Study Population
Data were obtained from one Spanish center, the Hospital Clinic of Barcelona, and two Dutch centers, the Gelderse Vallei Hospital and the Amsterdam UMC at AMC. LapTME was first introduced at the Hospital Clinic in 1994. In November 2011, TaTME became the standard of care for all patients presenting with rectal cancer that did not require abdominoperineal resection or pelvic exenteration. In February 2017, the transanal approach became standard for patients requiring an abdominoperineal resection. Gelderse Vallei Hospital is a high-volume rectal cancer institution in which TaTME was first used in 2012. At the Amsterdam UMC TaTME became the standard procedure for patients presenting with mid- and low rectal cancer from 2014 onwards. All patients with histologically proven rectal adenocarcinoma treated by TaTME were prospectively registered in a local standardised database or in the International TaTME Registry (11). Consequently, a multicenter database was created, which included the TaTME cohort and a cohort of patients treated by LapTME between January 2000 and February 2018, through a retrospective analysis of clinical records. All three hospitals used TaTME as a standard procedure for patients with mid- and low rectal cancer until their recent participation in the COLOR III trial, a randomised study in which participants are allocated to either TaTME or LapTME (12).
For this analysis, adult patients with a solitary locally advanced rectal adenocarcinoma, according to the ACOSOG Z6051 definition (cT3/cT4, or cN1/cN2 with any cT) detected by magnetic resonance imaging (MRI) with or without transrectal ultrasonography, within 12 centimetres of the anal verge treated with TaTME or LapTME were included (13). The exclusion criteria were: patients with cTisN0 or cT1-2N0; pelvic malignancy within five years; severe, incapacitating disease, i.e. American Society of Anaesthesiologists (ASA) classification IV-V; procedures performed in an emergency setting; tumours previously treated by local excision; unknown cT or cM; metastatic tumours (M1); synchronous tumours; active Crohn’s or ulcerative colitis; familial risk-colorectal cancer syndromes; and patients with 30-day mortality when it was judged to have occurred as a direct result of a major active postoperative complication, which is not of primary interest. The Institutional Ethics Committees approved the TaTME and LapTME techniques years prior to this study in the three institutions noted, and the current study protocol was assessed and accepted by the local Institutional Review Boards. Patients provided written informed consent.
Endpoints
The primary endpoint was three-year locoregional recurrence. Secondary endpoints included systemic recurrence, disease-free survival, and overall survival.
Procedures and definitions
The specific staging, classification methods, and surgical procedures have been described in more detail previously (14–16). Tumours were considered high if the distal border of the tumour was > 10 cm from the anal verge, mid if it was between 5 and 10 cm, and low in case of a distal border < 5 cm. Patients were eligible for neoadjuvant therapy in cases of cT3b-d/cT4 or cN-positive tumours below the peritoneal reflection, or if the circumferential resection margin was threatened or involved, although other factors such as extramural venous invasion were also taken into account and discussed by a multidisciplinary team. The indication to receive radiotherapy alone or in combination with chemotherapy was given depending on the institution-specific protocols. Short-course one-week radiotherapy was administered by 25 Gy in five daily fractions. Neoadjuvant long-course chemoradiotherapy was administered by continuous 5-fluorouracil (5-FU) infusion (225 mg/m2 for five days per week) or capecitabine (825 mg/m2 twice daily for five days per week), and a total dosage of 45 Gy, by a weekly dose of 9 Gy divided in five days each week, for a total of five weeks. The interval between completion of long-course chemoradiotherapy and surgery was five to seven weeks at the beginning of the LapTME cohort recruitment, and, in accordance with current guidelines, was subsequently extended up to 12 weeks and associated with appropriate restaging (1).
The mesorectal specimen was analysed on the basis of four major pathological factors: the integrity of the mesorectum, graded following the Quirke method: complete, near-complete, or incomplete (5); the circumferential resection margin, considered to be positive when the distance between the deepest portion of the tumour and the resection margin was ≤ 1 mm, or in the case of a positive lymph node at ≤ 1 mm of the radical dissection plane; the distal resection margin, considered to be positive if tumour cells were present ≤ 1 mm from the lower border of the tumour to the cut edge of the specimen; and the number of lymph nodes harvested. Pathological tumour response to neoadjuvant therapy was scored by the Ryan tumour regression grade (three-point TRG): TRG 1, no viable cancer cells, or single cells or small groups of cancer cells; TRG 2, residual cancer outgrown by fibrosis; and TRG 3, significant fibrosis outgrown by cancer, or no fibrosis with extensive residual cancer (17).
Follow-up
Until follow-up was completed after five years, patients visited every three to six months during the first two years and every six to 12 months during the remaining three years. Visits included a history, physical evaluation with digital rectal examination, and determination of the carcinoembryonic antigen level. In Barcelona, imaging studies with thoracic and abdominopelvic CT scans were requested every six months during the first two years and annually during the remaining three years. In both Gelderse Vallei and Amsterdam UMC, imaging study was based on liver ultrasound, and CT scan was performed in case of suspicion of local recurrence or distant metastasis. Pelvic MRI and/or transrectal ultrasound-guided needle biopsy were requested when pelvic recurrence was suspected. Locoregional recurrence was defined as any recurrence in the pelvic area and had to be confirmed at least on imaging.
Statistical analysis
Qualitative variables were expressed as absolute frequencies and percentages. Quantitative variables were reported as means or medians with their 95% confidence intervals (CI), except for follow-up periods, which were expressed as median with range. To allow for an unbiased comparison, an inverse probability of treatment weights approach was used (18). A logistic regression model was applied, including demographic and clinical preoperative variables such as hospital, age, gender, ASA classification, body mass index (BMI), the distance of the tumour from the anal verge, cT and cN stage, and baseline threatened or involved circumferential resection margin. As the goal was to develop a balanced population that was independent of the outcome assessment, postoperative variables were not included, except for relevant variables that were assessable only after surgery (i.e., pT and pN stage, and pathological response to neoadjuvant therapy). The covariate radiologic extramural venous invasion was excluded from the model due to a significant amount of missing data because this information was not routinely reported until recent years. The covariate type of surgery (categorised as sphincter-saving or abdominoperineal resection) could not be included in the propensity score calculation due to the small number of abdominoperineal resections in the TaTME group.
A well-balanced distribution of the covariates in the weighted sample was confirmed by means standardised differences meeting a standard objective of ± 0.10 (19). The only exception was the covariate hospital, in which a standardised difference of 0.12 was achieved. Since some authors consider the cut-off point for standardised differences to be ± 0.20, and given the extensive homogeneity of the rest of covariates (with standardised differences of less than ± 0.02), this was finally accepted (20).
The estimation of the survival functions was carried out using the Kaplan-Meier method. The estimation of the effect of surgical procedure was performed using Cox hazard models weighted by the inverse probability of treatment weight adjusting by preservation of the sphincter (sphincter-saving surgery or abdominoperineal resection) with a cut-off at three years. Additional analyses using Accelerated Failure Time models were used for the analysis of time to event data in order to estimate the time ratio (TR) for the effect of the surgical procedure on acceleration in the time to the event (21).
Statistical tests were two-sided with a 5% type I error. All the analyses were carried out using SPSS version 25 (IBM) or SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina).