More than 40 years ago, 5-Fluorouracil (5-FU), was

Transcription

1 GASTROENTEROLOGY 2004;126: RAPID COMMUNICATIONS Use of 5-Fluorouracil and Survival in Patients With Microsatellite-Unstable Colorectal Cancer JOHN M. CARETHERS,*,,, E. JULIETA SMITH, CYNTHIA A. BEHLING, LANCHINH NGUYEN,* AKIHIRO TAJIMA,* RYAN T. DOCTOLERO, BETTY L. CABRERA, AJAY GOEL,* CHRISTIAN A. ARNOLD,* KATSUMI MIYAI,, and C. RICHARD BOLAND*, *Department of Medicine, University of California, San Diego, California; Cancer Center, University of California, San Diego, California; Veterans Administration Research Service, San Diego, California; Veterans Medical Research Foundation, San Diego, California; and Department of Pathology, University of California, San Diego, California Background & Aims: 5-Fluorouracil improves mortality in stage III colorectal cancer patients. In vitro studies suggest that microsatellite instability influences cell survival after 5-fluorouracil treatment. We investigated the survival influence of 5-fluorouracil in patients with microsatellite instability high tumors. Methods: We collected data and tumors on 204 consecutive stage II and III colorectal cancer patients from registries at the University of California and Veterans Administration hospitals in San Diego, California, from 1982 to Archival DNA was extracted, and microsatellite instability was assessed by National Cancer Institute recommended markers. Cox proportional hazard modeling was used to determine survival associations for microsatellite instability and 5-fluorouracil treatment status. Results: We identified 36 microsatellite instability high (17.6%) and 168 non microsatellite instability high tumors (82.4%). Microsatellite instability high tumors were significantly associated with proximal colon location, presence of mucin, and surrounding lymphoid reaction. Univariate and multivariate analyses showed no survival difference between microsatellite instability high and non microsatellite instability high groups (hazard ratio, 1.04; P 0.88). Dichotomized by use of 5-fluorouracil, there was increased risk of death in patients who received no adjuvant chemotherapy (hazard ratio, 2.02; P 0.02). However, the benefit of 5-fluorouracil was different between microsatellite instability high and non microsatellite instability high groups. Patients with non microsatellite instability high tumors who received 5-fluorouracil had better survival compared with patients who were not treated (P < 0.05). Conversely, patients with microsatellite instability high tumors who were treated with 5-fluorouracil had no survival difference compared with patients without treatment (P 0.52). Conclusions: There is improved survival in patients with non microsatellite instability high tumors after 5-fluorouracil based chemotherapy that does not extend to patients with microsatellite instability high tumors. The microsatellite instability status of a patient s colorectal cancer may indicate differences in 5-fluorouracil based chemosensitivity; this is consistent with in vitro studies. More than 40 years ago, 5-Fluorouracil (5-FU), was introduced, and it remains the mainstay of chemotherapeutic treatment of colorectal cancer. Almost all adjuvant chemotherapy for advanced-stage colorectal cancer involves 5-FU, typically in combination with levamisole or leucovorin. 1 In particular, 5-FU based chemotherapy improves survival in patients with stage III colon cancer 2 4 and with stage II and III rectal cancer. 1 Colorectal cancers develop as a consequence of genomic instability. One form, chromosomal instability, is the genetic reason for tumor formation in approximately 80% 85% of colorectal cancers. These colorectal cancers have allelic loss of certain tumor suppressor genes after an initial allele has been mutated (e.g., adenomatous polyposis coli or P53), as originally predicted by Fearon and Vogelstein. 5 A second form of genomic instability, called microsatellite instability (MSI), is involved in the genesis of 15% 20% of colorectal cancers. 6 8 MSI does not involve allelic loss of tumor suppressor genes 9 ; rather, these tumors have multiple errors in repetitive DNA sequences, termed microsatellites, throughout their genome. This results from a failure to edit errors made during DNA replication; this occurs when the DNA mismatch repair (MMR) system is defective. 10 In most sporadic MSI colorectal cancers, epigenetic hypermeth- Abbreviations used in this paper: 5-FU, 5-fluorouracil; MMR, mismatch repair; MSI, microsatellite instability; MSS, microsatellite stable; PCR, polymerase chain reaction by the American Gastroenterological Association /04/$30.00 doi: /j.gastro

