Effect of Fruquintinib vs Placebo on Overall Survival in Patients With Previously Treated Metastatic Colorectal Cancer: The FRESCO Randomized Clinical Trial

Key Points

Question  Does fruquintinib prolong overall survival in patients with metastatic colorectal cancer (CRC) who have tumor progression following at least 2 lines of chemotherapy, targeted treatment, or both?

Findings  In this randomized clinical trial involving 416 patients in China with metastatic CRC who had tumor progression following at least 2 lines of chemotherapy, treatment with fruquintinib resulted in a statistically significant increase in overall survival compared with placebo (median survival time, 9.3 vs 6.6 months).

Meaning  Fruquintinib may prolong survival in patients with metastatic colorectal cancer who had tumor progression after previous treatment, although the efficacy of this therapy remains to be assessed outside of China.

Importance  Patients with metastatic colorectal cancer (CRC) have limited effective and tolerable treatment options.

Objective  To evaluate the efficacy and safety of oral fruquintinib, a vascular endothelial growth factor receptor (VEGFR) inhibitor, as third-line or later therapy in patients with metastatic CRC.

Design, Setting, and Participants  FRESCO (Fruquintinib Efficacy and Safety in 3+ Line Colorectal Cancer Patients) was a randomized, double-blind, placebo-controlled, multicenter (28 hospitals in China), phase 3 clinical trial. From December 2014 to May 2016, screening took place among 519 patients aged 18 to 75 years who had metastatic CRC that progressed after at least 2 lines of chemotherapy but had not received VEGFR inhibitor therapy; 416 met the eligibility criteria and were stratified by prior anti-VEGF therapy and K-ras status. The final date of follow-up was January 17, 2017.

Interventions  Patients were randomized in a 2:1 ratio to receive either fruquintinib, 5 mg (n = 278) or placebo (n = 138) orally, once daily for 21 days, followed by 7 days off in 28-day cycles, until disease progression, intolerable toxicity, or study withdrawal.

Main Outcomes and Measures  The primary end point was overall survival. Key secondary efficacy endpoints were progression-free survival (time from randomization to disease progression or death), objectiveresponse rate (confirmed complete or partial response), and disease control rate (complete or partial response, or stabledisease recorded ≥8 weeks postrandomization). Duration of response was also assessed. Safety outcomes included treatment-emergent adverse events.

Results  Of the 416 randomized patients (mean age, 54.6 years; 161 [38.7%] women), 404 (97.1%) completed the trial. Median overall survival was significantly prolonged with fruquintinib compared with placebo (9.3 months [95% CI, 8.2-10.5] vs 6.6 months [95% CI, 5.9-8.1]); hazard ratio (HR) for death, 0.65 (95% CI, 0.51-0.83; P < .001). Median progression-free survival was also significantly increased with fruquintinib (3.7 months [95% CI, 3.7-4.6] vs 1.8 months [95% CI, 1.8-1.8] months); HR for progression or death, 0.26 (95% CI, 0.21 to 0.34; P < .001). Grades 3 and 4 treatment-emergent adverse events occurred in 61.2% (170) of patients who received fruquintinib and 19.7% (27) who received placebo. Serious adverse events were reported by 15.5% (43) of patients in the fruquintinib group and 5.8% (8) in the placebo group, with 14.4% (40) of fruquintinib-treated and 5.1% (7) of placebo-treated patients requiring hospitalization.

Conclusions and Relevance  Among Chinese patients with metastatic CRC who had tumor progression following at least 2 prior chemotherapy regimens, oral fruquintinib compared with placebo resulted in a statistically significant increase in overall survival. Further research is needed to assess efficacy outside of China.

