Microsatellite Instability in Colorectal Cancer: Prognostic, Predictive or Both?

To the Editor-in-Chief:

Guidoboni et al 1 recently described favorable clinical outcome for proximally-located colorectal cancers (CRC) showing microsatellite instability (MSI+). This is in agreement with some 2–4 but not all 5,6 previous studies on the prognostic significance of this genetic alteration. Clarification is urgently required because of the possible consequences for selection of patients to receive adjuvant chemotherapy. Unfortunately, most of the studies to date have been difficult to interpret because of differences in MSI assessment criteria, as well as the use of mixed tumor stages, selected patient populations and adjuvant chemotherapy. For example, both the Guidoboni et al 1 and an earlier study 4 used the Bethesda criteria to define tumors as being MSI-H (high instability), yet the former report an incidence of 43% in proximal tumors compared to 25% for the latter.

We recently found that MSI+, defined only as deletions in the mononucleotide repeat BAT-26, was prognostic for CRC patients treated by surgery and chemotherapy, but not for those treated with surgery alone. 7 The first observation was subsequently confirmed by Hemminki et al. 3 As with other studies, 2,4–6 we compared the survival of MSI+ patients to that of all MSI− patients. However, approximately 90% of sporadic MSI+ tumors are found in the proximal colon, 2–4,7 therefore the outcome of patients with these tumors should be compared to that of patients with MSI-proximal tumors, as in the report by Guidoboni et al. 1 Multivariate analysis for factors affecting survival in stage III proximal tumors showed that MSI+ was a strong prognostic indicator for patients (n = 94) treated with chemotherapy (RR = 0.15, 95% CI:[0.04–0.64], P = 0.010) but not for those (n = 228) treated by surgery alone (RR = 0.70, 95% CI:[0.44–1.14], P = 0.151). A trend for improved survival was evident in the latter group and therefore MSI status may show significant prognostic value when used in combination with activated cytotoxic lymphocyte counts. 1

By comparing the survival of CRC patients treated with or without fluoropyrimidine-based (5FU) chemotherapy, we found that MSI+ is a predictive factor for good survival benefit from chemotherapy. 7,8 This finding is indirectly supported by observations made with the p53 tumor suppressor gene. CRC patients with wild-type, but not mutant p53, gain significant survival benefit from chemotherapy. 8,9 Since almost all MSI+ tumors have wild-type p53, 8,10 it is perhaps not surprising that patients with these tumors should also show good survival benefit from chemotherapy. Validation of the predictive value of MSI+ should preferably be carried out in the context of prospective clinical trials that include adjuvant treatment and non-treatment arms. However, because of the widespread acceptance of 5FU in the treatment of stage III CRC, it will be increasingly difficult to include non-treatment arms in future trials. Clinical trials of stage II CRC where patients are first stratified according to MSI status before randomization into treatment and non-treatment arms could be considered, although these will require larger numbers than for trials of state III CRC. Alternately, large retrospective studies that use defined MSI assessment criteria, tumor stage, patient characteristics, and adjuvant therapy status should allow comparison of survival rates between MSI+ and MSI− proximal CRC patients (prognostic value) and between adjuvant-treated and non-treated MSI+ cases (predictive value). Such studies should finally allow resolution of whether MSI+ in CRC is prognostic, predictive or, as we suspect, both.

Microsatellite Instability in Colorectal Cancer: Prognostic, Predictive or Both?

Massimo Guidoboni, Alessandra Viel, Laura Del Tin, Mauro Boiocchi, Riccardo Dolcetti, Roberta Gafa, Giovanni Lanza, Claudio Doglioni, Ettore Macri, Antonio Russo, Alessandra Santini

Authors’ Reply:

The need for robust markers for a better prognostic definition of patients with colorectal cancer (CRC) has been increasingly pressing during the last decade. The introduction of adjuvant chemotherapy in the routine clinical management of patients with stage III CRC or in patients with high-risk stage II disease further enhanced the need for more precise prognostic indicators to efficiently direct the therapeutic choice. In the last years, several studies have investigated the prognostic significance of high-frequency microsatellite instability (MSI-H) in colon cancer, sometimes yielding conflicting results. 1–5 Notwithstanding, the positive prognostic value of MSI-H in CRC has been convincingly demonstrated by recent large population-based studies in which standardized criteria for MSI-H assessment have been used. 4,5 Besides confirming the favorable clinical outcome of MSI-H CRC, our results also demonstrated that the prognostic value of the MSI status alone is significantly enhanced by the combined evaluation of the number of intratumoral-activated cytotoxic lymphocytes. 6 This supports the hypothesis that MSI-H tumors may continuously produce new immunogenic epitopes as a consequence of the inherent defective DNA mismatch repair and may explain why patients with MSI-H CRC who are able to mount effective antitumor immune responses have a particularly favorable clinical outcome.

We fully agree with van Rijnsoever et al that MSI has relevant implications for the selection of CRC patients to receive adjuvant chemotherapy. Nevertheless, the putative role of MSI as a predictor of response to chemotherapy is still controversial. 4,7,8 Our recent results do not seem to support such a generalized role, since the large majority (79.8%) of patients from our series did not receive any additional therapy besides radical surgery, suggesting that adjuvant treatment could be useless in cases showing both MSI-H and high numbers of activated cytotoxic lymphocytes. 6 On the other hand, adjuvant chemotherapy could be beneficial to those MSI-H cases (24% in our series of proximal CRC) showing no evidence of local antitumor immune responses. In this respect, it should be considered that, besides direct cytotoxic activity, 5FU-based regimens may also have immunomodulatory effects that could contribute to enhance the responsiveness of this subset of tumors. 9–10

As pointed out by van Rijnsoever et al, validation of the predictive value of MSI-H should require prospective clinical trials including adjuvant treatment and non-treatment arms. Nevertheless, due to the difficulties to include non-treatment arms in future trials, particularly in stage III CRC, as a first approach, we would favor the re-evaluation of large retrospective studies comprising both treated and non-treated arms, using standardized MSI assessment criteria, as well as defined tumor stage, patients’ characteristics, and adjuvant therapy. In this respect, we also strongly recommend considering the number of activated cytotoxic lymphocytes infiltrating CRC, which may allow a more precise assessment of the prognostic and, perhaps, the predictive value of MSI-H.