2 February FU AND MSI COLORECTAL CANCER 395 ylation of the hmlh1 MMR gene occurs. 11,12 The hypermethylation silences gene transcription of hmlh1, inactivating the DNA MMR and allowing the occurrence of MSI. In addition to the genetic findings, a number of clinicopathologic features of MSI colorectal tumors segregate them from tumors with chromosomal instability. These include a correlation with the tumor s location in the proximal (right) colon and an inverse relation to allelic loss. 6 8 Additionally, MSI colorectal tumors tend to be diploid, tend to possess a mucinous histology, and have a lymphoid reaction surrounding the tumor. Chromosomally unstable colorectal tumors, for example, are aneuploid and have an overall poorer survival compared with MSI tumors. 13 Work with in vitro cell models of MSI has indicated a difference in response to chemotherapeutic agents that might be extended to the clinical situation. The alkylator N-methyl-N -nitro-n-nitrosoguanidine makes the principle adduct O 6 -methylguanine, which can be recognized by cells with a competent MMR (i.e., cells without MSI) to effect arrest at the G 2 phase of the cell cycle and prohibit growth. 14,15 Cells with MSI did not respond to the alkylation and continued to grow with active cell cycling. 15 Similarly, the nucleotide analogue 6-thioguanine, when incorporated into MMR-competent cells, arrested cells in the G 2 phase of the cell cycle, a feature absent in MSI cells incorporated with 6-thioguanine. 16 The MMR proteins recognize cis-platinum adducts. 17 Finally, in vitro work with 5-FU indicated that colon cancer cells with MSI might not respond readily to this agent. 18 Some 5-FU was incorporated into DNA, showing the possibility of recognition by the DNA MMR system. Cells that were proficient in MMR were killed by treatment with 5-FU, compared with the sparing of MSI colon cancer cells, as detected by 3 methodologies (cell enrichment assay, cell growth, and colony formation). 18 Additionally, re-expression of hmlh1 in an MSI cell line with hypermethylated hmlh1 overcame resistance to treatment with 5-FU. 19 These data led us to evaluate the efficacy of 5-FU in the treatment of patients with MSI colorectal cancers. Materials and Methods Patients and Specimens We collected the data of 204 consecutive patients with stage II and III colorectal cancers from the cancer registries at the University of California and Veterans Administration Medical Centers in San Diego, California, from 1982 to 1999, under institutional review board approval. Patients were included in the study only if pathologic material was available and if their chemotherapeutic regimen was known. Adjuvant chemotherapy was not administered routinely to patients at these institutions until after 1991, which allowed for the inclusion of patients before 1991 who did not receive chemotherapy. The outcome of each patient, including survival and other epidemiological data, was initially known to 1 investigator (L.N.), but this was withheld from investigators who performed the microsatellite analysis. The blinded epidemiological data and microsatellite analysis code were revealed only after the microsatellite analysis was complete, to avoid bias. Specimen slides were reviewed by a pathologist (C.A.B. or K.M.) who was not aware of the MSI status. Tumor location, patient age, patient sex, patient race, tumor stage, use of chemotherapy, and survival were derived from the surgical pathology report. Follow-up data were retrieved from the tumor registries from the respective institutions and confirmed by chart and autopsy report review, if available. Formalinfixed, paraffin-embedded pathology blocks from each of the patient s tumors were obtained from storage in the pathology departments of each institution, cut into 5- m slices, and placed onto slides for H&E staining and subsequent DNA extraction. DNA Extraction and Microsatellite Analysis DNA was extracted