Systematic Review: Prognostic Molecular Biomarkers in Wilms Tumors

Abstract

PURPOSE

Molecular biomarkers are increasingly used for risk stratification, particularly in up-front surgery settings (Children's Oncology Group trials), whereas in preoperative chemotherapy setting, the ongoing International Society of Pediatric Oncology (SIOP)-Renal Tumor Study Group-2016 UMBRELLA study aims to validate selected biomarkers for future risk-adapted treatment strategies. This systematic review summarizes all literature on the prognostic value of these biomarkers.

MATERIALS AND METHODS

A systematic literature review (PubMed and Embase; up to January 2025) included studies with ≥50 de novo Wilms tumors (WTs). Eligible biomarkers included copy number variations; 1q gain, 1p and/or 16q loss of heterozygosity (LOH)/loss, 12 gain, 14q loss, 22 loss, 11p15 LOH/loss of imprinting (LOI), and structural somatic variants (TP53 [and/or 17p loss], MYCN, FBXW7, WT1, WTX, SIX1/SIX2, DROSHA, DGCR8, AMER1, CTNNB1, GPC3, MLLT1, DICER1, DIS3L2). Outcome included relapse-free survival, event-free survival (EFS), and overall survival (OS). Risk of bias was assessed with quality in prognosis studies tool.

RESULTS

Low-bias multivariable/stratified analyses identified 1q gain as worse EFS and 1p and/or 16q LOH/loss as worse EFS/OS prognostic factors, in up-front nephrectomy settings. Preoperative chemotherapy settings revealed similar trends with lacking significance. TP53 and MYCN were adverse prognostic in univariate analyses. No prognostic data were available for the remaining variants.

CONCLUSION

1q gain and 1p and/or 16q LOH/loss emerge as independent prognostic biomarkers in up-front nephrectomy settings. Evidence remains limited in preoperative chemotherapy settings, particularly when using SIOP-oriented treatment algorithms. Prognostic value of TP53, MYCN, and 11p15 LOH/LOI warrants further validation in both settings. This highlights the need for adequately powered prospective studies, specifically in the preoperative chemotherapy setting, to establish reliable molecular biomarkers.

INTRODUCTION

Treatment of Wilms tumor (WT) currently exceeds overall survival (OS) rates of 90%.^1-3 Worldwide, two main treatment strategies are used: up-front nephrectomy (used in approximately 80% of children treated on Children's Oncology Group [COG] clinical trials) and preoperative chemotherapy (with delayed nephrectomy) used in approximately 20% of children treated on COG-WT clinical trials and as a dominant approach by the International Society of Pediatric Oncology Renal Tumor Study Group (SIOP-RTSG), such as in the current UMBRELLA SIOP-RTSG 2016 protocol (further referred to as UMBRELLA).^1,4,5 Preoperative chemotherapy has been shown to downstage the tumor, to reduce rupture risk, and to identify a high-risk subgroup, that is, blastemal-type WT.^4,6,7 Conversely, up-front nephrectomy allows prechemotherapy surgical staging, up-front identification of anaplasia, and early assessment of biological biomarker status in chemotherapy-naïve patients.⁷

Postoperative histology and treatment response-based stratification has evolved in both COG and SIOP trials,^1,2,4 shaped by the identification of independent determinants of adverse outcome such as age, stage, histological subtype, and to a certain extent, tumor volume.^4,5 Still, currently 15%-20% of patients with WT relapse. In SIOP, most of these relapses occur in the intermediate-risk histology group.^8,9 COG clinical trials, have now implemented prechemotherapy tumor biomarkers in risk stratification, which requires tissue acquisition (nephrectomy or biopsy) before chemotherapy initiation.^1,2,10,11 In SIOP, the prognostic value of postchemotherapy tumor biomarkers (on the basis of tissue acquired after neoadjuvant chemotherapy) in patients with WT treated with preoperative chemotherapy is currently being explored as a primary aim of the UMBRELLA study.^1,4,5 To our knowledge, to date, a comprehensive overview of identified relevant and robust independent prognostic molecular aberrations for WT is lacking.

Herein, this systematic review analyzes available WT studies that investigated the prognostic value of relevant molecular biomarkers, with the ultimate goal to guide biomarkers to be further implemented in future clinical trials.

MATERIALS AND METHODS

This study was conducted as a systematic review according to established definitions,¹² with study identification, selection, data extraction, and reporting performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement guidelines.¹³ Risk of bias was assessed with QUIPS (Quality in prognosis studies) tool.¹⁴ The detailed methodology is depicted in the Data Supplement (Materials S2 and S3).

