1616 Factors Affecting the Risk of Contralateral Wilms Tumor Development A Report from the National Wilms Tumor Study Group Max J. Coppes, M.D., Ph.D.1 Muoi Arnold, M.D.2 J. Bruce Beckwith, M.D.2 Michael L. Ritchey, M.D.3 Giulio J. D’Angio, M.D.4 Daniel M. Green, M.D.5 Norman E. Breslow, Ph.D.6 1 Children’s Cancer Program, Alberta Children’s Hospital and Tom Baker Cancer Centre, and Departments of Oncology and Paediatrics, University of Calgary, Calgary, Alberta, Canada. 2 Department of Pathology and Human Anatomy and Division of Pediatric Pathology, Loma Linda University, Loma Linda, California. 3 Department of Surgery, University of TexasHouston Medical School, Houston, Texas. 4 Department of Radiation Oncology, University of Pennsylvania, Philadelphia, Pennsylvania. 5 Department of Pediatrics, Roswell Park Cancer Institute, and the School of Medicine, State University of New York at Buffalo, Buffalo, New York. 6 Department of Biostatistics, University of Washington, Seattle, Washington. Supported in part by research grants CA54498 and CA42326 from the U.S. Public Health Service. The authors thank the investigators of the Pediatric Oncology Group (POG) and the Children’s Cancer Group (CCG) and the many pathologists, surgeons, pediatricians, radiation oncologists, nurses, and other health care professionals who managed the children entered on the National Wilms Tumor Study Group trials. Dr. Max J. Coppes is a Clinical Investigator of the Alberta Heritage Foundation for Medical Research. Address for reprints: Max J. Coppes, M.D., Ph.D., Children’s Cancer Program, Alberta Children’s Hospital, 1820 Richmond Road, SW, Calgary, Alberta, Canada T2T 5C7. Received July 7, 1998; revision received October 15, 1998; accepted December 15, 1998. © 1999 American Cancer Society BACKGROUND. Approximately 1% of children with unilateral Wilms tumor develop contralateral disease. The authors assessed the demographic and histologic features associated with metachronous bilateral Wilms tumor (BWT). METHODS. Characteristics of all children registered on the first four National Wilms Tumor Studies (NWTS) were recorded. The primary endpoint for evaluation was the first appearance of Wilms tumor in the remaining kidney. The cumulative risk of contralateral disease as a function of time since initial presentation was calculated as 1 minus the Kaplan–Meier estimate of remaining free of contralateral disease. A matched case control study was conducted to determine whether the presence and type of nephrogenic rests (NRs) were associated with metachronous BWT. RESULTS. Fifty-eight of 4669 registered children developed metachronous BWT; 38 of 2445 females (expected, 30.2) versus 20 of 2224 males (expected, 27.8) (P 5 0.04) developed BWT. The cumulative incidence of contralateral disease 6 years after initial diagnosis decreased from greater than 3% in the first NWTS to approximately 1.5% in the three subsequent studies (P 5 0.08). Patients with NRs had a significantly increased risk of metachronous BWT. This was particularly true for young children (20 of 206 age , 12 months compared with 0 of 304 age . 12 months). Data from the matched case control study confirmed the increased relative risk associated with young age and the presence of NRs. CONCLUSIONS. Children younger than 12 months diagnosed with Wilms tumor who also have NRs, in particular perilobar NRs, have a markedly increased risk of developing contralateral disease and require frequent and regular surveillance for several years. Surveillance is also recommended for those with NRs who are diagnosed after the age of 12 months. Cancer 1999;85:1616 –25. © 1999 American Cancer Society. KEYWORDS: pediatric oncology, kidney, pediatrics, chemotherapy, nephrogenic rests, risk factors, Wilms tumor. P atients with bilateral Wilms tumor (BWT) present a therapeutic challenge to clinicians. Unlike those with unilateral Wilms tumor, the incidence of renal insufficiency following otherwise successful therapy is high in patients with bilateral disease. Recently, investigators from the National Wilms Tumor Study Group (NWTSG) reported that 30 of 391 (7.7%) children with bilateral disease at onset (synchronous BWT) registered on the first four NWTSG studies ultimately developed renal failure, most often because bilateral nephrectomy became necessary.1 The risk of renal failure increases with time. The use of effective preoperative chemotherapy seems to have decreased the risk of renal failure by facilitating parenchymal sparing surgery, as illustrated by a decreased frequency of Risk Factors for Metachronous Bilateral Wilms Tumor/Coppes et al. renal failure in successive NWTSG studies, a function of time: 8 of 65 (12.3%) patients in NWTS study 1 (NWTS-1) and NWTS-2 together, 15 of 158 (9.5%) patients in NWTS-3, and 7 of 168 (4.2%) patients in NWTS-4 registered up to July of 1993.1 In contrast to children with synchronous BWT, patients with metachronous BWT initially