A randomised Phase II trial of carboplatin and gemcitabine ± vandetanib in first‐line treatment of patients with advanced urothelial cell cancer not suitable to receive cisplatin

To assess the efficacy and tolerability of the dual epidermal growth factor receptor/vascular endothelial growth factor receptor inhibitor, vandetanib, in combination with carboplatin and gemcitabine in the first‐line treatment of patients with advanced transitional cell carcinoma urothelial cancer (UC) who were unsuitable for cisplatin.


Introduction
There are~10 000 patients newly diagnosed with urothelial cancer (UC) in the UK [1] and 118 000 in Europe [2] per annum. Around 38% die within 1 year of diagnosis. The majority of UC deaths are caused by locally advanced or metastatic invasive bladder cancer. Advanced UC is a chemosensitive disease with response rates to cisplatincontaining regimens in previously untreated patients of~55% and with a median overall survival (OS) in the region of 14 months [3,4]. However, cisplatin-based chemotherapy is not suitable for~40% of patients [5], due to reasons such as insufficient renal function, performance status or comorbidity. Much of the focus of clinical trials has been on improving outcomes in the cisplatin-fit population, but there is also a need to improve outcomes in the sizable minority of patients currently treated with non-cisplatin containing regimens.
For many of these patients, the standard of care is a combination gemcitabine plus carboplatin (GC) chemotherapy, giving a median progression-free survival (PFS) of between 4.8 and 5.3 months [6][7][8][9]. UC frequently expresses a variety of growth factor receptors, including epidermal growth factor receptors (EGFRs) and vascular endothelial growth factor receptors (VEGFRs) [10][11][12]. Overexpression of VEGF and its receptors in UC was associated with poor prognosis [13,14], suggesting a role for VEGF/ VEGFRs in pathogenesis and potential clinical utility for molecularly targeted agents directed against these cell-surface receptors.
Vandetanib (ZD6474; Caprelsaâ) is an oral tyrosine kinase inhibitor selective for VEGFR-1, VEGFR-2, VEGFR-3, and EGFR. Preclinical data have shown that vandetanib induced cell death in vitro at clinically meaningful concentrations in several UC cell lines, and that this effect was synergistic with platinum-containing chemotherapy agents [15]. In vivo, pharmacological inhibition of EGFR or VEGFR had antitumoral effects in carcinogen-induced and orthotopic models of bladder cancer, respectively [16,17]. Vandetanib has shown efficacy as a single agent in clinical trials for medullary thyroid cancer [18], and in combination with docetaxel in the second-line treatment of locally advanced or metastatic nonsmall cell lung cancer [19].
Our hypothesis was that co-targeting both EGFRs and VEGFRs may improve survival outcomes in patients with advanced UC who are not suitable to receive cisplatin as firstline treatment. The primary goal of the TOUCAN trial was to establish whether vandetanib combined with GC chemotherapy is safe and gives sufficient activity to warrant a future Phase III trial in this patient group. Patients were eligible if they were aged ≥18 years; had histologically confirmed UC with TCC (pure or mixed histology); had radiologically measurable, locally advanced and/or metastatic disease [Response Evaluation Criteria In Solid Tumors (RECIST), version 1.1]; were not amenable to curative treatment with surgery or radiotherapy; and were not suitable for cisplatin. Unsuitability for cisplatin was defined as one or more of the following: (i) creatinine clearance <60 mL/min; (ii) Eastern Cooperative Oncology Group Performance Status (ECOG PS) =2 (N.B. patients are excluded if ECOG PS was >=3) ; (iii) clinically significant ischaemic heart disease; (iv) prior intolerance of cisplatin; (v) age >75 years; (vi) any other factor, which, in the opinion of the investigator indicated that cisplatin was not suitable. Patients were also ineligible for the trial if their creatinine clearance was <30 mL/min. All patients provided written informed consent.

Patients and methods
The patients were randomly assigned (1:1) either to vandetanib or placebo in addition to GC with stratification by institution, ECOG PS (0-1 and 2) and renal function (creatinine clearance <60 vs ≥60 mL/min, calculated using the Cockcroft and Gault formula) using a central interactive web response system. Clinical and radiological response assessments were performed at weeks 9, 18, 26, 39 and 52 after the commencement of treatment, with radiological response assessed by comparison with baseline data, according to RECIST, version 1.1.

