National cohort study on postoperative risks after surgery for submucosal invasive colorectal cancer

Background The decision to perform surgery for patients with T1 colorectal cancer hinges on the estimated risk of lymph node metastasis, residual tumour and risks of surgery. The aim of this observational study was to compare surgical outcomes for T1 colorectal cancer with those for more advanced colorectal cancer. Methods This was a population‐based cohort study of patients treated surgically for pT1–3 colorectal cancer between 2009 and 2016, using data from the Dutch ColoRectal Audit. Postoperative complications (overall, surgical, severe complications and mortality) were compared using multivariable logistic regression. A risk stratification table was developed based on factors independently associated with severe complications (reintervention and/or mortality) after elective surgery. Results Of 39 813 patients, 5170 had pT1 colorectal cancer. No statistically significant differences were observed between patients with pT1 and pT2–3 disease in the rate of severe complications (8·3 versus 9·5 per cent respectively; odds ratio (OR) 0·89, 95 per cent c.i. 0·80 to 1·01, P = 0·061), surgical complications (12·6 versus 13·5 per cent; OR 0·93, 0·84 to 1·02, P = 0·119) or mortality (1·7 versus 2·5 per cent; OR 0·94, 0·74 to 1·19, P = 0·604). Male sex, higher ASA grade, previous abdominal surgery, open approach and type of procedure were associated with a higher severe complication rate in patients with pT1 colorectal cancer. Conclusion Elective bowel resection was associated with similar morbidity and mortality rates in patients with pT1 and those with pT2–3 colorectal carcinoma.


Introduction
The introduction of population-based colorectal carcinoma screening programmes aims to reduce mortality from colorectal cancer. Screening-detected colorectal cancers have a more favourable stage distribution than those that are symptom-detected, but it remains unclear whether early diagnosis following screening results in better surgical outcomes 1 . In January 2014, a nationwide colorectal cancer screening programme was launched in the Netherlands. Individuals aged 55-75 years are offered a biennial faecal immunochemical test (FIT), and diagnostic colonoscopy when the FIT is positive 2 . the magnitude of these risks is sparse. Studies evaluating surgical morbidity and mortality of colorectal surgery consist mainly of patients with more advanced tumours 9 -11 . These risks cannot simply be extrapolated to patients with pT1 colorectal cancer as the clinical characteristics of patients with advanced colorectal carcinoma might be different 12 , few treatment alternatives are available, and the risk of cancer-related death is higher.
The aim of this study was to compare short-term postoperative outcomes after elective bowel resection in patients with pT1 and those with pT2-3 colorectal cancer, and to identify the key clinical features associated with severe complications after surgery for pT1 colorectal cancer from which a risk stratification table could be developed to help clinicians guide treatment decisions in patients with pT1 colorectal cancer.

Methods
This was a population-based cohort study of patients who underwent colorectal surgery for pT1-3 stage colorectal cancer between January 2009 and December 2016 in the Netherlands. The total population in the Netherlands was estimated as 16⋅6 million people in 2010, according to Statistics Netherlands. Patients were identified from the Dutch ColoRectal Audit (DCRA), formerly known as the Dutch Surgical Colorectal Audit. The DCRA is a web-based national audit, in which information on all patients undergoing surgery for a primary tumour is recorded prospectively 13 . The database has complete national coverage as the Dutch Health Inspectorate obliges inclusion of all surgically treated patients with colorectal cancer.
Patients who had an elective oncological resection were included in the study. Those who underwent neoadjuvant treatment, urgent or emergency surgery, or only a local procedure were excluded, as were patients with metastatic disease or synchronous colorectal cancer. Patients treated with a local surgical procedure before bowel resection were not excluded. As all data in the DCRA are coded, no ethical approval or informed consent was required for this study under Dutch law 14 .

Outcomes
Main outcome measures were overall, surgical and severe complications, and mortality. Definitions are shown in Table 1. The reason for selecting the combined outcome of severe complications (reintervention and/or mortality) in this study was because mortality alone was considered an underestimation of the total burden to the patient 15 . If no complication was registered, the authors assumed no

