|
| |||||||||||||||||||||||||||||||
Background Moderate increases in the initial doses of certain chemotherapeutic drugs, such as cisplatin and cyclophosphamide, may prolong overall survival in patients with limited small-cell lung cancer.
Methods We conducted a prospective study of 105 patients with limited small-cell lung cancer. The patients were randomly assigned to receive higher or lower initial doses of cisplatin (100 or 80 mg per square meter of body-surface area) and cyclophosphamide (300 or 225 mg per square meter daily for four days); all patients received the same doses of doxorubicin and etoposide. The first course of chemotherapy was followed by five additional courses and by three courses of radiotherapy. All patients received the lower doses of cisplatin and cyclophosphamide and the same doses of doxorubicin and etoposide from the second through the sixth cycle of chemotherapy.
Results The median follow-up was 33 months. The two-year survival rate for the 55 patients who received the higher doses of chemotherapy was 43 percent, as compared with 26 percent for the 50 patients who received the lower doses (P = 0.02). The rates of complete response at six months were 67 percent in the higher-dose group and 54 percent in the lower-dose group (P = 0.16). Disease-free survival at two years was 28 percent in the higher-dose group, as compared with 8 percent in the lower-dose group (P = 0.02). Side effects from treatment were not increased in the higher-dose group.
Conclusions Higher initial doses of cyclophosphamide and cisplatin improve disease-free and overall survival in patients with limited small-cell lung cancer.
Thoracic radiotherapy has generally been combined with chemotherapy to improve local control, but combined therapy has afforded only a moderate survival benefit10. In our centers, all patients with limited small-cell lung cancer (localized in only one hemithorax, the mediastinum, and the supraclavicular nodes) have been treated since 1982 according to one of three consecutive protocols in which thoracic radiotherapy was alternated with a chemotherapeutic regimen including doxorubicin, cyclophosphamide, etoposide, and cisplatin1. A retrospective analysis of prognostic factors in the first 52 patients enrolled suggested that higher initial doses of cyclophosphamide and cisplatin reduced the frequency of distant metastases and increased overall survival rates4. These results were corroborated by a larger retrospective analysis (131 consecutive patients) showing that a 20 percent increase in the initial doses of cyclophosphamide and cisplatin resulted in a survival rate that was as much as 20 percent higher at two years11. We believed that this difference in survival justified a randomized, prospective study to test the hypothesis that a moderate increase in the initial doses of cisplatin and cyclophosphamide can improve overall survival.
Methods
Trial Design
We studied 105 patients with limited small-cell lung cancer that had been histologically confirmed. Limited disease was defined as cancer limited to one hemithorax, the mediastinum, and the supraclavicular nodes, provided all volumes could be included in the same radiotherapy field as the primary tumor and there was no malignant pleural effusion. Eligibility criteria included an age of 70 years or less; a Karnofsky performance score over 50; no prior chemotherapy, radiation therapy, or thoracic surgery; no history of other cancers (except basal-cell skin carcinoma); and no hematologic, cardiac, renal, or liver abnormalities contraindicating chemotherapy. The study was approved by the local committees of participating hospitals according to current French regulations, and all patients gave informed consent to participate in the study.
Each patient underwent the following studies: chest radiography and fiberoptic bronchoscopy, a complete blood count and biochemical tests, electrocardiography, computed tomography of the thorax and brain, computed tomography or ultrasonography of the abdomen, a bone marrow biopsy, and a radionuclide bone scan in the case of symptoms or abnormal results on biochemical tests.
