Abstract
Background:
Circumferential resection margin is a key quality metric and predictor of oncologic outcomes and overall survival following surgery for rectal cancer. We aimed to develop a nomogram to identify patients at risk for a positive circumferential resection margin in the preoperative setting.
Material and Methods:
We performed a retrospective evaluation of the National Cancer Database from 2010–2014 for patients with clinical stage I-III rectal cancer who underwent total mesorectal excision. Patients were excluded for emergency operation, resection for cancer recurrence, palliative resection, transanal resection, and missing circumferential resection margin status. The primary outcome was positive circumferential resection margin. Secondary outcomes included overall survival.
Results:
There were 28,790 patients included. 2,245 (7.8%) had a positive circumferential resection margin. Higher tumor grade, lack of neoadjuvant chemotherapy, mucinous/signet tumor histology, open approach, abdominoperineal resection, higher T stage, lymphovascular invasion, and perineural invasion were all significantly associated with positive circumferential resection margin (p <0.05) and were included in the nomogram. The C-statistic was 0.703, suggesting a good predictive model.
Conclusions:
Positive circumferential resection margin is associated with specific patient demographics and tumor characteristics. These factors can be used along with pre-operative MRI to predict circumferential resection margin positivity in the preoperative period and plan accordingly.
Keywords: Rectal Cancer, Colorectal Surgery, Surgical Oncology
Introduction
With 151,030 reported cases in 2021 in the United States and over 52,000 deaths, colorectal cancer continues to be one of the most common and deadly malignancies.1,2 A negative circumferential resection margin (CRM) after surgical resection for rectal cancer has been shown to be crucial in decreasing local recurrence.3,4 More recently, with data from the Swedish Cancer Registry, the risk of local recurrence was found to be 17% with a positive CRM and 3.3% with a negative CRM.5 Additionally, there have been multiple studies showing positive CRM is associated with a significantly decreased rate of 5-year overall survival when compared to negative CRM.6
Circumferential resection margin (CRM) is a key quality metric and predictor of oncologic outcomes and overall survival following surgery for rectal cancer. Thus, the risk of positive CRM and programs aimed to reduce the rate of positive margins has been a focus of rectal cancer research in the last decade. For instance, the Dutch Surgical Colorectal Audit demonstrated a positive CRM rate of 8.5% in 2013, decreased from a prior rate of 14% after initiation of an audit review and installment of a quality improvement program.7 There have been multiple studies aimed at describing risk factors for positive CRM, but application of these factors to aide in operative decision-making is difficult in the pre-operative setting.8–10 Despite the importance of multidisciplinary tumor board and MRI imaging, these studies highlighted a need for a well-constructed tool for the clinician to use in their assessment of the CRM. This could be used to provide valuable information to clinicians when formulating plans and help tailor treatments (e.g. Total Neoadjuvant Treatment, long vs. short course radiation) to decrease the risk of a positive pathologic CRM.
The aim of our current study was to develop and validate a nomogram to identify patients at risk for a positive CRM.
Materials and Methods
Study Design & Data Source
This was an original retrospective cohort study with the goal of creating and validating a nomogram for positive circumferential margins in rectal cancer resection. The Participant User File of the National Cancer Database (NCDB) was used as the data source. The NCDB is a clinical oncology database acquired from cancer center registry information collected in more than 1,500 Commission on Cancer (CoC) accredited facilities. Data represents more than 70 percent of newly diagnosed cancer cases within the CoC system and more than 34 million historical records. Authorization to use the NCDB was obtained from both Vanderbilt University Medical Center and the CoC. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) guidelines were followed in performing this study.11 This project was reviewed and approved by Vanderbilt University Medical Center’s Institutional Review Board (Study Number 190240) with a waiver of informed consent.
Study population
The NCDB from 2010–2014 was queried for all adult patients (age > 18) with a diagnosis of primary rectal adenocarcinoma (International Classification of Diseases for Oncology, 3rd Edition SEER Topography Codes C20.9 with Histology Code 8140, 8210, 8211, 8261, 8262, 8480, or 8481) designated clinically as Stage I, II, or III. Exclusion criteria included: emergency operation, recurrent malignancy, surgery performed outside the reporting facility, metastatic disease (Stage IV), procedure for palliative intent, any surgery other than radical resection, and patients without data on circumferential resection margin (CRM). 28,790 patients were included in the study population after application of these criteria within the NCDB as shown in Figure 1.