Endpoints
The primary endpoint was PFS, defined as the time from randomisation to disease progression and/or death. Those still alive and progression free were censored at the date last seen. Secondary endpoints included: safety, assessed via real-time serious adverse event (AE) reporting and at patients visits using the Common Terminology Criteria for Adverse Events (CTCAE), version 4.0; tolerability and feasibility of use of treatment, which was assessed by calculating the number of treatment reductions, delays and treatment withdrawals; objective response, which was derived from radiological assessments and; OS, defined as the time from randomisation to death from any cause, censoring for those still alive at time last seen.

Statistical analysis
This was a placebo-controlled, randomised screening Phase II trial. The median PFS in patients with advanced UC not suitable for cisplatin was estimated to be~5.3 months with chemotherapy. The sample size was calculated assuming 24month recruitment, with 80% power and a one-sided a (type I error) of 0.2. Allowing a drop-out rate of 5%, 82 participants and 62 (PFS) events were required to demonstrate a hazard ratio (HR) of 0.65, based on the logrank test. The data were analysed after 65 events were observed. A Phase III confirmatory trial was to be planned if there was statistical significance at the 10% level. In the event of statistical significance between 10.1% and 20%, a confirmatory trial was planned only if secondary endpoints indicated benefit.
At the end of the trial, analyses were performed on both intention-to-treat (ITT) and planned-per-protocol analysis (PPA) basis. The PPA excluded patients found to be ineligible or who did not start their trial medication during cycle 1.
Kaplan-Meier curves of PFS and OS were plotted and these were used to calculate the median PFS and OS for each arm. The Mantel-Cox version of the log-rank test (unadjusted) was used to assess the effect of vandetanib on PFS and OS. In addition, a planned adjusted analysis for the primary endpoint of PFS using a Cox proportional hazards model, including the stratification factors used in the randomisation, was performed (i.e., ECOG PS and renal function, with institution included in the model as a shared frailty). The secondary endpoints were presented as the proportion (and 95% CI) of patients in each treatment arm with: (i) an objective disease response (based on RECIST v1.1); (ii) Grade ≥3 toxicity; and (iii) a treatment reduction, delay and treatment withdrawal. No subgroup analyses were performed. There were no pre-defined early stopping guidelines.
An independent safety committee reviewed the trial throughout; including formal safety reviews after the first 10 and 20 patients in each arm had been recruited.

Baseline patient and tumour characteristics
The 82 patients were randomised from 16 hospitals across the UK, between April 2011 and December 2014. A Consolidated Standards of Reporting Trials (CONSORT) diagram is shown in Fig. 1. Baseline patient and tumour characteristics were similar in the two treatment arms and are presented in Table 1.

Treatment tolerability and feasibility of use
In all, 13 (33%) of the 40 participants assigned to receive vandetanib and 25 (60%) of the 42 assigned to placebo received all six cycles of GC. The median (interquartile range [IQR]) numbers of cycles received were 5 (3.5-6) and 6 (4-6), in the vandetanib and placebo arms, respectively. Patients received a median (IQR) of 88 (46-116) days of vandetanib and 105 (63-126) days of placebo. Eight (20%) of the 40 participants in the vandetanib arm received all six cycles of vandetanib; 22 (55%) failed to do so because of toxicity, three (8%) because of progression, three (8%) due to patient choice, two (5%) due to death, and two (5%) for unknown reasons. In all, 17 (41%) of the 42 patients in the placebo arm received all six cycles of placebo; 14 (33%) did not do so due to toxicity, seven (17%) due to disease progression, one (2%) by patient choice, one (2%) due to death, and two (5%) for unknown reasons.

Safety
Similar rates of treatment-emergent AEs were seen in the two arms, with Grade 3-4 toxicities seen in 80% of patients in the vandetanib arm and 76% in the placebo arm. The commonly occurring toxicities are summarised in Table 2.
In all, 32 patients (80%) in the vandetanib arm and 38 (90%) in the placebo arm were evaluable for radiological response.
On an ITT basis, responses (complete response + partial response) were seen in 20 of 40 patients (50%) in the vandetanib arm and 23 of 42 (55%) in the placebo arm. The change in size of measurable lesions at first protocolmandated response assessment (week 9 after commencement of treatment; 64 evaluable patients) is presented as a waterfall plot in Fig. 3. A total of 48 patients (58%) had died by the time of analysis: 42 deaths were disease related, two were treatment related, and four due to other reasons. The median OS for the vandetanib arm was 10.8 months (95% CI 8.0-13.0) and for the placebo arm was 13.8 months (95% CI 11.1-16.6). Comparison of OS gave an unadjusted HR of 1.41 (95% CI 0.79-2.52; one-sided P = 0.9), using an ITT analysis, which represents a 41% increase in risk of death after the addition of vandetanib (Fig. 2B). There was a corresponding absolute reduction of 10.6% in the 1-year OS in the vandetanib arm (54.4% in the vandetanib arm vs 65.0% in the placebo arm) (95% CI 31.2-6.0%).

