Efficacy of the nanoparticle–drug conjugate CRLX101 in combination with bevacizumab in metastatic renal cell carcinoma: results of an investigator-initiated phase I–IIa clinical trial

The combination of the campothecan-containing nanoparticle–drug conjugate, CRLX101 and bevacizumab is well tolerated, safe, and active in a heavily pretreated population of patients with metastatic renal cell carcinoma of clear and non-clear cell histology.

Abstract

Background

Anti-angiogenic therapies are effective in metastatic renal cell carcinoma (mRCC), but resistance is inevitable. A dual-inhibition strategy focused on hypoxia-inducible factor (HIF) is hypothesized to be active in this refractory setting. CRLX101 is an investigational camptothecin-containing nanoparticle–drug conjugate (NDC), which durably inhibits HIF1α and HIF2α in preclinical models and in gastric cancer patients. Synergy was observed in the preclinical setting when combining this NDC and anti-angiogenic agents, including bevacizumab.

Patients and methods

Patients with refractory mRCC were treated every 2 weeks with bevacizumab (10 mg/kg) and escalating doses of CRLX101 (12, 15 mg/m²) in a 3 + 3 phase I design. An expansion cohort of 10 patients was treated at the recommended phase II dose (RP2D). Patients were treated until progressive disease or prohibitive toxicity. Adverse events (AEs) were assessed using CTCAE v4.0 and clinical outcome using RECIST v1.1.

Results

Twenty-two patients were response-evaluable in an investigator-initiated trial at two academic medical centers. RCC histologies included clear cell (n = 12), papillary (n = 5), chromophobe (n = 2), and unclassified (n = 3). Patients received a median of two prior therapies, with at least one prior vascular endothelial tyrosine kinase inhibitor therapy (VEGF-TKI). No dose-limiting toxicities were observed. Grade ≥3 AEs related to CRLX101 included non-infectious cystitis (5 events), fatigue (3 events), anemia (2 events), diarrhea (2 events), dizziness (2 events), and 7 other individual events. Five of 22 patients (23%) achieved partial responses, including 3 of 12 patients with clear cell histology and 2 of 10 patients (20%) with non-clear cell histology. Twelve of 22 patients (55%) achieved progression-free survival (PFS) of >4 months.

Conclusions

CRLX101 combined with bevacizumab is safe in mRCC. This combination fulfilled the protocol's predefined threshold for further examination with responses and prolonged PFS in a heavily pretreated population. A randomized phase II clinical trial in mRCC of this combination is ongoing.

introduction

Most new cases of renal cell carcinomas (RCCs) diagnosed in the United States are of clear cell (ccRCC) histology [1]. Inactivation of the von Hippel Lindau tumor suppressor gene in ccRCC leads to higher intracellular levels of hypoxia-inducible factors 1α and 2α (HIF1α and HIF2α), and in tumorigenesis and disease progression of many RCCs, aberrant angiogenesis is an early pathophysiologic step [2]. Multiple VEGF tyrosine kinase inhibitors (TKIs), mammalian target of rapamycin inhibitors, and bevacizumab, a monoclonal antibody to circulating VEGF, are approved for the treatment of metastatic RCC (mRCC) [3]. Although benefits of progression-free survival (PFS) and overall survival (OS) are achieved with these agents, patients with mRCC have a 5-year survival rate of <12% [1], and there remains a need for better therapies with novel mechanisms.

