Response of bilateral Wilms tumor to chemotherapy suggests histologic subtype and guides treatment

Abstract

Background

Patients with bilateral Wilms tumor initially receive neoadjuvant chemotherapy to shrink the tumors and increase the likelihood of successful nephron-sparing surgery. Biopsy of poorly responding tumors is often done to better understand therapy resistance. The purpose of this retrospective, single-institution study was to determine whether initial chemotherapy response is associated with tumor histology, potentially obviating the need for biopsy or change in chemotherapy.

Methods

Patients with synchronous bilateral Wilms tumors who underwent surgery at St Jude Children’s Research Hospital from January 2000 to March 2022 were considered for this study. A mixed-effects logistic regression model was used to evaluate the likelihood of the tumor being stromal predominant, as predicted by tumor response to neoadjuvant chemotherapy.

Results

A total of 68 patients were eligible for this study. Tumors that increased in size had an odds ratio of 19.5 (95% confidence interval [CI] = 2.46 to 155.03) for being stromal predominant vs any other histologic subtype. Age at diagnosis was youngest in patients with stromal-predominant tumors, with a mean age of 18.8 (14.1) months compared with all other histologic subtypes (χ² = 7.05, P = .07). The predictive value of a tumor growing combined with patient aged younger than 18 months for confirming stromal-predominant histology was 85.7% (95% CI = 57.18% to 93.5%).

Conclusions

Tumors that increased in size during neoadjuvant chemotherapy were most frequently stromal-predominant bilateral Wilms tumor, especially in younger patients. Therefore, nephron-sparing surgery, rather than biopsy, or extension or intensification of neoadjuvant chemotherapy, should be considered for bilateral Wilms tumors that increase in volume during neoadjuvant chemotherapy, particularly in patients aged younger than 18 months.

Wilms tumor comprises more than 90% of childhood renal tumors (1) and annually affects approximately 1 in 10 000 children worldwide (1-3). Approximately 4%-8% of patients with Wilms tumors present with synchronous bilateral Wilms tumors (ie, tumors arising in both kidneys at the same time) (4-6). Historically, patients with favorable histology bilateral Wilms tumors had an overall 4-year event-free survival of only 61% (7) and a 20-year overall incidence of end-stage renal disease of 12%-15% (4,8). However, the recent Children’s Oncology Group (COG) bilateral Wilms tumor trial, AREN0534, showed statistically significant improvement for these patients with a 4-year event-free survival of 82.1% (9).

Treatment for unilateral Wilms tumors usually includes radical nephrectomy. However, patients with bilateral Wilms tumors present the additional challenge of preserving renal function and therefore, generally, do not follow this treatment algorithm. Instead, neoadjuvant chemotherapy is administered prior to definitive surgical intervention to increase the feasibility of successful nephron-sparing surgery, which should be performed within 12 weeks of initiation of chemotherapy. This is followed by adjuvant postoperative chemotherapy and, in certain circumstances, radiation therapy (10).

Wilms tumor is generally very chemosensitive; thus, the initial response is typically substantial tumor volume reduction (11,12). Tumors that are not responsive to neoadjuvant chemotherapy were previously thought to likely have unfavorable (anaplastic) histology (13), but biopsy confirmation was recommended before chemotherapy was intensified, even though Wilms tumor biopsies are not very accurate in documenting the presence of unfavorable histology (14). However, some published data suggest that nonresponsiveness may be a consequence of having well-differentiated tumors (15).

The aim of this observational study was to examine the relationship between the volumetric response of synchronous bilateral Wilms tumor to neoadjuvant chemotherapy and the histologic subtype of the treated tumors to assist in guiding subsequent therapy. Our hypothesis was tumors that do not shrink in response to neoadjuvant chemotherapy, as defined by tumor volume reduction on cross-sectional imaging, should not be treated with extended or intensified chemotherapy, with or without biopsy, as the underlying histology is most likely stromal predominant. Instead, patients with chemotherapy-resistant bilateral Wilms tumors should undergo surgical resection, preferably bilateral nephron-sparing surgery, when feasible.

