Imaging Characteristics of Nephrogenic Rests Versus Small Wilms Tumors: A Report From the Children’s Oncology Group Study AREN03B2

Abstract

OBJECTIVE.

Distinguishing nephrogenic rests from small Wilms tumors can be challenging. This retrospective study was performed to determine if imaging characteristics can be used to distinguish nephrogenic rests from Wilms tumors.

MATERIALS AND METHODS.

All cases of pathologically confirmed nephrogenic rests and Wilms tumors smaller than 5 cm in maximum dimension on imaging in patients younger than 5 years old were identified from the Children’s Oncology Group AREN03B2 study (July 2006–August 2016). Exclusion criteria were chemotherapy before pathologic evaluation or more than 30 days between imaging and surgery; in addition, patients with nephrogenic rests occurring within or juxtaposed to a Wilms tumor and patients with diffuse hyperplastic perilobar nephroblastomatosis were excluded. Two radiologists who were blinded to pathology results assessed all lesions. The two-sample t test was used for continuous variables, and the Fisher exact test was used for categoric variables. ROC analysis was performed to determine the optimal size cutoff for distinguishing between nephrogenic rests and Wilms tumors.

RESULTS.

Thirty-one pathologically confirmed rests (20 perilobar, 11 intralobar) and 26 Wilms tumors smaller than 5 cm met the eligibility criteria for study inclusion. The median diameter of the nephrogenic rests was 1.3 cm (range, 0.7–3.4 cm) and the median diameter of the Wilms tumor was 3.2 cm (range, 1.8–4.9 cm) (p < 0.001). Imaging findings supportive of Wilms tumors were spherical (p < 0.001) and exophytic (p < 0.001) lesions. Perilobar rests (17/20) were more likely to be homogeneous than intralobar rests (3/11) or Wilms tumor (3/26) (p < 0.001). ROC analysis showed that the optimal size cutoff for distinguishing between nephrogenic rests and Wilms tumors was 1.75 cm.

CONCLUSION.

In children younger than 5 years old, the diagnosis of a Wilms tumor should be favored over a nephrogenic rest when a renal mass is spherical, exophytic, or larger than 1.75 cm. Homogeneity favors the diagnosis of perilobar nephrogenic rests, whereas intralobar rests and Wilms tumors are more likely to be inhomogeneous.

Nephrogenic rests represent residual primitive metanephric tissue that persists past 36 weeks’ gestational age [1]. Although nephrogenic rests can have variable progression to sclerosis, dormancy, or hyperplasia, neoplastic transformation is a known complication. Approximately 28–40% of unilateral Wilms tumors and 90–100% of bilateral Wilms tumors are associated with nephrogenic rests [2, 3]. There are two major types of nephrogenic rests: perilobar and intralobar. These two types of nephrogenic rests differ in their location within the kidney (perilobar, along the periphery of the renal lobe; intralobar, within the renal lobe); cell type composition; and distribution based on age, sex, syndrome associations, and risk of malignant transformation [1].

Although accurate distinction between a nephrogenic rest and a Wilms tumor is highly clinically relevant, this distinction is challenging with both imaging findings and needle biopsy results [4]. Pathologic differentiation of a Wilms tumor from a nephrogenic rest relies largely on the evaluation of the interface of the lesion with the adjacent kidney [5]. Perilobar nephrogenic rests have an abrupt interface with the adjacent kidney and lack a pseudocapsule, whereas Wilms tumors reliably show a pseudocapsule separating the lesion from the adjacent kidney [5]. In addition, perilobar nephrogenic rests are homogeneous, often with the same lobulated appearance that is characteristic of the developing kidney, whereas Wilms tumors lack this subtle evidence of organization. Intralobar nephrogenic rests intermingle with the normal renal parenchyma at the periphery of the lesion, which is in contrast to the abrupt interface seen with a perilobar nephrogenic rest and in contrast to the encapsulation characteristic of a Wilms tumor [3]. Other pathologic features commonly associated with nephrogenic rests are an oblong contour rather than a spherical contour, although this association is not universal. It is noteworthy that the key features distinguishing nephrogenic rests from Wilms tumors depend on the availability of information about the interface between the lesion and the adjacent kidney, and this information is reliably present only in the setting of an excisional biopsy [6]. Finally, we should note that hyperplasia within both types of nephrogenic rests may increase the difficulty in differentiation from a Wilms tumor. However, the nature of the interface with the adjacent kidney, as described, remains a reliable feature even in the presence of hyperplasia [5].

