Detection of Preoperative Wilms Tumor Rupture with CT: A Report from the Children’s Oncology Group

The presence of ascites beyond the cul-de-sac, irrespective of attenuation, is the most useful CT indicator of preoperative Wilms tumor rupture; in the presence of ascites, fat stranding around the tumor and the presence of retroperitoneal fluid are also highly predictive of rupture.

Abstract

Purpose:

To retrospectively determine the diagnostic performance of computed tomography (CT) in identifying the presence or absence of preoperative Wilms tumor rupture.

Materials and Methods:

The cohort was derived from the AREN03B2 study of the Children’s Oncology Group. The study was approved by the institutional review board and was compliant with HIPAA. Written informed consent was obtained before enrollment. The diagnosis of Wilms tumor rupture was established by central review of notes from surgery and/or pathologic examination. Seventy Wilms tumor cases with rupture were matched to 70 Wilms tumor controls without rupture according to age and tumor weight (within 6 months and 50 g, respectively). CT scans were independently reviewed by two radiologists, and the following CT findings were assessed: poorly circumscribed mass, perinephric fat stranding, peritumoral fat planes obscured, retroperitoneal fluid (subcapsular vs extracapsular), ascites beyond the cul-de-sac, peritoneal implants, ipsilateral pleural effusion, and intratumoral hemorrhage. All fluids were classified as hemorrhagic or nonhemorrhagic by using a cutoff of 30 HU. The relationship between CT findings and rupture was assessed with logistic regression models.

Results:

The sensitivity and specificity for detecting Wilms tumor rupture were 54% (36 of 67 cases) and 88% (61 of 69 cases), respectively, for reviewer 1 and 70% (47 of 67 cases) and 88% (61 of 69 cases), respectively, for reviewer 2. Interobserver agreement was substantial (ĸ = 0.76). All imaging signs tested, except peritoneal implants, intratumoral hemorrhage, and subcapsular fluid, showed a significant association with rupture (P ≤ .02). The attenuation of ascitic fluid did not have a significant correlation with rupture (P = .9990). Ascites beyond the cul-de-sac was the single best indicator of rupture for both reviewers, followed by perinephric fat stranding and retroperitoneal fluid for reviewers 1 and 2, respectively (P < .01).

Conclusion:

CT has moderate specificity but relatively low sensitivity in the detection of preoperative Wilms tumor rupture. Ascites beyond the cul-de-sac, irrespective of attenuation, is most predictive of rupture.

© RSNA, 2012

Supplemental material: http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.12120670/-/DC1

Introduction

The overall cure rate of Wilms tumor has risen to more than 90% (1). Contemporary issues in the management of Wilms tumor involve risk stratification of children to avoid overtreatment and resultant morbidity in low-risk patients while providing intensive treatment to high-risk patients. This requires determination of criteria that imply a poor prognosis (2). Tumor spillage of Wilms tumor increases the risk of abdominal recurrence to 20% (3). Tumor spillage is considered to be present if there has been preoperative tumor rupture, intraoperative tumor spill, or a tumor biopsy. Although intraoperative tumor spill and tumor biopsy are intervention-related events, preoperative tumor rupture can be spontaneous or occur after trauma (3,4).

Because of the increased risk of abdominal recurrence after tumor rupture, all cases with preoperative tumor rupture are designated as stage III disease and treated with three-drug therapy and whole abdominal radiation. This intensive treatment has both acute side effects and long-term consequences, such as secondary malignancy and reduced fertility (5,6). Hence, the diagnosis of preoperative tumor rupture must be based on robust criteria. Clinical signs and symptoms that can suggest a diagnosis of preoperative tumor rupture include abdominal pain, anemia, and shock (7). However, these findings have limited accuracy in the diagnosis of preoperative tumor rupture (8,9). Imaging findings that have been suggested as indicative of preoperative tumor rupture include hemoperitoneum and enhancing solid masses in the peritoneum, mesentery, and/or omentum (10,11). The accuracy of these findings, however, has not been evaluated in a large series of patients. In addition, the clinical significance of intratumoral, subcapsular, or perirenal hemorrhage remains unclear (10).

Central archiving of the baseline imaging studies of all children enrolled in the ongoing Renal Tumor Classification, Biology, and Banking Study (AREN03B2) of the Children’s Oncology Group and central review of surgical notes and pathology specimens gave us the unique opportunity to evaluate the imaging findings associated with preoperative Wilms tumor rupture. The objective of our study was to retrospectively determine the diagnostic performance of computed tomography (CT) in identifying the presence or absence of preoperative Wilms tumor rupture.

