Table 16.1 lists the most relevant studies which have attempted to answer the utility of chemotherapy in stage II colon cancer. While not a perfect study, only the QUASAR trial [3] comes close to being a truly randomized trial of stage II colon cancer patients with reasonable power to answer the question of a chemotherapy benefit. All the other studies are either underpowered or are pooled subset analyses of randomized trials. Most of these are very well done scientifically, including a meta-analysis done by the Cochrane group [4], but suffer from biases inherent in pooled analysis. To this mix we also add registry data which are the weakest of all the study types in the table because of uncontrolled threats to internal validity. In the QUASAR trial, 5-FU chemotherapy resulted in a statistically significant improvement in overall survival and disease free survival. The absolute magnitude of the survival benefit was 3.6 % (95 % CI: 1–6 %) meaning you would have to treat 28 patients with chemotherapy to save one life. No other study confirms this survival advantage statistically, but most suggest a magnitude of benefit which is not inconsistent with the QUASAR results either for overall survival (OS) or at least for disease free survival (DFS) [4–11]. The exception to this are 2 registry studies from the United States and British Columbia which fail to show any advantage to chemotherapy and may even suggest it is detrimental [12, 13]. The two trials utilizing more modern oxaliplatin-based chemotherapy do not appear to show a significant improvement over 5-FU alone for stage II colon cancer patients, although they are underpowered to answer this question with any certainty [6, 11].

It is important to clarify terminology surrounding risk stratification, namely the distinction between a prognostic versus a predictive factor. Prognostic factors relate to the expected outcome of patients with those factors. Predictive factors are ones which determine how well a patient will respond to a particular therapy or intervention. A common fallacy to which we are all susceptible is that patients with the worst prognosis are the ones most likely to benefit from aggressive treatment. It is sometimes the case but often it is not. In stage II colon cancer, the most commonly recognized prognostic factors are as follows: T4 disease, inadequate lymph node sampling (<12 lymph nodes), poorly differentiated histology (in MSI-L/S patients), perforation, obstruction, lymphovascular invasion, perineural invasion, and positive resection margins [14]. It is very important that the reader pay special attention to the high grade tumor histology and the importance of interpreting this in the context of the mismatch repair (MMR) or microsatellite instability (MSI) status of the tumor. As we will go into detail below, tumors with MMR deficiency or MSI-H phenotype are often high grade yet have an excellent prognosis.

Whether these adverse risk factors are predictive of benefit to chemotherapy is less clear. The strongest data to suggest a benefit of chemotherapy in high risk groups comes from the British Columbia Cancer Agency (BCCA) registry which found a significant survival advantage for patients with T4 tumors who received 5-FU chemotherapy (HR 0.5 95 % CI: 0.33–0.77) [12]. In contrast neither a US Intergroup meta-analysis nor a SEER registry study could discern any differential chemotherapy advantage for high versus low risk groups [7, 13]. In the MOSAIC study utilizing oxaliplatin-based therapy, there was a suggestion of a disease free survival advantage in the high risk stage II group with 5 year DFS of 82.3 % versus 74.6 % (HR 0.72; 95 % CI: 0.5–1.02) for FOLFOX versus infusional 5-FU alone [6].

The strongest data for both a prognostic and predictive factor exists for a deficiency in the mismatch repair pathway. It is beyond the scope of this chapter to explain the nuances of MMR deficiency and testing for it; but in brief, patients with defective MMR tumors either have a germline loss of one of the MMR proteins (MLH1, MSH2, MSH6, PMS2) or epigenetic silencing of the MLH1 promoter [15]. The former is associated with Lynch syndrome and the later often in the setting of a CpG Island methylator phenotype (CIMP). Defective MMR tumors can either be tested for using a PCR panel of 5 reference microsatellite sites; if at least 2 show instability then the tumor is characterized as MSI-H. More often in the clinical setting, immunohistochemistry testing (IHC) is used to stain for the presence or absence of one of the MMR proteins. By convention in the literature, we call a tumor as defective MMR (dMMR) if they are either MSI-H or have an absence of an MMR protein by IHC.

Table 16.2 lists the major studies which have explored the prognostic and predictive value of MMR testing in stage II colon cancer. The majority of these studies clearly show that those with dMMR stage II tumors have a superior prognosis compared to those with pMMR with hazards of recurrence or death often 50 % lower [16–20]. In the study by Sargent, et al. [19] patients with dMMR tumors who received chemotherapy had a hazard of death which was nearly three times those who were on observation (HR 2.95; 95 % CI: 1.02–8.54). The reason why this may be the case is speculative, but we know patients with dMMR often have an intense immune response to their tumors and are in fact the only colon cancer cohort to date where immune checkpoint inhibitors appear to be effective in the metastatic setting [21]. It is suggested that chemotherapy may blunt this immune response. This hypothesis is further corroborated by recent data suggesting that if there is a chemotherapy benefit for these patients, it is only for those with germline tumors which tend not to express the hyper-mutated phenotype [22]. These findings of a detrimental impact have not been corroborated by the other studies listed in Table 16.2. However, no study has found a clearly beneficial impact of chemotherapy for this cohort, who have an otherwise excellent prognosis. It is important to note that all of these studies are severely limited by power to test for the interaction between dMMR status and chemotherapy effect.