Health-related quality of life in rectal cancer patients undergoing neoadjuvant chemoradiation with delayed surgery versus short-course radiotherapy with immediate surgery: a propensity score-matched cohort study

Background: Neoadjuvant chemoradiation with delayed surgery (CRT-DS) and short-course radiother- apy with immediate surgery (SCRT-IS) are two commonly used treatment strategies for rectal cancer. However, the optimal treatment strategy for patients with intermediate-risk rectal cancer remains a discussion. This study compares quality of life (QOL) between SCRT-IS and CRT-DS from diagnosis until 24 months after treatment. Methods: In a prospective colorectal cancer cohort, rectal cancer patients with clinical stage T2-3N0- 2M0 undergoing SCRT-IS or CRT-DS between 2013 and 2017 were identified. QOL was assessed using EORTC-C30 and EORTC-CR29 questionnaires before the start of neoadjuvant treatment (baseline) and at 3, 6, 12, 18 and 24 months after. Patients were 1:1 matched using propensity sore matching. Between- and within-group differences in QOL domains were analyzed with linear mixed-effects models. Symptoms and sexual interest at 12 and 24 months were compared using logistic regres- sion models. Results: 156 of 225 patients (69%) remained after matching. The CRT-DS group reported poorer emotional functioning at 3, 6, 12, 18 and 24 months (mean difference with SCRT-IS: (cid:3) 9.4, (cid:3) 12.1, (cid:3) 7.3, (cid:3) 8.0 and (cid:3) 7.9 respectively), and poorer global health, physical-, role-, social- and cognitive functioning at 6 months (mean difference with SCRT-IS: (cid:3) 9.1, (cid:3) 9.8, (cid:3) 14.0, (cid:3) 9.2 and (cid:3) 12.6, respectively). Besides emotional functioning, all QOL domains were comparable at 12, 18 and 24 months. Within-group changes showed a significant improvement of emotional functioning after baseline in the SCRT-IS group, whereas only a minor improvement was observed in the CRT-DS group. Symptoms and sexual interest in male patients at 12 and 24 months were comparable between the groups. Conclusions: In rectal cancer patients, CRT-DS may induce a stronger decline in short-term QOL than SCRT-IS. From 12 months onwards, QOL domains, symptoms and sexual interest in male patients were comparable between the groups. However, emotional functioning remained higher after SCRT-IS than after CRT-DS.


Introduction
Neoadjuvant therapy followed by surgery is the cornerstone of treatment in most patients with rectal cancer [1]. According to the Dutch rectal cancer guideline, patients with high-risk, locally advanced rectal cancer (including irresectable tumors, four or more suspicious regional lymph node metastases and/or suspicious extramesorectal lymph nodes) undergo chemoradiation -45-50 Gy in fractions of 1.8-2 Gy in 5 weeks with concurrent chemotherapyfollowed by delayed surgery (CRT-DS) usually after 8-12 weeks [2]. CRT-DS was designed to downsize high-risk rectal tumors and so to achieve curative resection and to decrease the risk of local recurrence [3][4][5]. Patients with intermediate-risk, resectable rectal cancer, receive short-course radiotherapy -25 Gy in 5 fractionsfollowed by immediate surgery (SCRT-IS) usually within 10 days after the start of radiotherapy. SCRT-IS has shown to improve local control as well as cancer-specific survival in patients with negative resection margins compared to surgery alone [6,7]. However, the optimal treatment strategy for intermediate-risk rectal cancer remains a discussion [8], partly because organ-sparing treatment options are not feasible in patients undergoing SCRT-IS in constrast to patients with a clinical complete response (cCR) following CRT [9].
Literature on SCRT-IS versus CRT-DS showed comparable survival, local and distant recurrence rates, sphincter preservation rate, radical resection rate and late toxicity in patients with various disease stages [10,11]. Nonetheless, a higher rate of acute toxicity after CRT-DS was observed in a Polish randomized trial including patients with a cT3-4 rectal tumor (18% versus 3% after SCRT-IS) [12]. Postoperative complications, however, did not differ significantly between the groups [13]. The Stockholm III trial, in which patients with resectable rectal cancer were randomized between SCRT-IS, SCRT with delayed surgery and long-course radiotherapy with delayed surgery (without concurrent chemotherapy), showed a higher, although non-significant, postoperative complication rate in the SCRT-IS group (50% versus 39% in the long-course radiotherapy group versus 38% in the SCRT with delayed group) and comparable oncological outcomes [14].
Nevertheless, CRT-DS involves a more extensive treatment including chemotherapy and a longer treatment duration than SCRT-IS. On the other hand, CRT may give the opportunity to opt for organ-sparing treatment approaches in case of a cCR [9,15,16], which would be suitable in approximately 15-20% of the LARC patients (based on pathological complete response rates) [17]. Patients undergoing SCRT-IS do not have the chance to achieve cCR and to proceed to organ-preservation because of the short time interval between neoadjuvant therapy and resection. Replacement of SCRT-IS by CRT-DS in patients with intermediate-risk rectal cancer may render patients with a complete response to become eligible for organ-sparing treatment. However, this may have implications for patients' health-related quality of life (QOL) during and after rectal cancer treatment.
Literature on the effect of SCRT-IS versus CRT-DS on patients' QOL during treatment is scarce [18]. A randomized trial from Australia compared SCRT with CRT up to 12 months after surgery and observed no differences [18]. Nevertheless, in this study, the longer treatment duration of CRT-DS was not considered because date of surgery was set as baseline. Based on our previous observational study on QOL during rectal cancer treatment, we noticed worse QOL in patients undergoing CRT-DS compared with SCRT-IS at 6 and 12 months after the start of treatment [19]. However, these findings were based on univariable analyses. In the present study, we, therefore, aimed to compare QOL between SCRT-IS and CRT-DS from start of treatment until 24 months after in a cohort using propensity score matching.

