Rectal cancer is one of the most common cancers worldwide; in 2020, colorectal cancer accounted for approximately 10% of all newly diagnosed cancers (1). With advances in total mesorectal excision (TME) surgery and multidisciplinary therapy for rectal cancer, the rate of locally recurrent rectal cancer (LRRC) has improved from approximately 30% to around 10%. However, LRRC remains one of the major challenges after rectal cancer surgery because of impaired quality of life and limited long-term survival (2-5).

Recent guidelines and contemporary benchmark data emphasize that management of LRRC should be centralized and multidisciplinary, integrating high-quality imaging, appropriate classification of pelvic compartments, and individualized selection of surgery, (re-)irradiation, and systemic therapy (6). A complete resection (R0) is widely considered a key determinant of treatment and the 2024 Japanese Society for Cancer of the Colon and Rectum (JSCCR) Guidelines weakly recommend performing surgery when an R0 resection is considered achievable (7). Because reoperation is often accompanied by severe adhesions and fibrosis, and because the extent and location of tumor spread are highly variable, frequently necessitating multivisceral resection (MVR), the indication for surgery and the operative approach must be tailored to each individual case.

The Memorial Sloan Kettering Cancer Center (MSKCC) pelvic compartment classification categorizes LRRC by recurrence location (axial, anterior, posterior, lateral) and provides a practical framework to anticipate resectability (8). However, real-world data from single-center practice describing treatment selection and prognosis according to this classification remain limited. Therefore, we retrospectively evaluated LRRC without distant metastasis treated at our institution, focusing on treatment patterns, surgical outcomes, and survival stratified by MSKCC recurrence location.

Patients. We retrospectively reviewed consecutive patients diagnosed with LRRC without distant metastasis treated at Kyushu University between January 2005 and December 2025. LRRC was assessed via computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography/computed tomography (PET/CT), or by endoscopic examination where applicable. Data on clinicopathological parameters at the time of initial primary tumor resection (age, sex, tumor location, preoperative treatment and therapy, surgical approach, type of surgery, pathological stage, resection margins, and adjuvant therapy) and LRRC (the first confirmed evidence of recurrent cancer, recurrent tumor localization, and information regarding recurrent cancer treatment and survival data) were obtained by reviewing medical and pathological reports. The localization of recurrence was classified utilizing the MSKCC Classification System (8). Additionally, data on perioperative outcomes such as operating time, amount of blood loss during surgery, postoperative complications, duration of hospital stay following surgery were collected. Complications were defined as those in which the Clavien-Dindo (CD) grading system was ranked as grade IIIb or higher. All the patients provided written informed consent for participation in this study. This study was approved by the Institutional Review Board for Studies in Humans of Kyushu University (approval number #23394). This study was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments or comparable standards.

Follow-up. Physical examination and blood testing were part of the follow-up, which was performed every three months for the first three years after surgery and every six months thereafter. CT scans were performed every six months. Overall survival (OS) was defined as the interval from the start date of treatment for LRRC (surgery or initiation of nonsurgical therapy) to death from any cause or last follow-up. Re-recurrence-free survival (RRFS) was defined as the interval from the date of salvage surgery for LRRC to documented re-recurrence or last follow-up without re-recurrence; RRFS was evaluated only in surgically treated patients.

Treatment. The primary tumor resection was performed for radical rectal resections. All patients underwent central vascular ligation, and all lymph nodes surrounding the tumor-supplying vessels and the entire mesocolon were excised. The choice between open, laparoscopic, or robotic approach for primary tumor resection and LRRC surgery was made by the operating surgeon and the team. Laparoscopic surgery was performed using the conventional five-port technique, whereas robotic surgery was performed using the da Vinci Si or Xi surgical systems. The surgeon made the intraoperative decision to convert to open surgery. Multidisciplinary management for LRRC consisted of short-course radiotherapy (25.0 Gy in five fractions) or long-course radiotherapy (50.4 Gy in 28 fractions) combined with an oral 5-fluorouracil-based agent. Chemotherapy consisted of Capecitabine plus Oxaliplatin (CAPOX), S-1 plus Oxaliplatin (SOX), or 5-Fluorouracil, Leucovorin, and Oxaliplatin (FOLFOX) with or without targeted antibodies such as bevacizumab.

