Biomarker-Based Phase II Trial of Savolitinib in Patients With Advanced Papillary Renal Cell Cancer

Purpose Patients with advanced papillary renal cell carcinoma (PRCC) have limited therapeutic options. PRCC may involve activation of the MET pathway, for example, through gene amplification or mutations. Savolitinib (AZD6094, HMPL-504, volitinib) is a highly selective MET tyrosine kinase inhibitor. We report results of a single-arm, multicenter, phase II study evaluating the safety and efficacy of savolitinib in patients with PRCC according to MET status. Patients and Methods Patients with histologically confirmed locally advanced or metastatic PRCC were enrolled and received savolitinib 600 mg orally once daily. MET-driven PRCCwas defined as any of the following: chromosome 7 copy gain, focal MET or HGF gene amplification, or MET kinase domain mutations. Efficacy was assessed according to MET status. Safety, toxicity, and patient-reported health-related quality-of-life outcomes were assessed in all patients. Results Of 109 patients treated, PRCCwasMET driven in 44 (40%) andMET independent in 46 (42%); MET status was unknown in 19 (17%). MET-driven PRCC was strongly associated with response; there were eight confirmed partial responders with MET-driven disease (18%), but none with METindependent disease (P = .002). Median progression-free survival for patients with MET-driven and MET-independent PRCC was 6.2 months (95% CI, 4.1 to 7.0 months) and 1.4 months (95% CI, 1.4 to 2.7 months), respectively (hazard ratio, 0.33; 95% CI, 0.20 to 0.52; log-rank P , .001). The most frequent adverse events associated with savolitinib were nausea, fatigue, vomiting, and peripheral edema. Conclusion These data show activity and tolerability of savolitinib in the subgroup of patients with MET-driven PRCC. Furthermore, molecular characterization of MET status was more predictive of response to savolitinib than a classification based on pathology. These findings justify investigating savolitinib in MET-driven PRCC. J Clin Oncol 35. © 2017 by American Society of Clinical Oncology


INTRODUCTION
Renal cell carcinoma (RCC) is a heterogeneous disease comprising several histologic subtypes with different genetic and biochemical characteristics; clear cell RCC is the most frequent, accounting for 75% to 90% of renal malignancies. 12][3] In 2017, it is estimated 64,000 new cases of RCC will be diagnosed in the United States, equating to up to 6,400 cases of PRCC. 4 Somatic PRCC is conventionally classified into two histologic subtypes (type 1 and type 2), with a worse prognosis reported for type 2. [5][6][7][8][9] Molecular profiling of PRCC corroborates disease linkage demonstrated by studying rare hereditary syndromes that lead to RCC.MET mutations are associated with hereditary papillary renal carcinoma and phenocopy PRCC type 1 histologies, whereas fumarate hydratase mutations in hereditary leiomyomatosis RCC are associated with a subset of papillary type 2 histologies. 10olecular classification based on MET or hepatocyte growth factor (HGF) aberrations may provide more clinically relevant and reproducible indicators of tumor behavior than histologic subtype.Two recent cohorts of molecular analyses of PRCC have confirmed the utility of molecular features to reclassify this disease. 11,12ctivation of the HGF/MET axis triggers tumor growth, promotes angiogenesis, and induces metastasis. 13][19][20][21] Germline mutations in the MET gene, found on chromosome 7q31, were first described in patients with hereditary PRCC.In addition to MET mutations, [22][23][24] copy number gain of chromosome 7 (containing loci of both the MET receptor gene, MET, and its ligand, HGF) is common, occurring in 45% to 75% of sporadic PRCC cases, and copy number alterations of MET occur in 81% of type 1 and 46% of type 2 PRCCs. 11,12urrently, prognosis for patients with advanced PRCC is poor, because of the limited efficacy of currently available therapies, which were mainly developed for clear cell RCC.1][32][33] The activity of foretinib, a multikinase inhibitor targeting MET, vascular endothelial growth factor, RON, AXL, and TIE-2 receptors, against PRCC was assessed in a phase II study. 34umor responses were reported in five of 10 patients with germline MET mutations, compared with five of 57 patients without such a mutation, suggesting potential benefit for patients with METdriven PRCC.
Savolitinib (AZD6094, HMPL-504, volitinib) is a potent, selective MET inhibitor being investigated as a targeted therapy for patients with non-small-cell lung cancer in combination with osimertinib, 35 as well as for patients with advanced or metastatic PRCC.On the basis of preclinical data and a phase I study showing partial responses in three of six patients with PRCC, 36,37 a singlearm, multicenter, phase II study was conducted to evaluate the safety and efficacy of savolitinib in patients with PRCC and to correlate savolitinib activity with MET pathway alterations.

Patients
Eligible patients had histologically confirmed locally advanced or metastatic PRCC, with central and/or local laboratory-confirmed pathology on pretreatment archival tumor tissue.Papillary histology was independently reviewed by two pathologists; if a consensus was not reached, a third review was performed by a pathologist at an independent institute.Other inclusion criteria included: Eastern Cooperative Oncology Group performance status of 0 or 1, life expectancy $ 12 weeks, age $ 18 years, and adequate hematologic, hepatic, and renal function.Exclusion criteria included prior or current MET inhibitor treatment, at the discretion of the study monitor.One patient who had received foretinib 3.5 years earlier was included.The first patient was dosed on May 21, 2014.
All patients provided written informed consent.The study was performed in accordance with ethical principles originating in the Declaration of Helsinki and was consistent with International Conference on Harmonisation/Good Clinical Practice guidelines, applicable regulatory requirements, and the AstraZeneca policy on bioethics and human biologic samples.

Study Design and Objectives
This single-arm, multicenter, global, phase II study evaluated the safety and efficacy of savolitinib in patients with PRCC irrespective of prior treatment.The primary objective was to assess savolitinib antitumor activity in patients with PRCC and by MET status, as measured by investigator assessment of ORR according to Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1). 38Secondary objectives included change in target lesion tumor size from baseline and PFS (time from first dosing until objective disease progression or death resulting from any cause) in all patients and by MET status according to RECIST (version 1.1).

Study Drug Administration
Patients received savolitinib 600 mg orally once daily, until RECIST (version 1.1) -defined progression or treatment discontinuation criteria were met.A treatment cycle was defined as 21 days.

Study Assessments
Screening and baseline assessments were obtained # 28 days before the first savolitinib dose.After baseline evaluation, objective tumor assessments were performed every 6 weeks (6 7 days) for the first 12 months and every 12 weeks thereafter until disease progression.Response to treatment was assessed by ORR, stable disease, and PFS.Blood samples for pharmacokinetic, pharmocodynamic, germline DNA, and pharmacogenetic analyses were collected.
Patient-reported health-related quality-of-life outcomes and diseaserelated symptoms were collected at the start of the first three treatment cycles and then every 6 weeks, up to and including at discontinuation.The Functional Assessment of Cancer Therapy (FACT) -General, FACT Kidney Symptom Index-19 (FKSI-19), and European Quality of Life 5-Dimensions 5-Levels (EQ-5D-5L) were used.
The safety and tolerability of savolitinib were assessed according to Common Terminology Criteria for Adverse Events (version 4.03).Adverse events (AEs) and medical/surgical history were classified according to the Medical Dictionary for Regulatory Activities (version 18.1) and recorded from on or after the first dose until 30 days after the last dose of savolitinib.End of study for ORR was January 29, 2016; data cutoff for PFS was June 27, 2016.

Assessment of MET Pathway Status
Next-generation sequencing of archival tumor tissue was analyzed using a targeted 400-gene panel (version T7; Foundation Medicine, Cambridge, MA) as previously described. 39PRCC was confirmed as MET driven by identification of MET copy number gain (either chromosome 7 gain or a focal MET amplification of $ six copies), HGF gene amplification ($ six copies), or MET kinase domain mutations (allele frequency .5%).The Appendix (online only) provides more information on assessment of MET status.Focal MET amplifications were confirmed by fluorescent in situ hybridization.

Statistical Considerations
The trial size was designed to detect a response rate (ORR) of greater than 10% in patients with MET-driven disease while accounting for the estimated prevalence of this target population.A sample size of 50 patients with MET-driven PRCC allowed this signal detection at a 90% two-sided confidence level with at least 80% power assuming the true response rate was 25% or better.Analyses of outcome measures were descriptive, and tests for significant differences were conducted between patients with MET-driven and MET-independent PRCC.
PRCC was MET driven in 44 patients (40%) and MET independent in 46 (42%).MET status was unknown in 19 patients (17%).Baseline characteristics of patients with MET-driven and -independent disease were generally similar (Table 1).In 25 patients (23%), it was not possible to define the tumor as type 1 or 2, partly as a result of limited tissue available for central review, because typically only a single block from each patient case is provided.The proportions classified as type 1 and type 2 PRCC by central review differed according to MET status; among patients with MET-driven PRCC, 12 (27%) were classified as type 1 and 23 (52%) as type 2, whereas among patients with MET-independent  PRCC, the proportions were two (4%) and 37 patients (80%), respectively (Table 1).

