Phase I Trial of DNA Methyltransferase Inhibitor Guadecitabine Combined with Cisplatin and Gemcitabine for Solid Malignancies Including Urothelial Carcinoma (SPIRE)

Urothelial carcinoma accounts for approximately 550,000 new diagnoses and 200,000 deaths annually (1). Cisplatin-based combination chemotherapy is the standard-of-care therapy for urothelial carcinoma, for both radical perioperative treatment, and as palliative first-line treatment for advanced disease (2–4). Standard regimens for urothelial carcinoma combine cisplatin with gemcitabine, or methotrexate, vinblastine, and doxorubicin (5). For metastatic urothelial carcinoma, this results in a median survival and time to progressive disease of approximately 14 and 7 months, respectively (4). For locally advanced muscle invasive bladder cancer (MIBC), cisplatin-based chemotherapy contributes an absolute survival advantage of 5%–6% to overall cure rates (3). Cisplatin resistance remains a critical barrier to therapeutic advance in urothelial carcinoma (6). For example, in a key randomized trial comparing cisplatin-based regimens for advanced disease, by 3 years, only 13% were alive and progression free and 17% had primary refractory disease (7). Urothelial carcinoma progression or relapse is associated with a dismal prognosis. Second-line immunotherapy, or chemotherapy, after prior platinum-based treatment, results in median survival outcomes under 1 year (5).

Altered cancer gene expression may arise through structural genomic change, or as a result of reversible epigenetic regulation. Epigenetic control includes biochemical modifications, both to histone proteins within chromatin and also to DNA itself (8). DNA CpG dinucleotide methylation is the most widely studied cancer epigenetic change. Dysregulation of CpG methylation in cancer cells leads to genomic instability, activation of previously silent oncogenes, or silencing of tumor suppressor genes (TSGs). In solid malignancies, many genes undergo promoter hypermethylation. Hypermethylation reversal, for example, through DNA methyltransferase inhibition, allows TSG reexpression with potential for anticancer therapy.

Abnormal DNA methylation patterns exist in urothelial carcinoma, associated with disease phenotype (stage, grade, and histology) and clinical outcomes (9). Hyper- and hypomethylation are associated, respectively, with invasive and noninvasive tumors, potentially through FOXA1 activation, indicating an epigenetic divergence, in addition to a genetic distinction, between lethal and nonlethal urothelial carcinoma (10–12). Various gene targets, miRNAs, and mirtrons have been associated with cisplatin resistance and a poor prognosis when hypermethylated in urothelial carcinoma (13–16). An epigenetic field defect, characterized by hypermethylation, has also been described in normal bladder from patients with urothelial carcinoma that is hypothesized to predispose to carcinogenesis (12). DNA methylation patterns are also linked to cisplatin resistance in preclinical urothelial carcinoma models, and in other cancers, and have been validated in translational studies (14, 17–20). Cisplatin resistance through epigenetic mechanisms may be associated with specific marks, such as HOXA9 promoter methylation, and to cell subset phenotype, such as “stemness” of the urothelial carcinoma stem cell population (18, 21). Furthermore, genetic silencing in preclinical models, resulting from acquired cisplatin resistance, has been demonstrated to be reversible through DNA methyltransferase inhibition with reinstatement of cisplatin responsiveness (17, 19, 21, 22). The DNA methyltransferase inhibitors, decitabine, azacitidine, and zebularine, have single-agent activity in multiple urothelial carcinoma cell line and xenograft models (22–27). Synergistic inhibition of cell proliferation, and reversal of cisplatin resistance, occurs with coadministration with cisplatin in urothelial carcinoma cell line models (19, 21, 26). Data also support investigation of a DNA hypomethylating agent with gemcitabine, including in urothelial carcinoma (21, 28–32).