Statistical analysis. All statistical analyses were performed using JMP Pro 16 software (SAS Institute, Cary, NC, USA). Continuous variables are presented as medians and interquartile ranges. OS and RRFS were estimated using the Kaplan-Meier method, and differences between groups were assessed using the log-rank test. Statistical significance was set at p<0.05. Due to the limited sample size, multivariable analysis was not performed in the present study.

Patient characteristics at primary tumor resection. The median age at primary tumor resection was 65 years [interquartile range (IQR)=61-73], and 15 patients (75.0%) were male (Table I). Preoperative therapy for the primary tumor was administered in four patients (20.0%) (chemoradiotherapy, n=1; chemotherapy, n=3). The primary surgical approach was open in seven patients (35.0%), laparoscopic in 10 (50.0%), and robotic in three (15.0%). Postoperative major complications (Clavien-Dindo ≥3b) occurred in two patients (10.0%) due to anastomotic leakage. Pathological stage was II in six patients (30.0%), III in 10 (50.0%), and IV in four (20.0%). Adjuvant therapy was administered in 10 patients (50.0%), and lateral pelvic lymph node dissection was performed in two (10.0%).

Patient characteristics at diagnosis of LRRC. The median time from primary tumor resection to LRRC diagnosis was 30 months (IQR=22-40) (Table II). Recurrence location was categorized using the MSKCC pelvic compartment classification (axial, anterior, posterior, and lateral) (Figure 1). According to this classification, 10 patients (50.0%) had axial recurrence, two (10.0%) anterior, two (10.0%) posterior, and six (30.0%) lateral recurrence (Table II, Figure 1). With respect to multidisciplinary management for LRRC, eight patients (40.0%) underwent surgery alone. Nine patients (45.0%) received preoperative therapy followed by surgery, including chemoradiotherapy plus surgery in two (10.0%), short-course radiotherapy plus surgery in one (5.0%), and chemotherapy plus surgery in six (30.0%). The remaining three patients (15.0%) did not undergo surgery and received definitive chemoradiotherapy (n=1), chemotherapy (n=1), or best supportive care (n=1).

Surgical outcomes for LRRC. Seventeen patients underwent salvage surgery for LRRC (Table III). The surgical approach was open in seven patients (41.2%), laparoscopic in nine (52.9%), and robotic in one (5.9%). Procedures included low anterior resection in two patients (11.8%), abdominoperineal resection in 10 (58.8%), abdominoperineal resection with sacrectomy in one (5.9%), Hartmann’s procedure in one (5.9%), and pelvic exenteration in three (17.6%). The median operative time was 477 min (IQR=413-515), median blood loss was 592 ml (IQR=257-1832), and median postoperative length of stay was 29 days (IQR=21-38). Major postoperative morbidity (Clavien-Dindo ≥3b) occurred in one patient (5.9%) due to bladder injury, and there was no postoperative mortality. Regarding margin status, R0 resection was achieved in 12 patients (70.5%), whereas R1 and R2 resections were observed in four (23.6%) and one (5.9%), respectively. Additional radiotherapy was delivered to one patient with R1 resection and one patient with R2 resection; however, the remaining three patients with R1 resection did not receive additional local therapy due to postoperative conditions or early re-recurrence.