Material and methods
The Dutch Prospective Data Collection Initiative on Colorectal Cancer (PLCRC) cohort includes adult patients with colorectal cancer and has been approved by the Medical Research Ethics Committee of the University Medical Centre (UMC) Utrecht, the Netherlands [20]. Within PLCRC, participants gave informed consent to the collection of clinical outcomes and optionally consented to questionnaires on patient reported outcomes (PROs). For the present study, we selected PLCRC participants of the Utrecht RECTal cancer (U-RECT) sub cohort, that includes patients referred for radiotherapy of rectal cancer to the Radiation-Oncology Department of the UMC Utrecht. We included patients with a cT2-3N0-2M0 enrolled between February 2013 and August 2017, who underwent routine SCRT-IS or CRT-DS with curative intent, who gave informed consent for PROs and who responded to at least one questionnaire ( Figure 1). Patients diagnosed with a cT4 stage (N ¼ 34) or with synchronous distant metastases (N ¼ 11) were excluded because these patients may have poorer short-term prognosis which could influence QOL.
All patients were treated in accordance with the Dutch guideline and underwent intensity-modulated radiotherapy (IMRT) [2]. CRT is delivered to patients with LARC (cT4 or cT3 with a distance to the mesorectal fascia (MRF) of 1 mm, and/or cN2 and/or suspicious extramesorectal lymph node metastases) and consists of 25 Â 2 Gy with concurrent oral Capecitabine (825 mg/m 2 twice a day) followed by delayed surgery. SCRT is administered in patients with intermediate-  risk disease (cT3c-dN0 or cT1-3N1 with a distance to the MRF of >1 mm, and cT2-3N0 before the implementation of the current guideline in 2014) and consists of 5 Â 5Gy followed by immediate surgery. Surgery is performed by the principles of a total mesorectal excision (TME), including low anterior resection (LAR) with or without temporary deviating stoma, abdominoperineal resection (APR) with permanent colostomy or rectosigmoid resection with permanent colostomy (Hartmann's procedure). Adjuvant therapy is not routinely administered. QOL was assessed using the European Organization for Research and Treatment of Cancer (EORTC) core QOL questionnaire (QLQ-C30) [21] and colorectal-specific questionnaire (QLQ-CR29) [22] before the start of neoadjuvant treatment (baseline) and at 3, 6, 12, 18 and 24 months after. The EORTC QLQ-C30 consists of 5 functioning domains (physical, role, emotional, cognitive and social functioning), a global health score and cancer-related symptoms [21]. The EORTC QLQ-CR29 comprises colorectal cancer-specific scales and symptoms [22]. For this study, we presented prevalent rectal cancer treatment-related symptoms [19] including fatigue, insomnia and pain of the EORTC QLQ-C30 and bowel-related items (stool frequency, flatulence and fecal incontinence) and genitourinary-related items (urinary frequency, urine incontinence, impotence and sexual interest) of the EORT QLQ-CR29. Questionnaires were provided online or on paper and collected within the Patient Reported Outcomes Following Initial treatment and Long-term Evaluation of Survivorship (PROFILES) registry [23]. Patient and treatment characteristics were collected from patients' medical files.

