Factors associated with relapse and survival in Wilms
tumor: a multivariate analysis
Joshua N. Honeymana, Barrie S. Richa, Maureen P. McEvoya, Molly A. Knowlesa,
Glenn Hellerb, Edward Riachya, Rachel Kobosc, Neerav Shuklac, Suzanne L. Woldend,
Peter G. Steinherzc, Michael P. La Quagliaa,⁎
aDepartment of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA
bDepartment of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY, USA
cDepartment of Pediatrics, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA
dDepartment of Radiation Oncology, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA
Received 2 March 2012; accepted 6 March 2012
Lymph node metastasis;
Purpose: Lymph node metastasis and anaplasia predict relapse-free survival in Wilms tumor. We
performed a multivariate analysis of our institutional database to identify factors independently
associated with relapse-free and overall survival.
Methods: We retrospectively reviewed cases of confirmed Wilms tumor diagnosed between 1990 and
2010 and treated at our institution. The log-rank test was used to screen variables for consideration in the
proportional hazards model.
Results: A total of 95 patients were treated at our institution during the study period, with a median
follow-up of 3.3 years. Factors correlated with overall survival in the univariate analysis were local
disease, metastasis, tumor size, anaplasia, renal vein tumor thrombus, inferior vena cava tumor
thrombus, lymph node positivity, and tumor rupture. On multivariate analysis, factors associated with
increased risk of death were lymph node positivity and anaplasia. Factors correlated with probability of
relapse in the univariate analysis were lymph node positivity, anaplasia, and female sex. All 3 of these
factors were also independently significant on multivariate analysis.
Conclusion: Lymph node involvement and anaplasia are significantly correlated with probability of
relapse and overall survival, reemphasizing the strong recommendation to sample regional lymph nodes
during Wilms tumor resection.
© 2012 Elsevier Inc. All rights reserved.
Wilms tumor is the second most common intraabdominal
solid tumor of childhood, accounting for approximately 14%
of pediatric cancers . Over the past 50 years, outcomes
well as cooperative group trials run under the auspices of the
National Wilms' Tumor Study Group (NWTSG). Currently,
the 5-year overall survival for patients with Wilms tumor is
⁎Corresponding author. Pediatric Surgical Service, Department of
Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY 10065.
Tel.: +1 212 639 7002; fax: +1 212 717 3373.
E-mail address: email@example.com (M.P. La Quaglia).
0022-3468/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
Journal of Pediatric Surgery (2012) 47, 1228–1233
combined International Society of Pediatric Oncology/
German, Swiss, and Austrian Society of Paediatric Oncology
and Haematology trials was 12%, with an overall survival in
relapsed patients of 48% . In patients without metastatic
disease at presentation, approximately 75% of all recurrences
occur within 1 year , and in patients surviving greater than
5 years after diagnosis, approximately 5% will experience a
Despite overall improved outcomes in Wilms tumor,
certain populations of patients continue to experience poor
survival and increased rates of relapse. Reports from the
National Wilms' Tumor Studies (NWTS) identified anapla-
sia and lymph node metastases as the most significant factors
predicting relapse and survival [6-8]. Surgical factors have
been associated with increased risk of local recurrence .
In this study, we performed a multivariate analysis of our
Wilms tumor database to identify factors associated with
time to relapse and overall disease-specific survival.
After obtaining a waiver of authorization from the
institutional review board, we performed a retrospective
2010. A total of 95 consecutive patients with histologically
confirmed Wilms tumor were identified. Fifty-eight patients
were newly diagnosed and had their first operation at our
institution. The remainder presented after receiving initial
care at an outside institution.
for data on demographics, medical history, diagnosis,
treatment, complications, and outcomes. Pathology reports
were reviewed for single-dimension tumor size (grouped as
b4, 4-8, 8-12, and N12 cm), histologic subtype (triphasic,
blastemal-predominant, stromal-predominant, epithelial-pre-
dominant, mixed), anaplasia, adjacent nephroblastomatosis,
capsular invasion, renal vein tumor thrombus, inferior vena
metastasis. All 95 patients were included in the analysis.
1.2. Data analysis
The probability of disease-specific survival was generated
using the Kaplan-Meier method; Wilms tumor was the only
cause of death in this cohort. The probability of relapse was
estimated using the cumulative incidence function. Factors
included in the univariate analysis included sex, race,
abdominal pain, abdominal mass, hematuria, fever, local
disease, laterality, metastasis at diagnosis, neoadjuvant
chemotherapy, tumor size, anaplasia, histologic subtype,
renal vein thrombosis, IVC thrombosis, positive margins,
lymph node positivity, biopsy, and rupture. The log-rank test
was used as a screen to determine the factors that would be
considered jointly in the proportional hazards model. A
P value of .05 was considered statistically significant.