Figure 1.
Query from NCDB with our basic inclusion/exclusion criteria applied. This excludes metastatic disease (stage IV), palliative surgery, emergency operation, recurrent malignancy, surgery outside the reporting facility, and missing data on CRM.
Outcome and Variables of Interest
The outcome of interest was positive circumferential resection margin (CRM) on surgical pathology. We defined a positive margin as evidence of adenocarcinoma ≤ 1 mm within the resection margin as per the current NCCN definition of positive CRM12. To create the nomogram, we included all clinicopathologic variables available in the NCDB that would be available to a provider prior to surgery and potentially related to circumferential resection margin. The variables included in the nomogram included basic demographic information and other factors selected a priori as clinically relevant based on existing literature8–10. Variables that would not be readily available to providers at a pre-operative clinical setting were excluded from the nomogram, including facility volume, local education rates, and median income for the patient’s home zip code.
The primary outcome was binary CRM status. The following risk factors were pre-specified based on clinical knowledge and literature review: age (years), gender (female, male), race (white, black, other), comorbidity (Charlson-Deyo Score 0, 1, 2+), tumor size (mm), tumor differentiation (Grade I, II, III, IV), histology (adenocarcinoma, mucinous, signet), neoadjuvant chemotherapy (yes, no), neoadjuvant chemoradiation (yes, no), facility volume, facility location (New England, Mid Atlantic, South Atlantic, East North Central, East South Central, West North Central, West South Central, Mountain, Pacific), facility type (community, comprehensive community, academic, integrated network), income (<$38,000; $38,000–47,999; $48,000–62,999; >$63,000), population density (metro, urban, rural), TNM clinical T stage (T1, T2, T3, T4), TNM clinical N stage (N0, N1, N2, Nx), lymph-vascular invasion on pre-operative biopsy (yes, no), perineural invasion on pre-operative biopsy (yes, no), carcinoembryonic antigen levels (CEA: elevated, normal, borderline), and anticipated surgical approach (laparoscopic/open, abdominoperineal resection (APR)/low anterior resection (LAR)).
In the data set, tumor sizes were coded from 0–989 mm. To exclude inaccurate tumor sizes from our analysis, we defined tumor size > 100 mm as missing size data.
Statistical Analysis
Patients and tumor characteristics were summarized using medians with quartiles (continuous) and frequencies with percentages (categorical) and compared between CRM groups (Positive CRM vs. Negative CRM) using Wilcoxon rank-sum (continuous) and chi-squared (categorical) tests.
Logistic regression was used to build a full-scale risk prediction model with all above candidate predictors. To develop a more parsimonious model for convenient use in a clinical setting, the full model has been reduced by removing risk factors that may not be readily available in a timely fashion, including facility volume, location, type, income, and population density. The likelihood ratio test was performed to compare nested models for parsimony. For all missing covariates data, multiple imputation was performed using chained equation (MICE) procedure with the aregImpute function in Hmisc R-package.
Internal validation was completed using bootstrapping (300-iteration) to assess the model discrimination and calibration. Discrimination was quantified using the bias-corrected area under the receiver operator characteristic (ROC) curve (AUC), also known as the C-statistic Calibration was assessed graphically with a calibration plot. All p values were two-sided with significance determined with p < 0.05. All statistical analyses were performed with R version 4.1.2.
Results
Participants and Primary Outcome
After applying inclusion and exclusion criteria, 28,790 patients with radically resected rectal cancer were available for analysis (Figure 1). Table 1 reports the baseline patient, disease, hospital and perioperative variable. In our study, patients had a median age of 61 years, 61% were men, and 87% were white.
