Background
A common presenting symptom of oesophageal cancer that significantly affects health-related quality of life is dysphagia [
1]. Dysphagia can be alleviated and palliated by several different means such as local endoscopic measures with stenting, internal or external radiotherapy and/or chemotherapy [
2]. Chemotherapy leads to partial or complete resolution of dysphagia in most patients [
3‐
5]. Self-expanding metallic stents are effective for short-term relief from dysphagia, although not without complications [
5‐
7]. Radiotherapy, either external beam radiotherapy (EBRT) or internal high dose-rate (HDR) brachytherapy, is an effective palliative treatment of dysphagia [
8,
9] and has a more prolonged effect in comparison to stents [
10]. Radiotherapy alone for dysphagia relief is as effective as concomitant palliative radio- and chemotherapy, however the latter comes at the cost of increased toxicity [
11]. In summary, there are several ways to relieve malignant dysphagia, which one is preferred depends on tumour and patient characteristics, the discretion of the treating physician and availability of the different treatment modalities.
The prognosis of incurable oesophageal cancer is dismal, with a 5-year survival of 4—6% [
12‐
14]. Overall survival of oesophageal cancer is marginally improved with palliative chemotherapy and targeted treatments compared to best supportive care, with a median overall survival of 4.7 months versus 4.2 months, respectively (HR 0.81; 95% CI 0.71–0.92) [
15]. Considering the expected short survival time, palliative treatments should ideally be efficient and convenient with a minimum of side-effects. Different radiotherapy regimens are used according to local practice to palliate oesophageal cancer. The aim of this study was to analyse the effectiveness of two different palliative EBRT schedules: short-course radiotherapy; 4 Gy daily given 5 workdays in a row (SR) and long-course radiotherapy; 3 Gy daily for 10—13 consecutive workdays (LR), at a single institution, Karolinska University Hospital, which serves the whole Stockholm county.
Discussion
Our results demonstrate equal effectiveness of SR and LR for short-term relief of dysphagia caused by oesophageal cancer. No difference was seen in survival. However, acute toxicity was more frequent and more severe after LR than after SR. These results emphasize the importance of careful consideration when choosing a radiotherapy schedule. Longer radiotherapy schedules, in a patient group with dismal prognosis, might lead to more side-effects without greater symptom relief or better survival.
Previously, radiotherapy has been shown to relieve dysphagia in 41—75% of patients [
9,
20‐
23]. No statistically significant difference was seen between SR and LR in dysphagia relief in our study. Vermeulen et al. [
20] did a cohort study in the Netherlands on low-dose radiotherapy (4 Gy × 5 EBRT) compared to high-dose radiotherapy (3 Gy × 10 EBRT and 12 Gy single dose HDR-brachytherapy). Dysphagia improvement was reported in 50% of patients treated with low-dose radiotherapy and in 66% of patients treated with high-dose radiotherapy, the difference was not statistically significant. In another Dutch study, Walterbos et al. [
21] did a retrospective cohort study on three different EBRT schedules: 4 Gy × 5, 3 Gy × 10 and 3 Gy × 13. In this study, the dysphagia score improved in 41% of patients and no difference was found in dysphagia relief between the three schedules. Thus, neither our nor the Dutch results suggest that higher radiotherapy doses are more likely to lead to short-term relief from malignant dysphagia than lower radiotherapy doses.
Short-term symptom relief from radiotherapy can be hampered by acute side-effects, although previous studies have reported acute toxicities of grade ≥ 3 from palliative radiotherapy as uncommon (3–3.9%) [
21‐
23]. In our cohort, 9% of the patients experienced grade ≥ 3 acute toxicities and this was less frequent after SR than after LR. Vermeulen et al. [
20] reported severe adverse events in 15% of the patients and no difference between low-dose (4 Gy × 5 EBRT) and high-dose radiotherapy (3 Gy × 10 EBRT and 12 Gy single dose HDR-brachytherapy). We chose to separate acute toxicity from other adverse events, which could possibly explain the differences between our results and Vermeulen’s. In our study, acute toxicity had a clear causal relationship to the treatment and occurred within 6 weeks. Other adverse events had a possible and perhaps a more probable relation to locoregional tumour burden. Like Vermeulen’s results, we saw no statistically significant differences between the two studied treatment schedules in other adverse events of grade ≥ 3. Considering the poor prognosis of incurable oesophageal cancer and the more moderate acute toxicity profile of SR compared to LR presented in our study, SR might be the preferred treatment schedule.