Guadecitabine (SGI-110, Astex Pharmaceuticals) is a DNA methyltransferase inhibitor prodrug, composed of a decitabine and deoxyguanosine dinucleotide to allow for optimized drug-like properties. An MTD of 90 mg/m² daily, on a 28-day cycle, for patients with myelodysplastic syndrome, was established in a first-in-human study, but was not reached in patients with acute myeloid leukemia (AML; ref. 33). DNA demethylation was dose dependent, but plateaued at 60 mg/m² daily, for 5 days, which was designated as the biologically effective dose recommended for phase II development. Febrile neutropenia, pneumonia, thrombocytopenia, anemia, and sepsis were the most frequent grade ≥3 adverse events. Clinical activity was demonstrated in this setting from monotherapy doses as low as 6 mg/m² for 5 days, and in a subsequent trial in AML (33, 34).

We hypothesized that cisplatin resistance might be reversed through coadministration with a DNA hypomethylating agent, such as guadecitabine. SPIRE was a phase Ib/IIa trial, in urothelial carcinoma and other solid malignancies, to determine a safe dose and schedule of guadecitabine in combination with gemcitabine and cisplatin.

SPIRE was an open-label trial comprising a dose-escalation phase Ib component for advanced solid cancers, followed by a randomized dose expansion phase IIa component as neoadjuvant treatment for MIBC. Patients eligible for the dose-escalation phase had incurable, histologically or cytologically confirmed, locally advanced or metastatic solid cancer, for which gemcitabine and cisplatin was clinically appropriate treatment. Any number of prior systemic chemotherapy lines were permitted.

The dose expansion included patients with T2–4a N0 M0 MIBC, planned for neoadjuvant gemcitabine and cisplatin prior to a planned radical cystectomy. Key inclusion criteria for both phases included Eastern Cooperative Oncology Group (ECOG) performance status 0 or 1, 16 years or older, glomerular filtration rate (GFR) estimation ≥60 mL/minute, hemoglobin ≥90 g/L, neutrophil count ≥1.5 × 10⁹/L, platelets ≥100 × 10⁹/L, bilirubin ≤ 1.5 × the institutional upper limit of normal (ULN), alanine transaminase and alkaline phosphatase ≤2.5 × ULN (ALP ≤ 5 × ULN if caused by liver or bone metastases), and life expectancy over 3 months. Key exclusion criteria included unresolved toxicities from prior therapy greater than Common Terminology Criteria for Adverse Events (CTCAE) v4.03 grade 1 (except alopecia), prior radiotherapy to >30% of bone marrow and major surgery, or an investigational medicinal product within 30 days. Full eligibility criteria are within the protocol (Supplementary Data) and as described previously (35). The study was undertaken in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines, and approved by North West–Haydock Research Ethics Committee (15/NW/0936). Patients provided written informed consent.

Baseline assessment included physical examination, full blood count, serum biochemistry (renal, liver, and bone profiles), and GFR estimation. Disease evaluation was undertaken in accordance with local routine practice for the relevant cancer type. Treatment cycle assessments were as per the baseline visit, plus assessment of adverse events (CTCAE v4.03) and blood sampling for pharmacodynamics, and in the dose-escalation phase, guadecitabine pharmacokinetics analyses. Dose modifications for predefined adverse event parameters are described in the protocol (Supplementary Data).

In all patients, gemcitabine and cisplatin was given as a 21-day cycle (Supplementary Fig. S1) with cisplatin 70 mg/m² on day 8 and gemcitabine 1,000 mg/m² on days 8 and 15 by intravenous infusions. Supportive medication, including antiemetics and an intravenous hydration schedule, was administered according to local institutional policy. Guadecitabine was administered by subcutaneous injection, preferably within the abdominal area. For the dose-escalation phase, up to four dose-level patient cohorts of guadecitabine were planned, of 20, 30, 45, and 60 mg/m², on each of days 1–5, for up to six treatment cycles. In the randomized dose-expansion phase, patients were allocated to gemcitabine and cisplatin chemotherapy alone, or in combination with guadecitabine at the recommended phase II dose (RP2D). Planned treatment duration was prospectively defined for either three or four cycles according to individual institutional practice.

For the dose-escalation phase, an evaluable patient was defined as one that, during cycle 1, completed study assessments, received guadecitabine on days 1–5, and cisplatin and gemcitabine on day 8, and where applicable, at least one dose of G-CSF and/or had experienced a dose-limiting toxicity (DLT). DLT was defined as any of the following occurring during cycle 1, if deemed definitely or probably related to treatment: >14-day delay in cycle 2 due to treatment-induced toxicity, grade 4 neutropenia ≥7 days, grade 3 or 4 neutropenia and temperature ≥38.5°C, grade 3 or 4 neutropenia and bacteriologically proven sepsis, grade 4 thrombocytopenia ≥7 days, grade 3 thrombocytopenia and nontraumatic bleeding, and other clinically significant grade ≥3 events, except nausea or vomiting (35). Dose-level cohorts enrolled 3–6 evaluable patients with a modified rolling 6 design (36). Initial dose-level cohorts did not include G-CSF prophylaxis, however, if DLTs occurred, specifically due to neutropenia or its complications, the protocol allowed for repeat of the current dose-level cohort with G-CSF at 300 μg s.c., daily, on days 15–21 in all future patients. If none of three, or one of six, patients experienced a DLT, then escalation to the next dose cohort was permitted. If ≥ two patients experienced a DLT, then this dose was deemed not tolerated. If ≤1 of six evaluable patients experienced a DLT and higher doses were not tolerable, then this dose level was deemed the MTD. The RP2D incorporated the MTD with consideration given by a safety review committee to the maximally biologically effective dose (MBED) based on circulating cell-free DNA (cfDNA) LINE-1 promoter methylation and hemoglobin F (HbF) reexpression. Full criteria for dose-escalation decisions, and determination of MTD and RP2D, are within the protocol (Supplementary Data) and are described previously (35). The RP2D was expanded to include six patients with advanced urothelial carcinoma. In the dose-expansion phase, patients were randomly allocated (1:1) to gemcitabine and cisplatin alone, or combined with guadecitabine at the RP2D.

Translational blood samples for pharmacodynamics effect of guadecitabine were taken on days 1, 8, and 15 in the escalation phase and on days 1 and 8 in the expansion phase. Promoter methylation status of LINE-1, NBL2, D4Z4, SAT2, and LTR12C was determined by EpigenDx through pyrosequencing of bisulphite-treated cfDNA, and experimental details are provided in the Supplementary Data. HbF level, as a percentage of total hemoglobin, was determined by high-performance liquid chromatography using the VARIANT II Hemoglobin Testing System (Bio-Rad) according to the manufacturer's instructions within a United Kingdom Accreditation Service accredited UK National Health Service Department of Haematology & Blood Transfusion.

The primary objective was to determine a guadecitabine RP2D when combined with gemcitabine and cisplatin for future investigation. Primary endpoints were the MTD based on defined criteria for DLT assessed by CTCAE v4.03 and the MBED based on circulating cfDNA LINE-1 methylation and HbF reexpression. Secondary endpoints included the toxicity profile (CTCAE v4.03) of guadecitabine combined with gemcitabine and cisplatin, including the RP2D within a randomized comparison with gemcitabine and cisplatin alone, pharmacokinetics of guadecitabine when combined with gemcitabine and cisplatin, the pathologic complete response (pCR) rate of patients with bladder cancer enrolled in the dose expansion phase of the trial (the trial was not formally statistically powered for this), and selected pharmacodynamics endpoints.

Statistical analyses were specified a priori in the statistical analysis plan. The dose-escalation phase analysis focused on DLT incidence, summarized by dose cohort, within the evaluable patient population. A descriptive summary of the cycle number received, dose intensity, and dose modification is presented by treatment allocation with adverse events summarized by CTCAE grade. Pharmacokinetics analysis are presented by dose received for guadecitabine and decitabine, including AUC, C_max, and T_max. The randomized dose-expansion phase was not powered for formal statistical comparisons of efficacy and sample size was set at 20. Analysis was conducted within the intention-to-treat population comprising all randomized patients. pCR was determined by local specialist uropathologist assessment summarized by treatment arm. Pharmacodynamics data are presented in graphs as mean change from cycle 1, day 1, by treatment allocation. Analyses were done with SAS (version 9.4) and Stata (version 16.0). Data were analyzed by Southampton Clinical Trials Unit statisticians. L. Day and G. Saunders had full access to the raw data. S.J. Crabb and G. Griffiths had final responsibility for the decision to submit for publication.