from formalin-fixed, paraffin-embedded tissues of each patient s colorectal tumor and the surrounding noncancerous tissue. Each area to be microdissected was identified on a reference H&E-stained slide. Unstained tissue slices adjacent to the reference slide were microdissected according to areas identified on the reference slide by using a surgical scalpel blade. The dissected specimen was deparaffinized in a microfuge tube with xylene, and the DNA was purified with ethanol and GeneReleaser (BioVentures, Murfeesboro, TN) according to the manufacturer s recommendations. Subsequently, the samples were treated with 200 g/ml of proteinase K (Sigma, St. Louis, MO) and incubated at 55 C for 5 hours. The proteinase K was destroyed by heating the sample to 95 C for 15 minutes, and the samples were immediately iced and stored for polymerase chain reaction (PCR) analysis. 20,21 For determination of MSI, we used the National Cancer Institute recommended panel of 5 microsatellite markers (BAT25, BAT26, D5S346, D2S123, and D17S250) to classify the tumor as MSI-high (associated with inactivation of DNA MMR) and MSI-low, or microsatellite stable (MSS), which are not associated with DNA MMR inactivation. 22 MSI-high was defined as 2 or more markers showing novel alleles compared with nontumor tissue, MSI-low was defined as 1 marker with a novel allele, and MSS was defined as no marker with novel alleles. Because of the similar features of MSI-low and MSS tumors, these 2 groups were included together in our cohort as non MSI-high tumors. One primer from each microsatellite marker pair was radio-end-labeled with 32 P in a reaction containing the primer, kinase buffer, T4 polynucleotide kinase, and 32 P-adenosine triphosphate. PCR was performed on the microdissected template DNA in a reaction containing pmol each of the end-labeled and cold primers, 0.25

3 396 CARETHERS ET AL. GASTROENTEROLOGY Vol. 126, No. 2 U of Taq DNA polymerase, 40 mol/l deoxynucleoside triphosphate stock solution, and a final concentration of mol/l magnesium. PCR products were denatured in 95% formamide and electrophoresed on a 6% polyacrylamide gel containing 7.5 mol/l urea. The gels were then dried and exposed by autoradiography to x-ray film or in a PhosphorImager (Molecular Dynamics, Sunnyvale CA). Statistical Analysis Statistical analyses were performed by the University of California, San Diego, Comprehensive Cancer Center Biostatistic Shared Resource. The following variables were assessed: age, sex, race, location of the tumor, stage, use of 5-FU therapy, follow-up time, and vital status (alive or dead). The location of the tumor was classified as right if the tumor was at or proximal to the splenic flexure. Left-sided tumors were classified as distal to the splenic flexure. Disease stage was classified at surgery. Vital status was classified as overall survival. Pathologic variables included the presence or absence of mucin in the tumor, the presence or absence of lymphoid cells surrounding the tumor, and the histological grade of the tumor (well, moderately, or poorly differentiated). Statistical analyses of these descriptive values were as follows: for continuous variables, the t test; for categorical values, the Fisher exact test; and for pathologic values, 2 analysis. Survival analyses were performed after patients were classified on the basis of their MSI status and whether they received 5-FU chemotherapy as part of their treatment. To examine differences in survival rates, a Cox proportional hazard function was used for both univariate and multivariate analyses, and Kaplan Meier curves were plotted. The Cox proportional hazard function allowed calculation of the relative hazard risk ratio. Univariate and multivariate survival distributions were compared with the use of the log-rank test. Multivariate analysis was used to examine the effects of MSI status (controlled for 5-FU treatment, age, and stage) and use of 5-FU (controlled for age and stage) on survival. Significance for all statistics the likelihood of a difference between groups was recorded if the P value was Results We identified 204 patients with colorectal cancer from the cancer registries at the University of California and Veterans Administration Hospital in San Diego from 1982 to 1999 for whom pathologic