|
Correction
to
Kemeny et al.,
|
| |||||||||||||||||||||||||||||||||||
Kemeny et al. state that the infusion pump was filled with "0.25 mg of floxuridine per kilogram of body weight per day for 14 days in combination with 20 mg of dexamethasone; 50,000 U of heparin, and enough normal saline to result in a volume of 50 ml." This does not mean that the dose delivered to patients was 0.25 mg per kilogram per day for 14 days. Because the average flow rate for these devices is approximately 2.0 ml per day, the dose of floxuridine that each patient received was about 0.14 mg per kilogram per day.
Given the risk of hepatobiliary toxicity and the narrow therapeutic index for this treatment,2 the dose of hepatic intraarterial floxuridine is a critical factor in the safe administration of regional chemotherapy. The dose of floxuridine should be based on the flow rate of the pump used to deliver the drug.
Alan P. Venook, M.D.
Betsy Althaus, Pharm.D.
Robert S. Warren, M.D.
University of California, San Francisco
San Francisco, CA 94143
References
More bothersome is the fact that although the treatment groups appear to have been well balanced with respect to the base-line characteristics enumerated by the authors, the difference in the survival rate at two years was significant only after adjustment by multivariate analysis with the use of the "best subgroup-selection method." The authors provide no reference for this method, but their description suggests a stepwise approach that may have diminished the value of the multivariate analysis.1 Their inclusion of some variables in the models for disease progression but not in the model for survival is counterintuitive. Certainly, the length of time since the diagnosis of primary cancer, the length of time since the last course of chemotherapy, and the lactate dehydrogenase level are all potentially important enough to be included in a multivariate model of survival. It is particularly disturbing that the location of the primary tumor, the size of the largest liver metastasis, the length of time since the diagnosis, and the length of time since the last course of chemotherapy were not reported as base-line characteristics. In addition, the authors do not mention any tests for the assumption of proportional hazards.2
Kemeny et al. suggest that hepatic arterial infusion of floxuridine improves the outcome for patients with colorectal cancer and resected liver metastases. However, the authors' data seem unconvincing. Even assuming the appropriateness of their multivariate adjustment, the difference in survival between the treatment groups appears to be small in relation to the morbidity and inconvenience associated with hepatic arterial infusion.
Carl D. Atkins, M.D.
242 Merrick Rd.
Rockville Centre, NY 11570
References
In our opinion, these results are partly biased by the heterogeneity of the patients enrolled in the study and by the difference in the duration of treatment. Since 21 patients (13 percent) had positive margins, the surgery they underwent was palliative.1 More than half the patients had previously received chemotherapy, and the median interval from resection of the primary tumor to the development of liver metastases was short (6.8 months in the combined-therapy group and 9.1 months in the monotherapy group). These two points constituted poor prognostic factors with respect to a response to intravenous chemotherapy and survival.2 On the other hand, hepatic arterial infusion has been recommended as an effective second-line therapy in patients with liver metastases that are resistant to systemic chemotherapy,3 and some of the patients enrolled in the recent study by Kemeny et al. were probably in this category.
We suggest that the improvement in survival at two years in the combined-therapy group, reported by Kemeny et al., may have been due to the advantage of hepatic arterial infusion in the subgroup of patients with positive margins. We understand that the data were analyzed on an intention-to-treat basis, but we would like to know either the rates of survival at two years for the subgroups of patients in whom surgery was palliative or curative or the results of a test of the interaction between treatment and positive or negative surgical margins in the multivariate analysis. The key point is whether there is a survival advantage with the use of adjuvant hepatic arterial infusion after curative resection. In addition, the duration of treatment was longer for the combined-therapy group than for the monotherapy group (35 weeks vs. 21 weeks), a difference that may have been particularly important for the subgroup of patients in whom surgery was palliative.
The initiation of postoperative chemotherapy was delayed in some patients, and others did not receive it because of surgical complications, refusal, or technical problems with hepatic arterial infusion. Therefore, the results of this interesting study should also prompt an investigation of the use of fluoropyrimidine-based preoperative chemotherapy, including new drugs such as irinotecan or oxaliplatin, in order to increase both the number of patients in whom subsequent surgery is curative and the overall survival rate.
Thierry Conroy, M.D.
Frédéric Marchal, M.D.
Centre Alexis Vautrin
54511 Vand
uvre-lès-Nancy, France
Serge Briançon, M.D.