Trial Registration  ClinicalTrials.gov Identifier: NCT02314819

Colorectal cancer (CRC) is the third most common cancer worldwide, causing more than 800 000 deaths in 2015.¹ The annual incidence and mortality of CRC in China has increased steadily since 2000, with 376 300 new cases and 191 000 deaths reported in 2015.²

Fluorouracil with leucovorin plus either irinotecan³ or oxaliplatin⁴ are the standard chemotherapeutic regimens for treating metastatic CRC.^5,6 To improve patient outcomes, chemotherapy can be combined with bevacizumab^7,8 or aflibercept⁸ to target the vascular endothelial growth factor (VEGF) pathway or with cetuximab or panitumumab to target the epidermal growth factor receptor (EGFR).^9,10 Although patients who progress after receiving 2 lines of systemic chemotherapy may still have good performance status, third-line treatment options are limited. There is therefore a strong unmet clinical need for treatment options in the third-line setting for metastatic CRC, especially in China.

The VEGF pathway is vital to the neoangiogenesis associated with tumor proliferation.¹¹ Antiangiogenic agents targeting the VEGF pathway include those that inhibit the ligand (ie, VEGF inhibitors) or its receptor (ie, vascular endothelial growth factor receptor [VEGFR] inhibitors). These agents inhibit new blood vessel growth and lead to vascular regression, tumor vessel normalization, and constriction; offsetting chemotherapy’s induction of VEGF.¹¹ The rationale of continuing to target the VEGF pathway in metastatic CRC was confirmed when third-line monotherapy with the VEGFR inhibitor regorafenib improved survival in global¹² and Asian¹³ study populations. However, the efficacy of regorafenib is limited and its adverse effects, particularly hepatotoxicity and fatigue, may be difficult to manage.¹²

Quiz Ref IDFruquintinib is a VEGFR inhibitor that blocks new blood vessel growth associated with tumor proliferation.¹⁴ It is a potent, highly selective small-molecule inhibitor of VEGFR-1, -2, and -3.¹⁵ This phase 3 study was conducted to assess its efficacy and adverse event profile in patients with metastatic CRC that had progressed after second-line or subsequent treatment.

The study protocol (Supplement 1) was approved by the independent ethics committee/institutional review board of each participating center. This study was conducted in accordance with the Declaration of Helsinki¹⁶ and Guidelines for Good Clinical Practice, as well as the local laws and regulations of China. An independent data monitoring committee, comprising 3 oncologists and 1 statistician, ensured the overall integrity of the trial and safety of the participants.

Quiz Ref IDEligible patients were aged 18 to 75 years, weighed at least 40 kg, and had to provide written informed consent prior to enrollment. They had to have histologically and/or cytologically confirmed metastatic CRC that progressed following at least 2 standard chemotherapy regimens. Patients had to have evidence of disease progression during or within 3 months after the last administration of standard treatment or to have stopped treatment because of unacceptable toxic effects.

Other eligibility criteria included an Eastern Cooperative Oncology Group performance status of 0 or 1; left ventricular ejection fraction of 50% or greater; measurable disease by Response Evaluation Criteria in Solid Tumors version 1.1; adequate bone marrow, liver, and renal function; and a life expectancy of at least 12 weeks.

Prior treatment with VEGF inhibitors (eg, bevacizumab and aflibercept) or EGFR inhibitors was permitted and was used as one of the stratification factors in efficacy analysis. However, because fruquintinib is a VEGFR inhibitor, patients who received prior treatment with other VEGFR inhibitors (eg, sorafenib, sunitinib, axitinib, regorafenib, ramucirumab, apatinib, axitinib, famitinib, or other tyrosine kinase inhibitors) were excluded. Full inclusion and exclusion criteria are listed in eTable 1 in Supplement 2.

FRESCO (Fruquintinib Efficacy and Safety in 3+ Line Colorectal Cancer Patients) was a randomized, double-blind, placebo-controlled, multicenter, phase 3 trial in patients with metastatic CRC who had tumor progression following treatment regimens that included fluoropyrimidine, oxaliplatin, and irinotecan.

Patients were randomly assigned by an interactive web response system in a 2:1 ratio to receive either oral fruquintinib (5 mg/d) or matching placebo, both in combination with best supportive care (Figure 1). Randomization was stratified by prior use of VEGF inhibitor treatment (yes vs no) and K-ras mutational status (wild type vs mutated). The investigators, sponsor, and patients were blinded to treatment allocation until database lock (sponsor) or study completion (investigators).