Because the management of a Wilms tumor is distinct from the management of a nephrogenic rest, characterization of radiologic features that could be used to distinguish the two may be helpful in avoiding unnecessary surgical interventions, especially in the setting of predisposition syndromes in which preservation of renal parenchyma is paramount [7].

To our knowledge, there are three previous studies that have investigated the imaging appearance of nephrogenic rests [8-10]. These studies have suggested that the primary imaging finding that differentiates nephrogenic rests from Wilms tumors is homogeneity of the lesion and elliptic shape of perilobar nephrogenic rests. The limitations of these studies include small sample size and pathologic evaluation after initiation of chemotherapy, which can alter the histologic characteristics of the neoplastic cells. In addition, these studies evaluated Wilms tumors that occurred within or juxtaposed to nephrogenic rests. The Children’s Oncology Group (COG) has an ongoing study titled the “Renal Tumors Classification, Biology, and Banking Study,” which is also known as the “AREN03B2 trial” (National Clinical Trials identifier NCT00898365 on Clinical Trials.gov). Central review of imaging and pathology samples in the AREN03B2 trial gave us the opportunity to investigate the diagnostic imaging findings of nephrogenic rests versus small Wilms tumors in a cohort of patients whose lesions had been surgically sampled and pathologically evaluated before any medical intervention.

Materials and Methods

The study cohort was composed of patients with pathologically proven nephrogenic rests and Wilms tumors who were enrolled in the AREN03B2 trial between July 2006 and August 2016. The AREN03B2 trial is an ongoing HIPAA-compliant study approved by the National Cancer Institute Pediatric Central Institutional Review Board as well as by local institutional review boards of all participating institutions. All subjects or their guardians signed informed assent or consent as appropriate. As of August 2016, 5530 renal tumor cases of all types had been enrolled in the AREN03B2 trial. Inclusion criteria for this ongoing study are a first-time occurrence of a renal tumor in a child or young adult younger than 30 years old. At the time of enrollment, the institution is required to submit a contrast-enhanced CT study or a contrast-enhanced MRI study of the abdomen and pelvis for central imaging review by study radiologists. Local institutions are also required to submit operative notes and pathology slides from biopsy or nephrectomy specimens for central review by the study surgeons and pathologists, respectively.

Inclusion criteria for our project were all cases for which the diagnosis of an intralobar or perilobar nephrogenic rest was made at the time of central pathology review, using the criteria outlined earlier, and a comparison group of all small Wilms tumors (maximum diameter on imaging, < 5 cm) in patients younger than 5 years old enrolled in the AREN03B2 trial. A size cutoff of less than 5 cm on imaging for Wilms tumors was used because prior studies have shown that nephrogenic rests are typically approximately 2 cm with a maximum reported dimension of 5 cm [9]. An age cutoff of 5 years was used because nephrogenic rests are unusual beyond this age [2]. The following patients were excluded: patients who received chemotherapy before pathologic evaluation and surgical sampling, patients with nephrogenic rests that occurred within or juxtaposed to a Wilms tumor, and patients who underwent surgical sampling more than 30 days after imaging. In addition, we excluded patients with diffuse hyperplastic perilobar nephroblastomatosis because this entity has a very distinct imaging appearance with a rind of abnormal nephrogenic tissue that surrounds the renal cortex like a mantle while preserving the reniform shape [10]. All nephrogenic rests were further pathologically classified into subtypes of hyperplastic, regressing, or adenomatous using the published criteria for pathologic evaluation [3, 5].

