Predicting treatment outcomes in metastatic colorectal cancer : heterogeneity in patient characteristics, treatment efficacy and molecular diagnostics

In recent years, it has become evident that metastatic colorectal cancer (mCRC) is a heterogeneous and molecularly complex disease. Prognosis and treatment response are being influenced by a combination of clinical, pathological and molecular features. Since few biomarkers are currently available, almost all systemic treatments are administered with a one-size-fits-all approach, with only a subset of patients experiencing benefit. The research presented in this thesis focuses on prognostic and predictive markers in order to improve treatment outcomes in patients with mCRC. Although randomised controlled trials are considered the gold standard to evaluate the efficacy of new treatment strategies, they often show heterogeneity in response and survival rates. This could be partly explained by differences in patient characteristics, since many are of prognostic value. Uniform trial reporting of patient characteristics and use of stratification factors is essential to enable a valid comparison of treatment arms, and to evaluate whether study populations are representative of the general patient population. However, there is marked inconsistency in the reporting of patient characteristics in mCRC trials, which was first described by Sorbye et al. in 2007. For example, although several studies have reported that synchronous metastases are associated with worse outcome compared with metachronous metastases, mCRC studies often do not report the distribution, prognosis and a (uniform) definition of synchronous versus metachronous metastases. We observed persistent heterogeneity in the reporting of patient characteristics and use of stratification in 67 first-line phase 3 trials published between 2005-2016. Using a two-round Delphi survey, we have developed the first consensus recommendation among 30 mCRC experts from 15 different countries on essential patient characteristics and stratification factors in mCRC trials. Implementing this minimum set of essential baseline characteristics and stratification factors in study protocols and final reports of mCRC trials will improve interpretation of trial results and cross-study comparisons. The phase 3 CAIRO3 and AIO 0207 trials showed that fluoropyrimidine and bevacizumab maintenance treatment is the preferred strategy in mCRC patients with stable disease or better after induction treatment with a fluoropyrimidine, oxaliplatin and bevacizumab, as it maintains disease control and quality of life without relevant toxicity. However, not all patients may benefit from this strategy. To improve personalised medicine, we aimed to identify patient subgroups according to clinical, pathological and molecular characteristics that benefit most from fluoropyrimidine and bevacizumab maintenance treatment or observation. We combined individual patient data of the CAIRO3 and AIO 0207 trials and found that maintenance treatment was effective, regardless of relevant clinical and pathological subgroups. In a post hoc analysis of the CAIRO3 study, we found that capecitabine and bevacizumab maintenance treatment was more effective compared with observation, regardless of RAS/BRAF wild-type, RAS-mutant or V600EBRAF-mutant subgroups. Within the CAIRO3 study, we also investigated whether quantification of the mutational burden, i.e. the mutant allele fraction (MAF), could be used as independent prognostic factor in mCRC patients with KRAS-mutant tumours. Our findings suggest that KRAS MAFs or MAFs adjusted for tumour purity (adjMAFS) are not independently associated with prognosis in mCRC patients with KRAS-mutant tumours