Circumferential resection margin positivity due to direct or indirect tumour involvement in rectal cancer – a call for better stratification

A positive circumferential resection margin (CRM) after rectal cancer surgery, which can be the result of direct or indirect tumour involvement, has consistently been associated with increased local recurrence and poorer survival. However, little is known of the differential impact of the mode of tumour involvement on outcomes.


Introduction
The last three decades have seen significant improvements in outcomes following rectal cancer surgery, with significantly lower local and distant recurrence rates. 1 This is despite the widespread adoption of sphincter-preserving surgery.Although these gains can be attributed to better preoperative staging and escalation in the use of neoadjuvant radiotherapy, the quality of surgical resection remains the major determinant of oncological outcomes in rectal cancer. 2In particular, meticulous dissection in embryological planes (total mesorectal excision) and ensuring a microscopically clear circumferential resection margin (CRM) have been widely described as key factors affecting local recurrence and survival. 3,4 microscopically positive CRM (R1 resection) has repeatedly been implicated as a major risk factor for local recurrence. 5In their seminal study, Quirke et al. demonstrated a local recurrence rate of 80% for patients undergoing curative resection if the CRM was positive, compared with no recurrence in the negative CRM group (R0) at a median follow-up of 23 months. 2Similar findings have been independently reported for patients undergoing total mesorectal excision (TME) (16.4% for R1 vs. 5.8% for R0) and abdominoperineal resection (APR) or pelvic exenteration. 4,6urthermore, R1 resection has been associated with increased rates of distant metastasis (37.6% vs. 12.7% for R0) and reduced survival. 7Therefore, R1 resection has become a widely used prognostic marker after rectal cancer surgery.
Classically, a positive CRM is caused by direct tumour extension at or near the resection margin.More rarely, however, it may be caused by a positive lymph node or the presence of tumour cells within or around lymphovascular or neural structures at the surgical margin. 8,9Studies investigating the role of R1 resection have so far assumed that these two distinct causes of a positive CRM are sufficiently similar that they have been considered as a single entity in the vast majority of cases.The very few reports that have drawn a distinction have had conflicting conclusions, probably because of the small number of cases considered. 4,10,11Importantly, TNM-8 does not distinguish between these two possible causes of R1 resection, a fact reflected in histopathological reports. 12Therefore, our aim was to determine whether patients with positive CRM due to direct tumour extension (R1-tumour) had a different outcome from those with lymph nodes or alternative causes of R1 resection (R1-other).

Methods
Patients who underwent rectal resection at a single high-volume tertiary hospital between 2003 and 2018 were identified from a colorectal cancer database.This colorectal cancer database has previously been described and combines prospectively collected operative, pathological, and clinical data for patients undergoing colorectal cancer resection at Norfolk & Norwich University Hospitals NHS Foundation Trust, United Kingdom, from 2003 onwards. 13The database integrates data on patient demographics, tumour characteristics, neoadjuvant and adjuvant therapies, and oncological outcomes.Survival is regularly updated through linkage with the data held on the NHS Spine.Ethical approval was granted by the UK Health Research Authority (IRAS/REC reference: 18/NI/0138).

Patient selection
Patients who underwent resection of a primary rectal cancer between February 2003 and November 2018 were included if they had biopsy-proven invasive adenocarcinoma with the distal edge of the tumour within 15 cm of the anal verge.
Exclusion criteria were patients undergoing only endoscopic or transanal resections, multi-visceral resections or exenterative procedures, panproctocolectomy, or other colonic resections, including synchronous colon cancers, recurrent cancers, and patients with pathologies other than primary adenocarcinoma.Patients undergoing palliative resection or R2 resection (macroscopic tumour involvement of the resection margin) were excluded from the main analysis, as were those with sigmoid colon cancer.Patients with positive distal resection margins were excluded.

