Effectiveness evaluation of adjuvant concurrent chemoradiotherapy for patients with positron emission tomography‐staged esophageal squamous cell carcinoma after complete resection: A population‐based cohort study

Abstract Background The role of adjuvant concurrent chemoradiotherapy (ACCRT) is unclear for patients with esophageal squamous cell carcinoma (ESCC) who receive esophagectomy with clean margins. We compared the survival of the ACCRT versus observation groups for these patients staged with positron emission tomography (PET) via a population‐based approach. Methods Eligible patients with locally advanced ESCC diagnosed between 2011 and 2017 were identified via the Taiwan Cancer Registry. We used propensity score (PS) weighting to balance observable potential confounders between groups. The hazard ratios (HR) of death and incidence of esophageal cancer mortality (IECM) were compared between the ACCRT and observation groups. We also evaluated overall survival (OS) in subgroups of either with or without lymph node metastases. Results Our primary analysis consisted of 105 patients in whom the covariates were well balanced after PS weighting. The HR for death when ACCRT was compared with observation was 0.58 (95% confidence interval 0.28–1.21, p = 0.15). The results were also not significantly different for IECM or in the subgroup analyses. Conclusion We found that for patients with PET‐staged ESCC who received esophagectomy with clean margins, the survival was not statistically different between ACCRT and observation. Further studies (randomized or larger sample size) are needed to clarify this issue.


INTRODUCTION
Esophageal cancer is one of the leading causes of cancer mortality around the world, including Taiwan. 1 The most common Hsin-Yuan Fang, Yu-Sen Lin, Chien-Kuang Chen, Jian-Xun Chen, Ting-Yu Lu, Tzu-Min Huang, Te-Chun Hsieh, and Yu-Cheng Kuo are equal contributors. histology is adenocarcinoma in Western countries, whereas it is squamous cell carcinoma in Asian countries. 1,2 For patients with esophageal squamous cell carcinoma (ESCC) who receive esophagectomy with involved margins, adjuvant concurrent chemoradiotherapy (ACCRT) is commonly suggested in the treatment guidelines. 3 However, the optimal adjuvant treatment for those with clear margins is less straightforward. Although ACCRT is not favored in the North American guidelines, 3,4 there was only one relevant randomized controlled trial (RCT) according to a systematic review published in 2020. 5 This RCT was published a decade ago in Asia and showed favorable outcomes for those treated with ACCRT. 6 However, this study 6 involved staging by computed tomography (CT), whereas the use of positron emission tomography (PET) rather than CT-only has been strongly favored in the modern era. 3,7 When we searched PubMed further using the keywords "(esophageal cancer) AND (squamous cell carcinoma) AND (adjuvant) OR (post operative) OR (postoperative) AND (concurrent chemoradiotherapy) AND (positron emission tomography)" in January 2022, we could not find relevant studies.
Because of the above-mentioned paucity of data regarding ACCRT, especially in the modern era, we compared the survival of ACCRT versus observation for patients with PET-staged ESCC treated via esophagectomy with clear margins based on the modern cancer registry data from Taiwan.

Data source
In this retrospective cohort study based on registry data, the analyzed data with personal identifiers removed were obtained from the Health and Welfare Data Science Center (HWDC) database, which included the Taiwan Cancer Registry (TCR), death registration, and reimbursement data for the entire Taiwan population provided by the Bureau of National Health Insurance (NHI). The TCR is a high-quality database 8 that provides comprehensive information such as patient demographics and patient and disease characteristics. This study was approved by the Central Regional Research Ethics Committee at China Medical University Taichung Taiwan (CRREC-108-080 [CR2]).

