Total mesorectal excision (TME) remains the gold standard for treatment for rectal cancer. The focus on organ preservation in the management of rectal cancer has resulted in local excision as a viable alternate treatment in selected patients with early rectal cancer.

Despite the use of preoperative staging with endoscopic rectal ultrasound (ERUS), magnetic resonance imaging (MRI), and biopsies of rectal lesions before selection for local treatment or radical resection, both over- and understaging and occurrence of high-risk histological features is commonly seen [1,2,3,4,5]. Also, benign biopsies of the lesion does not rule out malignancy, as they may miss the invasive component of the lesion, especially in larger tumors [6,7,8,9]. Several minimally invasive techniques, such as endoscopic mucosa resection (EMR), endoscopic submucosal dissection (ESD) and transanal endoscopic microsurgery (TEM) can be used for excision of rectal tumors. Treating endoscopic benign rectal lesions with EMR, often accomplished by piece meal resection in larger tumors, may lead to difficulties regarding correct histological classification and evaluation of resection margins in cases of malignancy. In terms of long-term results, full-thickness local excision by TEM of early rectal cancer (pT1) with no adverse features has equivalent outcomes to radical resection [10,11,12,13]. It is furthermore associated with good quality of life, functional outcomes, and lower intra- and postoperative morbidity and mortality, as compared to conventional radical rectal surgery, [10, 13,14,15], which is especially desirable especially in the elderly and more fragile patient. However, both unexpected finding of adenocarcinoma in a polyp, found with increasing incidence since implementation of colorectal cancer screening [16,17,18], and adverse histological features or non-radical resection following TEM, may require completion surgery (cTME).

Concerns regarding cTME have been raised in the recent years. Although the literature is limited, there are conflicting results on outcomes, with possible increase in morbidity, poorer TME specimen quality, higher abdominoperineal resection (APR) rates, along with an increased risk of intraoperative rectal perforation [19,20,21,22,23], which is an important predictor of local recurrence and impaired survival [24, 25]. Furthermore, the most optimal timing for cTME after TEM remains uncertain [22, 26].

This study aims to compare short-term clinical and pathological outcomes (primary outcome) as well as long-term oncological outcomes (secondary outcome) of cTME with primary TME (pTME) in patients with rectal cancer.

Methods

We performed a nationwide retrospective controlled cohort study with data from the Danish Colorectal Cancer Group (DCCG) database, which is a prospectively maintained and validated clinical database with predefined clinical, surgical, radiological, and pathological variables, and contains 99% of all patients with colorectal cancer in Denmark [27]. For the cTME group, the patients were selected by extraction of data on all patients with transanal endoscopic surgery for rectal cancer prior to total mesorectal excision. Since local excision before radical resection was not registered in the DCCG database prior to year 2014, the patients with TEM before TME were identified by a surgical code search in the Danish National Patient Registry. The data from DCCG and Danish National Patient Registry were supplemented with the data from National Pathology Data Bank (Patobank) for all included patients (both cTME and pTME group). For detailed information on preoperative imaging, specifics of TEM- or TME procedure, tumor location, and late surgical and oncological outcome, data from DCCG database, Danish National Patient Registry and Patobank was cross-checked and supplemented with manual reviews of hospital charts. The study period ranged from January 2005 to December 2015.

Inclusion criteria for the study were (1) patients ≥ 18 years old with rectal cancer; (2) TEM with curative intent in the cTME group (preoperatively assessed lesion as adenoma or T1 cancer without unfavorable histological risk factors (high tumor level, poor tumor differentiation, lymphatic-, venous-, or perineural invasion, and tumor budding)); and (3) cTME within 12 weeks of TEM (cTME group). Exclusion criteria were radio—chemotherapy before TEM and/or between TEM and cTME, or prior to pTME; (2) patients with FAP (Familial Adenomatous Polyposis) or HNPCC (Hereditary Non-Polyposis Colorectal Cancer); (3) distant metastasis at time of primary surgery; (4) pathological T stage 4 (pT4); (5) synchronous cancer; (6) previous colorectal cancer or current cancer in other organs; (7) active inflammatory bowel disease.