Patients were treated with alternating courses of radiotherapy and chemotherapy in a schedule similar to that used at the Institut Gustave-Roussy and cooperating centers since 19801. This treatment consisted of six courses of chemotherapy and three courses of thoracic radiotherapy delivered as shown in Figure 1. The patients were randomly assigned to receive 300 mg of cyclophosphamide per square meter of body-surface area daily for four days and 100 mg of cisplatin per square meter on day 2 (higher-dose group) or 225 mg of cyclophosphamide per square meter daily for four days and 80 mg of cisplatin per square meter on day 2 (lower-dose group). Doxorubicin and etoposide were administered in the same doses in both groups. All drugs were given intravenously. The lower doses of cyclophosphamide and cisplatin were given from the second through the sixth course of chemotherapy in both groups. The radiotherapy schedule was the same in the two groups: each of the first two series consisted of 20 Gy in eight fractions given over 12 days to the tumor site, mediastinum, and supraclavicular areas. The third series delivered 15 Gy in six fractions given over nine days in lateral or oblique fields to avoid the spinal cord.
|
Statistical Analysis
A centralized telephone-call procedure was used to assign patients randomly to treatment groups, and allocations to each group were made from a computer-generated list stratified according to center. The doses of chemotherapeutic drugs were calculated according to the concentration of the commercial drug solution and rounded off to 1 mg for doxorubicin, 10 mg for etoposide, 5 mg for cisplatin, and 10 mg for cyclophosphamide.
The main end point was overall survival. The investigation was designed as a one-sided sequential trial with a sequential log-rank statistic and a triangular test14,15. The analysis was based on an anticipated increase in the two-year survival rate of from 20 percent to 40 percent with a power of 90 percent and a type I error of 5 percent. In this design, the data were to be analyzed after every 20 deaths. If, in any analysis, the estimated z statistic (observed minus expected number of deaths) was above the upper boundary, then the results favored the higher-dose group and were significant at the P = 0.05 level. If the z statistic was below the lower boundary, the difference in survival rates between the two treatment groups was not likely to be as large as the 20 percent anticipated. The triangular test used to monitor the progressive accumulation of data allows type I and II errors to be controlled for and can therefore be considered a group-sequential method16. All randomized patients were included in the analysis. The results were reviewed periodically by the statistician in charge of the study and presented to an independent committee of two statisticians, one oncologist, and one thoracic physician. This group discussed the results of each sequential analysis, with the understanding that a recommendation to stop the trial would be made only if the decision to do so was unanimous.
The results in the two groups were compared by the chi-square test, Student's t-test, and the Wilcoxon test. Survival was calculated from the date of randomization to the date of death or the last follow-up visit. The Kaplan-Meier method17 and the log-rank test18 were used to estimate and compare survival curves. A multivariate analysis19 was also performed, allowing adjustments for small imbalances in prognostic factors between the two treatment groups. All reported P values are two-sided.
Results
Clinical Findings
From October 1988 to June 1991, 105 patients were enrolled at 15 centers; 77 percent of the patients were enrolled at 5 centers. Fifty-five patients were assigned to the higher-dose group, and 50 to the lower-dose group. The demographic and clinical characteristics of the patients are shown in Table 1. There were no significant differences between the two groups in terms of any of these characteristics.
|
All patients received the prescribed doses of drugs during the first course of chemotherapy. The doses for the five remaining cycles were slightly lower than planned because of toxicity but were the same in the two groups. Eighty percent of the patients in the higher-dose group and 78 percent of those in the lower-dose group received at least four courses of chemotherapy. Ninety-one percent of the patients in the higher-dose group and 90 percent of those in the lower-dose group received all three courses of thoracic radiotherapy.
Tumor Response and Causes of Failure
The complete-response rate six months after treatment was 67 percent in the higher-dose group and 54 percent in the lower-dose group (P = 0.16). Eight patients in the higher-dose group had an isolated local recurrence after a complete response, as compared with four patients in the lower-dose group. One patient in the higher-dose group and 2 in the lower-dose group had simultaneous local and distant recurrences; 21 and 27 patients in the respective groups had distant metastases alone. Three patients had a new cancer: two patients in the higher-dose group (one had ovarian cancer, and the other laryngeal cancer) and one in the lower-dose group (who had a melanoma). The median duration of a complete response was 540 days in the higher-dose group and 358 days in the lower-dose group (P = 0.06).
