Surufatinib in Advanced Well-Differentiated Neuroendocrine Tumors: A Multicenter, Single-Arm, Open-Label, Phase Ib/II Trial

This article is featured in Highlights of This Issue, p. 3471

Neuroendocrine tumors (NET) are rare neoplasms arising from diffusive neuroendocrine cells of various organs. In the past four decades, the age-adjusted incidence rate increased 6.4-fold in the United States, from 1.09 per 100,000 persons in 1973 to 6.98 per 100,000 persons in 2012 (1). In China, a similar trend was identified by a hospital-based, nationwide, retrospective epidemiologic study of gastroenteropancreatic neuroendocrine neoplasms (2). Treatment options for advanced, low, or intermediate grade NETs include somatostatin receptor-targeting therapeutics, peptide receptor radionuclide therapy, systemic chemotherapies, targeted agents including sunitinib in pancreatic NETs, and everolimus in pancreatic, gastrointestinal, and bronchopulmonary NETs, and local-regional treatments (3). Nearly half of all patients with NETs have distant metastasis at initial diagnosis (3). The median survival time for patients with well-differentiated to moderately differentiated distant stage NETs varies by tumor origins, ranging from 103 months for small intestinal origin, to 60 months for pancreatic origin, and 14 months for colonic origin (1).

Although NETs are highly vascularized neoplasms, those originating from diverse organs respond to antiangiogenesis treatment differently (4–6). In a phase III pivotal study, sunitinib significantly prolonged progression-free survival (PFS) in patients with pancreatic NETs compared with placebo [11.4 months vs 5.5 months; HR, 0.42; 95% confidence intervals (CI) 0.26–0.66; P < 0.001; ref. 5], but failed to demonstrate efficacy in extrapancreatic NETs (6). Bevacizumab, a potent antivascular endothelial growth factor (VEGF) agent, has been studied in combination with several approved medications for treating NETs. In a phase II randomized study comparing bevacizumab combined with everolimus to everolimus alone in treating pancreatic NETs, a higher response rate of 31% versus 12% (P = 0.005) was observed in the bevacizumab arm (7). However, the median PFS was similar between the two arms: 16.7 versus 14 months (HR, 0.80; 95% CI, 0.55–1.17, P = 0.12; ref. 7). A similar trend in response rates and PFS was shown in a phase III study comparing the combination of bevacizumab plus octreotide versus interferon alfa-2b plus octreotide [response rate (RR): 12% vs. 4%, P = 0.008; PFS: 16.6 vs. 15.4 months; HR 0.93; 95% CI, 0.73–1.18; P = 0.55; ref. 8]. However, in spite of all the previous efforts investigating VEGF pathway inhibition, to date, no antiangiogenic treatment is approved for NETs originating outside of the pancreas. Fibroblast growth factor (FGF) 2 was shown to be a potent mediator in antiangiogenesis resistance development, and inhibiting FGF receptor signaling could overcome resistance (9). Preclinical cancer models also showed that macrophages, usually recruited and activated by colony-stimulating factor 1 receptor (CSF-1R), played a proangiogenic role in the tumor microenvironment. Furthermore, eliminating tumor-associated macrophages by inhibiting CSF-1R led to decreased neoangiogenesis (10). Therefore, inhibiting these targets simultaneously could be a promising antiangiogenic strategy.

Surufatinib (HMPL012, previously named as sulfatinib) is a potent, small-molecule tyrosine kinase inhibitor (TKI), selectively targeting VEGF receptors (VEGFR) 1, 2, and 3, FGFR 1, and CSF-1R. In a phase I dose-finding study in patients with advanced solid tumors, surufatinib demonstrated antitumor activity in hepatocellular carcinoma and NETs, both highly vascularized tumors (11). The objectives of this phase Ib/II study were to further evaluate the efficacy, safety, and tolerability of surufatinib in patients with advanced NETs of diverse origins.

