MET nucleotide variations and amplification in advanced ovarian cancer: characteristics and outcomes with c-Met inhibitors.

PURPOSE
MET alterations including amplifications and nucleotide variations have been associated with resistance to therapy and aggressive clinical behavior.


EXPERIMENTAL DESIGN
The medical records of patients presenting to the University of Texas MD Anderson Cancer Center Phase I Clinic with relapsed or metastatic ovarian cancers and known MET nucleotide variation or amplification status were reviewed retrospectively (n=178). Categorical and continuous clinical and molecular characteristics were compared using Fisher's exact and Wilcoxon rank-sum tests, respectively. Univariate and multivariate survival were assessed via Kaplan-Meier and Cox regression analysis, respectively.


RESULTS
MET amplification occurred in 4 (3.5%) of 113 patients, whereas nonsynonomous nucleotide variations were present in 9 (7.4%) of 122 patients. No patients exhibited concomitant amplification and variation. MET variations were observed only in white women with high-grade ovarian tumors, whereas amplifications were observed in both black and white women with high-grade serous ovarian primary tumors. No patients (n=4) exhibiting a MET alteration achieved an objective response when treated on a c-Met inhibitor phase I trial. In addition, ovarian cancer patients treated with a c-Met inhibitor with multikinase activity trended towards a longer time-to-failure compared with those treated with a c-Met-specific inhibitor (median: 1.5 vs. 4.5 months, p=0.07).


CONCLUSIONS
MET alterations occur in a minority of patients with ovarian cancer. c-Met inhibitors with multikinase activity may exhibit less activity in ovarian cancer than c-Met specific drugs. These findings warrant further investigation.


INTRODUCTION
The c-Met tyrosine kinase receptor (TKR), upon activation by its cognate antigen, hepatocyte growth factor (HGF), generates proliferation, migration, and survival signals in numerous cancers [1,2]. This signaling cascade parallels, and at times can supplement, the activity of other oncogenic TKRs. As a result, the c-Met pathway has emerged as a resistance pathway in therapies targeting the epidermal growth factor receptor (EGFR) [3,4], B-Raf [5], and vascular endothelial growth factor receptor (VEGFR) Oncoscience 6 www.impactjournals.com/oncoscience [6], among others. Various MET alterations, including amplification and nucleotide variations, have been described and are associated with resistance to therapy and aggressive clinical behavior [3,4,7]. The pathologic implications of this important receptor has prompted the development of c-Met inhibitors, many of which are currently undergoing early phase trials in various cancers [2].
Ovarian, primary peritoneal, and fallopian tube cancers have similar ontological origins and clinical presentations. Their aggressive metastatic behavior and generally poor prognosis has prompted interest in developing therapies with TKR inhibitors, including those targeting EGFR and c-Met [8,9]. However, outside of bevacizumab, targeted therapies tested in early clinical trials have yet to gain widespread clinical success [8]. Despite understanding of the intrinsic biology of the c-Met pathway and its documented role in drug resistance, no substantive clinical series have assessed the effect of MET variations and amplifications in this disease [10][11][12]. We, therefore, investigated the clinical and molecular characteristics of patients with ovarian cancers referred to our Phase I Clinical Trials Program and their response to treatment on a phase I c-Met inhibitor trial.

Patient Characteristics
One-hundred-and-seventy-eight patients met study inclusion criteria, of whom 122 and 113 were tested for MET variations and amplification, respectively. Fifty-seven patients were tested for both variation and amplification. MET nucleotide variations were detected in 9 patients (7.4%): 6 with N375S and 3 with T1010I nonsynonomous variations. MET amplification was detected in 4 patients (3.5%), amplification gene copy numbers (in relation to CEP7) were 2.12, 2.27, 2.55, and 2.78. No patients exhibited concomitant MET nucleotide variation and amplification.

Characteristics associated with MET aberrations
No significant differences were noted between patient characteristics when stratified by MET variation or amplification status (all p>0.05). MET variations were detected only in white women with high-grade primary ovarian tumors (Tables 1 and 2). The histology of cancers with MET variations was predominately serous (74%), with one patient each having carcinosarcoma and clear cell carcinoma (Table 1). The median numbers of metastatic sites among patients with MET variations compared with those without were both 2. Among patients with MET variations, 44% had liver metastasis, a rate similar to liver metastasis in patients without MET variation (34%). Concomitant mutations included KRAS, BRCA1, ARID1A, and TP53 were identified in 1, 1, 1, and 2 patients, respectively. In addition, one patient had concomitant PTEN loss and another had weak PTEN staining (Table  2). Of note, no ovarian cancer patients within any stratum exhibited ALK rearrangement, BRAF, EGFR, or KIT mutations.
MET amplification occurred only in women with high grade serous ovarian cancer (Tables 1 and 2). Three out of 4 women were white and 1 was black. Similar to patients with MET variations, MET amplified patients had a median of 2 metastatic sites, with 1 patient exhibiting liver metastasis. No concomitant mutations were observed in MET amplified patients except for in 2 patients who exhibited loss of PTEN protein expression.

MET alterations and survival
OS in patients with a MET aberration (either variation or amplification, n=13) was compared with that of patients known to be negative for any aberrations (n=50). There was no significant different in median survival in patients exhibiting MET alterations ( Fig. 1

Treatment with c-Met inhibitors
All patients treated on a Phase I c-Met inhibitor clinical trial had a diagnosis of ovarian cancer. This included 4 patients with MET nucleotide variation and one with MET amplification (Table 3). No patients with a MET alteration achieved an objective response.
Among all patients with ovarian cancer treated on a c-Met inhibitor trial, 5 out of 18 (28%) exhibited a partial response (PR) or stable disease (SD) lasting ≥6 months. Of these, 2 of 18 patients (11%) achieved a best response of PR (Fig. 2, Table 4). Prolonged SD lasting ≥6 months was achieved in 3 patients with TTFs lasting 6, 7.8, and 29.8 months. The median TTF in patients treated with a c-Metspecific inhibitor (1.5 months, range 0.4-7.8 months, n=9) was less than patients treated with a multikinase inhibitor (n=8) or a multikinase inhibitor combined with a VEGFR2 inhibitor (n=1)(4.5 months, range 1.2-29.8 months). This difference trended towards significance (p=0.07). Interestingly, among 3 patients with known TP53 mutations treated on a c-Met inhibitor trial, the 2 patients treated with a c-Met inhibitor with multikinase activity exhibited objective responses, while the one Oncoscience 7 www.impactjournals.com/oncoscience Abbreviations: (Q3-Q1) = 75th to 25th percentile, AM-indian = American indian, NOS = not otherwise specified, Met = metastatic, CNS = central nervous system.

Figure 1: Kaplan-Meier plot of overall survival in ovarian cancer patients with MET variation or amplification (dashed-black line) compared with patients without MET variation or amplification (solid-gray line).
Oncoscience 8 www.impactjournals.com/oncoscience    Oncoscience 10 www.impactjournals.com/oncoscience patient who was treated with a c-Met specific inhibitor did not (Table 4).

DISCUSSION
The literature delineating the prevalence of MET amplification and variations in ovarian cancer is sparse. Yamato et al. reported on 5 patients, all with clear cell histology, identified with MET amplification out of 195 (2.5%) patients tested [16]. An earlier report using Southern blot analysis in 67 patients with ovarian cancer found no amplifications [12]. Two other series with 24 and 65 ovarian cancer patients found the incidence of MET variations to be 1 and 0, respectively [10,11]. The rates in our current cohort (nucleotide variations, 7.4%; and amplification, 3.5%) are comparable, albeit higher, than those reported in previous studies. A possible explanation is that our study population consisted entirely of patients with metastatic or relapsed disease, which may skew in favor of increased rates of MET aberrations. Of note, we did not detect any patients with concomitant MET amplification and variation.
The biologic activity of the MET nucleotide variations identified here and even whether these specific variations represent somatic mutations or germline polymorphisms have yet to be fully elucidated [17,18]. The role of MET amplification has been most extensively correlated with tumor invasion and aggressive metastatic behavior in gastrointestinal malignancies [19,20]. With regard to MET variations, the N375S nonsynonmous variation occurs in the extracellular semaphorin domain, whereas the activating T1010I variation occurs in the juxtamembrane domain [7,18]. The characteristics and clinical behavior of ovarian cancer in patients with MET variations have not been thoroughly described. However, data in renal cell and lung carcinomas suggest that somatic and germline MET polymorphisms may enhance c-Met TKR activity and even confer inhibitor resistance [17,21]. This second observation is corroborated by our data, in which no patients exhibiting a MET alteration achieved an objective response on a c-Met phase I inhibitor trial. Another possible explanation for the lack of an observed objective response are that most patients in this study were enrolled on a Phase I dose escalation trials and may not have received an efficacious study drug dose. Few preclinical data have tested the efficacy of these inhibitors in MET altered patients [22][23][24] and to our knowledge no associated clinical data have been reported.
A subset of all ovarian cancer patients achieved objective responses. Patients treated with a multikinase inhibitor trended toward a longer TTF compared with those treated with c-Met-specific inhibitors (median 1.5 vs. 4.5 months, p=0.07). A possible explanation is that activity against VEGFR2, a target with known clinical efficacy in ovarian cancer [14], may increase therapeutic efficacy. Interestingly 2 out of 3 patients exhibiting TP53 mutations exhibited objective responses to c-Met inhibitors and were both treated using a multikinase inhibitor. As such, observed responses may be due to effects on targets other than c-Met, including VEGFR2. This hypothesis is supported by recent evidence from our group showing that TP53 mutations predict responses to anti-VEGFR2 therapies [25]. In addition, these data also suggests that c-Met inhibition may be more clinically efficacious when utilized in a supporting role to block resistance pathways in other targeted agents rather than being utilized alone [1,2,4]. These data are preliminary findings that warrant further investigation in a randomized clinical trial setting.
We observed no difference in survival in patients with a MET alteration versus those without, although in multivariate analysis patients with MET alterations trending towards a worse OS (HR=1.8, p=0.13). However it is possible that separate analyses of MET variations and amplifications may yield different results. Due to the limited incidence, we did not observe any significant differences in the characteristics of patients with MET alterations. In addition, a number of concomitant mutations were identified in patients harboring MET variations, whereas there were none in patients with MET amplification other than PTEN loss.
Outside of limitations inherent to all retrospective reviews, the low frequency of MET variations and amplifications observed in this study did not provide sufficient statistical power to stringently assess demographic and patient characteristic differences. MET variation testing was also done with some heterogeneity. Although the majority of patients were tested at MD Anderson core laboratories, a minority was tested with outside platforms. In addition, mutations and amplifications were often grouped within a single stratum. With regard to amplifications, the relatively low gene copy numbers found in this study (range 2.12-2.78) may be inadequate to exhibit biological differences. Finally, the few number of patients, especially MET altered patients, treated on c-Met inhibitor phase I trials makes generalizing these observations into a larger patient populations difficult. However, despite these limitations, to our knowledge this is one of the first studies to substantially compare the demographic and clinical characteristics of ovarian cancer patients with MET altered patients and their response to treatment with c-Met inhibitors. The findings of this analysis provide insight into the clinical characteristics associated with MET alterations and, if validated in the prospective setting, may lead to improved strategies utilizing c-Met therapies in this prognostically poor patient population. www.impactjournals.com/oncoscience