Overall Survival
The median interval between randomization and the fourth (last) analysis was 33 months. None of the patients were lost to follow-up. At the time of the second analysis, 32 months after the start of the study, enrollment was closed at the recommendation of the monitoring committee. At that time, survival was significantly better in the higher-dose group (P = 0.001). To extend the follow-up period and confirm the results, additional analyses were performed 6 and 18 months later, after 69 and 85 patients, respectively, had died. These analyses confirmed a significant difference in favor of the higher-dose group (P = 0.02).
The overall survival is shown in Figure 2. The difference in favor of the higher-dose group was 17 percent at two years (P = 0.006 by the log-rank test, stratified according to center), and it remained significant after adjustment for the base-line Karnofsky score and the clinical stage of disease (P = 0.01 and P = 0.02, respectively). A multivariate analysis with adjustment for center, age, Karnofsky score, clinical stage, and weight loss also showed a significant difference (P = 0.008) in favor of the higher-dose group. The relative risk of death in the higher-dose group, as compared with the lower-dose group, was 0.6 (95 percent confidence interval, 0.4 to 0.9) with the triangular test and 0.5 (95 percent confidence interval, 0.3 to 0.8) with stratification and adjustment for prognostic factors by multivariate analysis.
|
The rate of disease-free survival (Figure 3) (i.e., survival with local control of the tumor and with no recurrence) was higher in the higher-dose group. At two years, disease-free survival was 28 percent in that group, as compared with 8 percent in the lower-dose group (P = 0.02).
|
Treatment Toxicity
More patients in the higher-dose than in the lower-dose group had severe granulocytopenia (fewer than 500 granulocytes per cubic millimeter) after the first cycle of chemotherapy (Table 2), but there was no difference between the two groups after subsequent cycles. One patient in the lower-dose group had a transient elevation of serum creatinine to over 2.8 mg per deciliter (250 µmol per liter), and five patients in each group had vomiting or diarrhea requiring intensive therapy. One patient in the lower-dose group had a myocardial infarction.
|
We found that higher doses of cyclophosphamide and cisplatin during the first cycle of chemotherapy in patients with limited small-cell lung cancer resulted in significantly higher survival rates. This result corroborates the hypothesis that gave rise to the trial4,11,20. A multivariate analysis of 131 patients11 predicted a relative risk of death of 0.6 with an increase of 20 mg per square meter in the initial dose of cisplatin and a relative risk of 0.7 with an increase of 300 mg per square meter in the initial dose of cyclophosphamide. The relative risk in this trial, calculated by multivariable analysis, was 0.5, and the absolute survival benefit at two years was 17 percent.
Rules for stopping a trial early may cause problems in the evaluation of a treatment effect; such rules should be established before the trial and strictly adhered to21. Since our rules for stopping early had been clearly established at the outset according to the hypothesis to be tested, interruption of the trial seemed justified.
The effect of drug dose on survival is one of the major questions being addressed by trials of cytotoxic chemotherapy for cancer22,23,24. Until now, dosage has been analyzed mainly in retrospective studies3,23 using the dose-intensity method, which takes into account the overall treatment time23. This method has been criticized24,25. Three randomized trials have corroborated the effect of dose intensity on survival: one in patients with breast cancer treated with fluorouracil, doxorubicin, and cyclophosphamide26; one in patients with ovarian cancer27; and one in patients with non-small-cell lung cancer28 treated with cisplatin. The effect of different doses of cyclophosphamide or cisplatin in patients with small-cell lung cancer has been evaluated in five trials5,6,7,8,9. Most of the patients had extensive disease, and only one trial included more than 100 patients with limited disease6. In that study, survival was prolonged by a 50 percent increase in the dose of cyclophosphamide in the patients with limited disease. In our trial, the median respective doses of cyclophosphamide and cisplatin were 194 and 18 mg per square meter per week in the higher-dose group and 176 and 16 mg per square meter per week in the lower-dose group (P = 0.02 for cyclophosphamide, and P = 0.1 for cisplatin). These differences in dose intensity reflect the differences in initial drug doses. Dose intensity is not an adequate method to describe variations in initial doses, since it is an average of the doses delivered throughout the treatment period.