This was a multicenter, single-arm, open-label, phase Ib/II trial. Patients with pancreatic or extrapancreatic NETs were enrolled in corresponding cohorts in 7 clinical centers across China. The primary objective of this study was to evaluate the efficacy and safety of surufatinib. Pharmacokinetics were evaluated as the secondary objective. Assessing the association between antitumor activities and VEGF/FGF pathway biomarker expression was an exploratory objective.

Eligible patients were ≥18 years, with pathologically confirmed low or intermediate grade (G1 or G2) inoperable or metastatic NETs, had failed prior systemic therapy, or were unable to receive standard treatments. Additional key eligibility criteria included measurable disease at baseline according to Response Evaluation Criteria in Solid Tumors (RECIST) version (v) 1.1 (12) and an Eastern Cooperative Oncology Group performance status of 0 or 1. Adequate bone marrow, liver, renal, and coagulation function were also required.

Key exclusion criteria included any diagnosed thrombosis 12 months prior to the study, significant bleeding 3 months prior, significant cardiovascular disease (e.g., myocardial infarction or unstable angina), gastrointestinal disease affecting medication absorption, untreated or unstable central nervous system metastases, or other malignancy. Patients receiving prior antitumor treatments, including systemic medication, surgery or radical radiotherapy within 4 weeks before surufatinib initiation, or prior palliative radiotherapy within 2 weeks before the first dose of surufatinib, were also excluded.

All patients provided written informed consent, and the study was conducted in accordance with the International Conference on Harmonization Good Clinical Practice guidelines, the Declaration of Helsinki, and applicable local laws and regulations. The study was also approved by the institutional review boards of participating centers. This study was registered at ClinicalTrials.gov (NCT02267967).

Patients received surufatinib 300 mg as an initial dose, once daily (QD) and continuously for every 28-day cycle until disease progression, intolerable toxicity, or withdrawal of consent. Treatment beyond progression was allowed at the discretion of investigators for patients who showed radiologic progression only, but who were otherwise experiencing clinical benefit. Tumor response was assessed by investigators, per RECIST v1.1, every 4 weeks for the cycles 1 to 2, then every 8 weeks for the first year, and every 12 weeks thereafter. Patients without disease progression upon surufatinib discontinuation were followed for tumor assessments until initiation of new antitumor treatment, loss to follow-up, withdrawal of consent, or death. Tumor responses were also retrospectively evaluated by a qualified independent radiologist.

Surufatinib treatment could be temporarily interrupted for up to 28 days if there were intolerable toxicities. Surufatinib dose could be reduced to 250 mg and then to 200 mg if grade ≥3 adverse events (AE) occurred. Concomitant antitumor medications were not allowed, except short-term somatostatin analogues for NET-related symptoms.

Clinical and laboratory evaluations were performed every 2 weeks for cycles 1 to 2, then every 4 weeks thereafter. Cardiographs and echocardiograms were conducted every 4 and 12 weeks, respectively. AEs were continually assessed using the Common Terminology Criteria for Adverse Events (CTCAE) v4.03.

Pharmacokinetic (PK) sampling was performed for all patients on days 1, 2, 14, and 15 of cycle 1, and on day 1 of cycles 5 and 7. Following consent, blood samples for biomarker analysis were collected at baseline and at each tumor evaluation visit. Plasma concentrations of VEGF-A (QVE00B, QuantiGlo), sVEGFR-2 (DY357), bFGF (HSFB00D), M-CSF (SMC00B; all R&D Systems), and FGF23 (CY-4000, Kainos) were measured by ELISA.

The primary endpoints were investigator-assessed safety and objective response rate (ORR). ORR was defined as the proportion of patients whose best tumor response was complete response (CR) or partial response (PR) during study therapy per RECIST v1.1. A confirmatory assessment of response was required ≥4 weeks after initial assessment. If a patient's initial CR or PR was not confirmed at the subsequent assessment, stable disease (SD) was assigned as his/her best overall response, provided SD had been demonstrated ≥6 weeks since first dosing. Safety was evaluated through AEs according to CTCAE v4.03.