present with unilateral Wilms tumor and subsequently develop contralateral disease. A recent literature review indicates overall survival rates of 49.1% and 47.2% at 5 years and 10 years, respectively.2 In these patients, treatment is dictated by the extent of disease and histopathologic characteristics at initial presentation and invariably includes unilateral nephrectomy, which is followed by adjuvant chemotherapy with or without radiation therapy.3 Parenchymal sparing surgery is then limited to salvaging nephrons of the second affected kidney, because the first kidney is removed at the time of initial diagnosis. The incidence of renal failure for this group of children, 6 of 20 (30%) patients in NWTS-1 and NWTS–2 and 3 of 23 (13%) patients in NWTS-3,1 is higher than that observed in children with synchronous BWT. The reported incidence of renal failure for the 17 metachronous BWT patients registered on NWTS-4 was zero, but the duration of follow-up was short. Initial parenchymal sparing surgery could be justified in a group of children with unilateral Wilms tumor if “. . . we could accurately predict which patients with unilateral disease are at risk for subsequent contralateral tumors.”1 One of the features associated with metachronous BWT is the presence of nephrogenic rests (NRs) in the normal tissue of the kidney with Wilms tumor.4 Previously, the presence of NRs usually was discovered during histopathologic examination following nephrectomy. In the past few years, however, we have come to better recognize the radiologic appearance of NRs.5,6 On magnetic resonance imaging, for example, NRs are best visualized evident on T1-weighted images post-gadolinium. Unlike Wilms tumors, which typically show inhomogeneous enhancement, NRs generally remain homogeneously hypointense.7 Because magnetic resonance imaging is not readily available for all patients with Wilms tumor, at present, histopathologic examination following nephrectomy remains an important tool to assess the presence of NRs. It would be helpful if additional clinicopathologic features associated with metachronous BWT could be identified. Here, we report on such factors: the combination of young age and the presence of NRs, gender, and initial chemotherapy. 1617 PATIENTS AND METHODS Study Population All children registered between October 1969 and December 1992 in the randomized or followed categories, as defined elsewhere,8,9 on the first four NWTSG studies were included in this review. Tumor specimens were reviewed centrally to confirm the diagnosis of Wilms tumor of either favorable or anaplastic histology. Patients with clear cell sarcoma, rhabdoid tumor of the kidney, renal cell carcinoma, and other renal neoplasms were excluded. Also excluded were those with bilateral disease at onset (synchronous BWT), those with a fused kidney or tumor originating from an extrarenal site, those diagnosed after age 15 years, and patients who received preoperative treatment with chemotherapy and/or radiation therapy. A total of 4669 children were eligible for analysis. According to the inclusion criteria, all had unilateral disease at diagnosis and were treated with nephrectomy and adjuvant chemotherapy with or without additional radiation therapy. Follow-Up and Evaluation Most patients (4643; 99.4%) were treated according to one of the standard NWTS regimens,3,8 –10 which included either single (dactinomycin [AMD] or vincristine [VCR]), double (AMD 1 VCR), triple (AMD 1 VCR 1 doxorubicin[DOX]), or quadruple (AMD 1 VCR 1 DOX 1 cyclophosphamide [CMD]) chemotherapy. Therefore, only 26 patients (0.6%), all of whom had been treated other than with one of the standard NWTS regimens, were excluded from analyses that used chemotherapy as a variable. Clinical details at diagnosis and information on treatment outcome were reported on check lists, narratives, and flow sheets submitted by the pediatric oncologist, surgeon, and pathologist involved in the actual care of the patients. Patients were followed on study indefinitely from the time of initial diagnosis. Median follow-up for the 4669 children was 6.8 years from initial diagnosis. All but 209 of the 4149 surviving patients were followed for at least 2 years. Estimation of the Cumulative Incidence of Contralateral Kidney Disease The primary endpoint for evaluation was the first appearance of Wilms tumor in the remaining kidney. In most cases, this was the first and only appearance of recurrent disease in an otherwise healthy child. In a few instances, however, the appearance of disease in the contralateral kidney followed earlier relapse in another site. The cumulative risk of contralateral disease as a function of time since initial presentation was calcu- 1618 CANCER April 1, 1999 / Volume 85 / Number 7 lated as 1 minus the Kaplan–Meier estimate11 of remaining free of contralateral disease. This estimate accounted for incomplete follow-up due to a patient’s death, loss to follow-up, or