Statistics
To decrease the risk of confounding bias in this observational study, patients in the SCRT-IS and CRT-DS group were matched using propensity score matching (PSM). PSM is a statistical matching technique using the probability of treatment assignment conditional on observed covariates [24]. Propensity scores were estimated by logistic regression, with treatment strategy group (CRT-DS versus SCRT-IS) as dependent variable and age (continuous), sex, presence of at least one comorbidity, and tumor location as independent variables. Matching was performed according to the 'nearest neighbour' method using a caliper width of 0.55 times the standard deviation of the logit of the propensity score and 1:1 ratio. Patients were not matched on cT-stage, cN-stage, and MRF involvement as these variables are used as selection criteria for SCRT-IS and CRT-DS. Baseline characteristics Table 1. Baseline characteristics before and after propensity score matching of rectal cancer patients treated with neoadjuvant shortcourse radiotherapy with immediate surgery (SCRT-IS) or neoadjuvant chemoradiation with delayed surgery (CRT-DS).

Original cohort
Matched cohort #  were described before and after matching with use of summary statistics. QOL outcomes were handled according to the EORTC manual and linearly transformed into scores between 0 and 100 [25]. Higher scores indicate better functioning, better global health, and higher symptom severity. To compare betweengroup differences in QOL domains of the EORTC QLQ-C30 at the different time points, linear mixed-effects models were used to take into account the intra-subject correlation between the repeated measurements and included a random intercept, time (as factor), treatment strategy group, adjusted for baseline QOL score and surgical procedure (LAR without stoma, LAR with temporary stoma, Hartmann's procedure and APR) with an autoregressive covariance structure of the first order (AR1) assuming that the correlation systematically decreases with increasing distance between time points [26]. The outcomes were presented as mean differences (MD) with the 95% confidence intervals (CI). Standardized effect sizes (ES) were calculated by dividing the MD by the pooled standard deviation of the baseline score and were categorized into no (ES <0.2), small (ES =0.2-0.4), moderate (ES =0.5-0.7), and large difference (ES >0.7), according to Cohen [27]. Withingroup differences were obtained with linear mixed-effects models including the baseline measurement as outcome variable and stratified by treatment strategy group, presenting the change in QOL relative to baseline for the treatment groups apart. Symptoms were categorized into none (score =0), mild (score =1-49), moderate (score =50-99) and severe complaints (score =100), corresponding to the original 4-point Likert scale answer options (for sexual interest these categories were 'not at all', 'a little', 'quite a bit' and 'very much', respectively). The effect of treatment strategy on symptoms (moderate/severe versus none/mild) and sexual interest (no versus yes) was estimated using univariable logistic regression models. As we were primarily interested in differences in late toxicity between the groups, we formally tested for differences only at 12 and 24 months after the start of treatment. The outcomes were presented as odds ratios (OR) and 95% confidence intervals. Significance level was set at p < 0.05. Statistical Package for Social Sciences (SPSS) version 23 was used and 'MatchIt' and 'opmatch' packages in R version 3.4.1.

Results
Of the 225 patients whom met the inclusion criteria, 106 (47.1%) patients underwent SCRT-IS and 119 (52.9%) CRT-DS ( Figure 1). The CRT-DS group included younger patients with lower rectal tumors, higher cT-stages, higher cN-stages, more MRF involvement, and more APR procedures compared to the SCRT-IS group (Table 1). After matching, 156 (61.2%) patients remained (78 patients in each group). Patients in the CRT-DS and SCRT-IS group were well balanced in terms of the matched variables and surgical procedure ( Table 1). As expected, more patients in the CRT-DS group were diagnosed with a cT3 stage (96.2% versus 64.1% in the SCRT-IS group), positive suspected lymph nodes (cN1 or cN2: 94.9% versus 75.6% in the SCRT-IS group), and distance to the MRF of 1 mm (65.4% versus 1.3% in the SCRT-IS group). Also,   Table S1.

