Abstract
Background: Microsatellite Instability (MSI) status is a good prognostic factor for colorectal cancer (CRC) but its predictive value for chemosensitivity remains controversial. A previous meta-analysis (MA) in the adjuvant setting showed that MSI-high (H) status did not predict the efficacy of chemotherapy. The predictive value of MSI status on the effect of metastatic chemotherapy was investigated by MA. Patients and Methods: Studies were identified by electronic search through PubMed, Embase and ASCO proceedings online databases, using several key words (colorectal cancer, chemotherapy, microsatellite instability). For each study, the ratio of response rate (RR), complete (CR) and partial response (PR) divided by stable disease and progression was calculated. From 190 articles and 100 abstracts, only eight independent studies were selected. The data were analysed with a random-effect model (due to heterogeneity between studies) using EasyMA software. Statistical calculations were performed on six studies representing 964 patients (mean age 63 years; 91 MSI-H; 873 microsatellite stable (MSS) tumours). A total of 287 patients received 5-fluorouracil (5FU)-based chemotherapy, whereas 678 patients received combinations of 5FU or capecitabine with oxaliplatin and/or irinotecan. Results: No benefit of metastatic chemotherapy in terms of RR for MSI-H patients compared with MSS patients was found. The global hazard ratio (HR) for RR was 0.82 (95% confidence interval, CI: 0.95; 0.65-1.03; p=0.09). Conclusion: MSI status does not predict the effect of chemotherapy which is similar in MSI-H and MSS metastatic CRC tumours.
Colorectal cancer (CRC) is the third most common cancer and the fourth most frequent cause of cancer death worldwide (1). Chemotherapy is obviously the cornerstone of treatment of metastatic CRC. In metastatic disease, 5-fluorouracil (5FU) is usually combined with oxaliplatin or irinotecan. However, to improve management of CRC and to better select those patients who will respond to chemotherapy, predictive factors are urgently needed.
Among the genetic abnormalities involved in carcinogenesis, microsatellite instability (MSI) is a major pathway of cancer development (2). MSI corresponds to a dysfunction of the mismatch repair (MMR) system resulting in a reduction in the length of highly repeated DNA sequences termed microsatellites. The MMR system corrects mismatches of DNA nucleotides occurring during replication. A germinal mutation inactivating one of the MMR genes (MLH1, MSH2, MSH6, PMS1 and PMS2) (3) may lead to a hereditary form termed hereditary non-polyposis colon cancer (HNPCC). A previous meta-analysis (MA) found that MSI status was a prognostic factor in CRC (4). However the relative benefits of chemotherapy (meaning predictive value) among MSI heterogeneous (MSI-H) and microsatellite stable (MSS) patients were not assessed. The issue of the comparative benefits of chemotherapy according to MSI status is crucial since no biological marker has so far been proved effective in predicting response to chemotherapy (5). This is of importance since MSI is not only found in the rare HNPCC forms (Lynch syndrome), but also in 15% of sporadic forms of CRC (6).
Several studies have investigated the relationship between MSI status and chemotherapy efficacy in metastatic CRC, i.e. between MSI status and response rate (RR), relapse-free survival (RFS) or overall survival (OS). The relationship between MSI status and the efficacy of chemotherapy differed among the studies. Therefore a MA of all published studies was performed to determine the relevance of MSI status as a predictive factor of chemotherapy efficacy. The main goal of the MA was to assess the predictive value of MSI-H compared to MSS status among patients receiving chemotherapy for metastatic CRC.
Patients and Methods
Publication selection. The MA was performed according to a predefined written protocol. To be eligible, the studies had to deal with colon or rectum cancer and to assess the relationships between MSI, chemotherapy and RR for metastatic patients. Studies (full articles) were identified by an electronic search using online PubMed, with a set of key words “colorectal cancer, chemotherapy, microsatellite instability” used simultaneously. The last query was updated on November 9th, 2008. An electronic search with the same key words was also performed using online EMBASE. Examination of the Cochrane database of systematic reviews did not retrieve additional pertinent references. Our initial selection of articles relied on careful reading of their abstracts. Abstracts concerning microsatellite instability were also retrieved from the ASCO (American Society of Clinical Oncology) proceedings of the annual meetings from 1998 to 2008 available online. References from the relevant literature were also screened, including all of the identified studies, but also reviews and editorials for additional information (4). Studies written in English or French were included as full papers or as abstracts from the ASCO proceedings were included. The duplication of data was carefully avoided, by examining, for each publication, the names of all the authors and the different medical centres involved. Studies in which survival data were not available were excluded. In some cases, the authors were contacted to obtain additional data allowing statistical calculations.
Methodological assessment. Information was carefully extracted from all the publications in duplicate by two readers (Dr Gaëtan Des Guetz and Dr Bernard Uzzan), using a standardised data collection form, including the following items: complete reference of the publication; original publication or update of a former publication; mode of making up of the series of cases; prospective study; inclusion of consecutive cases; randomised controlled trials (RCT); median duration of follow-up; number of patients included in the study; mean or median age; gender; anticancer treatment(s) during follow-up; histological type (adenocarcinoma or mucinous); tumour size; stage of disease; grade (good, moderate or poor differentiation) and nodal status. Assessment of the methods used to determine MSI status, immunohistochemistry of MLH1, MSH2 (mutL homolog 1, mutS homolog 2) or the molecular methods (number and type of MSI; i.e. NCI (National Cancer Institute) reference or pentaplex panels (7, 8) was conducted by Dr Olivier Schischmanoff. To simplify the analysis of the data, MSI-low (L) (9) and MSS patients were always pooled.
The choice of drugs in the metastatic setting was broad but was exclusively based on 5FU up to 2004. The RR to chemotherapy and RFS and OS were chosen as evaluation criteria in the studies. Due to the lack of results available for OS and RFS, only the RR ratio was analysed for comparing the MSI-H and MSS patients. To summarise the information provided by the metastatic studies, the complete responders (CR) and partial responders (PR) were considered as responders and stable or progressive disease as non-responders.
Disagreements were resolved by consensus between the three readers. No predefined minimal number of patients was not set for a study to be included in the MA, nor a minimal duration of median follow-up were set. Individual studies were not weighed by a quality score, because no such score has received general agreement for use in a MA, especially of observational studies, making more difficult the evaluation of its usefulness (10). When duplicate studies were retrieved, only the study involving the highest number of patients from which data could be extracted (usually the latest) was included in the systematic review. This was done to avoid overlapping between cohorts. In each study, the MSI determination was performed retrospectively. The studies only included patients receiving chemotherapy in the metastatic setting. Although their methodological quality was rather similar, their design (phase II/III) and the reliability of their conclusions were variable.
Statistical methods. In each study, the relationship between MSI status and response rate was considered significant when the p-value for the statistical test comparing survival distributions between the MSI and MSS groups was below 0.05 in univariate analysis (two-tailed test).
For each trial, the hazard ratio (HR) was estimated by a method depending on the data provided in the publication. The simplest method consisted of the direct collection of HRs, or odds ratios, and their 95% confidence interval (CI) from the original article. If not available, the total numbers of responses and the numbers of patients at risk in each group were used to determine the RR ratio estimate.
Thus, a pooled HR estimate and its 95% confidence interval were calculated by using a random-effect model (Der Simonian and Laird method) due to heterogeneity between the studies. By convention, a HR lower than 1 for MSI-H compared with MSS corresponded to a better survival. The statistical calculations for the MA were performed with EasyMA.net, (http://www.spc.univ-lyon1.fr/easyma.net/) online available application (Department of Clinical Pharmacology, Cardiology Hospital, Lyons, France). The statistical analysis was performed by Dr Patrick Nicolas, who chose the best statistical exploitation of the data, or decided on the exclusion of studies from the MA because their data could not be exploited statistically.
Results
The electronic data search using online PubMed retrieved a total of 190 references including 58 reviews. The EMBASE query did not provide any additional references. After exclusion of the references which were out of the scope of this MA (by reading the abstracts), there remained 7 studies dealing with MSI status (11-17). No study was published in duplicate. One hundred abstracts concerning microsatellite instability were found in the ASCO proceedings (1998-2008). For chemotherapy of metastatic CRCs, additional data corresponding to only one recent abstract were retrieved (18). This study could be included in the MA after the authors of this article responded to some enquiries (by e-mail) (18). Two articles, the latter provided by screening of recent literature on biomarkers, were excluded due to inability by the authors themselves to obtain HR for the treated patients and MSI status (17, 19). Two other articles were excluded due to the MSI methods immunohistochemistry without molecular techniques (13, 20). Information about the inclusion process is provided in the flow-chart of the MA (Figure 1).
The number of microsatellite markers analysed differed greatly among the studies (from 5 to 10). However, the Bethesda markers were used in four studies (11, 14-16). Four studies included immunohistochemistry of MMR proteins (one MLH1 and MSH2, and two MLH1, MSH2, MSH6, PMS2 and one MLH1, MSH2 and MSH6 (14, 15,18) (11) (Table I).
Flow-chart of the different steps of the meta-analysis (MA).
The MA was performed for six studies in the metastatic setting (11, 12, 14-16, 18). The patients had been treated with different schedules of newly marketed treatments now considered FOLFIRI (combination of folinic acid, 5FU and irinotecan) (14), FOLFOX or FUFOX (combination of folinic acid, 5FU and oxaliplatin) (11, 12), XELOX (combination of capecitabine and oxaliplatin) (11, 18) or only classical 5FU (15,16). The global number of metastatic patients was 964 with a mean age of 63 years and 91 MSI-H and 873 MSS tumours. Due to the lack of results available for RFS or OS, only the RR ratio was analysed, comparing MSI-H with MSS patients. A major statistical heterogeneity was found between the studies using a fixed effect model making it unsuitable. The RR ratio was not statistically different from one when the random effect model was used. No benefit of chemotherapy in terms of the RR was observed for the MSI-H compared with the MSS patients: 0.82 (CI: 0.95; 0.65-1.03; p=0.09) (Figure 2).
