Efficacy of early PET-CT directed switch to carboplatin and paclitaxel based definitive chemoradiotherapy in patients with oesophageal cancer who have a poor early response to induction cisplatin and capecitabine in the UK: a multi-centre randomised controlled phase II trial

Oesophageal cancer causes substantial global morbidity and mortality, with two predominant histologies: oesophageal adenocarcinoma (OAC) and oesophageal squamous cell carcinoma (OSCC). Definitive chemoradiotherapy (dCRT) is standard of care for locally advanced OSCC and an option for OAC when surgery is not undertaken. The systemic component of dCRT improves outcomes but is also a major source of toxicity, and systemic failure rates of 30–50% persist. Historically, fluoropyrimidine–platinum doublets have been used, though carboplatin–paclitaxel can yield similar outcomes with lower toxicity. Biomarkers to guide systemic regimen choice in dCRT are lacking. In neoadjuvant settings, 18F-FDG PET-CT–based early metabolic response after initial chemotherapy cycles has been explored to predict outcomes and guide therapy changes. In the UK, two cycles of induction chemotherapy are commonly used before concurrent dCRT. The study hypothesised that early PET-CT assessment (day 14 of cycle 1) of response to induction cisplatin/capecitabine could guide selection of the optimal concurrent chemotherapy (either continue cis/cap or switch to carboplatin/paclitaxel) to improve outcomes in OAC and OSCC. This hypothesis was embedded as a phase II PET-directed substudy within the SCOPE2 trial, which also evaluates radiotherapy dose escalation.

Evidence before this study: dCRT is an alternative to surgery in OSCC and an option for non-surgical OAC, but systemic failure remains common (30–50%). Metabolic response assessment after induction components has been associated with outcomes and used to individualise therapy in neoadjuvant contexts. Early (day 14) PET-CT assessment predicting outcomes has been suggested by prospective non-randomised trials in neoadjuvant chemotherapy, but its value to direct therapy during induction for dCRT was unknown. Prior trials such as MUNICON I/II and AGITG DOCTOR in OAC used PET response to guide subsequent therapy, with varying impacts on pathological response and survival. CALGB 80803 demonstrated prognostic PET response during neoadjuvant CRT in OAC and suggested regimen-specific benefits to switching. Retrospective data in OSCC suggested no benefit to changing chemotherapy based on PET response during CRT. Added value of this study: Demonstrates that a ≥35% reduction in SUVmax at day 14 is not prognostic in OSCC treated with dCRT, and switching from cis/cap to car/pac in non-responders leads to inferior outcomes in both OSCC and OAC. Implications: Early PET-CT–directed individualisation of systemic therapy in dCRT cannot be recommended for OAC or OSCC; further work is needed to optimise systemic therapy and identify predictive biomarkers.