RESULTS

The database searches identified 1,140 studies, of which, 1,008 were removed before screening because of duplication or failure to meet basic eligibility criteria. Title and abstract screening of the remaining 132 records resulted in a total of 28 cohort studies that met the inclusion criteria for full-text screening (Data Supplement, Fig S1). Of those, 13 of 28 biomarker studies included multivariable analysis (MVA)/stratified analyses (9 in the up-front nephrectomy setting, one in the preoperative chemotherapy setting, and three with a mixed cohort; Tables 1-3).

TABLE 1.

The Prognostic Value of CNV in WT (MVA) Biomarkers in Studies That Included Patients With Uup-front Nephrectomy

Author Group	n	Analysis Method CNV		Histologyd Stage	%	Variable	Multivariable Analysis/Stratified Analysis (95% CI)			Risk of Bias (QUIPS-Tool)a
							EFS		OS	1	2	3	4	5	6
Bown15 UK	195	Karyotyping 1q21-25 22		Favorable/unfavorable WT Stage I-V	18	1q gain	RR, 3.4 (1.5 to 8), P = .005		NA
					11	22 loss	RR, 5.9 (2.2 to 16.1), P < .001		NA
						Stage	Stage IV RR, 5.0 (2.2 to 11), P < .001		NA
Grundy16 NWTS-5	1,727	PCR—Microsatellite		All histologies Stage I-IV	11.3	LOH 1p all stages	F H	4 years RR, 1.25 (0.76 to 2.07), P = .39	4 years RR, 1.25 (0.54 to 2.93), P = .60
		1p36.13b 1p36.21 1p36.22 1p36.23 1p36.31 1p36.32 1p36.33	16q13 16q21 16q22.1 16q22.2 16q23.1 16q23.2 16q24.1 16q24.2 16q24.3			LOH 1p stage I and II		4 years 80%, RR, 2.19 (1.15 to 4.17), P = .02	4 years 91%, RR, 4.03 (1.20 to 12.43), P = .02
						LOH 1p stage III and IV		4 years 89%, RR, 0.69 (0.30 to 1.57), P = .37	4 years 97%, RR, 0.52 (0.12 to 2.14), P = .36
					17.4	LOH 16q All stages		4 years RR, 1.28 (0.78 to 1.86), P = 20	4 years RR, 1.00 (0.51 to 1.98), P = 1.00
						LOH 16q stage I and II		4 years 82%, RR, 1.91 (1.14 to 3.21), P = .01	4 years 98%, RR, 1.40 (0.40 to 4.95), P = .60
						LOH 16q stage III and IV		4 years 85%, RR, 0.89 (0.51 to 1.54), P = .67	4 years 92%, RR, 0.88 (0.39 to 1.97), P = .76
					4.6	LOH 1p and 16q All stages		4 years RR, 2.59 (1.62 to 4.15), P = .0001	4 years RR, 3.11 (1.52 to 6.37), P = .002
						LOH 1p and 16q Stage I and II		4 years 75%, RR, 2.8 (1.51 to 5.49), P = .001	4 years 90.5%, RR, 4.25 (1.37 to 13.19), P = .01
						LOH 1p and 16q Stage III and IV		4 years 65.9%, RR, 2.4 (1.20 to 4.82), P = .01	4 years 77.5%, RR, 2.66 (1.04 to 6.82), P = .04
Perotti17 AEIOP	77	WG SNP array 1q21.1-q44		FHWT and anaplastic Stage I-IV	16	1q gain	HR, 6.11 (1.70 to 21.94), P = .006		NA
						Age >24 versus 24 months	HR, 2.661 (0.557 to 12.70), P = .220		NA
						Stage III and IV versus I and II	HR, 1.094 (0.33 to 3.542), P = .881		NA
						DA	HR, 0.686 (0.077 to 6.106), P = .735		NA
Gratias18 NWTS4	212	MLPA 1q21-q25		FHWT Stage I-IV	27	1q gain	RR, 2.72, P = .0089		RR, 3.07, P = .067
Spreafico19 AIEOP TW 2003	125	PCR—microsatelliteb		FHWT Stage I-IV	19	LOH 1p	HR, 4.6 (1.7 to 12.2), P = .0011		NA
Fawzy20 Egypt	100	PCR—microsatelliteb		FHWT Stage I-IV	1p 14	LOH 1p and/or 16q. Stage I	5 years 83.3%, P = .56		5 years 66.7%, P = .005
					16q 13	LOH 1p and/or 16q. Stage II	5 years 85%, P = .13		5 years 71%, P = .26
					1p/16q 12	LOH 1p and/or 16q. Stage III	5 years 88%, P = .15		5 years 52%, P = .001
						LOH 1p and/or 16q. Stage IV	5 years 40%, P = .59		5 years 40%, P = .28
Gratias21 NWTS5	1,114	MLPA 1q21-q25		FHWT	28	1q gain	RR, 2.4, P < .001		NA
					9	1p loss	ns (details NA)		ns (details NA)
					13	16q loss	ns (details NA)		ns (details NA)
						Stage I	8 years 85%, P = .0052		8 years 90%, P = .0015
						Stage II	8 years 81%, P = .077		8 years 94%, P = .19
						Stage III	8 years 79%, P = .01		8 years 91%, P = .33
						Stage IV	8 years 64%. P = .0004		8 years 74%, P = .010
Haruta22 Japan	128	SNP array for LOH aCGH MLPAc		FHWT and anaplastic Stage I-IV	28	1q gain	HR, 5.3, P = .007		HR, 5.6, P = .25
					9	1p loss	Not evaluable because of collinearity		NA
					7	16q loss	HR, 5.21, P = .007		HR, 5.66, P = .025
					26	12 gain	HR, 0.23, P = .050		HR, 0.19, P = .112
Park23 K-PHOG	101	PCR—microsatelliteb		FHWT Stage I-IV	13	LOH 1p	HR, 0.83 (0.17 to 4.01), P = .817		NA
					13	LOH 16q	HR, 3.95 (1.08 to 14.39), P = .037		NA
						Age <2, 2-4, >4 years	2-4: HR, 0.84, P = .8/>4: HR, 2.4, P = .1		NA
						Stage III and IV versus I and II	HR, 1.11 (0.24 to 5.15), P = .893		NA