Previous trials of chemotherapy using high initial doses in small-cell lung cancer yielded impressive results in terms of tumor response but no survival benefit2,29. In these trials, the initial drug doses were very high (2 to 10 times the standard dose), and most of the patients needed bone marrow rescue. The risk with very high initial drug doses is that associated toxic effects may prevent completion of the induction regimen and thereby reduce its overall effectiveness. That was not the case in our study. The differences between the higher and lower initial doses of cisplatin and cyclophosphamide were only 20 and 25 percent, respectively -- differences that are frequently encountered in clinical practice. The absolute increase in overall survival, more than 15 percent at two years, is clinically relevant since long-term survival in patients with limited small-cell lung cancer is generally 5 to 15 percent10,29.
The results of this study emphasize the importance of initial doses of chemotherapeutic drugs in the management of small-cell lung cancer, which is a genetically unstable, heterogeneous, and potentially drug-resistant tumor,29,30,31 as predicted by a mathematical model of drug resistance32. If chemotherapeutic drugs effectively eliminate sensitive cells, as they do in the case of small-cell lung cancer, the value of a higher dose will be greater early in the treatment period. Thus, the delivery of higher initial doses could prevent the emergence of chemoresistant tumor cells. The results of our trial suggest that in clinical practice low initial doses should be avoided.
Supported in part by a grant (883063) from the Institut National de la Sante et de la Recherche Medicale/Caisse Regionale Assurance Maladie d'Ile de France.
We are indebted to Dr. Hugues de The and Mr. Florent de Vathaire for their previous studies; to the monitoring committee, which was composed of Mrs. A. Auquier and Drs. J.M. Cosset, B. Dautzenberg, and A. Laplanche; to Dr. R.L. Souhami for constructive criticism; and to Ms. Lorna Saint-Ange for editing the manuscript.
Source Information
From the Institut Gustave-Roussy, Villejuif (R.A., T.L.C., J.-P.P., M.T., P.R.); the Centre Francois-Baclesse, Caen (A.R.); the Centre Hospitalier Intercommunal, Creteil (I.M.); the Fondation Bergonie, Bordeaux (P.C.); and the Centre Paul Papin, Angers (C.T.) -- all in France.
Address reprint requests to Dr. Arriagada at the Institut Gustave-Roussy, Rue Camille Desmoulins, 94805 Villejuif CEDEX, France.
References
The following centers and specialists from the Federation Nationale des Centres de Lutte contre le Cancer and other cooperating hospitals participated in this research: Institut Gustave-Roussy, Villejuif (R. Arriagada, P. Baldeyrou, T. Le Chevalier, J.P. Pignon, P. Ruffie, and M. Tarayre); Centre Francois-Baclesse, Caen (A. Riviere); Centre Hospitalier Intercommunal, Creteil (H. de Cremoux, M. Martin, and I. Monnet); Fondation Bergonie, Bordeaux (P. Chomy); Centre Paul Papin and Centre Hospitalier Universitaire, Angers (C. Tuchais, E. Tuchais, and L. Savary); Clinique de Clermont-Ferrand (A. Tourreau); Centre Claudius-Regaud, Toulouse (C. Chevreau); Centre Rene-Gauducheau, Nantes (J.Y. Douillard); Hopital A. Beclere, Clamart (M.L. Cerrina); Hopital Charles-Nicolle, Rouen (L. Thiberville); Hopital Aiguelongue, Montpellier (J.L. Pujol); and Hopital de Rangueil, Toulouse (V. Fabre-Aldegheri).
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |