Cancer Cell Biology
High-frequency microsatellite instability predicts better chemosensitivity to high-dose 5-fluorouracil plus leucovorin chemotherapy for stage IV sporadic colorectal cancer after palliative bowel resection
Article first published online: 26 AUG 2002
Copyright © 2002 Wiley-Liss, Inc.
International Journal of Cancer
Volume 101, Issue 6, pages 519–525, 20 October 2002
How to Cite
Liang, J.-T., Huang, K.-C., Lai, H.-S., Lee, P.-H., Cheng, Y.-M., Hsu, H.-C., Cheng, A.-L., Hsu, C.-H., Yeh, K.-H., Wang, S.-M., Tang, C. and Chang, K.-J. (2002), High-frequency microsatellite instability predicts better chemosensitivity to high-dose 5-fluorouracil plus leucovorin chemotherapy for stage IV sporadic colorectal cancer after palliative bowel resection. Int. J. Cancer, 101: 519–525. doi: 10.1002/ijc.10643
- Issue published online: 17 SEP 2002
- Article first published online: 26 AUG 2002
- Manuscript Revised: 17 JUL 2002
- Manuscript Accepted: 17 JUL 2002
- Manuscript Received: 2 APR 2002
- National Science Council (Taiwan). Grant Number: NSC 90-2314-B-002-335
- microsatellite instability;
- high-dose 5-fluorouracil plus leucovorin chemotherapy;
- colorectal cancer
The influence of MSI on treatment outcome of colorectal cancers remains unclear and deserves further investigation. We recruited 244 patients with stage IV sporadic colorectal cancers for our study, based on appropriate eligibility criteria. Patients were nonrandomly allocated to 2 treatment groups of either with or without high-dose 5-FU plus leucovorin chemotherapy (HDFL, 5-FU 2,600 mg/m2 leucovorin 300 mg/m2 maximum 500 mg). Each treatment group was further divided into 2 subgroups according to high-frequency MSI (MSI-H) status. MSI-H was defined as the appearance of MSI in at least 2 of the 5 examined chromosomal loci (BAT-25, BAT-26, D5S346, D2S123, D17S250). We compared clinicopathologic parameters, p53 overexpression and overall survival between the groups. In addition, 4 subgroups were identified as follows: MSI-H+HDFL+, n = 35; MSI-H−HDFL+, n = 134; MSI-H+HDFL−, n = 17; MSI-H−HDFL−, n = 58. There was no significant difference of background clinicopathologic data between the HDFL+ and HDFL− treatment groups (p > 0.05). Survival analyses indicated that the patients of subgroup MSI-H+HDFL+ survived significantly longer than those of subgroup MSI-H−HDFL+, with median survival times of 24 (95% CI 20.2–27.9) and 13 (95% CI 11.6–14.4) months, respectively (p = 0.0001, log-rank test). In contrast, in patients without chemotherapy, the prognosis was poor irrespective of MSI status, with median survival times of 7.0 (95% CI 4.6–9.4) and 7.0 (95% CI 6.1–7.9) months in the MSI-H+HDFL− and MSI-H−HDFL− subgroups, respectively (p = 0.8205, log-rank test). MSI-H cancers responded significantly better to HDFL (p = 0.001), with a mean response rate of 65.71% (95% CI 49.98–81.44%) in subgroup MSI-H+HDFL+ compared to 35.07% (95% CI 26.99–43.15%) in subgroup MSI-H−HDFL+. There appeared to be no preferential metastatic site where response to HDFL can be predicted based on the MSI status of the primary tumor. Toxicity to HDFL was similarly minimal between MSI-H+ and MSI-H− patients (p > 0.05). Multivariate analysis of all patients further indicated that MSI-H and chemotherapy were independent favorable prognostic parameters (p < 0.05). Thus, the better prognosis of stage IV sporadic colorectal cancers with MSI-H may be associated with better chemosensitivity, rather than lower aggressiveness in biologic behavior. © 2002 Wiley-Liss, Inc.