Imaging Evaluation

All imaging studies for this retrospective review were reviewed independently by a pediatric radiologist with 14 years of experience and a senior radiology resident (postgraduate year 5); discrepancies were resolved by consensus before comparative analysis with pathology reports. This imaging review was performed blinded to the institutional report and institutional and central pathologic evaluations. In patients with multiple renal lesions, the radiology reviewers attempted to identify the index (i.e., pathologically evaluated) lesion in the kidney using the description of the treating surgeon in the operative note.

The radiology reviewers assessed the anatomic location, size, margin, shape, homogeneity, and contrast enhancement of the renal mass with respect to the normal renal parenchyma. Tumor dimensions were measured in the axial plane (longest diameter and orthogonal measurement) and maximum cepha-locaudal dimension. All measurements were initially made by the junior reviewer and were confirmed by the senior reviewer. For the purpose of this study, the shape of the mass was considered spherical if all three dimensions were within 20% of each other and nonspherical if there was a difference in dimensions of more than 20%. Lesions were characterized as having a sharp interface if the margin between the mass and normal renal parenchyma could be clearly delineated circumferentially and as being ill defined if the margins were not sharply demarcated circumferentially. Nonenhancing areas within the tumor were characterized as necrosis. The lesion was considered exophytic if the outer contour was past that of the normal renal cortex or if the lesion bulged into the renal sinus or pelvis. The location of the tumor with respect to the pelvicaliceal system (cortex, corticomedullary junction, or medulla) was also noted.

Statistical Analysis

The two-sample t test was used to compare continuous characteristics, and the Fisher exact test was used to assess associations between categoric variables. ROC analysis was performed to determine the optimal size cutoff for differentiating nephrogenic rests from Wilms tumors. We used the R software package (version 3.6.1, The R Foundation) for all analyses.

Results

A total of 52 pediatric patients enrolled in the AREN03B2 trial met the inclusion criteria. Thirty-one rests were identified from 29 patients, and 26 Wilms tumors were identified from 25 patients. Two patients had both one rest and one Wilms tumor (separately excised) included in the study. The imaging modality available for central review was as follows: contrast-enhanced CT of the abdomen and pelvis in 34 cases, unenhanced and contrast-enhanced MRI in 10 cases, and both CT and MRI in eight cases. DWI was included in five of the 18 cases with MRI studies that were submitted for central review.

For the 26 Wilms tumors, 22 had a solitary lesion in the ipsilateral kidney, one case had two lesions (both Wilms tumors), and three cases had three lesions (multifocal tumors). Regarding the contralateral kidney in the 26 Wilms tumor cases, there was no lesion in the contralateral kidney in 20 Wilms tumor cases; one lesion in the contralateral kidney in five cases (a Wilms tumor in one case, a rest in two cases, a scar in one case, and no surgical correlate in one case); and three lesions in the contralateral kidney, one of which was surgically excised and shown to be a rest, in one case.

For the 31 nephrogenic rests, 16 had a solitary lesion in the ipsilateral kidney, nine had two lesions in the ipsilateral kidney, two had three lesions in the ipsilateral kidney, and four had more than five lesions in the ipsilateral kidney. The number of lesions biopsied was at the discretion of the treating surgeon, but none of the nephrogenic rest cases had ipsilateral Wilms tumors. Regarding the contralateral kidney in the nephrogenic rest cases, eight rests had no lesion in the contralateral kidney, 11 had a single lesion in the contralateral kidney (10 Wilms tumors and one tumor that was not biopsied), three had two lesions in the contralateral kidney (Wilms tumors in two cases and nephrogenic rests in one case), two had three lesions in the contralateral kidney (Wilms tumors in both cases), and four had four or more lesions in the contralateral kidney (Wilms tumors in two cases, lesion not biopsied in one case, and rest in one case). The contralateral kidney was absent in one patient. In summary, 17 of 29 cases with rests had a synchronous contralateral Wilms tumor, and the remaining 12 cases did not have a synchronous Wilms tumor.