NOTE. Colors indicating the risk of bias according to above categories: : low/: moderate/: high. Results with statistical significance (P ≤ .05) are highlighted in bold.

Abbreviations: aCGH, microarray-based comparative genomic hybridization; AEIOP, Italian Association of Pediatric Hematology and Oncology; CNV, copy number variation; EFS, event-free survival; FH, favorable histology; HR, hazard ratio; IR, intermediate risk; Loc., localized; LOH, loss of heterozygosity; LR, low risk; Microsat., microsatellites; MLPA, multiplex ligation-dependent probe amplification; MVA, multivariable analysis; NA, not available; NWTS, National Wilms Study Tumor; OS, overall survival; PCR, polymerase chain reaction; QUIPS, quality in prognosis studies; RR, relative risk; RTSG, Renal Tumor Study Group; SIOP, International Society of Pediatric Oncology; SNP, single-nucleotide polymorphism; UKCCSG, United Kingdom Childhood Cancer Study Group; WG, whole genome; WT, Wilms tumors.

^aRisk bias (QUIPS-tool): 1-study participation, 2-study attrition, 3-prognostic factor measurement, 4-outcome measurement, 5-study confounding, 6-statistical analysis and reporting.

^b PCR microsatellite as described by Grundy et al.¹⁶

^c MLPA as described by Chagtai et al.²⁴

^d According to risk group definition of indicated Study groups.

TABLE 3.

The Prognostic Value of CNV in WT (MVA) Biomarkers in Studies That Included Patients Treated With Up-Front Nephrectomy and PreOperative Chemotherapy

Author Group	n	Analysis Method CNV	Histologya Stage	%	Variable	Multivariable Analysis/Stratified Analysis (95% CI)		Risk of Bias (QUIPS-Tool)b
						EFS	OS	1	2	3	4	5	6
Messahel25 UKCCSG UKW 1-3	456 Direct 269 Deferred 157	PCR—microsatellitec	FHWT and anaplastic Stage I-IV	10.3	LOH 1p	HR, 1.44 (0.56 to 3.72), P = .447	HR, 1.18 (0.3 to 4), P = .8
				14.6	LOH 16q	HR, 2.69 (1.47 to 4.92), P= .001	HR, 2.67 (1.17 to 6.06), P = .02
				2.6	LOH 1p and 16q	HR, 3.86 (1.35 to 11), P = .011	HR, 0.92 (0.12 to 6.8), P = .94
					Age: 0-2/2-4/>4 years	2-4 HR, 1.56 (0.77 to 3.19), P = .22 >4 HR, 1.28 (0.61 to 2.68), P = .51	NA
					Stage I-II versus III-IV	HR, 1.35 (0.78 to 2.31), P = .28	HR, 9.13 (3.12 to 26.76), P = .00
Segers26 UKCCSG SIOP-2001	331 Direct 243 Deferred 88	Karyotyping All length	FHWT and anaplastic LR IR High- risk Stage I-IV	19	1q gain	HR, 2.45 (5 1.17 to 5.15), P = .02	HR, 4.28 (1.59 to 11.53), P = .004
				7	1p loss	HR, 0.37 (0.1 to 1.37), P = .14	HR, 0.14 (0.05 to 1.1), P = .07
				12	16q loss	HR, 1.83 (0.8 to 4.16), P = .15	HR, 1.55 (0.56 to 4.33), P = .4
				27	12 gain	HR, 0.53 (0.22 to 1.24), P = .14	HR, 0.44 (0.14 to 1.38), P = .16
				4.9	14q loss	HR, 4.0 (1.05 to 15.32), P = 0.04	HR, 3.12 (0.77 to 12.73), P = .11
				5	22 loss	HR, 2.56 (0.95 to 6.87), P = .06	HR, 2.64 (0.79 to 8.86), P = .12
					Age	P = .47	P = .76
					Stage III-IV	Stage III HR, 2.2 (1.01 to 5.2), P = .05 Stage IV HR, 3.63 (1.62 to 8.13), P = .002	Stage III HR, 4.55 (1.42 to 14.6), P =.01 Stage IV HR, 8.43 (2.63 to 27), P< .001
					Anaplasia	HR, 2.55 (0.89 to 7.3), P = .08	HR, 6.13 (1.75 to 21.47), P = .005
Evageliou27 COG AREN03B2 AREN0532 Subsequent report of Fernendez et al28	668 Direct 500 Deferred 135	PCR—microsatellitec	FHWT Stage III	9	LOH 1p	4 years 77.5%, HR, 2.18, P = .0066	P = .24
				20	LOH 16q	4 years 83.5%, HR, 1.72, P = .042	P = .87
					LN status +	4 years 83.5%, HR, 2.78, P = .00017	4 years 95%, HR, 2.50, P = .054
					LN +/LOH 1p	4 years 73%, HR, 2.64, P = .016	4 years P = .24
					LN +/LOH 16q	4 years 79.3%, P = .09	4 years P = .34
					LN +/LOH 1p or 16q	4 years 76.9%, HR, 2.67, P = .0016	4 years P = .11
					LN– & LOH+ versus LN– & LOH–	4 years HR, 3.04 (1.29 to 7.16) P < .0001	ns data NA
					LN+ & LOH– versus LN– & LOH–	4 years HR, 3.57 (1.66 to 7.67)	ns data NA
					LN+ & LOH+ versus LN– & LOH–	4 years HR, 6.33 (2.73 to 14.66)	NA

