1Klinik für Hämatologie und Onkologie, Klinikum der Philipps Universität, Marburg, Germany; 2Cancer Division Medical Research Council Clinical Trials Unit, Cambridge, UK; 3Medizinische Klinik III, Klinikum Grosshadern der Ludwig Maximilians Universität, München, Germany; 4Department of Medical Oncology and Radiotherapy, The Norwegian Radium Hospital, Oslo, Norway; 5Medizinische Klinik II, Humboldt Universität zu Berlin, Berlin, Germany
Received 30 May 2001; revised 1 October 2001; accepted 23 October 2001.
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The purpose of this study was to compare high-dose chemotherapy (HDCT) with conventional-dose chemotherapy (CDCT) as first-salvage treatment in patients with relapsed or refractory non-seminomatous germ-cell tumors (NSGCT).
Patients and methods
One hundred and ninety-three patients with relapsed or refractory NSGCT, between 1981 and 1995, were identified from two large databases. In 74 of these, intensification of first-salvage treatment by HDCT was planned. Patients were matched based on primary tumor location, response to first-line treatment, duration of this response and serum levels of the tumor markers, human chorionic gonadotrophin (HCG) and -fetoprotein (AFP). Multivariate analyses were performed using event-free survival and overall survival as primary endpoints.
Results
Full matches on all five factors were found for 38 pairs of patients; for a further 17 pairs, matches on at least four factors could be identified. Hazard ratios in favor of HDCT were obtained between 0.72 and 0.84 [confidence interval (CI) 0.591.01] for event-free survival and between 0.77 and 0.83 (CI 0.600.99) for overall survival, depending on the type of analysis.
Conclusions
The current analysis suggests a benefit from HDCT, with an estimated absolute improvement in event-free survival of between 6 and 12% and in overall survival of between 9 and 11% at 2 years. This benefit is lower than expected from previous phase I/II studies.
Key words: hematopoietic stem cell transplantation, matched-pair analysis, non-seminoma, prognosis, salvage therapy, seminoma
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Initial studies from Indiana University reported long-term remissions, even in patients with multiple relapses and/or refractory NSGCT [10]. More recently, the same group has updated their results on high-dose chemotherapy (HDCT) given as intensification of first-salvage treatment in NSGCT and seminoma [11, 12]. It remains unknown, however, whether HDCT is more effective than conventional-dose chemotherapy (CDCT) in this setting. Several phase I/II studies suggested a superiority of HDCT over CDCT, but phase I/II trials are notoriously biased by patient selection and tend to overestimate a treatment effect. Pending the results of a multicenter European trial, we performed a retrospective analysis of patients with or without HDCT as part of their first-salvage treatment and matched these patients for known prognostic factors.
![]() |
Patients and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Patients with CDCT were either treated within centers participating in clinical trials of the British Medical Research Council (MRC) or at the Klinikum Grosshadern, Munich, Germany. Patients with HDCT were all treated at the Virchow Klinikum, Berlin, Germany. All patients who fulfilled the eligibility criteria in the two databases were included on an intention-to-treat basis, irrespective of whether or not they received the salvage treatment as scheduled or their response to salvage treatment.
Different combinations and schedules were used for CDCT. In addition to cisplatin, 75 out of 119 patients (63%) received etoposide and 43 patients (36%) received ifosfamide as part of their salvage treatment. Patients with HDCT as first-salvage treatment were scheduled for two to three cycles of conventional-dose cisplatin, etoposide and ifosfamide followed by one cycle of high-dose carboplatin, etoposide and ifosfamide, plus re-infusion of autologous bone marrow or peripheral blood progenitor cells [13]. Modifications of this schedule could be made in patients depending on previous treatment, response to CDCT, availability of HDCT and the occurrence of nephrotoxicity. The characteristics of all patients eligible for matching and included in the analysis are shown in Table 1. Patients with tumor marker-negative partial remissions after salvage treatment were scheduled for postchemotherapy surgical resections of residual tumors.
|
Matching factors
Details of the identification of prognostic factors for treatment response after CDCT or HDCT have been published elsewhere [7, 9]. In the present analysis we used the following factors for matching: (i) site of primary tumors; (ii) response to first-line treatment; (iii) relapse-free interval; (iv) level of HCG at relapse; (v) level of AFP at relapse. Each of the matching factors could have several categories (Table 2). Thereafter, a CDCT control patient was sought for each HDCT patient, identical in all categories of the five matching factors. Where more than one possible match was available, the control patient with the closest year of salvage treatment was chosen.
|
Statistical analysis
Survival curves were produced using the KaplanMeier method [14]. To avoid potential bias resulting from the unequal follow-up times in the HDCT and CDCT groups, all patients surviving >3 years were censored at this point. This timepoint was chosen both for its clinical relevance and because it was the minimum follow-up time for surviving patients (with only three exceptions).
