An evaluation of five different methods for estimating proliferation in human colorectal adenocarcinomas

Five different methods of determining cell proliferation have been compared in samples taken from a group of 125 human colorectal tumours labelled in vivo with iododeoxy-uridine (IUdR). The labelling index (LI) was obtained immunocytochemically using monoclonal antibodies against proliferating cell nuclear antigen (PCNA), the Ki67 antigen and IUdR (IUdRimm). Incorporation of IUdR was also determined flow cytometrically (IUdRfcm) and PCNA expression was measured in both formalin- and methanol-fixed tissue (PCNAf and PCNAm respectively). There was significant variation in the results obtained both within and between the different assays. Paired analysis of the data showed that the correlation between the different methods of determining the LI was poor. However, the IUdRfcm LI was significantly correlated with both IUdRimm (r = 0.39; n = 78; P < 0.001 by Spearman's test) and Ki67 LIs (r = 0.32; n = 87; P < 0.001). The IUdRimm LI was also significantly related to the Ki67 LI (r = 0.44; n = 60; P < 0.001). The median IUdRfcm and IUdRimm LIs were significantly higher in the aneuploid vs. the diploid tumours (17.4% vs. 6.2% for IUdRfcm; 23.2% vs. 18.9 for the IUdRimm; P < 0.001 and P = 0.014 respectively by Mann-Whitney U-test) but none of the other proliferative indices showed this relationship. Finally, none of the LIs showed a significant association with the clinical characteristics of the tumours such as stage, grade, age, sex or fixity. The findings of this investigation highlight the need for carefully controlled studies when assessing the value of proliferation markers in solid human tumours.