Comparative Efficacy of Adjuvant Chemotherapy in Patients With Dukes B Versus Dukes C Colon Cancer: Results From

Transcription

1 Comparative Efficacy of Adjuvant Chemotherapy in Patients With Dukes B Versus Dukes C Colon Cancer: Results From Four National Surgical Adjuvant Breast and Bowel Project Adjuvant Studies (C-01, C-02, C-03, and C-04) By Eleftherios Mamounas, Samuel Wieand, Norman Wolmark, Harry D. Bear, James N. Atkins, Kyunghee Song, Judy Jones, and Howard Rockette Purpose: Although the benefit from adjuvant chemotherapy has been clearly established in patients with Dukes C colon cancer, such benefit has been questioned in patients with Dukes B disease. To determine whether patients with Dukes B disease benefit from adjuvant chemotherapy and to evaluate the magnitude of the benefit, compared with that observed in Dukes C patients, we examined the relative efficacy of adjuvant chemotherapy according to Dukes stage in four sequential National Surgical Adjuvant Breast and Bowel Project trials (C-01, C-02, C-03, and C-04) that compared different adjuvant chemotherapy regimens with each other or with no adjuvant treatment. Patients and Methods: The four trials included Dukes B and C patients and were conducted between 1977 and The eligibility criteria and follow-up requirements were similar for all four trials. Protocol C-01 compared adjuvant semustine, vincristine, and fluorouracil (5-FU) (MOF regimen) with operation alone. Protocol C-02 compared the perioperative administration of a portal venous infusion of 5-FU with operation alone. Protocol C-03 compared adjuvant 5-FU and leucovorin (LV) with adjuvant MOF. Protocol C-04 compared adjuvant 5-FU and LV with 5-FU and levamisole (LEV) and with the combination of 5-FU, LV, and LEV. ALTHOUGH THE BENEFIT from adjuvant chemotherapy has been clearly established in Dukes C colon cancer patients, many oncologists still question the worth of such therapy in patients with Dukes B disease. In 1990, on the basis of information available at the time, a National Institutes of Health Consensus Development Conference on colorectal adjuvant therapy recommended that patients with stage III (Dukes C) colon cancer should receive adjuvant chemotherapy with fluorouracil (5-FU) and levamisole (LEV). 1 This recommendation was primarily based on results from intergroup study 0035, 2 which indicated a significant survival benefit from adjuvant 5-FU LEV in patients with Dukes C colon cancer. The same conference did not recommend any specific adjuvant therapy for patients with stage II (Dukes B) colon cancer outside of clinical trials. Subsequent updates of results from the intergroup 0035 trial 3,4 failed to demonstrate a survival benefit from adjuvant chemotherapy in Dukes B patients, although, in these patients, a reduction in recurrence comparable with that documented for Dukes C patients was also observed. Results: Forty-one percent of the patients included in these four trials had resected Dukes B tumors. In all four studies, the overall, disease-free, and recurrencefree survival improvement noted for all patients was evident in both Dukes B and Dukes C patients. When the relative efficacy of chemotherapy was examined, there was always an observed reduction in mortality, recurrence, or disease-free survival event, irrespective of Dukes stage, and in most instances, the reduction was as great or greater for Dukes B patients as for Dukes C patients. When data from all four trials were examined in a combined analysis, the mortality reduction was 30% for Dukes B patients versus 18% for Dukes C patients. The mortality reduction in Dukes B patients occurred irrespective of the presence or absence of adverse prognostic factors. Conclusion: Patients with Dukes B colon cancer benefit from adjuvant chemotherapy and should be presented with this treatment option. Regardless of the presence or absence of other clinical prognostic factors, Dukes B patients seem to benefit from chemotherapy administration. J Clin Oncol 17: by American Society of Clinical Oncology. Several possible reasons for the discrepancy in clinical benefit have been entertained. Because of the lower recurrence rate in Dukes B patients, most adjuvant clinical trials contain an insufficient number of such patients to be able to address with adequate statistical power the study s primary questions in the subpopulation of Dukes B patients. The relatively large number of noncancer deaths in this patient population further complicates the problem. From the National Surgical Adjuvant Breast and Bowel Project Operations and Biostatistical Centers, Pittsburgh, PA. Submitted May 4, 1998; accepted December 16, Supported by public health service grants from the National Cancer Institute (NCI-U10-CA-12027, NCI-U10-CA-37377, and NCI-U10- CA-39086) and by a grant from the American Cancer Society (ACS-R- 13). Address reprint requests to Norman Wolmark, MD, National Surgical Adjuvant Breast and Bowel Project, 4 Allegheny Center, 5th Floor, 320 E. North Ave, Pittsburgh, PA by American Society of Clinical Oncology X/99/ Journal of Clinical Oncology, Vol 17, No 5 (May), 1999: pp

