INTRODUCTION

Clear-cell sarcoma of the kidney (CCSK) was first reported by Kidd in 1970 [1]. This tumor is characterized by its bone metastasizing tendencies and propensity for late recurrences [2-4]. The investigations of National Wilms' Tumor Study (NWTS)-1, NWTS-2, and NWTS-3 in this group of patients suggested that the addition of doxorubicin to the combination of vincristine plus dactinomycin improved the 6-year relapse-free survival (RFS) rate of children [5]. In NWTS-3, the addition of cyclophosphamide in the schedule and the dose used in the study did not improve the 6-year RFS of children with CCSK. NWTS-4 was designed to compare the efficacy and toxicity of a new schedule of administration of dactinomycin and doxorubicin. Detailed toxicity, hospitalization, and outcome data for the primary randomization between divided-dose, or standard (STD), and single-dose, or pulse-intensive (PI), drug administration have been previously published [6,7]. We now report the results of the second randomization for patients with CCSK, which took place after 5.5 months of treatment. Patients were randomly assigned to discontinue chemotherapy after a 6-month treatment period (short regimen) or continue therapy for an additional 9 months (long regimen).

PATIENTS AND METHODS

NWTS-4 was a multi-institutional, randomized clinical trial of different treatment regimens for patients younger than 16 years at diagnosis with certain renal tumors, who were diagnosed between August 1986 and September 1994. Microscopic slides from all tumors from patients on NWTS-4 were reviewed by National Wilms' Tumor Study Group pathologists. The protocol was approved by all institutional review boards of all institutions that entered patients onto NWTS-4. All parents or guardians signed informed consent before study enrollment. The staging of patients in NWTS and the designation of patients as either randomized or observed has been discussed previously [6]. Chest x-ray films, skeletal survey, and computed tomography (CT) or magnetic resonance imaging of the brain were required for staging. Bone scans and bone marrow aspirates and biopsies were not required. Patients who had hepatic metastases were required to have CT scans of the liver. All patients with stages I to IV CCSK were randomly assigned to receive vincristine, dactinomycin, and doxorubicin using ST (regimen DD) or PI (regimen DD-4A) schedules, with a second randomization for treatment to be continued beyond or discontinued at week 28 (regimen DD) or week 26 (regimen DD-4A) or observed on this study. The treatment regimens are shown in Figure 1. All patients received postoperative radiation therapy to the tumor bed and other infradiaphragmatic and metastatic sites as necessary. Treatment outcomes were evaluated according to the final treatment regimen assigned to the patients. The number of patients assigned to the two groups is not balanced because the Data and Statistics Center used a randomized block design, blocking by institution but without regard to histology. Therefore, CCSK patients were not randomly assigned as a separate population.

The quantitative measure of therapeutic outcome used for this study design was the percentage of patients who were alive and disease-free 5 and 8 years after the second randomization. Patients who relapsed, had progression of disease that necessitated a change in therapy, or died as a result of toxicity or other causes without prior relapse were considered as having treatment failure.

Two separate analyses were performed. The first included all patients as originally randomly assigned, based on institutional assignment of stage and histology, with no exclusions for violations, such as failure to follow the assigned regimen. The second analysis excluded those patients who had serious protocol violations, such as a change from one regimen to the other. The results for the two analyses were similar, and only the results based on the original randomization are reported here.

Data were analyzed using standard statistical methods, which included the log-rank test [8,9]. Relative risk (hazard ratio) estimates were based on the Cox proportional hazards model [10].

RESULTS

There were a total of 86 (59 males and 27 females) CCSK patients in NWTS-4; the numbers of patients assigned to final stage I, II, III, and IV were 35, 21, 28, and two, respectively. No CCSK patient had bilateral disease (stage V) or positive CT or magnetic resonance imaging of the brain at diagnosis. Only one patient relapsed within 6 months of initial registration. This patient was randomly assigned to the PI chemotherapy at the initial randomization and to short-duration chemotherapy approximately 4.5 months after first randomization. However, the patient had a pulmonary relapse approximately 5 months after the first randomization. This patient was included in the initial randomization analysis but not in the analysis of the second randomization.

The initial randomization took place when the patient registered for NWTS-4. Fifty-three patients were randomly assigned, 29 patients potentially eligible for random assignment were observed, and four patients were not studied. Among the 53 patients who were randomly assigned, four were switched from their original regimen to the other regimen. Two of the changes in regimen occurred because of family preference, one was a protocol error, and one occurred because the physician preferred another treatment. In the initial randomization, 27 patients were assigned to ST chemotherapy, and 26 were assigned to PI chemotherapy. The 8-year RFS rates for patients randomly assigned to the PI arm and the ST arm were 71.8% and 69.6%, respectively (P = .81). The 8-year survival rates for patients randomly assigned to the PI arm and the ST arm were 87.3% and 83.7%, respectively (P = .65). There was no significant difference between ST and PI chemotherapy. There were seven observed relapses on the PI arm and eight on the ST arm [6].