The concordance probability estimate  was computed
to assess the predictive accuracy of this model. The
concordance probability measures the degree to which the
risk assessment ordering for any pair of subjects is in
agreement with the survival time ordering. The concordance
probability estimate ranges between 0.5 and 1.0, with 1.0
representing perfect concordance between the risk assess-
ment and survival time.
2.1. Patients and demographics
A total of 95 patients were treated for Wilms tumor at our
institution during the study period, with a median follow-up
Demographics and medical history
No. (range or
Median age3.3 y
(12 d–20.5 y)
History of mesoblastic
Denys-Drash syndrome 1
Presenting symptoms Palpable mass
Stage at diagnosis
1229 Relapse and survival in Wilms tumor
of 3.3 years (mean ± SD, 5.1 ± 4.6 years). The median age at
diagnosis was 3.3 years (range, 12 days–20.5 years). The
male-to-female ratio was 1:1.4. The stage distribution at
diagnosis was as follows: 9 patients had stage I disease, 20
had stage II disease, 30 had stage III disease, 22 had stage IV
disease, and 14 had stage V disease. Three patients presented
with extrarenal Wilms tumor (Table 1).
All patients received chemotherapy; 76% of patients
received radiotherapy, including 65% of newly diagnosed
patients. Six patients developed chronic renal insufficiency,
2 of whom required dialysis after bilateral nephrectomies. In
addition, 4 patients had respiratory dysfunction in the setting
of previous whole-lung irradiation, 2 patients developed
scoliosis, and 1 patient was diagnosed as having a secondary
renal cell carcinoma (Table 2).
2.2. Survival analysis
Fifty-eight patients presented with newly diagnosed
Wilms tumor, and 37 were transferred to our institution
after receiving initial surgical care elsewhere. The 5-year
overall survival was 89% for newly diagnosed patients (Fig.
1A) and 59% for transferred patients. Patients who were
transferred had higher rates of anaplasia (27% vs 12%),
lymph node positivity (22% vs 17%), and advanced stage
disease (81% had stage ≥III disease vs 62%).
Stratifying patients based on the decade of diagnosis
found that patients diagnosed in the 2000s had better overall
survival than patients diagnosed in the 1990s (84% vs 66%,
P = .045). From 1990 to 2000, although rates of lymph
node involvement remained stable, rates of anaplasia in our
study population decreased from 26% to 14%. Patients
without anaplasia did not have different survival rates
between the 2 decades.
In the entire study population, there were 17 disease-
related deaths, with a 5-year overall survival of 77% (Fig.
1B). Factors demonstrating significantly inferior overall
survival by the log-rank test were non–stage I/II disease (P =
.01), metastasis at diagnosis (P b .01), tumor size (P = .03),
anaplasia (P b .01), renal vein tumor thrombus (P b .01),
IVC tumor thrombus (P b .01), lymph node positivity (P b
.01), and rupture (P b .01) (Table 3).
There were 28 relapses, with a 5-year probability of
relapse equal to 0.39 (Fig. 2A). Factors significantly
correlated with relapse included lymph node positivity (P b
.01), anaplasia (P b .01), and female sex (P = .03) (Table 3).
A review of patients with neither anaplasia nor positive
lymph nodes identified 3 patients who died during the study
period. Two patients had stage IV disease at diagnosis. The
third presented with stage II disease but rapidly progressed in
the lungs after 3 months. A review of this patient's staging
scans demonstrated a small pulmonary nodule that was likely
an early metastatic lesion.
2.3. Multivariate analysis
On multivariate analysis (Table 4), 2 factors were
independently correlated with overall survival: lymph node
metastasis and anaplasia (relative risk [RR], 11.2 and 7.5,
respectively) (Fig. 1C, D).
Three factors were independently correlated with time
to relapse: lymph node metastasis (RR, 7.4), anaplasia
(RR, 4.2), and female sex (RR, 2.6) (Fig. 2B-D).
The concordance probability estimates for the assessment of
overall survival and time to relapse were 0.70 and 0.72,
2.4. Lymph nodes
Given that the presence of histopathologically positive
lymph nodes at diagnosis was a strong predictor of poor
outcome in both univariate and multivariate analyses, we
analyzed this subgroup to determine whether the location of
involved lymph nodes was of further prognostic value. Two
patients were excluded from this analysis because the medical
record contained no further data on the positive lymph nodes.