MSI results from genetic mutation or epigenetic silencing of MMR genes.1 Approximately 15% of sporadic colorectal cancers harbored this genomic alteration in most studies.2 Numerous investigators have attempted to correlate this important molecular marker with clinical outcome of colorectal cancer. However, results to date have been inconclusive. Many researchers have advocated the association of MSI with better prognosis,3, 4, 5, 6, 7 but others have not found this relation.8, 9, 10 The inconsistency of the clinical relevance of MSI in different studies might result from different methodology and interpretation criteria used for the assessment of MSI,11 different clinical treatment modalities used for patients and variations in the clinicopathologic characteristics of the included patients, particularly with regard to pTNM, stage grouping and residual tumor (R) classification.12 Furthermore, biologically, tumor prognosis is determined by intrinsic aggressiveness and/or potential sensitivity to chemotherapy. However, currently, although some authors strongly advocate that colorectal cancers with MSI benefit from adjuvant chemotherapy,13, 14, 15, 16, 17, 18, 19 we are not fully convinced that this is due to better chemosensitivity and/or lower biologic invasiveness. In vitro studies have provided conflicting results as to whether colorectal cancer cells with deficient MMR genes are more resistant to 5-FU-based treatment.20, 21 Therefore, the clinical implications of MSI remain obscure and deserve further investigation. Clarification of the prognostic significance of MSI status will rely on the implementation of large, population-based studies and prospective clinical trials.22, 23, 24, 25 In our study, we determined the clinical relevance of MSI-H in stage IV sporadic colorectal cancer, based on a nonrandomized, prospective study. We explored whether MSI-H was associated with chemosensitivity and/or biologic aggressiveness in predicting the clinical outcome of stage IV sporadic colorectal cancer.
MATERIAL AND METHODS
Patients met the following eligibility criteria: (i) the sporadic primary colorectal cancer could be palliatively resected and pathologically confirmed as adenocarcinoma; (ii) metastatic lesions were measurable but unresectable; (iii) Karnofsky performance status was ≥50%; (iv) life expectancy was >12 weeks; (v) WBC count was ≥4,000/μl, platelet count was ≥100,000/μl, serum bilirubin was ≤2.0 mg/dl and serum glucose and electrolytes were normal. Patients with a family history of colorectal cancer suggestive of FAP or HNPCC were excluded. Moreover, patients with evident carcinosis peritonitis were excluded because their bowel function could not be restored through palliative surgery and their prognosis was very poor. Tumor location was classified as right-sided or left-sided colon cancer and rectal cancer. Right-sided colon cancer was defined as tumors proximal to the splenic flexure of the colon. Tumors at the splenic flexure, descending colon and sigmoid colon were classified as left-sided colon cancers. Informed consent was obtained from all patients. Details of the treatment protocol were explained to all patients. However, some patients rejected HDFL and favored only supportive care. Therefore, allocation of patients to treatment arms was nonrandomized. The MSI status of each patient was determined by analyses of stored specimens from a palliatively resected primary colorectal cancer. According to the National Cancer Institute criteria,11 the MSI status of tumors was classified into 3 categories: MSS (0 of the 5 examined loci showed MSI), MSI-L (MSI detected in 1 of the 5 examined loci) and MSI-H (MSI detected in 2 or more of the 5 examined loci). Tumors with MSS or MSI-L were together categorized as MSI-H−. Patients were then stratified by whether or not they were treated with HDFL and by MSI-H status, resulting in 4 subgroups: MSI-H+HDFL+, MSI-H−HDFL+, MSI-H+HDFL− and MSI-H−HDFL−. Clinicopathologic data and overall survival were recorded, analyzed and compared among the groups.