Materials and Methods

The cohort for our retrospective study was derived from the ongoing AREN03B2 study of the Children’s Oncology Group. The study was approved by the institutional review boards of all participating institutions and was compliant with the Health Insurance Portability and Accountability Act. Written informed consent was obtained from all parents or guardians before enrollment in AREN03B2. Patients were enrolled from 204 institutions between February 2006 and December 2010. Per the study protocol, all children with a newly diagnosed renal mass are eligible for the study. At the time of enrollment, the institution is required to submit a chest CT scan and a contrast material–enhanced CT scan or magnetic resonance (MR) image of the abdomen and pelvis for central imaging review. Local institutions are also required to submit surgical notes and pathology slides from biopsy and/or nephrectomy specimens for central review by the study surgeons and pathologists, respectively.

In AREN03B2, the presence or absence of Wilms tumor rupture was determined by central review of surgical and pathologic findings. Preoperative tumor rupture is considered to be present if the surgical note documents the presence of capsular breach, peritoneal implants, or hemorrhagic fluid in the peritoneal cavity at laparotomy, before intervention on the tumor. Pathologic findings supporting the presence of rupture include identification of gross evidence of rupture by the institutional pathologist or histologic evidence of perirenal soft-tissue reaction and/or hemorrhage and tumor fragmentation on central review, in confirmation with central surgical review. As of December 2010, 1853 patients with Wilms tumor were enrolled in AREN03B2. Preoperative tumor rupture was present at central surgical and pathology review in 88 of the 1853 subjects. To perform a matched-pair analysis, we attempted to match each case (Wilms tumor with rupture) with a control (Wilms tumor without rupture at both surgical and pathology review) according to age (within 6 months) and tumor weight (within 50 g). This resulted in a total of 70 matched pairs. Contrast-enhanced CT scans of the abdomen and pelvis were not available for central radiology review in four of these subjects, resulting in a sample size of 136, with 67 cases and 69 controls.

CT Technique and Image Evaluation

In the AREN03B2 study protocol, all institutions are advised to perform contrast-enhanced CT of the abdomen and pelvis during the portal venous phase of contrast material injection. All submitted CT scans had been performed with intravenous contrast material administration, and 104 of the 136 subjects had received intraluminal bowel contrast material as well. Portal venous phase images were available for review in all 136 cases, and additional multiphase imaging had also been performed in 37 subjects (precontrast imaging, n = 12; arterial phase imaging, n = 8; and delayed phase imaging, n = 22). The distribution of the reconstructed section thickness of the CT scans submitted for central review was as follows: 5 mm (67 subjects), 3–4.9 mm (46 subjects), less than 3 mm (22 subjects), and more than 5 mm (one subject). Axial images and multiplanar reconstructions (sagittal and coronal) were available for review in 107 subjects, whereas only axial images were available in 29 subjects.

The radiology review was independently performed by two pediatric radiologists with certificate of added qualification in pediatric radiology (G.K. and F.H., with 13 and 31 years of experience, respectively). The radiology reviewers had institutional review board approval to perform central imaging review. The reviewers were blinded to all patient information, including clinical history, report of the institutional radiologist, and surgical and/or pathologic findings. The scans were reviewed in chronological order on the basis of the subject number. If a subject had undergone multiphase CT, images from all phases of the CT examination were reviewed. Each CT scan was rated on a five-point scale for the presence or absence of rupture, as follows: 1, definitely no rupture; 2, probably no rupture; 3, uncertain; 4, probable rupture; and 5, definite rupture. The presence or absence of each of the following CT findings was assessed in each subject to determine the utility of these imaging findings in identifying preoperative tumor rupture: poorly circumscribed mass (the edge of the mass cannot be clearly delineated), fat stranding around tumor (linear areas of soft-tissue attenuation in the fat around the tumor), fat planes with adjacent viscera obscured, retroperitoneal fluid (fluid in the subcapsular, perirenal, or pararenal space), ascites (considered present only if peritoneal fluid was present beyond the cul-de-sac), peritoneal implants (enhancing soft-tissue masses lining the peritoneal cavity or surface of viscera), ipsilateral pleural effusion, and intratumoral hemorrhage (fluid-fluid levels, or high-attenuation areas with attenuation of 30–200 HU within the tumor on precontrast images, with attenuation less than that of calcification). In addition, if retroperitoneal fluid was present, it was further classified as subcapsular (crescent-shaped fluid collection following the contour of the kidney) or extracapsular. If fluid was present in the ipsilateral pleural cavity, the retroperitoneal space, or the peritoneal cavity beyond the cul-de-sac, the attenuation of the fluid was determined to differentiate between hemorrhagic and nonhemorrhagic fluid. This was done by drawing a 5-mm-diameter region of interest over the most hyperattenuating part of the fluid. The fluid was classified as nonhemorrhagic if the attenuation was less than 30 HU and hemorrhagic if the attenuation was 30 HU or greater.

Statistical Analysis

The sensitivity and specificity of CT in the detection of preoperative tumor rupture were calculated for each reviewer. For this purpose, the CT readings were made dichotomous by classifying a rating of 1–3 as negative and a rating of 4–5 as positive. The sensitivity and specificity calculations were done by comparing the CT reading (positive vs negative) with the findings from central pathology and/or surgical review (case vs control) by using 2 × 2 contingency tables. Interobserver agreement was analyzed by calculating the Cohen κ statistic in two ways: a simple κ using the dichotomous CT readings (negative vs positive) and a weighted κ using the raw CT ratings of 1–5. A weighted κ was of interest because it takes into account the degree of disagreement between observers and assigns weights to agreement according to how far apart categories are (12). A κ value of less than 0.20 was indicative of slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.00, almost perfect interobserver agreement.

The McNemar test was used to assess the relationship between the occurrence of individual CT findings and preoperative tumor rupture. To account for case-control matching by age and tumor weight, univariate conditional logistic regression models were fit (13). Multivariate, conditional, logistic regression (including all imaging criteria) with backward elimination was used to determine the most parsimonious model. In cases with retroperitoneal fluid, the relationship between the occurrence of preoperative rupture and location of fluid (subcapsular vs extracapsular) was assessed with a univariate conditional logistic regression model. To determine if the attenuation of fluid was further predictive of increased risk of preoperative tumor rupture within the subgroup of patients in whom fluid was detected, additional univariate logistic regression models were fit with the attenuation of the fluid as both a continuous and a dichotomous (<30 HU, ≥30 HU) variable. Analyses were performed for each reviewer individually. All analyses were performed by using software (SAS version 9.2; SAS, Cary, NC). P < .05 was considered indicative of a statistically significant difference.

Results

The distribution of age, sex, tumor laterality, and stage of disease for cases and controls is summarized in Table 1. The median interval between CT and primary nephrectomy was 3 days (mean, 4 days; range, 0–60 days). The respective sensitivity and specificity of CT in the detection of preoperative tumor rupture were 54% (36 of 67 cases; 95% confidence interval [CI]: 41.1%, 66.0%) and 88% (61 of 69 cases; 95% CI: 78.4%, 94.9%) for reviewer 1 and 70% (47 of 67 cases; 95% CI: 57.7%, 80.7%) and 88% (61 of 69 cases; 95% CI: 78.4%, 94.9%) for reviewer 2 (Table 2). Reviewer 1 correctly classified 71% of cases (97 of 136 cases) as to the presence or absence of tumor rupture, and reviewer 2 was accurate in 79% (108 of 136 cases). Both simple and weighted κ values indicated that there was substantial agreement in CT readings of tumor rupture between the two reviewers, with a simple κ statistic of 0.76 (95% CI: 0.65, 0.87; P < .0001) and a weighted κ statistic of 0.66 (95% CI: 0.58, 0.73; P < .0001) (Tables 3, E1 [online]).

Table 1.

Patient Characteristics for Cases and Controls

Note.—Except where indicated, numbers in parentheses are percentages.

^*Numbers in parentheses are ranges.

^†Disease stage was missing for one case because the CT scan of the chest was not submitted.

Table 2.

Cross Tabulation of CT Readings and Findings at Central Pathology and/or Surgical Review

Note.—Sensitivity and specificity were 54% (36 of 67 cases; 95% CI: 41.1%, 66.0%) and 88% (61 of 69 cases; 95% CI: 78.4%, 94.9%), respectively, for reviewer 1 and 70% (47 of 67 cases; 95% CI: 57.7%, 80.7%) and 88% (61 of 69 cases; 95% CI: 78.4%, 94.9%) for reviewer 2.

Table 3.

Cross Tabulation of CT Readings from Reviewers 1 and 2 with Use of Dichotomous Values and Simple κ

Note.—There was substantial agreement in CT readings of tumor rupture between the two reviewers, with a simple κ statistic of 0.76 (P < .0001, from the hypothesis test with null hypothesis of no agreement [κ = 0]).

The presence of a poorly circumscribed mass, fat stranding around the tumor, retroperitoneal fluid, ascites beyond the cul-de-sac, and ipsilateral pleural effusion were all significantly related to the presence of preoperative tumor rupture for both reviewers (P ≤ .02 for all) (Tables 4, 5; Figs 1, 2). For both reviewers, the odds for the presence of preoperative tumor rupture were highest in the presence of ascites beyond the cul-de-sac (reviewer 1: OR = 24.0, P = .0018; reviewer 2: OR = 18.0, P < .0001). Further classification of retroperitoneal fluid as subcapsular or extracapsular showed that extracapsular extension of retroperitoneal fluid was strongly predictive of rupture for both reviewers (reviewer 1: OR = 8.0, reviewer 2: OR = 21.8; P < .0001 for both). Although the presence of subcapsular fluid increased the odds of rupture for both reviewers, subcapsular fluid was not a statistically significant predictor of rupture for either reviewer (reviewer 1: OR = 2.2, P = .3184; reviewer 2: OR = 2.0, P = .3515). A backward-selected multivariate conditional logistic regression model for occurrence of preoperative rupture was fit separately for each reviewer. The final model for reviewer 1 indicated that ascites and fat stranding around the tumor were predictive of preoperative tumor rupture, with the presence of each corresponding to 18.4 (P = .0085) and 10.6 (P = .0045) times greater odds of occurrence, respectively. The order of removal was poorly circumscribed mass, ipsilateral pleural effusion, presence of retroperitoneal fluid, and fat planes with adjacent viscera obscured, with the least significant term dropping out at each step. The final model for reviewer 2 indicated that the presence of ascites and retroperitoneal fluid were predictive of preoperative tumor rupture, with the presence of each corresponding to 8.4 (P = .0058) and 4.9 (P = .0056) times greater odds of occurrence, respectively. The order of removal was fat stranding around tumor, presence of peritoneal implants, poorly circumscribed mass, and intratumoral hemorrhage.

Table 4.

Relationship between Preoperative Tumor Rupture and CT Findings

Note.—Data were determined with the McNemer test. NA = not applicable; that is, we were unable to perform the McNemar test owing to sparse data in some combinations of predictor and response.

Table 5.

Results of Univariate Conditional Logistic Regression Models with CT Findings as Predictors of Preoperative Tumor Rupture

Note.—NA = not applicable (we were unable to fit a logistic regression model because of sparse data in some combinations of predictor and response), OR = odds ratio.

^*Indicates the reference level for each variable. The OR is the increased odds of preoperative tumor rupture in comparison to this reference level.

Figure 1a:

Contrast-enhanced CT scans of abdomen in 3-year-old girl who presented with abdominal pain. (a) Scan shows a poorly circumscribed mass arising from upper pole of right kidney (arrow). (b, c) There is associated retroperitoneal fluid, both subcapsular (*) and extracapsular (arrowheads), with ascites in lower abdomen (arrows). Hemoperitoneum was encountered at laparotomy. Pathologic examination confirmed a ruptured Wilms tumor penetrating the renal capsule, with tumor extension into perinephric hematoma.

Figure 2:

Contrast-enhanced CT scan of abdomen in 2-year-old patient with palpable abdominal mass shows a well-circumscribed left renal mass with no associated fat stranding or retroperitoneal fluid. There was no ascites in abdomen or pelvis. Patient underwent complete left nephrectomy. Pathologic examination confirmed Wilms tumor with no evidence of rupture.

Figure 1b:

Contrast-enhanced CT scans of abdomen in 3-year-old girl who presented with abdominal pain. (a) Scan shows a poorly circumscribed mass arising from upper pole of right kidney (arrow). (b, c) There is associated retroperitoneal fluid, both subcapsular (*) and extracapsular (arrowheads), with ascites in lower abdomen (arrows). Hemoperitoneum was encountered at laparotomy. Pathologic examination confirmed a ruptured Wilms tumor penetrating the renal capsule, with tumor extension into perinephric hematoma.