CRLX101 is an investigational self-assembling nanoparticle–drug conjugate (NDC) containing the payload camptothecin. CRLX101 preferentially accumulates in tumor tissue through the enhanced permeability and retention effect through which nanoparticles accumulate inside tumors due to their leaky vasculature and reduced lymphatic drainage [4]. Once inside tumor cells, camptothecin is released from CRLX101 over several days with a plasma half-life of ∼24 h compared with <2 h for naked camptothecin [5]. The slow release of CRLX101 provides prolonged drug exposure within tumors and sustained topoisomerase-I inhibition, while limiting the level of freely circulating camptothecin and thereby reducing systemic toxicity [5]. Of note, durable inhibition of topoisomerase-1 has been shown preclinically to result in the inhibition of HIF1α [6]. For example, a clinically relevant dose of CRLX101 inhibited both HIF1α and HIF2α in a sustained fashion, a result that was not possible with a high dose of topotecan [4]. Furthermore, CRLX101 is synergistic with bevacizumab, aflibercept, and pazopanib in preclinical models, improving both tumor growth inhibition and survival versus the respective monotherapies [5]. Approximately 350 patients have been treated with CRLX101 alone or in combination in different tumor types with evidence of antitumor activity and acceptable toxicity [7, 8]. These preclinical and clinical data led to the investigation of combination of CRLX101 and bevacizumab in mRCC refractory to targeted therapy.

methods

patients

Patients were recruited from the Abramson Cancer Center at the University of Pennsylvania (Philadelphia, PA, USA) and the Kimmel Cancer Center at Thomas Jefferson University Hospital (Philadelphia, PA, USA). Eligible patients 18 years and older and with ECOG performance status 0 or 1 had histologically confirmed metastatic or locally advanced unresectable RCC. Tumors needed to be measurable by the Response Evaluation Criteria in Solid Tumors guidelines (RECIST) v1.1 with disease sites assessed within 4 weeks of study entry [9]. No prior bevacizumab or topoisomerase-1 inhibitor was allowed and patients must have received at least one prior conventional therapy (defined as pazopanib, sorafenib, sunitinib, temsirolimus, axitinib, or everolimus) in a non-adjuvant setting. A 2-week washout period was required between the last dose of prior therapy and baseline correlative studies (including bone marrow biopsy as applicable) and toxicities from prior therapy must have declined to ≤grade 1 before initiation of study therapy. Prior treatment with surgery, radiation, conventional chemotherapy, or cytokine therapy was allowed. Patients with treated brain or spinal-associated metastases were eligible but must have had their disease controlled, off steroids, and be neurologically stable.

study design and treatment

Patients in this open label, industry-supported, investigator-initiated, single-arm phase I–IIa clinical trial were enrolled in a non-randomized sequential fashion. The first in human phase I trial explored CRLX101 monotherapy weekly at 6, 12, and 18 mg/m² and later bi-weekly at 12, 15, and 18 mg/m² i.v. [10]. Starting at a CRLX101 dose one level below the single-agent recommended phase II dose level of 15 mg/m² i.v. bi-weekly, we planned in the phase I dose-finding stage to enroll a minimum of three patients per dosing cohort at the following dose levels: 1 (bevacizumab 10 mg/kg i.v. and CRLX101 12 mg/m² i.v. every 2 weeks) and 2 (bevacizumab 10 mg/kg i.v. and CRLX101 15 mg/m² i.v. every 2 weeks). No dosing of CRLX101 above 15 mg/m² i.v. every 2 weeks (the drug's monotherapy MTD) was planned.

Before any dose escalation, at least three patients needed to be evaluable for the occurrence of a dose-limiting toxicity (DLT). Dose-limiting was defined as non-hematologic toxicity including grade ≥3 nausea, vomiting, and/or diarrhea, despite maximal prophylaxis and/or treatment, and any grade 3 or 4 non-hematologic toxicity not easily corrected with medical management; hematologic toxicity including grade 4 neutropenia of duration ≥5 days or grade ≥3 febrile neutropenia or any other grade 4 hematologic adverse events (AEs); a study treatment delay of ≥7 days due to an AE or inability to commence cycle 2 within 14 days of the planned start date due to an AE; and a treatment-emergent AE warranting a dose reduction that, in the opinion of the investigator, was of clinical significance such that further dose escalation would expose patients to unacceptable risk.