Methods

Patients

After receiving institutional review board approval, a retrospective review was conducted of all patients with synchronous bilateral Wilms tumors during the period January 2000 to March 2022, whose treatment at St Jude Children’s Research Hospital included definitive surgical resection of the primary tumor(s). The data collected included patient demographics, type and duration of preoperative chemotherapy, tumor measurements obtained before and after neoadjuvant chemotherapy, surgical procedure(s) performed, tumor histology, associated genetic predisposition to Wilms tumors, time to relapse, survival, and duration of follow-up.

The initial query of institutional electronic medical records returned 97 patients with bilateral Wilms tumors from which exclusions are detailed in Supplementary Figure 1 (available online). Of these, 8 were not synchronous and, therefore, were excluded. Of the 89 cases of synchronous bilateral Wilms tumors, 21 were excluded because of a lack of complete information from the referring institutions. Two patients underwent an upfront unilateral radical nephrectomy at another institution before any induction chemotherapy was administered, resulting in the exclusion of those kidneys from the analysis, although the treatment of the tumors in the remaining, contralateral kidneys was included.

Data collection

For this observational study, all tumor measurements were extracted from radiology reports if 3-dimensional measurements were described. If 1 or more dimensional measurements were missing, volumes were calculated by measurements obtained directly from images. Prechemotherapy measurements were obtained from cross-sectional images (magnetic resonance imaging or computed tomography) obtained closest to the start of treatment. Postchemotherapy data were obtained from imaging studies performed 1-4 days prior to surgery. All pathology information was derived from the final surgical pathology report. Measurements and pathologic features of only the largest tumor were extracted if 2 or more lesions were removed from the same side.

For the purposes of analyses, tumors were grouped on the basis of the International Society of Pediatric Oncology (SIOP) classification of risk (low, intermediate, or high) (16), except the stromal-predominant Wilms tumor, which was considered an independent group. This resulted in 4 distinct posttherapy tumor groups. Each tumor was categorized individually, on the basis of the predominant histologic component (>67% of the residual tumor displayed the histological feature) and the percentage of residual viable tumor of the largest lesion removed, if multiple tumors were removed from a single kidney.

Tumor volumes and response classification

Baseline tumor volume (Vol_BL) and posttreatment tumor volume (VOL_PT) were measured on images captured before and after neoadjuvant chemotherapy. Tumor volume was calculated as the maximum dimension in 3 orthogonal axes (cephalo-caudal, anterior-posterior, and lateral-medial) multiplied together. These imaging measures were then used to calculate the relative volume change of each tumor as [(Vol_PT−Vol_BL)/Vol_BL] * 100. A positive value indicated tumor size increase during chemotherapy, and a negative value indicated tumor shrinkage. We used these values to classify chemotherapy responses into 3 categories: good response (shrinkage >50%), poor response (shrinkage 0% to <50%), and progression (increase in size).

Statistical analyses

Distributions of all volume measures were visually inspected, and potential influence points or outliers were double-checked for validity. It is important to note that because of the bilateral nature of this study, some patients exhibited 2 separate tumor types, thereby introducing a potential violation of the independence assumption necessary for traditional hypothesis testing. In such cases, to achieve statistical feasibility, we assumed the more aggressive tumor type to be designated primary tumor classification at the patient level. For instance, a single patient may present with both stromal-predominant and SIOP low risk, in which case stromal predominant becomes the primary tumor type for the patient. Thereby, demographics and clinical characteristics were expressed only as frequencies and percentages or mean and standard deviation with no hypothesis testing to compare between group differences. Hypothesis testing was used only to compare age at diagnosis, in which case the more aggressive tumor type in discordant bilateral cases was assumed as the patient’s primary tumor type.

We used mixed-effects logistic regression to analyze the utility of tumor response to predict the odds of a tumor being stromal predominant vs any other subtype. Three separate mixed-effects logistic regression models were run with the tumor-response predictor, as either dichotomized (model 1: tumor decrease vs increase) or trichotomized (model 2: good response vs poor response; model 3: good response vs poor response vs progression). We corrected for the random effects of repeated measurements within an individual (ie, 2 kidneys per patient) by adding a patient identifier as a mixed-effects variable. All models were adjusted for age at surgery and Vol_BL. Marginal R² was used to assess model fit and to select the best-fitting models. Additionally, log-rank tests and Kaplan–Meier curves were done to compare survival between tumor groups. No missing data were reported for any of the study variables. R statistical software (v4.1.2) was used to run all analyses. Two-sided P values less than .05 were considered statistically significant.