Secondary endpoints included disease control rate (DCR), duration of response (DoR), PFS and PKs. DCR was defined as the proportion of patients who had a CR, PR, or SD as best response. DoR was defined as the time from the first documented evidence of CR or PR until the time of the first documented disease progression or death (any cause), whichever occurred first. PFS was defined as the interval between the first surufatinib dose and the earliest date of disease progression or death (any cause).

The recruitment plan was to enroll 80 patients (40 patients in each cohort). For a given AE with a true rate of 10%, 5%, or 1%, the probability of observing ≥1 such AE in 80 patients is >99%, 98%, and 55%, respectively. For efficacy evaluation, further investigation was deemed worthy if the observed ORR was 15% in the pancreatic NET cohort and 10% in the extrapancreatic NET cohort, according to the prior literature in this population (5, 13, 14). With a sample size of 40 patients in each cohort, observed ORRs of 15% and 10% would result in exact binomial 95% CI of 5.7%–29.8% and 2.8%–23.7%, respectively.

Patients who received ≥1 dose of surufatinib were included in safety and efficacy analyses. Descriptive statistics and graphs were used to summarize demographics, tumor characteristics, toxicity, and tumor response. PFS and DoR were analyzed using the Kaplan–Meier method. PK parameters were calculated by Pharsight Phoenix WinNonlin (Certara) software using a noncompartmental analysis model. A Wilcoxon signed-rank test compared changes in soluble protein levels. Multiple covariate analyses were performed for the association between baseline biomarker levels and PFS. Stepwise selection was carried out at the entry and retaining level of 0.1, the candidate covariates included disease category (i.e., pancreatic NETs and extrapancreatic NETs), all biomarker tested, and interactions between each biomarker with disease category. Statistical analyses of safety, efficacy, and the association between PFS and biomarkers were performed using SAS software (v9.3).

Between November 6, 2014, and January 8, 2016, 81 patients were enrolled into the trial. Forty-two patients were diagnosed with pancreatic NETs and 39 with extrapancreatic NETs. The median follow-up was 14.7 months (range, 0.6–30.2) as of the cutoff date (August 23, 2017), when the maturity of PFS data was 60% [47 progressive disease (PD) and two death events]. At data cutoff, 16 (20%) patients remained on study treatment (11 in the pancreatic NET cohort, five in the extrapancreatic NET cohort). Reasons for treatment discontinuation included PD (41 patients, 51%), intolerable AEs (18 patients, 22%), withdrawal of consent (5 patients, 6%), and major protocol deviation (1 patient, 1%). All 81 patients received study medication and were included in the safety and efficacy analyses (Supplementary Fig. S1).

Patient demographics and baseline disease characteristics in each cohort are listed in Table 1. Overall, the median age was 49 years (range, 20–71), 44 (54%) patients were men, 65 (80%) had grade 2 NETs, and 79 (98%) had distant metastases. Fifty-seven (70%) patients had received systemic antitumor drugs prior to enrollment. Most patients had PD, with 64 (79%) patients recording disease progression within 1 year before enrollment; the other 17 (21%) patients experienced clinical deterioration, but without documented radiologic progression because of new diagnosis or irregular follow-up in the prior year.

The median treatment duration was 13.7 months (range, 0.1–30.2) for all patients [17.9 months (range, 0.1–29.5) in the pancreatic NET cohort and 9.2 months (range, 0.5–30.2) in the extrapancreatic NET cohort). During treatment, 3 patients concomitantly received octreotide for symptom control: 2 patients used short-acting octreotide for 3 and 4 days, respectively, and one used octreotide acetate microspheres (long-acting release) for 2 cycles.