Forty eligible patients were enrolled between May 2016 and September 2019 (Supplementary Fig. S2). Three patients in the dose-escalation phase became nonevaluable due to rapid disease progression leading to death (one before treatment allocation and two during treatment cycle 1), and were replaced. The remaining 17 patients represent the evaluable patient population within the dose-escalation phase. Twenty patients with MIBC were randomly assigned within the dose expansion phase.

Baseline characteristics are presented in Table 1. Median age was 59 years (range, 38–76) in the dose-escalation phase and 68 years (range, 34–76) in the dose expansion phase. Eleven (64.7%) and 19 (95.0%) were male, and nine (52.9%) and 13 (65.0%) had an ECOG performance status of 0, respectively. Patient characteristics within the dose expansion phase were balanced between treatment arms.

DLTs occurring within the dose-escalation phase are summarized in Table 2. As three patients within cohort 1 (20 mg/m² guadecitabine, days 1–5) experienced a DLT, and two of these were related to neutropenia, this dose level was repeated (cohort 2) with G-CSF prophylaxis in all subsequent patients (and remaining treatment cycles within cohort 1). Following one DLT in initial six patients recruited to cohort 2, guadecitabine dose was escalated to 30 mg/m², days 1–5 (cohort 3). Three patients experienced at least one DLT in cohort 3, which was, therefore, deemed not tolerated. Cohort 2 was designated as the MTD, and after review of pharmacodynamics and pharmacokinetics data (described below), also the MBED. Per protocol, cohort 2 was expanded to eight patients to include six patients with advanced urothelial carcinoma. Adverse events within the dose-escalation phase for all treatment cycles are shown in Table 3 (grade 3 or higher) and Supplementary Tables S1–S5. Higher grade (grade ≥3) toxicities were predominantly hematologic and related to neutropenia and thrombocytopenia. Treatment duration and all dose alterations are indicated by treatment cohort in Fig. 1A and Supplementary Table S6. Of the 17 patients in the dose-escalation group, seven patients (41%) discontinued because of treatment-related toxicity, 16 (94%) had a delay of at least one treatment dose, and dose reductions were required for guadecitabine in seven patients (41%), gemcitabine in seven patients (41%), and cisplatin in two patients (12%). Dose delays and alterations were almost entirely related to hematologic toxicity through neutropenia and thrombocytopenia. Two patients treated within cohort 2 with refractory germ cell cancer, having received multiple prior lines of chemotherapy (for each including two separate cisplatin-containing regimens and carboplatin-based high-dose chemotherapy with autologous stem cell transplantation), achieved clinical stabilization of disease and tumor marker responses, as reported previously (37). Patients with urothelial carcinoma in the escalation phase who were cisplatin naïve (n = 4) had time to disease progression of between 10 and 46 months, and for those with prior cisplatin exposure (n = 7) of between 2 and 10 months (Supplementary Table S7). There were no treatment-related deaths.

