Fax: (206) 616-2724
Doxorubicin for favorable histology, Stage II–III Wilms tumor
Results from the National Wilms Tumor Studies
Article first published online: 16 JUL 2004
Copyright © 2004 American Cancer Society
Volume 101, Issue 5, pages 1072–1080, 1 September 2004
How to Cite
Breslow, N. E., Ou, S.-S., Beckwith, J. B., Haase, G. M., Kalapurakal, J. A., Ritchey, M. L., Shamberger, R. C., Thomas, P. R. M., D'Angio, G. J. and Green, D. M. (2004), Doxorubicin for favorable histology, Stage II–III Wilms tumor. Cancer, 101: 1072–1080. doi: 10.1002/cncr.20433
- Issue published online: 18 AUG 2004
- Article first published online: 16 JUL 2004
- Manuscript Accepted: 12 MAY 2004
- Manuscript Revised: 7 MAY 2004
- Manuscript Received: 5 FEB 2004
- United States Public Health Service. Grant Numbers: CA 42326, CA 54498
- congestive heart failure;
- historical controls;
After randomized trials in the 1980s, doxorubicin (DOX) was added to dactinomycin plus vincristine as standard chemotherapy for patients who had Stage III Wilms tumor (WT) of favorable histology (FH). Double-agent chemotherapy was retained for patients with Stage II disease. In this study, the authors reevaluated the efficacy of DOX using extended follow-up and additional patients.
The relative risks (RR) (DOX vs. no DOX) of disease recurrence and mortality were estimated for patients with Stage II–III/FH WT who were enrolled in the third and fourth National Wilms Tumor Studies (NWTS-3 and NWTS-4). The risk of congestive heart failure (CHF) was estimated for all patients who received DOX.
No statistically significant effects of DOX were found for patients with Stage II tumors. For patients with Stage III tumors, the 8 year recurrence-free survival (RFS) and overall survival (OS) rates for randomized patients on NWTS-3 were 84% and 89%, respectively, for patients who received DOX (n = 130) and 74% and 83%, respectively, for patients who did not receive DOX (n = 118). Including all patients with Stage III disease who received DOX (n = 678) and did not receive DOX (n = 138), the RRs of recurrence were 0.47 (P = 0.007) and 0.40 (P = 0.011), and local recurrence respectively, after adjustment for radiation therapy (RT) dose, whereas the RR of mortality adjusted for RT and study was 0.68 (P = 0.17). The 20-year risk of CHF after primary DOX treatment on NWTS-3 and NWTS-4 was 1.2%.
The inclusion of data from nonrandomized patients yielded estimates of DOX treatment effects for Stage III/FH WT that were stronger, albeit more susceptible to bias, than reported previously. Despite a lower reported risk of CHF, conclusive evidence that frontline therapy with DOX definitively improves survival remains elusive. Cancer 2004. © 2004 American Cancer Society.
The prognosis for children with Wilms tumor (WT) improved dramatically during the past 3 decades. One factor believed responsible for this improvement was the introduction of anthracycline therapy for patients with tumors of unfavorable histology or advanced-stage disease of favorable histology (FH).1 The addition of doxorubicin (DOX) to dactinomycin and vincristine after radiation therapy (RT) led to improved recurrence-free survival (RFS) for patients with Stage II–III/FH disease who were randomized on the second National Wilms Tumor Study (NWTS-2), although overall survival (OS) changed little.2 Because of concern regarding chronic toxicity, NWTS-3 reevaluated therapy for FH disease using a randomized, factorial design that involved treatment with two or three agents and an RT dose reduction (Stage III) or elimination (Stage II).3 DOX did not improve RFS for patients with Stage II/FH disease. A moderate improvement in RFS observed for patients with Stage III/FH disease was statistically significant (P = 0.04) only in subgroup analyses that excluded patients who were not treated as randomized. The recommendation that DOX be used with reduced RT (10 grays [Gy]) for patients with Stage III/FH disease was based partly on the observation that 3 of 4 intraabdominal recurrences in patients with Stage III/FH disease occurred among 71 patients who received reduced RT and no DOX.3