Figure 1c:

Contrast-enhanced CT scans of abdomen in 3-year-old girl who presented with abdominal pain. (a) Scan shows a poorly circumscribed mass arising from upper pole of right kidney (arrow). (b, c) There is associated retroperitoneal fluid, both subcapsular (*) and extracapsular (arrowheads), with ascites in lower abdomen (arrows). Hemoperitoneum was encountered at laparotomy. Pathologic examination confirmed a ruptured Wilms tumor penetrating the renal capsule, with tumor extension into perinephric hematoma.

The attenuation of pleural effusion, retroperitoneal fluid, or ascites did not have a statistically significant correlation with rupture either as a dichotomous or as a continuous variable (Table 6).

Table 6.

Results of Univariate Conditional Logistic Regression Models with Attenuation of Fluid Detected at CT as Predictor of Preoperative Tumor Rupture

Note.—NA = not applicable; we were unable to fit a logistic regression model, either because of sparse data in some combinations of predictor and response or the optimization procedure did not converge.

^*Indicates the reference level for each variable; the OR is the increased odds of preoperative tumor rupture in comparison tothis reference level.

Discussion

Preoperative tumor rupture is a significant risk factor for abdominal recurrence in children with Wilms tumor. In North America, the diagnosis of preoperative tumor rupture is based on initial surgical evaluation, whereas in Europe it is based on baseline imaging and surgical and/or histologic findings after neoadjuvant chemotherapy. Recent data from the International Society of Pediatric Oncology suggest that all cases with baseline imaging findings of tumor rupture should be classified as stage IIIc and treated with whole abdominal radiation (14). However, the diagnostic performance of abdominal CT in the detection of preoperative Wilms tumor rupture or the imaging features that can be used to establish a diagnosis of tumor rupture have not been established in a large series with surgical confirmation. Because the Children’s Oncology Group’s recommended initial treatment of Wilms tumor is primary nephrectomy, this gave us the unique opportunity to test baseline imaging findings for associations with surgical and/or histologic findings of tumor rupture.

We found that two experienced pediatric radiologists were able to correctly identify the presence or absence of preoperative tumor rupture on the basis of CT findings in 71% and 79% of cases. Although the specificity of the two radiologists was high (88%), the sensitivity was relatively low (54% and 70%) in identifying Wilms tumor cases with preoperative tumor rupture. The interobserver agreement for the diagnosis of preoperative tumor rupture between the two radiologists was substantial.

In our study, several imaging findings were significantly associated with preoperative tumor rupture for both radiologists. These included a poorly circumscribed mass, fat stranding around the tumor, extracapsular retroperitoneal fluid, ascites beyond the cul-de-sac, and ipsilateral pleural effusion. For both reviewers, the odds of rupture were highest in the presence of ascites beyond the cul-de-sac. In our study, ascites was considered present only when there was free peritoneal fluid beyond the cul-de-sac. We emphasize that a small amount of fluid in the cul-de-sac can often be seen at diagnosis and should not be misinterpreted as tumor rupture (10). Although other authors have suggested that a hyperattenuating hemoperitoneum is indicative of tumor rupture, in our study the attenuation of the ascitic fluid did not have a significant association with tumor rupture (10). Regression analysis showed that, in the presence of ascites, fat stranding around the tumor and the presence of retroperitoneal fluid were other significant CT features for reviewers 1 and 2, respectively.

We did not find an association between the presence of peritoneal implants and preoperative tumor rupture. The imaging findings of peritoneal implants in the setting of stage III Wilms tumor have been described in a series of four cases (11). In our study, only 12 cases of preoperative rupture were noted to have peritoneal implants on the basis of surgical evaluation. Both radiologists independently identified only two of 12 cases with peritoneal implants. Reviewer 1 did not classify any of the controls as having peritoneal implants, so the statistical significance of this CT feature could not be tested. For reviewer 2, however, the presence of peritoneal implants was not significantly correlated with a diagnosis of preoperative rupture. Studies in adults have shown that CT has limited sensitivity in the detection of peritoneal spread of malignancy (15,16). We speculate that CT detection of peritoneal implants would be even less accurate in young children compared with adults owing to the absence of significant intraabdominal fat, which aids delineation of normal anatomy at CT. The lack of intraluminal bowel contrast in children who underwent imaging while under sedation and the anatomic distortion caused by the relatively large tumor size to abdominal cavity size ratio in children makes CT detection of peritoneal tumor nodules even more challenging.