the end of the study. Contralateral disease incidence rates in different patient subgroups were compared by using the log-rank test statistic.12 Multiple regression analysis was conducted by using the Cox proportional hazards model.13 control study. All surgical records from the first surgery were reviewed, and contralateral exploration was recorded as “no exploration of the contralateral kidney,” “exploration without visualization,” and “visualization of the posterior surface.” Relative risk of contralateral involvement according to the degree of surgical exploration subsequently was calculated by using conditional logistic regression.15 Matched Case-Control Study Prior to 1980, renal lesions thought to be potential precursors of Wilms tumor, which are found in 30 – 40% of kidneys removed for Wilms tumor,4 were recorded as “blastemal rests.” Starting with NWTS-3, the term for these lesions changed to nephrogenic rests (NRs).4 In addition, the NWTSG Pathology Center started to recognize the importance of the localization of NRs in the renal lobe. Consequently, NRs were categorized as being perilobar (PLNRs) if they occurred at the surface of the renal lobe or intralobar (ILNRs) if the they occurred in the deep medulla, medullary cortex, renal sinus, or pelvicalyceal wall.4 NWTSG investigators have suggested previously that the presence of ILNRs or PLNRs is associated with a risk of subsequent tumor development in the contralateral kidney.4 Hence, a substudy of slides archived at the NWTSG Pathology Center was conducted to ensure a uniform evaluation of the NRs in tumor specimens collected throughout all four NWTSG studies. For each metachronous BWT case, four controls were selected at random from among children registered on the same NWTSG study. Each control had to have survived at least as long as it took for the corresponding case to develop contralateral disease. In other words, the four controls were selected at random from “risk sets” created during a Cox proportional hazards regression analysis with stratification on study.14 The list of metachronous BWT cases and randomly selected controls, identified only by NWTSG identification number, was submitted to the NWTSG Pathology Center for histopathologic reevaluation of the corresponding diagnostic slides. For the matched case-control study, odds ratios, which approximate relative risks of contralateral disease, were estimated from the matched sets by using conditional logistic regression.15 This produced estimates of relative risk comparable to those that would have been obtained from Cox proportional hazards analysis of the entire data set had the slides been reevaluated for all patients. Although there is some loss of efficiency, we nevertheless obtained 80% efficiency for testing the null hypothesis,15 having selected four controls for each study case. We also analyzed the extent of surgical exploration of the contralateral kidney in our nested case- RESULTS Among the 4669 children, 58 (1.2%) developed metachronous Wilms tumor. Thirty-eight were female, and 20 were male. In 50 patients, contralateral tumor development occurred prior to or simultaneously with any other relapse, whereas eight occurred after relapse to another site. Thirteen cases occurred in the first year following initial nephrectomy, 18 in the second year, 12 in the third year, 5 in the fourth year, 4 in the fifth year, and 5 in the sixth year. The last case occurred 13.1 years from diagnosis of the initial Wilms tumor. The actuarial percentage of patients who developed contralateral disease was 1.5% 5 years after initial diagnosis. However, the cumulative incidence of contralateral disease at 6 years from initial diagnosis decreased from greater than 3% in NWTS-1 to approximately 1.4% in NWTS-2 and remained in that same range in NWTS-3 and NWTS-4 (Fig. 1). There was some evidence that patients with an initial tumor localized to the kidney and completely resected, i.e., Stage I disease, had a higher incidence of contralateral kidney involvement than children with more advanced disease at diagnosis (Table 1). Further analysis, however, suggested that this was explained largely by treatment following initial diagnosis and by the association of stage and NRs rather than disease extent per se. Among 265 children treated with a single chemotherapeutic agent (AMD or VCR), 9 developed contralateral disease (expected, 3.6) compared with 38 of 2808 patients treated with VCR 1 AMD (expected, 35) and 10 of 1570 children treated with AMD 1 VCR 1 DOX (expected, 19.84; P 5 0.002). The cumulative incidence for children treated with four drugs (AMD 1 VCR 1 DOX 1 CMD) also appeared to be intermediate, although this result was based on only three cases occurring among 152 patients and therefore is statistically unreliable. Finally, children receiving radiation therapy to the flank as part of their treatment for the initial Wilms tumor were not found to have an incidence