The sixth study of WT by the International Society of Pediatric Oncology (SIOP-6), likewise, involved a randomized comparison of 2 drugs versus 3 drugs for patients with Stage II N1 or Stage III disease, regardless of histology4 (these 2 stages jointly meet the NWTS Stage III definition; however, in SIOP-6, they represented residual disease after preoperative therapy.) Improvement in the 2-year RFS rate (from 49% to 77%; P = 0.03), but no difference in OS, was observed for small numbers of patients who were treated without DOX (n = 43) or with DOX (n = 40) in addition to dactinomycin, vincristine, and RT after preoperative chemotherapy. Separate results were not reported for the even smaller numbers of patients with FH disease (n = 35 and n = 33, respectively). Subsequent SIOP protocols recommended 3-agent chemotherapy, including an anthracycline, either DOX or epirubicin, and 15-Gy RT for these patients.5
The current reanalysis of NWTS data was motivated by the weak statistical evidence from randomized clinical trials supporting the use of DOX for patients with Stage III/FH WT. Concern is increasing about the effects of DOX on rates of congestive heart failure (CHF) and, in combination with RT, on second malignancies in WT survivors.6, 7 NWTS-4 omitted DOX and RT for patients with Stage II/FH disease while mandating 10-Gy flank RT and DOX for patients with Stage III/FH disease. Short, single-dose treatment regimens yielded RFS and OS rates equivalent to those obtained with standard, divided-dose regimens while reducing hematologic toxicity and treatment costs.8–10 Because those regimens were comparable to the regimens used in the same stage/histology subgroups on NWTS-3, this offered the opportunity to revisit the therapeutic comparisons of NWTS-3 with larger patient numbers and longer follow-up. The objective was to minimize selectivity and sampling errors by analyzing all eligible patients from both studies, whether or not their treatment was randomized. Results previously reported from NWTS-1–4 on DOX-induced CHF were updated to better understand the trade-off between risks and benefits.7
MATERIALS AND METHODS
NWTS-3 and NWTS-4 were multiinstitution, randomized clinical trials for patients age < 16 years at the time they were diagnosed with WT, with or without anaplasia, or with clear cell sarcoma or malignant rhabdoid tumor of kidney. Patients with tumors that were judged inoperable without pretreatment were ineligible. Details of the eligibility criteria, diagnoses, staging, treatment, and therapeutic outcomes have been published.3, 9–11 Patients with Stage II/FH disease were randomized on NWTS-3 between no flank RT or 20-Gy RT and between treatment with DOX or without DOX in a factorial design. Patients with Stage III/FH disease were randomized between 10-Gy RT or 20-Gy RT and between treatment with DOX or without DOX. For NWTS-4, the use of RT and DOX was determined by histology and stage. Treatment modality (single vs. divided dose) and duration (approximately 6 months vs. 15 months) were randomized in certain patient subgroups; however, because neither affected outcome for patients with FH disease, these factors were not considered.9, 11 The analyses included patients whose treatment assignment was randomized and patients who were followed. The latter were eligible patients who were not randomized for various reasons but who were treated on protocol regimens and had the same requirements for data submission as randomized patients.3 Protocol deviations were allowed if they were due to changes in disease stage or histology.
Between October, 1979 and August, 1986, 789 eligible patients with Stage II–III/FH WT were entered on NWTS-3. Among these, 29 patients were excluded for lack of baseline surgery or pathology records, and 5 patients were excluded because they had tumors in a solitary or fused kidney. Between August, 1986 and September, 1994, 1079 eligible patients with Stage II–III/FH WT were entered on NWTS-4. Among these, 54 patients were excluded for lack of baseline records, and 11 patients were excluded because they had tumors in a solitary or fused kidney. This left 755 patients on NWTS-3 and 1014 patients on NWTS-4 for statistical analysis. The NWTS protocols and informed consent documents were approved by each Institutional Review Board, and informed consent was obtained from the parents of all patients before participation in this study.
Treatment factors were evaluated using RFS and OS as endpoints. RFS considers the time to the first of recurrence, metastasis, progression, or death without disease; and OS considers the time to death. All deaths were counted in both endpoints, because the intent was to evaluate both therapeutic and toxic effects of RT and DOX. A third endpoint, local recurrence, was defined as an initial recurrence in the renal fossa (flank), elsewhere in the abdomen (exclusive of the contralateral kidney), or in the pelvis. All such recurrences also were counted as failures in the RFS analyses.
