Laparoscopic curative surgery for T3 rectal cancer with uninvolved circumferential resection margin on magnetic resonance imaging without neo-adjuvant treatment: Findings from a 3-year follow-up study

Settings and patients

This longitudinal cohort study,a clinical intervention without a control group, was conducted at the University Medical Centre, a teaching hospital of the University of Medicine & Pharmacy at Ho Chi Minh City, from January 2018 to December 2022. Patients were included retrospectively before and prospectively after ethical approval, all treated under a standardised protocol to ensure a consistent cohort and minimise bias.

Inclusion criteria comprised pathologically confirmed rectal adenocarcinoma (≤ 12 cm from the AV), staged T3a/b/c (excluded T3c at low rectum), uninvolved mrCRM, negative EMVI, based on pre-operative enhanced rectal MRI with the following sequences: (1) Coronal T2 HASTE 5 mm, (2) Sagittal T2W 3.5 mm, (3) Coronal, and axial high-resolution T2W 3 mm, (4) T1W 3.5 mm.

Exclusion criteria were individuals with a history of hereditary nonpolyposis colorectal cancer, familial adenomatous polyposis, or who had undergone surgery for other cancers. Patients who were unable to contact, non-compliant with treatment, or had synchronous disease. Metastatic rectal cancer and inflammatory bowel disease were also excluded.

Patients had pre-operative assessment with digital rectal examination, colonoscopy, biopsy, thoraco-abdominal-pelvic computed tomography scan, pelvic magnetic resonance imaging (MRI), and carcinoembryonic antigen (CEA) levels, as well as other laboratory tests. These results, along with the performance and financial status of each patient, were discussed in a multi-disciplinary team (MDT) meeting to decide on initial treatment and another MDT consultation after surgery with ana-pathology results to establish the judiciousness of adjuvant therapy.

Adjuvant therapy was guided by pathological risk: pT3N0 patients with high-risk features (low tumour, poor differentiation, lymphovascular or perineural invasion, or <12 lymph nodes) were advised pelvic radiotherapy with capecitabine. Patients with pT4 tumours, nodal metastasis, or involved pCRM received chemoradiotherapy with XELOX and pelvic radiation. Final regimens were tailored based on MDT recommendations, physician judgment, and patient compliance.

Follow-up included clinical exam, CEA, chest X-ray, and abdominal ultrasound every 3–6 months for 2 years, then every 6 months until year 5. Thoraco-abdominal–pelvic computed tomography(CT) (and pelvic MRI if indicated) was performed every 6–12 months. Colonoscopy was done at 1 year and every 3–5 years thereafter.

Operative procedures

All laparoscopic rectal resections were radical, adhering to total mesorectal excision (TME) principles or partial mesorectal excision (PME). PME was defined as transverse resection of mesorectum to the rectal wall with a 2 to 4 cm distal margin for middle (approved by MDT consensus) and upper rectal cancers.^[11]The technique involved a medial to lateral approach; the inferior mesenteric artery was ligated by the root, and the inferior mesenteric vein was ligated below the lower border of the pancreas. Then, dissection was continued by a lateral to medial approach, mobilising the rectum, denuding the rectal wall, and transecting the rectum by the appropriate distal margin.

Study outcomes

The primary endpoint was LR, while secondary endpoints included DM, OS, and DFS. Pelvic and peri-anastomotic recurrence were defined as LR, whereas recurrences outside the pelvis were classified as DM. OS referred to the duration from the date of surgery to death from any cause or to the last follow-up. DFS denoted the period from the date of surgery to LR, DM, death due to the disease, or the last follow-up.

Independent variables included perioperative data and pathological findings. Perioperative data encompassed age, sex, body mass index (BMI), history of morbidity, ASA score, CEA level, differentiation, tumour location, tumour diameter, mrCRM, clinical tumour–node–metastasis (TNM), operative procedure, anastomosis, and ileostomy. The primary surgeon assessed the macroscopic specimen, evaluating mesorectal integrity based on the Enker classification^[12], and measured the lengths of proximal and distal margins. Pathological findings, such as the number of lymph nodes harvested, lymph node metastasis, resection margin, pCRM, vascular invasion, and neural invasion, were determined through microscopic assessment by pathologists.

