA and other oncogene mutations in lung adenocarcinoma-rationale for comprehensive mutation profiling Mol Cancer Ther 11:485–491,2012 Crossref, Medline, Google Scholar |
| 27. | LV Sequist, BA Waltman, D Dias-Santagata, etal: Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors Sci Transl Med 3:75ra26,2011 Crossref, Medline, Google Scholar |
| 28. | MG Kris, BE Johnson, LD Berry, etal: Using multiplexed assays of oncogenic drivers in lung cancers to select targeted drugs JAMA 311:1998–2006,2014 Crossref, Medline, Google Scholar |
| 29. | S Ichihara, S Toyooka, Y Fujiwara, etal: The impact of epidermal growth factor receptor gene status on gefitinib-treated Japanese patients with non-small-cell lung cancer Int J Cancer 120:1239–1247,2007 Crossref, Medline, Google Scholar |
| 30. | W Pao, TY Wang, GJ Riely, etal: KRAS mutations and primary resistance of lung adenocarcinomas to gefitinib or erlotinib PLoS Med 2:e17,2005 Crossref, Medline, Google Scholar |
| 31. | H Linardou, IJ Dahabreh, D Kanaloupiti, etal: Assessment of somatic k-RAS mutations as a mechanism associated with resistance to EGFR-targeted agents: A systematic review and meta-analysis of studies in advanced non-small-cell lung cancer and metastatic colorectal cancer Lancet Oncol 9:962–972,2008 Crossref, Medline, Google Scholar |
| 32. | J Lefebvre, F Ancot, C Leroy, etal: Met degradation: More than one stone to shoot a receptor down FASEB J 26:1387–1399,2012 Crossref, Medline, Google Scholar |
| 33. | SM Dhanasekaran, OA Balbin, G Chen, etal: Transcriptome meta-analysis of lung cancer reveals recurrent aberrations in NRG1 and hippo pathway genes Nat Commun 5:5893,2014 Crossref, Medline, Google Scholar |
| 34. | JS Seo, YS Ju, WC Lee, etal: The transcriptional landscape and mutational profile of lung adenocarcinoma Genome Res 22:2109–2119,2012 Crossref, Medline, Google Scholar |
| 35. | K Okuda, H Sasaki, H Yukiue, etal: Met gene copy number predicts the prognosis for completely resected non-small cell lung cancer Cancer Sci 99:2280–2285,2008 Crossref, Medline, Google Scholar |
| 36. | R Onozato, T Kosaka, H Kuwano, etal: Activation of MET by gene amplification or by splice mutations deleting the juxtamembrane domain in primary resected lung cancers J Thorac Oncol 4:5–11,2009 Crossref, Medline, Google Scholar |
| 37. | KM Weidner, N Arakaki, G Hartmann, etal: Evidence for the identity of human scatter factor and human hepatocyte growth factor Proc Natl Acad Sci U S A 88:7001–7005,1991 Crossref, Medline, Google Scholar |
| 38. | M Stoker, E Gherardi, M Perryman, etal: Scatter factor is a fibroblast-derived modulator of epithelial cell mobility Nature 327:239–242,1987 Crossref, Medline, Google Scholar |
| 39. | RW Jenkins, GR Oxnard, S Elkin, etal: Response to crizotinib in a patient with lung adenocarcinoma harboring a MET splice site mutation Clin Lung Cancer [epub ahead of print on February 7, 2015] Google Scholar |
| 40. | L Wang, H Hu, Y Pan, etal: PIK3CA mutations frequently coexist with EGFR/KRAS mutations in non-small cell lung cancer and suggest poor prognosis in EGFR/KRAS wildtype subgroup PLoS One 9:e88291,2014 Crossref, Medline, Google Scholar |
| 41. | DR Camidge, S-HI Ou, G Shapiro, etal: Efficacy and safety of crizotinib in patients with advanced c-MET-amplified non-small cell lung cancer (NSCLC) J Clin Oncol 32:506s,2014 (suppl 15s; abstr 8001) Link, Google Scholar |
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or jco.ascopubs.org/site/ifc.