Sources of Data
Surgical checklists and operative notes submitted from over 200 participating institutions were reviewed and edited by NWTS staff and surgeons. Pathology checklists, narratives, and slides were reviewed by the NWTS Pathology Center to confirm FH WT.1 RT doses were abstracted from checklists by staff and were reviewed by NWTS radiation oncologists. Sites of recurrent or progressive disease and causes of death were abstracted by staff from clinical records and were reviewed by NWTS oncologists. Disease stage was assigned by the institution, usually after input from the Pathology Center.
The analysis of CHF incidence used additional follow-up and corrections to the records of patients from NWTS-1–4.7 The cohort of patients who first received DOX for treatment of recurrent disease, originally restricted to NWTS-1–3, was augmented with patients who had recurrent disease from NWTS-4. Inclusion/exclusion criteria for patients and CHF outcomes were as described previously.7
The time to event distributions and standard errors were estimated by actuarial methods.12, 13 Differences among patient subgroups were evaluated using the log-rank test.14, 15 Estimates of relative risk (RR) were based on the Cox model.16 Statistical adjustments of tests and estimates were effected by stratification on potential confounding factors.
Patients were classified according to whether or not they received DOX during initial therapy and by the RT dose to the renal fossa: none (0 Gy), low (0.1–14.9 Gy), and high (≥ 15 Gy). Substantial treatment variation on NWTS-3 reflected the fact that the DOX and RT doses were randomized for most patients (73%). Among the patients with Stage II disease, 41% received DOX, and 42% received RT, most at high doses. Among patients with Stage III disease, 64% received DOX, and 98% received RT, half at high doses and half at low doses. Among nonrandomized patients, 59% of patients who were treated with DOX received low-dose RT, and 35% received high-dose RT; whereas, among patients who were not treated with DOX, the percentages who received low-dose and high-dose RT were 31% and 62%, respectively. On NWTS-4, by contrast, the use of DOX and RT was determined largely by disease stage: Among patients with Stage II disease, 98% received no RT and no DOX; whereas, among patients with Stage III disease, 92% received DOX and low-dose (10 Gy) RT. The few patients who, although they were treated on a protocol regimen, did not receive the regimen specified for their disease stage usually had a change in stage or histology after treatment commenced.
The relative frequency of the 2 disease stages differed between studies, with equal numbers of patients with Stage II disease (n = 378) and Stage III disease (n = 377) on NWTS-3 but with larger numbers of patients with Stage II disease (n = 580) than Stage III disease (n = 434) on NWTS-4. This difference was due primarily to the upstaging to Stage II, as permitted by protocol, of > 100 patients on NWTS-4 who had tumor in the renal vein or sinus beyond the renal hilar plane but who, otherwise, satisfied Stage I criteria. In addition, a larger fraction of patients with Stage III disease on NWTS-4 (19%) than on NWTS-3 (10%) received preoperative treatment. Therefore, as many as 50 patients with Stage III disease may have been excluded as ineligible on NWTS-4 who would not have been excluded on NWTS-3. The exact number is uncertain, because posttreatment and pretreatment stages are not comparable.
Postrecurrence therapy also differed between NWTS-3 and NWTS-4. Retrieval therapy data were reviewed for 185 patients with recurrent disease of all stages from NWTS-3 and for 212 patients with recurrent disease of all stages from NWTS-4. The percentages of patients who received chemotherapy in each study were as follows: etoposide, 28% versus 77%; cisplatin, 31% versus 12%; and ifosfamide, 13% versus 52% for NWTS-3 versus NWTS-4, respectively.
Outcomes by Stage for NWTS-3 vs. NWTS-4
Table 1 presents the RFS and OS rates at 8 years from diagnosis. The risk of recurrence was lower on NWTS-3 for patients with Stage II disease, whereas the risk was lower for patients with Stage III disease on NWTS-4. Survival outcomes were the same for patients with Stage II disease, whereas survival improved on NWTS-4 for patients with Stage III disease. Patients on NWTS-4 who were assigned Stage II by their institution, but who actually had Stage I disease according to pathology and surgical checklists, had a 20% reduction in the risk of recurrence compared with patients who had Stage II disease according to strict NWTS-3 criteria; however, this difference was not statistically significant.
|Disease stage||No. of patients||Recurrence-free survival||Overall survival|
|No. of recurrences||% ± SE||No. of deaths||% ± SE|
|Stage II||378||44||89.5 ± 1.6||31||93.2 ± 1.3|
|Stage III||377||75||80.2 ± 2.1||50||87.0 ± 1.8|
|Stage II||580||95||83.6 ± 1.0||36||93.8 ± 1.1|
|Stage III||434||47||88.9 ± 1.5||30||93.0 ± 1.3|
The differences in RFS and OS between NWTS-3 and NWTS-4 were explained largely by differences in treatment. Ninety-five recurrences were observed among patients on NWTS-4 with Stage II disease. The number of recurrences that were expected, assuming no true difference in the recurrence rate between studies, was 80.6 recurrences (P = 0.012) before statistical adjustment for DOX and RT and 88.7 recurrences (P = 0.16) after adjustment. Forty-seven disease recurrences were observed in patients with Stage III disease on NWTS-4, compared with 66.1 expected recurrences (P = 0.0005) before adjustment and 54.3 expected recurrences (P = 0.06) after adjustment. There was no difference in the survival of patients with Stage II disease, with or without adjustment. Among patients with Stage III disease, 30 deaths were observed on NWTS-4 compared with 42.2 expected deaths (P = 0.006) before adjustment and 33.1 expected deaths (0P =.30) after adjustment.
Effects of DOX on Disease Recurrence and Mortality
Part A of Table 2 presents the RFS and OS estimates ± standard errors at 8 years. For patients with Stage II disease, the percentage of patients who developed recurrent disease or who died within 8 years of diagnosis were 11% with DOX and 15% without DOX, whereas the respective mortality rates were 9% and 6%. For patients with Stage III disease, 13% of those who received DOX and 26% of those who did not receive DOX developed recurrent disease, whereas the respectively mortality rates were 8% and 17%. These raw percentages, however, do not account for confounding caused by the correlation of DOX with RT and study. They are presented only to provide information about the absolute magnitude of the risks of treatment failure. Statistical adjustment for confounding is accomplished best through analysis of RR.
|Treatment group||No. of patients||Recurrence-free survival||Overall survival|
|No. of recurrences||Mean % ± SE||No. of deaths||Mean % ± SE|
|Part A: All available patients from NWTS-3 and NWTS-4|
|DOX||160||19||89.3 ± 2.5||16||91.0 ± 2.3|
|No DOX||798||120||85.3 ± 1.3||51||94.1 ± 0.9|
|DOX||673||85||87.1 ± 1.3||57||91.7 ± 1.1|
|No DOX||138||37||73.8 ± 3.8||23||83.1 ± 3.2|
|Part B: Randomized patients from NWTS-3 who were treated according to randomization|
|DOX||117||15||88.7 ± 2.9||12||91.3 ± 2.6|
|No DOX||113||11||91.9 ± 2.6||8||94.5 ± 2.2|
|DOX||130||20||84.3 ± 3.2||15||88.9 ± 2.8|
|No DOX||118||30||74.3 ± 4.1||20||82.7 ± 3.5|
The effects of DOX on disease recurrence and death are summarized in Table 3. For each endpoint, RRs are shown without adjustment, with adjustment for RT only, and with adjustment for both RT and study. DOX had no statistically significant effect on the risk of disease recurrence or mortality for patients with Stage II disease. For patients with Stage III/FH WT, DOX was associated with reductions > 50% in the rates of general and local recurrence, even after adjusting for RT or for RT and study, respectively. Twenty-eight local recurrences occurred among 673 patients who received DOX, and 12 local recurrences occurred among 138 patients who did not receive DOX. The same reductions in the risk of local recurrence (RR, 0.42 vs. 0.47) and general recurrence (RR, 0.49 vs. 0.45) were observed when the analysis was restricted to subgroups of patients who received 10-Gy RT versus 20-Gy RT, respectively. Although the survival results tended in the same direction, suggesting a reduction ≥ 32% in mortality, they were statistically significant only in the unadjusted analysis.
|Stage||Relative risk (P value)|
|Local recurrence||General recurrence||Death|
|No adjustment||RT alone||RT and study||No adjustment||RT alone||RT and study||No adjustment||RT alone||RT and study|
|II||0.54 (0.18)||0.76 (0.59)||0.90 (0.84)||0.73 (0.20)||0.82 (0.46)||1.02 (0.94)||1.39 (0.25)||1.31 (0.41)||1.39 (0.36)|
|III||0.45 (0.019)||0.40 (0.011)||0.43 (0.037)||0.44 (0.001)||0.47 (0.007)||0.56 (0.009)||0.51 (0.006)||0.61 (0.060)||0.68 (0.173)|
Parallel RFS and OS analyses were conducted using only randomized NWTS-3 patients who were treated as randomized (Table 2, Part B). For patients with Stage II disease, the RR of recurrent disease associated with DOX treatment was 1.32 (P = 0.48), whereas the RR for survival was 1.48 (P = 0.39). For patients with Stage III disease, the RR for local recurrence was 0.57 (P = 0.29; 6 events among 130 patients who were treated with DOX; 9 events among 118 patients who were not treated with DOX), the RR for general recurrence was 0.57 (P = 0.05), and the RR for death was 0.69 (P = 0.27).
Figure 1 graphs the cumulative risk of CHF for patients who received DOX as primary treatment by study and by time since WT diagnosis. NWTS-1 and NWTS-2 were combined (NWTS-1,2), because only 17 patients on NWTS 1 received DOX initially. NWTS-3 and NWTS-4 also were combined (NWTS-3,4), because the results were similar, and patients on NWTS-4 had less follow-up. The median follow-up for 298 patients on NWTS-1,2 was 20.7 years, with 152 patients (51%) who were followed beyond 20 years. The cumulative risk of CHF at 20 years was 4.5% ± 1.8% based on 10 events; the 11th and last event occurred at 24.3 years. The median follow-up for 2376 patients on NWTS-3,4 was 10.9 years, with 179 patients (8%) who were followed beyond 20 years. The risk of CHF at 20 years was 1.2% ± 0.3% based on 16 events, and the last event was at 15.6 years with 570 patients at risk. The risk was greater for patients with left-sided tumors (Fig. 2). Restricting attention to patients who had known unilateral disease, 8 CHF events occurred among 133 patients who had left-sided tumors compared with 2 CHF events among 136 patients who had right-sided tumors on NWTS-1,2. In NWTS-3,4, the corresponding fractions were 9 CHF events among 1026 patients with left-sided tumors and 2 CHF events among 959 patients with right-sided tumors. In a second cohort of patients who were treated with DOX for the first time at disease recurrence (Fig. 3), the risk of CHF at 20 years for 126 patients with recurrent disease on NWTS-1,2 was 22.9% ± 7.4% based on 10 events, with the last event at 18.2 years. For patients on NWTS-3,4 the risk was 3.2% ± 1.4% based on 6 events among 268 patients, with the last at 8.4 years, when 108 patients remained alive and under observation. The CHF risk also was elevated for patients in this second cohort who had left-sided tumors (Fig. 4).
Very few patients experienced CHF who were included in the preceding analysis of DOX efficacy for patients on NWTS-3,4 with Stage II–III/FH WT. Only one such patient received DOX for initial treatment. She developed CHF at 1.3 years and was alive and well at 19.1 years (2 other patients with Stage II–III/FH WT developed CHF but were excluded from analysis because of preoperative therapy or lack of baseline records). Five patients who received DOX first for the treatment of recurrent disease and, thus, were categorized as no DOX in the preceding analysis, developed CHF; and two of those patients consequently died.
The results of this reanalysis of NWTS-3,4 data are in broad agreement with results reported for the NWTS-3 randomized trial but substantially strengthen the statistical precision through longer follow-up and greater patient numbers. When the analysis was restricted to randomized patients on NWTS-3 who were treated as randomized, the RFS and OS rates were almost exactly as reported originally using intent-to-treat analyses in which all randomized patients were analyzed according to their assigned regimen.3 The RFS and OS rates for patients with Stage II disease were slightly lower if they received DOX, and the differences were not statistically significant. DOX reduced the risk of recurrence by ≈ 40% for randomized patients with Stage III disease, and the difference was of borderline statistical significance. The parallel survival comparison, with a 30% reduction in mortality, failed to achieve statistical significance.