Discussion
This MA on metastatic chemotherapy showed no difference in RR ratios corresponding to MSI-H and MSS, but found a major statistically significant heterogeneity, which might be explained by the difference in chemotherapy regimens between and even within the studies and by the large differences in the numbers of patients included in the studies. For instance Liang et al. used high doses of 5FU (16) whereas Koopman et al. pooled different regimens of chemotherapy with capecitabine, irinotecan and XELOX (18). A MA based on individual data and stratified according to the chemotherapy protocol would be time consuming and might not help to clarify this issue since multiplication of subgroup analyses (e.g. according to the dose of 5FU, combination with or without folinic acid) would limit their statistical power.
Main characteristics of the studies included in the meta-analysis.
The studies included in the present MA included RCT aiming to evaluate the best treatment in the metastatic setting considering doses of 5FU (15, 16), combination of chemotherapy (11, 18) and small phase II studies (12, 14). The stratification according to MSI status was always completed retrospectively, leading to various biases. A MA could eventually permit an overview of the publications and an analysis of these differences and similarities.
MSI-H status is a relatively rare biological event, with frequencies of 5% for genetic abnormalities (corresponding to HNPCC syndrome) and 15% for sporadic tumours. The evaluation of MSI-H as a predictive factor in the metastatic setting is particularly difficult because metastatic progression is a rare event for MSI-H tumours (21, 22). The addition of many patients from various studies was deemed useful to answer the issue of chemotherapy treatment benefit for MSI-H patients . In the present MA more than 10% MSI-H patients were found (91 MSI-H patients/964) and for the first time metastatic treatment was compared for many patients (the median number of patients with MSI-H tumours in individual studies is nine). Popat et al. previously found only two references (15, 16) in a metastatic setting and it was not possible to draw conclusions and their study mainly dealt with the prognostic use of MSI status and not with its predictive value on the effect of chemotherapy. Four additional studies (11, 12, 14, 18), were included in the present MA.
At present, combined chemotherapy has replaced 5FU monotherapy both in adjuvant (FOLFOX) and metastatic settings (FOLFOX, FOLFIRI). However, until now, no prospective study randomised chemotherapy according to MSI status and compared its efficacy among MSI-H and MSS patients. A major ongoing RCT, in the United-States, including more than 3000 patients (stage II, colon cancer) chose to randomise MSS patients between adjuvant chemotherapy alone and with bevacizumab, whereas the MSI-H patients will have no chemotherapy (23).
Since the study of Ribic et al. published in 2003, updated in 2008 at the last ASCO meeting, a general opinion was that MSI-H patients were not sensitive to adjuvant chemotherapy and moreover, chemotherapy could have by itself a detrimental effect on survival (24) (25). In the adjuvant setting, chemotherapy was essentially based on 5FU alone when the studies were began. Our previous MA studied for the first time the value of MSI status to predict the response to chemotherapy and we confirmed that MSI-H patients had a better overall and relapse-free survival than MSS patients in the adjuvant setting but found that MSI status did not predict the efficacy of adjuvant chemotherapy (26). In the present MA no difference was found in the metastatic setting. These differences show the difficulties of comparing adjuvant and metastatic settings especially considering the heterogeneity of the disease biology and types of treatments. In the metastatic setting, RR is the direct effect of chemosensitivity, with variation of tumour size according to treatment efficacy. However, in the present MA, the different regimens used in the different studies did not allow to evaluate the relative benefit of each drug. Now target therapies (such as anti-EGFR; or anti-angiogenic) are indicated in the metastatic setting either alone or more frequently in combination with chemotherapy. Ki-ras testing seems particularly useful for anti-EGFR antibodies and for anti-VEGF treatments, many studies are performed to determine predictive markers, but for chemotherapy, molecular markers are always needed.
Forest plot of the studies assessing Response Rate ratio in MSI-H metastatic patients compared with MSS patients. By convention, a Hazard Ratio <1 corresponded to a better Response rate for MSI-H patients. Microsatellite Instability high (MSI-H), MSS: microsatellite stable, HR (hazard ratio) (confidence interval, CI).
To conclude chemotherapy is not associated with a significantly improved response rate in MSI-H compared with MSS patients, in metastatic CRC. MSI-H status does not seem to be a good predictive marker for evaluating chemosensitivity in the metastatic setting
- Received December 10, 2008.
- Revision received January 25, 2009.
- Accepted February 2, 2009.
- Copyright© 2009 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- Microsatellite instability and sensitivity to fluoropyrimidine and oxaliplatin containing first-line chemotherapy in metastatic colorectal cancer
- TAS-102 and the quest for predictive biomarkers
- Molecular Dissection of Microsatellite Instable Colorectal Cancer
- Clinical Effects of Adjuvant Active Specific Immunotherapy Differ between Patients with Microsatellite-Stable and Microsatellite-Instable Colon Cancer