of subsequent contralateral tumor involvement that was significantly different from those who did not receive radiation therapy after initial nephrectomy: 22 of 2170 patients who received radiation therapy as part of treatment for their initial Wilms tumor developed metachronous BWT (expect- Risk Factors for Metachronous Bilateral Wilms Tumor/Coppes et al. 1619 FIGURE 1. Probability of contralateral tumor development for patients with unilateral Wilms tumor registered on the first four National Wilms Tumor Studies as a function of time. The total number of patients included per study, the number of patients who actually developed metachronous bilateral Wilms tumor, the number expected, and the relative risk (RR) with 95% confidence interval (CI) also are provided (P 5 0.08). TABLE 1 Observed versus Expected Number of Patients who Developed Contralateral Wilms Tumor and Estimated Relative Risk with 95% Confidence Intervals in Relation to Stage at Initial Diagnosisa Stage N Observed Expected RR 95% CI I II III IV 1813 1254 1026 576 33 12 9 4 23.6 15.8 12.5 6.1 1.0 0.54 0.51 0.47 0.28–1.05 0.25–1.07 0.17–1.33 N: number; RR: relative risk; CI: confidence interval. a P 5 0.03. ed, 26.6) versus 36 of 2499 children who did not receive radiation therapy (expected, 31.4; P 5 0.23). There was no evidence that the incidence of metachronous BWT was associated with histopathologic features of the initial Wilms tumor. Fifty-six of 4378 patients with favorable histology at initial diagnosis developed metachronous BWT (expected, 55.4) versus 2 of 291 with anaplastic histology at first presentation (expected, 2.6; P 5 0.71). Similarly, there was no indication that ethnicity influenced the occurrence of contralateral disease. Gender, however, was found to be a factor. The incidence was significantly higher in girls compared with boys: 38 of 2445 females developed contralateral disease (expected, 30.2) versus 20 of 2224 males (expected, 27.8; P 5 0.04). There was a pronounced decline in the risk of contralateral Wilms tumor development with increasing age at diagnosis (Fig. 2). The risk was particularly elevated among children ages , 1 year when the initial Wilms tumor was found. Such children accounted for 21 of 58 cases of metachronous BWT. The cumulative incidence of contralateral tumor development in this particular age group was 4% at 6 years from diagnosis (Fig. 2). Several syndromes predispose to the development of Wilms tumor. These include aniridia, Beckwith– 1620 CANCER April 1, 1999 / Volume 85 / Number 7 FIGURE 2. Probability of contralateral tumor development for patients with unilateral Wilms tumor in each age group as a function of time. The total number of patients included in each age group, the number of patients who actually developed metachronous bilateral Wilms tumor, the number expected, and the relative risk (RR) with 95% confidence interval (CI) also are provided (P , 0.001). Wiedemann syndrome, Denys–Drash syndrome, and hemihypertrophy. We therefore determined whether the incidence of metachronous BWT was higher among patients with these syndromes compared with children without congenital anomalies. Children with Beckwith–Wiedemann syndrome, hemihypertrophy, or congenital aniridia appeared to be at higher risk for metachronous BWT, whereas children with other congenital anomalies were not. Fifty-three of the 4602 children without aniridia or Beckwith–Wiedemann syndrome developed metachronous BWT (expected, 57.14), as did 2 of 32 children with aniridia (expected, 0.41) and 3 of 35 patients with Beckwith–Wiedemann syndrome (expected, 0.45). Similarly, 5 of 144 children with hemihypertrophy developed metachronous BWT (expected, 1.83) compared with 53 of 4525 who did not have hemihypertrophy (expected, 56.17). The numbers of patients with anomalies were judged to be too small for reliable statistical inference. Thus, specific congenital anomalies were not among the predictors considered in the multiple regression analysis. Because the coding of NRs changed in 1980, as explained above, the association between NRs and metachronous BWT was determined separately for patients registered on NWTS-1 and NWTS-2 and those registered on NWTS-3 and NWTS– 4 (Table 2). Only in the latter group were we able to determine the incidence of metachronous BWT for those with intralobar nephrogenic rests (ILNRs) only or with perilobar nephrogenic rests (PLNRs) only (Table 2B). Regardless of the NWTSG study, patients with unilateral Wilms tumor and NRs had a significantly increased risk of developing metachronous BWT. In a further evaluation of the joint effects of NRs and age at diagnosis on predicting the occurrence of contralateral disease, patients were simply classified with regard to whether or not NRs had been noted. Patients who had inadequate tissue submitted for NR determination were classified as “unknown.” Table 3 shows the high risk of contralateral tumor development in patients with NRs whose Wilms tumor was diagnosed under the age of 12 months. Among those with NRs, 20 cases of metachronous BWT were reported, whereas no cases were observed among 304 children that did not have NRs at diagnosis (Table 3). The relative risk estimate in this group of patients is infinite (Table 4). Beyond 1 year of age, the relative risk associated with positive rest status stabilized at about 5. The actuarial estimate at 6 years was 15% for those with both PLNRs and ILNRs, 10% for those with PLNRs only, 5% for those with ILNRs only, and 0% for those without NRs (Fig. 3). Risk Factors for Metachronous Bilateral Wilms Tumor/Coppes et al. 1621 TABLE 2 Observed versus Expected Number of Patients with Metachronous Bilateral Wilms Tumor in Relation to Nephrogenic Rest Status with Estimated Relative Risk and 95% Confidence Intervalsa Rest status N Observed Expected RR 95% CI Negative Positive 841 231 3 15 14.1 3.9 1.0 18.3 5.3–63.4 Negative PLNR ILNR Both 1855 512 423 110 7 11 8 8 21.8 5.9 4.9 1.4 1.0 5.82 5.06 18.54 2.25–15.0 1.83–14.0 6.72–51.1 A B N: number; RR: relative risk; CI: confidence interval; PLNR: perilobar nephrogenic rest; ILNR: intralobar nephrogenic rest. a (A) Patients registered on National Wilms Tumor Study 1 (NWTS-1) and NWTS-2 (P , 0.0001). Thirty-seven patients were excluded from this particular analysis because their rest status was unknown. (B) Patients registered on NWTS-3 and NWTS-4 (P , 0.0001). Six hundred and sixty patients were excluded because their exact rest status could not be determined. TABLE 3 Observed versus Expected Number of Patients with Metachronous Bilateral Wilms Tumor in Relation to Nephrogenic Rest Status and Age at Diagnosis (P < 0.0001) Rest status Age (yrs) N Observed Expected Negative Negative Negative Negative ,1 $1 and ,2 $2 and ,4 $4 304 407 887 1098 0 2 4 4 3.98 5.34 11.04 13.38 Positive Positive Positive Positive ,1 $1 and ,2 $2 and ,4 $4 206 292 408 370 20 8 8 6 2.61 3.67 5.13 4.33 Unknown Unknown Unknown Unknown ,1 $1 and ,2 $2 and ,4 $4 106 124 241 226 1 0 4 1 1.23 1.57 3.02 2.69 4669 58 58 Total N: number. In an attempt to ascertain those factors that were most important in predicting contralateral Wilms tumor development, a multiple regression analysis using the Cox proportional hazards model was conducted among the 3949 patients who were treated with a standard NWTSG chemotherapy regimen for whom rest status was known. Fifty-one of 58 metachronous BWT cases occurred in this subgroup. The association between metachronous BWT and age at initial diagnosis noted in Table 3 was accommodated by including an interaction term for the logarithm of the age in months 1 1. Extrapolation from this model suggests that children whose Wilms tumor is diagnosed at birth in the presence of NRs would have 15 times the chance of developing metachronous BWT compared with those who do not have NRs. According to the model, the relative risk falls to 6 by age 4 years and to TABLE 4 Relative Risk for Metachronous Bilateral Wilms Tumor by Age Group in Patients who were Rest Positive at First Presentation Age (yrs) <1 >1 and <2 >2 and <4 >4 RR ` 5.7 4.5 4.6 RR: relative risk. 3 by age 16 years. Even accounting for these strong effects, girls still have twice the risk of boys. Matched Case-Control Study The main purpose of the matched case-control study was to determine systematically whether the presence of NRs was correlated with the development of metachronous BWT. With four controls matched to each case of metachronous BWT, there were 5 3 58 5 290 patients, one of whom was excluded later because the initial Wilms tumor originated from an extrarenal site. The presence of PLNRs or ILNRs both increased the risk of contralateral Wilms tumor development (Table 5). Patients with NRs of both types had the most dramatic risk of metachronous BWT (matched odds ratio for PLNRs 1 single ILNR 5 8.23; for PLNRs 1 multiple ILNRs 5 7.56). Data from this substudy confirmed our previous observation regarding the interactive effect of age at initial diagnosis and the presence of NRs. The diagnosis of PLNRs in very young children (,12 months of age) carried the most dramatic risk of contralateral disease (Table 5B). Evaluation of the degree of surgical exploration at the time of initial surgery (Tables 6 and 7), demonstrated a clear temporal change in surgical practice. Relatively few patients in NWTS-1, whether metachronous BWT patients or controls, had visualization of the posterior surface (Table 6). By contrast, virtually all 1622 CANCER April 1, 1999 / Volume 85 / Number 7 FIGURE 3. Probability of contralateral tumor development by nephrogenic rest status and type for children less than 12 months of age with unilateral Wilms tumor. PLNR: perilobar nephrogenic rest; ILNR: intralobar nephrogenic rest; NEG, negative. patients registered on NWTS-4 had such visualization. Because the case control analysis automatically adjusted for study number, the data shown in Tables 6 and 7 provide no reason to believe that surgical exploration of the contralateral kidney affected the risk of subsequent contralateral tumor involvement. Finally, Table 8 shows the results of the conditional logistic regression analysis. The model contained terms for the presence of PLNRs, age at initial diagnosis, the interaction between age and positive PLNRs, gender, and the number of chemotherapeutic agents used in the treatment of the first Wilms tumor. Once again, age was treated as a continuous variable by taking the logarithm of age in months 1 1. The relative risk estimated for the presence of PLNRs associated with Wilms tumor diagnosed at age 0 was 8.8 relative to that for a child without PLNRs. According to the model, the relative risk associated with positive PLNRs declines with increasing age to the point that there is no excess risk associated with positive rests for tumors diagnosed at age 16 years. There continued to be some suggestion of a reduction in risk associated with increasing amounts of chemotherapy, although this reduction was not statistically significant. Effects of study number on risk were accounted for through the formation of the matched sets. The slightly increased risk associated with female gender (Table 8) was not significant for the matched case-control study. DISCUSSION This study identified several demographic features associated with an increased risk for developing metachronous BWT that can be ascertained prior to surgery. For example, children diagnosed with Wilms tumor prior to their first birthday have an increased risk for developing contralateral disease. However, further analysis reveals that this is only true if these patients also have NRs, in particular, PLNRs (Table 5B). NWTSG pathologists previously reported that PLNRs were more commonly associated with synchronous BWT, whereas ILNRs were more preponderant in patients with metachronous BWT.4 This study confirms the association of metachronous BWT and NRs in general (Table 2), but the matched case-control study (Table 5A) suggests a particularly high risk for developing metachronous BWT in children ,12 months of age who are found to have PLNRs in the renal tissue removed for unilateral Wilms tumor. Because the NWTSG has demonstrated that the median age at diagnosis for children with Wilms tumor and PLNRs is substantially higher (36 months) than that of those with Wilms tumor and ILNRs (16 months),16 the occurrence of PLNRs in children ,12 months is rare. Nevertheless, it is precisely this combination that seems to carry the highest risk of contralateral disease development (Table 8). A second characteristic identified with increased risk of metachronous BWT is gender. The 1:2 male: female ratio found in this study was similar to that reported by investigators of the International Society of Paediatric Oncology.17 In the current study, however, we demonstrate that the risk of contralateral Wilms tumor development is higher in females even taking other risk factors into account. Our current knowledge of the molecular genetics of Wilms tumor does not provide an explanation for this observation, despite the recent implication of chromosome X in Wilms tumorigenesis. Some investigators have suggested that a locus on chromosome X might control parental allele specific imprinting (a genetic phenomenon affecting gene expression and therefore function) of genes on chromosome 11p15, a region that includes WT2, a putative second Wilms tumor suppressor gene. If WT2 function is indeed regulated by a gene or gene complex on the X chromosome, then the same locus might help explain the high incidence of BWT in females. To date, however, no “imprintor” gene has been identified on the X chromosome. The incidence of metachronous BWT decreased by half after the first NWTSG study (Fig. 1). A previous NWTSG report indicated that only 7 of 18 patients with metachronous BWT registered on NWTS-1 and NWTS-2 had adequate contralateral kidney exploration at the time of initial surgery.18 Therefore, we considered the possibility that a higher incidence of failure to recognize synchronous BWT at the time of first surgery during the first NWTSG study might explain the higher incidence of metachronous BWT in NWTS-1. The case-control study, however, provided no evidence for this. The extent of surgical exploration of the contralateral kidney at the time of initial surgery Risk Factors for Metachronous Bilateral Wilms Tumor/Coppes et al. TABLE 5 Number of Cases, Controls, and Estimated Relative Risk with 95% Confidence Interval According to the Presence of Perilobar or Intralobar Nephrogenic Rests and Age at Diagnosis Number Group A. PLNR/ILNRa Negative/negative Negative/single Negative/multiple Positive/negative Positive/single Positive/multiple Unknown/unknown B. Age (mos)/PLNR 0–11b Negative Positive $12c Negative Positive Unknown C. Age (mos)/ILNR 0–11d Negative Single Multiple $12e Negative Single Multiple Unknown Cases Controls TABLE 6 Degree of Surgical Exploration of the Contralateral Kidney for Cases of Metachronous Bilateral Wilms Tumor and Matched Controls by Study Matched set analysis RR Degree of exploration 95% CI Study 10 10 4 16 6 10 2 88 51 15 39 7 12 19 1.0 1.87 2.66 3.61 8.23 7.56 1.11 6 15 23 3 1.0 ` 18 17 2 132 55 18 1.0 2.18 1.02 1.01–4.68 0.20–5.10 4 6 11 9 13 4 1.0 0.31 2.65 0.03–3.16 0.25–27.5 22 10 3 2 118 46 23 18 1.0 1.22 0.95 0.65 0.72–4.91 0.71–10.0 1.50–8.70 2.05–33.1 2.59–22.0 0.22–5.53 1.66–` 1623 NWTS-1 Case Control NWTS-2 Case Control NWTS-3 Case Control NWTS-4 Case Control Total Case Control None Exploration without visualization of posterior surface Visualization of posterior surface Total 3 15 6 21 2 6 11 42 3 15 4 8 1 8 8 31 0 7 5 28 16 46 21 81 2 5 4 7 10 46 16 58 8 42 19 64 29 106 56 212 NWST: National Wilms Tumor Study. 0.52–2.90 0.25–3.60 0.14–3.08 RR: relative risk; CI: confidence interval; PLNR: perilobar nephrogenic rest; ILNR: intralobar nephrogenic rest. a P 5 0.001. b P 5 0.01. c P 5 0.11. d P 5 0.22. e P 5 0.89. did not affect the frequency of contralateral involvement (Tables 6 and 7). Moreover, review of the NWTSG data indicates that the incidence of synchronous BWT has not changed significantly over the years (5.8 %, 4.6%, 6.9%, and 6.6% for NWTS-1, NWTS-2, NWTS-3, and NWTS-4, respectively). Thus, there is nothing to suggest that the decrease in incidence of metachronous BWT can be attributed to more accurate diagnosis of contralateral tumor due to better visualization at the time of initial surgery. Following NWTS-1, the incidence of metachronous BWT has remained stable despite a dramatic increase in patient accrual and improved outcome for patients with unilateral Wilms tumor. Perhaps a fixed proportion of patients with Wilms tumor is biologically predisposed to develop contralateral disease despite receiving chemotherapy adequate to cure the TABLE 7 Relative Risk with 95% Confidence Interval of Contralateral Involvement According to the Degree of Surgical Exploration of the Opposite Kidney Estimated from the Matched Set Analysis Degree of exploration RRa 95% CI None Exploration without visualization Visualization of posterior surface 1.0 1.60 1.44 — 0.62–4.11 0.53–3.77 RR: relative risk; CI: confidence interval. a P 5 0.62. initial tumor. In other patients, however, chemotherapy may not only contribute to the patient’s ultimate cure but, in addition, obliterate cells that otherwise would have developed into macroscopic Wilms tumor. Univariate analysis suggests that triple chemotherapy with AMD, VCR, and DOX is more effective in inhibiting the transformation of a NR into a Wilms tumor (or, alternatively, eradicating an indolent Wilms tumor) than double agent therapy with AMD and VCR. Single chemotherapy regimens utilizing either VCR or AMD are even less effective. This could explain the higher incidence of metachronous BWT observed in NWTS-1. On the first NWTSG study, all patients with Group I disease and 60% of those with Group II and Group III disease were treated with a single chemotherapeutic agent (VCR or AMD) with or without radiation therapy, which did not include the contralat- 1624 CANCER April 1, 1999 / Volume 85 / Number 7 TABLE 8 Conditional Logistic Regression Analysis of Contralateral Kidney Involvement Variable Coefficient 6 SE Relative risk P value PLNR (age 5 0) PLNR 3 log (age 1 1) Female gender Chemotherapya 2.17 6 0.47 20.75 6 0.24 0.35 6 0.38 20.58 6 0.44 8.80 0.48 1.42 0.56 0.0 0.02 0.36 0.18 PLNR: perilobar nephrogenic rest; SE: standard error. a Code 0 for dactinomycin (AMD) or vincristine (VCR); code 1 for AMD 1 VCR; code 2 for AMD 1 VCR 1 doxorubicin 6 cyclophosphamide. eral kidney.8 In the three ensuing NWTSG studies, all patients received at least two chemotherapeutic agents, i.e., both AMD and VCR, as adjuvant therapy. Certain congenital syndromes known to predispose to the development of Wilms tumor (aniridia, Beckwith–Wiedemann syndrome, and hemihypertrophy) appeared to be at higher risk, but the number of patients identified with such anomalies were judged too small for reliable statistical inference. Children with Denys–Drash syndrome were not found to have an increased risk at developing metachronous BWT (data not shown). This is likely due to the fact that most if not all of these patients undergo bilateral nephrectomy once the disease is recognized. The follow-up recommendations that follow are based on the data presented. We recommend that children with NRs in the nephrectomy specimen undergo regular abdominal ultrasonography following surgery, with particular attention to the remaining kidney. For children ,48 months of age at initial diagnosis, abdominal ultrasonography should be performed at least every 3 months for 6 years. For children .48 months of age at initial diagnosis, we suggest abdominal ultrasonography at least every 3 months for 4 years.2 Because this study has not addressed specifically