Evaluation of Postoperative Chemotherapy Duration of Capecitabine-Alone Regimen for Surgery Treated Elderly Colon Cancer Patients : A Cohort Study


 Background: Only 50-70% elderly colon cancer patients could complete a six-month postoperative chemotherapy. It is unknown whether a shorter duration of postoperative capecitabine-alone chemotherapy would not compromise the survival. We thus conducted this study to analyze the association between postoperative chemotherapy duration of capecitabine-alone regimen and cancer specific survival (CSS) in the surgery treated elderly colon cancer patients.Methods: We performed a retrospective cohort study in surgery treated stage III and high-risk stage II colon cancer patients aged >= 70 from two medical centers. Cox proportional hazard regression models were utilized to calculate crude and adjusted hazard ratios (HRs). The non-linear relationship between postoperative chemotherapy duration and CSS was analyzed through the restricted cubic spline regression analysis, and the threshold effect was calculated by the two-piece-wise Cox proportional hazard model.Results: A total of 1,217 surgery treated colon cancer patients between August 1, 2013 and September 1, 2019 were reviewed and 257 stage III and high-risk stage II patients aged >= 70 were finally enrolled. Postoperative chemotherapy with capecitabine was delivered in 114 patients and 143 cases only received surgery. The capecitabine duration was generally associated with a 11% risk decrease in death (HR=0.89, 95% confidence interval (CI) 0.82 - 0.96). Non-linearity exploration suggested a threshold effect of capecitabine duration on CSS in stage III disease. The HR for death was 0.79 (95% CI: 0.68 - 0.92) with duration <= 16 (week) while 1.34 (95% CI: 0.91 – 1.97) with duration > 16 (week).Conclusions: The postoperative capecitabine duration was significantly associated with CSS in elderly colon cancer. However, a threshold effect of capecitabine duration on survival may suggest a short-term chemotherapy could be an alternative for the conventional six-month regimen in the elderly stage III colon cancer.


Introduction
Globally, colorectal cancer (CRC) is one of the ve most common causes of cancer death (1). Patients with stage III and high-risk stage II colon cancer have a risk of recurrence over 30% after surgery (2,3).
Postoperative chemotherapy has been clearly proven to decrease the recurrence risk and increase survival in patients with high-risk stage II and stage III colon cancer (4). For several decades, a six-month chemotherapy regimen has been the standard postoperative treatment in colon cancer (4,5). However, recent studies from International Duration Evaluation of Adjuvant Therapy collaboration (IDEA) suggested noninferiority of shorter duration (3 months) of chemotherapy, as compared with 6 months, was con rmed in most colon cancer patients with CAPOX (Oxaliplatin plus capecitabine) regimen, but not in those with FOLFOX (Oxaliplatin plus 5-uorouracil/leucovorin) regimen (6, 7). The duration effect appeared dependent on the chemotherapy regimen and it seems that the use of capecitabine could shorten postoperative chemotherapy duration without compromising survival.
So far, only a small fraction of ttest elderly was enrolled in randomized controlled trials (RCTs). Besides, the participants in the RCTs all received oxaliplatin-containing regimen, but it is still controversial to apply oxaliplatin-containing regimens for patients 70 years or older. A uoropyrimidine alone regimen is still the recommended postoperative protocol for the elderly in clinical practice (8,9). The limited number of elderly participants and the different recommended chemotherapy regimens for the elderly made it impossible to evaluate the effect of capecitabine duration on survival in the elderly from results of RCTs.
Additionally, more than 30%-50% of patients could not complete the whole duration (10,11). We were thus prompted to conduct this study to explore the association between the postoperative chemotherapy duration of capecitabine-alone regimen and cancer speci c survival (CSS) in elderly colon cancer patients.

Study population
The data of patients was collected from the electronic medical record system from Guizhou Province Cancer Hospital and A liated Hospital of Guizhou Medical University. Surgery treated patients aged > or = 70, with stage III and high-risk stage II colon cancer, were included. All patients enrolled were restaged based on the 8th edition of the Union for International Cancer Control / American Joint Cancer Committee TNM staging classi cation. Figure 1 shows the ow chart of enrollment. The inclusion criteria for this study were as follows: 1. All patients enrolled were pathologically diagnosed as colonic adenocarcinoma; 2. They were aged > = 70 years at the time of diagnosis and underwent radical surgery; 3. They were staged as high-risk stage II (risk factors de ned as tumor obstruction preoperatively, tumor perforation preoperatively, poorly differentiation, positive surgical margin, fewer than 12 lymph nodes detected, perineural in ltration, invasion of extramural venous or lymphatic vascular, or T4 disease) or stage III; 4.
The enrolled patients did not receive neoadjuvant treatment.
The local ethics committee of the Guizhou Province Cancer Hospital approved this study. Informed consent from enrolled patients was not required because this was a retrospective cohort study.