Design: Multi-centre, randomised, open-label, phase II PET-CT–directed substudy within the UK SCOPE2 phase II/III trial (2×2 factorial design). All patients were also randomised to standard-dose (50 Gy/25 fractions) vs high-dose (60 Gy/25 fractions) radiotherapy. Registration: ISRCTN 97125464; ClinicalTrials.gov NCT02741856. Approvals: UK MHRA and Wales Research Ethics Committee (15/WA/0395). Centres: 16 UK radiotherapy centres. Population: Histologically confirmed oesophageal carcinoma (OAC, OSCC, or undifferentiated) or Siewert I–II GOJ tumours (<2 cm into stomach), selected for dCRT; age ≥17; WHO PS 0–1; T1–4, N any (TNM7); total disease length ≤10 cm (≤13 cm from Feb 2019). Required adequate renal, hepatic, haematologic, pulmonary, and cardiac function. Staging with contrast-enhanced CT of thorax/abdomen/pelvis and PET-CT; EUS recommended. PET substudy eligibility: baseline PET <5 weeks pre–induction with SUVmax ≥5; repeat PET-CT on day 14 (−2/+3 days) of cycle 1 of induction cis/cap; all began with cis/cap. PET response definition: Responders ≥35% reduction in SUVmax from baseline to day 14; non-responders <35% reduction. Randomisation: Conducted after day 14 PET, before cycle 2. All patients randomised 1:1 to RT dose; non-responders further randomised 1:1 to continue cis/cap vs switch to car/pac. Stratified by hospital, reason for non-surgical management, and stage; separate randomisation in OAC and OSCC cohorts; minimisation with 80:20 random element; concealed allocation via CTR. Treatments: Induction cycle 1 for all: cisplatin 60 mg/m2 day 1 + capecitabine 625 mg/m2 twice daily days 1–21 (cis/cap). Responders: continued cis/cap for 2nd induction cycle and two further cycles given concomitantly with EBRT (cycles 3–4). Non-responders: randomised to either (a) continue cis/cap for one more induction cycle and two concurrent cycles with EBRT, or (b) switch to carboplatin AUC5 + paclitaxel 175 mg/m2 day 1 for induction cycle 2, then weekly carboplatin AUC2 + paclitaxel 50 mg/m2 concurrently with EBRT. Permitted substitutions: 5-FU 225 mg/m2/day days 1–21 if unable to swallow capecitabine; carboplatin AUC5 if cisplatin contraindicated. Radiotherapy: IMRT/VMAT planned from contrast-enhanced 3D-CT or 4D-CT (≤3 mm slices). GTV defined from multimodality imaging; target volumes per SCOPE2 RT guidance. Dose: 50 Gy in 25 fractions (standard) or 60 Gy in 25 fractions (high), 5 fractions/week; ICRU 50/62 prescription; QA per protocol. Assessments: Clinical review before each induction cycle and weekly during CRT; toxicity per CTCAE v4.03. End-of-treatment assessment at week 12; follow-up at weeks 15 and 24, then 9, 12, 16, 20, 24, 36, 48, 60 months. CT surveillance ± endoscopy/biopsy at week 24 (endoscopy allowed up to week 36 during COVID-19). Subsequent imaging based on symptoms; salvage therapy at clinician discretion. Endpoints: Primary (substudy): TFFS at week 24 in PET non-responders (alive without progression on CT ± endoscopy/biopsy). Secondary: acute toxicity (within 12 weeks of end of treatment), overall survival (OS), progression-free survival (PFS). Statistics: Separate powering for OSCC and OAC. Assumed 24-week TFFS 55% in non-responders; OSCC sample of 86 to detect improvement to 75% (87% power, one-sided alpha 0.20); anticipated 27 OAC to detect increase to 85% (80% power, one-sided alpha 0.20). Analyses in Stata 17 on intention-to-treat. OSCC primary analysed by chi-square and mixed-effects logistic regression including stratification factors (centre random effect); sensitivity analyses adjusted for RT dose, sex, WHO PS. OAC analyses exploratory (small n). Wilcoxon tests for SUVmax distributions vs TFFS. Chi-square for grade 3/4 acute toxicity. Survival by Kaplan–Meier; Cox regression for HRs: OSCC univariable by arm; combined OSCC+OAC uni- and multivariable including arm, age, sex, PS, stage, histology, disease length. Prognostic value of PET response assessed by comparing TFFS (chi-square) and OS (Cox) among responders (all cis/cap) vs non-responders randomised to cis/cap.

- Enrollment and PET response: 103 patients entered the PET-CT substudy (16 UK centres; 22 Nov 2016–1 Aug 2021). PET non-responders: 63/103 (61.2%; 52 OSCC, 11 OAC). Randomisation of non-responders: 31 to switch to carboplatin/paclitaxel (car/pac) and 32 to continue cisplatin/capecitabine (cis/cap). PET responders: 40 continued cis/cap. - Early closure: Substudy closed 1 Aug 2021 by IDMC for futility and potential harm. Interim OSCC TFFS at 24 weeks: 14/20 (70.0%) car/pac vs 22/24 (91.7%) cis/cap; conditional power 4%. - Primary endpoint (TFFS at 24 weeks in non-responders): OSCC cohort: 17/25 (68.0%) car/pac vs 25/27 (92.6%) cis/cap; chi-square p = 0.025; adjusted logistic regression p = 0.028; post hoc adjusted for RT dose p = 0.032; further adjusted for sex and WHO PS p = 0.039. Direction of effect consistent across subgroups. OAC cohort: n = 11; trend favoured cis/cap but numbers too small for firm conclusions. - Overall survival (OS): In OSCC, OS favoured cis/cap over car/pac (median 42.5 vs 20.4 months; adjusted HR 0.36; p = 0.018). - Progression-free survival (PFS): OSCC unadjusted HR 0.30 (95% CI 0.13–0.69), p = 0.005, favouring cis/cap. Combined OSCC+OAC: median PFS 34.6 months (cis/cap) vs 19.4 months (car/pac); unadjusted HR 0.54 (95% CI 0.27–1.08), p = 0.079; adjusted HR 0.58 (95% CI 0.28–1.21), p = 0.147. Post hoc excluding one patient who progressed before cycle 2: adjusted HR 0.42 (95% CI 0.20–0.91), p = 0.027. - Toxicity: Grade 3/4 acute toxicity rates similar between arms (car/pac 22/31 [71.0%] vs cis/cap 21/31; overall comparable). - PET metrics and prognosis: Neither baseline SUVmax nor % change to day 14 associated with TFFS at 24 weeks. Trend toward worse survival in cis/cap PET responders vs non-responders who continued cis/cap (median 33.6 vs 42.5 months; HR 1.43, 95% CI 0.67–3.08; p = 0.35). - Compliance: >80% received RT per protocol in both arms; high chemotherapy dose intensity across arms. - Interpretation: Early PET-CT–based metabolic response (≥35% SUVmax reduction at day 14) is not prognostic in OSCC treated with dCRT. Switching non-responders from cis/cap to car/pac leads to inferior TFFS and OS in OSCC and worse TFFS in OAC.