material was available and whose chemotherapeutic regimen was known. Thirty-six (17.6%) tumors from our cohort were MSIhigh, 43 (21.1%) tumors were MSI-low, and 125 (61.3%) lacked MSI (MSS). We compared patients with MSI-high tumors with those who had non MSI-high (MSI-low plus MSS) tumors. MSI-high tumors were associated with location in the right colon (P 0.001) and were associated with the tumor expressing mucin (P 0.04) and showing a lymphoid reaction around the tumor (P 0.03; Table 1). We found no significant differences between patients with MSI-high and non MSI-high tumors for 5-FU treatment, age, sex, race, tumor stage, follow-up period, vital status, or tumor grade (Table 1). In our cohort of 204 patients with colorectal cancer, there were 81 (39.7%) deaths over the mean follow-up time of 43.7 months. To assess the effect of MSI status on survival, we used a Cox proportional hazard model for univariate and multivariate analyses. We found no differences in overall survival on the basis of MSI status (P 0.99; Table 2). When comparing the MSI status in patients treated with 5-FU, we found no differences in survival (P 0.74). Similarly, we found no difference in survival when we compared the MSI status in patients who were not treated with 5-FU (P 0.998; Table 2). Kaplan Meier curves based on the MSI status of the patient s tumor were plotted from the Cox model (Figure 1). In multivariate analysis with adjustment for the significant variables disease stage and age, we found no differences in survival between the MSI-high and non MSI-high groups (P 0.88; Table 3). To assess the effect of the 5-FU treatment status on survival, we used the Cox proportional hazard model for analysis. We found a significant overall survival benefit among colorectal cancer patients who were treated with 5-FU based chemotherapy compared with those who were not thus treated (P 0.04; Table 2). When adjusted for disease stage and age in the multivariate analysis, 5-FU treatment remained significant (P 0.02; Table 3). When categorized by the MSI status and 5-FU treatment, there was a significant difference in survival. We found no difference in survival among patients whose tumor was MSI-high, irrespective of whether they received 5-FU based chemotherapy (P 0.52; Table 2). However, we found a significant difference in survival among patients who had non MSI-high tumors and who received 5-FU (P ; Table 2). Kaplan Meier curves based on 5-FU treatment were plotted from the Cox model (Figure 2). Hazard ratios were calculated from the Cox proportion hazard modeling. Three variables were determined to be significant in multivariate analysis: 5-FU treatment, disease stage, and age. In our cohort, the MSI status was not a significant factor for survival after Cox modeling and multivariate analysis with adjustment for disease stage, age, and 5-FU treatment. Patients not receiving 5-FU based chemotherapy are twice as likely to die compared with those who receive 5-FU (Table 3). Stage III patients have a 2.24-fold higher risk for death than stage II

4 February FU AND MSI COLORECTAL CANCER 397 Table 1. Patient Characteristics From our Study Characteristic All patients (n 204) Patients with MSI-high tumors (n 36) Patients with non MSI-high tumors (n 168) P value a Treatment, n (%) 5-FU 66 (32) 10 (28) 56 (33) 0.56 No 5-FU 138 (68) 26 (72) 112 (67) Age, yr (mean SD) Sex, n (%) Male 131 (64) 22 (61) 109 (65) 0.70 Female 73 (36) 14 (39) 59 (35) Race, n (%) Asian 12 (6) 2 (6) 10 (6) 0.71 Black 15 (7) 2 (6) 13 (8) Hispanic 21 (10) 4 (11) 17 (10) White 154 (76) 27 (75) 127 (76) Unknown 2 (1) 1 (3) 1 (1) Tumor location, n (%) Right 93 (46) 26 (72) 67 (40) Left 111 (54) 10 (28) 101 (60) Tumor stage, n (%) Stage II 105 (51) 22 (61) 83 (49) Stage III 97 (48) 14 (39) 83 (49) Unknown 2 (1) 0 (0) 2 (1) Follow-up (mean) (mo) Vital status, n (%) Alive 123 (60) 21 (58) 102 (61) 0.85 Dead 81 (40) 15 (42) 66 (39) Tumor grade, n (%) b Well 68 (42) 15 (45) 53 (41) 0.54 c Moderate 61 (38) 11 (33) 50 (39) Poor 32 (20) 7 (21) 25 (20) Mucinous tumor, n (%) b Present 60 (34) 15 (50) 45 (31) 0.04 Absent 117 (66) 15 (50) 102 (69) Lymphoid reaction, n (%) b Present 96 (68) 24 (86) 72 (64) 0.03 Absent 45 (32) 4 (14) 41 (36) a P values comparing patients between MSI-high tumors and non MSI-high tumors