Objective Disease Response
In the overall treatment population, the ORR was 7%.However, when assessed by MET status, the ORR was significantly higher in patients with MET-driven PRCC (eight partial responses [18%; PRs] among 44 patients) than with MET-independent PRCC (zero [0%] PRs among 46 patients; P = .002;Table 2).Categorization of MET-driven PRCC by histologic subtype showed two (17%) of 12 patients with type 1 and one (4%) of 23 patients with type 2 PRCC exhibited a PR.Five patients with a PR were not classified as having either type 1 or 2 disease by central review.

PFS and Duration of Response
In the treatment population, 82 patients (75%) experienced progression, died, or discontinued therapy, and 27 (25%) continued to receive study drug or remained in follow-up at end of study.As of the June 27, 2016, data cutoff, 41 patients had died and 19 were still receiving savolitinib.Disease progression occurred in 33 (75%), 44 (96%), and 14 patients (74%) with MET-driven, MET-independent, and MET-unknown PRCC, respectively.Patients with MET-driven PRCC had a significantly longer median PFS than those with MET-independent disease (6.2 months; 95% CI, 4.1 to 7.0 months v 1.4 months; 95% CI, 1.4 to 2.7 months, respectively).The PFS hazard ratio (HR) for patients with METdriven PRCC was 0.33 (95% CI, 0.20 to 0.52) compared with METindependent PRCC (log-rank P , .001;Fig 2).After discontinuation of savolitinib, 10 (23%) and 14 patients (30%) with MET-driven and MET-independent disease, respectively, received additional lines of therapy for PRCC.Of the eight patients exhibiting a PR, six were still responding to treatment at data cutoff, with a duration of response of 2.4 to 16.4 months.Two patients who achieved a PR subsequently experienced progressive disease after 1.8 and 2.8 months.

Activity by Lines of Prior Therapy
Of the patients achieving a PR, three had received no prior therapy and five had received one or more previous lines of treatment.For the 38 patients (35%) with a best response of stable disease, 24 had received no previous therapy and 14 had received one or more prior lines of treatment.Among patients with progressive disease, 26 had received no prior therapy and 22 had received one or more lines of treatment.Prior cancer therapies are listed in Appendix Table A1 (online only).

Patient-Reported Outcomes: Quality of Life
Quality-of-life data were collected via the FACT-G, FKSI-19, and EQ-5D-5L questionnaires.Overall, during savolitinib treatment, scores for symptoms and quality of life were maintained at similar values to those reported at baseline, irrespective of the questionnaire used (Appendix Table A2, online only).This suggests that, among patients completing the questionnaires, there was little change in these outcomes during the study.

Safety and Tolerability
The overall incidence of AEs and those considered treatment related occurring in $ 3% of patients who received at least one dose of savolitinib are listed in Table 3.Most patients (88%) experienced at least one AE considered by the investigator to be related to study drug.Abnormal liver function tests were reported in 20% of patients, irrespective of MET status.The most frequently observed hepatic AEs were increased blood AST in 12 patients (11%) and ALT in 11 patients (10%), which were grade 3 or higher in four (4%) and five patients (5%), respectively.Three patients reported serious AEs considered at least related to treatment; these were: grade 3 pneumonitis, grade 4 elevated transaminases, and grade 4 drug-induced liver injury, which led to death resulting from hepatic encephalopathy.
A total of 13 AEs in nine patients (6%) led to drug discontinuation.These AEs included increased ALT and peripheral edema (both in two patients) and individual events of increased AST, proteinuria, pain, nausea, vomiting, fatigue, and embolism.Dosing was delayed for 47 patients (43%); in 38 (35%) of these patients, the delay was because of an AE.Fourteen patients (13%) had dose reductions as a result of an AE at some time during the study.

DISCUSSION
To our knowledge, this is the largest study of patients with advanced or metastatic PRCC with central pathology review and biomarker analysis used to define MET-driven and METindependent PRCC.Patients experiencing responses in this study were found only in the group with MET-driven disease, indicating that savolitinib may suppress MET-driven tumor growth.Furthermore, PFS for those with MET-driven tumors was significantly longer (6.2 months) compared with those with METindependent disease (1.4  will allow further assessment of the difference in PFS and provide data on overall survival.