Treatment patterns and outcomes according to recurrence location. Treatment strategies differed by recurrence location (Table IV). Surgery alone was performed in seven patients (70.0%) with axial recurrence and one patient (50.0%) with anterior recurrence, whereas no patients with posterior or lateral recurrence underwent surgery alone. Preoperative therapy was more frequently applied in posterior recurrence (2/2) and in surgically treated lateral recurrence (3/3), reflecting the complexity of these locations. Notably, all three patients who did not undergo surgery had lateral recurrence. The distribution of surgical procedures also varied by recurrence location: pelvic exenteration was performed in two patients (20.0%) with axial recurrence and one patient (33.3%) with lateral recurrence, and abdominoperineal resection with sacrectomy was performed in one patient (50.0%) with posterior recurrence. R0 resection was achieved in seven patients (70.0%) with axial recurrence, all patients with anterior recurrence (2/2, 100%), one patient (50.0%) with posterior recurrence, and two patients (66.7%) with lateral recurrence. Re-recurrence after LRRC treatment was observed in four patients (40.0%) with axial recurrence, in none with anterior recurrence, and in all patients with posterior recurrence (2/2, 100%) and in all surgically treated patients with lateral recurrence (3/3, 100%). Re-recurrence sites included the lung, liver, local pelvis, and skin; in some patients, re-recurrence sites overlapped.

Survival according to recurrence location. Given the unfavorable clinical course observed in posterior and lateral recurrences, patients were grouped into an axial/anterior group and a posterior/lateral group for survival analyses (Figure 2). OS after treatment for LRRC was significantly worse in the posterior/lateral group than in the axial/anterior group (p=0.044) (Figure 2A). Similarly, RRFS after treatment for LRRC was significantly worse in the posterior/lateral group than in the axial/anterior group (p=0.0059) (Figure 2B).

In this single-center retrospective study of 20 patients with LRRC without distant metastasis, we found that recurrence location stratification using a pelvic compartment approach was clinically informative for treatment selection and prognosis. Posterior and lateral recurrences were associated with markedly unfavorable outcomes compared with axial/anterior disease, as demonstrated by significantly worse OS and RRFS in the posterior/lateral group. Recent national guidelines highlight that optimal management of LRRC requires a centralized, multidisciplinary approach supported by high-quality imaging, standardized pelvic compartment assessment, and individualized selection of surgery, (re-)irradiation, and systemic therapy (6). In line with this framework, our data suggest that recurrence location is a practical surrogate of anatomical complexity and can stratify prognosis in routine clinical practice. In particular, posterior and lateral disease showed unfavorable outcomes despite salvage surgery, underscoring the need for careful patient selection and consideration of alternative or intensified multimodal strategies when durable local control by surgery alone is uncertain.

Anatomical extent is a major determinant of resectability and oncologic outcomes in LRRC (8). Posterior and lateral recurrences frequently involve structures such as the sacrum/coccyx and the pelvic sidewall, including the internal iliac vessels, nerves, pelvic bones, and surrounding musculature, which often necessitate technically demanding multivisceral resections to achieve negative margins (8-10). In our cohort, all surgically treated posterior and lateral cases developed re-recurrence, highlighting the aggressive clinical behavior and limitations of local control in these compartments even when salvage surgery is attempted. Conversely, axial/anterior recurrences showed comparatively better outcomes, consistent with the concept that centrally located disease may be more amenable to complete resection and durable control.

Our results also underscore the need to consider multimodal treatment strategies, particularly for posterior and lateral LRRC. Ongoing prospective efforts are evaluating the role of preoperative chemoradiotherapy to improve resectability and local control in resectable LRRC, while advanced radiation modalities, including particle therapy, have been explored as alternatives or adjuncts when R0 resection is unlikely (11-13). In a recent single-center study of 21 patients who underwent surgery for postoperative pelvic recurrence of rectal cancer, Shoji et al. reported that preoperative chemotherapy or chemoradiotherapy was associated with a higher rate of negative resection margins than upfront surgery, and that postoperative recurrence was least frequent among patients treated with neoadjuvant chemoradiotherapy (14). These findings support the potential value of tailoring preoperative treatment according to the anatomical extent and anticipated difficulty of margin-negative resection. In our practice, preoperative therapy was more commonly used for posterior disease and for surgically managed lateral disease, reflecting the anticipated technical difficulty of achieving negative margins in these compartments. Although our study was not designed to compare specific regimens, the consistently poor outcomes observed in posterior/lateral recurrences suggest that treatment intensification and careful selection of local therapy modality –including definitive radiation approaches in selected patients– may be reasonable considerations.