Results

Patient characteristics

A total of 68 patients (134 kidneys) had all essential information and fit all inclusion criteria for the current study. Of these, 78 (58%) kidneys had multifocal involvement; for these, the average was 2.9 foci (range = 2-10 foci) per kidney. Patient demographics are summarized in Table 1.

Table 1.

Demographics and histology of 68 pediatric patients with bilateral Wilms tumors^a

Characteristic	No. of patients (%) (N = 68)
Mean age at diagnosis (SD), mo	27.3 (19.9)
Sex
Female	40 (58.8)
Male	28 (41.2)
Race
Asian	5 (7.4)
Black	7 (10.3)
Othera	5 (7.4)
White	51 (75.0)
Ethnicity
Hispanic	8 (11.8)
Non-Hispanic	60 (88.2)
Associated syndromes
Beckwith–Wiedemann	6 (8.8)
Denys–Drash	3 (4.4)
WAGR	1 (1.5)
Presenting with metastatic disease
Lung	10 (14.7)
Lung and liver	2 (2.9)
Histology	No. of tumors (%)
SIOP LR group (n = 47)
Cystic degeneration	7 (14.9)
Necrotic	11 (23.4)
Nephrogenic rests	26 (55.3)
Otherb	3 (6.4)
SIOP IR group (n = 36)
Triphasic	10 (27.8)
Biphasic	7 (19.4)
Epithelial	14 (38.9)
Focal anaplasia	4 (11.1)
Otherc	1 (2.8)
SIOP HR group (n = 25)
Blastemal predominant	17 (68.0)
Diffuse anaplasia	8 (32.0)
Stromal-predominant group (n = 26)
Stromal predominant	26 (100)
Histologic pairing when discordant between kidneys	No. of patients (%)
Stromal/SIOP LR	8 (29.6)
Stromal/SIOP IR	2 (7.4)
Stromal/SIOP HR	1 (3.7)
SIOP LR/SIOP IR	7 (25.9)
SIOP LR/SIOP HR	5 (18.5)
SIOP HR/SIOP IR	4 (14.8)

a

Other races included Pacific Islander, American Indian, and no response. HR = high risk; IR = intermediate risk; LR = low risk; SIOP = International Society of Pediatric Oncology.

b

Data categorized as Other represent pathology described as favorable with treatment effect.

c

Data categorized as Other represent pathology described as intermediate risk, with less than 33% viable tumor remaining and organized fibrinous tissue.

Neoadjuvant chemotherapy varied among patients

The 68 patients received neoadjuvant chemotherapy, either vincristine, dactinomycin and doxorubicin (n = 65, 95.6%), or vincristine and dactinomycin alone (n = 3, 4.4%) for an average of 11.1 weeks (range = 6-27 weeks) before surgery. Seven (10.2%) patients were treated for greater than 12 weeks prior to nephron-sparing surgery.

Surgical procedures and margin status

Among the patients, 55 (81%) underwent bilateral nephron-sparing surgery. The remaining 13 underwent unilateral radical nephrectomy and contralateral nephron-sparing surgery, including the 2 patients who underwent upfront unilateral nephrectomy before initial treatment and referral. Five patients underwent staged surgeries in which only 1 kidney was operated on during each procedure, again including the 2 patients who underwent upfront unilateral nephrectomy before referral.

Of the 134 tumors, 39 (29.1%) had viable tumor at the resection margin (local stage III), of which 27 received radiation. For tumors of stromal-predominant histology, 11 of 26 (42.3%) tumors had positive margins, a rate higher than for tumors of other histologic subtypes (28 of 108; 25.9%).

Volumetric responses to neoadjuvant chemotherapy may predict tumor histology

Tumor sizes were estimated before initiating neoadjuvant chemotherapy (Vol_BL) and again within 4 days prior to surgery (Vol_PT), using magnetic resonance imaging or computed tomography images to obtain measurements in 3 orthogonal axes. The relative change in tumor volume in response to neoadjuvant chemotherapy was then calculated (Figure 1, A). Tumor histologies overall and circumstances where there were discordant histologies between sides are detailed in Table 1.

Figure 1.