Hazard ratios for the treatment group effect estimates and 95% confidence intervals (CI) were obtained from multivariate analyses conducted using Coxs proportional hazards regression model [15]. Three types of models were fitted. The first model used all eligible patients and adjusted for the important prognostic factors, namely those chosen as matching factors. Although this provided a large patient cohort, imbalances in some of the factors across the treatment groups (e.g. the last response category and the first relapse-free interval category; see Table 1) are such that even adjusted analyses may not be adequate. This was therefore used as a control analysis, while the main analyses used only the matched patients. The stratified model used the matching process directly, by considering each matched-pair of patients as a separate stratum and fitting the treatment-group variable as the only covariate. Pairs of patients in which both patients are censored, or in which one patients survival time is censored before the event-time of its pair, do not contribute to this analysis. Therefore, further models (referred to as the adjusted analyses) were carried out in which the five factors used to match patients, as well as a variable indicating treatment group, were fitted simultaneously.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Matching
The factors chosen for matching are listed in Table 2. All factors (except one) were identified as being of independent prognostic significance for survival after conventional-dose salvage treatment in one recent retrospective analysis and were very similar to published prognostic factors for HDCT [9]. However, the presence of a primary mediastinal NSGCT was added as an additional matching factor, because primary mediastinal NSGCT are reported to respond poorly to conventional-dose salvage treatment [7, 16].
As a result both the CDCT and HDCT groups were highly comparable in histology, first-line treatment, response to first-line treatment and disease status at the time of relapse (Table 3). However, as HDCT only recently became available for the salvage treatment of NSGCT, patients who received CDCT were treated at an earlier date, compared with patients who received HDCT. Therefore, the control patient with the closest year of salvage treatment was chosen, where more than one possible match was available. Most likely for the same reason, etoposide, which was introduced into the first-line treatment of metastatic NSGCT in the mid-1980s, was used less often during first-line treatment in patients with CDCT compared with patients with HDCT (76 versus 97%).
|
Follow-up
All patients underwent follow-up at their respective institutions. The median follow-up time for patients alive after CDCT was 9 years (range 2.512.5 years) for the MRC patients and 7.5 years (range 13 months to 12.5 years) for the Munich patients. The median follow-up time for patients alive after HDCT was 5 years (range 12 months to 8 years).
Among the 55 matched patients who received CDCT, 49 (89%) had disease progression or relapse after initiation of salvage treatment, of whom 42 have died. Of the surviving patients, one was lost to follow-up after 28 months and one after 32 months. The remaining 11 patients all have a minimum follow-up time of 3 years after initiation of salvage treatment. Among the 55 matched patients scheduled to receive HDCT, 41 (74%) suffered disease progression or relapse after initiation of salvage treatment, of whom 37 have died. Of the surviving patients, one was lost to follow-up after 12 months, the remaining 17 patients also have a minimum follow-up time of 3 years after initiation of salvage treatment.
Multivariate analyses
The results of the multivariate analyses are summarized in Table 4 (overall survival) and in Table 5 (event-free survival). The results for overall survival were very consistent for both models (adjusted and stratified) and when using the 38 fully matched-pairs or the 55 pairs with at least four matching factors, ranged from 0.75 (adjusted, fully matched) to 0.79 (at least four factors matching, stratified model), where hazard ratios <1 indicate benefit from HDCT. The results for event-free survival were only slightly more varied, and slightly less extreme (with one exception) than for overall survival, ranging from 0.72 (fully matched, adjusted analysis) to 0.82 (fully matched, stratified analysis). The analyses of all potentially matchable patients also produced broadly similar results.
|
|
Additional analyses
As there are data to suggest that patients with a less effective first-line treatment may have a better chance of a good response to salvage treatment and as there was a slight excess of patients with CDCT who had received cisplatin, vinblastine and bleomycin rather than cisplatin, etoposide and bleomycin as first-line treatment, the above analyses were repeated only in patients who had received both cisplatin and etoposide during their first-line treatment. The survival rates after salvage treatment were slightly lower in patients who had received cisplatin and etoposide as first-line treatment compared with the corresponding groups in which all matched patients were analysed whether or not they had received etoposide.
