M E Kelly, R Glynn, A G J Aalbers, M Abraham-Nordling, W Alberda, A Antoniou, K K Austin, G L Beets, J Beynon, S J Bosman, M Brunner, M W Buchler, J W A Burger, N Campain, H K Christensen, M Codd, M Coscia, A J Colquhoun, I R Daniels, R J Davies, J H W de Wilt, C Deutsch, D Dietz, T Eglinton, N Fearnhead, F A Frizelle, J L Garcia-Sabrido, M L George, L Gentilini, D A Harris, D Harji, A G Heriot, W Hohenberger Brunner, J T Jenkins, Y Kanemitsu, K K L Chan, H Kim, C E Koh, N F Kok, C Kontovounisios, Y Kulu, W L Law, G N Lê, P Lee, M L Lydrup, A C Lynch, A Martling, J Meijerink, S Merkel, F D McDermott, J S McGrath, M B Nielsen Christensen, G A P Nieuwenhuijzen, M A Nordling, J M A Northover, P R O'Connell, D Patsouras, G Poggioli, R W Radwan, S Rasheed, P C Rasmussen, J Rothbarth, H J T Rutten, P M Sagar, A M P Schizas, D Shida, N J Smart, M J Solomon, L Stocchi, P P Tekkis, S Tsukamoto, W H Turner, J Tuynman, A Ulrich, A van Leeuwenhoek, G H van Ramshorst, W Vasquez-Jimenez, C Verhoef, M Versteegen, C Wakeman, S Warrier, J Yip, D C Winter

Pelvic exenteration for locally recurrent rectal cancer (LRRC) is associated with variable outcomes, with the majority of data from single-centre series. This study analysed data from an international collaboration to determine robust parameters that could inform clinical decision-making. Anonymized data on patients who had pelvic exenteration for LRRC between 2004 and 2014 were accrued from 27 specialist centres. The primary endpoint was survival. The impact of resection margin, bone resection, node status and use of neoadjuvant therapy (before exenteration) was assessed. Of 1184 patients, 614 (51·9 per cent) had neoadjuvant therapy. A clear resection margin (R0 resection) was achieved in 55·4 per cent of operations. Twenty-one patients (1·8 per cent) died within 30 days and 380 (32·1 per cent) experienced a major complication. Median overall survival was 36 months following R0 resection, 27 months after R1 resection and 16 months following R2 resection (P < 0·001). Patients who received neoadjuvant therapy had more postoperative complications (unadjusted odds ratio (OR) 1·53), readmissions (unadjusted OR 2·33) and radiological reinterventions (unadjusted OR 2·12). Three-year survival rates were 48·1 per cent, 33·9 per cent and 15 per cent respectively. Bone resection (when required) was associated with a longer median survival (36 versus 29 months; P < 0·001). Node-positive patients had a shorter median overall survival than those with node-negative disease (22 versus 29 months respectively). Multivariable analysis identified margin status and bone resection as significant determinants of long-term survival. Negative margins and bone resection (where needed) were identified as the most important factors influencing overall survival. Neoadjuvant therapy before pelvic exenteration did not affect survival, but was associated with higher rates of readmission, complications and radiological reintervention.