Prognostic biomarkers for stage II and III colon cancer

As we learn more and more about the underlying biology of colon cancer, biomarkers will play a prominent role in the improvement of the identification of patients with high risk of recurrence. Once the identification of the prognostic tumour characteristics has been optimized an appropriate therapy can be advised, whereas others may safely choose to refrain from adjuvant therapy. In this thesis we tried to optimize the process of identification of patients with poor prognosis, the efficacy, the safety and the cost-effectiveness of adjuvant therapies need to be addressed separately, and was beyond the scope of this thesis. In Chapter 2 we described the prognostic value of microvessel density (MVD) as a surrogate marker for angiogenesis, one of the hallmarks of cancer. As angiogenesis itself remains difficult to measure directly, MVD may provide insight in this microenvironmental process. This computerized morphological study showed that MVD increases with stage, which may explain the observation that stage II patients with high MVD had poor prognosis, biologically being a kind of pre-stage III. Actual stage III patients with high MVD, that were all treated with adjuvant 5-FU based chemotherapy, showed better prognosis on the contrary. We hypothesized that residual or recurrent tumour tissue in stage III patients, from primary tumours with high MVD, were better penetrable for chemotherapy, explaining the improved survival. In Chapter 3 we tried to validate CDX2 immunohistochemically, being a promising biomarker published in a high-impact journal by Dalerba et al. Our study showed the practical hurdles involved in validating biomarkers in other cohorts. For disease free survival we were not able to find significant differences between high and low expression of CDX2. However, for disease specific survival we did find comparable results. For a subset of 41 patients we analysed expression of CDX2 by both immunohistochemistry and mass spectrometry. Interestingly, discriminatory power of CDX2 as a prognostic biomarker detected by mass spectrometry outperformed the immunohistochemical detection method. This observation shows that in some cases the biomarker itself may be promising, but the method ought to be fine-tuned to benefit from that marker’s full potential. In chapter 4 we presented a new biomarker, obtained from promising research in mice. Expression of KCNQ1 and CD44, both regulated by Wnt-signalling, was analysed immunohistochemically in 386 stage II and III patients, and on mRNA in an external cohort of 90 patients. We concluded that KCNQ1 was a strong prognostic biomarker for disease recurrence. KCNQ1 may be particularly useful in stage II MSS patients, where the question remains which patients are at risk for recurrence and might benefit from adjuvant chemotherapy. In Chapter 5 MACROD2 was discussed, a relatively unknown gene, although a tumour suppressing function by activating PARP1 and DNA repair is suspected. We indeed found that low nuclear expression was associated with poor prognosis in stage III MSS colon cancer patients, treated with 5-FU based ACT. Even more, high expression of MACROD2 may serve as a predictive biomarker in stage III MSS tumours, favouring adjuvant treatment with 5-FU compared to no adjuvant treatment at all. In the final chapter, Chapter 6, we describe an ultimate attempt to improve the identification of high risk patients. Several biomarkers have been studied on our cohort over the years, some with promising and some with disappointing results. For this study we intended to analyse all these markers combined, to find out which (combination of) biomarkers was able to estimate prognosis best, and whether or not previous less important biomarkers became more interesting. Although other techniques were also used on (some subsets of) this cohort, i.e. mass spectrometry, next generation sequencing and morphological analysis (MVD), for this study we chose for a feasible and widely applicable technique like immunohistochemistry. Therefore, all immunohistochemical biomarkers examined previously on tissue micro arrays were included, in addition to all common clinical and pathological parameters available in our cohort. These were all included in a Classification and Regression Tree (CART) analysis, which showed that both lymphovascular invasions (LVI) and KCNQ1 were the key features for estimating prognosis in stage II and III colon cancer.