Study population and intervention
The inclusion criteria of our study population were: (a) patients with thoracic ESCC diagnosed within 2011-2017 from the TCR via the International Classification of Disease for Oncology 3rd edition (ICD-O-3) reference and histology codes; (b) had PET for staging; (c) received upfront esophagectomies with clear margins; (d) had locally advanced disease as pathological stage T3-4N0M0 or pT1-4 N1-3 M0 as defined by the 7th American Joint Committee on Cancer (AJCC); and (e) were 18-70 years old. The exclusion criteria were: (a) those with multiple treatment records in the TCR; and (b) those with prior cancer(s). These inclusion/exclusion criteria were modified from our clinical and research experience as well as previous studies. 6,9,10 Regarding intervention, we identified patients treated with either ACCRT (ACCRT group) using external beam radiotherapy 45-50.4 Gy in conventional fractionation according to the records in the TCR or no further systemic or radiotherapy (observation group).

Covariates
We included the following covariates as modified from recent relevant studies and our clinical and research experience. 6,9,10 Patient demographics (age, gender, and residency); patient characteristics (comorbidity, body mass index [BMI], drinking, and smoking); and disease characteristics (tumor size, tumor grade, tumor location, T-stage, and number of lymph node metastases as well as p-stage) were defined as follows. The patient residency region was classified as "northern Taiwan" or "non-north." Comorbidity was determined by the modified Charlson comorbidity index score 11 and classified as "with" or "without." The smoking and drinking were classified as "yes" or "no." The p-stage was classified as "2" versus "3." The pathological T-stage was classified as "1-2" or "3-4." Grade was classified as "well/moderately differentiated" or "poorly/undifferentiated." Tumor location was classified as "upper," "middle," or lower.

Analyses
The primary outcome of interest was overall survival (OS). We also evaluated the impact of intervention (ACCRT vs. observation) on incidence of esophageal cancer mortality (IECM). These endpoints were determined based on records in the TCR and death registry (censored on December 31, 2019). We adopted a propensity score (PS) approach and used propensity score weighting (PSW) as the framework for primary analyses as advocated in previous studies. [12][13][14][15] We estimated the probability of receiving ACCRT (vs. observation) with a logistic regression model based on all the above covariates, and then assessed the balance of covariates between groups after PSW using overlap T A B L E 1 Patient characteristics of the study population in the primary analysis Patient characteristics before PSW Patient characteristics (%) after PSW a ACCRT (n = 65) weight 16 via the standardized difference (SDif). 12 We compared the hazard ratio (HR) of death between the ACCRT and observation groups during the entire follow-up period via Cox proportional hazards model in the weighted sample for point estimation and used the bootstrap method to estimate the 95% confidence interval (95% CI). [17][18][19] We used an E-value to assess the robustness of our finding regarding potential unmeasured confounder(s), as suggested in the literature 20,21 because the PS approach can only be valid under the assumption of no unmeasured confounder(s). We took a competing risk approach to compare IECM between groups. 22 We performed the following supplementary analyses (SA) for subgroup analyses: SA-1 for patients with pathological lymph node metastases (pN+) and SA-2 for those without (pN0), whereas SA-3 for patients with pathological stage II (p-stage 2) and SA-4 for patients with pathological stage III (p-stage 3) because the role of adjuvant thoracic radiotherapy may vary according to lymph node status or pathological stage. 23 We used SAS v.9.4 software (SAS Institute) for statistical analyses.

Study population
We identified 105 patients (65 for ACCRT group and 40 for observation group) as our primary study population as shown in Figure 1. 24 We achieved covariate balance after PSW, although some imbalance was seen before PSW as shown in Table 1.

Primary analyses
After a median follow-up of 40 months (range, 1-104 months), 53 deaths were observed (32 and 21 patients for ACCRT and observation groups, respectively). The median follow-up was 72 months (range, 37-104) for survivors. The 5-year OS rates were 50% and 44% for the ACCRT and observation groups, respectively, in the unadjusted analysis (log-rank test, p = 0.56) (Figure 2(a)). The overlap weights adjusted OS curve is shown in Figure 2(b). The 3-5-year OS rates for both groups were 65%-57% (ACCRT group) and 47%-35% (observation group), respectively. When the ACCRT group was compared to the observation group, the HR of death was 0.58 (95% CI, 0.28-1.21, p = 0.15). The observed HR of 0.58 for OS could be explained by an unmeasured confounder that was associated with both selections of treatment (ACCRT vs. observation) and outcome (live vs. death) by a risk ratio of 2.27 (E-value)fold each, but weaker confounding could not do so. 20 The HR for IECM was 0.84 (95% CI, 0.31-2.29, p = 0.73).