Patients were followed-up until 31 December 2018 or until death. Follow-up was carried out in accordance with DCCG guidelines on colorectal cancer [28]. The 30-day morbidity variables were predefined from the DCCG database [27] and supplemented with information from hospital charts to acquire 90-day morbidity and mortality rates. Postoperative complications were graded according to the Clavien–Dindo score [29] and were registered as one complication (the most severe) per patient. If the TEM specimen showed unexpected adenocarcinoma, staging with thoracoabdominal computed tomography (CT) scan was performed, according to national guidelines [28]. Time to cTME was investigated as a risk factor for APR. Because there is no clear consensus regarding the ideal interval between TEM and cTME, we chose a cut off corresponding to the median time to cTME with short time to cTME defined as time (weeks) below the median value. The quality of the mesorectum was determined using pathology reports and scored using three grades according to Quirke’s mesorectal grading (complete (grade I), nearly complete (grade II), and incomplete (grade III)) [30]. A nonradical resection (R1) was defined as tumor tissue ≤ 1 mm from the circumferential resection margin (CRM) at the histopathological evaluation of the completion specimen and as tumor tissue ≤ 1 mm from the resection margin in TEM specimen. Local recurrence was defined as histologically verified adenocarcinoma at or near the previous resection site. Distant metastases were defined as recurrence in other organs.

This study was approved by the Danish Patient Safety Authority (No. 3-3013-2406/1) and reported to Danish Data Protection Agency (No. 2012-58-0004). The data were obtained with the permission and full support of the scientific committee of the Danish Colorectal Cancer Group.

Statistical analysis

Comparison between patients with cTME and pTME was done after propensity score-based analysis [31]. Variables included for propensity score matching were age, gender, tumor distance from anal verge, American Society of Anesthesiologists (ASA) score and American Joint Committee on Cancer (AJCC) stage. Following construction of propensity score, patients in the cTME group were matched with patients in the pTME group at a ratio of 1:2, with the nearest neighbor approach, and caliper of 0.2 times the standard deviation of the logit of the propensity score [32].

Categorical data are presented as absolute numbers and percentages, and continuous data as mean with standard deviation, unless stated otherwise. Differences between the two groups were calculated using a chi-square test or Fisher’s exact test, as appropriate, for categorical variables, and Student’s t test or Mann–Whitney U test, as appropriate, for continuous variables. Statistics of the risk factors that were measured was carried out by univariate and multivariate regression analysis. The variables potentially related to the risk measured with p value ≤ 0.2 in the univariate analysis were entered into a multivariate analysis. Comparisons of survival was performed with a Cox proportional hazard model with reporting of hazard ratios (HR), with a 95% confidence interval (95% CI). Survival curves were plotted using the Kaplan–Meier method. A p value < 0.05 was considered statistically significant. The data were collected and analyzed using the statistical package IBM® SPSS® Statistics version 26.0 for Windows.

Results

A total of 60 patients with rectal cancer and completion TME (cTME) within 12 weeks of TEM in the period between 2005 and 2015 were identified from the Danish Colorectal Cancer Group database. These patients were matched with 120 patients with primary TME (pTME) according to the preset variables stated in the above. Patient characteristics for the two groups, including definitive pathology following TEM, are presented in Table 1. In 54/60 patients (90%) a biopsy was taken prior to TEM, showing benign histology in 46 of the cases (85%). Diagnostic imaging was used in approximately 1/3 of the patients (27% and 32% for ERUS and MRI, respectively). TEM was performed as full-thickness excision in 80% of the patients (n = 48) and the defect was closed in 72%. TEM was performed as en bloc excision in 46 patients (77%). The tumor was located in the anterior rectal wall in 12 patients (20%), posterior wall in 15 (25%) and lateral rectal wall in 23 patients (38%). In the remaining cases, the tumor location was not specified in the surgical report. Histopathological evaluation showed a nonradical TEM resection (R1) in 39 patients (65%). The median time from TEM to cTME was 5 weeks.

Table 1 Patient characteristics
Full size table

Short-term outcomes

The results of surgery following TME for both groups are presented in Table 2. There was no significant difference between cTME and pTME regarding APR rate (28.3% vs. 17.5%, p = 0.121). Patients with cTME experienced more intraoperative complications as compared to pTME patients (18.3% vs. 6.7%, p = 0.021). However, intraoperative bowel perforation at or near the previous tumor site did not differ between the two groups (6.7% vs. 2.5%, p = 0.224). The intraoperative complications in the cTME group were as follows: 2 splenic lesions, 2 vaginal lesions, 2 sacral vein lesions, 2 lesions of the small bowel, and 4 bowel lesions at or near the previous TEM site. In the pTME group there were 2 lesions of the small bowel, 2 sacral vein lesions, 3 bowel lesions at the tumor site, and 1 large bowel lesion that was unrelated to the tumor site. All complications were managed intraoperatively. The 30-day morbidity and mortality rates were similar in both groups (Table 2). There were no further complications beyond 30 days, and thus the 90-day and 30-day morbidity and mortality rates were the same.