2 1350 MAMOUNAS ET AL The National Surgical Adjuvant Breast and Bowel Project (NSABP) has included Dukes B and C colon cancer patients in four adjuvant chemotherapy trials for which outcome results are available. These four trials (C-01, C-02, C-03, and C-04) compared different adjuvant chemotherapy regimens with each other or with no adjuvant treatment. Overall, 41% of the patients accrued in these four trials had Dukes B tumors. In all four trials, a disease-free survival (DFS) and/or survival benefit from chemotherapy (reaching or approaching statistical significance) has been demonstrated at 5 years of follow-up between at least two treatment arms. To address the question of whether Dukes B colon cancer patients benefit from adjuvant chemotherapy and to determine the magnitude of such benefit compared with that observed in Dukes C patients, we examined the relative efficacy of adjuvant therapy in these four NSABP trials according to Dukes stage. Preliminary results were previously reported in abstract form 5 ; the current report is the first full analysis of our experience on the effect of adjuvant chemotherapy according to Dukes stage. Eligibility PATIENTS AND METHODS The four trials were conducted between 1977 and The eligibility criteria were generally similar for all four trials and have been described in detail in previous publications. 6-9 In summary, in C-01, C-03, and C-04, eligible patients had adenocarcinoma of the colon resected with curative intent with no evidence of gross residual or metastatic disease at the time of laparotomy. Patients with pathologically confirmed tumor extension into adjacent organs were eligible provided that all tumor was removed en bloc with negative resection margins. In C-02, in which randomization occurred before operative exploration, eligible patients were required to have a potentially curable adenocarcinoma, as documented by barium enema or endoscopic biopsy. Those patients having intraoperative extent of disease consistent with Dukes D tumors did not receive the randomized treatment and were treated at the discretion of the participating investigator. In all four trials, patients were classified as having Dukes B tumors if, on pathologic examination, the tumor demonstrated full-thickness penetration of the bowel wall (through the serosa or into the pericolic fat) with no regional lymph node involvement. 10 Patients were classified as having Dukes C tumors if, on pathologic examination, there was evidence of involvement of the regional lymph nodes. In all four trials, patients presenting with obstruction or contained perforation were eligible, but patients presenting with free perforation were not. Finally, in all four trials, eligible patients were required to have adequate hepatic or renal function and adequate WBC counts and platelet counts as well as an Eastern Cooperative Oncology Group performance status of 0, 1, or 2. Treatment and Patient Information Protocol C-01 6 compared adjuvant semustine, vincristine, and 5-FU (MOF regimen) with operation alone. In that protocol, a third arm tested the worth of postoperative administration of Bacille Calmette-Guérin. From 1977 to 1983, 773 patients were accrued to the operation and the MOF arms (Table 1). The MOF regimen consisted of eight 10-week cycles of semustine (130 mg/m 2 orally on day 1), vincristine (1 mg/m 2 intravenously [isqb] IV [rsqb] on days 1 and 36), and 5-FU (325 mg/m 2 IV on days 1 through 5 and 375 mg/m 2 IV on days 36 through 40). Protocol C-02 7 compared the perioperative administration of a portal venous infusion (PVI) of 5-FU with operation alone. From 1984 to 1988, 1,158 patients were accrued; of these, 718 patients had Dukes B and C lesions (Table 1), and 440 patients had Dukes A or Dukes D lesions and are not included in this analysis. The perioperative PVI of 5-FU was administered in a dose of 600 mg/m 2 along with 5,000 units of heparin as a continuous 24-hour infusion for 7 consecutive days. Protocol C-03 8 compared adjuvant 5-FU and leucovorin (LV) with adjuvant MOF. From 1987 to 1989, 1,081 patients were accrued (Table 1). The MOF regimen was identical to that used in C-01 but was given only for five cycles every 10 weeks. Each cycle of the 5-FU