Treatment pathway and follow-up
All patients were discussed preoperatively at the local multidisciplinary tumour board for treatment pathway selection.In brief, patients in whom the surgical margin was predicted on imaging to be threatened or involved (either by nodal disease or direct tumour invasion) received neoadjuvant chemoradiotherapy (long-course chemoradiotherapy, LCCRT).Where there was evidence of systemic disease, upfront systemic chemotherapy was offered.Patients with MRI-predicted node-positive disease but with clear surgical margins would proceed straight to surgery without neoadjuvant therapy.Finally, within the unit, short-course radiotherapy (SCRT, 5 Â 5 Gy) was reserved for patients planned for abdominoperineal resections (APER) where there were potentially close margins at or near the sphincter complex.Neoadjuvant chemotherapy refers to systemic, full-dose chemotherapy, whereas neoadjuvant chemoradiotherapy consists of a sensitizing dose of chemotherapy with radiation.The use of neoadjuvant therapy was in line with the contemporaneous National Institute for Health and Care Excellence (NICE) guidance. 14ostoperative follow-up also conformed to NICE guidance and included serial monitoring of serum carcinogenic antigen (CEA) and at least three CT scans of the chest, abdomen, and pelvis for the first five post-operative years.

Definitions
An R1 resection was defined as a macroscopically clear resection, but with the microscopic presence of tumour cells within 1 mm of (but not at) the closest non-peritonealised margin of the surgical specimen at pathology. 9For the purposes of this study, the direct presence of tumour extension is termed R1-tumour.Conversely, the presence of tumour cells within a lymph node (encapsulated or otherwise) or extramural vascular invasion within 1 mm of a nonperitonealised margin was termed R1-other.
R2 resection was defined as where macroscopically visible primary tumour was left in the patient, as described by the operating surgeon.These patients were excluded from further analysis.The presence of residual distant metastatic disease did not influence the R status.
Definitions of local and distant metastases are as described by Patel et al., where local recurrence (LR) was defined as a tumour in the pelvis, at the anastomosis, or at the operative site following resection. 11Distant metastasis (DM) was defined as tumour spread to organs outside the pelvis.

Outcome measures
The main endpoints were overall survival and disease-free survival (DFS) (time between the operation date and the date of detection of local or distant metastasis, date of death, or last follow-up date, whichever occurred first), and patterns and timing of recurrence.Patients with metastatic disease at presentation were excluded from the DFS analysis but were included in the overall survival analysis only if the metastases were potentially resectable.Overall survival was defined as the time between the date of the operation and the date of death (or date of last follow-up for censored patients).Survival and recurrence data were collected in August 2023.

Statistical analysis
Student's t-test was used for group differences in continuous variables, whereas the chi-square test or Fisher's exact test was used for categorical variables.Survival curves were generated using Kaplan-Meier survival analysis, and a log-rank test was performed to detect differences between curves.Multivariate Cox regression analysis was performed to identify independent predictors of survival.To prevent overfitting of the regression model, only the factors that reached statistical significance in the univariate regression were included in the multivariate analysis.Statistical significance was set at P < 0.05.All analyses were performed using R Foundation Statistical software (R 4.2.3).

Results
A total of 1460 patients underwent rectal cancer resection during the study period.Of these, 98 (6.7%) had a R1 resection.The median follow-up duration in the R1 cohort was 140 months.Eighty patients had an R1 resection due to positivity of the CRM only, 13 due to distal resection margin (DRM), and 5 due to both CRM and DRM positivity.Given that the existing body of literature indicates a strong role for CRM positivity in dictating local recurrence and survival, patients with DRM-only positive margins were excluded from further analysis.Of the remaining patients, 69 had direct tumour extension (hereafter called R1-tumour).The rest consisted of eight cases with tumour cells within a lymph node, six with extramural vascular invasion, and two with perineural invasion within 1 mm of the non-peritonealised margin (termed R1-other).Except for age and extramural invasion, the two groups were comparable (Table 1).Importantly, there were no statistically significant differences in the rates of neoadjuvant and adjuvant therapies.

Overall survival and its determinants
Strikingly, R1-other was associated with better overall survival (OS) compared to R1-tumour (hazard ratio 0.40, log-rank P = 0.006) (Fig. 1a).The 5-year OS rates for R1-other and R1-tumour were 75% (95% confidence interval: 57%-100%) and 35% (26%-49%), respectively.More importantly, there was no statistical difference in the overall survival between the R1-other and R0 cohort.Univariate regression analysis of clinicopathological factors that are believed to have an impact on overall survival was conducted.This revealed that increasing ASA grade, N stage, M stage, number of positive nodes assessed, and type of R1 resection (that is R1-tumour vs. R1-other) were associated with poorer survival (Fig. 1b).On multivariate regression analysis, only ASA grade and R1 type were found to be independent predictors of OS (Fig. 1c).