were calculated by t test for continuous variables and the Fisher exact test for categorical values. For pathologic variables, 2 analysis was performed. b Data pertaining to the pathologic variables (tumor grade, mucinous tumor, and lymphoid reaction) were not available for all 204 patients. c For tumor grade, analysis was performed comparing poor-plus moderate-grade tumors vs. well-differentiated tumors. Because of rounding of the percentages, not all percentage values total 100. patients. Older patients had an increased risk for death (relative risk of 1.03 per year of age) in our cohort. Discussion Here we show in our retrospective cohort that the type of genomic instability within a colorectal cancer might dictate how the patient responds to 5-FU based chemotherapy. Whereas patients whose tumors are non MSI-high (MSI-low plus MSS) derive a benefit from 5-FU based chemotherapy, with an improvement in survival on univariate analysis, patients with MSI-high tumors do not. This type of study could not be performed in a prospective manner because potentially valuable therapy would be withheld in some patients. It should be noted that MSI-low tumors are identified by their biochemical definition and behave biologically similarly to MSS tumors. The frequency varies in the literature, and our findings are similar to other published results. 23 Our cohort showed commonly found features that segregate to patients with MSI-high tumors. We found a significant association of MSI-high tumors with a proximal colon location and with the presence of mucin and lymphoid cells around the tumor; these findings are consistent with other reports. 6 8,22,24 However, we did not find significant associations with earlier staging in our cohort compared with other cohorts. This may be in part due to the inclusion of only stage II and II patients, because a statistical significance in survival among patients with MSI-high tumors was found over all stages in patients younger than 50 years of age. 13 Although we did not have pathologic information on all of the tumors, we

5 398 CARETHERS ET AL. GASTROENTEROLOGY Vol. 126, No. 2 Table 2. Univariate Analysis of Survival Among Patients According to MSI Status or Treatment With 5-FU Analysis No. of patients P value a Based on MSI status MSI-H overall Non MSI-H overall 168 MSI-H with 5-FU Non MSI-H with 5-FU 56 MSI-H without 5-FU Non MSI-H without 5-FU 112 Based on 5-FU status No 5-FU overall FU overall 66 No 5-FU with MSI-H FU with MSI-H 10 No 5-FU with non MSI-H b 5-FU with non MSI-H 56 MSI-H, microsatellite instability high. a P values were calculated between the paired groups and derived from the Cox proportion hazard model. Survival times were different between subjects. b P value is based on the log-rank test; P value based on the likelihood ratio test is did not find an association with a poorer tumor grade. We found no difference in vital status between patients with MSI-high and non MSI-high tumors overall, as some other cohorts have found. 25 This could be due to the overall follow-up duration and differences in follow-up times, because this study was retrospective. A study performed on 244 stage IV colorectal cancer patients found no difference in survival between MSI-high and non MSI-high patients who did not receive 5-FU chemotherapy. 26 In patients who received 5-FU, those with MSI-high tumors had improved survival compared with non MSI-high patients; this suggests better chemotherapeutic sensitivity for these stage IV patients. 26 This was not observed for our stage II and III patients. Another study of 43 colorectal cancer patients, of whom 7 had MSI-high tumors, suggested that the MSI-high patients had a better 5-FU response rate and improved survival compared with MSS patients. 27 However, in this study, there was no control group of MSI-high patients who did not receive 5-FU for comparison. Figure 1. Kaplan Meier plots of overall survival based on the MSI status of patient colorectal tumors. Cox proportional hazard analysis indicated no survival difference between patients with MSI-high tumors compared with non MSI-high tumors (top). Comparison of subgroups who received 5-FU adjuvant chemotherapy (bottom left) and who did not receive adjuvant chemotherapy (bottom right) showed no difference in survival.