Table 2 Operative results following total mesorectal excision in the two groups
Full size table

Table 3 shows the univariate analysis of risk factors for APR. Of all the risk factors considered, age (> 70 years) and distance from anal verge (≤ 5 cm) showed statistical significance in the univariate analysis. By multivariate analysis age > 70 years (OR 2.78; 95% CI 1.07–7.19; p = 0.035) and ≤ 5 cm tumor distance from anal verge (OR 25.37; 95% CI 9.49–67.84; p < 0.001) were found to be the only independent predictors for APR (Table 3).

Table 3 Risk factors for abdominoperineal resection
Full size table

In a subgroup univariate analysis of cTME patients only, distance from anal verge and short time to completion surgery (< 5 weeks) were statistically significant risk factors for APR (p = 0.001 and p = 0.014 respectively), while leaving the TEM defect unsutured was not associated with an increased risk of APR (p = 0.163). Tumor location was an insignificant risk factor (p = 0.733). Following multivariate subgroup analysis of cTME patients only, low tumor distance from anal verge remained the only significant risk factor for APR (OR 14.3; 95% CI 2.75–74–38; p = 0.002): However, time to surgery < 5 weeks revealed an increased risk for APR (OR 5.15; 95% CI 0.98–27; p = 0.053) (Table 3).

Pathological findings and long-term outcomes

The histological examination of the TME specimen showed no residual tumor in 41/60 (68%) of the cTME patients. Table 4 shows the pathological findings. No differences were found regarding the completeness of the mesorectal fascia (MRF) or R1 resection rate. Univariate analysis showed that occurrence of intraoperative complications and low distance from anal verge (≤ 5 cm) were significant risk factors for incomplete MRF (Table 5). In the multivariate analysis, the occurrence of an intraoperative complication remained the only significant risk factor (OR 6.84; 95% CI 1.97–23.79; p = 0.003).

Table 4 Oncological outcomes
Full size table
Table 5 Risk factors for incomplete mesorectal fascia
Full size table

After a total mean follow-up of 6 years, there were no differences in the rate of local recurrence (Fig. 1) or distant metastases between the groups. Correspondingly, pTME patients did not have an improved overall survival when compared with cTME patients (OR 0.55; 95% CI 0.28–1.08; p = 0.081) (Fig. 2).

Fig. 1
figure 1

Recurrence rate. Kaplan–Meier curve with log rank test for recurrence rate (p = 0.975). cTME completion total mesorectal Excision, pTME primary total mesorectal excision

Full size image
Fig. 2
figure 2

Overall survival. Kaplan–Meier curve with log rank test for total overall survival (p = 0.081). cTME completion total mesorectal Excision, pTME primary total mesorectal excision

Full size image

Discussion

This nationwide register-based study suggests that completion TME following local excision for rectal cancer was not a statistically significant risk factor for APR. Furthermore, there was no difference in the rate of tumor perforation, quality of the mesorectal fascia, and long-term oncological outcomes between patients with completion- and primary TME.

Accurate preoperative staging of early rectal cancer is difficult. Imaging modalities such as EUS and MRI cannot always correctly distinguish between benign and malignant tumors and tend to overstage malignant lesions [3, 4]. Benign biopsies of the lesion will not rule out malignancy. Contrary to other local excision techniques, such as EMR, full-thickness TEM can provide both complete pathological staging and definite treatment in case of unexpected malignancy (malignant polyps). Therefore, TEM is a reasonable selection of treatment for certain rectal tumors, since it is curative for patients with early rectal cancer and favorable histological characteristics, and offers equivalent oncological results to radical resection [10, 11, 33, 34]. This approach offers an organ preserving option for patients with low rectal cancer.

When the TEM specimen shows nonradical resection and/or unfavorable histological parameters, completion TME is warranted. Several concerns regarding completion TME have been raised in previous studies. The main concern is an increased rate of APR with permanent colostomies compared to patients without a previous TEM [22].

In the present study, we found a significant difference in intraoperative complications between patients with cTME and pTME, but no difference in rate of perforations at or near the tumor/previous TEM site. The tumor was located in the posterior wall in 25% of the patients with TEM in our study population. It is possible that patients with a lesion located in the posterior wall of rectum have a reduced risk of perforation during cTME because the defect is covered by the mesorectum.

