LAUSR

Conventionally, lymphatic spread is regarded as the principal mechanism by which haematogenous metastasis occurs in colorectal cancer. The aim of this cross sectional study was to determine the relative strengths of direct tumour spread, the presence of lymph node metastasis and histologically demonstrated venous invasion as drivers of haematogenous metastasis diagnosed at the time of resection of colorectal cancer. The data were drawn from a hospital database of consecutive bowel cancer resections between 1995 and 2017 inclusive. The presence of haematogenous metastasis was determined at the time of surgery by imaging or other investigations or operative findings. Where possible, histological confirmation was obtained. Specimen dissection and reporting followed a standardised procedure. Tumour staging was according to the 7th edition of the UICC/AJCC pTNM system. Analysis was by multivariable logistic regression. After exclusions 3133 patients remained, among whom 380 (12.1%) had one or more haematogenous metastases. In bivariate analyses, the frequency of haematogenous metastasis was directly associated with increasing T status (p<0.001), increasing N status (p<0.001) and increasing extent of venous invasion (p<0.001) and with some other patient and tumour features. In a multivariable model, after adjustment for other features, associations with the occurrence of haematogenous metastasis were as follows: T3 odds ratio (OR) 4.41 (95% confidence interval 2.40-8.10), p<0.001; T4a OR 6.29 (3.27-12.10), p<0.001; T4b OR 5.50 (2.71-11.15), p<0.001; N1 OR 3.39 (2.47-4.64), p<0.001; N2 OR 4.59 (3.21-6.54), p<0.001; mural venous invasion OR 2.18 (1.14-4.16), p=0.018; extramural venous invasion OR 2.91 (2.21-3.83), p<0.001. Only three other features had significant, though weak effects in the model. These results led to the conclusion that venous invasion, demonstrated histologically and also inferred independently by the extent of direct tumour spread, made a greater contribution to the occurrence of haematogenous metastasis than did spread through lymphatics. Our approach and findings may have implications for other cancer sites apart from colorectal cancer.