Pathology and outcomes
No. (range or percentage)
Median tumor size (cm)
Renal vein thrombus
Lymph node (LN) positivity
Positive LN at diagnosis
Negative LN at diagnosis
Unknown LN status
8.75 (1.3-30) cm
1 (renal cell carcinoma)
1230 J.N. Honeyman et al.
There were 4 patients with positive renal hilar lymph
nodes, 10 patients with positive ipsilateral lymph nodes, 5
patients with positive interaortocaval nodes, and 3 patients
with metastases in remote lymph nodes (iliac, omental,
and retrocrural). There was no clear correlation between
lymph node location and either relapse-free survival or
We describe here a single institution analysis of all
patients with Wilms tumor during a 20-year period,
including newly diagnosed and relapsed cases. We found a
5-year overall survival of 89% for newly diagnosed patients,
which is less than that described nationally . However, we
found a higher-than-expected number of patients with stage
III and IV disease compared with the NWTSG data (55% vs
45%), as well as an increased incidence of anaplasia (18% vs
10.8%), which could account for the decrease in overall
On multivariate analysis, we identified 3 factors inde-
pendently correlated with the time to relapse in Wilms tumor
(anaplasia, lymph node metastases, and sex) and 2 factors
independently correlated with disease-specific survival
OVERALL SURVIVAL: NEW DIAGNOSIS
OVERALL SURVIVAL: STUDY POPULATION
OVERALL SURVIVAL: ANAPLASIA
OVERALL SURVIVAL: LYMPH NODE STATUS
05 10 1520
05 10 15 20
05 1015 2005 10 15 20
Negative lymph nodes
Positive lymph nodes
Overall survival for the study population was 77% at 5 years. Multivariate analysis identified 2 factors independently correlated with overall
survival: anaplasia (C) and lymph node status (D).
Overall 5-year survival. A, Five-year overall survival in patients presenting with a new diagnosis of Wilms tumor was 89%. B,
on the log-rank test
Results of the association of individual factors based
Stage I/II disease
Metastasis at diagnosis
Tumor size group
Renal vein thrombosis
Lymph node positivity
1231Relapse and survival in Wilms tumor
(anaplasia and lymph node metastases). All patients in our
study population who died either had positive lymph nodes
at diagnosis, anaplasia, or presented with distant metastases.
The NWTS has found no statistically significant
association between sex and outcome [6,7]. The results of
the NWTS-1 found no trend toward increased relapse in
females compared to males. However, a recent study did
find that females had increased mortality after 5-year
Unfavorable histology has a well-described correlation
with adverse outcomes in Wilms tumor [6,7,13]. Pathology
reviews, including results from the NWTSG, have identified
anaplasia as one of the strongest predictors of poor outcome.
This finding has been integrated with the staging of Wilms
tumor, and patients with anaplasia do better with more
aggressive therapy [11,14].
Lymph node metastasis has been repeatedly found to
correlate with relapse and mortality in patients with Wilms
tumor [15,16]. Lymph node sampling is an important part of
sampling has been associated with increased risk of local
recurrence , likely caused by the down-staging and
subsequent undertreatment of patients with undiagnosed
positive lymph nodes. A review of the National Cancer Data
Base found that 42% of patients had no lymph nodes
evaluated, and patients who had lymph nodes assessed had a
significantly better 5-year survival . Data from the
Surveillance, Epidemiology, and End Results database
showed that 22% of patients did not undergo lymph node
sampling; in this population, patients who did not have their
lymph nodes sampled had an increased risk of local
PROBABILITY OF RELAPSE: ALL PATIENTS
PROBABILITY OF RELAPSE: LYMPH NODE STATUS
PROBABILITY OF RELAPSE: ANAPLASIA
PROBABILITY OF RELAPSE: SEX
05 10 15 20
05 1015 20
independently correlate with relapse: anaplasia (B), positive lymph nodes (C), and female sex (D).
Probability of relapse. A, The probability of relapse by 5 years was 0.39. On multivariate analysis, 3 factors were identified that
Results of multivariate analysis
RR of death
Lymph node positivity
Abbreviation: CI, confidence interval.
1232 J.N. Honeyman et al.
recurrence. This was an independent indicator of decreased Download full-text
survival . A report from the NWTS-5 found that lymph
node sampling was not performed in 9% of patients. This
same report found that positive lymph nodes were the critical
staging criteria in 41% of patients with stage III disease .
Routine radical lymph node dissection has not been found
to decrease recurrence or to prolong survival . Intrao-
perative assessment of lymph nodes without histopathologic
confirmation is discouraged because lymph node assessment
by surgeons has a false-negative rate of around 31%, with a
predictive value of only 56% . For this reason, the
current recommendation is for routine intraoperative lymph
node sampling with pathologic review.