A) Waterfall plot demonstrating the change in each tumor’s volume after neoadjuvant chemotherapy. Each tumor is represented by a bar shaded on the basis of SIOP risk classification. B) Individual tumor volume trajectories in response to neoadjuvant chemotherapy were grouped into SIOP histologic subtypes. SIOP = International Society of Pediatric Oncology; Tx = treatment; vol = volume.

When grouped by SIOP classification, and excluding stromal-predominant tumors, all SIOP histology groups (low, intermediate, and high risk) showed a net decrease in tumor volume, with mean relative volume change ranging from −50% to −62.6%. In contrast, stromal-predominant tumors had a net increase with mean relative volume change of +50% (223%) after neoadjuvant chemotherapy. Of the 26 stromal-predominant tumors, 13 (50%) grew during neoadjuvant chemotherapy. Among all other groups, only 7 of 108 (6.4%) tumors grew: 2 (6.9%) in the SIOP high-risk category and 5 (10.6%) in the SIOP low-risk category. Thus, of the 20 tumors that grew during neoadjuvant chemotherapy, 13 (65%) were stromal predominant, 5 (25%) were in the SIOP low-risk category, and 2 (10%) were in the SIOP high-risk category (Figure 1, A). The SIOP high-risk tumors that grew were of diffuse anaplasia histology. Among the 5 SIOP low-risk tumors that grew, 2 were consistent with cystic degeneration, and 3 were composed of nephrogenic rests. None of the tumors that grew during neoadjuvant chemotherapy underwent a biopsy prior to definitive surgery.

Table 2 shows the relationship between patient age and primary tumor histology. Patients with stromal-predominant bilateral Wilms tumors had a younger mean and median age at diagnosis (18.8 [14.1] months and 13.0 months; min = 4, max = 49 months, respectively) than any other group (P = .07). The SIOP high-risk group had the oldest mean and median age at diagnosis (35.2 [20.2] months and 31.5 months; min = 11, max = 72 months, respectively).

Table 2.

Evaluation of association between age at diagnosis and primary tumor histology

Tumor subtype	No. of patients	Mean age at diagnosis (SD)a	Median age at diagnosis (min, max)a	Pb
SIOP LR	23	31.7 (20.2)	28.0 (0, 64.0)	.07
SIOP IR	16	26.1 (22.9)	18.0 (6.00, 90.0)
SIOP HR	10	35.2 (20.2)	31.5 (11.0, 72.0)
Stromal predominant	19	18.8 (14.1)	13.0 (4.00, 49.0)

a

All age data are given in months. HR = high risk; IR = intermediate risk; LR = low risk; max = maximum age; min = minimum age; SIOP = International Society of Pediatric Oncology.

b

Nonparametric Kruskal–Wallis test.

Table 3 shows the tumor volumes and a summary of volumetric responses after chemotherapy. The mean tumor volumes of stromal-predominant tumors before and after chemotherapy were 734 (712) mL and 813 (898) mL, respectively, with a mean relative volume change of 50% (223%) posttreatment. Thus, the stromal-predominant group was not only the only group to show a net increase in mean tumor volume after chemotherapy (Figure 1, B) but also had the greatest average tumor size at diagnosis. Subjectively, the stromal-predominant tumors also appeared more homogeneous and well circumscribed on preoperative imaging studies. A representative case of stromal predominant tumor growth can be seen in Figure 2.

Figure 2.

Computed tomography scan images of a patient with bilateral stromal-predominant Wilms tumor. Images shown were obtained in the transverse (A, B) and coronal (C, D) planes, before (A, C) and after (B, D) 6 weeks of neoadjuvant chemotherapy, which was administered per the DD4A regimen (24). The right tumor increased from 1370 mL to 2731 mL, a 99.3% increase, during treatment, and the left tumor increased from 177 mL to 659 mL, a 272.3% increase. This patient subsequently underwent a bilateral nephron-sparing surgery.

Table 3.