Ten patients per arm were allocated to gemcitabine and cisplatin alone, or gemcitabine and cisplatin combined with guadecitabine and G-CSF, for the dose expansion phase. Adverse events are presented in Table 4 and Supplementary Tables S8–S12. Again, high-grade toxicity was predominantly hematologic, but balanced between treatment arms, in terms of severity overall, the nature of the adverse events experienced, and for those events deemed as guadecitabine related. Intended cycle number, dose administration, and intensity by treatment arm are shown in Fig. 1B and Supplementary Table S13, indicating similar administration of cisplatin, but a modest reduction in gemcitabine dose intensity in the guadecitabine arm. The latter was primarily although omission of the day 15 gemcitabine dose, or gemcitabine dose reduction, on the basis of per protocol criteria for neutropenia and thrombocytopenia. All patients received at least three cycles of treatment, except one in the gemcitabine and cisplatin alone arm (who discontinued trial treatment per protocol after one cycle due to a GFR < 60 mL/minute, but continued through three further cycles of chemotherapy prior to cystectomy). However, one of four (25%) patients in the gemcitabine and cisplatin and guadecitabine arm, versus two of three (67%) patients in the gemcitabine and cisplatin alone arm, had been planned for four cycles. Therefore, a greater number of patients discontinued treatment prior to the planned duration of treatment if receiving guadecitabine. Of the 20 randomized expansion patients, one (10%) per treatment arm discontinued because of treatment-related toxicity and five (50%) per arm required delay of at least one treatment dose. Guadecitabine dose reduction was required for one (10%) patient. Gemcitabine dose reduction occurred in one (10%) patient in the gemcitabine and cisplatin alone arm and in four (40%) patients in the guadecitabine arm. No cisplatin dose reductions occurred in either arm.

Eight patients in each treatment arm proceeded to radical cystectomy. The remaining two patients in each arm opted for radical chemoradiotherapy. No patients were delayed in proceeding to either cystectomy or radiotherapy through addition of guadecitabine. One patient in each arm underwent cystectomy >90 days from trial treatment for reasons of patient choice (guadecitabine arm) and completion of gemcitabine and cisplatin off-trial because of a lowered GFR (control arm). Six of 16 patients had a pCR at cystectomy, two in the guadecitabine arm and four in the chemotherapy arm. Time from randomization to completion of radical treatment to the bladder and postcystectomy perioperative morbidity (Clavien–Dindo classification) were similar between treatment arms (Supplementary Table S14). All patients within the dose-expansion phase remain alive at a median duration of follow-up of 7.6 [inter quartile range (IQR), 6.7–11.5, gemcitabine and cisplatin and guadecitabine] and 8.6 months (IQR, 6.8–12.4, gemcitabine and cisplatin alone) by arm. One patient, in the gemcitabine and cisplatin alone arm, has had a urothelial carcinoma metastatic relapse diagnosed to date.

cfDNA LINE-1 promoter methylation and HbF reexpression status are shown for each trial phase in Fig. 2. Promoter methylation status for selected other genes is shown in Supplementary Fig. S3. Results were consistent with guadecitabine target effect. Pharmacokinetics parameters for guadecitabine are shown in Supplementary Fig. S4 and Supplementary Table S15, and were consistent with the single-agent experience to date for guadecitabine (33).

We have established a dose and schedule for the DNA methyltransferase inhibitor, guadecitabine, for combination with gemcitabine and cisplatin chemotherapy at conventional doses for urothelial carcinoma. As anticipated, addition of guadecitabine to gemcitabine and cisplatin produces some additional treatment-related toxicity, manifesting predominantly as neutropenia, thrombocytopenia, and complications of these adverse events. In respect to neutropenia, it is clear that G-CSF prophylaxis is required for all patients to reduce risk of infective complications and impact on dose intensity (which many oncologists would already consider a reasonable addition to gemcitabine and cisplatin alone). Beyond this, we detected limited evidence for additional symptomatic adverse events, above that which would be anticipated for gemcitabine and cisplatin alone. The randomized dose expansion phase of the trial, for neoadjuvant treatment of MIBC, allowed assessment of relative dose intensity. We found no impact on the cumulative delivery of cisplatin to patients, although gemcitabine dose intensity was reduced modestly. Similar numbers of patient received up to three cycles of treatment between arms. However, some patients planned for four cycles had the final cycle omitted, and so we cannot exclude a cumulative effect of treatment such that later cycles might be compromised. Future studies should assess carefully the impact on dose delivery over multiple cycles of treatment. Our data were reassuring within the dose expansion phase with respect to timeliness and completion of radical treatment options, with cystectomy and radical radiotherapy delivered to similar time lines despite addition of guadecitabine.