6 February FU AND MSI COLORECTAL CANCER 399 Table 3. Multivariate Analysis Using Cox Proportional Hazard Modeling to Identify Significant Variables and Generate Hazard Ratios Analysis Hazard ratio for death 95% Confidence interval P value MSI status (MSI-high vs. non MSI-high) FU treatment (none vs. 5-FU) Stage (III vs. II) Age NOTE. The total number of patients is 202 in the model because of unknown stage in 2 patients. The impetus for our evaluation was the observation of resistance to 5-FU and other agents in some in vitro studies with colorectal cancer cells ,18,19 These studies show that basic laboratory findings may indeed have a clinical application. Additionally, these cell models may provide clues to the mechanism of resistance of MSI tumors to 5-FU. For instance, with concentrations similar to in vivo tissue concentrations of 5-FU, 10% of 5-FU is incorporated into the DNA of colorectal cancer cells, where the DNA MMR system could potentially recognize the altered nucleotide. 18 How the DNA MMR proteins recognize 5-FU is currently not known. Certainly the lack of MMR, typical of MSI-high colorectal tumors, might allow incorporated 5-FU to cause harmful effects to the DNA synthesis and replication, but with no recognition by the dysfunctional MMR system and no inhibition of cell growth. A competent MMR system may trigger a cell death program that might be additive or independent of the effect of 5-FU on RNA metabolism and thymidylate synthetase and might be operative in MSS colorectal tumors. Selection of patients with colorectal cancer for 5-FU treatment has been based on the stage of tumor rather than the biology of the tumor. 4 Although surgical stag- Figure 2. Kaplan Meier plots of overall survival based on the 5-FU treatment status of patients with colorectal tumors. Cox proportional hazard analysis indicated a difference in survival between patients who received 5-FU and those who did not (P 0.04; top). When further stratified by tumor MSI status, a survival difference was found only among patients with non MSI-high tumors (P ; bottom right), and this did not extend to patients with MSI-high tumors (P 0.52; bottom left).

7 400 CARETHERS ET AL. GASTROENTEROLOGY Vol. 126, No. 2 ing is very predictive of survival, our study indicates that the form of genomic instability within a patient s colorectal tumor has clinical implications with regard to treatment with 5-FU. The suggestion here is that patients with MSI-high colorectal tumors do not benefit from 5-FU based chemotherapy and thus could avoid the potential side effects of 5-FU (e.g., nausea, diarrhea, stomatitis, dermatitis, alopecia, and neurological symptoms) that are found in half of patients treated with 5-FU. 3 However, because our study was retrospective, we urge caution in implementing these findings in clinical practice until prospective studies using effective drugs such as the topoisomerase I inhibitor camptothecin or others can be compared with 5-FU based chemotherapy with similar findings. This recommendation is despite the fact that during the preparation and submission of this article, another retrospective study based on colon cancer treatment trials found similar results to ours. 28 In future studies, identification of the form of genomic instability will be paramount for interpretation of the results. Microsatellite testing of colorectal tumors will need to be more commonplace to implement any findings. In summary, our retrospective