Abstract
Background/Aim: The effect of pelvic neoadjuvant radiotherapy (nRT) for stage M1a rectal adenocarcinoma patients treated with systemic therapy followed by proctectomy and metastasectomy was scarcely investigated in the literatures. Patients and Methods: The eligible rectal cancer patients diagnosed between 2011-2019 were identified via the Taiwan Cancer Registry. In the primary analysis, we used propensity score weighting to balance observable potential confounders and compared the hazard ratio (HR) of death for the nRT group vs. without RT group. We also compared the incidence of rectal cancer mortality (IRCM) and performed various supplementary analyses. Results: Our primary analyses included 145 patients. nRT was associated with improved OS (HR=0.51, p=0.01). The numerical trends remained similar for IRCM and in supplementary analyses. Conclusion: nRT was associated with improved OS in our study population.
Colorectal cancer (mostly adenocarcinoma) was a common cancer world-wide and pelvic neoadjuvant radiotherapy (nRT) was part of the standard of care for locally advanced rectal cancer (1, 2). However, its role in M1 stage was less clear because the primary therapy would be systemic therapy. For patients with limited metastases such as American Joint Committee on Cancer (AJCC) 7th or 8th edition (3, 4) resectable stage M1a (metastasis to one site or organ without peritoneal metastasis), nRT was still one component of the guideline-recommended therapy (2). This concept (aggressive local therapy for oligometastases) was compatible with the results from recent randomized controlled trials (RCTs) in general (not limited to colorectal cancer) (5).
Specifically for colorectal cancer, a meta-analysis had reported that primary tumor resection in stage IV colorectal cancer was associated with longer overall survival (OS) (6). Another systematic review (searched till Jun 30, 2018) reported significant benefit in local control and possible benefit in OS when nRT was compared to no radiotherapy (7). However, the evidence level (8) of the studies included in this systematic review (7) was low. There were no formal RCTs except a subgroup analysis (9) in this systematic review (7). There were also two population-based studies (10, 11) from USA and Sweden included in this systematic review (7). We updated the search strategy (7) in Pubmed until Jan 2023 and found no additional RCT or population-based study except one study from USA (12).
Given the above-mentioned paucity in evidence as well as population-based studies from Asia, we performed this retrospective cohort study via Taiwan Cancer Registry (TCR) to investigate the effect of pelvic nRT for stage M1a rectal adenocarcinoma patients treated with systemic therapy followed by proctectomy and metastasectomy.
Patients and Methods
Data source. In this retrospective cohort study, we used the Health and Welfare Data Science Center database [including the Taiwan Cancer Registry (TCR), death registration, and reimbursement data for the whole Taiwan population provided by the Bureau of National Health Insurance] with personal identifiers removed. The TCR provides comprehensive information and has been reported to be a high-quality cancer registry in the world (13). This study was approved by the Central Regional Research Ethics Committee of China Medical University Taichung Taiwan [CRREC-108-080 (CR2)].
Study design, study population, and intervention. As suggested in the STROBE statement (14), our study flowchart is shown in Figure 1. We identified stage M1a rectal adenocarcinoma patients diagnosed during the period 2011-2019, treated with systemic therapy followed by proctectomy and metastasectomy. We excluded patients with multiple treatment records or prior cancer(s) to ensure data quality. We further selected patients treated with pelvic nRT using an external beam radiotherapy dose (20-70 Gy) (12) or without radiotherapy (without RT).
STROBE study flowchart and the number of individuals at each stage of the study. 1We only included those treated (class 1-2) with a single record to ensure data consistency. 2Radiotherapy dose 20-70 Gy. 3Without missing information in the TCR and death registry regarding survival status, and cause of death.
The explanatory variable of interest (nRT vs. without RT), primary outcome [overall survival (OS)], and other supplementary outcome [incidence of rectal cancer mortality (IRCM)] were determined via the TCR or death registry recordings. We also defined the diagnosis date as the index date and calculated the OS/IRCM from the index date to the date of death or to Dec 31, 2020 (the censoring date of the death registry).
Covariates. We collected covariates to adjust for potential nonrandomized treatment selection. These used covariates were based on recent relevant studies and our clinical research experiences (10-12, 15-17). Patient demographics (age, sex, residency), patient characteristics (comorbidity), disease characteristics [clinical T- and N- stage, abnormally elevated baseline carcinoembryonic antigen (CEA) level], and treatment characteristics [type of surgery, at least twelve regional lymph nodes examined (ALTRLNE)] were defined as follows. Age was defined in years. The patient residency region was classified as Northern or non-North in Taiwan. Comorbidity was determined by the modified Charlson comorbidity index score and classified as with or without (18). The clinical T-stage was classified as 1-2 or 3-4. The clinical N-stage was classified as 0 or 1-3. CEA was defined as normal or abnormal. The type of surgery was classified as partial proctectomy/anal preservation or total proctectomy/exenteration. For ALTRLNE, patients with at least twelve regional lymph nodes examined (2) were classified as yes, whereas those with fewer numbers were classified as no.