No relationship to disclose
No relationship to disclose
Research Funding: Pfizer (Inst), AstraZeneca (Inst), Merck (Inst), Boehringer Ingleheim (Inst), Roche (Inst)
No relationship to disclose
No relationship to disclose
No relationship to disclose
No relationship to disclose
No relationship to disclose
Research Funding: Pfizer, AstraZeneca, Merck, Boehringer Ingelheim, Roche
No relationship to disclose
Fig A1. Detailed strategy for selection of significantly mutated genes. COSMIC, catalogue of somatic mutations in cancer.
Fig A2. Mutation summary in discovery set. Ten tumor samples are shown in conjunction with 12 known oncogene and tumor suppressor genes (TSGs) and 25 nonsynonymous somatic mutations identified by whole-exome sequencing and scoring algorithm ≥ 3. X denotes mutation in that sample; red indicates deleterious effect of mutation by PROVEAN; blue indicates neutral or tolerated prediction; uncolored X boxes are alterations without predicted effect; genes in yellow had mutations confirmed by either TruSeq cancer panel or Sanger sequencing. (*) Subsequently identified.
Fig A3. Summary of mutations identified by whole-exome sequencing (WES) in discovery set. (A) Total mutations per megabase (MB) and mutations subtypes identified by WES in each tumor of discovery set. (B) Percentages of different mutation subtypes in all mutations identified in 10 tumors of discovery set.
Fig A4. Sanger sequencing histograms of all MET mutations leading to exon 14 skipping. (A) Bidirectional Sanger sequencing in genomic DNA showing 5′ splice-site 16-bp (TCTCTCTGTTTTAAGA) deletion in tumor but not in adjacent normal tissue of patient 2. (B) Bidirectional Sanger sequencing in genomic DNA showing 5′ splice donor site 28-bp (TTTAACAAGCTCTTTCTTTCTCTCTGTT) deletion in tumor but not in adjacent normal tissue of patient 5. (C) Bidirectional Sanger sequencing in genomic DNA showing 3′ splice acceptor site c.G3028C p.D1010_splice point mutation in tumor but not in adjacent normal tissue of patients 6, 16, 31, and 32. (D) Bidirectional Sanger sequencing in genomic DNA showing 5′ splice donor site 11-bp (CTTTCTTTCTTTTCTC) deletion in tumor but not in adjacent normal tissue of patient 38. (E) Bidirectional Sanger sequencing in genomic DNA showing intron 14 + 3 A>G point mutation in tumor (red arrows) but not in adjacent normal tissue of patient 39.
Fig A5. Sanger sequencing histograms of exon 14 skipped MET in H596 and Hs746T cell lines and wild-type MET in Calu-3 and HCC827 cell lines. (A) Exon 14 skipped MET sequencing histograms of H596 and Hs746T cells. Top histograms show intron 14 + 1 G>T point mutation was identified in genomic DNA isolated from two cell lines via bidirectional Sanger sequencing (red arrows). Bidirectional Sanger sequencing in complementary DNA obtained by real-time polymerase chain reaction (RT-PCR) on RNA shows presence of MET exon 14 skipping in H596 and Hs746T cells (bottom histograms). (B) Bidirectional Sanger sequencing histograms of complementary DNA obtained by RT-PCR on RNA show absence of MET exon 14 skipping in Calu-3 and HCC827 cells.
Fig A6. MET mutations and clinicopathologic parameters. Patient cases with MET mutation were analyzed in comparison with mutational status, smoking status, and histopathologic parameters, including adenocarcinoma subtypes. Data are shown only for mutated patient cases. Fisher's exact test comparing mutated and nonmutated groups was performed. (*) P < .01. ADSQ, adenosquamous; LEP, lepidic; MP, micropapillary; PAP, papillary.