the risk of either contralateral disease or local relapse in children who do not have NRs in their nephrectomy specimens, we suggest that they be monitored as previously recommended.19 Briefly, for those with Stage I and II favorable histology Wilms tumors, abdominal ultrasonography yearly 3 3 is recommended. For those with Stage III favorable histology disease, abdominal ultrasonography 6 weeks and 3 months postoperative, then every 3 months 3 5, every 6 months 3 3, and finally yearly 3 2 is recommended. The latter group of patients requires additional surveillance to monitor for local relapse rather than contralateral disease. The use of abdominal ultrasonography for follow-up has two advantages: it is readily available and is relatively inexpensive.19 Finally, with regard to parenchymal sparing surgery, the low incidence of metachronous BWT in children .12 months of age with NRs does not support such an approach, and these children should be treated with nephrectomy up front. Nephron conserving surgery might be considered for a very selected group of infants with NRs demonstrated pre- or intraoperatively, given the higher frequency of metachronous disease in such children. However, pilot studies have shown that there are very few tumors located at one pole of the kidney that are circumscribed so clearly that sharp and complete separation from the remaining kidney is possible.20 Incomplete removal after injudicious attempts is a danger. In short, we have no data to support such a policy, nor do we have a specific treatment or series of treatments that might be appropriate if such an attempt were to be made. We feel strongly that such an approach should be studied formally, and the NWTSG recommendations for managing children with synchronous BWT might be consulted for guidance.21 REFERENCES 1. Ritchey ML, Green DM, Thomas PRM, Smith GR, Haase G, Shochat S, et al. Renal failure in Wilms tumor patients: a report from the National Wilms Tumor Study Group. Med Pediatr Oncol 1996;26:75– 80. 2. Paulino AC, Thakkar B, Henderson WG. Metachronous bilateral Wilms’ tumor. The importance of time interval to the development of a second tumor. Cancer 1998;82:415– 20. 3. Green DM, Thomas PRM, Shochat S. The treatment of Wilms tumor. Results of the National Wilms Tumor Studies. Hematol Oncol Clin North Am 1995;9:1267–74. 4. Beckwith JB, Kiviat NB, Bonadio JF. Nephrogenic rests, nephroblastomatosis, and the pathogenesis of Wilms’ tumor. Pediatr Pathol 1990;10:1–36. 5. White KS, Kirks DR, Bove KE. Imaging of nephroblastomatosis: an overview. Radiology 1992;182:1–5. 6. Babyn P, Owens C, Gyepes M, D’Angio GJ. Imaging patients with Wilms tumor. Hematol Oncol Clin North Am 1995;9: 1217–52. 7. Gylys-Morin V, Hoffer FA, Kozakewich H, Shamberger RC. Wilms tumor and nephroblastomatosis: imaging characteristics at gadolinium-enhanced MR imaging. Radiology 1993; 188:517–21. 8. D’Angio GJ, Evans AE, Breslow N, Beckwith B, Bishop H, Feigl P, et al. The treatment of Wilms’ tumor: results of the National Wilms’ tumor study. Cancer 1976;38:633– 46. 9. D’Angio GJ, Evans A, Breslow N, Beckwith B, Bishop H, Farewell V, et al. The treatment of Wilms’ tumor: results of the Second National Wilms’ Tumor Study. Cancer 1981;47: 2302–11. 10. D’Angio GJ, Breslow N, Beckwith JB, Evans A, Baum H, deLorimier A, et al. Treatment of Wilms’ tumor. results of the Third National Wilms’ Tumor Study. Cancer 1989;64: 349 – 60. Risk Factors for Metachronous Bilateral Wilms Tumor/Coppes et al. 11. Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 81. 12. Peto R, Peto J. Asymptotically efficient rank invariant test procedures. J Royal Stat Soc (A) 1972;135:185–92. 13. Cox DR. Regression models and life tables. J Royal Stat Soc (B) 1972;34:187–202. 14. Langholz B, Goldstein L. Risk set sampling in epidemiologic cohort studies. Stat Sci 1996;11:35–53. 15. Breslow NE, Day NE. Statistical methods in cancer research I: the analysis of case-control studies. Lyon: International Agency for Cancer Research, 1980. 16. Beckwith JB. Precursor lesions of Wilms tumor: clinical and biological implications. Med Pediatr Oncol 1993;21: 158 – 68. 17. Coppes MJ, de Kraker J, van Dijken PJ, Perry HJ, Delemarre 18. 19. 20. 21. 1625 JF, Tournade MF, et al. Bilateral Wilms’ tumor: long-term survival and some epidemiological features. J Clin Oncol 1989;7:310 –5. Jones B, Hrabovsky E, Kiviat N, Breslow N. Metachronous bilateral Wilms’ tumor: a National Wilms’ Tumor Study. Am J Clin Oncol 1982;5:545–50. D’Angio GJ, Rosenberg H, Sharples K, Kelalis P, Breslow N, Green DM. Position paper: imaging methods for primary renal tumors of childhood: costs versus benefits. Med Pediatr Oncol 1993;21:205–12. Pringle KC, Nicholls WD, Lewis ME, Hill RW. The role of delayed surgery and partial nephrectomy in Wilms tumor. Med Pediatr Oncol 1998;31:309. Ritchey ML, Coppes MJ. The management of synchronous bilateral Wilms tumor. Hematol Oncol Clin North Am 1995; 9:1303–15.