Between-group differences in QOL
Compared with the SCRT-IS group, patients in the CRT-DS group reported significantly poorer emotional functioning at 3 and 6 months with moderate ES (MDs À9.4 and À12.1, respectively) and at 12, 18, and 24 months with small ES (MDs À7.3, À8.0 and À7.9, respectively) ( Table 2). At 6 months, global health, physical-, role-, and cognitive functioning were significantly poorer in the CRT-DS group than in the SCRT-IS group with moderate ES (MDs À8.9, À9.9, À13.6 and À12.3, respectively) and social functioning was poorer with a small ES (MD À9.2). Besides emotional functioning, all functioning scores were comparable between the groups at 12, 18 and 24 months after the start of treatment.

Within-group differences in QOL
Within-group differences show the changes in QOL relative to baseline stratified by treatment strategy group (Figure 2). At 3 months, a significant decline was observed in all QOL domains, but emotional functioning, compared with baseline level in both treatment groups. In the SCRT-IS group, global health and cognitive functioning recovered to baseline level after 3 months, and role functioning after 6 months. Social functioning improved after 3 months but remained (borderline) significantly lower than baseline until 12 months. Emotional functioning showed significant improvement at all follow-up measurements. In the CRT-DS group, global health, social-, cognitive-and role functioning returned to baseline  level after 6 months. Emotional functioning showed an increasing trend but was only significantly improved since baseline at 18 months. In both groups, physical functioning remained significantly poorer up to 24 months compared with baseline level.

Symptoms and sexual interest
In Figure 3, longitudinal outcomes for the selected symptoms and for sexual interest in male patients are presented stratified by SCRT-IS and CRT-DS (differences not tested for  significance). At 12 and 24 months after the start of treatment, no significant differences in moderate/severe symptoms between the treatment strategy groups were observed (Table 3). Also, the probability for having no sexual interest was comparable between SCRT-IS and CRT-DS in male patients at 12 and 24 months after the start of treatment. For female patients, sexual interest was not presented because of the insufficient number of responses.