Difference in chemotherapy response on the basis of tumor histology

	SIOP low risk (n = 47)	SIOP intermediate risk (n = 36)	SIOP high risk (n = 25)	Stromal (n = 26)
VolBL, mL,a mean (SD)	204 (332)	555 (515)	428 (452)	734 (712)
VolPT, mL,b mean (SD)	71.9 (184)	136 (246)	264 (645)	813 (898)
Relative volume change (%),c mean (SD)	−62.6 (64.2)	−76.5 (28.1)	−21.2 (200)	50.0 (223)
Tumor response after treatment, No. (%)
Good response, >50% shrinkage	38 (80.9%)	31 (86.1%)	16 (64.0%)	9 (34.6%)
Poor response, <50% shrinkage	4 (8.5%)	5 (13.9%)	7 (28.0%)	4 (15.4%)
Progression, growth	5 (10.6%)	0 (0%)	2 (8.0%)	13 (50.0%)

a

Baseline tumor volume. SIOP = International Society of Pediatric Oncology.

b

Posttreatment tumor volume.

c

[(Vol_PT−Vol_BL)/Vol_BL] * 100.

We used 3 separate models to evaluate the utility of tumor response to predict the odds of stromal-predominant Wilms tumors vs any other type of SIOP histology risk classifications. The results of these models are presented in Table 4. Furthermore, we evaluated the positive predictive value of tumor size increase alone for the presence of stromal predominant Wilms tumor, which was 65% (95% confidence interval [CI] = 40.78% to 84.61%). The positive predictive value of age at diagnosis alone for the presence of stromal predominant Wilms tumor was 34.0% (95% CI = 22.1% to 47.4%). When tumor size increase and age at diagnosis are considered in combination, the positive predictive value increased to 85.7% (95% CI = 57.2% to 93.5%).

Table 4.

Multivariate mixed effects logistic regression of tumor response predicting the odds of being a stromal-predominant Wilms tumor vs any other histologic subtype

Model	Comparison(s)	OR (95% CI)d	P
1	Tumor decreased vs	Referent
	Tumor increased	28.32 (2.26 to 355.33)	.01
2	Good responsea vs	Referent
	Poor responseb	7.41 (1.41 to 38.96)	.018
3	Good responsea vs	Referent
	Poor responseb	1.64 (0.27 to 9.85)	.589
	Progressionc	19.54 (2.46 to 155.03)	.005

a

Good response was defined as more than 50% shrinkage in tumor volume after neoadjuvant chemotherapy. CI = confidence interval; OR = odds ratio; referent = reference value used to calculate the odds ratio.

b

Poor response was defined as less than 50% shrinkage in tumor volume after neoadjuvant chemotherapy.

c

Progression was defined as any tumor growth during neoadjuvant chemotherapy.

d

All models adjusted for age (at surgery) and baseline tumor volume (VolBL).

Volumetric responses to chemotherapy are shown in Table 5 for the 11 patients who had discordant histology, where 1 side was stromal predominant. Of these patients, 8 (72.7%) had a stromal-predominant tumor and a contralateral SIOP low-risk tumor. Only 1 patient had a contralateral SIOP high-risk tumor. This patient also had extensive lung and liver metastases and a significant delay in therapy initiation and died of disease. In each case, the nonstromal-predominant tumor shrunk with neoadjuvant chemotherapy as did 6 of the 11 (54.5%) stromal predominant tumors.

Table 5.

Volumetric responses of patients with discordant histology in which the largest tumor in 1 kidney was stromal-predominant^a

Case	Histologic subtype 1	% change	mL change	Contralateral histologic subtype 2	% change	mL change
1	Stromal predominant	64.1	1008.5	SIOP low risk	−91.7	−64.7
2	Stromal predominant	−78.5	−2070.1	SIOP low risk	−89.3	−41.6
3	Stromal predominant	−63.5	−379	SIOP low risk	−97.4	−6.3
4	Stromal predominant	−72	−664	SIOP low risk	−79	−3.8
5	Stromal predominant	65	313	SIOP low risk	−46.7	−1.2
6	Stromal predominant	60.4	318	SIOP low risk	−75.8	−23.7
7	Stromal predominant	17.2	356	SIOP low risk	−76.6	−14.3
8	Stromal predominant	168.8	412	SIOP low risk	−53.7	−5.1
9	Stromal predominant	−90	−197	SIOP intermediate risk	−65.9	−424
10	Stromal predominant	−78.5	−17	SIOP intermediate risk	−9.6	−136
11	Stromal predominant	−47.5	−114	SIOP high risk	−61.7	−170

^aSIOP = International Society of Pediatric Oncology.