LV regimen consisted of LV (500 mg/m 2 given as a 2-hour IV infusion and repeated weekly for six doses) and 5-FU (500 mg/m 2 given as an IV bolus, 1 hour after the LV infusion and repeated weekly for six doses). A total of eight cycles were given. Protocol C-04 9 compared the same adjuvant 5-FU LV regimen with 5-FU LEV (as used in the intergroup adjuvant trials) and with the combination of 5-FU LV LEV. From 1989 to 1990, 1,434 patients were accrued in the 5-FU LV and the 5-FU LEV arms of the protocol (Table 1). The 5-FU LEV regimen consisted of 5-FU (450 mg/m 2 IV on days 1 through 5 and repeated on day 29 and weekly thereafter for 1 year) and LEV (50 mg orally tid for 3 days and repeated every 2 weeks for 1 year). Follow-Up Requirements During chemotherapy, patients were evaluated at intervals according to the specific treatment regimen. The follow-up requirements were similar for all four trials. During the first 2 years after surgery, investigators were required to submit patient follow-up forms every 3 months, reporting the results of a physical examination, complete blood Table 1. Stage Distribution in NSABP C-01, C-02, C-03, and C-04 All C-01 C-02 C-03 C-04 No. % MOF OP PVI OP 5-FU LV MOF 5-FU LV FU LEV Total 4, Ineligible Eligible/no F/U Unable to determine Dukes B or C* Included in analysis 3, Dukes B 1, Dukes C 2, Abbreviations: OP, operation; F/U, follow-up. *Full-thickness tumor penetration of the bowel wall.

3 CHEMOTHERAPY FOR DUKES B COLON CANCER 1351 cell count, and chemistry profile, including liver function tests; a chest x-ray and carcinoembryonic antigen levels were required every 6 months (for C-04, a chest x-ray was required yearly), and a barium enema and/or colonoscopy was required yearly. During the third through fifth years after surgery, a physical examination, including weight and performance status assessment as well as complete blood cell count, chemistry profile including liver function tests, chest x-ray, and carcinoembryonic antigen levels, was required every 6 months (for C-04, a chest x-ray was required yearly); a barium enema and/or colonoscopy was required yearly; after the fifth year postsurgery, the status of disease was reported on a yearly basis. Statistical Methods All P-values for comparisons of survival, disease-free survival, and relapse-free survival between treatments or between groups were derived from a stratified log-rank statistic and are two-sided. For the analyses from individual protocols, the stratification factors were those used in the original reports. When analyses used data from combined protocols, the protocol was included as a stratification factor. The term absolute improvement in survival refers to the difference in the 5-year survival rates obtained using the life-table method of Cutler and Ederer. 11 Estimates for the reduction in event rates were obtained from computing the cumulative odds. 12 In all analyses, only data from the first 5 years of follow-up were used. The 5-year interval was chosen in order to be consistent across protocols. One set of analyses presented in this paper used data from all of the trials for which a chemotherapy regimen was shown to have a benefit. The purpose of this analysis was to take advantage of the additional information available from multiple trials in order to obtain estimates of treatment effect that were sufficiently stable to determine whether there was a differential treatment effect in Dukes B versus Dukes C patients. The method used computed the estimated cumulative odds for treatment benefit in Dukes B patients and the estimated cumulative odds for treatment benefit in Dukes C patients and took the ratio of these estimates (estimated ratio for Dukes B patients divided by the estimated ratio for Dukes C patients) and the 95% confidence interval (CI) for this ratio. The CI was obtained by determining the 95% CI for the difference in the logarithms of the cumulative odds and then taking the exponential value of the interval end points. If the upper bound for the CI for the ratio was less than 1, it would indicate that the observed reduction in risk was greater in the Dukes B patients than in Dukes C patients, ie, there was more of a treatment effect in Dukes B patients than Dukes C patients. If the lower bound was greater than one, this would indicate that the observed reduction in risk was less in the Dukes B patients than in Dukes C patients, ie, there was a greater treatment effect in Dukes C patients than in Dukes B patients. If the CI contained 1, this would indicate that there was no significant difference in the treatment effect according to Dukes class. RESULTS Of the 4,006 Dukes B and C patients