The study addressed whether early PET-CT–assessed metabolic response during induction chemotherapy can guide selection of systemic therapy for dCRT in oesophageal cancer. In OSCC, the strategy of switching PET non-responders from cis/cap to car/pac significantly worsened 24-week TFFS and overall survival, indicating that PET-directed regimen change at day 14 is detrimental rather than beneficial. Furthermore, a ≥35% SUVmax reduction at day 14 was not prognostic for TFFS or OS in OSCC, challenging the extrapolation of neoadjuvant PET-guided strategies from OAC to dCRT settings and across histologies. Although numbers were small for OAC, trends similarly did not support switching to car/pac. The findings suggest cis/cap provides superior disease control compared with switching to car/pac in non-responders within this dCRT context, aligning with some prior data indicating better outcomes for cisplatin/fluoropyrimidine-based regimens in CRT compared with carboplatin/paclitaxel in mixed histology cohorts. The lack of association between early SUVmax change and outcomes in dCRT-treated OSCC contrasts with neoadjuvant data in OAC (e.g., MUNICON, AGITG DOCTOR, CALGB 80803), underscoring biological and treatment-context differences. Overall, early PET-CT should not be used to personalise the systemic component of dCRT in OSCC; the role in OAC remains unproven within this design.

In this multi-centre, randomised phase II PET-CT substudy embedded within SCOPE2, early metabolic response (day 14 SUVmax reduction ≥35%) was not prognostic in OSCC treated with dCRT, and switching PET non-responders from cis/cap to car/pac resulted in inferior 24-week TFFS and overall survival. Early PET-CT–directed individualisation of systemic therapy during dCRT cannot be recommended for OSCC and is not supported for OAC based on available data. Future research should focus on optimising the systemic component of dCRT, identifying robust biomarkers (potentially beyond early SUVmax changes), and clarifying regimen-specific interactions with radiotherapy dose. Advances in PET technology and analytics may warrant re-evaluation of optimal timing and parameters for response assessment in prospective trials.

- The PET-CT substudy was stopped early for futility and potential harm, leading to smaller-than-planned sample sizes, particularly in OAC, limiting statistical power and generalisability. - Imbalances in baseline characteristics (e.g., WHO performance status and sex) and a higher proportion receiving high-dose RT in the cis/cap arm could confound results; however, adjusted and subgroup analyses maintained direction and significance in OSCC. - Only 16 of 29 participating centres enrolled into the PET-CT substudy due to funding/logistical issues, introducing potential selection bias. - PET-CT response assessment was based on local SUVmax without central review, though acquisition/reporting were standardised and high concordance with central quantification has been reported elsewhere. - Endoscopy/biopsy at 24 weeks was not completed in all patients (CT-only assessment accepted in some cases, including during COVID-19), which may affect TFFS ascertainment; however, most had endoscopic assessment and trends were consistent. - The chosen early timepoint (day 14) and metric (SUVmax reduction ≥35%) may not be optimal for dCRT settings; evolving PET technologies and analytic methods may impact predictive performance. - Optional participation in the PET substudy and allowance of chemotherapy substitutions (e.g., 5-FU for capecitabine, carboplatin for cisplatin when contraindicated) could introduce heterogeneity.