Quiz Ref IDEligible participants repeated the 28-day treatment cycle of 3 weeks on followed by 1 week off until disease progression, death, unacceptable toxicity, withdrawal of consent by the patient, or discontinuation by the physician. All patients received best supportive care excluding other investigational antitumor agents or anticancer treatments during the study period (from 28 days before the informed consent form date to end of treatment). Protocol-predefined dose reduction (in two 1-mg/d increments) was permitted to manage clinically significant treatment-related toxic effects (eTable 2 in Supplement 2). Treatment was discontinued permanently if the toxicity did not resolve after a 2-week treatment interruption or did not meet protocol-defined criteria after 2 dose reductions. No crossover between treatment groups was permitted.

The primary efficacy outcome was overall survival, defined as the time from randomization until death. For patients who were not reported to have died at the planned analysis cutoff, the final known date of survival was used as the censoring date.

Tumor assessment was performed every 8 weeks as defined by Response Evaluation Criteria in Solid Tumors version 1.1. Key secondary efficacy end points were progression-free survival (defined as time from randomization to disease progression or death), objective response rate (defined as a confirmed complete or partial response), and disease control rate (defined as a complete or partial response, or stable disease recorded ≥8 weeks after randomization). Duration of response was also assessed.

Safety assessments included adverse events, laboratory abnormalities (hematology, clinical chemistry, and urinalysis), vital signs, electrocardiograms, and echocardiographs. Treatment-emergent adverse events were graded using the National Cancer Institute’s Common Terminology Criteria for Adverse Events version 4.03. The duration, severity, and relation to study medication, based on the investigator’s clinical assessment, were recorded at scheduled study visits.

Taking into consideration the previous phase 1b and phase 2 efficacy data for fruquintinib,¹⁷ assuming the hazard ratio (HR) for death with fruquintinib over placebo was 0.7, this would correspond to a median overall survival of 9.0 months for fruquintinib and 6.3 months for placebo. Based on this assumption, it was calculated that 280 overall survival events would provide an 80% power to detect a difference in overall survival between the 2 treatment groups, with a 2-sided P value of .05 indicating statistical significance. To achieve this, the study was planned to randomize approximately 400 patients.

Statistical analyses were carried out using SAS version 9.2. The statistical analysis plan and tests for the proportional hazards assumption are detailed in Supplement 3 and Supplement 4. Overall survival and progression-free survival were compared using a stratified log-rank test. Hazard ratios with 95% CIs were estimated using the Cox model, adjusting for baseline stratification factors. To evaluate the effect of the different study sites, post hoc analyses of overall survival and progression-free survival were performed using stratified Cox models with site as a random effect. Kaplan-Meier curves and median survival were estimated for each treatment group. The Mantel-Haenszel test was used for comparisons of objective response rate and disease control rate. Subgroup analyses of overall survival and progression-free survival were conducted for the covariates using descriptive statistics and HR (95% CI). Post hoc tests for interaction of treatment by each covariate were performed by fitting Cox models, including treatment, covariate, and covariate by treatment interaction term. The significance level for the interaction test was set at 2-sided P = .05.

The assumption of proportionality was assessed first by examining plots of complementary log-log (event times) vs log (time), and if this raised concerns, by adding interaction of treatment with time to the stratified Cox model to assess the extent to which this represented random variation. If a lack of proportionality was evident, the variation in treatment effect would be described by presenting a piecewise HR calculated over distinct time periods from a post hoc analysis. In such circumstances, the HR can still be meaningfully interpreted as an average HR over time.^18,19

The efficacy analyses were based on the intention-to-treat population, which included all randomized patients. The per-protocol set included the patients who did not experience any major protocol deviations that may have influenced the overall survival evaluation and who completed at least 1 treatment cycle. The per-protocol set was only analyzed for overall survival. The safety analysis set included all randomized patients who received at least 1 dose of study treatment. Missing efficacy data were not imputed when performing statistical analyses.

Between December 8, 2014, and May 13, 2016, 519 patients from 28 study sites in China were screened; the 416 eligible patients who consented to participate were randomized to receive fruquintinib (n = 278) or placebo (n = 138) plus best supportive care (for both groups) and were included in the intention-to-treat population. Except for 1 patient in the placebo group, all patients received treatment as allocated and were included in the safety analysis. Participant flow through the study is shown in Figure 1. The median follow-up time was 13.3 months for the fruquintinib group and 13.2 months for the placebo group. The final date of follow-up was January 17, 2017.

Patient baseline characteristics are summarized in Table 1. Of the 416 randomized patients (mean age, 54.6 years; 161 [38.7%] women), 404 (97.1%) completed the trial. Most baseline demographics, disease characteristics, and prior treatments were similar between the treatment groups, except the proportion of men was higher in the placebo group than in the fruquintinib group. Most patients had multiple metastases, with liver metastases present in 185 of 278 (66.5%) in the fruquintinib group and in 102 of 138 (73.9%) in the placebo group. Between-group proportions were similar for patients who had previously received VEGF inhibitors (30.2% [84] in the fruquintinib group vs 29.7% [41] in the placebo group) or EGRF inhibitors (14.4% [40] in the fruquintinib group vs 13.8% [19] in the placebo group), and who had K-ras mutations (43.5% [121] in the fruquintinib group vs 46.4% [64] in the placebo group).

After disease progression or the end of study treatment, 188 patients (45.2%) received further systemic treatment (118 of 278 [42.4%] in the fruquintinib group and 70 of 138 [50.7%] in the placebo group), which included cytotoxic anticancer therapies, monoclonal antibodies, and kinase inhibitors (eTable 3 in Supplement 2).

At the planned cutoff date (January 17, 2017), after 297 deaths, the median overall survival was 9.30 (95% CI, 8.18-10.45) months in the fruquintinib group and 6.57 (95% CI, 5.88-8.11) months in the placebo group (HR for death, 0.65 [95% CI, 0.51-0.83]; log-rank test P < .001; Figure 2A). The overall survival findings in the per-protocol set were similar: 9.30 (95% CI, 8.18-10.45) months with fruquintinib and 6.80 (95% CI, 5.91-8.38) months with placebo (HR for death, 0.66 [95% CI, 0.52-0.85]; log-rank test P = .001). Sensitivity analyses of the overall survival data considering the effect of study site showed similar results, with HRs for death of 0.63 (95% CI, 0.49-0.81) in the intention-to-treat population and 0.65 (95% CI, 0.50-0.83) in the per-protocol set (eTable 4 in Supplement 2).

Median progression-free survival was also significantly increased with fruquintinib (3.71 [95% CI, 3.65-4.63] vs 1.84 [95% CI, 1.81-1.84] months with placebo; HR for progression or death, 0.26 [95% CI, 0.21-0.34]; P < .001; Figure 2B). Sensitivity analysis of progression-free survival considering site effect provided similar results (3.7 [95% CI, 3.7-4.6] months with fruquintinib vs 1.8 [95% CI, 1.8-1.8] months with placebo; HR for progression or death, 0.26 [95% CI, 0.20-0.33]; eTable 4 in Supplement 2). Given that the assumption of proportional hazards for the Cox model was violated, the piecewise HRs for progression-free survival were explored. The results showed a superior treatment effect of fruquintinib compared with placebo during both time intervals (HR for progression or death, 0.20 [95% CI, 0.14-0.29] during the interval from 0 to 1.85 months postrandomization; HR for progression or death, 0.12 [95% CI, 0.05-0.26] during the interval from 1.85 months postrandomization and thereafter). The cutoff point of 1.85 months postrandomization was the time point at which the 2 progression-free survival curves approached each other most closely.