accrued in the four studies, 162 (4.0%) were found to be ineligible, and six eligible patients (0.1%) had no follow-up. In 18 patients (0.4%) with full-thickness tumor penetration of the bowel wall, the nodal status could not be determined. Thus, 3,820 patients (95.4%) were available for analysis. Of these, 1,565 patients (41%) were Dukes B, and 2,255 (59%) were Dukes C (Table 1). The patient and tumor characteristics for the 1,565 Dukes B and 2,255 Dukes C patients are provided in Table 2. The distribution of age, sex, and tumor location was wellbalanced between the Dukes B and C patient cohorts. Whereas 26% of the Dukes B population had high-risk characteristics (defined as the presence of obstruction, bowel perforation [contained], or extension of tumor into adjacent organs, only 5% of patients were in the latter category. In each of the four trials, the 5-year results demonstrated a difference in overall survival for all patients for at least two of the arms (Table 3). In C-01, the administration of the MOF regimen resulted in a 7% absolute improvement in Table 2. Patient and Tumor Characteristics in Dukes B and Dukes C Patients According to Treatment Arm Dukes B (%) Dukes C (%) All (n 1,565) Treatment 1 (n 793) Treatment 2 (n 772) All (n 2,255) Treatment 1 (n 1,131) Treatment 2 (n 1,124) Sex Male Female Age, years Location Right colon Left colon Sigmoid/rectosigmoid Multiple Unknown 1 1 High-risk characteristics: presence of obstruction, perforation, or extension to adjacent organs No Yes Unknown

4 1352 MAMOUNAS ET AL Table 3. 5-Year Overall Survival Results in NSABP C-01, C-02, C-03, and C-04, According to Stage of Disease All Dukes B Dukes C Study No. Survival (%) P No. Survival (%) P No. Survival (%) P C-01 Operation MOF C-02 Operation PVI C-03 MOF FU LV C-04 5-FU LEV FU LV survival over operation alone (P.07); in C-02, perioperative PVI of 5-FU resulted in a 7% absolute improvement in survival over operation alone (P.08); in C-03, the administration of 5-FU LV resulted in a 10% improvement in survival over MOF (P.0008); and in C-04, the administration of 5-FU LV resulted in a 5% absolute improvement in survival over 5-FU LEV (P.06). Similar differences were also observed for disease-free and recurrence-free survival (not shown). It should be noted that protocol C-02 was designed to use a one-sided test for the final conclusions, and this is reflected in all previous reports of this study; to maintain consistency across protocols, a two-sided test for P values was used in the current analysis. A DFS event was defined as consisting of tumor recurrence, second primary cancer, or death. The 5-year outcome results according to stage of disease indicated that in all four studies the observed difference in overall survival was in the same direction for Dukes B and Dukes C patients (Table 3). In C-01, the administration of MOF, compared with operation alone, resulted in a 3% absolute improvement in survival in Dukes B patients (P.73) and a 9% absolute improvement in survival in Dukes C patients (P.05). In C-02, there was a 12% improvement in survival for Dukes B patients (P.005) and a 2% improvement for Dukes C patients (P.81) with the perioperative PVI of 5-FU, compared with operation alone. In C-03, there was an 8% improvement in survival in Dukes B patients (P.03) and an 11% improvement in Dukes C patients (P.003) with 5-FU LV, compared with MOF. Finally, in C-04, there was a 4% improvement in survival in Dukes B patients (P.25) and a 4% improvement in Dukes C patients with 5-FU LV, compared with 5-FU LEV (P.21). Again, similar differences were also noted for recurrence-free survival and DFS (not shown). As is demonstrated in Fig 1, there was always an observed reduction in mortality, recurrence, or DFS event rate from chemotherapy, irrespective of Dukes stage, and in most cases, the reduction was as great or greater for Dukes B patients as for Dukes C patients. In C-01, the administration of MOF resulted in a 7% reduction in mortality for Dukes B patients, compared with a 26% reduction for Dukes C patients. In C-02, 7 days of perioperative PVI of 5-FU resulted in a 51% reduction in mortality for Dukes B patients, compared with a 4% reduction for Dukes C patients. In C-03, 5-FU LV compared with MOF resulted in a 53% reduction in mortality for Dukes B patients, compared with a 31% reduction for Dukes C patients. Finally, in C-04, 5-FU LV compared with 5-FU LEV resulted in a 21% reduction in mortality for Dukes B patients, compared with a 14% reduction for Dukes C patients. The results were very similar for recurrence or DFS event (Fig 1). The above results are presented to demonstrate that in all four studies