University Hospital
54000 Nancy, France
References
To assess internal validity, we would like more information with regard to the Consolidated Standards of Reporting Trials (CONSORT) statement3; the method of masking treatment assignments and the details of the main analyses as indicated in the study protocol are particularly important. The projected sample size was based on survival rates at two years, rates of progression, and one-sided tests without adjustment for the censoring of data and for the use of two main end points. The use of two-sided tests in the final analyses seems to be more appropriate, but the authors did not adjust the sample size in order to achieve the desired statistical power when they changed the study protocol. The interpretation of the study findings as positive is based on the improvement in the results at two years in the combined-therapy group, although for the comparison of overall survival at two years (P=0.03), statistical significance was not actually demonstrated because there was no outline of the multiple-testing procedure.
The authors do not explain their rationale for selecting the main end points and the statistical methods used in their analyses. The differences in the event rates at two years are particularly important, if there is the possibility of a cure. However, the data shown in Figure 1 of the article do not support the hypothesis that additional treatment with hepatic arterial infusion has a curative potential. We were also unable to demonstrate a benefit in our study, even in the final, updated analyses according to the treatment received (Figure 1), despite a reduction in the recurrence of intrahepatic disease.1 In the analyses of end points that included the progression of disease, the data were censored for patients who died before the confirmation of progressive disease. Thus, the estimated rates of survival free of hepatic progression and of overall progression-free survival cannot be interpreted simply as a prognosis, but must instead be interpreted as a prognosis that is conditional on the patient's being alive. Whether progression-free survival, defined in this way, can be considered clinically relevant is questionable.
|
Matthias Lorenz, M.D.
Kathrin Hochmuth, M.D.
Johann Wolfgang Goethe University
60590 Frankfurt am Main, Germany
Hans-Helge Müller, Ph.D.
Philipps University of Marburg
35037 Marburg, Germany
References
To the Editor: In our study, the recommended dose of floxuridine was the total dose, and it was not adjusted for the flow rate. To make an adjustment for the flow rate, the dose should be 0.14 mg per kilogram multiplied by the pump volume and divided by the pump flow rate. We chose two-year end points because they were used in many other reports. The rate of survival at two years was significantly higher in the combined-therapy group than in the monotherapy group (86 percent vs. 72 percent, P=0.03). A study with many more patients than the 156 in our study would be necessary to show a significant increase in overall survival. The difference in the median period of survival (72.2 months in the combined-therapy group vs. 59.3 in the monotherapy group) may have clinical importance. The best subgroup-selection method is described by Furnival and Wilson.1 We chose the stepwise approach to pick the factors for the multivariate analysis because we believed that this approach would result in the selection of the best factors, which can differ in the models for overall and progression-free survival. With a multivariate analysis that included the potentially important variables suggested by Dr. Atkins, the risk ratio for death was 2.33 in the monotherapy group as compared with the combined-therapy group (P=0.028). We used Cox's proportional-hazards model for easy interpretation of the variables.
Conroy et al. suggest that we selected patients who had characteristics associated with resistance to systemic chemotherapy. If such patients can benefit from hepatic arterial infusion, then it is a useful treatment. The duration of treatment was longer for the combined-therapy group than for the monotherapy group. However, the patients in the combined-therapy group were treated every five weeks rather than every four weeks. We did not report the results of subgroup analyses of data at two years because we had too few patients to do so.
We believe hepatic arterial infusion with floxuridine is better than hepatic arterial infusion with fluorouracil plus leucovorin because of the higher hepatic extraction rate of floxuridine (minimizing toxicity elsewhere and allowing for combination with new agents such as irinotecan or oxaliplatin) and because of the poor results of the study by Lorenz et al.,2 in which only 30 percent of patients completed treatment.
As outlined in the CONSORT statement, all enrolled patients were eligible and were followed. As for the method of assignment to a treatment group, sealed envelopes were picked from boxes in the operating room; the envelopes were labeled according to the number of metastases (1, 2 through 4, or more than 4) after the surgeon had confirmed how many liver metastases were present.
A one-sided statistical test was initially chosen, because we believed the combined therapy would be better than systemic therapy alone. P values for one-sided tests in the final analyses would have made our already significant results even more pronounced. The study showed improved outcome at two years. No multiple-testing procedures were used. Data on a patient were censored if death occurred before the end point, without evidence of disease. Many patients would disagree with the comments of Lorenz et al. concerning the curative potential of treatment and would choose the treatment that offered a greater chance of being alive at two years, even if it were not curative.
We would also like to note that the first sentence of the Methods paragraph in the Abstract of our article should have referred to six cycles of similar systemic therapy, not six weeks.
Nancy Kemeny, M.D.
Ying Huang, Ph.D.
Yuman Fong, M.D.
Memorial Sloan-Kettering Cancer Center
New York, NY 10021
References
| |||||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