Inclusion of data for nonrandomized (for DOX) patients greatly increased patient numbers. The number of patients with Stage II disease treated without DOX increased seven-fold, and the number of patients with Stage III disease treated with DOX increased five-fold, with corresponding increases in the numbers of recurrences and deaths. This substantially improved the precision (standard errors) of estimated RFS and OS percentages for one of the two treatment groups used in each comparison. Thus, for example, for patients with Stage III disease who were treated with DOX, the RFS estimates changed from 84% ± 3% for the strictly randomized estimate to 87% ± 1% when all available data were used (Table 2). The OS estimates for these same patients changed from 89% ± 3% to 92% ± 1%. The increases in patient numbers in the other two subgroups were much more modest, because they involved primarily followed patients from NWTS-3. The analyses that were restricted to randomized patients and the analyses that used all available data both estimated 89% ± 3% RFS and 91% ± 3% OS for patients with Stage II disease who were treated with DOX and estimated 74% ± 4% RFS and 83% ± 3% OS for patients with Stage III disease who were treated without DOX (Table 2).
The hazards of scientific inference using observational data and historic comparisons are well known.17 In the current context, they include the fact that DOX was correlated with RT, with retrieval therapy, and possibly with other supportive care measures that affected outcomes. Nonrandomized patients on NWTS-3 who received DOX may have been selected on the basis of a better or worse prognosis a priori. Statistical adjustment can account for such imbalances to the extent that they are known. The major problem with nonrandomized comparisons is that it never can be ruled out that there may be confounding factors on which no data are available.
Because there were no major changes in prerecurrence therapy between NWTS-3 and NWTS-4 other than the use of DOX and RT, adjustment for study probably is not needed when examining the RR for disease recurrences. Conversely, changes between studies in retrieval therapy suggest that results for OS should be adjusted for both RT and study. Because little treatment variation occurred in NWTS-4, results adjusted for study are based predominately on within-NWTS-3 comparisons. Thus, it is not surprising that they most resemble the results from the randomized NWTS-3 analyses.
NWTS-3,4 enrolled a major fraction of the entire North American incidence of patients with WT. Eligibility criteria for the two studies were the same. Nevertheless, three important issues regarding selection bias remain. First, some patients assigned to Stage II on NWTS-4 had Stage I disease according to NWTS-3 criteria. Second, treatment assignments for nonrandomized patients on NWTS-3 may have been influenced by patient prognosis. Third, proportionally more patients on NWTS-4 than on NWTS-3 had Stage III disease after preoperative therapy and, thus, were ineligible. Unfortunately, it was not possible to identify and exclude patients on NWTS-4 who were upstaged deliberately. Historically, some 10–15% of patients on NWTS-3 who were assigned Stage II satisfied Stage I criteria according to surgery and pathology data.18 Such misclassification affects patients both on NWTS-3 and on NWTS-4. Deliberately upstaged patients on NWTS-4 may be expected to have a slightly more favorable prognosis; therefore, their removal from the no-DOX subgroup would increase slightly the unadjusted RRs for local and distant recurrence in the direction of no effect (Table 2). Similarly, it may be believed that patients on NWTS-3 who were selected for more intensive treatment with DOX had a worse prognosis than patients who were assigned to the no-DOX subgroup. However, it was found that DOX protected against disease recurrences. In both instances, therefore, accounting for possible selection bias would reinforce the results and conclusions. Conversely, it may be suspected that patients who were selected for preoperative therapy had a poorer prognosis compared with patients who were not selected. The NWTS-4 results for patients with Stage III/FH disease may have been worse and, thus, closer to the results for patients in NWTS-3 had those patients remained eligible for the study. The estimated number of such patients (n = 50) and the low rates of recurrence, however, suggest that any additional recurrences that resulted from this phenomenon would be relatively few and would be unlikely to affect the main conclusions. The RR adjusted for both RT and study would be unaffected.
Results from the randomized trial and from the current observational study both support the conclusion that treatment with DOX offers no benefit for patients with Stage II/FH WT. Both studies found a statistically insignificant increase in mortality for patients who were treated with DOX. The observational data suggest a slight RFS advantage for DOX, the randomized data suggest a slight disadvantage; however, neither RR was statistically significant. The retrieval of patients who were treated with DOX was particularly poor, which may be expected for patients who have disease that has developed resistance to a principal therapeutic agent. This also may be a statistical aberration in view of the small numbers of patients and disease recurrences.