The pathologic diagnoses of the 31 nephrogenic rests, as determined by central pathology review, were as follows: There were 20 perilobar rests and 11 intralobar rests. Of the 20 perilobar rests, 15 were hyperplastic; one, hyperplastic and adenomatous; one, regressing; and three, neither hyperplastic nor regressing Of the 11 intralobar rests, four were hyperplastic; one, hyperplastic and adenomatous; and six, neither hyperplastic nor regressing.

Clinical Features

The clinical data for both nephrogenic rests and Wilms tumors are summarized in Table 1. Cases with nephrogenic rests were younger than those with Wilms tumor (p = 0.002) with ages ranging from 0.2 to 4.0 years (median, 1.1 years) and from 0.4 to 4.5 years (median, 3.1 years) for nephrogenic rests and Wilms tumor, respectively. If excluding patients with underlying predisposition syndromes, cases with nephrogenic rests were still younger than those with Wilms tumors (p = 0.028), with ages ranging from 0.2 to 4.0 years (median, 1.9 years) and from 0.4 to 4.5 years (median, 3.3 years) for nephrogenic rests and Wilms tumors, respectively. Underlying predisposition syndromes were more commonly reported in patients with rests than in those with Wilms tumor (35.5% rests vs 11.5% Wilms tumor; p = 0.06). For rests, these cases included five cases of hemihypertrophy and four cases of Denys-Drash syndrome. For Wilms tumor, these cases included two cases of hemihypertrophy and one case of Denys-Drash syndrome. There was no sex predisposition of rests versus Wilms tumor (p = 0.79).

TABLE 1:

Clinical Characteristics of 31 Nephrogenic Rests and 26 Small Wilms Tumors

Clinical Characteristics	Nephrogenic Rests (n = 31)	Wilms Tumors (n = 26)	pa
Age (y)			0.002
Median	1.1	3.1
Minimum	0.2	0.4
Maximum	4.0	4.5
Sex			0.79
Female	19 (61.3)	17 (65.4)
Male	12 (38.7)	9 (34.6)
Predisposition syndrome			0.06
No	20 (64.5)	23 (88.5)
Yes	11 (35.5)	3 (11.5)
Beckwith-Wiedemann syndrome	1 (9.1)	0 (0)
Denys-Drash syndrome	2 (18.2)	1 (33.3)
Denys-Drash syndrome with neonatal acute kidney injury	2 (18.2)	0 (0)
Hemihypertrophy	5 (45.5)	1 (33.3)
Hemihypertrophy and hemangiomas	0 (0)	1 (33.3)
4q23-q25 deletion	1 (9.1)	0 (0)

Note—Thirty-one rests were identified from 29 patients, and 26 small Wilms tumors were identified from 25 patients. Two patients had both one rest and one Wilms tumor (separately excised) included in the study. Unless indicated otherwise, results are number (%) of lesions. Percentages may not add up to 100 because of rounding.

^aTwo-sample t test was used for continuous characteristics, and Fisher exact test was used for categoric characteristics.

Imaging Findings

Tables 2-4 summarize the imaging findings of our study. The maximum dimension of Wilms tumors on imaging was significantly larger than that of rests (p < 0.001): Wilms tumors ranged from 1.8 to 4.9 cm (median, 3.2 cm) and rests ranged from 0.7 to 3.4 cm (median, 1.3 cm) (Fig. 1). Because all the rests were 4 cm or less in maximum dimension, a subset analysis was performed after excluding the larger Wilms tumors (> 4 cm), and this difference remained statistically significant (Wilms tumors ≤ 4 cm vs nephrogenic rests: 1.8–3.9 cm vs 0.7–3.4 cm; p < 0.001).