DFS and its determinants
To assess any potential differences in disease recurrence, patients with metastatic disease at presentation were excluded, leaving 14 and 51 patients in the R1-other and R1-tumour groups, respectively.Overall, the rate of recurrence was 38.5% across the R1 group.Of the 11 patients with a positive DRM, 2 incurred a recurrence during the follow-up period.There was a trend for reduced cancer recurrence and, therefore, improved DFS in the R1-other   group compared to that in the R1-tumour group (log-rank P = 0.043) (Fig. 2a).Univariate analysis revealed that the number of positive lymph nodes was the only independent predictor of DFS (HR 1.1, P = 0.0039) (Fig. 2b).

Patterns of local and distant recurrences
There were significantly more overall recurrences in the R1-tumour group compared to the R1-other group (44.0% vs. 14.3%,P = 0.04) (Fig. 3a).More importantly, 27.2% of patients in the R1-tumour group developed local recurrence compared to no local recurrence in the R1-other group (Fig. 3b).The liver and lungs were the two most common sites of distant recurrence.

Discussion
Few studies have investigated the outcomes associated with the various subtypes of R1 resection.Our findings are consistent with those of Patel et al., who showed that in a smaller cohort of R1 resections, CRM tumour involvement was linked to higher local recurrence and reduced DFS. 11In contrast, a retrospective study of 165 R1 rectal cancer resections reported no difference in DFS between the specific modes of CRM involvement studied, although local recurrence rates were higher in cases of direct tumour involvement. 10This apparent discrepancy could be attributed to better preoperative staging in the modern era, which guides the use of neoadjuvant therapies.Indeed, our study suggests that finer stratification of R1 subtypes may better guide adjuvant treatment.In particular, reserving the use of adjuvant radiotherapy for cases of direct tumour CRM involvement might reduce overtreatment and prevent complications such as low anterior resection syndrome.However, given that R1 resections are relatively rare, prospective studies that aim to investigate this question by stratifying patients into different treatment modalities will be very difficult to establish, and this decision might therefore need to be based on pragmatism.This clinical question is further complicated by evidence suggesting that extracapsular lymph node involvement is associated with worse outcomes than a positive but encapsulated node. 15Hence, future tumour staging could consider a distinction between encapsulated nodes with R1 versus extracapsular nodes with R1, although randomized trials looking to confirm the prognostic impact of this distinction will again be difficult to power.
Our study has a few limitations that require mention.First, the retrospective single-centre nature of the data limits the generalizability of the conclusions.Second, no data was available for an assessment of the quality of the surgical resection, in particular of the plane of mesorectal excision, which would have been a useful factor to include in the multivariate analysis.All the resections at this institution were performed by a small group of colorectal specialist surgeons who have been trained in the principles of total mesorectal excision, thereby reducing some of the variability associated with surgical specimen quality.Furthermore, the impact of an involved distal resection margin was not investigated in the present study.The rationale for this was to try to isolate the effect of a positive CRM which has been more associated with local recurrence historically.Finally, while we were able to show that an R1-other resection leads to similar overall survival when compared to R0 resection, we did not have the required data to be able to compare recurrence rates between these two groups.

Conclusions
In summary, our data suggest striking differences in terms of both overall and DFS between patients with a positive CRM due to direct or indirect tumour extension.Additionally, the patterns of recurrence are different, with R1 resection due to tumour leading to higher rates of local recurrence.Our data highlight the potential prognostic value of a better discrimination of the mode of CRM involvement.More specifically, our study makes a case for the routine histopathological reporting of this distinction.

Fig. 1 .
Fig. 1.Overall survival differences and determinants.(a) Kaplan-Meier survival curves comparing R1_other to R1_tumour.Overlaid is the survival curve for the R0 group.(b) Summary of univariate Cox regression.Only ASA grade, N stage, M stage, R1 subtype, and the number of positive lymph nodes were statistically significant.(c) Forest plot showing results of Cox multivariate regression, taking only factors significant in univariate analysis.Significant P-values are indicated with asterisks.

Table 1
Baseline characteristics of the two groups