The weekly HDFL regimen consisted of 5-FU 2,600 mg/m2 and leucovorin 300 mg/m2 (maximum 500 mg) in a 24 hr i.v. infusion. 5-FU and leucovorin were mixed together to a final volume of 250 ml with 0.9% normal saline. An ambulatory Lifecare Provider 5500 infusion pump system (Abbott Laboratories, North Chicago, IL) was used to perform weekly 24 hr infusion via a single-lumen catheter in an outpatient setting. This method of drug administration has previously proven to be safe and without adverse precipitation.26, 27, 28 Chemotherapy was continued until objective evidence of disease progression or unacceptable toxicity developed. Weekly HDFL was discontinued temporarily when WBC ≤3,000/μl or platelet ≤100,000/μl and resumed at full dose if the hemogram returned to an acceptable level (WBC ≥3,000/μl, platelet ≥100,000/μl) within 1 week of the due day. If more than a 1-week delay was needed for the hemogram to return to normal, HDFL was decreased by 25% in the subsequent courses. Repeated 5-FU dose reduction was necessary if the above situation persisted in subsequent courses. When ECOG grade 3 or higher diarrhea or stomatitis developed, chemotherapy was discontinued temporarily.29 HDFL was then resumed with a prolonged interval after the diarrhea and stomatitis subsided (i.e., rest for 2 weeks after every 4 weekly HDFL treatments). All patients who received a minimum of 4 weeks of treatment were eligible for response evaluation.
Evaluation of response and toxicity
Complete blood count with WBC differential classification and a biochemical screening test were examined every 1–2 weeks. Chest X-ray and carcinoembryonic antigen level, if elevated, were studied every 4 weeks. Abdominal sonography and computed tomographic scanning were performed every 2–3 months. Whole-body bone scan and other necessary examinations (e.g., ascites or effusion cytology) were examined as indicated.
CR was defined as the disappearance of all objective evidence of disease, including all necessary imaging studies, lasting for >4 weeks. PR was defined as a decrease of >50% in the sum of the products of the diameters of all measurable lesions without evidence of new lesion(s), lasting for >4 weeks. Progressive disease was defined as an increase of >25% in the sum of the products of the diameters of all measurable lesions or the appearance of new lesion(s). All other patients were considered to have stable disease. The duration of follow-up was calculated from the date of entry to the cut-off date of the study. Duration of survival was the interval from the date of starting HDFL treatment to the date of death or last follow-up. For patients refusing HDFL, duration of survival was calculated from the date of the first postoperative visit to the outpatient clinic.
Treatment toxicity was recorded according to the ECOG criteria.29
Evaluation of MSI
Five microsatellite markers at distinct chromosomal loci (BAT-25, BAT-26, D5S346, D2S123, D17S250), as recommended by National Cancer Institute consensus on MSI,11 were used to determine the MSI status of all primary tumors (Fig. 1). All 5 sets of PCR primers were purchased from Research Genetics (Huntsville, AL). PCRs were performed on paired normal and tumor DNAs, as described previously.1, 2 All microsatellite markers were amplified with 35 cycles of PCR at 95°C for 30 sec, 60°C for 30 sec and 72°C for 30 sec. Products were electrophoresed on 6% polyacrylamide gels containing 8 M urea and then analyzed by autoradiography. Tumoral MSI for a given primer set was defined as the appearance of 1 or more PCR products either smaller or larger than those produced from normal mucosa (Fig. 1).
Immunocytochemic staining of P53
Specimens were fixed in 10% neutral formalin, processed and embedded in paraffin. Serial sections were prepared at 4 μm thickness for hematoxylin–eosin and immunocytochemic staining.30 The avidin-biotin-peroxidase technique was performed to demonstrate p53 using MAb DO1 (Signet, Dedham, MA). Peroxidase was visualized with 3′,3-diaminobenzidine tetrahydrochloride solution and counterstained by hematoxylin. Staining intensity was scored as follows: −, negative staining; ±, scattered positive cells (<10% of the specimen); +, intermediate positive staining (10–49% of the specimen); and ++, diffuse positive staining (≥50% of the specimen). Tumors with immunostaining of >10% of cells were considered positive for p53.
The prognostic significance of MSI-H, p53 overexpression and various clinicopathologic factors was evaluated by multivariate analysis using the Cox proportional hazards model. Background clinicopathologic variables of the 2 patient groups with or without HDFL were compared by χ2 or Fisher's exact test, as appropriate. Kaplan-Meier curves were constructed, with death as the primary end point. Differences in overall survival among subgroups were assessed by the log-rank test. p < 0.05 was considered statistically significant.
