Local versus radical surgery for early rectal cancer with or without neoadjuvant or adjuvant therapy

Total mesorectal excision is the standard of care for stage I rectal cancer. Despite major advances and increasing enthusiasm for modern endoscopic local excision (LE), uncertainty remains regarding its oncologic equivalence and safety relative to radical resection (RR).

To assess the oncologic, operative, and functional outcomes of modern endoscopic LE compared to RR surgery in adults with stage I rectal cancer.

We searched CENTRAL, Ovid MEDLINE, Ovid Embase, Web of Science - Science Citation Index Expanded (1900 to present), four trial registers (ClinicalTrials.gov, ISRCTN registry, the WHO International Clinical Trials Registry Platform, and the National Cancer Institute Clinical Trials database), two thesis and proceedings databases, and relevant scientific societies' publications in February 2022. We performed handsearching and reference checking and contacted study authors of ongoing trials to identify additional studies.

We searched for randomized controlled trials (RCTs) in people with stage I rectal cancer comparing any modern LE techniques to any RR techniques with or without the use of neo/adjuvant chemoradiotherapy (CRT).

We used standard Cochrane methodological procedures. We calculated hazard ratios (HR) and standard errors for time-to-event data and risk ratios for dichotomous outcomes, using generic inverse variance and random-effects methods. We regrouped surgical complications from the included studies into major and minor according to the standard Clavien-Dindo classification. We assessed the certainty of evidence using the GRADE framework.

Four RCTs were included in data synthesis with a combined total of 266 participants with stage I rectal cancer (T1-2N0M0), if not stated otherwise. Surgery was performed in university hospital settings. The mean age of participants was above 60, and median follow-up ranged from 17.5 months to 9.6 years. Regarding the use of co-interventions, one study used neoadjuvant CRT in all participants (T2 cancers); one study used short-course radiotherapy in the LE group (T1-T2 cancers); one study used adjuvant CRT selectively in high-risk patients undergoing RR (T1-T2 cancers); and the fourth study did not use any CRT (T1 cancers). We assessed the overall risk of bias as high for oncologic and morbidity outcomes across studies. All studies had at least one key domain with a high risk of bias. None of the studies reported separate outcomes for T1 versus T2 or for high-risk features. Low-certainty evidence suggests that RR may result in an improvement in disease-free survival compared to LE (3 trials, 212 participants; HR 1.96, 95% confidence interval (CI) 0.91 to 4.24). This would translate into a three-year disease-recurrence risk of 27% (95% CI 14 to 50%) versus 15% after LE and RR, respectively. Regarding sphincter function, only one study provided objective results and reported short-term deterioration in stool frequency, flatulence, incontinence, abdominal pain, and embarrassment about bowel function in the RR group. At three years, the LE group had superiority in overall stool frequency, embarrassment about bowel function, and diarrhea. Local excision may have little to no effect on cancer-related survival compared to RR (3 trials, 207 participants; HR 1.42, 95% CI 0.60 to 3.33; very low-certainty evidence). We did not pool studies for local recurrence, but the included studies individually reported comparable local recurrence rates for LE and RR (low-certainty evidence). It is unclear if the risk of major postoperative complications may be lower with LE compared with RR (risk ratio 0.53, 95% CI 0.22 to 1.28; low-certainty evidence; corresponding to 5.8% (95% CI 2.4% to 14.1%) risk for LE versus 11% for RR). Moderate-certainty evidence shows that the risk of minor postoperative complications is probably lower after LE (risk ratio 0.48, 95% CI 0.27 to 0.85); corresponding to an absolute risk of 14% (95% CI 8% to 26%) for LE compared to 30.1% for RR. One study reported an 11% rate of temporary stoma after LE versus 82% in the RR group. Another study reported a 46% rate of temporary or permanent stomas after RR and none after LE. The evidence is uncertain about the effect of LE compared with RR on quality of life. Only one study reported standard quality of life function, in favor of LE, with a 90% or greater probability of superiority in overall quality of life, role, social, and emotional functions, body image, and health anxiety. Other studies reported a significantly shorter postoperative period to oral intake, bowel movement, and off-bed activities in the LE group.

Based on low-certainty evidence, LE may decrease disease-free survival in early rectal cancer. Very low-certainty evidence suggests that LE may have little to no effect on cancer-related survival compared to RR for the treatment of stage I rectal cancer. Based on low-certainty evidence, it is unclear if LE may have a lower major complication rate, but probably causes a large reduction in minor complication rate. Limited data based on one study suggest better sphincter function, quality of life, or genitourinary function after LE. Limitations exist with respect to the applicability of these findings. We identified only four eligible studies with a low number of total participants, subjecting the results to imprecision. Risk of bias had a serious impact on the quality of evidence. More RCTs are needed to answer our review question with greater certainty and to compare local and distant metastasis rates. Data on important patient outcomes such as sphincter function and quality of life are very limited. Results of currently ongoing trials will likely impact the results of this review. Future trials should accurately report and compare outcomes according to the stage and high-risk features of rectal tumors, and evaluate quality of life, sphincter, and genitourinary outcomes. The role of neoadjuvant or adjuvant therapy as an emerging co-intervention for improving oncologic outcomes after LE needs to be further defined.