Short-term outcomes comprised intraoperative iatrogenic events, post-operative complications, mortality, and length of hospitalisation. Iatrogenic events were defined as complications arising during surgery, whereas postoperative morbidity and mortality included events occurring during the hospital stay or within 30 days postoperatively. Post-operative complications were categorised as Clavien– Dindo ^[13] grade II or higher, comprising wound infection, anastomotic leakage or stricture, intra-abdominal abscess, or ileus.

Data analysis

Data analyses were performed using R 4.4.2 (October 31, 2024 update). A significant level of 5% was employed. Categorical variables were represented using frequency and percentage, while numerical variables were represented using median and inter-quantile range (unless mean and standard deviation were specified, depending on variable distribution according to the Shapiro-Wilk test). Differences in categorical variables between groups were evaluated using the Chi-square and Fisher’s exact test, with the latter used when the expected frequency count was less than five.

The 3- and 5-year recurrence rates were estimated using Kaplan-Meier analysis, and the log-rank test was conducted to compare differences between stages II and III rectal cancer (analysed using the SPSS™ software package (SPSS Inc., Chicago, Illinois). Univariate Cox regression analysis was conducted for prespecified predictors of interest, with variables possessing a p-value of less than 0.1 selected for multivariate Cox regression. Ridge regression was used to estimate the effects of prespecified predictors on LR and DM.

Research ethics

The study received approval from the local ethics committee of the University Medical Centre on November 12, 2020 (number 50/GCN-HĐĐĐ and was performed in accordance with the Declaration of Helsinki. Written informed consent was obtained at enrolment for prospective patients and during follow-up for retrospective cases.

Short-term outcomes

Three iatrogenic complications (2.8%), comprising one case each of massive bleeding, small intestine perforation, and autonomic pelvic nerve injury, were observed. Postoperative complications occurred in 13 patients (11.9%): wound infection (5, 4.6%; Grade I), urinary infection (3, 2.8%; Grade II), ileus (1, 0.9%; Grade II), bowel obstruction (1, 0.9%; Grade IIIb), parastomal hernia (1, 0.9%; Grade IIIb), incisional hernia (1, 0.9%; Grade IIIb), and peroneal nerve paralysis (1, 0.9%; Grade II). Anastomosis was performed in 85 patients, of whom 3 developed anastomotic leakage. The median hospitalisation time was seven days (IQR: 6-8). Hospitalisation time exceeded ten days in patients with postoperative complications. A 75-year-old patient with a pre-history of chronic obstructive pulmonary disease(COPD) and myocardial ischemia required hospitalisation for 22 daysfollowing bowel obstruction proximal to the ileostomy.

Macroscopic mesorectum assessment showed 1 (0.9%) incomplete, 3 (2.8%) nearly complete, and 105 (96.3%) complete. The median number of lymph nodes harvested was 12 (IQR: 10-14), and the mean number of metastatic lymph nodes was 0.8 (SD = 1.65). There were 40 cases (36.7%) with a total of less than 12 lymph nodes investigated, and 34 cases (31.2%) had lymph node metastasis. LNR was 0.06. An R0 resection of both the proximal and distal margin was achieved in all 109 cases, while the pCRM(+) rate was 2.8%.

Oncologic outcomes

Oncologic and pathologic outcomes are provided in [Table 2]. The mean follow-up time is 42.5 months. The rates of LR and DM were 5.5% and 11%, respectively, with the majority being liver metastases. Among these, three cases of synchronous LR and DM were detected: (1) Liver and lung metastases with pelvic recurrence; (2) Peritoneal metastases with anastomotic recurrence; (3) Liver and brain metastases with pelvic recurrence. The 3-, 5-year cumulative rates of LR and DM were 4.8%, 5.9%, and 10.4%, 13.6%, respectively. There was no significant difference in terms of LR (p = 0.47), while there was a significant difference associated with DM (p = 0.004) between stage II and stage III [Figure 2a-b].