A final analysis was made to assess the impact of the type of salvage treatment on subgroups of patients with good prognosis or poor prognosis, according to the MRC prognostic score for conventional-dose salvage treatment [9]. A treatment effect in favor of HDCT was slightly more pronounced in poor prognosis patients, with a hazard ratio (stratified) for event-free survival of 0.72 (CI 0.461.12) compared with a hazard ratio for good prognosis patients of 0.84 (CI 0.651.15). For survival, the difference was a little more extreme with a hazard ratio (stratified) of 0.97 (CI 0.731.29) for good prognosis patients and 0.25 (CI 0.051.25) for poor prognosis patients. Although this suggests a greater benefit in event-free survival from HDCT in poor prognosis patients, the sample size for this analysis was small and a test for heterogeneity was not statistically significant.
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Comparisons between CDCT and HDCT are seriously limited, however, by a number of factors including: the small number of patients treated; the heterogeneity of prior treatments; large variations in inclusion criteria, such as inclusion of patients with seminoma and NSGCT and those scheduled for first as well as subsequent salvage treatments. The results from available phase I/II trials of the salvage treatment of relapsed or refractory NSGCT are scattered over a wide range and can hardly be compared.
We therefore performed a retrospective matched-pair analysis based on very stingent inclusion criteria to account for most of these confounding variables. First, only patients without any prior salvage chemotherapy were considered. Secondly, we limited the analysis to patients with NSGCT. Patients with pure seminoma rarely require salvage chemotherapy and will respond better to salvage treatment than patients with NSGCT [4, 12, 21]. Thirdly, we only included patients who had re-ceived cisplatin as a first-line as well as a first-salvage drug. Fourth, patients with a marker plateau after first-line treatment or with only radiological evidence of relapse or progression may not require salvage chemotherapy. We therefore limited the analysis to patients with unequivocal progression after first-line treatment, either defined by increasing tumor markers or by histological evidence of progressive undifferentiated tumors. Fifth, time-to-treatment and selection bias frequently interfere with the interpretation of trials that investigate intensification of conventional-dose treatment with HDCT. To ensure comparability of both cohorts of patients, we performed the analysis on an intention-to-treat basis. For this reason seven patients who were scheduled for, but did not receive, HDCT remained in the analysis sample.
The matching of patients was extensive, taking into account five variables with between two and four categories each that covered most of the known prognostic factors for CDCT and HDCT [7, 9]. As a result both the CDCT and HDCT groups were highly comparable in histology, first-line treatment, response to first-line treatment and disease status at the time of relapse, and were thus very similar in their expected treatment outcomes (Table 3).
We found consistent results in favor of HDCT using different multivariate approaches. The analysis in which each matched-pair of patients was considered as a separate stratum is generally the best analysis, but it can reduce the number of pairs that can be compared. However, we found similar estimates in the multivariate analyses that included matching factors and treatment groups as covariates, as well as in those that considered the entire group of all 193 eligible patients. In addition, we found consistent results when analyzing only those patients who received both etoposide and cisplatin as primary treatment.
Yet, the present analysis has several limitations that have to be considered in the interpretation of our findings. Due to the close matching, the number of patients from which the conclusions of the study are drawn is small, representing 76 and 110 patients, i.e. 38 and 55 pairs of patients, respectively. In addition, the two cohorts of patients, after CDCT and after HDCT, differed in several important aspects. Information about CDCT was obtained from various institutions cooperating within trials of the MRC and from an earlier time period than information about HDCT. The results with HDCT represented more recent and single center data. Treatment period as well as treatment at an expert center have both been shown to be of additional prognostic significance [22, 23]. Due to the differences in treatment periods, patients in the CDCT cohort may not have received what is considered optimal salvage treatment today. Only 82% of matched CDCT patients received etoposide as part of their first-line treatment and not all of these patients received ifosfamide during salvage. In contrast, these drugs were used in all HDCT patients. As a result, the diffences in favor of HDCT might in fact be smaller if more recent data about CDCT had been analyzed.
Given these limitations, how do the results of the present analysis contribute to the discussion about the optimal strategy for first-line salvage treatment in NSCGT? Based on an overall survival rate of 32% after conventional-dose salvage treatment, suggested by the results from one of the most recent trials, the hazard ratios found in our analysis translate to best estimates for an absolute benefit from HDCT of between 9 and 11% [4]. Similarly, given an event-free survival rate of 24% from the same trial, the best estimates for a benefit from HDCT range between 6 and 12%.
Retrospective analyses such as reported here serve only as estimates about possible treatment benefits and do not prove that HDCT is indeed superior to CDCT in the first-salvage setting. Moreover, the biases inherent in retrospective analyses tend to favor the most recently treated group and do not take into account all the complex and intuitive treatment decisions that have contributed to the success of salvage treatment in recent years. Without randomization it is impossible to exclude all possible imbalances and biases that may arise from retrospective analyses of different patient cohorts.Therefore, the results from the present analysis are no substitute for a prospective, randomized trial as is currently underway within the European Group for Bone Marrow Transplantion, but rather serve as a best estimate of a potential treatment benefit from HDCT and of the sample size required for a randomized trial.