Subgroup analyses of overall survival were generally consistent with benefit in the intention-to-treat population across nearly all subgroups (Figure 3). The overall survival benefit among the patients who had previously been treated with VEGF inhibitors (84 with fruquintinib and 41 with placebo) was similar to that in the intention-to-treat population, with an HR of 0.68 (95% CI, 0.45-1.03). Subgroup analyses revealed no significant difference between fruquintinib and placebo among women (HR for death, 0.85 [95% CI, 0.57-1.29]), patients aged 65 years and older (HR for death, 0.95 [95% CI, 0.55-1.63]), or among patients with right-sided primary tumors (HR for death, 0.96 [95% CI, 0.53-1.75]). Progression-free survival was superior for fruquintinib compared with placebo across all patient subgroups. The test for interaction of treatment by K-ras status was statistically significant (P = .04; see Figure 4), although the small patient numbers in each subgroup precluded firm conclusions.

Compared with placebo, patients treated with fruquintinib also demonstrated a significantly higher objective response rate (13 of 278 patients [4.7%] vs 0%; P = .01; treatment difference, 4.7% [95% CI, 2.1%-7.2%]) and disease control rate (62.2% vs 12.3%; P < .001; treatment difference, 49.9% [95% CI, 42.0%-57.8%]). With fruquintinib, a complete response was achieved by 1 patient (0.4%; treatment difference, 0.4% [95% CI, −0.3% to 1.1%]), and partial response was achieved by 12 patients (4.3%) in the fruquintinib group vs none in the placebo group (treatment difference, 4.3% [95% CI, 1.9%-6.7%]). Most of the patients who had a response were still on treatment without disease progression at the cutoff date, hence the median duration of response (5.6 months) was not reached and a 95% CI was not calculable.

The median treatment exposure was 3.7 months (range, 0.1-21.9 months) for fruquintinib and 1.8 months (range, 0.1-11.1 months) for placebo. The mean relative dose intensity was 92% in the fruquintinib and 98% in the placebo group; the median relative dose intensity was 100% in both groups.

In the safety analysis set, 274 (98.6%) of the 278 patients in the fruquintinib group and 121 (88.3%) of the 137 patients in the placebo group experienced at least 1 treatment-emergent adverse event; among these, 170 (61.2%) patients receiving fruquintinib and 27 (19.7%) receiving placebo experienced a treatment-emergent adverse event of grade 3 or higher severity. Serious adverse events were reported in 43 (15.5%) patients in the fruquintinib group and 8 (5.8%) patients in the placebo group. Forty (14.4%) of the 278 patients treated with fruquintinib and 7 (5.1%) of the 137 patients receiving placebo required hospitalization or prolongation of an existing hospital stay to manage study drug toxicity. The adverse event data are summarized in Table 2. The grade 3 to 4 severity adverse events most frequently reported with fruquintinib were hypertension in 59 of 278 patients (21.2%), hand-foot skin reaction in 30 patients (10.8%), and proteinuria in 9 patients (3.2%) (Table 2). Grade 3 hepatic toxicities occurred in 1.5% or less in both treatment groups.

Overall, 11 patients (9 patients receiving fruquintinib [3.2%]; 2 receiving placebo [1.5%]) had fatal treatment-emergent adverse events (Table 2). In the fruquintinib group, these treatment-emergent adverse events were gastrointestinal hemorrhage, death not otherwise specified, lung infection, fungal lower respiratory tract infection, multiple organ dysfunction syndrome, sudden death, bacterial infection, cerebral infarction, and hemoptysis. In the placebo group, these treatment-emergent adverse events were pulmonary embolism and shock (eTable 5 in Supplement 2).

Fifty patients (42 receiving fruquintinib [15.1%] and 8 [5.8%] receiving placebo) discontinued treatment due to treatment-emergent adverse events (Table 2). The treatment-emergent adverse event that most frequently led to fruquintinib discontinuation was proteinuria in 6 patients (2.2%). Treatment interruption or dose reduction due to a treatment-emergent adverse event was required by 131 (47.1%) fruquintinib-treated patients and 18 (13.1%) patients receiving placebo. The treatment-emergent adverse events that most commonly led to fruquintinib interruption or dose reduction were hand-foot skin reaction in 37 (13.3%), proteinuria in 27 (9.7%), and decreased platelet counts in 15 (5.4%) patients.