Table 1:
Demographics
| Negative (N=26545) |
Positive (N=2245) |
Total (N=28790) |
p value | ||
|---|---|---|---|---|---|
| AGE | 0.02 | ||||
| Median (Q1,Q3) | 61.0 (52.0, 71.0) | 62.0 (53.0, 72.0) | 61.0 (52.0, 71.0) | ||
| Gender | 0.75 | ||||
| Male | 16227 (61%) | 1380 (61%) | 17607 (61%) | ||
| Female | 10318 (39%) | 865 (39%) | 11183 (39%) | ||
| Race | < 0.01 | ||||
| White | 23029 (87%) | 1901 (85%) | 24930 (87%) | ||
| Black | 1992 (8%) | 206 (9%) | 2198 (8%) | ||
| Other | 1356 (5%) | 124 (6%) | 1480 (5%) | ||
| Comorbidity | 0.57 | ||||
| Charlson-DeyoScore 0 | 20052 (76%) | 1718 (77%) | 21770 (76%) | ||
| Charlson-DeyoScore 1 | 5093 (19%) | 412 (18%) | 5505 (19%) | ||
| Charlson-DeyoScore 2+ | 1400 (5%) | 115 (5%) | 1515 (5%) | ||
| Tumor size (mm) | < 0.01 | ||||
| Median (Q1,Q3) | 36.0 (24.0, 50.0) | 40.0 (28.0, 60.0) | 37.0 (24.0, 50.0) | ||
| Tumor Differentiation | < 0.01 | ||||
| Grade I (well) | 2257 (10%) | 167 (8%) | 2424 (9%) | ||
| Grade II (moderately) | 18574 (79%) | 1418 (70%) | 19992 (78%) | ||
| Grade III (poorly) | 2382 (10%) | 374 (18%) | 2756 (11%) | ||
| Grade IV (undifferentiated) | 296 (1%) | 67 (3%) | 363 (1%) | ||
| Histology | < 0.01 | ||||
| Adenocarcinoma | 25231 (95%) | 1966 (88%) | 27197 (94%) | ||
| Mucinous | 1193 (4%) | 236 (11%) | 1429 (5%) | ||
| Signet | 121 (0%) | 43 (2%) | 164 (1%) | ||
| Neoadjuvant Chemotherapy | < 0.01 | ||||
| No | 9325 (35%) | 850 (38%) | 10175 (35%) | ||
| Yes | 17220 (65%) | 1395 (62%) | 18615 (65%) | ||
| Neoadjuvant Chemoradiation | 0.03 | ||||
| No | 9257 (35%) | 835 (37%) | 10092 (35%) | ||
| Yes | 17288 (65%) | 1410 (63%) | 18698 (65%) | ||
| Adjuvant Chemotherapy | < 0.01 | ||||
| No | 17844 (67%) | 1366 (61%) | 19210 (67%) | ||
| Yes | 8701 (33%) | 879 (39%) | 9580 (33%) | ||
| Adjuvant Radiation | < 0.01 | ||||
| No | 24488 (92%) | 1906 (85%) | 26394 (92%) | ||
| Yes | 2057 (8%) | 339 (15%) | 2396 (8%) | ||
| Type of Operation | < 0.01 | ||||
| LAR | 19725 (74%) | 1412 (63%) | 21137 (73%) | ||
| APR | 6820 (26%) | 833 (37%) | 7653 (27%) | ||
| Surgical Approach | < 0.01 | ||||
| Open | 14219 (54%) | 1374 (61%) | 15593 (54%) | ||
| Minimally Invasive | 12326 (46%) | 871 (39%) | 13197 (46%) | ||
| Facility Volume | < 0.01 | ||||
| Median (Q1,Q3) | 11.0 (5.0, 18.0) | 10.0 (5.0, 18.0) | 11.0 (5.0, 18.0) | ||
| Facility Location | < 0.01 | ||||
| 1 (New England) | 1376 (5%) | 114 (5%) | 1490 (5%) | ||
| 2 (Mid Atlantic) | 3530 (14%) | 272 (13%) | 3802 (14%) | ||
| 3 (South Atlantic) | 5261 (21%) | 459 (21%) | 5720 (21%) | ||
| 4 (East North Central) | 4943 (19%) | 406 (19%) | 5349 (19%) | ||
| 5 (East South Central) | 1495 (6%) | 141 (7%) | 1636 (6%) | ||
| 6 (West North Central) | 2506 (10%) | 149 (7%) | 2655 (10%) | ||
| 7 (West South Central) | 2094 (8%) | 181 (8%) | 2275 (8%) | ||
| 8 (Mountain) | 1334 (5%) | 178 (8%) | 1512 (5%) | ||
| 9 (Pacific) | 3004 (12%) | 254 (12%) | 3258 (12%) | ||
| Facility Type | 0.52 | ||||
| Community | 1934 (8%) | 165 (8%) | 2099 (8%) | ||