METHODS Patients
We retrospectively reviewed the medical records of all consecutive patients presenting to our Phase I Clinic starting in May 2010 to November 2012 (n=3607). Patients presented for treatment of recurrent or metastatic disease, usually after multiple prior treatments (range 0-13). Eligibility criteria for inclusion in this study were a histologic diagnosis of ovarian, fallopian tube, or primary peritoneal carcinoma and testing for MET nucleotide variations and/or amplification (n=178). This study was approved by the MD Anderson Institutional Review Board and patient confidentiality was maintained following Health Insurance Portability and Accountability Act guidelines.

Tissue samples and molecular analysis
MET nucleotide variation and amplification status were tested using archival formalin-fixed, paraffinembedded tissue blocks in addition to biologic material from fine needle aspiration biopsies or excised primary or metastatic tumors from diagnostic and/or therapeutic procedures. Histology was centrally reviewed in the MD Anderson Department of Pathology. MET nucleotide variations were assessed in several Clinical Laboratory Improvement Amendment-certified laboratories using a single test or as part of a gene panel. The majority of patients (n=160) were analyzed at the MD Anderson core laboratories. Of the remaining samples, 15 were tested at Knight Diagnostics (Portland, OR), 2 at the Baylor College of Medicine (Houston, TX), and 1 using a Foundation One platform (Foundation Medicine, Cambridge, MA) as previously described [13].
MET amplification status was analyzed via fluorescence in situ hybridization (FISH) following institutional protocols at MD Anderson. Copy numbers were expressed as a gene copy number in relation to CEP7, a gene located near the centrosome of the same chromosome. Gene copy numbers of 2-fold or greater were considered amplified.

Treatment and evaluation
All patients were enrolled when possible in a Phase I clinical trial(s) judged to be clinically appropriate by the patient's attending physician. Because confidentiality agreements with the providing pharmaceutical company stipulate that c-Met inhibitor identities cannot be disclosed, c-Met agents were classified as being in one of three categories: c-Met-specific inhibitor (3 separate trials), c-Met inhibitor with multikinase activity (2 separate trials), and c-Met inhibitor with multikinase kinase activity in combination with a VEGFR2 kinase inhibitor (1 trial). All c-Met inhibitors with multikinase activity also targeted the VEGFR2 receptor, a molecule associated with known clinical efficacy in ovarian cancer [14]. Patients were treated until clinical or radiologic disease progression, development of unacceptable toxicities or death, clinical judgment necessitating patient removal, or withdrawal of patient consent.
Clinical assessments were performed according to the specific requirements of individual protocols, typically once prior to the initiation of treatment and then at least every treatment cycle. Treatment responses were primarily assessed using computed tomography scans, magnetic resonance imaging, and/or positron emission tomography conducted prior to therapy and every 2 cycles thereafter (6-8 weeks). Radiographs were read in the Department of Radiology at MD Anderson and reviewed by physicians in our Phase I Clinical Trials Program. Objective responses were determined according to RECSIT 1.0 or 1.1 criteria, as specified by individual protocols, and the best responses achieved were recorded [15].

Statistical Methods
Patient demographics and tumor molecular and histologic characteristics were summarized in relation to MET nucleotide variation and amplification status. Time to treatment failure (TTF) was defined as the time from the initiation of therapy to its cessation for any reason. Categorical and continuous variables, including TTF and stratification by MET alteration were compared utilizing Fisher's exact and Wilcoxon rank-sum tests, respectively. Overall survival (OS) was assessed starting from the date of the first appointment in the Phase I Clinic using Kaplan-Meier analysis with comparisons via the partial likelihood ratio test. Multivariate and univariate hazard ratios (HR) were calculated via Cox regression. Wald p-values were reported for multivariate analyses. All tests were two-sided when appropriate and considered significant at p<0.05. Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).