TABLE 2:

Imaging Characteristics of Nephrogenic Rests and Wilms Tumors

Imaging Characteristic	Nephrogenic Rests (n = 31)	All Wilms Tumors (n = 26)	Wilms Tumors ≤ 4 cm (n = 19)	pa
Tumor diameter (cm)				< 0.001
Median	1.3	3.2	2.9
Minimum	0.7	1.8	1.8
Maximum	3.4	4.9	3.9
Shape				< 0.001
Spherical	6 (19.4)	18 (69.2)
Not spherical	25 (80.6)	8 (30.8)
Exophytic				< 0.001
No	19 (61.3)	3 (11.5)
Yes	12 (38.7)	23 (88.5)
Interface of lesion with adjacent kidney				0.42
Indistinct	15 (48.4)	9 (34.6)
Sharp	16 (51.6)	17 (65.4)

^aTwo-sample t test was used for continuous characteristics, and Fisher exact test was used for categoric characteristics.

TABLE 4:

Imaging Characteristics of Hyperplastic and Nonhyperplastic Intralobar and Perilobar Rests

Imaging Characteristic	Intralobar Rests (n = 11), No. (%)		pa	Perilobar Rests (n = 20), No. (%)		pa
	Hyperplastic	Nonhyperplastic		Hyperplastic	Nonhyperplastic
Density			0.55			1
Homogeneous	2 (40.0)	1 (16.7)		13 (81.3)	4 (100)
Inhomogeneous	3 (60.0)	5 (83.3)		3 (18.8)	0 (0)
Location			0.74			0.37
Cortex	1 (20.0)	1 (16.7)		15 (93.8)	3 (75.0)
Corticomedullary junction	2 (40.0)	4 (66.7)		1 (6.3)	0 (0)
Medulla	2 (40.0)	1 (16.7)		0 (0)	1 (25.0)
Shape			1			1
Spherical	2 (40.0)	2 (33.3)		2 (12.5)	0 (0)
Not spherical	3 (60.0)	4 (66.7)		14 (87.5)	4 (100)
Exophytic			0.24			0.12
No	2 (40.0)	5 (83.3)		8 (50.0)	4 (100)
Yes	3 (60.0)	1 (16.7)		8 (50.0)	0 (0)
Interface of lesion with adjacent kidney			0.24			0.014
Indistinct	2 (40.0)	5 (83.3)		4 (25.0)	4 (100)
Sharp	3 (60.0)	1 (16.7)		12 (75.0)	0 (0)

^aFisher exact test.

Fig. 1—

Box-and-whisker plot shows size distributions of Wilms tumors versus nephrogenic rests using largest dimension of lesion on imaging. Boxes show spread of middle 50% of data, whiskers extend to lower and upper limits, and circles show outliers.

In this cohort of patients less than 5 years old and with lesions smaller than 5 cm, ROC analysis shows an optimal maximal diameter cutoff between rests and Wilms tumors of 1.75 cm (Fig. 2). This lesion size cutoff corresponds to a sensitivity of 100% and a specificity of 81.3% for identifying Wilms tumors. A size cutoff of less than 1.75 cm had a negative predictive value of 100% and a positive predictive value of 81.0% for the diagnosis of a Wilms tumor.

Fig. 2—

ROC curve shows optimal lesion size cutoff between nephrogenic rests and Wilms tumors. Size cutoff of 1.75 cm yielded highest diagnostic accuracy for identifying Wilms tumors (sensitivity, 100%; specificity, 81.0%).

A spherical shape of the lesion was significantly more common in Wilms tumors (69.2%), followed by intralobar rests (36.4%) and perilobar rests (10.0%) (p < 0.001) (Fig. 3 and Tables 2 and 3). Nonspherical-shaped rests were most commonly noted to be either elliptical or triangular (Fig. 4). Exophytic lesions were more likely (p < 0.001) to be Wilms tumor (88.5%) than rests (all rests, 38.7%; intralobar, 36.4%; perilobar, 40.0%) (Fig. 5). Exophytic lesions were larger than nonexophytic lesions, with a median size of 2.7 cm (range, 1.0–4.9 cm) and 1.2 cm (range, 0.7–4.7 cm), respectively (p < 0.001). The imaging interface between the lesion and adjacent kidney (sharp vs indistinct) was not a statistically significant discriminator between nephrogenic rests versus Wilms tumors (p = 0.42). A possible, but not definite, pseudocapsule was identified by imaging in four of the Wilms tumor cases only on MRI, seen as a T2-hypointense rim around the tumor. A pseudocapsule was seen in none of the rests by either radiology reviewer.