NOTE. Colors indicating the risk of bias according to above categories: : low/: moderate/: high. Results with statistical significance (P ≤ .05) are highlighted in bold.

Abbreviations: CNV, copy number variation; COG, Children Oncology Group; EFS, event-free survival; FH, favorable histology; HR, hazard ratio; IR, intermediate risk; LN, lymph node; Loc., localized; LOH, loss of heterozygosity; LR, low risk; Microsat., microsatellites; MVA, multivariable analysis; NA, not available; OS, overall survival; PCR, polymerase chain reaction; QUIPS, quality in prognosis studies; SIOP, International Society of Pediatric Oncology; UKCCSG, United Kingdom Childhood Cancer Study Group; WT, Wilms tumors.

^a According to risk group definition of indicated Study groups.

^b Risk bias (QUIPS-tool): 1-study participation, 2-study attrition, 3-prognostic factor measurement, 4-outcome measurement, 5-study confounding, 6-statistical analysis and reporting.

^c PCR microsatellite as described by Grundy et al.¹⁶

In the studies that included MVA/stratified analysis, the risk of bias in study participation and for prognostic factor assessment was low in all studies (13/13); study attrition bias was high in 9/13 (70%) and moderate in 4/13 (30%) of studies; outcome measurement bias was low in 10/13 (77%) and moderate in 3/13 (23%), confounding was low in 11/13 (85%) and moderate in 2/13 (15%), and statistical analyses and reporting bias was low in 8/13 (62%) and moderate in 5/13(38%); (Data Supplement, Fig S2). Risk of bias assessment of studies that only performed UVAs is shown in the Data Supplement (Tables S1-S3).

The methodology for biomarker classification in the various studies is schematically illustrated in Figure 1.

FIG 1.

Summary of methodological approaches that had been used in the published manuscripts. aCGH, array comparative genomic hybridization; amp, amplification; cnLOH, copy-neutral LOH; CNV, copy number variations; F, fluorescence; LOH, loss of heterozygosity; LOI, loss of imprinting; MLPA, multiplex ligation-dependent probe amplification; RT-PCR, reverse transcription polymerase chain reaction; SNP Array, single nucleotide polymorphism array; SNV, single nucleotide variant; STR, short tandem repeat; WES, whole exome sequencing; WGS, whole genome sequencing. ^*Depending on SNP resolution, ^#H19 (imprint control region 1) and KvDMR1 (imprint control region 2) on chromosome 11p15. Created in BioRender. Van den Heuvel-Eibrink, M. (2025) https://BioRender.com/dxp3t3j.

CNVs as Prognostic Factors From Studies That Were Able to Pursue Multivariable and/or Stratified Retrospective and/or Ad Hoc Analyses

Dividing by up-front treatment approach, studies that allowed MVA and/or stratified analyses to confirm the independent predictive value of biomarkers were analyzed (Tables 1-3).^15-27,29 Results of the risk-adapted prospective AREN0532 and AREN0533 trials, which include both up-front nephrectomy and preoperative chemotherapy cohorts, and which incorporated prospective biomarker-based therapeutic intensification, are presented separately. Furthermore, note that biomarkers in National Wilms Study Tumor (NWTS) and COG publications were derived from prechemotherapy tumor tissues regardless of timing of nephrectomy (biopsy was required up-front if nephrectomy was not done before chemotherapy), whereas biomarkers in SIOP publications were generally obtained from postchemotherapy tumor tissues.