The second randomization was conducted approximately 5.5 months after the first randomization. Patients who relapsed before 6 months after the first randomization were not eligible for subsequent randomization. Among the 53 patients with CCSK who received the initial randomization, only one patient was ineligible because of prior relapse. Forty patients were randomized, and 12 patients were eligible for randomization but were placed in the category of observed. This was a result of institutional or family preference in 11 patients and a change in stage and histology in one patient. All 12 patients were treated on the long-duration chemotherapy arm. Of those 12 patients, three patients relapsed (8-year RFS, 75%). The overall 8-year RFS for patients (randomized and observed) treated on the long-duration chemotherapy arm was 80.7%. The earliest relapse occurred at approximately 2 years after the first randomization.

Among the 40 randomized patients (11 females and 29 males), the median age was 29 months, with a range of 5 months to 91 months. There were 30 white patients, four Hispanics, five African-Americans, and one other who were randomized subsequently. The demographics for the randomized patients and all patients on NWTS-4 are listed in Table 1.

Twenty-three patients were randomly assigned to short-duration chemotherapy, and 17 patients were randomly assigned to long-duration chemotherapy. There were eight observed relapses on the short arm and three observed relapses on the long arm. The 5- and 8-year RFS rate for patients who were randomly assigned to the long arm was 87.8% for both time points. For patients randomly assigned to the short arm, the 5- and 8-year RFS rates were 65.2% and 60.6%, respectively (P = .08). The 5- and 8-year survival rate for patients randomly assigned to the long arm was 87.5% for both time points versus 5- and 8-year RFS rates of 95.5% and 85.9%, respectively, for patients randomly assigned to the short arm (P = .99). Figures 2 and 3 show the RFS and overall survival for patients. The difference for overall survival between short-duration and long-duration chemotherapy was not statistically significant, and survival curves overlapped. Patients randomly assigned to long-duration chemotherapy had a higher estimated RFS rate than the short-duration group at both 5 and 8 years.

A total of 23 patients (observed and randomized) relapsed. Fifteen of the patients underwent the first randomization, whereas 11 of them underwent the second randomization. The median time to relapse was 1.4 years, with a range of 5 months to 3.06 years. One event in the short-duration arm was the death of a patient at 7.93 years as a result of chronic myelogenous leukemia as a second malignancy. The initial sites of relapse are listed in Table 2. The lungs were the most common site of relapse, followed by bone and tumor bed and abdomen. The median follow-up time for all CCSK patients was 8.94 years (range, 0.57 to 14.6 years).

There were eight relapses among the 32 stage I CCSK patients on NWTS-4 who were observed for a median of 10.3 years. The RFS at 2 years was 84.4%. All but one of the relapses of stage I patients occurred within the first 2 years.

There were three second malignancies observed. These included one acute lymphoblastic leukemia, one chronic myelogenous leukemia, and one acute myelomonocytic leukemia. The patient with chronic myelogenous leukemia was one of 23 patients treated on the short-duration arm, and the other two patients were among 17 patients on the long-duration arms. Other reported late effects include renal insufficiency, renal tubular acidosis (n = 3), chronic persistent hepatitis with portal fibrosis, mild scoliosis (n = 2), and primary ovarian failure.

Overall, there were a total of 86 CCSK patients on NWTS-4, of whom 23 relapsed and 13 died. On NWTS-3, there were 90 patients, of whom 35 relapsed and 30 died. The RFS and overall survival rates for NWTS-4 are 71.6% and 83.0% at 8 years compared with 60.2% and 66.9% in NWTS-3, respectively. The improvement in overall survival from NWTS-3 to NWTS-4 is statistically significant (P < .01, Fig 4).

DISCUSSION

The present analysis was undertaken to compare the RFS and overall survival rates of patients with stage I to IV CCSK treated with conventional and ST versus PI chemotherapy and short-duration versus long-duration chemotherapy entered onto NWTS-4. Previous NWTSG studies suggested an improvement in RFS from the addition of doxorubicin to the treatment of this group of patients. The results of the first randomization of NWTS-4 have been previously reported and did not show any difference between ST and PI therapy [6]. The results of the second randomization suggest that the group of patients receiving long-duration chemotherapy had higher estimated rates of RFS than the short-duration group at 5 and 8 years. However, because of the small sample size and small number of events, the possibility that the true difference (in a large population) is zero at the conventional level (P = .05) of statistical significance cannot be excluded. The estimated relative risk for long-duration versus short-duration chemotherapy with respect to the RFS is 0.28 (95% CI, 0.04 to 1.13). This is in contrast to the findings in favorable-histology Wilms' tumor, where there was no difference in efficacy based on the duration of therapy [11].