Fig. 3—

3-year-old girl with known hemihypertrophy noted to have new left renal mass on sonography. A and B, Axial contrast-enhanced CT (A) and unenhanced T2-weighted MR (B) images show spherical, inhomogeneous 3.0-cm lesion at upper pole of left kidney. These imaging findings favor diagnosis of Wilms tumor. Subtle T2-hypointense pseudocapsule (arrow, B) can also be seen on MR image. Wilms tumor was confirmed by pathologic evaluation after partial nephrectomy.

TABLE 3:

Imaging Characteristics of Intralobar Rests, Perilobar Rests, and Wilms Tumors

Imaging Characteristic	Intralobar Rests (n = 11)		Perilobar Rests (n = 20)		Wilms Tumors (n = 26)		pa
	No.	%	No.	%	No.	%
Density							< 0.001
Homogeneous	3	27.3	17	85.0	3	11.5
Inhomogeneous	8	72.7	3	15.0	23	88.5
Location							0.44
Cortex	2	18.2	18	90.0	13	50.0
Corticomedullary junction	6	54.5	1	5.0	7	26.9
Medulla	3	27.3	1	5.0	6	23.1
Shape							< 0.001
Spherical	4	36.4	2	10.0	18	69.2
Not spherical	7	63.6	18	90.0	8	30.8
Exophytic							< 0.001
No	7	63.6	12	60.0	3	11.5
Yes	4	36.4	8	40.0	23	88.5
Interface							0.28
Indistinct	7	63.6	8	40.0	9	34.6
Sharp	4	36.4	12	60.0	17	65.4

^aFisher exact test.

Fig. 4—

1-year-old boy with known hemihypertrophy with new renal mass identified on renal sonography. A and B, Axial contrast-enhanced CT (A) and contrast-enhanced T1-weighted fat-saturated MR (B) images show ellipsoid, homogeneous intensity lesion (maximum dimension, 1.6 cm) along posterior cortex of right kidney. Lesion does not bulge beyond renal cortex. Imaging findings support diagnosis of perilobar nephrogenic rest, which was confirmed by pathologic evaluation after partial nephrectomy.

Fig. 5—

3-year-old girl with incidentally detected renal mass on sonography during follow-up for vesicoureteral reflux.

A and B, Axial (A) and sagittal (B) contrast-enhanced CT images of abdomen in portal venous phase show exophytic and fairly well-circumscribed mass (maximum dimension, 2.0 cm) arising from right kidney mid pole. Exophytic appearance favors Wilms tumor over nephrogenic rest. Inhomogeneous attenuation also favors Wilms tumor over perilobar rest; however, lesion is not spherical. Diagnosis of Wilms tumor was confirmed at partial nephrectomy.

Evaluation of attenuation and signal intensity of the lesions showed that homogeneity was more often seen with perilobar rests (85.0%), whereas inhomogeneity of lesion was more commonly associated with intralobar rests (72.7%) and Wilms tumors (88.5%) (p < 0.001) (Figs. 5 and 6 and Table 3).

Fig. 6—

8-month-old boy with incidentally detected left renal mass during follow-up for vesicoureteral reflux and pelvocaliectasis.

A and B, Axial T2-weighted (A) and contrast-enhanced T1-weighted fat-saturated (B) MR images show inhomogeneous 1.85-cm left renal mass. Imaging findings of lesion inhomogeneity and appearance of lesion being mildly exophytic into renal sinus favor diagnosis of Wilms tumor. Left nephrectomy led to diagnosis of intralobar nephrogenic rest.