Quiz Ref IDThis study showed that oral fruquintinib compared with placebo resulted in a statistically significant increase in overall survival among Chinese patients with metastatic CRC who had tumor progression following at least 2 prior chemotherapy regimens.

Fruquintinib is an oral VEGFR inhibitor that has the same mechanism of action as regorafenib, which was approved for third-line treatment of metastatic CRC by the US Food and Drug Administration in September 2014. When this study was conducted, regorafenib was not available in China, and no participants received this drug or any other VEGFR inhibitor prior to or during the study.

In contrast to the treatment patterns in North America and Europe, in China, the VEGF inhibitors bevacizumab and aflibercept are not routinely integrated into first- or second-line therapy. This raises the possibility that results achieved in this study with a VEGFR inhibitor might not generalize to contexts where all patients receive VEGF inhibitors early in their treatment course. The CORRECT trial (regorafenib monotherapy for previously treated metastatic colorectal cancer) compared regorafenib to placebo among 760 patients who had received at least 2 prior chemotherapy regimens including a VEGF inhibitor (typically bevacizumab). In that study, a median survival of 6.4 months was obtained with regorafenib vs 5.0 months with placebo, with an HR for death of 0.77 (95% CI, 0.64-0.94).¹² The CONCUR trial (regorafenib plus best supportive care vs placebo plus best supportive care in Asian patients with previously treated metastatic colorectal cancer) compared regorafenib and placebo as third-line treatment for 204 Asian patients with metastatic CRC, not all of whom had previously received a VEGF inhibitor, and that study also demonstrated better survival with regorafenib (median 8.8 months, [95% CI, 7.3-9.8]) than placebo (median 6.3 months, [95% CI, 4.8 to 7.6] and an HR for death of 0.55 [95% CI, 0.40-0.77]).¹³ In the present study, 30% of patients had received prior VEGF inhibitor therapy, and even in this subgroup, fruquintinib compared with placebo had an HR for death of 0.68 (95% CI, 0.45-1.03) and an HR for progression or death of 0.24 (95% CI, 0.15-0.38).

Systemic chemotherapy can induce VEGF-A and VEGF-C.¹¹ Currently available VEGF inhibitors (eg, bevacizumab and aflibercept) shut down tumor angiogenesis by inhibiting the VEGF-A/VEGFR2 pathway while having little effect on VEGFR3 activation and lymphangiogenesis induced by VEGF-C. Fruquintinib has potent activity against all 3 isoforms of VEGFR 1, 2, and 3²⁰ and at doses of 5 mg once daily, the steady state average trough drug plasma concentration provided a complete and sustained VEGFR inhibition,¹⁵ potentially leading to simultaneous inhibition of tumor angiogenesis and lymphangiogenesis.

Quiz Ref IDThe overall incidence of adverse events and serious adverse events was higher in the fruquintinib treatment group than in the placebo group. The most frequently reported adverse events of grade 3 to 4 severity in the fruquintinib treatment group included hypertension, hand-foot skin reaction, proteinuria, and diarrhea. These adverse events have been commonly observed with other VEGF or VEGFR inhibitors.^12,21 Although the incidence of these adverse events in this study was higher with fruquintinib than placebo, most occurred during the first 2 cycles of treatment and could be managed with supportive care and dose adjustment. Furthermore, the duration of fruquintinib treatment was twice as long as that of placebo, hence the adverse event observation periods differed between the 2 treatment groups. This may have contributed to the relatively higher incidence of adverse events and serious adverse events observed with fruquintinib.

This study has several limitations. First, the outcomes were evaluated in a purely Chinese population, and further studies will be needed to confirm fruquintinib’s efficacy and tolerability in other populations. Second, metastatic CRC standard of care in China differs from that in the Western world in that only one-third of patients had received prior treatment with anti-VEGF or anti-EGFR antibodies. Third, because neither regorafenib nor TAS-102 (tipiracil hydrochloride, a combination of trifluridine and tipiracil hydrochloride)²² were available in China during the time of study conduct, it was not possible to compare the efficacy of fruquintinib directly with those of regorafenib or TAS-102. Fourth, microsatellite instability status, which can influence prognosis and response to immunotherapy, was not defined.

