Oesophago-gastric cancer: assessment and management in adults

Reason for the exceptional review

The purpose of this exceptional review was to examine any impact on NICE's guideline on oesophago-gastric cancer following the publication of the National Institute for Health Research's report on palliative radiotherapy combined with stent insertion to reduce recurrent dysphagia in oesophageal cancer patients: the ROCS (radiotherapy after oesophageal cancer stenting) RCT (randomised control trial).

Methods

The exceptional surveillance process consisted of:

For further details about the process and the possible update decisions that are available, see ensuring that published guidelines are current and accurate in developing NICE guidelines: the manual.

Methods

Participants (n=220) were recruited from 23 cancer centres and acute hospitals across England, Scotland and Wales. Participants were included if they:

Both arms of the study received SEMS for primary treatment of dysphagia related to oesophageal cancer, and randomisation post stenting was permitted following the initial study pilot. The intervention arm of the study also received EBRT, the protocol for which stated that EBRT should begin within 4 weeks, but preferably 2 weeks following stent insertion. The intervention arm had 2 treatment doses, either 20 Gy in 5 fractions over 1 week or 30 Gy in 10 fractions over 2 weeks using daily fractionation. The choice of treatment dose was determined by the patient's clinical oncologist. If patients missed 7 consecutive calendar days of radiotherapy, they were withdrawn from the trial.

All patients were followed up at 1 week after stent insertion, where QoL data was collected using the World Health Organization (WHO) performance status, questionnaires including EORTC QLQ-C30, EORTC QLQ-OG25, EQ-5D-3L, and toxicity assessment (CTCAE). Following the 1‑week check, all patients were followed up 4‑weekly for up to 1 year, until all participants had completed 12 weeks of follow up. Data on serious adverse events was monitored throughout.

The primary outcome for the ROCS study was measure of dysphagia symptoms, which was patient reported at each follow up, and also included death up to the 12‑week time point. The dysphagia score taken at 1 week post-SEMS insertion was taken as time zero, as some relief in dysphagia was expected in all patients. Cost effectiveness was assessed using cost–utility analysis with a time horizon of 12 weeks in the base case as well as 12 months in the sensitivity analysis. There was also an embedded qualitative study designed to explore patient experience and feasibility of recruitment to the study, and also to understand patient experience of living with advanced oesophageal cancer and dysphagia.

The sample size for this study was originally calculated to detect patient deterioration and increased dysphagia at 4 weeks. This was revised to include dysphagia deteriorations at the 12‑week time point. The authors highlight that missing data increased significantly after 12 weeks of participation, thought to be due to the increasing frailty of the study population by this stage in their disease. The revised calculations required 220 participants to be randomised, to allow for a 25% loss to follow up, and achieve 80% power to detect an increase in median time to self-reported dysphagia deterioration. An increase in time to dysphagia deterioration of 4 weeks was determined to be clinically meaningful. The authors state this was based on similar studies, expert multidisciplinary clinical and service user opinion, as median survival in this patient group was approximately 4 months.

Results

Baseline characteristics were similar and well-balanced following randomisation for both the usual care group (SEMS, n=112) and intervention group (SEMS plus EBRT, n=108).

For the primary outcome, the modified intention-to-treat population included 108 patients from the usual care group and 97 patients from the SEMS plus EBRT group, of which 74 and 75 patients respectively had complete data at the primary end point of 12 weeks post-stent insertion. No between-group differences were seen for the dysphagia-free survival up to week 12 post stent for usual care versus SEMS plus EBRT respectively (36/74 [48.6%] versus 34/75 [45.3%]; adjusted odds ratio [OR] 0.82; 95% confidence interval [CI] 0.40 to 1.68; p=0.587). The best (p=0.612) and worst (p=0.345) case scenario for missing data also showed no between-group differences for the primary end point.

This finding was also seen when the per-protocol population was considered, with 66 (usual care) versus 64 (SEMS plus EBRT) patients having a complete data set at 12 weeks post stent (28/66 in the usual care group [42.4%] versus 26/64 in the SEMS plus EBRT group [40.6%]; adjusted OR 0.99; 95% CI 0.45 to 2.14; p=0.972). No evidence of between-group difference was seen when the sensitivity analysis was performed, treating death by week 12 as no event (19/66 usual care [28.8%] versus 19/64 SEMS plus EBRT [29.7%]; adjusted OR 1.20; 95% CI 0.53 to 2.71; p=0.666).

A secondary analysis of dysphagia deterioration-free survival found the median time to dysphagia event or death was 13.1 weeks (95% CI 10.0 to 17.9 weeks) in the usual care arm (SEMS only) and 14.7 weeks (95% CI 12.1 to 17.4 weeks) in the SEMS plus EBRT arm, with an overall median time to dysphagia for the whole study population of 14.6 weeks (95% CI 12.1 to 17.4 weeks).

No evidence of a between-group difference was seen in the secondary outcome of overall survival (19.7 weeks for the usual care group [95% CI 14.4 to 27.7 weeks] versus 18.9 weeks for the EBRT group [95% CI 14.7 to 25.6 weeks], p=0.700).

No evidence of time versus treatment interactions was seen for the subscales of the EORTC QLQ-C30 and EORTC QLQ-OG25 questionnaires scales and WHO performance status (global health, odynophagia, pain/discomfort, eating restrictions) with all p values above 0.05. A time to treatment interaction however was seen for dysphagia at week 4 where it was higher in the SEMS plus EBRT group, but by week 8, there were no longer any between-group differences. The authors stated previously that initial increase in discomfort is expected from radiotherapy treatment and as such, this finding was not unexpected. The mean pain score was also higher in the SEMS plus EBRT group compared with the usual care group (p=0.005) at weeks 8, 12 and 16; similarly, interactions for constipation (p=0.009) were also seen at those time points for the SEMS plus EBRT group.

Higher median scores for cognitive function at weeks 12 and 16 (p=0.007), body image at weeks 8 and 16 (p=0.011), anxiety at weeks 12 and 16 (p=0.002) and trouble swallowing saliva at week 16 (p=0.025) were seen in the SEMS plus EBRT group. However, no evidence for time-treatment interactions for fatigue (p=0.522), weight loss (p=0.053) or WHO performance status (p=0.565) were seen.

Gastrointestinal (GI)-related bleeding events were observed initially up to week 16, where 19 patients (18.6%) versus 10 patients (10.3%) had an upper GI-related bleed event (SEMS alone versus SEMS plus EBRT). At week 52, upper GI-related bleed events were seen in 28.4% and 16.5% of patients in the usual care and SEMS plus EBRT groups respectively, up to week 52, giving a number needed to treat of 8.4. Median time to first bleeding event (upper GI-related bleed event or upper GI-related hospital admission) was 49.0 weeks (95% CI 33.3 to not reached weeks) in the SEMS alone arm versus 65.9 weeks (95% CI 52.7 to not reached weeks) in the SEMS plus EBRT arm. In a post-hoc subgroup analysis, no evidence of an interaction between group allocation, tumour length and time to first bleed was found (p=0.947). The treatment effect was consistent in both the less than 6 cm (p=0.216) and more than 6 cm (p=0.265) baseline tumour length subgroups indicating that SEMS plus EBRT had a positive impact on bleeding events.

Analysis of toxicity found the SEMS plus EBRT group experienced more vomiting at week 4, and more fatigue at week 12, more nausea, anorexia and stent-related pain compared with the usual care group; however, the SEMS plus EBRT group had less anaemia and upper GI haemorrhage. Additionally, 29 patients received palliative radiotherapy outside of the study protocol, 20 in the SEMS alone group and 9 in the SEMS plus EBRT group.

Discussion

This study was a well-designed RCT that is directly relevant to the UK population. Limitations were seen in consistent guidance on the trial itself from a patient perspective, with patients also raising that although stenting is commonly used to relieve symptoms of advanced disease such as dysphagia, it does not address the likely course of the disease. Difficulties in recruitment were also noted, with 1,252 people being assessed for eligibility, of whom 546 were eligible to enter the trial. The additional symptoms patients live with that may affect their QoL may have been improved by additional interventions such as dietary advice and pain relief information. The authors also highlight that the assumption that stent insertion addresses all issues of food intake and nutrition may inadvertently obstruct additional treatment options. The study found no evidence that SEMS plus EBRT was beneficial compared with SEMS plus usual care for any outcome except for upper GI bleeding events, which were fewer in the SEMS plus EBRT group. No significant difference was found between groups for the primary outcome of dysphagia-free survival. This finding is not consistent with recommendation 1.5.11 in the NICE guideline, which states 'Consider external beam radiotherapy after stenting for people with oesophageal and gastro‑oesophageal junctional cancer, for long-term disease control'.

Health economics evaluation

A cost–utility analysis, using a combined decision tree and Markov model, was performed to assess the cost effectiveness of SEMS plus EBRT on patient-reported dysphagia, and dysphagia-related events at 12 weeks in the base case and at 12 months in sensitivity analysis, following stent insertion. The key model assumptions are:

The cost–utility analysis produced incremental cost-effectiveness ratios (ICERs) expressed as cost per quality-adjusted life year (QALY) gained, following a UK NHS and personal social services perspective for analysis of health outcomes. The implementation cost of EBRT and cost of subsequent healthcare use were included for all patients where 4 follow-up points were available for data collection. These implementation costs were added to determine a total mean cost per patient at the 12‑week and 12‑month time points.