Study design and variables
This retrospective cohort study aimed to assess association between postoperative capecitabine duration and CSS in elderly colon cancer patients. The postoperative capecitabine duration was described as the time between the last cycle of chemotherapy accomplished and the rst cycle of chemotherapy initiated.
We obtained CSS as nal outcome variable. CSS was de ned as the interval from diagnosis to the date of cancer-attributable mortality. The Charlson comorbidity index (CCI) has been proposed to evaluate the prognosis of CRC patients with comorbidities and frequently used in the Western countries (12)(13)(14). CCI was calculated using data regarding complications (15). We utilized clinical data from electronic medical record system to calculate CCI of each participant. Moreover, mismatch repair (MMR) status has been con rmed to be associated with response to uoropyrimidines and recommended testing in CRC patients (16). Therefore, apart from demographic factors, CCI and clinicopathological features, we obtained MMR status as covariates. Finally, the variables used to establish multivariable adjusted models included: (1) Table 1. Strati ed analyses were performed by Cox proportional hazard models. Cox proportional hazards regression model with cubic spline functions and smooth curve tting were employed to explore nonlinear relationship between postoperative chemotherapy duration and CSS. If nonlinearity was detected, then the in ection point was calculated by recursive algorithm, and a two-piecewise Cox proportional hazard model was established on both sides of the in ection point. In the end, which model was more suitable for tting the association between target independent variable and outcome variable was mainly determined by log likelihood ratio test.

Sensitivity analysis
For sensitivity analysis, we rst set up model II which adjusted for gender, age, stage, CEA (elevated, normal), CA199 (elevated, normal), CCI ( < = 4, > 4), histological type and MMR status ( Table 3). The results were consistent with those of crude mode and model I. We further used different covariates as the strati cation variables to observe the trend of HR among different subgroups (Fig. 2). The core results were consistent with our initial cohort. The tests for interactions were not statistically signi cant for age, gender, stage, histological type, CEA, CA199 or MMR status (P > 0.05).

Non-linear relationship of postoperative chemotherapy duration with CSS in stage III patients
In our study, the nonlinearity of postoperative capecitabine duration on CSS was observed in stage III (Fig. 3a, P for nonlinearity = 0.003), but not in high-risk stage II (P for nonlinearity = 0.24). After adjusting for gender, age, T stage, CA199, CEA, histological type, CCI and MMR status, smooth curve analysis and Cox proportional hazards regression model with cubic spline functions indicated that the relationship between the postoperative capecitabine duration and CSS was still non-linear and presented as L-shape in stage III elderly patients (Fig. 3b, P for nonlinearity = 0.01). Both Cox proportional hazard model and two-piecewise Cox proportional hazard model were applied to t the association. We selected the best t model according to P for log likelihood ratio test. Because the P for log likelihood ratio test was < 0.05, we selected the two-piecewise Cox proportional hazard model for tting the association between postoperative capecitabine duration and CSS in stage III elderly patients. The in ection point was calculated as 16 by using the recursive algorithm and two-piecewise Cox proportional hazard model ( Table 4). The HR for the death was 0.79 (95% CI: 0.68-0.92) with postoperative capecitabine duration < = 16 (week) while 1.34 (95% CI: 0.91-1.97) with duration > 16 (week). However, in high-risk stage II patients, we didn't discover this non-linear association.

Discussion
In this retrospective cohort study, we investigated the association of postoperative chemotherapy duration of capecitabine-alone regimen to CSS for surgery treated elderly colon cancer patients. In general, the postoperative capecitabine duration was associated with a 11% decrease of death risk in the elderly, which was consistent with the ndings mainly obtained from younger patients (17,18). Further analyses in the stage III group unveiled some novel ndings. Non-linearity exploration suggested that a Lshape relationship between the postoperative capecitabine duration and CSS. After adjusting potential confounding factors, the non-linear relationship between capecitabine duration and CSS was still signi cant (Fig. 3b, Table 4). The HR for the death was 0.79 (95% CI: 0.68-0.92) with chemotherapy duration < = 16 (week) while 1.34 (95% CI: 0.91-1.97) with duration > 16 (week).
The duration effect appeared dependent on the use of capecitabine since patients with pT4 or pN2 tumor were still recommended with 6-months FOLFOX regimen, while CAPOX duration could be shorten to 3 months regardless of high-risk factors (17). Although capecitabine-alone regimen remains to be one of the recommended postoperative protocol for the elderly, studies reported less than 50%-70% of patients could complete the whole duration. Our study found that in stage III elderly colon cancer patients, the effect of capecitabine duration on CSS changed at the turning point of duration as 16 weeks. That is, after 16-week use of capecitabine, the declining trend of death risk as increasing in chemotherapy duration disappeared. To the best of our knowledge, this study rstly demonstrated the threshold effect of postoperative capecitabine duration on CSS in the elderly colon cancer. A short-term duration of capecitabine chemotherapy may be an alternative for the conventional six-month postoperative chemotherapy regimen in the elderly stage III colon cancer.
The threshold effect of capecitabine duration on survival which we discovered in stage III group was not detected in stage II group. The prognosis of stage II colon cancer varies greatly according to different risk factors. The 5-year survival rate of patients with stage IIB/IIC (T4) colon cancer was signi cantly lower than that of patients with stage IIIA (19). In addition, the prognosis of stage II patients with residual tumor was also worse than that of stage III patients, and the incidence of postoperative tumor residual in IIB/IIC stage was signi cantly higher than that in stage IIIA (20). Therefore, patients with certain high-risk factors in stage II may need longer capecitabine duration. A multicenter large cohort study may be needed to identify the effect of postoperative capecitabine duration on survival in terms of different high-risk factors in stage II.
Despite the International Society of Geriatric Oncology (SIOG) has recommended postoperative chemotherapy for elderly CRC patients, they tend to receive postoperative chemotherapy less frequently than non-elderly patients. In our study, 42.2% with high-risk stage II and 46.5% with stage III disease received postoperative therapy. Similarly, previous studies reported the proportion of elderly patients who received postoperative chemotherapy ranged from 15-57% (14,(21)(22)(23). These proportions were much lower than those in younger patients. Reasons for nontreatment and treatment interruptions in the elderly included comorbidities, side effects and patient choice (23). Although SIOG has recommended the performance of a comprehensive geriatric assessment (CGA) before cancer treatment (24), CGA is not frequently used in clinical practice in China. Considering the comorbidities and tolerance of side effects in the elderly may affect the postoperative chemotherapy duration and outcome, we utilized clinical data from electronic medical record system to calculate CCI and then adjusted CCI in following Cox proportional hazard models.
Our study has some strengths. To our knowledge, this is the rst study to evaluate the role of postoperative capecitabine duration for the elderly patients with stage III and high-risk stage II colon cancer. We rstly reported a signi cant association between postoperative capecitabine duration and CSS in the surgery treated elderly colon cancer. Furthermore, a L-shape relationship between capecitabine duration and CSS in stage III elderly colon cancer was discovered, and the threshold effect of capecitabine duration on CSS in this group was found. Strict statistical adjustment was applied to minimize residual confounders in this study.
There are still some limitations included in our study. Firstly, we could not evaluate the effect of postoperative chemotherapy duration of 5-uorouracil-alone or other regimens since the regimen applied for elderly colon cancer was capecitabine in both two medical centers. A multi-center study is needed to extract more conclusions. Secondly, bias could be caused by unmeasured confounders and residual, but the sensitivity analysis indicated that the association between postoperative chemotherapy duration and CSS was robust. Thirdly, dose reduction or signi cant delay was not reported. We have not found available published data to assess the in uence of dose reductions or treatment delays on survival in postoperative chemotherapy yet. Actually, a trial is ongoing to provide related knowledge (25).

Conclusions
In conclusion, the postoperative capecitabine duration was signi cantly associated with a decrease of death risk in elderly stage III and high-risk stage II colon cancer. However, a threshold effect of capecitabine duration on survival was found with the turning point as 16 weeks. A short-term duration of capecitabine chemotherapy may be an alternative for the conventional six-month postoperative chemotherapy regimen in the elderly stage III colon cancer. Subgroup analyses by gender, age, stage, histological type, CEA, CA199, MMR status in stage III and highrisk stage II colon cancer.