Results from the randomized trial and the current analysis do suggest that DOX is efficacious for the treatment of patients with Stage III/FH WT. The RFS rate of 87% for patients who received DOX in the observational analysis is < 1 standard error from the 84% estimated in the randomized analysis: well within the bounds of statistical variation. However, it results in a change in the unadjusted RR for disease recurrence from 0.57 (P = 0.05) to 0.44 (P = 0.001). Adjustment for RT changes the latter estimate only slightly. The RR of 0.56 (P = 0.009) adjusted for both RT and study is nearly identical to the RR from the randomized trial but is estimated more precisely due to inclusion of followed patients from NWTS-3. The RR for local recurrence is affected little by adjustment and indicates that DOX lowers the risk substantially.
The survival results are more difficult to interpret. The advisability of adjusting for study leads to an RR estimate of 0.68 (P = 0.17), nearly identical to the RR of 0.69 (P = 0.27) from the randomized analysis and still not statistically significant. Hence, it cannot be concluded definitively that the survival outcome is improved by using DOX as part of the initial treatment regimen instead of holding it in reserve for use as a retrieval agent. Patients who have survived free from disease recurrence without its use, however, also are free from the long-term risks of anthracycline-induced CHF and secondary malignancies.
It turned out that no deaths attributable to CHF occurred among patients who received DOX for initial treatment and who met the criteria for inclusion in the current analysis of patients with Stage II–III/FH WT. Two CHF deaths occurred among patients who first were treated with DOX at the time they developed recurrent disease. The failure to demonstrate a statistically significant survival advantage for frontline DOX therapy, therefore, cannot be ascribed currently to a higher risk of CHF-associated mortality after such treatment. The cumulative risk of CHF now estimated for all patients from NWTS-3,4, 1.2% at 20 years, is consistent with that shown in Figure 3 from the original report.7 The cumulative risk of 4.4% reported for NWTS-1–4 in the abstract of that report was over-influenced by a single CHF event that occurred when few patients still were under observation (see Fig. 2 from the original report) and that was included mistakenly in the analysis before confirmation of digoxin treatment. (This patient from NWTS-2 reportedly had undergone heart transplantation and subsequently died of disseminated candidiasis, but relevant clinical records could not be obtained.) It is now estimated that only patients from NWTS-1,2 have such high risk: 4.5% at 20 years (Fig. 1), and 5.1% if the unconfirmed CHF event is included. These patients generally received more abdominal RT and had DOX administered in larger doses compared with the patients on NWTS-3,4.7
The cumulative incidence curves shown in Figures 2 and 4 reinforce the finding from Table 3 of the original report that abdominal irradiation for left-sided tumors, but not right-sided tumors, increased the risk of CHF markedly.7 Pinkel and colleagues,19 using case reports and the observation that left-side abdominal radiation fields often included the lower portion of the heart, noted previously that children with left-sided WT were at particular risk of DOX-induced CHF. Protection of the heart during RT to the left hemiabdomen should decrease the risk of CHF in patients who receive DOX. Because current treatment protocols do not require the RT field to extend to the diaphragm in all patients, the risk of CHF may be lower than in the past.
In summary, the results of this study reinforce the results of the NWTS-3 randomized trial vis-à-vis local and distant recurrences that were used to recommend DOX for the treatment of patients with Stage III/FH WT. Another randomized trial with a substantially greater sample size will be needed to prove definitively that frontline therapy with DOX improves survival.
The authors thank the investigators in the Pediatric Oncology Group and the Children's Cancer Group as well as the health professionals who managed the care of children enrolled in the studies conducted by the National Wilms Tumor Study Group.
- 8Relationship between dose schedule and charges for treatment on National Wilms' Tumor Study-4. A report from the National Wilms' Tumor Study Group. J Natl Cancer Inst Monogr. 1995; 19: 21–25., , , et al.
- 13The natural duration of cancer. In: United Kingdom Ministry of Health. Reports on public health and medical subjects. Volume 33. London: His Majesty's Stationery Office, 1926: 1–26..
- 16Regression models and life-tables (with discussion). J R Stat Soc B. 1972; 34: 187–220..