Statistical analysis
Baseline characteristics were compared between patients with pT1, pT2 and pT3 colorectal cancer using the χ 2 test for categorical variables and the Kruskal-Wallis test for continuous variables. Missing data were assumed to be missing at random. For all logistic regression analyses, multiple imputation using a Markov chain Monte Carlo method was performed to adjust for missing values (10-imputation data sets, 25 iterations) 21,22 . The association between pT category (pT1 versus pT2-3 colorectal cancer) and short-term postoperative outcomes was evaluated with univariable logistic regression analysis, expressed as odds ratios (ORs) with 95 per cent confidence intervals. Multivariable logistic regression analysis was performed to adjust for possible confounding factors. Age, BMI and year of surgery were analysed continuously in regression analyses; the remaining variables were analysed as categorical.
To identify risk factors associated with severe complications after elective surgery for pT1 colorectal cancer, logistic regression analyses were performed. Independent variables with P < 0⋅050 in univariable analysis were entered into the multivariable logistic regression model. A risk stratification table was developed for severe complications after surgery for pT1 colorectal cancer, stratified for sex (men versus women), type of operation (right colectomy versus left colectomy versus sigmoid resection versus LAR versus APR) and ASA grade (I-II versus III-V).
Bootstrapping was performed to calculate 95 per cent confidence intervals.
GraphPad Prism ® version 7.02 (GraphPad Software, La Jolla, California, USA) and Microsoft Visio ® version 2010 (Microsoft, Redmond, Washington, USA) were used to draw figures. All analyses were performed in IBM SPSS ® version 23.0 software (IBM, Armonk, New York, USA). Statistical significance was defined as P < 0⋅050.

Results
Of 51 470 surgically treated patients with colorectal cancer identified, 39 813 fulfilled the inclusion criteria ( Fig. 1). Some 5170 (13⋅0 per cent) were diagnosed with pT1, 9701 (24⋅4 per cent) with pT2 and 24 942 (62⋅6 per cent) with pT3 colorectal carcinoma. The mean age of the cohort was 71 years and 54⋅4 per cent were men. Baseline characteristics are shown in Table 2. Patients with T1 CRC were significantly younger, more often men, and had a lower ASA grade (all P < 0⋅001). pT1 cancers were more often screening-detected, more frequently diagnosed in 2015-2016 and more often located in the rectum (all P < 0⋅001). Patients with pT2-3 tumours more often had preoperative complications and underwent open surgery more frequently (both P < 0⋅001). Patients treated with a local surgical procedure before bowel resection accounted for 1⋅3 per cent of the complete cohort. Ileocaecal and transverse resections accounted for 0⋅6 and 2⋅1 per cent of operations respectively; these were recategorized as right colectomies. Panproctocolectomy and subtotal colectomy accounted for 0⋅3 and 1⋅3 per cent respectively, and were recategorized as (sub)total colectomies.

Time trends
An increase in the absolute number of patients treated surgically for colorectal cancer was observed over time, from 3139 in 2009 to 6864 in 2016. The proportion of pT1 cancer increased from 8⋅1 per cent in 2009 to 17⋅7 per cent in 2016 (P < 0⋅001) (Fig. 2). The steepest increase was between 2014 and 2015 (+4⋅4 per cent), with 2014 being the year in which the colorectal cancer screening programme was introduced in the Netherlands. The proportion of screening-detected pT1 tumours among all pT1 colorectal cancers increased from 34⋅6 per cent in 2014 to 61⋅3 per cent in 2016 (P < 0⋅001) (Fig. 3).

Risk stratification in patients with pT1 colorectal cancer
Factors associated with severe complications after surgery for pT1 colorectal cancer are shown in   . Increasing risk is indicated by change in colour from dark green to light green to yellow to orange to red. Values in parentheses are 95 per cent confidence intervals. *Includes ileocaecal resection and transverse resection. n.a., Not applicable (sample size too small); LAR, low anterior resection; APR, abdominoperineal resection risk of severe complications. Sigmoid resection was associated with a lower risk of severe complications (versus right colectomy; adjusted OR 0⋅67, 0⋅52 to 0⋅87). Using these risk factors, severe complication risk was stratified (Fig. 4).
Women with ASA grade I-II and pT1 disease who underwent right colectomy or sigmoid resection had the lowest risk of severe complications (5 per cent or less), whereas men with ASA grade III-IV and pT1 disease treated with right or left colectomy had the highest risk of severe complications (more than 19 per cent).
Severe complication risks of surgery for pT2-3 colorectal cancer stratified for the same risk factors showed similar results. Women with ASA grade I-II who underwent sigmoid resection had a 5 per cent risk of severe complications and men with ASA grade III-IV treated with left colectomy had an 18⋅8 per cent risk (Fig. 4).