Statistical analysis and supplementary analyses. As advocated in the literature (19-23), we adopted the propensity score (PS) approach to balance the measured potential confounders. In the primary analysis (PA), we used PS weighting (PSW) (24-26) as the primary analysis framework. We estimated the probability of receiving nRT (vs. without RT) as the PS via a logistic regression model based on the above covariates. After PSW using the overlap weight (24), we used standardized difference (SDif) to assess the balance in covariates between groups (27, 28). During the entire follow-up period, we compared the hazard ratio (HR) of death between groups and obtained point estimation via the Cox proportional hazards model in the weighted sample. We adopted the bootstrap method to estimate the 95% confidence interval (95%CI) (29-31). The E-value was also used to evaluate the robustness of our findings regarding potential unmeasured confounder(s) (32). The IRCM between groups was evaluated by the competing risk approach (33) between groups in the weighted sample.
In the supplementary analyses (SA), we performed three analyses to clarify the robustness of our findings. In SA-1, we used an alternative approach (PS matching, PSM) within the primary study population to construct a subgroup (1:1 PS matched cohort without replacement) and compared the HR of death via a robust variance estimator (34) between groups. We used the subdistribution hazard ratio among the clustered Fine-Gray model to evaluate IRCM (35). In SA-2, we limited the radiotherapy dose to common regimens (25 Gy in 5 fractions or 45-50.4 Gy in 25-28 fractions) (2) for the nRT group in PA to explore the impact on OS via PSW. In SA-3, we performed the analysis among the nRT group in SA-2 to compare patients with short (5 fractions) or long course (25-28 fractions) RT (36).
The statistical analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC, USA) and R version 4.1.3 (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria).
Results
Study population. As shown in Figure 1, the main study population consisted of 145 eligible rectal adenocarcinoma patients who received nRT (88 patients) or not (57 patients) between 2011-2019. These patient characteristics are described in Table I. Some covariates (age, clinical T-stage, CEA, ALTRLNE) were imbalanced before PS weighting (28), but all covariates achieved balance (standardized differences ≈ 0) after PS weighting via the overlap weights.
Patient characteristics of the study population in the primary analysis.
Primary analysis. After a median follow-up of 39 months (range=1-111 months), 92 deaths were observed (50 and 42 patients for the nRT and without RT groups, respectively). The median follow-up was 55 months (range=13-111 months) for survivors. In the unadjusted analysis, the 5-year OS rates were 53% and 26% for the nRT and without RT groups, respectively (log-rank test, p<0.01; Figure 2). The overlap weight-adjusted OS curve is shown in Figure 3. The 5-year PSW-adjusted OS rates were 51% (nRT group) and 28% (without RT group). The PSW-adjusted HR of death was 0.51 (95%CI=0.31-0.83, p=0.01) when the nRT group was compared to the without RT group. The observed HR of 0.51 for OS could be explained by an unmeasured confounder associated with the selection of treatment (nRT or without RT) and survival by a risk ratio of 2.56 (E-value)-fold each, but weaker confounding factors could not. The result for the IRCM was in favor of nRT with borderline significance (HR=0.54, p=0.07).
Kaplan–Meier unadjusted overall survival curve (in years) in the primary analysis.
The overlap weights adjusted overall survival curve (in years) in the primary analysis.
Supplementary analysis. In the SA-1, the constructed PS-matched subgroup (n=94) is shown in Table II, and all covariates were balanced after PSM. The Kaplan–Meier OS curve is shown in Figure 4. The 5-year OS rates were 46% (nRT group) and 24% (without RT group). There was a statistically significant difference between the groups (HR=0.48, 95%CI=0.30-0.77, p<0.01). The results for the IRCM were also similar to those of the primary analyses (HR=0.59; 95%CI=0.35-1.01, p=0.06).
Patient characteristics of the PS-matched subgroup.
Kaplan–Meier survival curve (in years) for the propensity score-matched subgroup in the SA-1. RT: Radiotherapy; nRT: neoadjuvant RT.