| Comprehensive community | 11269 (44%) | 974 (45%) | 12243 (44%) | ||
| Academic | 9612 (38%) | 776 (36%) | 10388 (38%) | ||
| Integrated network | 2728 (11%) | 239 (11%) | 2967 (11%) | ||
| % No High School Degree in zip code | 0.07 | ||||
| 1 (>21%) | 4341 (16%) | 400 (18%) | 4741 (17%) | ||
| 2 (13–20.9%) | 6791 (26%) | 600 (27%) | 7391 (26%) | ||
| 3 (7–12.9%) | 8931 (34%) | 704 (31%) | 9635 (34%) | ||
| 4 (<7%) | 6420 (24%) | 534 (24%) | 6954 (24%) | ||
| Median Income Quartile | 0.33 | ||||
| 1 (<$38,000) | 4289 (16%) | 393 (18%) | 4682 (16%) | ||
| 2 ($38,000–47,999) | 6342 (24%) | 535 (24%) | 6877 (24%) | ||
| 3 ($48,000–62,999) | 7205 (27%) | 605 (27%) | 7810 (27%) | ||
| 4 (>$63,000) | 8641 (33%) | 701 (31%) | 9342 (33%) | ||
| Population Density | 0.15 | ||||
| metro | 21083 (81%) | 1761 (81%) | 22844 (81%) | ||
| urban | 4211 (16%) | 374 (17%) | 4585 (16%) | ||
| rural | 606 (2%) | 39 (2%) | 645 (2%) | ||
| Distance from facility (miles) | 0.30 | ||||
| Median (Q1,Q3) | 11.6 (5.0, 29.3) | 11.6 (4.9, 28.4) | 11.6 (5.0, 29.2) | ||
| TNM Clinical T Stage | < 0.01 | ||||
| T1 | 3534 (13%) | 228 (10%) | 3762 (13%) | ||
| T2 | 4689 (18%) | 258 (12%) | 4947 (17%) | ||
| T3 | 16880 (64%) | 1401 (63%) | 18281 (64%) | ||
| T4 | 1350 (5%) | 349 (16%) | 1699 (6%) | ||
| TNM Clinical N Stage | < 0.01 | ||||
| N0 | 15976 (60%) | 1227 (55%) | 17203 (60%) | ||
| N1 | 8523 (32%) | 749 (33%) | 9272 (32%) | ||
| N2 | 1689 (6%) | 233 (10%) | 1922 (7%) | ||
| Nx | 263 (1%) | 27 (1%) | 290 (1%) | ||
| Lymph-Vascular Invasion | < 0.01 | ||||
| No | 18584 (84%) | 1300 (68%) | 19884 (82%) | ||
| Yes | 3619 (16%) | 615 (32%) | 4234 (18%) | ||
| Perineural Invasion | < 0.01 | ||||
| No | 21912 (91%) | 1505 (73%) | 23417 (90%) | ||
| Yes | 2183 (9%) | 562 (27%) | 2745 (10%) | ||
| CEA (carcinoembryonic antigen) | < 0.01 | ||||
| Elevated | 6634 (38%) | 769 (52%) | 7403 (39%) | ||
| Normal | 10779 (62%) | 710 (48%) | 11489 (60%) | ||
| Borderline | 98 (1%) | 6 (0%) | 104 (1%) | ||
| Follow-up Time (months) * | 33.6 (0, 71.4) | 28.2 (0, 69.5) | 33.1 (0, 71.4) | <0.01 | |
| Vital Status | |||||
| Alive | 17648 (85%) | 1197 (68%) | 18845 (83%) | <0.01 | |
| Dead | 3184 (15%) | 568 (32%) | 3752 (17%) |
Presented are median with range for patient’s follow-up months, which were defined as from the date of surgery to the date of death or date of last contact for those who were alive.
Overall, the rate of CRM positivity was 7.8%. 2,245 patients had positive CRM and 26,545 patients had negative CRM. The median overall survival for the CRM positive group is 58.6 months; the median survival is not reached for the CRM negative group. The difference in OS between the two groups is significant (p<0.0001). The adjusted hazard ratio for CRM positive subjects = 1.82 (95% CI: 1.66, 2.01). Results of the multivariate analysis of the association of selected variables and CRM positivity, determined after the bootstrapping method described above, are shown in Table 2.