In terms of location within the kidney, there was no significant difference when comparing intralobar rests, perilobar rests, and Wilms tumors (p = 0.44). Intralobar rests were more likely to be located at the corticomedullary junction or medulla (81.8%) when compared with perilobar rests (10.0%) and Wilms tumors (50.0%). Further evaluation of imaging findings of hyperplastic rests versus nonhyperplastic rests showed no statistically significant differences in rest attenuation and signal intensity, shape, exophytic appearance, or interface with adjacent renal parenchyma (all, p > 0.01) (Table 4).

Of the four Wilms tumor cases with DWI included on the MRI study, all four showed restricted diffusion subjectively. The single case of a rest with DWI available did not show diffusion restriction; however, the lesion was 6 mm in diameter at the superior pole of the kidney and was likely too small to resolve on axial DW images.

Discussion

In our study of 52 young children (< 5 years old) with nephrogenic rests and small Wilms tumors, we have shown that there is overlap in imaging findings between nephrogenic rests and Wilms tumors and no single imaging finding can fully differentiate a nephrogenic rest from a Wilms tumor. This is especially true for intralobar nephrogenic rests, which are more often inhomogeneous and spherical as compared with perilobar rests (72.7% and 36.4%, respectively, in this series). Although a pseudocapsule is a reliable differentiating factor between Wilms tumors and nephrogenic rests at pathology, a pseudocapsule could be identified on imaging in only four of the 26 Wilms tumors and in none of the nephrogenic rests. Prior studies comparing the imaging findings of nephrogenic rests versus those of Wilms tumors have shown that a capsulelike interface may be inconsistently visualized in Wilms tumors [9]. We used the margin of the lesion (sharp vs indistinct) as a surrogate marker to represent the presence of a pseudocapsule and found this imaging feature to be a poor discriminator between nephrogenic rests versus Wilms tumors on both CT and MRI. This is likely because the pseudocapsule is a microscopic finding and is below the spatial resolution of anatomic imaging [11, 12].

Our study cohort was selected from a large database of renal tumors with more than 5000 cases. Comparing the size of nephrogenic rests with the size of small Wilms tumors (< 5 cm), we propose that a size cutoff of 1.75 cm provides the highest accuracy in differentiating between nephrogenic rests and Wilms tumors. None of the Wilms tumors in our cohort was smaller than 1.75 cm in maximum dimension, whereas 81.0% of the rests were less than or equal to 1.75 cm in maximum dimension. Although these results support the current conservative approach in the COG renal tumor studies in which all lesions larger than 1 cm are treated as a potential tumor [13], our results suggest that a size cutoff of 1.75 cm provides higher accuracy in differentiating between nephrogenic rests and Wilms tumors. Because nephrogenic tumors are precursor lesions for Wilms tumors, it is intuitive that the mean size of nephrogenic rests is smaller than that of Wilms tumors even when limiting to a cohort of small (< 5 cm) Wilms tumors.

In a study by Rohrschneider et al. [9], most nephrogenic rests were smaller than 2 cm, whereas the smallest Wilms tumors in that study was 3 cm in diameter. That study [9], however, did report nephrogenic rests measuring up to 5 cm, which is the basis for our rationale to use a 5-cm cutoff for Wilms tumor in our current investigation. A limitation of that study study was that all patients were treated according to the International Society of Pediatric Oncology protocol, which meant that they received chemotherapy for 6 weeks before surgical sampling. Hence, there was a minimum of a 6-week interval between the imaging and surgical sampling, and the histology of the lesions could have evolved secondary to chemotherapy-induced changes [9].

Prior studies have suggested that Wilms tumors are more likely to be spherical and that nephrogenic rests are more often plaque-like, ovoid, or lenticular [8]. It is postulated that clonal expansion and rapid proliferation of cells in a neoplasm accounts for their spherical shape as opposed to the plaquelike appearance of rests [5]. These observations are supported by our results that showed a spherical shape to have a strong association with a Wilms tumor over a rest. A novel finding in our study is that a surface or peripheral lesion bulging beyond the renal contour (exophytic) is more likely to be a Wilms tumor than a rest. This association is likely because tumors form in precursor lesions secondary to proliferation of neoplastic cells that result in the bulging contours of the tumors and protrusion of the tumors beyond the surface of the renal contour. This is supported by our finding that exophytic lesions were significantly larger than nonexophytic lesions (p < 0.001).