Close

Figure 2:

Cumulative local recurrence (a) Distant metastasis (b) Overall survival and (c) Disease survival (d). Cumulative 3- and 5-year rates: LR 4.8% and 5.9%, DM 10.4% and 13.6%. LR did not differ between stage II and III (p=0.47), whereas DM was significantly higher in stage III (p=0.004). OS was 93.5% and 88.6% at 3 and 5 years, DFS 91.5% and 73.7%; both were significantly worse in stage III (OS p=0.009; DFS p=0.005). DM: Distant metastasis, LR: Local recurrence, OS: Overall survival, DFS: Disease-free survival.

Export to PPT

No trocar-site recurrence was detected during the follow-up period. Additional outcomes on demographic, clinical characteristics, and short-term pathologic-oncologic outcomes are provided in [Supplementary Table 1 and 2].

We used a univariate screening approach to identify predictors with a p-value of less than 0.1, which were then fitted in a Multivariate Cox regression.

Our multivariate analysis indicated Hartmann’s surgery (p = 0.04), anastomosis leakage (p = 0.02), and violated mesorectum (p < 0.005), to be significant predictors of LR. Using Ridge regression to allow for full model fitting and variable interaction, the final model revealed that among the prespecified variables of interest, pCRM (HR:3.28) and violated mesorectum (HR:2.02) had a significant effect on the outcomes, along with anastomosis leakage being mildly related to LR (HR:1.86) [Table 3]. Meanwhile, multivariate regression indicated CEA level equal to or greater than 5 ng/ml (p=0.01) and pN(+) (p=0.01) to be statistically significant predictors of DM. Full model fitting with prespecified variables yielded a violated mesorectum (HR:4.434), and involved pCRM (HR:5.393) as a favourable predictor despite coefficient regularisation with Ridge regression [Table 4].

Additional results on univariate analysis for screening independent factors (p< 0.1) affecting LR and DM are presented in the [Supplementary Tables 3-4].

Survival outcomes

The 3-year and 5-year OS rates were 93.5% and 88.6% while the 3-year and 5-year DFS rates were 91.5% and 73.7%, respectively. There was a significant difference in terms of OS (p = 0.009) and DFS (p =0.005) between stage II and stage III [Figure 2c-d].

Oncologic outcomes

During the mean follow-up period of 42.5 months, the present study revealed an LR rate of 5.5% and a DM rate of 11%. These outcomes are encouraging and alignwith similar population studies, as Ptok et al.^[16] (5.3%, 28.9%; 48.9 months follow-up), Kulu et al.^[17] (4.5%, 10%; 29 months follow-up), and prospective randomised comparison studies between the TME alone group and the nCRT+TME group, conducted by Fan et al.^[18] (4.3%, 12.8%) and Wang et al.^[19] (5%, 15.7%). According to Ptok et al., local control can be maintained without nCRT before TME.^[16] In a propensity score matching study (2024) for T3, mid-low rectal cancer with uninvolved mrCRM, Zeng et al., reported a 3.3% local recurrence rate in the surgery-alone group and concluded that long-term outcomes were comparable to those with nCRT + TME, suggesting potential overtreatment with nCRT in selected patients.^[20]

In the present cohort, the 3- and 5-year cumulative rates of LR were 4.8% and 5.9%, respectively, which were higher than the corresponding rates of 1.2% and 2% reported in the OCUM study.^[21] In contrast, the 3- and 5-year cumulative rates of DM were 10.4% and 13.6%, respectively, comparable to the OCUM study (8.9% and 14.4%)^[21], but notably lower than the 19% reported for the T3 subgroup in the MERCURY study.^[9]These differences likely reflect variations in study populations: OCUM^[21]included T2 and only T3a/b tumours while excluding low rectal T3 lesions, and MERCURY^[9] further restricted enrolment to T2–T3a/b tumours.