In SA-2, among 82 patients who received a relatively common RT dose [25 Gy in 5 fractions (short course) or 45-50.4 Gy in 25-28 fractions (long course)] from PA, we found that the distribution of covariates between the common RT dose group and the group without RT could be balanced after PSW (Table III), and the PSW-adjusted HR of death was 0.51 (p=0.01) when the nRT group was compared to the group without RT.
Patient characteristics of the study population when comparing the common RT dose group vs. the group without RT.
In SA-3, the distribution of covariates between the long course (n=63) and short course (n=19) RT groups was also balanced after PSW (Table IV), and the PSW-adjusted HR of death was 0.78 (p=0.58) when long course RT was compared to short course RT.
Patient characteristics of the study population when comparing short vs. long radiotherapy courses.
Discussion
In this retrospective cohort study for stage M1a rectal adenocarcinoma patients treated with systemic therapy followed by proctectomy and metastasectomy in Taiwan, we found that nRT was associated with improved OS (HR=0.51, p=0.01). The numerical trends remained similar for the other endpoint and in supplementary analyses. To the best of our knowledge, this was the 1st population-based study from Asia.
Our results were in the same trend as reported in the abovementioned systematic review and population-based studies from Western countries (7, 10-12) in that nRT was associated with improved outcome. Agas et al. reported “Pooled 5-year overall survival showed a statistically significant benefit with neoadjuvant radiotherapy [risk ratio (RR)=1.47; 95%CI=1.14-1.89, p=0.003]” in their systematic review and meta-analyses for stage IV rectal cancer, although the RR was 1.31 in favor of nRT with a p-value 0.11 for those who received metastasectomy (7). Wu et al. used “the Surveillance, Epidemiology, and End Results database” from the USA and reported an adjusted HR of death of 1.3 (p<0.01) when surgery + chemotherapy was compared to neoadjuvant chemoradiotherapy for stage IV rectal cancer patients (10). Khani et al. used the Swedish Rectal Cancer Registry and reported an adjusted HR of death of 1.32 (p<0.01) when (no radiotherapy) compared to nRT in stage IV rectal cancer patients (11). Renz et al. used the National Cancer Database from the USA and reported “radiotherapy was associated with a statistically significant reduction in risk of death (HR=0.718; 95%CI=0.661-0.780)” for metastatic rectal adenocarcinoma patients treated with surgery (12). However, all the above four studies did not specifically analyzed patients with M1a disease, whereas our study focused on M1a disease (metastasis to one site or organ), which may be one of the reasons why the effect of radiotherapy was slightly stronger than that of the above studies.
On the contrary, the role of pelvic radiotherapy in rectal cancer is not without debate. The FOWARC study had reported no survival benefit of nRT when compared to neoadjuvant systemic therapy only (37). Given the non-randomized nature of our study, the interpretation of our results should be cautious and randomized controlled trials are eagerly awaited. The implication of our finding in the era of immunotherapy also remains unclear (38).
There were also additional limitations to our study. Firstly, potential unmeasured bias is always possible in non-randomized studies like ours. For example, the importance regarding the sequence of proctectomy and metastasectomy was debated (39), and the details of systemic therapy (40, 41) may also be important, but this information was not available in TCR and not included in our analyses, therefore our results might be biased if these factors were not balanced between groups. Therefore, we used the E-value advocated in the literature (32) to assess the robustness of our finding. Secondly, other endpoints such as quality of life (42) might be more relevant than OS investigated in our study; however, it was not investigated due to limitation of data availability.
In conclusion, nRT was associated with improved OS. The numerical trends remained similar for the other endpoint and in supplementary analyses. Further studies are needed to confirm our results.
Acknowledgements
The data analyzed in this study were provided by the Health and Welfare Data Science Center, Ministry of Health and Welfare, Executive Yuan, Taiwan. The Authors are grateful to Health Data Science Center, China Medical University Hospital for providing administrative, technical, and funding support. The Authors thank Mrs. Li CC for her help during this study.
Footnotes
Authors’ Contributions
Chien CR participated in the concept and design, analysis, and interpretation of data, and drafting of the manuscript. Ke TW, Chang SC, Chen HC, Wang HM, Chen WTL, Chen YC, Hsieh MH, Tsai YY, Huang CW, Liao YM, Lin CH, Huang CL, and Chen LC participated in the concept and design, interpretation of data, and drafting of the manuscript. All Authors read and approved the final manuscript.
Conflicts of Interest
The Authors declare that they have no competing interests in relation to this study.
- Received January 30, 2023.
- Revision received February 13, 2023.
- Accepted February 16, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