Among Chinese patients with metastatic CRC who had tumor progression following at least 2 prior chemotherapy regimens, oral fruquintinib compared with placebo resulted in a statistically significant increase in overall survival. Further research is needed to assess efficacy outside China.

Corresponding Author: Shukui Qin, MD, Department of Medical Oncology, Nanjing Chinese Medicine University Affiliated Bayi Hospital, No. 34, 34 Biao, Yanggongjing, Nanjing, Jiangsu, 210001, China (qinsk@csco.org.cn).

Accepted for Publication: May 21, 2018.

Author Contributions: Drs Jin Li and Shukui Qin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: J. Li, Qin, R. Xu, Shen, J. Xu, Bai, Yang, Z. Chen, Pan, Guo, Shu, Zhou, Liu, Wu, W. Li, Yu, H. Chen, Fan, Su.

Acquisition, analysis, or interpretation of data: J. Li, Qin, R. Xu, Shen, J. Xu, Bai, Yang, Deng, Z. Chen, Zhong, Guo, Shu, Yuan, Zhou, N. Xu, Ma, Wu, Cheng, D. Chen, W. Li, Sun, Yu, Cao, J. Wang, S. Wang, H. Wang, Fan, Hua.

Drafting of the manuscript: J. Li, Qin, R. Xu, J. Xu, Bai, Yang, Deng, Z. Chen, Pan, Guo, Shu, Zhou, Liu, Wu, W. Li, Yu, S. Wang, Fan, Su.

Critical revision of the manuscript for important intellectual content: J. Li, Qin, Shen, J. Xu, Zhong, Yuan, N. Xu, Liu, Ma, Cheng, D. Chen, Sun, Cao, H. Chen, J. Wang, H. Wang, Hua, Su.

Statistical analysis: Yuan, H. Chen, Hua.

Obtained funding: R. Xu, Bai, Yang, Z. Chen, Pan, Shu, Wu, W. Li, Yu, Fan, Hua.

Administrative, technical, or material support: J. Li, Qin, R. Xu, Shen, J. Xu, Bai, Yang, Z. Chen, Zhong, Guo, Shu, Zhou, N. Xu, Liu, Ma, Wu, Cheng, W. Li, Sun, Yu, Cao, J. Wang, H. Wang, Fan, Hua, Su.

Supervision: J. Li, R. Xu, J. Xu, Bai, Yang, Z. Chen, Shu, Liu, Ma, Wu, W. Li, Yu, Fan, Hua, Su.

Drs Jin Li and Shukui Qin contributed equally to this study.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr R. Xu reported receipt of personal fees and nonfinancial support from Roche and Merck; and personal fees from Bristol-Myers Squibb and Bayer. Drs Fan, Hua, and Su reported receipt of personal fees as employees of Hutchison MediPharma during the conduct of the study but outside the submitted work. No other disclosures were reported.

Funding/Support: This study was sponsored by Hutchison MediPharma, and was supported by grants from the Ministry of Science and Technology of the People's Republic of China (2013ZX09101105) and the Science and Technology Development Fund of Pudong Shanghai China (PKJ2014-S01).

Role of the Funder/Sponsor: All of the sponsors led, participated in, and funded the design and conduct of the study in collaboration with the academic authors. The sponsors funded and participated in the collection, management, analysis, and interpretation of data. All authors, including those employed by the sponsor, participated in the preparation and review of the manuscript. All authors, including those employed by the sponsor, provided approval of the manuscript. Although the sponsor participated in the decision to submit the manuscript for publication, it did not have the ability to veto that decision.

Meeting Presentation: This study was presented at the annual meeting of the American Society of Clinical Oncology; June 5, 2017; Chicago, Illinois.

Additional Contributions: We are grateful to all patients and their families for their participation in this study, the study sites, and the study monitors. Editorial assistance with preparation of the manuscript was provided by Samantha Santangelo, PhD, of Nucleus Global, Shanghai, China, funded by Hutchison MediPharma; and by Mengye Peng and Wei Gong of Hutchison MediPharma.