The DLT served as the basis for dose escalation and the determination of a maximum tolerated dose or recommended phase II dose (MTD/RP2D). For all dosing cohorts evaluated, if a single DLT was observed among the initial three patients evaluated, then additional three patients would be enrolled. If one or more of the additional three patients experienced a DLT, subsequent dose escalation was stopped. If none of the additional three patients experienced a DLT, dose escalation was continued or an MTD established (if already at dose level 2). If two or more DLTs were observed at dose level 1, an examination of dose level 1 (bevacizumab 10 mg/kg i.v., and CRLX101 9 mg/m² i.v. every 2 weeks) was anticipated. Intra-patient dose escalation to a higher dose level or to the RP2D could occur at the commencement of the following treatment cycle if maximum toxicity experienced by the patients at the current dose level was grade ≤2 and at least three patients at the higher dose level had completed at least one cycle of treatment without DLT. Dose-escalated patients could not contribute DLT events at the higher dose level.

At least, 12 patients were required to be treated in the dose-finding stage. If at least 4 of these 12 patients achieved PFS of 16 weeks or greater, an additional 10 patients were treated with the RP2D regimen.

Patients who signed an informed consent document and received study treatment were considered enrolled. Replacement of patients could occur if a consented patient discontinued from study treatment for non-safety reasons before commencement of treatment with a second cycle of therapy. Patients who failed to complete cycle 1 because of treatment-related AEs were not replaced.

end points and statistical considerations

The primary objective of this clinical trial was to determine the recommended phase II dose (RP2D) of CRLX101 combined with standard dose bevacizumab for the treatment of patients with mRCC. Secondary objectives included describing the safety and tolerability of this combination and estimating the PFS and objective response rate.

PFS end points and assumptions on this study were based on the RECORD-1 trial demonstrating a PFS in second-line mRCC of 4.0 months [11]. Thus, an end point assessing the fraction of patients treated with the combination of CRLX101 and bevacizumab surviving without RECIST-based progression of disease at 16 weeks from commencement of treatment (16-week PFS) was selected. The null hypothesis was that the fraction of patients experiencing 16-week PFS would be 50% or less. A preliminary test for futility was used employing data from the dose-finding stage (initial 12 patients treated on study). Using a modified Gehan two-stage design, the study would be terminated for lack of efficacy if three or fewer of the initial 12 patients achieved 16-week PFS, giving an upper 85% confidence boundary for 16-week PFS of <50%. Were ≥4/12 patients to achieve ≥16-week PFS, an additional 10 patients would be enrolled and treated at the RP2D for a total of 22 assessable patients for the 16-week PFS end point. With this sample size, the 95% confidence interval (CI) for 16-week PFS would be assumed to extend no further than ±11% from the observed rate. The combination would be viewed as active and worthy of further examination if the estimated rate of 16 week PFS is ≥50%. An additional secondary analysis of the median PFS time would be carried out through an estimation of PFS distribution by the Kaplan–Meier curves.

exploratory biomarker analysis

Bone marrow aspirates and biopsies were obtained at baseline and after 8 weeks of combined therapy in the first six patients enrolled (phase I cohort 3 at 12 mg/m² and 3 at 15 mg/m² i.v.). Blood was collected from all patients before treatment initiation and on day 3 post-dosing (cycle 1, days 1 and 3), mixed thoroughly, and centrifuged at 2500 g for 15 min. Platelet-poor plasma was aliquoted, snap-frozen, and stored at −80°C until use. Plasma samples were analyzed for a panel of circulating biomarkers using a proprietary CIRAplex immunoassay platform (Aushon BioSystems). See supplementary methods, available at Annals of Oncology online.

results

patients

Twenty-seven patients were enrolled (22 men, 5 women) with an average age of 63.2 years. No DLTs occurred in the six-patient phase I cohort. Five patients were evaluated for safety and toxicity but excluded from the efficacy analyses because they were either treated with a suboptimal dose of CRLX101 at 12 mg/m² following identification of the higher RP2D of 15 mg/m² (n = 3) or failed to complete one cycle of therapy for reasons unrelated to the drug or drug toxicity (n = 2). Thus, 22 fully assessable patients were included in both the safety and efficacy analyses. Complete demographics for these patients are provided in Table 1. RCC histologies included clear cell (n = 12 patients), papillary (n = 5), chromophobe (n = 2), and unclassified (n = 3). Patients represented a refractory treatment population: 18 of 22 patients had received at least 2 prior regimens of therapy for their mRCC disease and 9 of 22 patients had received at least 3 prior regimens.