Although previous studies have suggested that homogeneity of the lesion suggests a nephrogenic rest [8], we note that intralobar rests were more likely (72.7% in our series) to be inhomogeneous in signal intensity and attenuation, just like Wilms tumors. Hence, attenuation and signal intensity of lesions cannot be used to differentiate between intralobar nephrogenic rests and Wilms tumors—a novel finding in our study. The heterogeneous attenuation and intensity of intralobar rests correlates with their histologic appearance because they are composed of stroma, blastemal cells, and epithelial cells [1, 14]. Perilobar rests are typically stroma-poor and consist predominantly of blastemal and epithelial cells, which accounts for their more homogeneous appearance on imaging [1, 14].

Distinguishing between perilobar rests and intralobar rests is important because their syndromic associations and risk for malignant transformation are different [3]. Perilobar nephrogenic rests are associated with overgrowth syndromes including hemihypertrophy and Beckwith-Wiedemann syndrome, whereas intralobar rests are commonly found in WAGR (Wilms tumor, aniridia, genital anomalies, mental retardation) and Denys-Drash syndromes. The risk for developing a Wilms tumor is much higher in patients with Denys-Drash and WAGR syndromes than in most children with overgrowth syndromes (30–95% vs 5%, respectively) [15].

Previous studies have shown that the median age of patients presenting with a unilateral Wilms tumor is higher than that in that patients presenting with bilateral Wilms tumors or with nephroblastomatosis [16]. Patients with predisposition syndromes generally undergo careful surveillance imaging (i.e., typically sonography every 3 months), which results in earlier detection of precursor lesions. In our study, earlier age at presentation for rests (rests vs Wilms tumors: median age, 1.1 vs 3.1 years) can only be partially explained by the use of screening sonography in patients with predisposition syndromes, as a trend toward significant difference in age between patients with Wilms tumors and patients with rests was seen even when restricting the analysis to nonsyndromic cases (p = 0.028).

Our study has limitations. First, it was not feasible to compare the imaging findings and histology results of every lesion in patients with multifocal lesions because not all lesions were surgically sampled. The number of lesions individually excised in a kidney with multiple lesions was at the discretion of the treating institutional surgeon. According to our review of the institutional operative reports, surgeons typically excised one or two lesions per kidney and tended to excise the lesions that had the most concerning findings on preoperative imaging. In patients with multiple lesions, we attempted to identify the index lesion by carefully comparing the institutional operative notes with the images. Second, in spite of selecting from a large cohort of patients, the total numbers of cases with rests or small Wilms tumors are small because of our strict inclusion criteria. Another limitation of this study is the variability in technique, such as the variability in MRI acquisitions and lack of DWI in all cases, between participating institutions. However, nephrogenic rests can show diffusion restriction, and Platzer et al. [17] have described overlapping apparent diffusion coefficients between nephrogenic rests and Wilms tumors. Although we have not investigated the role of ¹⁸F-FDG PET, avid uptake of FDG by normal renal parenchyma limits the role of PET in the evaluation of renal lesions [18]. Finally, we did not evaluate interobserver variability in assessment of imaging findings because we used consensus interpretation of two radiology readers in our data collection and analysis.

In summary, our study shows that in children younger than 5 years old and for lesions smaller than 5 cm, a diagnosis of a Wilms tumor should be favored over a nephrogenic rest when a renal mass is spherical, exophytic, and larger than 1.75 cm in maximal diameter. Homogeneity is associated more often with perilobar nephrogenic rests, whereas intralobar rests and Wilms tumors are more often inhomogeneous. Imaging is not effective in identifying a pseudocapsule, which is a key pathologic discriminator between a nephrogenic rest versus a Wilms tumor.