Discussion
This population-based cohort study demonstrates that patients undergoing elective bowel resections for pT1 colorectal cancer have similar risks for surgical complications, severe complications and mortality as those undergoing elective bowel resections for pT2-3 colorectal carcinoma. The absolute difference in overall complication rate following pT1 versus pT2-3 resection was, although statistically significant, considered minor and therefore of little clinical relevance. Implementation of colorectal cancer screening aims to increase cancer-specific survival by diagnosing disease at an earlier stage, but also introduces treatment dilemmas. Early-stage tumours do not necessarily lead to safer surgical procedures.
The risks of postoperative complications after elective surgery for pT1 colorectal cancer have not been well described in previous studies. This is surprising because this type of surgery is frequently performed in clinical practice. Existing literature has focused mainly on advanced stage tumours in patients undergoing emergency surgery, and includes limited analysis of mortality with no morbidity estimates. In the present study an overall postoperative 30-day mortality rate of 2⋅4 per cent was observed for all patients, comparable with previous population-based studies 11,12,23 -25 evaluating mortality risk in patients undergoing elective colorectal cancer resection (1⋅8-3⋅5 per cent). Previous reported relaparotomy rates after surgery for colorectal cancer range from 5⋅8 to 7⋅2 per cent 26 , in accordance with the present study. A recently published study 27 on surgical risks after surgery for non-malignant colorectal polyps showed a low overall 30-day mortality rate of 0⋅7 per cent and a postoperative adverse event rate of 14 per cent. This, however, might be an underestimation as the American College of Surgeons' National Surgical Quality Improvement Program is not representative of all hospitals in the USA. A recently published multicentre study 28 from the Netherlands with more than 900 patients undergoing surgery for benign colorectal polyps showed a 30-day mortality rate of 1⋅4 per cent, which is more in line with the present findings.
Risk factors for severe complications after pT1 colorectal cancer surgery included sex, ASA grade, previous abdominal surgery, type of procedure and type of surgery. This is in line with previous publications, as these factors are frequently used in prognostic scoring for colorectal cancer surgery 16,18,19,29,30 . Most of these existing scoring systems have been based on data of patients with more advanced colorectal carcinoma and include factors such as urgency, perioperative contamination, disseminated cancer, ascites and signs of hypovolaemic shock, which are irrelevant in most early-stage colorectal cancers 29 . The predictive model of the Association of Coloproctology of Great Britain and Ireland was based on a cohort in which 90 per cent of patients had advanced colorectal cancer 31 . The data used to produce the colorectal (CR)-POSSUM model were taken from a wide range of procedures, and more than 30 per cent of the 6790 included procedures were non-elective 32 . In the present study, patient factors such as age, co-morbidity, BMI, tumour location, screening status and pN status were not predictive for severe complications. There has been long-standing controversy about whether age and higher BMI are associated with worse perioperative outcomes. A recent meta-analysis 10 of the effect of BMI failed to show significant influence on overall mortality or reoperation/reintervention rate after laparoscopic colorectal surgery.
A major strength of this study is its nationwide population-based design. Data are compared annually with those in the National Cancer Registry, and show nearly 100 per cent completeness 13,14 , thereby reflecting daily clinical practice. It should be emphasized that patients who had neoadjuvant treatment or were operated on in the emergency setting were not included to avoid major confounding of postoperative outcomes. Several limitations should be mentioned. Inherent to a retrospective analysis, unmeasured confounding could be a source of bias. Although adjustment was made for possible confounders in multivariable analyses including screening status, a healthy user bias cannot be excluded. In previous papers, common factors such as educational level and regular check-up experience were identified as determinants of participation in colorectal cancer screening 33 . Therefore, screened participants could be less vulnerable for postoperative complications, regardless of pT status. The stratified risk model might slightly overestimate the actual risk, because of the decline of short-term mortality after colorectal surgery in the past decade, which was shown in this study as well as in other population-based studies 24 . Finally, the proportion of patients with pT1 colorectal cancer that was clinically staged correctly was not known. Diagnosis by endoscopy or imaging can be misleading and either overestimate or underestimate the actual tumour stage. This may influence surgical risks and oncological benefit in either direction.
Screening programmes target a population regardless of life expectancy. Additional surgery in patients with high-risk pT1 colorectal cancer should be well considered. Clinicians should estimate the patient's competing risks of morbidity and mortality. The risk stratification (Fig. 4) helps to estimate individual risks of significant morbidity and can be used before surgery in shared decision-making of whether or not to perform completion surgery for pT1 colorectal cancer.

Disclosure
The authors declare no conflict of interest.