1q Gain

In the up-front nephrectomy setting, five (multiplex ligation-dependent probe amplification [MLPA], karyotyping, whole-genome sequencing, single-nucleotide polymorphism [SNP] array) studies assessed the prognostic value of 1q gain, in 11%-28% of favorable histology WT (FHWT; Table 1).^{15,17,18,21,22} All five found a significant adverse prognostic value of 1q gain for event-free survival (EFS). In one study, 1q gain was associated with significant inferior EFS in stages I, III, and IV (the negative impact in stage II did not reach statistical significance).²¹ Three (MLPA) studies (n = 212, n = 1,114, n = 128)^18,21,22 analyzed 1q gain as an independent prognostic factor for OS, including the largest series with >1,000 children. Although the two smaller studies did not find a statistically significant association of 1q gain with OS,^18,22 the largest (MLPA) study described poor OS in stage I and IV WT patients with 1q gain.²¹ In the preoperative chemotherapy setting, one (MLPA) study (n = 586) reported 1q gain in low-, intermediate-, and high-risk WT²⁴ to be associated with lower EFS but not OS rates, but this study did not reach sufficient power (Table 2). One (karyotyping) study (n = 331),²⁶ which analyzed the prognostic value of 1q gain in a combined up-front nephrectomy (n = 243) and preoperative chemotherapy-treated (n = 88) patients cohort, found associations with adverse EFS and OS (Table 3).

TABLE 2.

The Prognostic Value of CNV in WT (MVA) Biomarkers in Studies That Included Patients Treated With PreOperative Chemotherapy

Author Group	n	Analysis Method CNV			Histology	%	Variable	Multivariable Analysis/Stratified Analysis (95% CI)		Risk of Bias (QUIPS-Tool)a
								EFS	OS	1	2	3	4	5	6
Chagtai24 SIOP-RTSG	586	1q21.2 1q23.3 1q32.1 1q32.2 1q43	MLPA 1p12 1p22.1 1p36.12 1p36.22 1p36.23 1p36.31 1p36.33	16q12.1 16q13 16q21 16q22.1 16q23.3 16q24.3	LR IR High-risk Stage I-IV	28	1q gain	HR, 1.98 (1.27 to 3.07), P = .002	HR, 1.61 (0.83 to 3.15), P = .16
						8	1p loss	HR, 0.98 (0.5 to 1.91), P = .95	HR, 0.67 (0.24 to 1.89), P = .45
						28	16q loss	HR, 1.14 (0.68 to 1.91), P = .63	HR, 1.37 (0.67 to 2.83), P = .39
							Age	HR, 1.01 (1 to 1.01), P = .06	HR, 1 (0.99 to 1.01), P = .48
							Sex	HR, 0.99 (0.65 to 1.51), P = .98	HR, 0.93 (0.49 to 1.74), P = .81
							Stage IV versus I	HR, 4.58 (2.58 to 8.15), P < .001	HR, 21.6 (6.93 to 67.66), P < .001

NOTE. Colors indicating the risk of bias according to above categories: : low/: moderate/: high. Results with statistical significance (P ≤ .05) are highlighted in bold.

Abbreviations: CNV, copy number variation; EFS, event-free survival; HR, hazard ratio, IR, intermediate risk; LOH, loss of heterozygosity; LR, low risk; MLPA, multiplex ligation-dependent probe amplification; MVA, multivariable analysis; OS, overall survival; PCR, polymerase chain reaction; QUIPS, quality in prognosis studies; RTSG, Renal Tumor Study Group; SIOP, International Society of Pediatric Oncology.

^aRisk bias (QUIPS-tool): 1-study participation, 2-study attrition, 3-prognostic factor measurement, 4-outcome measurement, 5-study confounding, 6-statistical analysis and reporting.