LR was significantly associated with anastomosis leakage and violated mesorectum, with each factor increasing the risk of LR approximately twofold. Meanwhile, Hartmann's surgery showed inconsistent results between the multivariate regression with variable selection and the Ridge regression models. One of the reasons may have stemmed from the effect of collinearity between Hartmann's surgery and other variables, masking its effect in a full model approach. Furthermore, among the two patients who received Hartmann's surgery, one suffered from LR, while the other did not. The second patient's pathology result was pT3 with uninvolved pCRM, which did not support the intraoperative conclusion of pelvic wall involvement leading to Hartmann's surgery. Therefore, we suspect the relationship between Hartmann's surgery and LR may not have been accurately depicted and may require further investigation. Additionally, although multivariate regression suggested that involved pCRM was not a significant predictor for LR, Ridge regression, by allowing for variable interactions, revealed that patients with involved pCRM had a threefold higher risk of LR. This result concurs with the work of Räsänen et al., as their study showed pCRM to be significantly associated with LR, along with tumourlocation <6 cm from the AV.^[22] Nonetheless, our study did not find a significant relation between tumourlocation and LR. Theoretically, anastomotic leakage is not a direct cause of LR, but difficult pelvic dissection with breach of the rectal wall may increase both leakage risk and subsequent local failure, while also delaying adjuvant therapy. Similarly, a violated mesorectum or involved pCRM can disseminate tumourcells within the pelvis or bowel lumen, predisposing to LR.

With respect to DM, elevated pre-operative CEA (≥5 ng/mL) and pN(+) status were significantly associated in univariate analysis and remained predictors in subsequent multivariate regression. However, their effects diminished when Ridge regression was applied, likely due to collinearity with other covariates. Notably, Ridge regression revealed nearly a fourfold increased risk of DM in patients with mesorectal violation and a fivefold increase in those with pCRM involvement. Although these factors were not significant in univariate screening, the full-model approach of Ridge regression accounted for variable interactions and may have provided a more accurate representation of their relationship with DM. In comparison, Tang et al.’s study demonstrated similar independent risk factors of DM, including pre-operative CEA level and pN staging, while detecting tumourdiameter, positive EMVI, and the number of lymph nodes harvested as important predictors of DM.^[23]

Survival outcomes

In the present study, the 3- and 5-year OS rates were 93.5% and 88.6%, while the corresponding DFS rates were 91.5% and 73.7%. Comparable outcomes have been reported in similar populations, selected cohorts: the OCUM study^[21]reported 84.9% and 76% of 3- and 5-year DFS rates. Ptok et al.^[16] reported a 5-year DFS rate of 71.1%, Zeng et al.^[20] observed 3-year OS and DFS rates of 90.3% and 78.3%; and Kulu et al.^[17] documented 5-year OS and DFS rates of 88.2% and 79.1%, respectively. Our findings for 3- and 5-year OS rates were comparable to those in the studies by Fan et al.^[18] and Wang et al.^[19](90.7%, and 86.5%), while there was a difference in the 3- and 5-year DFS rates (85.7% and 84.3%). Notably, Wang et al.^[19], extending the follow-up of Fan et al.^[18], concluded that long-term outcomes did not clearly support a survival benefit from nCRT, a finding also echoed by Kulu et al.^[17]

Short-term outcomes

The complications were slightly higher in our study: 2.8% of iatrogenic complications, 11.9% of post-operative complications, 3.5% of anastomotic leakage, and 0% of mortality, compared to Fan et al.^[18] (13.9% in overall morbidity). Our findings were equivalent to or lower than those of Zeng et al.^[20]results (11% of post-operative complications, 6.3% of anastomotic leakage, 0.2% of mortality), Kulu et al.^[17] (51.7% morbidity, 0% mortality), and Cho et al.^[24](19%). Compared to Cho et al.^[24] and Fan et al.^[18], the mean hospitalising time was relatively shorter (7.4 versus 12.7 and 9 days).

Compared to some studies with similar selected samples, such as Kulu et al.^[17], Ptok et al.^[16], and Zeng et al.^[20], the median number of lymph nodes harvested in the present study was relatively lower (12 versus 16, 18.4, 19.6), leading to a lower rate of mean number of metastatic lymph nodes (31.2% vs 40% in Ptok et al.^[16] study, and 50.7% as per Zeng et al.^[20] study). These may be due to the unstandardised lymph node harvesting process, as the lymph node was dissected by hand and naked eye, as well as the mesentery fatty tissue dissolving agent, like Xylene, was not available at that time.