Table 1.

Patient and tumor characteristics

Clinical characteristics	Total, n = 22
Age in years, median (range)	63 (41, 79)
Gender
Female, n (%)	4 (18%)
Male, n (%)	18 (82%)
White race, n (%)	18 (82%)
Histology, n (%)
Clear cell	12 (55%)
Non-clear cell	10 (45%)
Papillary	5 (23%)
Chromophobe	2 (9%)
Other/non-specified	3 (14%)
Performance status, n (%)*
0	13 (59%)
1	9 (41%)
No. of prior therapies, n (%)
1	4 (18%)
2	10 (45%)
>2	8 (36%)

*ECOG performance status.

safety and identification of recommended phase II dose

CRLX101 15 mg/m² administered i.v. every 2 weeks in combination with standard dose bevacizumab 10 mg/kg i.v. was identified as the recommended phase II dose (RP2D) and further evaluated for safety in phase IIa. Among patients evaluable for safety (n = 27), 13 patients (48%) experienced at least one related grade 3–4 AE. These are given in Table 2. Grade 3–4 AEs considered by investigators to be at least possibly related to CRLX101 included non-infectious cystitis (5 events among 5 patients), fatigue (3 events among 3 patients), anemia (2 events among 2 patients), diarrhea (2 events among 2 patients), dizziness (2 events among 2 patients), and abdominal pain, dehydration, dental carries, edema, nausea, neutropenia, and vomiting (1 event each). All other AEs considered at least possibly related to CRLX101 were grade 1 or 2.

Table 2.

Adverse events possibly related to CRLX101

Preferred terms	G3 and 4, n = 27 (%)	All grade, n = 27 (%)
Fatigue	3 (11.1%)	12 (44.4%)
Cystitis non-infective	5 (18.5%)	10 (37.0)%
Headaches		6 (22.2%)
Vomiting	1 (3.7%)	5 (18.5%)
Weight loss		5 (18.5%)
Constipation		4 (14.8%)
Diarrhea	2 (7.4%)	4 (14.8%)
Nausea	1 (3.7%)	4 (14.8%)
Edema limbs	1 (3.7%)	4 (14.8%)
Anemia	2 (7.4%)	3 (11.1%)
Anorexia		3 (11.1%)
Dizziness	2 (7.4%)	3 (11.1%)
Dyspnea		3 (11.1%)
Pruritus		3 (11.1%)
Abdominal pain	1 (3.7%)	2 (7.4%)
Urinary tract infection		2 (7.4%)
Back pain		2 (7.4%)
Hearing impaired		1 (3.7%)
Dry eye		1 (3.7%)
Bloating		1 (3.7%)
Dental caries	1 (3.7%)	1 (3.7%)
Dyspepsia		1 (3.7%)
Flatulence		1 (3.7%)
Mucositis oral		1 (3.7%)
Other—blood in stool		1 (3.7%)
Other—dry heaves		1 (3.7%)
Other—thrush		1 (3.7%)
Chills		1 (3.7%)
Fever		1 (3.7%)
Other—smell changes		1 (3.7%)
Neutrophil count decreased	1 (3.7%)	1 (3.7%)
WBC decreased		1 (3.7%)
Dehydration	1 (3.7%)	1 (3.7%)
Hypoalbuminemia		1 (3.7%)
Hypocalcemia		1 (3.7%)
Extremity pain		1 (3.7%)
Dysgeusia		1 (3.7%)
Agitation		1 (3.7%)
Confusion		1 (3.7%)
Hematuria		1 (3.7%)
Proteinuria		1 (3.7%)
Urinary frequency		1 (3.7%)
Hiccups		1 (3.7%)
Pleural effusion		1 (3.7%)
Voice alteration		1 (3.7%)
Dry skin		1 (3.7%)
Skin hyperpigmentation		1 (3.7%)

Myelosuppression was infrequent. A single grade 3 neutropenia event occurred in one patient for 7 days without fever and without intervention. No cases of thrombocytopenia were observed. Finally, bone marrow aspirates and biopsies obtained at baseline and after 8 weeks of combined therapy revealed no cellularity, morphology, or other immunohistochemical differences between patients treated with CRLX101 12 versus 15 mg/m².