Furthermore, the rate of APR did not differ between patients with cTME and pTME (28.3% vs. 17.5%, p = 0.121). However, on multivariate analysis, previous TEM, but not intraoperative morbidity, showed tendency as a risk factor for APR. This is in accordance to the few other studies on cTME that have reported morbidity rates [21,22,23]. In a study by Morino et al. where 17 patients with laparoscopic TME after TEM were compared with 34 patients undergoing primary TME, previous TEM was the only independent predictor for APR (OR 4.13; 95% CI 1.09–15.55; p = 0.046) [22]. Their finding has not been supported by others, but methodological issues with matching based on procedure type may have contributed to the conflicting findings in other studies [21, 23]. Although small, recent reports have not shown higher APR rates in patients with cTME [19, 35]. The potentially higher rates of APR in patients with cTME may be due to inflammatory changes leading to scarring and fibrotic changes in the mesorectum surrounding the previous TEM site, resulting in severe adhesions and challenges for surgical dissection in the correct anatomic planes and performing anastomosis. Although not fully investigated yet, it is possible that completion surgery with transanal TME (taTME) may result in higher sphincter preserving rates in patients with low rectal tumors, as these inflammatory changes are more evident closer to the anal verge [36]. At the moment, only three studies with small sample sizes have described outcomes of patients with completion taTME, reporting higher sphincter preserving rates, lower perforation rates and better mesorectal quality when compared with conventional cTME [37,38,39]. Full-thickness local excision with other platforms, such as TEO or TAMIS may achieve similar results [35, 36].

Leaving the rectal wall defect unsutured after TEM has been proposed to contribute to higher local complications and APR rates [21]. In our subgroup analysis of cTME patients only, we did not find this to be a risk factor for APR. An interesting finding was however, that a short time interval to cTME may increase the risk for APR (OR 5.15; 95% CI 0.98–27; p = 0.053). So far only two previous studies have reported outcomes based on the time interval to cTME. Hompes et al. showed poorer specimen quality after an interval from TEM to cTME of more than 7 weeks [26]. However, Morino et al. did not find any difference in outcomes among patients operated within 30 days vs more than 30 days after TEM [22]. The optimal timing for completion surgery needs to be further investigated in larger studies.

The local inflammation following TEM and subsequent surgical challenges in completion surgery has also raised a concern regarding the impact of the quality of the mesorectal fascia. Poor mesorectal quality along with iatrogenic rectal perforation is proven to worsen the oncological outcome after TME [24, 40, 41]. The results regarding long-term oncological results in patients with cTME are, however, very limited [23, 35]. A meta-analysis showed that incomplete mesorectal excision rate was significantly higher in patients with a previous transanal excision (OR 5.74; 95% CI 2.24–14.75; p = 0.0003) [20]. The meta-analysis included only 3 studies with a total of 56 patients with a previous TEM, and one of the included studies consisted of patients with preoperative chemo- or radiotherapy, and the defect in the rectal wall was left unsutured in all patients, which limits the accuracy of the results [21]. Multivariate analysis in our study failed to identify previous TEM as a risk factor for poor specimen quality. Although not investigated with multivariate analysis, recent smaller studies have shown similar results [19, 35]. Our study shows no difference in rate of local recurrence, distant metastases or survival between patients with cTME and pTME after a follow-up of at least 5 years.

There are limitations to this study. First, the nonrandomized nature and retrospective design constitutes a potential risk of selection bias. We performed propensity score matching to increase comparability between patients with cTME and pTME. Secondly, diagnostic imaging prior to TEM was used in only 1/3 of the patients and we lacked information on the size of the tumor and the defect created by TEM. It is possible that a cohort with large defects would have yielded different results. Also, the lack of information on tumor size prevented us from adding this to the variables included for the propensity score matching. Thirdly, there was a small number of patients in the cTME cohort which may limit the interpretation of the results and generate a type II statistical error. This is reflected in the wide confidence intervals in our multivariate analysis. However, the sample size of 60 patients is to our knowledge the largest number of patients with cTME in the literature so far. Furthermore, this is a nationwide study with 100% follow-up completeness due to manual reviews of charts and cross-checking with National Patient Registry and National Pathology Data Bank, which we believe strengthens the conclusions regarding long-term outcomes.

In conclusion, the present study did not demonstrate significantly higher rates of APR in patients with cTME following TEM. The intraoperative perforation rates and specimen quality was comparable between the two groups. Furthermore, in patients with cTME the long-term oncological outcomes were similar to pTME.