Tumor evaluations are listed in Table 2. In the pancreatic NET cohort (n = 42), 8 patients had confirmed PR and 30 had SD, producing an ORR of 19% (95% CI, 9%–34%) and a DCR of 91% (95% CI, 77%–97%). In the extrapancreatic NET cohort (n = 39), 6 patients had confirmed PR and 30 had SD, producing an ORR of 15% (95% CI, 6%–31%) and DCR of 92% (95% CI, 79%–98%). Of the 6 patients who achieved PR in this cohort, the primary tumor origins were stomach, duodenum, jejunum, liver, rectum, and lung, respectively. For patients with PRs, the median DoR was 21.1 months (95% CI, 1.5–21.1) and 12.0 months (95% CI, 6.8–12.9) in the pancreatic NET and extrapancreatic NET cohorts.

Most patients experienced tumor shrinkage from baseline (Fig. 1), with a decrease of target lesions >10% in 25 (61%) and 22 (55%) patients in the pancreatic and extrapancreatic NET cohorts, respectively. Overall, of the 13 patients who progressed on previous antiangiogenesis treatment (i.e., sunitinib or famitinib), 2 had confirmed PR and 11 had SD (including one unconfirmed PR), with an ORR of 15% and DCR of 100%, and the median PFS was 13.8 months (95% CI, 5.6–24.8), all of which were comparable with those reported in the remaining patients.

The median PFS was 21.2 months (95% CI, 15.9–24.8) and 13.4 months (95% CI, 7.6–19.3) in the pancreatic NET and extrapancreatic NET cohorts (Fig. 2A). Among patients with radiologic disease progression within 1 year before treatment (n = 32 per cohort), the PFS was 21.2 months (95% CI, 15.9–24.8) and 11.1 months (95% CI, 7.2–16.6) in the pancreatic and extrapancreatic NET cohorts, respectively (Fig. 2B).

Tumor responses assessed retrospectively by the independent reviewer were consistent with investigators’ evaluations (Table 2). The ORR was 12% (95% CI, 4%–25%) and 10% (95% CI, 3%–24%) in the pancreatic and extrapancreatic NET cohorts, respectively; median PFS was 19.4 months (95% CI, 14.8–24.8) and 13.6 months (95% CI, 7.6–19.3), respectively.

All patients had at least one AE of any grade. In all 81 patients, the most common treatment-related AEs were proteinuria (81%), diarrhea (72%), hypertension (60%), elevated blood thyroid-stimulating hormone (52%), asthenia (44%), and elevated aspartate aminotransferase (41%; Table 3). Dermatologic reactions were less common; rash was observed in 3 (4%) patients, hand and foot syndrome in 2 (2%) patients and other dermatologic manifestations were only reported in one (1%) patient separately. The most common treatment-related grade ≥3 AEs were hypertension (33%), proteinuria (12%), hyperuricemia (10%), hypertriglyceridemia and diarrhea (6% each), and increased alanine aminotransferase (5%).

Dose interruption and reductions due to AEs occurred in 54 (67%) and 26 (32%) patients, respectively. AEs that resulted in treatment discontinuation were reported in 18 (22%) patients, including proteinuria (5%), abnormal hepatic function (2%), and other AEs (1%). Serious adverse events (SAE) were reported in 22 (27%) patients. SAEs experienced by more than one patient were abnormal hepatic function (5%), intestinal obstruction (4%), anemia, pancreatitis, and upper gastrointestinal hemorrhage (2% each).

Three (4%) patients experienced fatal AEs. One patient with rectal NET discontinued treatment due to PD on day 14 and died from hypovolemic shock, possibly related to surufatinib treatment 4 days after the last dosing. Fatal AEs in the other 2 patients were considered unlikely to be related to surufatinib; a patient with gastric NET died from a biliary tract infection and a patient with pancreatic NET died of multiorgan failure, 28 and 3 days after the last dosing, respectively.

The PKs of surufatinib were evaluated in 81 patients for day 1 and in 78 patients for day 14 (due to withdrawal of 3 patients; Supplementary Table S1). Following an oral dose of 300 mg on day 1, the geometric mean maximum concentration (C_max) of surufatinib was 376 ng/mL, and the median time to maximum concentration (T_max) was 2.0 hours. On day 14, the geometric mean C_max of surufatinib was 487 ng/mL, and the median T_max was 2.1 hours. The geometric mean areas under the concentration curve (AUC_0–24) were 2,770 and 4,810 hour × ng/mL, for days 1 and 14, respectively. The mean accumulation ratio was 1.91.