Between 1 January 1994 and 30 June 1999, a total of 244 patients were enrolled into our study, with n = 35 in group MSI-H+HDFL+, n = 134 in group MSI-H−HDFL+, n = 17 in group MSI-H+HDFL− and n = 58 in group MSI-H−HDFL−. All patients were followed up until October 2001. There was no significant difference of background clinicopathologic data between the HDFL+ and HDFL− groups (p > 0.05) (Table I). However, MSI-H tumors (n = 52, 21.3%) were significantly associated with young age at tumor onset (21.2% vs. 5.2%, p = 0.001), right-sided colon cancers (48.1% vs. 29.7%, p = 0.007), mucin production in histology (44.2% vs. 16.7%, p < 0.001) and normal p53 expression (84.6% vs. 28.1%, p < 0.001) compared to MSS or MSI-L tumors. Kaplan-Meier survival curves indicated that patients of group MSI-H+HDFL+ survived significantly longer (p = 0.0001, log-rank test) than those of group MSI-H−HDFL+, with median survival times of 24.0 (95% CI 20.2–27.9) and 13.0 (95% CI 11.6–14.4) months, respectively (Fig. 2). In contrast, there was no significant difference of survival (p = 0.8205, log-rank test) between groups MSI-H+HDFL− and MSI-H−HDFL−, with median survival times of 7.0 (95% CI 4.6–9.4) and 7.0 (95% CI 6.1–7.9) months, respectively. These findings implied that among patients with HDFL therapy the MSI-H+ group had better survival than the MSI-H− group. However, in patients without HDFL therapy, the prognosis was similarly poor regardless of MSI-H status. Thus, the prognostic significance of MSI-H for stage IV sporadic colorectal cancer may lie in its prediction of better chemosensitivity rather than the lower biologic aggressiveness. Better chemosensitivity in cancers with MSI-H was further demonstrated by direct evidence that the response rate to HDFL was significantly higher (p = 0.001) in group MSI-H+HDFL+ (mean 65.71%, 95% CI 49.98–81.44%) than in group MSI-H−HDFL+ (mean 35.07%, 95% CI 26.99–43.15%) (Table II). Multivariate analysis of all patients indicated that poor differentiation, mucin production, carcinoembryonic antigen level > 100 ng/ml and p53 overexpression were significant independent poor prognostic factors for survival (Table III). In contrast, MSI-H and chemotherapy were significant favorable prognostic factors (p < 0.05). Survival was not significantly affected by age, gender, tumor location, performance status, lymphatic/vascular permeation and the number of organs with metastasis (p > 0.05) (Table III). Remarkably, when the influence of true tumor burden on the chemotherapeutic effect was considered in the data analysis, i.e., when the MSI-H+HDFL+ and MSI-H−HDFL+ groups were further broken down into subcategories of liver metastasis only, lung metastasis only, liver plus lung metastasis and all other metastases and compared, there was no significant difference (p > 0.05) in response rate (Table II). Thus, there appeared to be no preferential metastatic site where response could be predicted based on the primary tumor MSI status. Additionally, toxicity to HDFL was minimal, and there was no correlation of chemotherapeutic toxicity with MSI-H status (p > 0.05) (Table II).
|Factor||HDFL − (n = 75)||HDFL + (n = 169)||p1|
|MSI-H+ (n = 17)||MSI-H− (n = 58)||Total||MSI-H+ (n = 35)||MSI-H− (n = 134)||Total|
|Performance status (Karnofsky)|
|Sites of measurable metastasis|
|Bone and spine||1||5||6||3||12||15|
|≥ 3 organs||3||5||8||4||12||16|
|Response1||MSI-H+ (n = 35)||MSI-H− (n = 134)||p|
|Total||Liver||Lung||Liver + Lung||Others||p||Total||Liver||Lung||Liver + Lung||Others||p|
|Overall response rate (95% CI)||65.71% (49.98–81.44%)||35.07% (26.99–43.15%)||0.001|
|ECOG grade of toxicity||1||2||3||4||1||2||3||4||NS|
|Factor||Number of patients||p||Hazard ratio||95% CI|
|Performance status (Karnofsky)|
|Sites of measuable metastases|
With the progress of molecular biology, more and more molecular targets of chemotherapeutic or chemopreventive agents have been recognized.24, 25, 31, 32 The clinical applicability of these molecular markers should be confirmed only through evidence-based medicine, i.e., population-based, well-controlled, prospective studies. MSI is one of the molecular markers whose prognostic significance has been intensively studied. However, to the best of our knowledge, there is no published report based on a well-controlled prospective study. The lack of objective data reflects that the implementation of clinical trials for molecular markers has been a challenging task. The success of our study was potentiated by the following 2 favorable factors. Firstly, it was not until 8–9 December 1997 that The International Workshop on MSI and RER Phenotypes in Cancer Detection and Familial Predisposition was held and a consensus regarding screening methodology and interpretation of MSI established.11 We took advantage of this consensus on MSI screening and classification. Secondly, the study population had stage IV colorectal cancer. It is well known that the mainstay treatment of stage IV colorectal cancer is chemotherapy and the role of surgery is palliation only. Namely, the biologic characteristics, rather than the surgical resection, determine overall survival. Therefore, the influence of surgery on survival was eliminated in our study. Moreover, the prognosis for stage IV colorectal cancer is generally very poor. Thus, follow-up is not long. However, one may argue that the impact of the study was reduced because the allocation of patients into treatment arms was not randomized.23 Actually, because HDFL is generally accepted as the standard treatment for patients with metastatic colorectal cancer,26, 28 randomization of patients was ethically impossible in our clinical setting. Nevertheless, all patients subjected to either treatment arm had the same eligibility criteria. Actually, there was no significant difference in various clinicopathologic factors between the 2 treatment arms: HDFL+ (n = 169) and HDFL− (n = 75). Moreover, our results parallel those of previous studies of HDFL in the treatment of metastatic colorectal cancer.27, 28, 33 Therefore, we believe that the potential bias of case selection has been minimized. However, comparisons of chemotherapeutic effect on the metastatic sites were based on the MSI status of primary colorectal cancers. It is generally accepted that there are sequential accumulations of genetic alterations during the evolution of a benign colorectal tumor to an invasive or metastatic cancer. Conceivably, the MSI status of metastatic lesions might be not the same as that of a primary colorectal cancer. Therefore, further correlation of MSI status between a primary tumor and the corresponding metastases should be done in a future study, to clarify the chemotherapeutic implications of MSI.
The mechanism underlying the association between MSI-H and better sensitivity to 5-FU-based treatment is thought-provoking. Theoretically, the chemosensitivity of colorectal cancer was determined by the magnitude of apoptotic signals induced by this chemotherapeutic agent, the expression level of TS34 and the cellular concentration of the 5-FU-metabolizing enzyme DPD.35 MSI has been recognized as a mutator phenotype of tumor cells.36 That is, tumor cells with MSI are predisposed to mutations in multiple genes, especially in those with repetitive DNA sequences. For example, colorectal cancers with MSI-H have been reported to harbor more frequent mutations of the β-catenin (CTNNB1) gene,37, 38 the transforming growth factor β receptor type II gene,39 the IGF-II receptor gene and the proapoptotic gene BAX.40 Moreover, MSI-H is also associated with the methylator phenotype, characterized by frequent hypermethylation of CpG sites within promoter regions and often resulting in transcriptional silencing.41 As indicated by the previous reports, simultaneous methylation of p161 and hMLH142, 43 was a frequent finding in colorectal cancers