Quality of total mesorectal excision, quality of care and prognostic factors of rectal cancer

The mainstay of rectal cancer treatment still relies on surgical resection, which has evolved from blunt and blind dissection to total mesorectal excision (TME) using sharp dissection under direct vision. The TME technique, popularized by Sir Heald, aims at removing the rectum and its lymphatic drainage together with the surrounding fascia recti, resulting at best in a surgical specimen with a smooth mesorectal surface. Oncological results have improved dramatically since the implementation of TME and have been shown to be influenced by the quality of TME reflected in the aspect of the specimen. In Chapter 1, based on a subgroup of 266 patients registered in the PROCARE database, we found that TME quality depends on patient, surgical and pathological factors. Univariate analyses showed that the surgeon, female gender, pathological BMI, negative clinically assessed nodal status (cN), a lower limit of the rectal cancer, cT3 to cT4 tumours not responding to neoadjuvant chemoradiation, laparoscopic resection, and APR were all significantly associated with incomplete mesorectal excision. Pathologic BMI, the absence of downstaging after long-course chemoradiation and laparoscopic resection were identified as independent prognostic factors. Even if the role of surgery remains a treatment cornerstone, modern rectal cancer management is multimodal, i.e. multidisciplinary. Consequently, in Chapter 2 the potential effect of surgical experience and volume was analysed in the broader context of hospital volume. Using a sample of 1469 patients, we investigated the impact of experience or volume of centres on the performance of every discipline involved in the management of rectal cancer as well as on the whole team's performance. Quality of care indicators and long-term oncological outcome were the end-points. The conclusions of Chapter 2 were in favour of a modest but significant effect of patient volume on surgical outcome, specifically on the rates of sphincter saving procedure and radical (R0) resections. An association was also found between hospital volume and neodjuvant treatment for cTNM stage II-III, pCRM reporting and number of lymph nodes examined in the TME specimen after CRT. Morbidity and mortality were not associated with volume. In the PROCARE database, 5-year local and overall recurrence rates were not associated with hospital volume, either before or after adjustment for patient or tumour characteristics. A sensitivity analysis performed on the population-based data set from the Belgian Cancer Registry (BCR) differed from the PROCARE results since volume was significantly associated with adjusted 30-day mortality and overall survival, but not with the sphincter preservation rate. This difference could, at least partially, be explained by the larger number of patients in the BCR and by the registration bias of PROCARE. Based on these results, we concluded that some modest volume effects existed, but that caution was warranted in their interpretation. Such findings, therefore, suggest that in the Belgian context, audit is more important for quality assurance than centralization based on volume criteria alone. Obesity, an additional potential predictor of TME quality and prognostic factor of outcome was specifically investigated in Chapter 3. Unlike other publications focusing on BMI, we also measured obesity using the waist and hip circumferences, taking advantage of its greater ability to evaluate abdominal obesity. The potential influence of obesity on short- and long-term outcome was tested. An effect of the waist/hip circumference ratio on TME quality was observed whereas BMI was not significantly linked to TME quality in this series of 295 patients. A similar effect was demonstrated on surgical morbidity and the risk of re-intervention. Conversely, recurrence and survival were not influenced by any obesity parameter. TME quality is commonly evaluated through a three-graded scale. For the purpose of analysis, two of the three grades were merged and compared to the third one in several reports. In Chapter 4, we formally assessed the reliability of a simplified two-graded scoring system. The adjusted hazard ratio (HR) for distant metastases reached a level of statistical significance when good TME quality was compared to suboptimal resection planes. Goodness-of-fit and performance of this two-graded system was as good as that of the three-graded scoring method. Similarly, the adjusted HR for death during follow-up reached a level of statistical significance for the same two-graded system distinguishing mesorectal from the combined intramesorectal and muscularis propria resection planes. This two-graded scoring system performed better than the alternative two-graded score and could therefore suitably replace the classic three-graded system without losing discriminatory or predictive power. These findings indicate that intramesorectal resection is to be considered suboptimal surgery, increasing the risk for adverse outcome, and that it should not be combined with mesorectal resection for outcome analysis. In Chapter 5 the weight of factors in predictive statistical models of survival and recurrence was scrutinized. Beyond the typical statistically significant relationship between clinically relevant variables and outcome, this work aimed at estimating their unique, individual and relative contribution, i.e. the quantitative contribution or the “effect size” of patient-, tumour- and treatment-related factors to distant metastasis rate and overall survival after rectal cancer resection. The unique contribution of the variables was very modest, ranging from 0.1% to 3.1%. For TME quality for example the unique contribution was 0.8% for distant metastases and 1.2% for overall survival. The unique contribution of all variables was lower than their individual contribution, indicating the presence of some overlap. Overlap explained 6.6% to 9.4% of outcome. More than 80% of outcome variation remained unexplained for each outcome measure. The limited contribution of outcome prognosticators is not inconsistent or incompatible with current principles of good clinical practice for the management of rectal cancer. Causal relationships and predictive power should not be conflated, and the strength of a causal relationship does not imply its predictive power. In fact, unpredictable variability is accountable for individual differences in outcome after optimal treatment of patients with a comparable severity of disease. We concluded that despite the clinical relevance of the variables in the current models, future studies should explore the potential additive value or superiority of new variables, e.g. based on molecular and genetic mapping. Eventually, improved prediction will help physicians to better personalize treatment and follow-up strategies in daily practice. In Chapter 6, prognosticators of previous chapters were further tested on our referral single-centre rectal cancer series. Overall and recurrence-free survival were tested. Local recurrence rate being low (2.2%), further analysis was not deemed possible (instable models). Multivariable analysis indicated that lymph node ratio, bad TME quality, absence of symptoms, peri-neural invasion and age were significantly associated with overall survival, whereas lymph node ratio and extramural vascular invasion were associated with recurrence-free survival. Lymph node ratio displayed the highest hazard ratio for both outcome. These results, based on a smaller single-center series were in line with the previous findings of this work based on PROCARE data.