Intervention implementation costs for planning EBRT was costed at £377.15, with a cost per fraction of radiotherapy delivered of £155.58 (mean fractions per patient 6.06). Based on all available cases, the mean EBRT cost per patient was £1,304.42.

Total healthcare costs (including the cost of all primary, secondary and social care, and any medications prescribed) at the 12‑week period were not significant between groups despite the SEMS plus EBRT group having higher healthcare costs in the first 8 weeks, and lower costs in the remaining 4 weeks. After adding the intervention implementation costs, the SEMS plus EBRT arm accrued a mean cost of £5,854.10 (standard deviation [SD] £4,768.79), compared with £5,699.02 (SD £5,024.49) in the SEM alone arm.

Total healthcare costs at the 12-month period were based on cases who were able to contribute 1 healthcare data set at 1 follow‑up point within the trial period (n=102 SEMS, n=104 SEMS plus EBRT). The total 12‑month healthcare costs in this population were £7,440.43 (SD £7,643.21) and £9,087.14 (SD £8,890.75) in the SEMS plus EBRT and SEMS alone arm respectively, with a mean difference of -£1,646.70 (95% CI -£3,922.89 to £629.49), which was not statistically significant (p=0.155). When the EBRT implementation costs were added at an individual patient level, the cost difference decreases to -£568.05 (95% CI -£2,858.80 to £1,722.71; p=0.625).

No statistically significant difference in QALYs between the SEMS alone and SEMS plus EBRT group were seen at baseline, 12‑week or 12-month time points (p=0.1345).

After 1,000 simulated patients in the cost–utility analysis, the difference in QALYs between arms was minimal over 12 weeks (total QALYs, SEMS plus EBRT 53.993, SEMS alone 55.32). However, the total cost of the SEMS plus EBRT arm were over £750,000 higher compared with SEMS alone.

This resulted in an incremental cost of £549,200 per QALY gained at 12 weeks for SEMS plus EBRT compared with SEMS alone, highlighting that SEMS plus EBRT is less effective than SEMS alone and more costly. The one-way sensitivity analyses conducted at 12 months found no change to the base-case conclusion, SEMS plus EBRT was dominated and as such found to not be cost effective for end stage treatment of oesophageal cancer.

The impact of the significant reduction in bleeding events for the SEMS plus EBRT group did not result in a decrease in costs associated with hospital admission or healthcare interventions. However, the potential for EBRT to reduce bleeding in those patients who are clinically more likely to experience bleeding events could be further examined.

At a willingness to pay threshold of £20,000 per QALY gained, the use of SEMS plus EBRT for advanced oesophageal cancer is not a cost-effective option.

Studies included from the literature review

A systematic review and meta-analysis by Lai et al (2018) examined 8 RCTs and 732 patients with oesophageal cancer who received palliative care for the primary outcomes of mean change in dysphagia score, QoL and overall survival. Adverse events were also examined as secondary outcomes.

Three combinations of therapy were compared, of which 1 is relevant to this review: stents alone versus stents combination therapy (combination therapy included radiotherapy). In the stents combination therapy group, improvements were seen in mean dysphagia score (particularly in those patients 3 months post insertion), overall survival, QoL, lower risk of stent migration, aspiration pneumonia and restenosis when compared with the stents alone group. However, risk of fistula formation, haemorrhage and severe pain were higher in the stent combination group compared with the stent alone group. The authors conclude that whereas there were no immediate short-term differences, for patients living longer than 3 months, there was a significant improvement in dysphagia score for the stents combination group, as well as improvements in overall survival and QoL scores.

A systematic review and meta-analysis (Tinusz et al. 2020) of 17 studies compared stents alone (control group) with stents combined with any active oncological treatment (intervention group) including radio-, chemo- or photodynamic therapy, for efficacy and safety as treatment for people with oesophageal cancer receiving palliative care. Dysphagia grade, survival, late dysphagia, medical costs and oesophageal perforation were examined by the review. There were no significant differences seen in any complications between the control group (stents only, n=629 participants) and the intervention group (stents plus active oncological treatment, n=548). Of the 13 studies that reported on survival between the 2 groups (mean or median), 8 studies showed significantly longer life expectancy in the intervention group compared with the control group; however, the remaining 5 studies found no between-group difference in survival. The authors conclude that the benefit of additional oncological treatment alongside stenting was unclear; however, it was not associated with any increase in adverse effects.