Limitations of tissue micro array in Duke's B colon cancer

Tissue micro array (TMA) is widely used in cancer research in search of new predictive and prognostic markers. Colon cancer is known to be heterogeneous and the present study addresses some methodological aspects using cores of different size and analysing markers with different cellular distribution. We selected 61 paraffin-embedded tissue blocks representing patients diagnosed with Dukes B colon cancer. Two 1 mm and two 2 mm cores were taken from both the centre and the invasive front of the tumour respectively. The immunostaining included MLH1, MSH2, PMS2, p53, COX-2, TIMP and Betacatenin. Twenty-five percent of the cores taken from paraffin blocks less than 0.5 cm was lost and the total loss was 8%. The homogeneous stains (MLH1, MSH2 and PMS2) all showed high agreement between TMA and whole tissue stains (kappa = 0.96,1 and 1 respectively). The COX-2, p53 and Betacatenin illustrated moderate to high agreement (kappa = 0.54-0.9) whereas TIMP-1 had the lowest score (kappa 0.19-0.25). The application of TMA in Dukes B colon cancer has several pitfalls and depends substantially on the immunohistochemical marker in question. Therefore a validation study seems justified before applying large scale TMA in this setting.