![]() |
Acknowledgements |
---|
|
![]() |
Footnotes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Harstrick A, Schmoll HJ, Wilke H et al. Cisplatin, etoposide, and ifosfamide salvage therapy for refractory or relapsing germ cell carcinoma. J Clin Oncol 1991; 9: 15491555.[Abstract]
3. McCaffrey JA, Mazumdar M, Bajorin DF et al. Ifosfamide- and cisplatin-containing chemotherapy as first-line salvage therapy in germ cell tumors: response and survival. J Clin Oncol 1997; 15: 25592563.[Abstract]
4. Loehrer PJ Sr, Gonin R, Nichols CR et al. Vinblastine plus ifosfamide plus cisplatin as initial salvage therapy in recurrent germ cell tumor. J Clin Oncol 1998; 16: 25002504.[Abstract]
5. Motzer RJ, Geller NL, Tan CCY et al. Salvage chemotherapy for patients with germ cell tumors. The Memorial Sloan-Kettering Cancer Center experience (19791989). Cancer 1991; 67: 13051310.[ISI][Medline]
6. Gerl A, Clemm C, Schmeller N et al. Prognosis after salvage treatment for unselected male patients with germ cell tumours. Br J Cancer 1995; 72: 10261032.[ISI][Medline]
7. Beyer J, Kramar A, Mandanas R et al. High-dose chemotherapy as salvage treatment in germ cell tumors: a multivariate analysis of prognostic factors. J Clin Oncol 1996; 14: 26382645.[Abstract]
8. Schmoll HJ, Beyer J. Prognostic factors in metastatic germ cell tumors. Semin Oncol 1998; 25: 174185.[ISI][Medline]
9. Fossa SD, Stenning SP, Gerl A et al. Prognostic factors in patients progressing after cisplatin-based chemotherapy for malignant non-seminomatous germ cell tumours. Br J Cancer 1999; 80: 13921399.[ISI][Medline]
10. Nichols CR, Tricot G, Williams SD et al. Dose-intensive chemotherapy in refractory germ cell cancera phase I/II trial of high-dose carboplatin and etoposide with autologous bone marrow transplantation. J Clin Oncol 1989; 7: 932939.[Abstract]
11. Broun ER, Nichols CR, Gize G et al. Tandem high dose chemotherapy with autologous bone marrow transplantation for initial relapse of testicular germ cell cancer. Cancer 1997; 79: 16051610.[ISI][Medline]
12.
Bhatia S, Abonour R, Porcu P et al. High-dose chemotherapy as initial salvage chemotherapy in patients with relapsed testicular cancer. J Clin Oncol 2000; 18: 33463351.
13. Rick O, Beyer J, Kingreen D et al. High-dose chemotherapy in germ cell tumours: a large single centre experience. Eur J Cancer 1998; 34: 18831888.[ISI][Medline]
14. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457481.[ISI]
15. Cox DR. Regression models and life-tables. J R Stat Soc (B) 1972; 34: 187220.[ISI]
16. Saxman SB, Nichols CR, Einhorn LH. Salvage chemotherapy in patients with extragonadal nonseminomatous germ cell tumors: the Indiana University experience. J Clin Oncol 1994; 12: 13901393.[Abstract]
17. Levine EG, Raghavan D. Treatment of refractory testis cancer: salvage or savage chemotherapy? Eur J Cancer 1991; 27: 932936.[ISI][Medline]
18. Williams SD. High dose therapy in germ cell tumors: when, what, and how much? Ann Oncol 1992; 3: 780781.[ISI][Medline]
19. Horwich A. Salvage therapy of germ cell tumours. Br J Cancer 1995; 71: 901903.[ISI][Medline]
20. Josefsen D, Ous S, Hoie J et al. Salvage treatment in male patients with germ cell tumours. Br J Cancer 1993; 67: 568572.[ISI][Medline]
21. Miller KD, Loehrer PJ, Gonin R, Einhorn LH. Salvage chemotherapy with vinblastine, ifosfamide, and cisplatin in recurrent seminoma. J Clin Oncol 1997; 15: 14271431.[Abstract]
22. Einhorn LH. Complicated problems in testicular cancer. Semin Oncol 1988; 15 (Suppl 3): 915.[ISI][Medline]
23.
Collette L, Sylvester RJ, Stenning SP et al. Impact of the treating institution on survival of patients with "poor-prognosis" metastatic nonseminoma. European Organization for Research and Treatment of Cancer Genito-Urinary Tract Cancer Collaborative Group and the Medical Research Council Testicular Cancer Working Party. J Natl Cancer Inst 1999; 91: 839846.