Table 2:
Multivariate Analysis of Predictors for CRM Positivity
| Variable | OR | 95% CI | P |
|---|---|---|---|
| AGE(71 vs. 52) | 1.04 | 0.98 – 1.12 | 0.21 |
| Tumor.size (50 vs. 24) | 1.06 | 0.99 – 1.14 | 0.091 |
| Gender - Female:Male | 0.97 | 0.89 – 1.07 | 0.558 |
| Race - Black:White | 1.1 | 0.94 – 1.29 | 0.231 |
| Race - Other:White | 1.06 | 0.87 – 1.29 | 0.586 |
| Comorbidity - Charlson-DeyoScore 1:0 | 0.93 | 0.83 – 1.04 | 0.211 |
| Comorbidity - Charlson-DeyoScore 2:0 | 0.93 | 0.76 – 1.14 | 0.474 |
| Tumor Differentiation - I:II | 1 | 0.84 – 1.18 | 0.969 |
| Tumor Differentiation - III:II | 1.54 | 1.35 – 1.75 | <0.001 |
| Tumor Differentiation - IV:II | 1.99 | 1.49 – 2.66 | <0.001 |
| HISTOLOGY - Mucinous:Adenocarcinoma | 2.07 | 1.77 – 2.42 | <0.001 |
| HISTOLOGY - Signet :Adenocarcinoma | 1.76 | 1.19 – 2.61 | 0.005 |
| Neoadjuvant Chemo - No:Yes | 1.2 | 1 – 1.44 | 0.047 |
| Neoadjuvant ChemoXRT - No:Yes | 1.16 | 0.97 – 1.39 | 0.108 |
| Type of Operation - APR:LAR | 1.35 | 1.22 – 1.49 | <0.001 |
| Surgical Approach - MinimallyInvasive:Open | 0.84 | 0.77 – 0.92 | <0.001 |
| TNM T Stage - T1:T3 | 0.83 | 0.7 – 0.98 | 0.029 |
| TNM T Stage - T2:T3 | 0.7 | 0.6 – 0.81 | <0.001 |
| TNM T Stage - T4:T3 | 2.37 | 2.06 – 2.73 | <0.001 |
| TNM T Stage - T.Others:T3 | 1 | 0.9 – 1.11 | 0.971 |
| TNM N Stage - N1:N0 | 1.11 | 0.94 – 1.32 | 0.198 |
| TNM N Stage - N2:N0 | 1.15 | 0.76 – 1.74 | 0.511 |
| TNM N Stage - Nx:N0 | 1.49 | 1.31 – 1.7 | <0.001 |
| Lymphovascular invasion - Yes:No | 2.54 | 2.25 – 2.88 | <0.001 |
| Perineural invasion - Yes:No | 1.42 | 1.26 – 1.6 | <0.001 |
| CEA - Elevated:Normal | 0.89 | 0.42 – 1.88 | 0.755 |
| CEA - Borderline:Normal | 1.04 | 0.98 – 1.12 | 0.21 |
Model Development, Specification, and Performance
A nomogram predicting the risk of CRM positivity was constructed based on these variables (Figure 2). Model performance was internally validated for discrimination and calibration. Discrimination, as measured by the bootstrap, optimism-corrected C-statistic was 0.703. The calibration curve of the model is shown in Figure 3. Hosmer-Lemeshow tests showed no lack of fit.
Figure 2.
Nomogram for the prediction model of positive circumferential resection margins.
Figure 3.
Calibration curve for predictive modeling of circumferential resection margin status.
To apply the nomogram, (Figure 2) a vertical line can be drawn from each predictor to the associated value on the “Points” section. Then adding together a collective score finds the “Total Points” for an individual patient and the corresponding result in the “Predicted Risk of Positive CRM” section. The final prediction model is available both with the convenient nomogram and as a user-friendly web-based tool at https://statez.shinyapps.io/CRMPred/.
Discussion
A positive circumferential resection margin in rectal cancer is associated with increased local recurrence and decreased overall survival. While multiple publications have identified risk factors and imaging findings associated with positive CRM, there is not currently a user-friendly clinical tool aimed to assist with prediction of positive CRM in the pre-operative setting.8–10,13–15 We developed a nomogram to predict positive CRM utilizing baseline patient and facility demographics, tumor characteristics, and planned operative approach/details. This tool will be helpful in assisting providers and patients to make informed treatment decisions in the pre-operative setting prior to planned rectal cancer resection.