A 28-day treatment cycle has been used in all prior clinical investigations of guadecitabine, either as monotherapy or in therapeutic combinations (33, 38, 39). Our schedule utilized day 1–5 guadecitabine administration, but for the first time within a 21-day chemotherapy cycle, to accommodate a typical gemcitabine and cisplatin dosing schedule for urothelial carcinoma, and to maintain cisplatin dose intensity which is considered critically important for this disease (5). This may have been relevant to the need to incorporate G-CSF, and the impact on gemcitabine delivery on day 15 in some patients. We did not test guadecitabine doses below 20 mg/m², day 1–5. This was primarily guided by monotherapy pharmacodynamics data that guadecitabine reliably depletes LINE-1 promoter methylation from 18 mg/m², day 1–5 (28-day cycle) and upwards, and maximally at around day 8 for cisplatin administration in our schedule (33). We acknowledge that lower, less myelosuppressive, guadecitabine dosing remains a hypothesis to explore for a cisplatin response optimization strategy, although with a potential sacrifice of guadecitabine monotherapy efficacy seen in preclinical urothelial carcinoma models (22–27).

We undertook pharmacodynamics evaluation of guadecitabine effect in cfDNA. LINE-1 promoter methylation has been used most frequently in this setting, as an on-target effect of DNA methyltransferase activity. We found this to decrease in a cyclical fashion, with a nadir at around day 8–15 of treatment, and meeting our intention to coincide this with cisplatin administration. One question for future study will be the degree to which magnitude of demethylation correlates to treatment efficacy. Whether this effect requires dose to be escalated to tolerance in chemotherapy combinations remains open to clinical evaluation (21). Promoter methylation of a panel of other genes demonstrated similar patterns of cyclical demethylation with, subjectively, greatest magnitude between days 8 and 15. We did see a greater variability around the timing of this nadir, and perhaps its duration, than for guadecitabine monotherapy. This is despite pharmacokinetics parameters for guadecitabine that were unaltered compared with prior data as a result of this chemotherapy combination (33). Arguably, there may be benefit in guadecitabine effect covering all three chemotherapy doses in this combination, however. Further assessment of this issue would require tumor biopsies, if feasible, in future investigation. We also assessed HbF reexpression as a readily measurable DNA methyltransferase inhibitor effect. Our findings suggest that, although there were 2- to 6-fold increases seen in HbF percentage in blood, the impact within the randomized expansion was subjectively similar within the control arm. This endpoint would, therefore, seem to have less utility to monitor treatment-induced target effect, at least in a chemotherapy combination.

This trial was not intended to formally assess clinical efficacy and it would be premature to form firm conclusions regarding this surrogate endpoint. With this caveat, we did establish clinical benefit in some patients within the dose-escalation phase of the trial. Of note, as described previously, two patients with multiple pretreated, platinum-resistant, germ cell cancers achieved significant clinical benefit that warrants consideration for development in this rare disease (37). This is consistent with preclinical, and limited clinical data, in germ cell cancer that imply supporting this strategy (40–42). In addition, we saw comparable pCR rates between the two arms of the MIBC dose expansion phase. Elsewhere, clinical data supporting a similar approach have recently been presented for a study of guadecitabine with carboplatin in platinum-resistant ovarian cancer. This randomized phase II trial did not meet its progression-free survival primary endpoint (16.3 vs. 9.1 weeks, for a chemotherapy of choice control arm; P = 0.07). However, the 6-month progression-free rate was significantly higher in the guadecitabine with carboplatin group (37% vs. 11%; P = 0.003). Questions remain for the development of a DNA methyltransferase/platinum combination regarding optimal dose and schedule, optimal platinum agent, optimal degree of demethylation impact, and its measurement.

A further practical aspect of this combination regimen is an increase in drug administrations over chemotherapy alone, with multiple subcutaneous administrations of guadecitabine and G-CSF. G-CSF was self-administered at home, whereas guadecitabine required clinic attendance for research nurse administration on days 1–5. We found this to be acceptable to patients and we did not find skin reactions, or multiple subcutaneous administrations, to be problematic. However, patient acceptability and the option of guadecitabine self-administration should be evaluated in future studies.