the treatment effect was similar between Dukes B and Dukes C patients. Because there was a limited number of Dukes B and Dukes C patients in each of these trials, in any one trial individually one could not rule out with confidence a substantial difference in treatment effect according to Dukes stage. To address this specific question, we combined the data from these four trials into two treatment groups. Treatment 1 included the treatment groups from each trial with the inferior overall, disease-free, and recurrence-free survival for all patients (operation groups in C-01 and C-02, MOF group in C-03, and 5-FU LEV group in C-04). Treatment 2 included the treatment groups from each trial with the superior overall, disease-free, and recurrence-free survival for all patients (MOF group in C-01, perioperative PVI of 5-FU in C-02, 5-FU LV in

5 CHEMOTHERAPY FOR DUKES B COLON CANCER 1353 Fig 1. Reduction in cumulative odds of death, recurrence, and DFS event according to Dukes stage for the four clinical trials (with 95% CIs). C-03 and C-04) (Fig 2). There were no significant differences in patient and tumor characteristics between treatment 1 and treatment 2 and for Dukes B and Dukes C patients (Table 2). To estimate the differential effect of treatment according to Dukes stage, we calculated the cumulative odds of death in the better treatment group (treatment 2) relative to the poorer treatment group (treatment 1) for both Dukes B and Dukes C patients (Fig 3). The cumulative odds of death in the Dukes B patients was 0.70 (indicating that at any point during the 5 years of follow-up, a Dukes B patient receiving treatment 2 was estimated to be 0.70 times as likely to die as a Dukes B patient receiving treatment 1) and was 0.82 in Dukes C patients (indicating that a Dukes C patient receiving treatment 2 was estimated to be 0.82 times as likely to die as a Dukes C patient receiving treatment 1). It should be noted that a smaller value of the cumulative odds of death represents a greater benefit from treatment. The ratio of these cumulative odds was 0.86 (0.701/0.816) (95% CI, ). Finally, the mortality reduction was examined in Dukes B patients according to the presence or absence of clinical adverse prognostic factors. The effectiveness of adjuvant chemotherapy in Dukes B patients was evident whether patients presented with or without adverse prognostic factors (Fig 4). Twenty-six percent of the Dukes B cohort possessed high-risk characteristics. Patients who did not have a high-risk characteristic had a 32% reduction in mortality (cumulative odds, 0.68; 95% CI, ; P.01), whereas those with one or more high-risk characteristic had a 20% reduction in mortality (cumulative odds, 0.80; 95% CI, ; P.26). This reduction in mortality was translated into an absolute improvement in survival of 5% in each risk category (treatment 2, 87%, v treatment 1, 82%, in the low-risk category and treatment 2, 75%, v treatment 1, 70%, in the high-risk category). DISCUSSION Two major arguments are given by those who do not recommend administration of adjuvant chemotherapy in patients with Dukes B colon cancer. The first relates to the Fig 2. Combined-analysis treatment arms. Fig 3. Reduction in cumulative odds (treatment 2 v treatment 1) of death, recurrence, and DFS event according to Dukes stage in the combined analysis (with 95% CIs).

6 1354 MAMOUNAS ET AL Fig 4. Reduction in cumulative odds (treatment 2 v treatment 1) of death, recurrence, and DFS event according to the presence of high-risk characteristics in the combined analysis of Dukes B patients (with 95% CIs). relatively good prognosis of these patients after curative resection alone that could minimize any potential gains, particularly in light of the toxicity and cost of adjuvant chemotherapy. The second relates to the possibility of differential effectiveness of adjuvant chemotherapy between Dukes B and Dukes C colon cancer patients, because one cannot assume that the biology of tumors confined to the bowel wall is the same as that of a tumor that involves the regional nodes. These results, from a large, well-controlled population of Dukes B patients, contradict both of these arguments. The results demonstrate that the 5-year survival of patients with Dukes B colon cancer treated with surgery alone, at least those entered onto the NSABP trials, is such that effective adjuvant chemotherapy would be desirable. In other malignancies, such as node-negative breast cancer, adjuvant chemotherapy is widely administered in patients with recurrence rates lower than those observed in our trials for Dukes B colon cancer patients. This is mainly because available data have convinced the oncologic