Our analysis of data from the National Cancer Database showed that positive CRM was associated with higher tumor grade, mucinous or signet tumor histology, lack of neoadjuvant chemotherapy, higher T stage, perineural invasion, lymphovascular invasion, open operations when compared to minimally-invasive approaches, and those who underwent APR (abdominoperineal resection) when compared to LAR (low anterior resection). Surprisingly, there was not a significant association between positive CRM and increased patient age or higher Charlson-Deyo Score indicating more severe medical comorbidities. The increased risk of positive CRM with an open operative approach, when compared to laparoscopic approach, may reflect selection bias of more difficult cases. Similarly, the association of increased risk for positive CRM with APR, when compared to LAR, also may be related to APR being a more appropriate strategy for surgical resection in low rectal cancers or tumors that involve the anal sphincters.
Our study builds on the 2013 study published by Russell, et al. detailing development of a predictive nomogram for positive circumferential resection margins in rectal cancer.16 Our study has several additional, relevant features. We evaluated patients from a more recent data range of 2010–2014, which likely differs in treatment options and pathways from the prior study’s population which was collected from 1998–2007, and reflects the modern approach to neoadjuvant treatment of rectal cancer following the German Rectal Cancer Trial published in 2004.17 As surgical techniques have further developed during this time, including minimally invasive approaches, updated data collection from more recent surgical interventions will be more representative of current practices. For example, in our study, nearly half of cases (46%) were performed with a minimally invasive approach. The Russell et al. study also focused on patient demographics and baseline tumor characteristics in their nomogram and reported specific facility outliers for high or low rates of margin positivity, whereas our study evaluated additional relevant factors, such as prior treatments (neoadjuvant therapy), facility type, operative approach, CEA levels, and tumor characteristics including lymphovascular invasion, perineural invasion, and analysis of individual clinical TNM stages. Another important difference, was the Russell et al. study was designed with the goal of calculating margin positivity per hospital. The goal of the current study, is to make a predictive tool for the clinician to help individualize care for each patient.
The nomogram in our study incorporates a wide variety of patient, tumor, treatment, operative, and facility factors. This is beneficial to clinicians who are making active treatment decisions and working to identify distinct factors associated with an increased risk of positive CRM. This nomogram provides a defined likelihood and confidence interval for each individual patient. Once the tool has been applied to the patient, if they are deemed to be high risk for a positive CRM (higher than the published average near 8%), this may guide their provider to pursue different or additional treatment options. Some possible alterations include total neoadjuvant therapy over standard neoadjuvant regimens, more aggressive surgical resections (such as consideration of exenteration or extralevator vs. standard APR), intraoperative radiation therapy, or even prolonged time between neoadjuvant radiotherapy and surgical resection to allow further treatment response. Many of these treatment options have been shown to be beneficial for certain high-risk patients with rectal cancer, but have adverse effects that limit their current use.15,17–21
Magnetic resonance imaging (MRI) has become one of the most useful tools in the pre-operative staging of rectal cancer.22 Within the dataset used for our study, documentation of a threatened circumferential margin by imaging is not available; however, the T and N stages are clinical values included in our analysis and in the United States the standard for obtaining this information is MRI or endorectal ultrasound for staging. While we did not have access to the direct MRI data, the inclusion of clinical T stage provides other useful information regarding the tumor involvement of these patients. Careful review of the MRI with a radiologist will be an important adjunct to the nomogram.
There are limitations of this study inherent to the use of a large database. There is limited ability to obtain missing information from the NCDB. Patients with missing or unreported pathologic margins were excluded and this may confound our analysis. The database is limited in that it does not contain information regarding surgeon training or experience, distance of the tumor from the anal verge, or other factors that may influence operative decision-making. Patients who undergo open operations or an APR may be more likely to have more aggressive tumors or have other underlying factors that put them at higher risk for positive CRM. This confounding could be investigated further by studying the nomogram in a population with more information regarding tumor location and surgeon preferences/background. Additionally, we utilized the NCDB to internally validate our nomogram. As the NCDB is a large and robust database that encompasses patients from across the United States, internal validation is an accepted validation method. External validation with a smaller data set would add little to the analysis and the development of our nomogram. Lastly, subjects included in our study were diagnosed with rectal cancer from 2010–2014. Due to developments in the field of rectal cancer care and other trends within operative practices for rectal cancer surgery, this may limit the generalizability of the nomogram.