1p LOH/Loss

1p loss of heterozygosity (LOH), found in 10%-14% of WT samples, was evaluated for prognostic value in five up-front nephrectomy studies (Table 1).^{16,19,22,23,27} Of these, three (polymerase chain reaction [PCR] microsatellite-based) FHWT studies (n = 1,727, n = 125, n = 655) reported a significant adverse association with EFS,^16,19,27 but in the largest study (n = 1,727),¹⁶ only in stage I-II disease. The prognostic value of 1p LOH, in two studies, demonstrated, only the largest study (n = 1,727), a prognostic significance for OS, again in patients with stage I-II.¹⁶ Two (PCR microsatellite and aCHG-based) cohort studies (n = 101, n = 128) did not find a significant association between 1p LOH/loss and EFS and OS.^22,23 In the preoperative chemotherapy setting, the only available (MLPA-based) study (n = 586; all histology subtype WTs) was not able to identify 1p loss as an independent prognostic factor for EFS or OS²⁴ (Table 2). Three (PCR microsatellite and karyotyping-based) studies included patients with both treatment approaches. One cohort identified 1p LOH/loss as a prognostic factor only for EFS.²⁷ In the other two, there was no association with adverse EFS or OS (Table 3).^25,26

16q LOH/Loss

16q LOH, found in 7%-18% of WT samples, was investigated for prognostic value in three (2 PCR microsatellite and one aCGH-based) studies in patients with de novo FHWT (n = 1,727, n = 101, n = 128) undergoing up-front nephrectomy.^16,22,23 All studies identified 16q LOH as independent adverse prognosticator for EFS, but one of them (n = 1,727)¹⁶ limited to patients with stage I-II disease. In the aCGH study (n = 128),²² 16q loss was associated with lower OS, in contrast to the 2 PCR microsatellite studies (Table 1).^16,23 In the preoperative chemotherapy setting, one (MLPA-based) study (n = 586)²⁴ did not find an independent prognostic value of 16q loss for EFS or OS (Table 2). The three aforementioned studies included patients with both treatment approaches,^25-27 identified 16q LOH/loss as an adverse prognostic factor in 2 PCR-based studies only, for EFS in both^25,27 and OS in one²⁵ (Table 3).

1p and/or 16q LOH/Loss

The combination of the prognostic relevance of these biomarkers has been investigated through various approaches. Combined 1p and 16q LOH was evaluated in one up-front nephrectomy setting (PCR microsatellite-based) study (n = 1,727).¹⁶ The frequency of combined 1p and 16q LOH was 4.6% in FHWT and identified as an independent adverse prognostic factor for both EFS and OS across all disease stages. In one (PCR microsatellite-based) study (n = 452), including both up-front nephrectomy and preoperative approaches treated patients with WT (n = 426 FHWT, n = 26 anaplastic WT), combined 1p and 16q LOH was identified as an independent prognostic factor for EFS, but not for OS²⁵ (Table 3). There are currently no studies available that explored the prognostic value of combined 1p and 16q LOH specifically in the preoperative chemotherapy setting alone.

1p and/or 16q LOH (whether occurring individually or in combination) was evaluated in a (PCR microsatellite-based) study of patients with de novo FHWT treated with up-front nephrectomy (n = 100) with analyses stratified by stage.²⁰ In this study, no significant association with EFS, but an association with OS was found, limited to stages I and IV (Table 1). Singular 1p or 16q LOH, in patients with or without lymph node involvement ( LN+, LN–, respectively) was evaluated for prognostic relevance in two subsequent (PCR microsatellite based) studies in patients with de novo stage III FHWT (n = 535, n = 635),^27,28 including approximately 80% up-front nephrectomy and 20% delayed nephrectomy cases, wherein isolated LOH at either 1p or 16q was associated with inferior EFS, but not with OS in both studies. Patients with 1p or 16q LOH and LN+ had a markedly worse EFS, but not a lower OS (Table 3 and Data Supplement, Table S3). Currently, no studies are available that explored the prognostic value of 1p and/or 16q LOH in the preoperative chemotherapy setting.

Other CNVs

Other CNVs that were evaluated for prognostic value by MVA included gain of chromosome 12, 14q, and loss of chromosome 22. In the up-front nephrectomy setting, one (MLPA-based) study (n = 128), showed that gain of chromosome 12 was present in 26% with a favorable independent prognostic impact on EFS, but not on OS in FHWT.²² Chromosome 12 was also analyzed in a preoperative chemotherapy (karyotyping-based) study (n = 331; frequency: 27%), but no significant association with EFS or OS was found.²⁶ Loss of 14q, present in 3% of patients with low- and intermediate-risk WT, was only assessed in a (karyotyping-based) preoperative chemotherapy setting study (n = 331)²⁶ revealed an adverse prognostic impact on both EFS/OS. Loss of chromosome 22, found in 5% of cases in a merged UK WT cohort treated with up-front nephrectomy (n = 195)¹⁵ and preoperative chemotherapy (n = 331)²⁶ revealed inferior EFS, but not OS (karyotyping-based study; Tables 1 and 2). No studies included 11p15 LOH/loss of imprinting (LOI) in their MVA and/or stratified analyses. A single (methylation RT-PCR-based) study using UVA, analyzed the prognostic value of 11p15 LOH/LOI in the very low-risk WT (VLRWT) (11% of the cases; n = 108) in the up-front nephrectomy setting.³⁰ These patients with VLRWT tumor (age <2-year, stage I, FH, nephrectomy weight <550 g, absence of a WT predisposition syndrome) are usually treated with surgery only in COG trials. Harboring 11p15 LOH/LOI was, in the UVA, associated with 20% (LOH)-25% (LOI) relapse rate, significantly higher than observed in other tumors without 11p15 LOH or LOI (3.3%; P = .011). This, however, did not affect OS, given the excellent response to 3-drug salvage therapy³⁰ (Data Supplement, Table S1).