Survival of patients with bilateral Wilms tumor

The survival rate of our patient cohort was 89.7%, with a mean follow-up of 5.0 years (range = 0-16.3 years). A total of 27 patients were followed for at least 5 years; 7 died, and 13 were lost to follow-up. All known deaths occurred within 4 years of the initial surgical procedure at St Jude Children’s Research Hospital. Of the 68 patients, 13 (19.1%) experienced a relapse, with an average time to relapse of 15.2 months. All relapses occurred within 3 years (range = 3-29 months) of completing treatment. Three patients who experienced relapse died of disease.

Of the remaining 4 patients, 3 had metastatic disease at presentation, and 1 had a pathogenic germline BRCA2 variant and died of secondary medulloblastoma. Of the 7 deceased patients, 5 (71.4%) had SIOP high-risk bilateral Wilms tumors. One patient with Beckwith–Wiedemann syndrome had SIOP low-risk bilateral Wilms tumor that either recurred or in whom a new Wilms tumor developed. The remaining deceased patient had stromal-predominant bilateral Wilms tumor that was metastatic at presentation. On the basis of primary tumor histology, survival was 100% for patients with SIOP intermediate-risk bilateral Wilms tumors (range = 0-173 months), 93% for those with SIOP low-risk bilateral Wilms tumors (range = 0-173 months), 80% for patients with SIOP high-risk bilateral Wilms tumors (range = 0-195 months), and 94% for patients with stromal-predominant bilateral Wilms tumors (range = 0-160 months). Log-rank test was statistically nonsignificant (P = .09), demonstrating no differences in overall survival between tumor types. Kaplan–Meier survival curves are shown in Supplementary Figure 2 (available online).

Discussion

The goal of this study was to elucidate the relationship between the volumetric response of bilateral Wilms tumors to neoadjuvant chemotherapy and the pathologic subtype. The ability to predict the histologic subtype based on imaging-assessed volumetric response to chemotherapy, particularly when there is an increase in tumor volume, could help guide appropriate treatment decision making. The prediction, specifically, would inform the decision of whether to perform a tumor biopsy, give additional or alternative neoadjuvant chemotherapy, or attempt surgical resection. The concern about a tumor that increases in size during neoadjuvant chemotherapy is that it might be anaplastic and, therefore, high-risk disease. However, here we have shown that stromal-predominant bilateral Wilms tumors not only presented with the highest initial volume but also was the histology most likely to grow during chemotherapy, and yet was associated with excellent survival. Overall survival for patients with stromal-predominant bilateral Wilms tumors in our series was 93.7%, which is consistent with the excellent survival for these patients in the COG AREN0534 study (17).

Some previous studies have shown that poor response to chemotherapy, as assessed by radiologic imaging, may not indicate a poor overall outcome for bilateral Wilms tumors (13,18,19). Chemotherapy resistance of some Wilms tumors has been attributed to the stromal-predominant subtype, including rhabdomyoblastic elements (19,20). A single-institution review of 11 patients with rhabdomyoblastic bilateral Wilms tumors with poor radiologic response, treated between 1985 and 1995, revealed an overall survival of 63.6% (21). A time period–comparable overall survival for bilateral Wilms tumors treated before 2000 was reported at 67% (12). Similarly, a poor response to preoperative chemotherapy was noted in stromal- and epithelial-predominant Wilms tumors that did not correlate with a negative outcome (19). In a study of 36 unilateral stromal-predominant cases of Wilms tumors, 22 showed little change in volume in response to chemotherapy, and none shrunk by more than 50% (22). Additionally, a study of 132 unilateral stromal-predominant Wilms tumors found a median volume reduction of 33.4%. This result differed substantially from all other intermediate-risk Wilms tumors, which showed a median volume reduction of 67.0% (23).