This study does not address implementation of the nomogram into clinical use. Future study should include applying this nomogram in an updated patient sample outside of the NCDB, utilizing implementation science frameworks to optimize the process. These future studies will elucidate how the use of the nomogram may change clinical decision-making and what further information is needed about the patient or tumor to help clinicians apply the nomogram, such as other predictive tools for overall survival rate or risk of local recurrence. Another point of interest is the interpretation of CRM risk by clinicians in practice. What increase in percentage risk of CRM (from the reported baseline risk of ~8%) would trigger providers to change their practice?10
Future directions from this study include study of the nomogram used in conjunction with MRI and integration into clinical practice. For practice integration, the online tool is user friendly and can be accessed through a computer or a smartphone. The data points should be available to the clinician prior to the patient visit and the calculated risk can then be used as a decision aid when discussing additional treatment versus proceeding with surgery. Risk calculators have been shown to be important for individual treatment and shared decision making and this could be the topic of further study to evaluate the role this nomogram tool plays in altering decision making pathways.23
Conclusions
Positive CRM is associated with specific patient demographics and tumor characteristics. Given the known significant increase in local recurrence and decrease in overall survival, interventions aimed at decreasing rates of positive CRM in rectal cancer are critical. There is currently a lack of clinical tools available to use along with MRI to identify patients at high risk for positive CRM. Our nomogram addresses this knowledge gap and can be used in the preoperative period to assist providers in determining appropriate treatment plans.
Supplementary Material
Supplemental Figure 1: Kaplan Meier Curve
Supplemental Figure 2: Final Model
Acknowledgements, Funding Support, Disclosures
Dr. Hawkins’ work on this manuscript was supported by the National Institute of Diabetes and Digestive and Kidney Disease of the National Institutes of Health [award number K23DK118192]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The project described was supported by the National Center for Research Resources [Grant UL1 RR024975-01] and is now at the National Center for Advancing Translational Sciences [Grant 2 UL1 TR000445-06]. Additional support was provided by CTSA [award No. UL1 TR002243] from the National Center for Advancing Translational Sciences.
Dr. Shroder’s work on this manuscript was supported by the office of Academic Affiliations, Department of Veterans Affairs. VA national Quality Scholars Program and with use of facilities at the VA Tennessee Valley Healthcare system, Nashville Tennessee.
The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
This work has not previously been published. Our findings were presented as an ePoster Presentation for the 2022 ASCRS Annual Scientific meeting in Tampa, Florida (April 30th – May 4th, 2022) and as a podium presentation for the TN ACS meeting in Nashville, Tennessee (August 11th – 14th, 2022).
Declarations of Interest/Commercial Interests: None
References
- 1.Zhang GQ, Sahyoun R, Stem M, et al. Operative Approach Does Not Impact Radial Margin Positivity in Distal Rectal Cancer. World J Surg. 2021. [DOI] [PubMed] [Google Scholar]
- 2.Lorenzo Linan MA, Garcia Armengol J, Martin Martin GP, Martinez Sanjuan V, Roig Vila JV. Validation of pelvic magnetic resonance imaging as the method of choice to determine the distance to the anal margin in rectal cancer. Cir Esp. 2021. [DOI] [PubMed] [Google Scholar]
- 3.Quirke PDP, Dixon MF, Williams NS, et al. Local recurrence of rectal adenocarcinoma due to inadequate surgical resection. Histopathological study of lateral tumour spread and surgical excision. Lancet 1986;2(8514):996–999. [DOI] [PubMed] [Google Scholar]
- 4.Quirke PDP, Dixon MF, Williams NS, et al. Local recurrence of rectal adenocarcinoma due to inadequate surgical resection. Histopathological study of lateral tumour spread and surgical excision. Lancet. 1986;2(8514):996–999. [DOI] [PubMed] [Google Scholar]