Discussion
The goal of this trial was to assess the safety and efficacy of vandetanib in combination with GC chemotherapy. Based on this trial, although this combination was found to be safe, there was no evidence that this combination improved clinical outcomes in this cohort of patients with advanced UC who were unsuitable for cisplatin. These data are consistent with a  previous report that vandetanib does not improve efficacy when combined with docetaxel in patients receiving secondline treatment of advanced UC [20]. Our present results are in contrast to the small benefits seen with vandetanib in combination with docetaxel in non-small cell lung cancer [19] and with a recently reported randomised Phase III trial using another anti-VEGFR agent, ramucirumab (a fully human anti-VEGFR2 monoclonal antibody), which demonstrated improved PFS and response rates in combination with docetaxel for patients with UC in the second-line setting [21]. However, the data presented in the present study are consistent with other trials exploring the efficacy of drugs targeting the VEGFR pathways in UC [22], with most agents tested having been found to have insufficient activity to take to Phase III. Recent results of a randomised Phase III trial of gemcitabine plus cisplatin with or without bevacizumab in the first-line setting have shown an improvement in PFS with the addition of bevacizumab, but no improvement in the primary endpoint of OS (ClinicalTrials.gov Identifier: NCT00942331; www.clinicaltrials.gov) [23]. Similarly, EGFR-targeted therapies have shown insufficient activity in both biomarkerselected and -unselected patients with UC. Notably, a Phase III trial of the dual EGFR inhibitor, lapatinib (LaMB), in patients selected for EGFR and/or human epidermal growth factor receptor 2 (HER2) expression, failed to show activity in advanced UC following first-line chemotherapy [24].
One explanation for the lack of efficacy of vandetanib in the present trial might have been failure to select patients appropriately. Predictive markers for VEGF-targeted therapies have, to date, been elusive in other disease and valid predictive markers for EGFR-targeted therapies are not sufficiently prevalent in UC [25].
Despite the failure to demonstrate incremental benefit from vandetanib, the overall outcomes for patients in the present trial were better than expected. Notably, in both arms of the present study the median PFS (6.8 vs 8.8 months) and OS (10.8 vs 13.8 months) were better than seen in a previous Phase III trial of GC chemotherapy in a similar group of patients (PFS 5.8 months and OS 9.3 months) [6]. This may reflect patient selection, as patients needed to be considered suitable for combination treatment by investigators or could be due to the use of a wider definition of 'unsuitable for cisplatin'. We cannot be certain as to the reasons for the apparent trend towards poorer survival amongst those receiving vandetanib, but it could have been due to reduced exposure to chemotherapy seen in this arm. In addition, there were some imbalances in baseline characteristics between the arms.
After many years in which various combinations of smallmolecule cytotoxic drugs and/or molecularly targeted drugs have failed to achieve substantial improvements in survival outcomes in advanced UC, significant interest has recently been generated in the use of immunotherapies, including a report of improved survival with the anti-programmed cell death-1 (PD-1) monoclonal antibody, pembrolizumab, as second-line therapy for metastatic UC [26]. Our response rate (50%) and OS data (median OS 10.8 months) in the vandetanib arm are not dissimilar to those seen in a trial of pembrolizumab in the same setting (first-line treatment of patients with metastatic UC unsuitable for cisplatin), which reported an objective response rate (ORR) of 24% and a median OS of 11.5 months [27] and with a single-arm trial of the anti-PD-L1 monoclonal antibody, atezolizumab, which demonstrated an ORR of 23% and a median OS of 15.9 months [28]. Several Phase III trials comparing GC with immunotherapy in the population unsuitable for cisplatin are currently unreported (ClinicalTrials.gov Identifier: NCT02516241; NCT02853305; NCT02807636; www.clinica ltrials.gov).
In conclusion, there is no evidence that the addition of vandetanib to GC chemotherapy improves clinical outcomes. Our present findings do not support a Phase III study or its use as first-line treatment in patients with UC who are unfit for cisplatin.