Supplementary analyses
In both SA-1 and SA-2, we achieved covariate balance after PSW although some imbalance was seen before PSW as shown in Tables 2 and 3. When the ACCRT group was compared to the observation group, the HR of death was 0.38 (95% CI, 0.04-3.44, p = 0.39) for SA-1 (pN+) and 1.79 (95% CI, 0.07-44.81, p = 0.36) for SA-2 (pN0), respectively. The overlap weights adjusted OS curves are shown in Figure 3(a),(b). In both SA-3 (p-stage 2) and SA-4 (p-stage 3), we also achieved covariate balance after PSW although some imbalance was seen before PSW (see Tables S1 and S2). There was also no statistical difference in PSW-adjusted OS for p-stage 2 (p = 0.4) or p-stage 3 (p = 0.39).

DISCUSSION
In this population-based study, we found that for patients with PET staged ESCC who received esophagectomy with clear margins, the survival was not statistically different between adjuvant concurrent chemoradiotherapy and observation. To the best of our knowledge, our study was first to address this topic. The numeric trend found in our study (in favor of ACCRT) was compatible with previous RCT or observational studies without mandatory PET. Lv et al. 6 reported 3-year OS of 63% for ACCRT versus 51% for observation, whereas 65% versus 47% was observed in our study after PSW adjustment. In another recently published RCT (also limited by lack of mandatory PET), Ni et al. 25 reported 3-year OS of 66.5% for ACCRT versus 48% for observation (p = 0.016). Hsu et al. 26 reported the HR of death to be 0.63 when ACCRT was compared to observation, whereas we found the HR of death to be 0.58 in our study. However, statistical significance was not reached in our study. We summarized our finding with the available RCTs in the Table 4.  Therefore, our finding (no statistically significant difference between ACCRT vs. observation) was also compatible with current guidelines, 3,4 which do not recommend routine ACCRT for these patients.
Further studies, such as RCT or studies with larger sample sizes are needed to clarify this issue. However, when we searched the clinical trial registry (https://clinicaltrials.gov/) using keywords "Esophageal Squamous Cell Carcinoma j esophagectomy concurrent chemoradiotherapy j positron emission tomography," we did not find any relevant RCT.
In addition to the limitation of a moderate study sample size as mentioned above, there were several other limitations in our study. First, the nonrandomized nature of our study made potential unmeasured confounder(s) (such as systemic therapy or radiotherapy details or micro-metastases status) 27,28,29 always an issue although we used a PS method to adjust for observable covariates. Therefore, we reported the E-value to assess the robustness of our results to potential unmeasured confounder(s). Second, we did not investigate other endpoints like disease control, toxicity, or quality of life because of  uncertainty in data quality or accessibility. Furthermore, the generalizability of our study may be limited by the wide-spread use of neoadjuvant concurrent chemoradiotherapy and the increasing role of immunotherapy. For example, adjuvant immunotherapy had been reported to be beneficial for complete resected stage II or III esophageal cancer that had received neoadjuvant chemoradiotherapy and had residual pathological disease. 30 Therefore, the role of adjuvant radiotherapy should be further evaluated in the era of immunotherapy.

CONCLUSION
We found that for patients with PET-staged ESCC who received esophagectomy with clear margins, the survival was not statistically different between adjuvant concurrent chemoradiotherapy and observation. Further studies such as RCT or studies with larger sample sizes are needed to clarify this issue.

ACKNOWLEDGMENTS
The data analyzed in this study were provided by the Health and Welfare Data Science Center, Ministry of Health and Welfare, Executive Yuan, Taiwan. We are grateful to Health Data Science Center, China Medical University Hospital for providing administrative, technical, and funding support. We thank The Charlesworth Author Services Team for their help in English language editing.

CONFLICT OF INTEREST
All authors declare no conflict of interest. The exact numbers were not reported because of a Health and Welfare Data Science Center (HWDC) database center policy to avoid numbers in single cells (≤2).