- 5.Agger EA JF, Lydrup MA, et al. Risk of local recurrence of rectal cancer and circumferential resection margin: population-based cohort study. Br J Surg. 2020;107(5):580–585. [DOI] [PubMed] [Google Scholar]
- 6.Patel SH HC, Massarweh NN, et al. Circumferential Resection Margin as a Hospital Quality Assessment Tool for Rectal Cancer Surgery. J Am Coll Surg. 2020;230(6):1008–1018 e1005. [DOI] [PubMed] [Google Scholar]
- 7.Van Leersum NJ SH, Henneman D, et al. The Dutch Surgical Colorectal Audit. European Journal of Surgical Oncology (EJSO). 2013;39(10):1063–1070. [DOI] [PubMed] [Google Scholar]
- 8.Simon HL dPT, Profeta da Luz MM, Kiran RP, Keller DS. Predictors of Positive Circumferential Resection Margin in Rectal Cancer: A Current Audit of the National Cancer Database. Dis Colon Rectum. 2021;64(9):1096–1105. [DOI] [PubMed] [Google Scholar]
- 9.Pasch JA ME, Pasch LB, Premaratne C, Fok KY, Kotecha K, El Khoury T, Barto W. Clinicopathological factors associated with positive circumferential margins in rectal cancers. ANZ J Surg. 2019;89(12):1636–1641. [DOI] [PubMed] [Google Scholar]
- 10.Rullier AG-BS, Jarlier M. Predictive Factors of positive circumferential resection margin after radiochemotherapy for rectal cancer: The French randomised trial ACCORD12/0405 PRODIGE 2. European Journal of Cancer. 2013;40(1):82–89. [DOI] [PubMed] [Google Scholar]
- 11.Collins GS RJ, Altman DG, Moons KG. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement. Ann Intern Med. 2015;162(1):55–63. [DOI] [PubMed] [Google Scholar]
- 12.National Comprehensive Cancer Network. Rectal Cancer (Version 2.2022) https://www.nccn.org/professionals/physician_gls/pdf/rectal.pdf. Accessed October 18, 2022.
- 13.Rickles AS DD, Chang GJ. High Rate of Positive Circumferential Resection Margins Following Rectal Cancer Surgery: A Call to Action. Ann Surg. 2015;262(6):891–898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Petroudi SBG, Bond S, Brady M. Circumferential resection margin assessment on MRI of rectal cancer. Conf Proc IEEE Eng Med Biol Soc. 2006;2006:4881–4884. [DOI] [PubMed] [Google Scholar]
- 15.Patel SSV, Rohila J, Bankar S, Desouza A, Saklani A. Patterns of failure and outcomes of rectal cancer patients who had a positive pathological circumferential resection margin (pCRM). European Journal of Surgical Oncology (EJSO). 2021;47(2):e24. [Google Scholar]
- 16.Russell MC YY, Hu CY, et al. A novel risk-adjusted nomogram for rectal cancer surgery outcomes. JAMA Surg. 2013;148(8):769–777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sauer RLR, Merkel S, et al. Preoperative Versus Postoperative Chemoradiotherapy for Locally Advanced Rectal Cancer: Results of the German CAO/ARO/AIO-94 Randomized Phase III Trial After a Median Follow-Up of 11 Years. Journal of Clinical Oncology. 2012;30(16):1926–1933. [DOI] [PubMed] [Google Scholar]
- 18.Tao YHJ, Wang Z. Extralevator abdominoperineal excision for advanced low rectal cancer: Where to go. World Journal of Gastroenterology. 2020;26(22):3012–3023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Fahy MKM, Power Foley M, Nugent T, Shields C, Winter D. The role of intraoperative radiotherapy in advanced rectal cancer: a meta-analysis. Colorectal Disease. 2021;23(8):1998–2006. [DOI] [PubMed] [Google Scholar]
- 20.Liu SJT, Xiao L, et al. Total Neoadjuvant Therapy (TNT) versus Standard Neoadjuvant Chemoradiotherapy for Locally Advanced Rectal Cancer: A Systematic Review and Meta-Analysis. The Oncologist. 2021;26(9):e1555–e1566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Foster JJE, Falk S, Cooper E, Francis N. Timing of Surgery After Long-Course Neoadjuvant Chemoradiotherapy for Rectal Cancer. Dis Colon Rectum. 2013;56(7):921–930. [DOI] [PubMed] [Google Scholar]
- 22.Patel UB TF, Blomqvist L, et al. Magnetic resonance imaging-detected tumor response for locally advanced rectal cancer predicts survival outcomes: MERCURY experience. J Clin Oncol. 2011;29(28):3753–3760. [DOI] [PubMed] [Google Scholar]
- 23.Mansmann URA, Strahwald B, Crispin A. Risk calculators-methods, development, implementation, and validation. Int J Colorectal Dis. 2016;31(6):1111–1116. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Figure 1: Kaplan Meier Curve
Supplemental Figure 2: Final Model