community of the benefit from adjuvant chemotherapy for node-negative breast cancer. Thus, the question of chemotherapy effectiveness in Dukes B patients becomes the more important one in considering adjuvant therapy for these patients. Again, the results from this analysis indicate that adjuvant chemotherapy is as effective in patients with Dukes B tumors as in those with Dukes C tumors. We believe that these results provide important supplemental information that was unavailable at the time of the last National Institutes of Health Consensus Development Conference on the role of adjuvant chemotherapy in colorectal cancer. We recognize the limitations in combining several randomized studies that were conducted in different time periods with changing standards of care; nevertheless, there are several factors that lend credibility to these results. The eligibility criteria were similar across the studies, and there was uniformity in follow-up procedures; there was balance in patient and tumor characteristics between the two treatment arms for both Dukes B and Dukes C patients. We excluded from the present analysis the Bacille Calmette-Guérin group in protocol C-01, because it does not represent chemotherapeutic intervention. We also excluded the 5-FU LV LEV group from protocol C-04, because it contained both LV and LEV, and it would be difficult to select a control group for comparison. This group has demonstrated disease-free and overall survival equivalent to that of the 5-FU LV group. When the mortality reduction observed with 5-FU LV LEV versus 5-FU LEV for all patients (13%) was examined according to Dukes stage, similar results were observed (23% reduction for Dukes B patients and 8% reduction for Dukes C patients). Since the initial presentation of our results in 1996, 5 two other studies have examined the relative efficacy of chemotherapy in Dukes B patients. The first study, 13 a metaanalysis of 4,000 patients in 10 studies evaluating the efficacy of short, continuous infusion of portal vein chemotherapy, demonstrated that the observed treatment benefit was present both in Dukes A/B patients as well as in Dukes C patients. The second study, 14,15 a pooled analysis of five randomized trials evaluating the efficacy of adjuvant 5-FU and folinic acid (LV) in patients with Dukes B colon cancer, demonstrated a modest improvement in event-free survival and overall survival. The 5-year event-free survival was 73% for the untreated control group versus 76% in the group receiving adjuvant 5-FU and folinic acid (hazards ratio, 0.83; 90% CI, ); the 5-year overall survival was 80% for the untreated control group versus 82% for the group receiving chemotherapy (hazards ratio, 0.86; 90% CI, ). Dukes B patients were compared only with untreated controls, and data on relative risk comparing Dukes B and C patients were not provided. Whereas the authors concluded that these differences were not significant, we believe that they are not inconsistent with our own results. Moertel et al 4 reported the results of an intergroup trial in which 318 Dukes B2 (stage II) patients were randomized to 5-FU LEV or observation only (INT-0035). At a follow-up time of 7 years, 5-FU LEV reduced the recurrence rate by 31%; this reduction was not statistically significant (P.10). It should be emphasized that this study, by design, was underpowered to detect reductions in recurrence of less than 50%. Thus, the results from our analysis and those of the intergroup trial are not discordant in terms of colon cancer recurrence. In the intergroup study, although there was no difference in overall survival, there was a nonsignificant 20% reduction in the rate of colon cancer related deaths in the group receiving 5-FU LEV. The lack of an overall survival benefit may have been due to the relatively high non cancer-related death rate in Dukes B patients. Some investigators who oppose the routine administration of adjuvant chemotherapy in all Dukes B patients agree that such therapy may be indicated in a subset of patients presenting with high-risk prognostic characteristics. Our