The study cohort showed encouraging outcomes with 2.8% involved pCRM vs 1.8%, 4.5%, and 0.8% in studies of Kulu et al.^[17], Ptok et al.^[16], and Zeng et al.^[20] In terms of mesorectal integrity, the complete-nearly complete-incomplete rate was 96.3% - 2.8% - 0.9%, which is reasonable as compared to those in studies of Ptok et al.^[16] (84% - 16% - 0%), and Zeng et al. (91.2% - 8.8% - 0%). These positive outcomes could be attributed to several factors involving both the patients and the participation of the multi-disciplinary team (MDT). The tumourwas relatively small and less invasive, with T3a/b/c staging. The involvement of well-trained surgeons, dedicated radiologists, and pathologists might have contributed to the consistency and quality of the surgery.

LNR is the most significant prognostic factor for OS and DFS, especially in patients with fewer than 12 lymph nodes harvested.^[25,26] Lee et al. proved that a higher LNR was predictive of poorer DFS and OS.^[27] Our study showed a relatively low LNR of 0.06, which explains why OS and DFS rates in our study were quite high, while 36.7% of cases had fewer than 12 lymph nodes harvested.

Adjuvant treatment

Despite recommendations from the MDT to undergo adjuvant therapy, treatment adherence was suboptimal. Specifically, 42 patients (38.5%) did not receive any adjuvant treatment, 40 (36.7%) received combined chemoradiotherapy, 26 (23.6%) received chemotherapy alone, and one patient (0.9%) underwent radiotherapy. These treatment patterns were not consistent with standard adjuvant protocols. Possible explanations include patient refusal, socioeconomic barriers, limited access to oncology services, and concerns about treatment-related toxicity, all of which may have influenced adherence and ultimately affected long-term outcomes. The median number of harvested lymph nodes was acceptable at 12, although this figure remains lower than that reported in some other studies such as Kulu et al.^[17], Ptok et al.^[16], and Zeng et al.^[20]. Additionally, 40 (36.7%) cases with fewer than 12 lymph nodes harvested may contribute to 69.3% of cases without lymph node metastasis. Although the indication for adjuvant therapy is not determined solely by nodal status, inadequate lymph node retrieval can lead to under-staging and potential omission of adjuvant treatment. Consequently, surgery played an even greater role in controlling LR and DM in this study.

Research implications

In developing countries as Vietnam, radiation infrastructure is markedly deficient, with the number of linear accelerators limited by need. Moreover, colorectal cancer screening is not routine, leading to advanced-stage diagnoses.[2] Cancer patients in Vietnam often present with advanced tumoursand an overwhelmed healthcare infrastructure.^[28,29] The scarcity of healthcare facilities and human resources for radiation therapy results in treatment delays and increased waiting time, which heightens the risk of tumourprogression and non-compliant treatment, reducing opportunities for radical treatment.

Therefore, in Vietnam’s realistic situation, considering the aforementioned difficulties and advantages, first-line surgery might be a beneficial option for T3 rectal cancer patients with uninvolved mrCRM and a low risk of LR. However, accurate staging, precise pre-operative evaluation, and meticulous dissection are essential. Furthermore, these patients require pathological reassessment for TNM stage, resection margin, pCRM, lymph node involvement, and other factors to determine the need for adjuvant chemoradiotherapy.

Results of the study may contribute to establishing the criteria for patient selection as well as the indication of upfront radical surgery for T3, and uninvolved mrCRM rectal cancer without nCRT.

Research limitations

This study has several limitations, including its design as a single-centre, longitudinal cohort study, and the major drawback was the lack of a control group. Adjuvant treatment was not performed entirely and did not strictly comply with protocol due to subjective and objective reasons.

Research strengths

Each patient was pre-operatively discussed in an MDT consultation for treatment strategy and followed for at least 3 years after surgery.