Most reports concerned with the relation between stromal-predominant subtype and volumetric response to neoadjuvant chemotherapy over the past 20 years have involved unilateral Wilms tumors. Therefore, this study increases the understanding of and will help guide treatment for patients with synchronous bilateral Wilms tumors that are chemotherapy resistant but should generally not be managed by radical nephrectomy. Our results show tumors that continue to increase in volume during chemotherapy are most often stromal-predominant Wilms tumors. However, these tumors have favorable outcomes; recurrence and survival are comparable with those of SIOP low- or intermediate-risk classifications. There is no evidence that stromal-predominant tumors warrant intensified or prolonged neoadjuvant chemotherapy to reduce their volume before surgical resection. In fact, our study revealed the opposite—such tumors will most likely continue to grow if chemotherapy is extended or intensified. Therefore, transitioning toward surgical removal, even after only 2 cycles of neoadjuvant chemotherapy, would benefit patients with bilateral Wilms tumors that grow during initial chemotherapy, as further tumor growth may cause the nephron-sparing surgery to be more difficult. Although tumors of stromal-predominant histology were generally the largest among the various histologic subtypes, nephron-sparing surgery was feasible given their well-circumscribed growth pattern, making enucleation feasible. Finally, a positive margin consisting of stromal-predominant cells may not benefit from or require adjuvant radiation therapy. This hypothesis is being tested in an ongoing clinical trial at St Jude Children’s Research Hospital, SJWT21 (ClinicalTrials.gov ID NCT04968990). If the likelihood of the tumor being stromal-predominant histology is high, enucleation is acceptable and most likely feasible. Although concern for performing an enucleation, and the increased possibility of positive margins, for a tumor of anaplastic histology exists, in our series, only 2 of 25 (8.0%) tumors of high-risk histology grew during the administration of neoadjuvant chemotherapy, and only 2 of 20 (10%) tumors that grew were of high-risk histology. This is consistent with the COG AREN0534 experience in which there were 25 cases where the tumor histology was anaplastic (8 focal and 17 diffuse). In only 3 (12%) cases was tumor progression observed during neoadjuvant chemotherapy, and in each case, there was rapid progression of disease (9).

An exciting, emerging technology that may be useful in determining renal tumor histology in the future without performing an invasive biopsy that samples only a limited area of a tumor is the use of liquid biopsies to evaluate circulating tumor DNA (ctDNA). An early study by Madanat-Harjuoja et al. (24) found that ctDNA could be detected in the serum of 41 of 50 (82%) patients with Wilms tumors, and agreement between serum ctDNA and tumor sequencing results was more than 70% for prognostic copy number changes. This methodology may be able to detect anaplastic histology Wilms tumor on the basis of the presence of mutant p53 in ctDNA (25), thereby further helping inform the decision to perform early nephron-sparing surgery or intensify therapy to poorly responding tumors.

Some limitations of this study that should be considered are the relatively small sample size and retrospective approach. Additionally, it is uncertain whether the results can be extrapolated to patients with unilateral Wilms tumor who receive neoadjuvant chemotherapy or patients treated with bilateral Wilms tumors on a SIOP protocol, both of which generally start with 2-drug chemotherapy (vincristine and actinomycin-D, omitting doxorubicin).

In conclusion, our experience suggests that patients treated for synchronous bilateral Wilms tumors that progress during neoadjuvant chemotherapy are most likely of stromal-predominant histology, especially if the patient is aged younger than 18 months at diagnosis and the tumor is large and appears well circumscribed. Therefore, such patients should undergo definitive surgery, preferably nephron-sparing surgery, after 6 weeks of neoadjuvant chemotherapy rather than undergoing biopsy or additional or intensified chemotherapy, and they can expect to have a good oncologic outcome.

Data availability

All original data are available by written request through the St Jude Department of Surgery at the time of publication.

Author contributions

Colton Duncan, BA (Data curation; Formal analysis; Investigation; Methodology; Writing—original draft; Writing—review & editing), Suraj Sarvode Mothi, MPH (Formal analysis; Software; Writing—original draft; Writing—review & editing), Teresa Santiago, MD (Methodology; Writing—review & editing), Jordan Coggins, MS (Data curation; Writing—review & editing), Dylan Graetz, MD, MPH (Writing—review & editing), Michael Bishop, MD, MS (Writing—review & editing), Elizabeth Mullen, MD (Writing—review & editing), Andrew Murphy, MD (Writing—review & editing), Daniel Green, MD (Writing—review & editing), Matthew Krasin, MD (Writing—review & editing), and Andrew Davidoff, MD (Conceptualization; Investigation; Methodology; Project administration; Supervision; Writing—original draft; Writing—review & editing).

Funding

The work was supported by the Cancer Center Support Grant (P30 CA21765 to CR) from the National Cancer Institute.

The funder did not play a role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.

Conflicts of interests

The authors declare no potential conflict of interest.

Acknowledgements

We thank Angela McArthur, PhD, Director, Scientific Editing, St Jude Children’s Research Hospital, for her thorough review of the manuscript.