with MSI-H. Therefore, it is conceivable that the expression of genes related to the apoptotic pathways and the metabolism of 5-FU might be changed through the mechanisms of genomic instability and/or aberrant promoter methylation, thus contributing to the better chemosensitivity of MSI-H colorectal cancers. In this context, it would be of particular interest to look at the status of TS, DPD and even the apoptosis regulatory proteins in the metastases and to correlate this with MSI status and chemotherapeutic response. Actually, 2 reports are helpful in this respect. Villafranca et al.44 reported that TS repetitive-sequence polymorphisms are predictive of the effectiveness of preoperative 5-FU-based chemoradiation. Esteller et al.45 reported that the O6-methylguanine-DNA methyltransferase promoter in gliomas is a useful predictor of the responsiveness of tumors to alkylating agents. Remarkably, however, our previous report indicated that p53 overexpression predicts poor chemosensitivity to HDFL for stage IV colorectal cancer.33 In the present study, we further found that p53 overexpression and MSI-H were mutually exclusive. Therefore, it is probable that the better sensitivity to 5-FU-based treatment for MSI-H tumors found in the current study may be partly attributed to normal p53 expression. However, based on our data, both MSI-H+ and MSI-H− colorectal cancer patients appeared to benefit from HDFL chemotherapy. This implies that HDFL can also act on colorectal cancers through some other apoptotic pathways that are MSI-independent. Additionally, HDFL was an effective and safe regimen for stage IV colorectal cancer patients.27, 28, 33
Despite the encouraging explanations cited above, the better sensitivity of MSI-H tumors to 5-FU-based chemotherapy was not fully supported by the in vitro studies.46, 47 Actually, results have been inconclusive regarding the sensitivity of MSI+ cancer cell lines to various chemotherapeutic agents. Some have reported that MSI+ cells had increased vulnerability to camptothecin and halogenated thymidine sensitizers48, 49, 50 and appeared to retain their sensitivity to oxaliplatin, topotecan and irinotecan.51, 52, 53 In contrast, others have pointed out that cell lines with deficient MMR mechanisms were more resistant to many chemotherapeutic drugs, including the methylating agents procarbazine and temozolomide, the alkylating agent busulfan, the platinum-containing drugs cisplatin and carboplatin, the antimetabolite 6-thioguanine and the topoisomerase II inhibitors etoposide and doxorubicin.46, 47 It is well known that MMR proteins play a role in the recognition of DNA mismatch caused by various types of biosynthetic and photochemic damage. Therefore, MMR-deficient cells may fail to recognize DNA damage and subsequently fail to induce apoptosis.54 Reports comparing the sensitivity of MSI+ and MSI− colorectal cancer cell lines to 5-FU-based treatment remain scarce and appear to contradict each other. Aebi et al.20 used a cytotoxicity assay to evaluate the efficacy of 5-FU on 2 MSI+ cell lines and reported no reduction of effect. Carethers et al.21 used 1 of these 2 lines for an enrichment assay in the presence of 5-FU and found that MSI+ cells increased by 21% in comparison with MSI− cells after 5 days of incubation. Since HDFL for patients in our study was totally different from the 5-FU treatment of cell lines, more elaborate investigation and assessment are necessary to clarify the inconsistency between in vivo and in vitro results.
In conclusion, the better prognosis of stage IV sporadic colorectal cancers with MSI-H may be associated with better chemosensitivity. This finding could prove to be clinically applicable. However, its underlying mechanism remains to be further elucidated.
- 11A National Cancer Institute workshop on microsatellite instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res 1998; 58: 5248–57., , , , , , , , , , .
- 35Relationship between dihydropyrimidine dehydrogenase activity and plasma 5-fluorouracil levels with evidence for circadian variation of enzyme activity and plasma drug levels in cancer patients receiving 5-fluorouracil by protracted continuous infusion. Cancer Res 1990; 50: 197–201., , , .
- 40Genetic progression in microsatellite instability high (MSI-H) colon cancers correlates with clinicopathological parameters: a study of the TGFβRII, BAX, HMSH3, HMSH6, IGFIIR and BLM genes. Int J Cancer 2000; 89: 230–5., , , , , , , , , .
- 48Increased sensitivity to camptothecin in human colon carcinoma cell lines with mismatch repair deficiency [abstract]. Proc Am Assoc Cancer Res 1996; 37: 431., , , .
- 53Oxaliplatin (OxPt) effects on naked and intracellular DNA [abstract]. Proc Am Assoc Cancer Res 1997; 38: 311., , , .