A review of the surgical outcomes of the NWTS-1 found
that relapse and mortality increased in patients with both
hilar and aortic nodes positive compared with those with
hilar nodes alone . The NWTS-4 also found that patients
with positive aortic nodes had a higher risk of local
recurrence than did patients with positive hilar nodes .
The results of the present study confirm the poor outcome
previously reported for patients with anaplastic histology
Wilms tumor. Furthermore, they reemphasize the strong
recommendation to sample regional lymph nodes in patients
with Wilms tumor.
 Johnson KA, Aplenc R, Bagatell R. Survival by race among children
with extracranial solid tumors in the United States between 1985 and
2005. Pediatr Blood Cancer 2011;56:425-31.
 Smith MA, Seibel NL, Altekruse SF, et al. Outcomes for children and
adolescents with cancer: challenges for the twenty-first century. J Clin
 Reinhard H, Schmidt A, Furtwängler R, et al. Outcome of relapses of
nephroblastoma in patients registered in the SIOP/GPOH trials and
studies. Oncol Rep 2008;20:463-7.
 Sutow WW, Breslow NE, Palmer NF, et al. Prognosis in children with
Wilms' tumor metastases prior to or following primary treatment:
results from the first National Wilms' Tumor Study (NWTS-1). Am J
Clin Oncol 1982;5:339-47.
 Termuhlen AM, Tersak JM, Liu Q, et al. Twenty-five year follow-up
of childhood Wilms tumor: a report from the Childhood Cancer
Survivor Study. Pediatr Blood Cancer 2011;57:1210-6.
 Breslow NE, Palmer NF, Hill LR, et al. Wilms' tumor: prognostic
factors for patients without metastases at diagnosis: results of the
National Wilms' Tumor Study. Cancer 1978;41:1577-89.
 Breslow N, Churchill G, Beckwith JB, et al. Prognosis for Wilms'
tumor patients with nonmetastatic disease at diagnosis—results of the
second National Wilms' Tumor Study. J Clin Oncol 1985;3:521-31.
 Breslow N, Sharples K, Beckwith JB, et al. Prognostic factors in
nonmetastatic, favorable histology Wilms' tumor. Results of the Third
National Wilms' Tumor Study. Cancer 1991;68:2345-53.
 Shamberger RC, Guthrie KA, Ritchey ML, et al. Surgery-related
factors and local recurrence of Wilms tumor in National Wilms Tumor
Study 4. Ann Surg 1999;229:292-7.
 Gonen M, Heller G. Concordance probability and discriminatory
power in proportional hazards regression. Biometrika 2005;92:965-70.
 Dome JS, Cotton CA, Perlman EJ, et al. Treatment of anaplastic
histology Wilms' tumor: results from the fifth National Wilms' Tumor
Study. J Clin Oncol 2006;24:2352-8.
 Cotton CA, Peterson S, Norkool PA, et al. Early andlate mortality after
diagnosis of Wilms tumor. J Clin Oncol 2009;27:1304-9.
 Beckwith JB, Palmer NF. Histopathology and prognosis of Wilms
tumors: results from the First National Wilms' Tumor Study. Cancer
 Bonadio JF, Storer B, Norkool P, et al. Anaplastic Wilms' tumor:
clinical and pathologic studies. J Clin Oncol 1985;3:513-20.
 Jereb B, Tournade MF, Lemerle J, et al. Lymph node invasion and
prognosis in nephroblastoma. Cancer 1980;45:1632-6.
 Spreafico F, Gandola L, D'Angelo P, et al. Heterogeneity of disease
classified as stage III in Wilms tumor: a report from the Associazione
Italiana Ematologia Oncologia Pediatrica (AIEOP). Int J Radiat Oncol
Biol Phys (in press).
 Raval MV, Bilimoria KY, Bentrem DJ, et al. Nodal evaluation in
Wilms' tumors: analysis of the national cancer database. Ann Surg
 Zhuge Y, Cheung MC, Yang R, et al. Improved survival with lymph
node sampling in Wilms tumor. J Surg Res 2011;167:e199-203.
 Ehrlich PF, Ritchey ML, Hamilton TE, et al. Quality assessment for
Wilms' tumor: a report from the National Wilms' Tumor Study-5.
J Pediatr Surg 2005;40:208-12.
 Othersen HB, DeLorimer A, Hrabovsky E, et al. Surgical evaluation of
lymph node metastases in Wilms' tumor. J Pediatr Surg 1990;25:
 Leape LL, Breslow NE, Bishop HC. The surgical treatment of Wilms'
tumor: results of the National Wilms' Tumor Study. Ann Surg
1233 Relapse and survival in Wilms tumor