In conclusion, we have defined a recommended dose and schedule for guadecitabine in combination with gemcitabine and cisplatin. This modestly increases the adverse event profile, but appears deliverable over at least three to four cycles of treatment. Pharmacodynamics parameters are supportive of on-target effect for guadecitabine. Future studies are now warranted to formally test the efficacy of this combination in both platinum refractory and platinum-naïve patients.

S.J. Crabb reported grants and nonfinancial support from Astex Pharmaceuticals during the conduct of the study, personal fees and nonfinancial support from MSD, Bayer, Roche, and Janssen Cilag, personal fees from Astellas, Pfizer, and BeiGene, grants and personal fees from AstraZeneca, and grants from Clovis Oncology outside the submitted work. S. Danson reported grants from Cancer Research UK & the Department of Health and NIHR during the conduct of the study. J.W.F. Catto reported personal fees from AstraZeneca, Roche, Janssen, BMS, Ferring, Nucleix, and MAD during the conduct of the study. S. Hussain reported personal fees from AstraZeneca, Roche, Merck, Janssen, BMS, GlaxoSmithKline, Pfizer, Eisai, and Pierre Fabre outside the submitted work. D. Dunkley reported grants from Astex Pharmaceuticals during the conduct of the study. N. Downs reported grants from Astex Pharmaceuticals during the conduct of the study. E. Marwood reported grants and nonfinancial support from Astex Pharmaceuticals during the conduct of the study. L. Day reported grants from Astex Pharmaceuticals during the conduct of the study. D. Ellis reported grants from Astex Pharmaceuticals during the conduct of the study. A. Birtle reported other from Janssen, Roche, and MSD outside the submitted work. R. Huddart reported grants from Astex Pharmaceuticals during the conduct of the study, grants, personal fees, and nonfinancial support from MSD and Roche, personal fees and nonfinancial support from Janssen and Nektar Pharmaceuticals, and personal fees from Bayer and Astellas outside the submitted work, and is a partner in Cancer Centre London. G. Griffiths reported grants from AstraZeneca and personal fees from AstraZeneca during the conduct of the study, grants from Janssen Cilag, AstraZeneca, Novartis, Astex Pharmaceuticals, Roche, Heartflow, BMS, and BioNTech and personal fees from Celldex outside the submitted work. No disclosures were reported by the other authors.

S.J. Crabb: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft, writing–review and editing. S. Danson: Conceptualization, funding acquisition, investigation, writing–review and editing. J.W.F. Catto: Conceptualization, funding acquisition, investigation, writing–review and editing. S. Hussain: Investigation, writing–review and editing. D. Chan: Formal analysis, writing–review and editing. D. Dunkley: Project administration, writing–review and editing. N. Downs: Project administration, writing–review and editing. E. Marwood: Project administration, writing–review and editing. L. Day: Formal analysis, methodology, writing–review and editing. G. Saunders: Formal analysis, methodology, writing–review and editing. M. Light: Formal analysis, methodology, writing–review and editing. A. Whitehead: Formal analysis, methodology, writing–review and editing. D. Ellis: Data curation, software, writing–review and editing. N. Sarwar: Investigation, writing–review and editing. D. Enting: Investigation, writing–review and editing. A. Birtle: Investigation, writing–review and editing. B. Johnson: Investigation, writing–review and editing. R. Huddart: Investigation, writing–review and editing. G. Griffiths: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.

We thank the patients and their families for participating in this study, along with all investigators and site personnel. We thank Cancer Research UK and Astex Pharmaceuticals for provision of research funding to undertake this trial. The authors also wish to acknowledge Prof. John Staffurth, Dr. Anthony Kong, and Jacqueline Birks as members of the independent data monitoring and ethics committee. R. Huddart and B. Johnson acknowledge the support of the Institute of Cancer Research/Royal Marsden FT NIHR Biomedical Research Centre. This work was supported by Cancer Research UK (grant No. C9317/A19903) and Astex Pharmaceuticals.

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.