7 CHEMOTHERAPY FOR DUKES B COLON CANCER 1355 results indicated that the benefit of adjuvant therapy in Dukes B patients was not related to the presence or absence of high-risk characteristics. The 5-year survival for patients in the treatment 1 category who received less effective therapy and who did not possess high-risk characteristics was only 82%, underscoring our contention that a conservative approach in this group is unwarranted. In summary, our results indicate that patients with Dukes B colon cancer benefit from adjuvant chemotherapy to a similar extent as do those with Dukes C tumors. These results further demonstrate that the benefit from adjuvant chemotherapy seen in Dukes B patients is not confined to those patients at high risk for recurrence but extends to those patients with none of the clinical adverse prognostic factors, in whom the benefit is comparable. Furthermore, the results indicate that the prognosis of Dukes B patients with tumors demonstrating none of these known clinical adverse prognostic factors is not good enough to exclude consideration of systemic adjuvant chemotherapy. With the emergence of molecular and genetic prognostic markers such as 18q chromosomal deletion, 16 DNA mismatch repair gene mutations, 17,18 thymidylate synthase levels, 19 and p53 mutations, 20,21 it may become possible to identify subgroups of Dukes B patients with such a good prognosis that adjuvant chemotherapy can be avoided. However, until such biomarkers become validated in prospective studies, all Dukes B patients should be considered for adjuvant chemotherapy after discussion of the risk and benefit of such treatment. 1. NIH Consensus Conference: Adjuvant therapy for patients with colon and rectal cancer. JAMA 264: , Moertel CG, Fleming TR, Macdonald JS, et al: Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 322: , Moertel CG, Fleming TR, Macdonald JS, et al: Fluorouracil plus levamisole as effective adjuvant therapy after resection of stage III colon carcinoma: A final report. Ann Intern Med 122: , Moertel CG, Fleming TR, Macdonald JS, et al: Intergroup study of fluorouracil plus levamisole as adjuvant therapy for stage II/Dukes B2 colon cancer. J Clin Oncol 13: , Mamounas EP, Rockette H, Jones J, et al: Comparative efficacy of adjuvant chemotherapy in patients with Dukes B vs Dukes C colon cancer: Results from four NSABP adjuvant studies (C-01, C-02, C-03, C-04). Proc Am Soc Clin Oncol 15:205a, 1996 (abstr 461) 6. Wolmark N, Fisher B, Rockette H, et al: Postoperative adjuvant chemotherapy or BCG for colon cancer: Results from NSABP protocol C-01. J Natl Cancer Inst 80:30-36, Wolmark N, Rockette H, Wickerham DL, et al: Adjuvant therapy of Dukes A, B, and C adenocarcinoma of the colon with portal-vein fluorouracil hepatic infusion: Preliminary results of National Surgical Adjuvant Breast and Bowel Project protocol C-02. J Clin Oncol 8: , Wolmark N, Rockette H, Fisher B, et al: The benefit of leucovorinmodulated fluorouracil as postoperative adjuvant therapy for primary colon cancer: Results from National Surgical Adjuvant Breast and Bowel Project protocol C-03. J Clin Oncol 11: , Wolmark N, Rockette H, Mamounas EP, et al: The relative efficacy of 5-FU leucovorin (FU-LV), 5-FU levamisole (FU- LEV), and 5-FU leucovorin levamisole (FU-LV-LEV) in patients with Dukes B and C carcinoma of the colon. Proc Am Soc Clin Oncol 15:205a, 1996 (abstr 460) 10. Dukes CE: The classification of cancer of the rectum. J Pathol 35: , 1932 REFERENCES 11. Cutler SJ, Ederer F: Maximum utilization of the life table method in analyzing survival. J Chron Dis 8: , Mantel N, Haenszel W: Statistical aspects of the analysis of the data from retrospective studies of disease. J Natl Cancer Inst 22: , Portal vein chemotherapy for colorectal cancer: A meta-analysis of 4000 patients in 10 studies Liver Infusion Meta-Analysis Group. J Natl Cancer Inst 89: , Erlichman C, Marsoni S, Seitz JF, et al: Event free and overall survival is increased by FUFA in resected B colon cancer: A pooled analysis of five randomized trials (RCTS). Proc Am Soc Clin Oncol 16:280a, 1997 (abstr 991) 15. International Multicentre Pooled Analysis of B2 Colon Cancer Trials (IMPACT B2) Investigators: Efficacy of adjuvant fluorouracil and folinic acid in B2 colon cancer. J Clin Oncol 17: , Jen J, Kim H, Piantadosi S, et al: Allelic loss of chromosome 18q and prognosis in colorectal cancer. N Engl J Med 331: , Bronner CE, Baker SM, Morrison PT, et al: Mutation in the DNA mismatch repair gene homologue hmlh1 is associated with hereditary non-polyposis colon cancer. Nature 368: , Nicolaides NC, Papadopoulos N, Liu B, et al: Mutations in two PMS homologues in hereditary nonpolyposis colon cancer. Nature 371:75-80, Johnston PG, Fisher ER, Rockette HE, et al: The role of thymidylate synthase expression in prognosis and outcome of adjuvant chemotherapy in patients with rectal cancer. J Clin Oncol 12: , Bosari S, Viale G, Possi P, et al: Cytoplasmic accumulation of p53 protein: An independent prognostic indicator in colorectal adenocarcinomas. J Natl Cancer Inst 86: , Zeng Z-S, Sarkis AS, Zhang Z-F, et al: p53 nuclear overexpression: An independent predictor of survival in lymph node positive colorectal cancer patients. J Clin Oncol 12: , 1994