months). Continued long-term follow-up
The analysis of MET mutations, copy number gain, or gain of chromosome 7 in patients defined as having type 1 or type 2 PRCC on the basis of pathology challenges the view that aberrations in MET are mainly associated with type 1 PRCC.Our study found MET copy number gain (either chromosomal or focal) in 72% of type 1 PRCCs and 46% of type 2 PRCCs, which is comparable to previous reports of MET copy number gain in 81% and 46% of type 1 and type 2 PRCCs, respectively. 11The frequency of MET kinase domain mutations identified here is also similar to data from The Cancer Genome Atlas Research Network of 15%, 2%, and 12% in type 1, type 2, and unclassified PRCCs, respectively. 12Of note, the The Cancer Genome Atlas publication reported any MET gene mutation, whereas our study considered only kinase domain mutations.This may explain the slightly reduced frequency of MET mutations in our study and suggests the molecular features of PRCC reported here are comparable to those of previously published cohorts.Overall, METstatus was more predictive of response to savolitinib in our study than a classification based on pathology; all partial responders had archival tumor samples that harbored a copy number gain in the MET pathway (HGF, MET, or chromosome 7), some in combination with a MET kinase domain mutation.This suggests that genomic profiling can better identify patients who may respond to savolitinib than can the type 1 or type 2 histologic subtype, adding relevance to the use of genomic profiling in such studies.MET pathway PRCC tumors have also been identified by others to not only be associated with type 1 histology. 12verall, savolitinib was generally well tolerated, with the three most common AEs (nausea, fatigue, and vomiting) also commonly reported previously by patients with PRCC receiving foretinib. 34Other AEs occurring in $ 10% of patients in our study were peripheral edema, ALT and AST increases, serum creatinine increase, and decreased appetite.Increased ALT and AST have been reported in patients with PRCC treated with foretinib (in 22% and 24% of patients, respectively) and sunitinib (up to 74% of patients). 26,27,34A majority of these events, as in our study, were grade 1 or 2. Dosing of savolitinib was only reduced in four patients because of abnormal liver function, and two patients discontinued treatment because of abnormal AST or ALT levels.
Following the history of evaluating MET inhibitors for the treatment of RCC, these results with savolitinib are encouraging.There have been numerous reasons why earlier MET inhibitors have failed during drug development.For example, tivantinib was reported to be a selective MET inhibitor, but a subsequent study reported similar suppression of both MET-dependent and METindependent tumor cell lines, via inhibition of microtubule polymerization. 40Less selective multikinase inhibitors may not achieve the extent or duration of MET pathway attenuation necessary for suppression of MET-mediated migration and invasion at tolerated doses.Most importantly, patients have not been appropriately selected in the past; it is now increasingly recognized that high MET protein levels (usually detected by immunohistochemistry) do not always correlate with response or survival outcomes. 41In a phase II study of foretinib, tumor responses were reported in five of 10 patients with germline MET mutations compared with 9% of patients without MET mutations. 34Recently, in a small study of type 1 PRCC, patients with MET-driven disease (MET mutation positive [n = 4] and MET amplification [n = 2]) responded to treatment with crizotinib, an inhibitor of MET, ALK, and ROS1. 42The current interest in MET inhibitors for the treatment of PRCC is demonstrated by the initiation of a randomized phase II trial comparing the efficacy of several MET kinase inhibitors, including savolitinib (ClinicalTrials.govidentifier: NCT02761057).In that cooperative group study, patients are not being selected based on MET mutation status, although tumor response by MET mutation and expression level is an additional outcome.
Limitations of our study include the single-arm design, and therefore the inherent lack of a comparator group, and the relatively small number of patients with MET-driven disease.Prognostic information for MET in PRCC is also lacking because of the single-arm design.
In summary, these results confirm that savolitinib, a potent and selective small-molecule MET kinase inhibitor, holds promise as a personalized treatment for patients with metastatic MET-driven PRCC.Our study identified a defined molecular group and highlights the prevalence of MET-driven disease, including patients with ligand-dependent (ie, HGF amplification) and -independent PRCC who responded to treatment.These data support the hypothesis that savolitinib has antitumor activity in patients with MET-driven PRCC and justifies the recently launched phase III trial comparing savolitinib with sunitinib in a population of patients with MET-driven PRCC (ClinicalTrials.gov identifier: NCT03091192)

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Disclosures provided by the authors are available with this article at jco.org.

Fig 1 .
Fig 1. Best percentage change in tumor size from baseline according to MET status.Investigatorassessed measurements of target lesion size using RECIST (version 1.0) in patients with papillary renal cell carcinoma (PRCC) and MET status assessment with measurable disease at baseline and at least one postbaseline measurement.(A) MET-driven PRCC (n = 40); (B) MET-independent PRCC (n = 41); and (C) MET status unknown (n = 16).

Fig 2 .
Fig 2. Kaplan-Meier estimate of progressionfree survival (PFS), defined as time from the date of first dosing until the date of objective disease progression or death resulting from any cause, in patients with papillary renal cell carcinoma by MET status (treatment population).(+) indicates censored event.

Table 1 .
Patient Demographic and Baseline Clinical Characteristics

Table A2 .
Change From Baseline of Overall FACT-G, FKSI-19, and EQ-5D-5L Scores by MET Status at Selected Time Points