Competing risks analysis of the effect of local residual tumour on recurrence and cancer-specific death after resection of colorectal cancer: implications for staging

The pTNM staging system for colorectal cancer (CRC) is not entirely effective in discriminating between potentially curative and non-curative resections because it does not account for local residual tumour in patients with stages I, II or III. This study aimed to evaluate the prognostic importance of histologically verified tumour in any line of resection of the bowel resection specimen (TLR) in relation to pTNM stages and to demonstrate how TLR may be integrated into pTNM staging. Information on patients in the period 1995 to 2010 with complete follow-up to the end of 2015 was extracted from a prospective database of CRC resections. The outcome variables were the competing risks incidence of CRC recurrence and CRC-specific death. After exclusions, 2220 patients remained. In 1930 patients with pTNM stages I-III tumour, recurrence was markedly higher in those with TLR than in those without (HR 6.0, 95% CI 4.2-8.5, p < 0.001) and this persisted after adjustment for covariates associated with recurrence. CRC-specific death was markedly higher in the presence of TLR (HR 7.7, CI 5.3-11.2, p < 0.001), which persisted after adjustment for relevant covariates. These results justify removing patients with TLR from pTNM stages I to III and placing them in stage IV, thereby allowing the categorisation of all patients with any known residual tumour into three prognostically distinct groups. This study demonstrates how TLR may be integrated into pTNM staging, thus improving the definition of the three stages which are considered potentially curable (I, II and III).