Discussion
This study showed that global health, physical-, role-, cognitive-and emotional functioning were poorer in the CRT-DS group than in the SCRT-IS group at 6 months after the start of neoadjuvant treatment with moderate effect sizes. Social functioning at 6 months and emotional functioning at 12, 18 and 24 months were poorer in CRT-DS with small effect sizes. Besides better emotional functioning in the SCRT-IS group, all other QOL domains were comparable with CRT-DS on longer-term. Within-group QOL changes showed that in both groups physical functioning was significantly lower at all follow-up measurements compared with baseline. Symptoms of fatigue, insomnia, pain, stool frequency, flatulence, fecal incontinence, urinary frequency, urine incontinence and impotence as well as sexual interest in male patients were comparable between the groups at 12 and 24 months after the start of treatment. Several other studies compared QOL between SCRT-IS and CRT-DS [18,[28][29][30]. As mentioned earlier, an Australian trial longitudinally assessed QOL in 297 rectal cancer patients (cT3N0-2M0) randomized between SCRT-IS plus 6 courses of adjuvant chemotherapy and CRT-DS plus 4 courses of chemotherapy at seven time points up to 12 months after randomization [18]. Similar to our findings, a decline in QOL was observed in both treatment groups shortly after surgery with gradually improvement up to 12 months with the most severely affected domains/symptoms including physical-and role functioning, fatigue, pain, impotence and sexual functioning. In contrast to our study, no significant differences in short-term QOL were observed between SCRT-IS and CRT-DS. This could be explained by the re-arrangement of QOL measurements in the Australian study with date of surgery taken as baseline to align treatment duration. However, in our view, the difference in treatment duration is inherent to SCRT-IS and CRT-DS and forms an essential difference between the treatment strategies that, as suggested by our results, may affect QOL. Our aim was therefore to compare the treatment strategies including surgery and not solely radiotherapy regimens. Besides, in contrast to our study, patients in the Australian trial received adjuvant chemotherapy which may likely affect QOL.
A Dutch cross-sectional study compared QOL at a median follow-up of 58 months after diagnosis between 85 patients routinely treated with CRT-DS and 306 patients treated with SCRT-IS in the TME trial [30]. No significant differences were found in global health and functioning. A Polish cross-sectional study neither observed significant differences in QOL and sexual functioning in 222 cT3-4 rectal cancer patients randomized to SCRT-IS or CRT-DS at 12 months after surgery [28]. In a German cross-sectional study with a median followup of 67 months after diagnosis, no difference was found in QOL between 108 patients treated with SCRT-IS and 117 patients with CRT-DS, except for better physical functioning in the CRT-DS group [29]. These studies support our conclusion that longer-term QOL is comparable between SCRT-IS and CRT-DS.
As shown by the within-group QOL changes, patients undergoing CRT-DS took longer time to recover to pretreatment levels than patients undergoing SCRT-IS. This could be related to the longer neoadjuvant treatment duration, Table 3. Results of univariable logistic regression models on the association between treatment strategy and symptoms (moderate/severe versus none/mild) and sexual interest (no versus yes) at 12 and 24 months after the start of treatment in a matched cohort of rectal cancer patients. chemotherapy administration and/or the timing of surgery. Within 24 months, however, all QOL domains have returned to baseline level or above, except for physical functioning which remained lower compared with baseline in both groups. This is in line with a study that investigated recovery of physical functioning after hospital discharge in colorectal cancer patients that showed that half of the patients had not recovered to baseline function at 6 months after diagnosis and that this was more common in rectal cancer patients [31]. This study suggested that an increase in physical activity after surgery is associated with enhanced recovery of physical functioning. More research should focus on improving physical functioning in rectal cancer patients after treatment. Our findings suggest that emotional functioning is better in patients treated with SCRT-IS than with CRT-DS. Patients in the SCRT-IS group improved to above baseline level, equal to the level of the Dutch normative population (mean score of 89 at 24 months in patients and of 88 in the Dutch population with age of 55-75 years, based on normative population data of PROFILES), which was not the case in the CRT-DS group (mean score of 82 at 24 months). The better emotional functioning in the SCRT-IS group could be related to the shorter treatment duration. Nevertheless, this effect warrants further investigation.
This study has several limitations. First, patients were not randomized to one of the neoadjuvant treatment groups. To minimize the risk of confounding bias, we matched the groups based on their propensity score for treatment conditional on baseline characteristics that may have affected treatment choice. However, this could only be performed for known covariates. Residual confounding can, therefore, not be ruled out. Also, the groups were not matched on clinical disease stage as this was highly correlated with treatment indication. We, therefore, excluded patients with most advanced disease (cT4 and/or M1) and corrected the outcomes on QOL domains for baseline scores, surgical procedure and stoma presence. Besides, the study of the Dutch TME trial, earlier discussed, reported that distance to the MRF, tumor and nodal stage were not associated with QOL in their study population [30]. Also, oncological outcomes, such as recurrence rate, are approximately comparable between resectable and locally advanced rectal cancer patients on the short-term [14,32]. We, therefore, assumed that the differences in disease stage between the groups would not influence QOL during the first 24 months after start of treatment to an important extent. Second, to keep sufficient sample size after matching, the caliper width was set at 0.55 which is wider than recommended in literature [33]. Still, matching was considered successful as the differences in baseline characteristics between the groups were small. Third, the proportion of questionnaire non-responders increased over the time and was slightly higher in the SCRT-IS group, which could have introduced non-response bias of the QOL outcomes, meaning that those who respond to the questionnaire differ from those who do not respond.
Besides the use of CRT to become eligible for organ preservation in intermediate risk rectal cancer patients, SCRT with delayed surgery has been proposed as alternative to SCRT-IS and was investigated by the Stockholm III trial [14] and by a prospective non-randomized study in elderly patients [34]. The Stockholm III trial showed comparable oncological outcomes between SCRT with delayed surgery and SCRT-IS [14]. Despite more radiation-induced toxicity after SCRT with delayed surgery, significantly less postoperative complications were observed in this group compared with SCRT-IS. Nonetheless, concerns about delaying surgery after SCRT may include the risk of tumor repopulation [34,35]. Furthermore, the effect of SCRT with delayed surgery on QOL remains yet unknown. Studies investigating the optimal fractionation of neoadjuvant radiotherapy, with or without use of additional chemotherapy, and the optimal time interval to surgery to allow for organ preservation in intermediate risk rectal cancer are warranted. Besides, larger series of QOL following organ-sparing approaches are needed to support the assumption that patients' QOL after organ preservation is indeed better than after radical surgery [36].
In conclusion, this study suggests that the treatment strategy including CRT with delayed surgery may stronger impair patients' QOL shortly after the start of rectal cancer treatment than SCRT with immediate surgery. However, longerterm QOL seems comparable between both groups, except for a slightly better emotional functioning after SCRT-IS. Furthermore, we showed that patients of both treatment strategies have poorer physical functioning up to 24 months compared with pretreatment status. These results emphasize and stimulate the need for shared-and evidence-based decision making regarding neoadjuvant rectal cancer treatment and its purposes.