Non-infectious cystitis was the AE of interest. Among 10 non-infectious cystitis events observed on study, all events were considered related to CRLX101 and these included grade 1 (n = 2), grade 2 (n = 3), and grade 3 (n = 5) events. Two of eight patients who stayed on study had recurrent episodes of cystitis. The onset of cystitis occurred as early as day 6 of cycle 1 and as late as day 21 of cycle 10. All cystitis events required CRLX dosing suspension and this occurred for an average of 39 days. Bevacizumab was continued throughout CRLX101 dosing suspensions for all patients. Two patients ultimately required study discontinuation: cystitis did resolve within a month of discontinuing therapy. All patients on this trial received vigorous hydration (>2 l). The cystitis was non-hemorrhagic and treated with hydration, antispasmodics, and non-steroidal anti-inflammatory agents. In patients (n = 1) undergoing cystoscopy, non-hemorrhagic inflammation was noted. We surmise that predisposition could be pharmacologic and that continuation of bevacizumab may have prolonged recovery from the chemical insult.

A complete listing of related AEs is provided in Table 2.

phase IIa and overall efficacy results

Twenty-two efficacy-evaluable patients received an average of 5.4 treatment cycles. Among these patients, 19 of 22 (86%) experienced either partial response (PR) or stable disease for more than 3 months, as their best response per RECIST v1.1. A waterfall plot demonstrating best tumor reductions is provided in Figure 1. Five patients (23%) achieved PR status (3/12 with clear cell histology and 2/5 with papillary histology). A single PR was observed among 4 patients receiving CRLX101 + bevacizumab as their second-line therapy for mRCC, while 4 of 18 third- to sixth-line patients experienced PRs (supplementary Table S1, available at Annals of Oncology online). Among all PRs, the best net tumor reductions ranged from 30% to 39% with an average response duration of 3.5 months (Figure 2). A spider plot describing the evolution of tumor reduction over time is provided in Figure 2.

Figure 1.

Waterfall plot of best net tumor change. Only 17 out of 22 patients are depicted in this plot as these patients have both baseline and at least one CT while on therapy.

Figure 2.

Spider plot of patients evaluable for response on therapy. Note that only 17 out of 22 patients are depicted in this plot, as these patients have both baseline and at least one CT scan while on therapy.

The secondary end point was PFS. As of the date of final PFS analysis (7/22/15), the median follow-up was 19 months. Eight patients came off the study before 16 weeks (the target PFS event) due to drug-related toxicity (n = 6) or fatigue intolerance (n = 2); one patient had rapid clinical progression. One additional patient remains on study with >1 year of PFS time at the time of final PFS analysis.

Among the patients evaluable for PFS (n = 22), the median time was 9.9 months: 11.2 months among second-line patients (n = 4) and 7.6 months among third- to sixth-line patients (n = 18). Of additional interest, 12 of 22 patients (55%) achieved PFS treatment time of 16 weeks or greater. Finally, multiple patients achieved tumor reductions and duration of treatment exceeding their experience with earlier lines of RCC therapy (supplementary Figure S1, available at Annals of Oncology online).

correlative studies

Pre- and post-treatment plasma samples were collected from all patients. Net reductions in levels of various proteins associated with poor outcome (including osteopontin, platelet-derived growth factor BB, Fibroblast Growth Factor-b, and TIE-2) were observed in patients who had decreases in tumor volume. Pharmacodynamic changes in these markers occurring between days 1 and 3 of cycle 1 are presented in supplementary Figures S2 A & B and S3, available at Annals of Oncology online.