Plasma samples were available for 36 patients at baseline and posttreatment (17 patients with pancreatic NETs and 19 patients with extrapancreatic NETs). Surufatinib treatment induced significant increases in plasma VEGF-A, FGF23, and macrophage colony-stimulating factor (M-CSF), and decreases in soluble VEGFR-2 (sVEGFR-2) compared with baseline, indicating a potent blockade of VEGFR, FGFR1, and CSF-1R signaling (Table 4). No significant change in basic FGF (bFGF) levels was determined at the time of best response or PD compared with baseline. Interestingly, higher baseline levels of sVEGFR-2 and lower baseline levels of bFGF were significantly associated with prolonged PFS (Supplementary Table S2), but not in other biomarkers investigated.

In this single-arm, phase Ib/II study, surufatinib demonstrated manageable toxicity and promising antitumor activity in patients with NETs of diverse tumor origins. Surufatinib, as a single-drug therapy, is the first antiangiogenic drug to show robust antitumor activity in patients with extrapancreatic NETs. Patients with extrapancreatic NETs obtained a durable response, despite diverse primary tumor origins that correlate with poor prognosis (1, 15). Furthermore, 77% of patients in this cohort had grade 2 NETs and 25% had a Ki 67 >10%. These characteristics suggest that tumors in this study might have more malignant behavior as compared with those from previous clinical trials involving everolimus, octreotide, and other drugs. In those studies, the majority of extrapancreatic NETs were grade 1 originating from the small intestine, typically correlating with a better prognosis (1, 4, 13, 14, 16).

Additionally, surufatinib demonstrated potent antitumor activity in pancreatic NETs, including patients who progressed on prior VEGFR inhibitors, possibly owing to its different mechanism of action on the anti-VEGF pathway, anti-FGF pathway, and its ability to regulate the tumor immune microenvironment. The ORR appeared higher and PFS longer in the pancreatic NET cohort than the extrapancreatic NET cohort in this study, which has also been observed in other studies of antiangiogenic treatments (4, 6). To date, investigations characterizing the molecular abnormalities driving the growth of NETs have been limited. In pancreatic NETs, VHL alterations lead to increased levels of HIF1α and secondary overexpression of VEGF, indicating the importance of the VEGFR pathway, and mutations in TSC1/2 and PTEN result in activation of the PI3K/mTOR pathway, providing the rationale for mTOR-directed therapies (17). Furthermore, the genes implicated in chromatin remodeling have been found to be frequently mutated in pancreatic NETs, such as MEN1, DAXX, or ATRX, suggesting that DNA methylation and histone modifications may present new potential targets in pancreatic NETs (18, 19). In contrast, most small-intestine NETs are characterized by a loss of tumor suppressor genes on chromosome 18 and no other frequent genetic alterations or putative affected pathways have yet been implicated in the tumorigenesis of small-intestine NETs (18, 20). In general, the varied molecular profiles, biological and clinical behaviors of NETs from diverse tumor origins may provide an explanation for the different treatment responses, but further research is required to elucidate the mechanisms.

Efficacy assessments were mostly similar between investigators and the independent radiologist, although ORRs assessed by the independent radiologist were slightly lower in both cohorts. NETs usually present with multiple liver metastases, making evaluations challenging (21, 22). In this study, 93% and 82% of patients in both cohorts had liver metastases, most of which were multiple. Different target lesion selection for tumor assessment may have contributed to the discordance between the investigators and the independent radiologist. Despite the difference in ORRs, the antitumor activity of surufatinib was well demonstrated in both assessments. Given that prior studies have reported ORRs between 2% and 9%, DCRs between 69% and 83% and PFS of approximately 11 months (5, 13, 16, 23), surufatinib could present a promising candidate for the treatment of advanced NETs.

The safety profile of surufatinib was similar to other oral antiangiogenesis inhibitors (24), with no unexpected safety signals. Frequently reported grade ≥3 treatment-related AEs were hypertension (33%), proteinuria (12%), and hyperuricemia (10%), all manageable with standard supportive care and/or dose modifications. Treatment discontinuation due to AEs occurred in 22% of patients, due to hypertension and proteinuria in 1% and 5% of patients, respectively. No patient discontinued treatment due to hyperuricemia. It is worth noting that 22.2% of patients had hypertension, 39.5% proteinuria, and 6.2% hyperuricemia recorded as medical history at baseline. The safety and tolerability profiles of surufatinib seem acceptable in patients with advanced NETs, who usually have a long treatment period (median treatment duration of 13.7 months in this study).

Five circulating proteins were evaluated for target inhibition and associations with clinical outcomes. Changes in VEGF-A, sVEGFR-2, FGF23, and M-CSF with surufatinib are in line with previous reports regarding other selective VEGF, FGF, and CSF-1 pathway inhibitors (25–27), supporting the notion that these proteins are putative pharmacodynamic biomarkers of the related targets. Notably, bFGF levels did not significantly change after treatment. This contrasts with observations of significant bFGF elevation at tumor progression following treatment with other antiangiogenesis inhibitors, albeit in different cancer types, suggesting FGF/FGFR activation as a possible mechanism for acquired resistance (28–30). As surufatinib strongly attenuated FGFR1, we speculate that FGF/FGFR signaling might not be the dominant mechanism for resistance to surufatinib. The higher sVEGFR-2 levels and lower bFGF levels at baseline were associated with longer median PFS, which has also been reported in previous studies of other anti-VEGF/VEGFR agents in several cancer types, including metastatic NETs (31–34). Due to the exploratory and retrospective nature of the biomarker analysis, validating these results in a larger population is required.

This trial had several limitations. It was a single-arm study without a control group; hence, these findings must be interpreted with caution. Post-study antitumor treatments and survival status data after documented radiologic progression were not collected; consequently, overall survival information for this population is not available. Finally, the small sample size makes it difficult to correlate biomarkers with clinical efficacy for each cohort.

In summary, surufatinib demonstrated promising antitumor activity with an acceptable safety profile in patients with advanced, well-differentiated NETs, regardless of tumor origins. The treatment benefit of surufatinib will be confirmed in two phase III double-blinded, randomized, placebo-controlled trials in patients with pancreatic (NCT02589821) and extrapancreatic NETs (NCT02588170).

C. Qi, Y. Hua, Y. Ren, and W. Su hold ownership interest (including patents) in Hutchison MediPharma. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Xu, C. Bai, Y. Hua, C. Qi, Y. Ren, W. Su

Development of methodology: J. Xu, C. Bai, C. Qi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Xu, J. Li, C. Bai, N. Xu, Z. Zhou, Z. Li, C. Zhou, R. Jia, M. Lu, Y. Cheng, C. Mao, W. Wang, C. Su, C. Qi

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Xu, C. Zhou, Y. Hua, C. Qi, J. Li, W. Wang, K. Li, Q. Sun, Y. Ren, W. Su

Writing, review, and/or revision of the manuscript: J. Xu, J. Li, N. Xu, Z. Zhou, C. Zhou, R. Jia, Y. Cheng, W. Wang, C. Su, Y. Hua, J. Li, K. Li, Y. Ren, W. Su

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Z. Zhoui, C. Zhou, R. Jia, Y. Cheng, W. Wang, K. Cheng, C. Su

Study supervision: J. Xu, J. Li, N. Xui, C. Zhou, C. Qi

We thank the patients for participating in this trial and all the investigators and site personnel involved in the study. Editorial assistance in manuscript preparation was provided by Nucleus Global, Shanghai, China, funded by Hutchison MediPharma. This study was supported by Hutchison MediPharma (Shanghai) Ltd.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.