Because it was not shown to be statistically significant, explanations for the difference in disease control between the two arms include chance, the duration of therapy, and increased doses of the three chemotherapeutic agents. When comparing the therapeutic arms, there is a difference in cumulative doses of anthracycline. The short-duration arm of regimen DD contains 120 mg/m² compared with 150 mg/m² on regimen DD-4A. The long-duration arms of both regimen arms include 300 mg/m² of doxorubicin. This could be a factor in the difference between event-free survival between the short and long arms [12]. In addition, the risk of long-term effects, notably congestive heart failure, and possible oncogenesis needs to be considered in the evaluation, particularly in the context of the lack of overall survival benefit [13]. Although patients who relapse can receive additional salvage therapy, the additional therapy toxicities, cost, and emotional burden cannot be quantitated. These factors are not examined in this article.

Other cooperative groups have included CCSK patients in their therapeutic trials. Results of the second United Kingdom Children's Cancer Study Group Wilms' Tumor Study-2 with CCSK (n = 18) demonstrated a 4-year event-free survival (defined as time from diagnosis to first relapse or progression, time to death, or date of last follow-up) of 88%. Treatment included three chemotherapeutic agents (intensive vincristine, dactinomycin, and doxorubicin) administered for 12 months. Stage I and II patients did not receive irradiation to the tumor bed [14]. The International Society of Pediatric Oncology (SIOP) reported a 2-year event-free survival for CCSK (n = 14) of 75% and overall 5-year survival of 88% for patients treated on SIOP-9. SIOP-9 therapy included a randomization between 4 or 8 weeks of preoperative chemotherapy with vincristine and dactinomycin. After nephrectomy, patients received an additional 36 weeks of chemotherapy (dactinomycin, vincristine, anthracycline, and ifosfamide) in addition to radiotherapy. The entire duration of therapy ranged between 40 and 44 weeks, with short follow-up for this disease [15]. The cumulative anthracycline doses were 360 mg/m² (300 mg/m² for those undergoing pulmonary radiation therapy) on the United Kingdom Children's Cancer Study Group Wilms' Tumor Study-2 and 250 mg/m² on SIOP-9. The European experience provides further support for the benefit of anthracycline in disease control as well as longer duration of treatment.

The stage distribution for CCSK patients in the analysis reported here is similar to previous studies, with only two patients presenting with stage IV disease compared with four patients in NWTS-3. The occurrence of brain metastasis as the first site of recurrence was lower compared with earlier studies. In this report, only one of 21 patients (randomized and observed patients) developed a brain metastasis compared with 11% of patients in earlier studies. This metastasis was in addition to other sites, whereas there was a higher incidence of isolated brain metastases in earlier studies [4,16]. More than 30% of the recurrences occurred between 2 years and 37 months; however, none have been reported after 37 months. This differs from a previous experience [12]. Bone scans were required at diagnosis for patients on NWTS-5 based on the findings in this study.

The overall RFS for patients with CCSK has improved from NWTS-3. Patients treated on NWTS-3 had an 8-year RFS of 60.2% and overall survival of 66.9%. Patients treated on arm DD-RT of NWTS-3 had a 64.6% 6-year RFS. This arm was identical to the long-duration arm DD of NWTS-4 (including radiation therapy). One possible explanation for the improvement in outcome on NWTS-4 is the PI administration of dactinomycin and doxorubicin. A statistically significant improvement is present in overall survival between NWTS-3 and NWTS-4 (66.9% v 83.0%, respectively; P < .01). The improvements observed in NWTSG studies are shown in Figure 4. This could be related to the use of newer agents for salvage therapy, such as ifosfamide, carboplatin, and etoposide.

CCSK patients administered a longer course of therapy (15 months) using vincristine, doxorubicin, and dactinomycin exhibit better RFS results. Whether this improvement is a result of the longer duration or higher cumulative doses of chemotherapy, particularly anthracycline given as a pulse, is not known. This does not lead to an improvement in overall survival. The addition of newer agents into the treatment for this renal tumor may produce superior RFS and overall survival. This will be studied in the next NWTSG therapeutic trial for patients with CCSK.

We conclude that a longer course of treatment (15 months) with vincristine, doxorubicin, and dactinomycin for patients with CCSK provides better RFS than a shorter course (6 months). This does not lead to an improvement in overall survival. Although patients who have experienced relapse may be administered salvage treatment, it is not without the costs of additional therapy and toxicities.

Authors' Disclosures of Potential Conflicts of Interest

The authors indicated no potential conflicts of interest.