analysis indicates that 5-FU based chemotherapy does not improve survival among stage II and III patients with MSI-high colorectal tumors a result that was suggested from in vitro studies. Tailoring 5-FU treatment on the basis of the biology of a patient s tumor, rather than just the stage, may become more frequent as we come to understand the clinical differences among the genetic causes for colorectal cancer development. References 1. Boland CR, Sinicrope FA, Brenner DE, Carethers JM. Colorectal cancer prevention and treatment. Gastroenterology 2000;118: S115 S Laurie JA, Moertel CG, Fleming TR, Wieand HS, Leigh JE, Rubin J, McCormack GW, Gerstner JB, Krook JE, Mailliard J, Twito DI, Morton RF, Tschetter LK, Barlow JF. Surgical adjuvant therapy of large-bowel carcinoma: an evaluation of levamisole and the combination of levamisole and fluorouracil. J Clin Oncol 1989;7: Moertel CG, Fleming TR, Macdonald JS, Haller DG, Laurie JA, Goodman PJ, Underleider JS, Emerson WA, Tormey DC, Glick JH, Veeder MH, Mailliard JA. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 1990;322: Moertel CG, Fleming TR, Macdonald JS, Haller DG, Laurie JA, Tangen CM, Ungerleider JS, Emerson WA, Tormey DC, Glick JH, Veeder MH, Mailliard JA. Fluorouracil plus levamisole as effective adjuvant therapy after resection of stage III colon carcinoma: a final report. Ann Intern Med 1995;122: Fearon ER, Vogelstein B. A genetic model for colorectal tumorigenesis. Cell 1990;61: Ionov Y, Peinado MA, Malkhosyan S, Shibata D, Perucho M. Ubiquitous somatic mutations in simple repeated sequences reveal a new mechanism for colonic carcinogenesis. Nature 1993;363: Thibodeau SN, Bren G, Schaid D. Microsatellite instability in cancer of the proximal colon. Science 1993;260: Aaltonen LA, Peltomaki P, Leach FS, Sistonen P, Pylkkanen L, Mecklin JP, Jarvinen H, Powell SM, Jen J, Hamilton SR, Petersen GM, Kinzler KW, Vogelstein B, de la Chapelle A. Clues to the pathogenesis of familial colorectal cancer. Science 1993;260: Lengauer C, Kinzler KW, Vogelstein B. Genetic instability in colorectal cancers. Nature 1997;386: Carethers JM, Boland CR. Neoplasia of the gastrointestinal tract. In: Yamada T, Alpers DH, Kaplowitz N, Laine L, Owyang C, Powell DW, eds. Textbook of gastroenterology. Vol 1. 4th ed. Philadelphia: Lippincott-Raven, 2003: Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa J-PJ, Markowitz S, Willson JKV, Hamilton SR, Kinzler KW, Kane MF, Kolodner RD, Vogelstein B, Kunkel TA, Baylin SB. Incidence and functional consequences of hmlh1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci U S A 1998;95: Viegl ML, Kasturi L, Olechnowicz J, Ma AH, Lutterbaugh JD, Periyasamy S, Li G-M, Drummond J, Modrich PL, Sedwick WD, Markowitz SD. Biallelic inactivation of hmlh1 by epigenetic gene silencing, a novel mechanism causing human MSI cancers. Proc Natl Acad Sci U S A 1998;95: Gryfe R, Kim H, Hsieh ET, Aronson MD, Holowaty EJ, Bull SB, Redston M, Gallinger S. Tumor microsatellite instability and clinical outcome in young patients with colorectal cancer. N Engl J Med 2000;342: Koi M, Umar A, Chauhan DP, Cherian SP, Carethers JM, Kunkel TA, Boland CR. Human chromosome 3 corrects mismatch repair deficiency and microsatellite instability and reduces N-methyl-N nitro-n-nitrosoguanidine tolerance in colon tumor cells with homozygous hmlh1 mutation. Cancer Res 1994;54: Carethers JM, Hawn MT, Chauhan DP, Luce MC, Marra G, Koi M, Boland CR. Competency in mismatch