Prognostic Relevance of Somatic Mutations and Gene Alterations

The search included studies on the prognostic value of molecular relevant mutations in TP53 (and/or 17p loss), MYCN, FBXW7, WT1, WTX, SIX1/SIX2, DROSHA, DGCR8, AMER1, CTNNB1, GPC3, MLLT1, DICER1, and DIS3L2. Only studies reporting on the prognostic value of TP53 (and/or 17p loss), MYCN, FBXW7, WT1, WTX, DROSHA, DGCR8, and SIX1/SIX2 were found. The prognostic value has been assessed through available studies of which the design allowed only UVA (Data Supplement, Tables S1-S3).

TP53 and/or Segmental Copy Number loss (17p Loss)

The prognostic value of TP53 mutations and/or segmental copy number loss (17p loss) in the up-front nephrectomy setting was available from two studies (n = 128, n = 118).^22,31 TP53 alterations were present in 6/128 (4.6%) patients in one (MLPA and sequencing based)study, of which 3/6 had clear DA histology.²² In the second (MLPA and sequencing based) study,³¹ including only patients with DA-WT, TP53 mutations occurred in 48% (56/118) and 17p loss in 11% (12/118) and reported an association between the presence of TP53 alterations in stage III/IV and poor EFS.³¹ Both studies suggest TP53 alterations as an independent prognostic factor for OS, but sample size was limited, allowing only UVA. TP53 mutations and/or 17p loss were evaluated in two (MLPA-based) studies in the preoperative chemotherapy setting (n = 586, n = 55), both of which included tumors with all histologic WT subtypes.^24,32 One study reported TP53 and/or 17p loss in 44/586 (7.5%) of WT cases overall (23/44 (57%) were DA-WT and 21/44 FHWT).²⁴ The second study, including only stage IV tumors, describes TP53 alterations in 9/55 (16.4%) patients with WT, of which 7/9 (77%) were high-risk histology (DA and/or Blastemal type).³² Both studies suggest TP53 alterations to be prognostic factors for adverse EFS and OS in UVA (Data Supplement, Tables S1 and S2). None of the studies analyzed the independent prognostic value of TP53 alterations in non-DA specifically.

MYCN Gain and/or Mutations

MYCN gain and mutations were studied as potential prognostic factor in one (sequencing-based) study in the up-front nephrectomy setting (n = 128) and detected in, respectively, 7% and 0.7% of the patients.²² No aberration revealed an association with adverse EFS or OS. In the preoperative chemotherapy setting, three (MLPA and WES based) studies showed MYCN gain (n = 586, n = 55, n = 293),^24,32,33 to occur in 10%-12.7% of all patients with WT, and in 18% of patients with stage IV. In two/three cohorts, MYCN aberrations were found to be associated with adverse EFS and OS.^24,33 MYCN mutations, only analyzed in two (sequencing-based; n = 293, n = 810)^33,34 studies, were found in 3% and to be associated, in UVA, with adverse EFS in both studies, and with adverse OS in one³³ (Data Supplement, Table S2).

Other Gene Aberrations

The prognostic value of FBXW7 mutations or deletions (4q loss) had not been studied in the up-front nephrectomy setting. In the preoperative chemotherapy setting, FBXW7 deletion (MLPA-based) study occurred in 4% (n = 586), associated with adverse EFS and OS²⁴ in UVA. The only up-front nephrectomy (MLPA and sequencing based) study including WTX alterations (deletion or mutation) identified this aberration in 26% of all WTs (n = 128) and showed no association with EFS or OS.²² In the preoperative chemotherapy setting, two (MLPA and sequencing based) studies (n = 586, n = 219) evaluated the prognostic value of WTX aberrations, and no association with relapse or survival was found.^24,35 WT1 alterations (mutations and small insertions/deletions) were examined in two up-front nephrectomy setting (MLPA and sequencing based) studies (n = 108, n = 128), identifying WT1 mutations in 12.4% of VLR-WT and 24% of FH/anaplastic WT patients, and they did not find an association with EFS or OS.^22,30 One (MLPA-based) study in the preoperative chemotherapy setting (n = 586) studied WT1 deletion(11p loss), reported a frequency of 8% of WTs, and did not find prognostic significance (EFS nor OS).²⁴ Mutations in SIX1/SIX2, DGCR8, and DROSHA were analyzed altogether in one study in the up-front nephrectomy setting (n = 534), with a frequency of 1.8%, the co-occurrence of these four mutations was associated with an increased risk of relapse and worse OS³⁶ (Data Supplement, Tables S1-S3).