Previous investigators have stated that intratumoral hemorrhage is indicative of tumor rupture (7). However, intratumoral hemorrhage was not a significant predictor of tumor rupture in our study. Intratumoral hemorrhage is typically most evident on noncontrast studies as an area of increased attenuation or as an area of T1 high signal intensity on MR images. Intratumoral hemorrhage has been reported in up to 27% of Wilms tumors at MR imaging, which is higher than the reported 7.5%–13% prevalence of preoperative tumor rupture (17). In our study, we did not evaluate MR images of the abdomen because most institutions perform abdominal CT at baseline imaging and MR imaging of the abdomen was available in only five of 136 subjects. Precontrast CT scans of the abdomen were available in our cohort in only 12 of 136 subjects, which limits our ability to detect intratumoral hemorrhage. However, routine performance of multiphase and precontrast images is not recommended in pediatric oncology patients to limit radiation exposure (18). In the study by Brisse et al (7), all patients with intratumoral hemorrhage at imaging were classified as having retroperitoneal rupture. They defined hemorrhage as a spontaneously hyperattenuating area on unenhanced CT scans. However, it is unclear as to how many of their 57 cases had precontrast images. All of their cases with intratumoral hemorrhage received 4–6 weeks of neoadjuvant chemotherapy before surgical and/or histologic evaluation. Our cohort consists of cases with primary nephrectomy, hence the surgical and/or histologic confirmation of the imaging findings.

This study had a number of limitations. The CT scans were obtained from multiple institutions, resulting in variations in scanning parameters, the administration of oral contrast material, and the availability of multiplanar reconstructions. However, this limitation is inherent in a multi-institutional study in which an imaging question is not the primary objective. Second, the reference standard for the determination of tumor rupture relied on the central review of local surgical notes. Although the study protocol provides guidelines to be used by institutional surgeons for the documentation of findings supporting rupture, these documentations are based on the subjective assessments and expertise of the institutional surgeons. To overcome this limitation, we incorporated central review of the pathology specimens in the reference standard and identified cases and controls based on the presence or absence of tumor rupture at both central surgical and pathologic reviews. Third, our study cohort included only patients who underwent primary nephrectomy. Patients with obvious clinical and/or radiologic signs of rupture could potentially have been excluded if the institution decided to treat them with initial chemotherapy followed by nephrectomy. This would result in underestimation of the sensitivity of CT in the detection of preoperative tumor rupture.

In conclusion, CT has high specificity but relatively low sensitivity in the detection of preoperative rupture in patients with Wilms tumor. The presence of ascites beyond the cul-de-sac, irrespective of attenuation, was the most useful CT indicator of preoperative Wilms tumor rupture. In the presence of ascites, fat stranding around the tumor and the presence of retroperitoneal fluid were also highly predictive of rupture. According to the guidelines of the Children’s Oncology Group, surgical evaluation remains the standard of reference in the detection of Wilms tumor rupture in North America. However, CT detection of preoperative Wilms tumor rupture may aid the surgeon in preoperative planning before laparotomy. Further research is needed to determine whether CT findings of rupture at initial presentation have prognostic significance in the long-term follow up of these patients.

Advances in Knowledge.

• • CT has moderate specificity (88%, 61 of 69 cases) but relatively low sensitivity (54% and 70%, 36 and 47 of 67 cases, respectively) in the detection of preoperative Wilms tumor rupture.

• • Ascites beyond the cul-de-sac, irrespective of attenuation, is most predictive of preoperative Wilms tumor rupture; in the presence of ascites, fat stranding around the tumor and the presence of retroperitoneal fluid are also highly predictive of rupture.

• • Extracapsular retroperitoneal fluid is more likely to be associated with rupture than subcapsular retroperitoneal fluid.

Implication for Patient Care.

• • The reported CT findings of Wilms tumor rupture can be applied in clinical practice to suggest the presence of preoperative tumor rupture in patients with Wilms tumor.

Disclosures of Conflicts of Interest: G.K. No relevant conflicts of interest to disclose. A.N. No relevant conflicts of interest to disclose. F.H. No relevant conflicts of interest to disclose. E.M. No relevant conflicts of interest to disclose. J.G. No relevant conflicts of interest to disclose. E.J.G. No relevant conflicts of interest to disclose. P.F.E. No relevant conflicts of interest to disclose. E.J.P. No relevant conflicts of interest to disclose. N.R. No relevant conflicts of interest to disclose. P.G. No relevant conflicts of interest to disclose. J.S.D. No relevant conflicts of interest to disclose.