Though the prognosis is generally dismal for incurable oesophageal cancer, a few patients live long enough for the short-term symptom relief to subside and some require re-interventions. Ideally, the length of dysphagia improvement should be recorded with a follow-up dysphagia score. However, the short survival of the patients limited the available data on the length of dysphagia relief. On the other hand, re-intervention reflects an important part of the outcome of the treatment, both for patients and health care providers. Walterbos’ [
21] study on three different EBRT schedules found that 24% of patients underwent re-intervention in the group as a whole; 35% of patients after radiotherapy with 4 Gy × 5, 22% of patients after radiotherapy with 3 Gy × 10 and 17% of patients after radiotherapy with 3 Gy × 13. Furthermore, treatment schedule was the only prognostic factor related to time to re-intervention. Similarly, Vermeulen et al. [
20] found that re-treatment was indicated in 23% of patients after low-dose radiotherapy (4 Gy × 5) and 17% of patients after high-dose radiotherapy (3 Gy × 10 and 12-Gy single-dose HDR-brachytherapy). In our cohort, there was a greater need for a re-intervention after SR than after LR (32% vs 19%), though the difference was not statistically significant. The median time from radiotherapy to re-intervention in our study was similar in the two groups, 122 days after SR and 119 after LR. Altogether, these three retrospective studies indicate that the need for re-intervention after radiotherapy with lower doses is greater than after radiotherapy with higher doses. Whether this outweighs the longer treatment times with increased risk of toxicity remains to be established.
Increased temporary acute toxicity might be acceptable if a lesser need for re-intervention and longer survival were to be expected. In our study, survival was similar after treatment with higher and lower radiotherapy doses. However, other observational studies have shown that higher radiotherapy doses in palliative settings might improve survival for patients with oesophageal cancer. Guttman et al. [
24] did a cohort study on radio- and chemotherapy in patients with metastatic oesophageal cancer based on the National Cancer Database in the USA. Their results demonstrated better OS after higher doses (≥ 50.4 Gy) of radiotherapy compared to lower doses (< 50,4 Gy), median overall survival was 11.3 months and 7.5 months, respectively. Vermeulen et al. [
20] reported better survival in patients treated with EBRT of 3 Gy × 10 and 12-Gy single-dose HDR-brachytherapy compared to patients treated with EBRT of 4 Gy × 5, median survival was 177 days versus 88 days. Lastly, survival after EBRT with 3 Gy × 13 was superior to survival after 4 Gy × 5 in the study of Walterbos et al., where median OS was 9.7 months versus 4.6 months [
21]. A possible explanation of superior survival after longer radiotherapy schedules in retrospective studies could be a selection bias. In the study by Walterbos et al. [
21] patients selected for the different treatment schedules differed in baseline characteristics. For example, patients with distant metastases were more often treated with lower radiotherapy doses, which was associated with worse survival in their study as well as in ours. Vermeulen et al. [
20] and Guttmann et al. [
24] used propensity score to account for confounding, however residual confounding cannot be adjusted for.
Nevertheless, there seem to be groups of patients with more favourable prognosis who might well benefit from higher radiation doses and longer therapies. Women with localised SCC and ≥ 55 years of age were demonstrated to have better survival in two cohort studies, one based on data from the Surveillance, Epidemiology and End Results Program in the USA [
25] and the other a nationwide Swedish cohort study [
26] on the prognosis of patients treated with surgery for oesophageal cancer. Similarly, female sex was shown to be a positive prognostic survival factor in a Chinese study on the impact of sex on the prognosis of SCC in the oesophagus treated with definitive radiotherapy [
27]. In our subgroup analysis, we saw a tendency towards better survival for women, older patients (> 70 years), SCC and patients with localised disease. These patient groups might benefit from longer radiotherapy treatments, though further research is needed before applying this in clinical practice.
Our study has some limitations that should be addressed. Firstly, the inherent selection bias of a retrospective study. The baseline characteristics of the two radiotherapy groups differed, although the only variable that differed significantly was distant metastases. To minimize bias and improve internal validity a randomized controlled trial would be ideal. Still, to our knowledge, a randomized controlled trial on different palliative EBRT schedules for oesophageal cancer has not been done and the results from our study contribute to the evidence on which we can base current treatment decisions. Secondly, the retrospective nature of the study led to limitations in registered data and missing values. Missing data would have been minimized by prospectively collected standardised data. Though, the consistency of data was strengthened by the availability of all medical charts, no loss to follow-up and because all medical charts were screened by one person, the first author. Thirdly, the precision of the results would have improved with a larger study population, though the real-life nature of the study, demonstrating clinical practice in a high-volume tertiary cancer centre, strengthens the study’s generalisability.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.