discussion

The primary objectives of this study were to define the recommended phase II dose (RP2D) of CRLX101 (15 mg/kg) when administered in combination with standard dose bevacizumab (10 mg/kg) and to establish the safety of this drug–drug combination. Importantly, no additive toxicity of the combination was observed: concomitant bevacizumab administration did not exacerbate CRLX101-associated myelosuppression. The incidence of non-infectious cystitis from CRLX101 was similar in frequency and severity to those rates observed with CRLX101 monotherapy [10]. In this study, the cystitis was not gender-specific, nor related to the number of prior therapies, histology, or duration of therapy. Interruption of both bevacizumab and CRLX101 during cystitis might accelerate recovery and should be considered in future studies.

One mechanism of resistance to anti-angiogenic therapy is the induction of tumor hypoxia and up-regulation of HIF1α and HIF2α. The latter are correlated with epithelial mesenchymal transition and promotion of tumor stem cell production, angiogenesis, tumor invasiveness, and metastasis formation [2, 12–14]. Analogs of the topoisomerase-1 inhibitor, camptothecin, such as topotecan, administered with low daily dosing, have been shown to inhibit HIF1α protein levels [6, 15, 16] but are poorly tolerated. Alternatively, pegylated, liposomal, and polymeric nanoparticles of topoisomerase inhibitors, such as CRLX101, may provide a means of sustained drug exposure and extended inhibition of HIF1α protein accumulation [4]. Admittedly, another mechanism of activity could be direct DNA topoisomerase inhibition of rapid proliferation of cancer cells by CRLX101, similar to DNA damage seen with the use of gemcitabine or 5-Fluorouracil in mRCC [17, 18].

VEGF-targeted drugs demonstrate only modest success when administered to patients with progression through multiple lines of prior therapy [19]. For example, in a randomized phase III study comparing dovitinib to sorafenib in refractory mRCC, the median PFS was 3.7 months (95% CI 3.5–3.9) for dovitinib patients and 3.6 months (3.5–3.7) for sorafenib patients (hazard ratio 0·86, 95% CI 0.72–1.04; one-sided P = 0.063) [20]. The heavily pretreated patient population in our study demonstrated an ORR of 23% and an mPFS of 9.9 months. Also, nearly half of the patients (9/22) evaluable for efficacy achieved a PFS time which exceeded performance on their prior regimens of therapy. These results must be viewed in the context of the study's small sample size and uncontrolled design. The availability of other treatments for advanced RCC, including the immune checkpoint inhibitors [21] and dual met and VEGF inhibitors [22] makes selection of first- and second-line therapy quite complicated. Nevertheless, the combination of CRLX101 and bevacizumab is an attractive choice for patients with rapidly proliferating RCC or those intolerant of or progressing on VEGF-TKIs or who are not candidates for the immune checkpoint inhibitors. The activity across both clear and non-clear cell RCC is also attractive, and while the number of papillary RCC patients in our study was small, this population could be further explored.

The decrease in VEGFR resistance biomarkers supports the biologic hypotheses for combining CRLX101 and bevacizumab (see supplementary material, available at Annals of Oncology online). Furthermore, these results support that the addition of CRLX101 does not adversely modulate the benefits of bevacizumab monotherapy.

Based on the experience described here, a well-controlled, randomized phase II clinical trial comparing the combination of CRLX101 and bevacizumab with the investigator's choice of standard of care in advanced, unresectable mRCC patients post 2 or 3 regimens of prior therapy was launched in the second half of 2014 [23].

funding

This investigator-initiated trial was supported by Cerulean Pharmaceuticals. RM is supported by National Institutes of Health (grant K23-CA 187185).

disclosure

The following authors are employees of Cerulean Pharmaceuticals: SE, EGG, MH, LJ, AS, AP, and ABT. SMK is an employee of Merck & Co., Inc. All remaining authors have declared no conflicts of interest.