repair prohibits clonal expansion of cancer cells treated with N-methyl-N -nitro-n-nitrosoguanidine. J Clin Invest 1996;98: Hawn MT, Umar A, Carethers JM, Marra G, Kunkel TA, Boland CR, Koi M. Evidence for a connection between the mismatch repair system and the G2 cell cycle checkpoint. Cancer Res 1995;55: Fink D, Nebel S, Aebi S, Zheng H, Cenni B, Nehme A, Christen RD, Howell SB. The role of DNA mismatch repair in platinum drug resistance. Cancer Res 1996;56: Carethers JM, Chauhan DP, Fink D, Nebel S, Bresalier RS, Howell SB, Boland CR. Mismatch repair proficiency and in vitro response to 5-fluorouracil. Gastroenterology 1999;117: Arnold CN, Goel A, Boland CR. Role of hmlh1 promoter hypermethylation in drug resistance to 5-fluorouracil in colorectal cancer cell lines. Int J Cancer 2003;106: Carethers JM, Hawn MT, Greenson JK, Hitchcock CL, Boland CR. Prognostic significance of allelic loss at chromosome 18q21 for stage II colorectal cancer. Gastroenterology 1998;114: Gervaz P, Cerottini JP, Bouzourene H, Hahnloser D, Doan CL, Benhattar J, Chaubert P, Secic M, Gillet M, Carethers JM. Comparison of microsatellite instability and chromosomal instability in predicting survival of patients with T3N0 colorectal cancer. Surgery 2002;131: Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR, Burt RW, Meltzer SJ, Rodriguez-Bigas MA, Fodde R, Ranzani GN, Srivastava S. The international workshop on microsatellite instability and RER phenotypes in cancer detection and familial predisposition. Cancer Res 1998;58: Goel A, Arnold CN, Niedzwiecki D, Chang DK, Ricciardiello L,

8 February FU AND MSI COLORECTAL CANCER 401 Carethers JM, Dowell JM, Wasserman L, Compton C, Mayer RJ, Bertagnolli MM, Boland CR. Characterization of sporadic colon cancer by patterns of genomic instability. Cancer Res 2003;63: Jass JR, Do KA, Simms LA, Iino H, Wynter C, Pillay SP, Searle J, Radford-Smith G, Young J, Leggett B. Morphology of sporadic colorectal cancer with DNA replication errors. Gut 1998;42: Watanabe T, Wu TT, Catalano PJ, Ueki T, Satriano R, Haller DG, Benson AB III, Hamilton SR. Molecular predictors of survival after adjuvant chemotherapy for colon cancer. N Engl J Med 2001; 344: 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, Chang K-J. 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 2002;101: Brueckl WM, Moesch C, Brabletz T, Koebnick C, Riedel C, Jung A, Merkel S, Schaber S, Boxberger F, Kirchner T, Hohenberger W, Hahn EG, Wein A. Relationship between microsatellite instability, response and survival in palliative patients with colorectal cancer undergoing first-line chemotherapy. Anticancer Res 2003;23: Ribic CM, Sargent DJ, Moore MJ, Thibodeau SN, French AJ, Goldberg RM, Hamilton SR, Laurent-Puig P, Gryfe R, Shepherd LE, Tu D, Redston M, Gallinger S. Tumor microsatellite-instability status as a predictor of benefit from fluorouracil-based adjuvant chemotherapy for colon cancer. N Engl J Med 2003;349: Received August 22, Accepted December 4, Address requests for reprints to: John M. Carethers, M.D., GI Section, 111D, VA San Diego Healthcare System, 3350 La Jolla Village Drive, San Diego, California jcarethers@ucsd.edu; fax: (858) Supported by the California Department of Health Services Cancer Research Program (grant ; to J.M.C.). This work was presented in abstract form at Digestive Diseases Week 2002 in San Francisco, California. Dr. Boland s current affiliation is Baylor University Medical Center, Dallas, Texas.