An important and somewhat unexpected observation was that margin status (R0 vs. R1/R2) did not translate into statistically significant differences in OS or RRFS in this cohort (Figure 3). Recent Japanese multicenter data have nevertheless supported the oncological value of complete resection. Maruyama et al. evaluated 18 patients who underwent R0 resection for colorectal cancer local recurrence and reported a 5-year recurrence-free survival rate of 39.4% and a 5-year overall survival rate of 52.2%, although 44.4% of patients developed re-recurrence (15). These findings indicate that R0 resection can provide meaningful long-term survival in selected patients, while also illustrating the substantial residual risk of re-recurrence even after curative surgery. Because R1/2 resection is widely recognized as a strong adverse prognostic factor after salvage surgery for LRRC, this finding requires cautious interpretation (15-18). Several explanations may account for the lack of separation between groups in our analysis. First, the sample size was small (R0, n=12; R1/R2, n=5), resulting in limited statistical power. Second, substantial treatment-selection bias is unavoidable: anatomical complexity and tumor biology likely influenced both the ability to achieve R0 resection and the choice/intensity of perioperative therapies. Finally, post-re-recurrence management may have confounded survival comparisons. In our series, patients with re-recurrence after R0 resection often underwent systemic therapy or reoperation but eventually died, whereas a subset of patients with re-recurrence after R1/2 resection received additional treatments (chemotherapy, radiotherapy, or surgery) and remained alive at last follow-up. This imbalance in subsequent therapy intensity and effectiveness could attenuate survival differences attributable solely to surgical margin status.

Study limitations. This study has limitations inherent to its retrospective, single-center design and small cohort size. The heterogeneity of treatment approaches across two decades, variability in preoperative regimens, and unmeasured confounders (e.g., prior radiation exposure, detailed imaging-based invasion patterns, and performance status) further limit causal inference. Nevertheless, our real-world data suggest that recurrence location is a clinically relevant factor for risk stratification in LRRC. In particular, posterior and lateral recurrences were associated with a high risk of re-recurrence and poor survival, supporting the need for careful patient selection for salvage surgery and consideration of intensified multimodal strategies. Future multicenter studies with standardized treatment algorithms are warranted to refine location-based decision-making and optimize outcomes for patients with LRRC.

Recurrence location stratification was useful for risk stratification and treatment planning in LRRC. Posterior and lateral recurrences showed poor outcomes even after salvage surgery, suggesting that careful patient selection and consideration of intensified multimodal strategies are warranted.

The Authors declare no conflicts of interest in relation to this study.

Sho Nambara conceived and designed the study, collected and curated the data, performed the statistical analyses, prepared the tables and figures, and drafted the manuscript. Koji Ando contributed to the study design, supervised the data interpretation, and critically revised the manuscript. Tetsuro Kawazoe, Shotaro Korehisa, Yasuo Tsuda, and Tomonori Nakanoko were responsible for patient enrollment, surgical treatment, and data acquisition at each participating institution, and they reviewed and edited the manuscript. Eiji Oki and Tomoharu Yoshizumi supervised the overall conduct of the study, contributed to the study design and interpretation of the data, and critically reviewed the manuscript for important intellectual content. All Authors read and approved the final version of the manuscript and agree to be accountable for all aspects of the work.

The Authors are grateful to the medical, nursing, and administrative staff at Kyushu University Hospital for their dedicated support in patient care and data collection. The authors also thank the pathologists and radiologists at each participating institution for their assistance with the diagnostic evaluations.

This study did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.

During the preparation of this manuscript, a large language model (ChatGPT, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.