Outcome Results of Currently Available Studies That Have Implemented Biomarker-Based Stratification for Treatment Intensification

Two COG reports are currently available from studies that included treatment intensification on the basis of biomarker implementation in their clinical trials.

The AREN0532 and AREN0533 studies reported results of intensified treatment chemotherapy for FHWT patients with stage I-IV harboring combined (by PCR microsatellite assessed) 1p and 16q LOH (n = 131/2511)³⁷ (Table 4). Augmented therapy improved EFS for patients with FHWT and combined 1p and 16q LOH compared with the historical NWTS-5 study, for all WT stages. In a previous report focusing on patients with stage IV-only WT (n = 292), approximately 6% had combined 1p and 16q LOH.³⁸ Treatment intensification was applied on the basis of lung nodule response as well as the presence of combined 1p and 16q LOH after 6 weeks of vincristine, dactinomycin, and doxorubicin. Therapeutic intensification for those stage IV patients with combined 1p and 16q LOH from a 3-drug regimen to a 5-drug regimen, regardless of lung response, demonstrated a significantly improved EFS compared with NWTS-5 treatment (3-drug therapy only), suggesting that therapeutic intensification can overcome the prognostic impact of these biomarkers (Table 5). The AREN0532 and AREN0533 studies included overall 21%,²⁷ with approximately 35%³⁹ and 68%⁴⁰ of patients, treated with preoperative chemotherapy and delayed nephrectomy in patients with localized stage III FHWT (n = 668), stage IV-pulmonary only metastases (n = 102/288), and stage IV-extrapulmonar metastases (n = 98/145), respectively. However, no separate analyses were performed focusing solely on those treated with preoperative chemotherapy. No implementation studies have been conducted in the preoperative chemotherapy setting alone.

TABLE 4.

Studies That Used Biomarkers for Stratifying

Stage I-IV
Author Group	n	Stage	Biomarker (PCR microsatellitea)	%	Treatment		4-Year EFS (95% CI)			4-Year OS (95% CI)			Risk of Bias (QUIPS-Tool)c
					Standard Therapy	Augmented Therapy	AREN0532 AREN0533 (%)	NWTS-5 (%)	P	AREN0532 AREN0533 (%)	NWTS-5 (%)	P	1	2	3	4	5	6
Dix37 AREN0532 AREN0533	1,147	I/II	LOH 1p and 16q	2.7	EE4A	DD4A/No RT	87.3 (75.1 to 99.5)	68.8 (55.2 to 82.3)	.042	100	91 (83.6 to 99.6)	.096
	1,364	III/IV	LOH 1p and 16q	3.7	DD4A	Regimen M/RTb	90.2 (81.7 to 98.6)	61.3 (44.9 to 77.6)	.001	96.1 (90.5 to 100)	86 (74.5 to 97.5)	.087

NOTE. Colors indicating the risk of bias according to above categories: : Low/: Moderate/: High. Results with statistical significance (P ≤ .05) are highlighted in bold.

Abbreviations: CR, complete response; EFS, event-free survival; IR, incomplete response; LOH, loss of heterozygosity; NWTS, National Wilms Study Tumor; OS, overall survival; PCR, polymerase chain reaction; QUIPS, quality in prognosis studies; regimen DD4A, vincristine, dactinomycin, and doxorubicin; regimen EE4A, vincristine, dactinomycin; regimen M, vincristine, dactinomycin, and doxorubicin alternating with cyclophosphamide and etoposide.

^a PCR microsatellite as described by Grundy et al.¹⁶

^b Lungs/flank/whole abdomen per study guidelines.

^c Risk bias (QUIPS-tool): 1-study participation, 2-study attrition, 3-prognostic factor measurement, 4-outcome measurement, 5-study confounding, 6-statistical analysis and reporting.

TABLE 5.

Studies That Used Biomarkers for Stratifying

Stage IV LOH 1p/16q
Author Group	n	Stage	Standard Therapy 6 weeks	Lung Nodule Response	n	Biomarker (PCR microsatellitea)	Augmented Therapy	4-Year EFS	4-Year OS	Risk of Bias (QUIPS-Tool)b
								AREN0533 (%)	AREN0533 (%)	1	2	3	4	5	6
Dix38 AREN0533	292	IV	DD4A	CR	132	LOH 1p	DD4A/No RT	75	NA
						LOH 16q		66.7	NA
						LOH neither locus		83.4	94.4
						LOH 1p or 16q		70.3	100
				IR	141	LOH 1p	Regimen M/RT	90	NA
						LOH 16q		75	NA
						LOH neither locus		92.4	95.7
						LOH 1p or 16q		79	94.7
				Independent of response	20	LOH 1p and 16q	Regimen M/RT	95	100

NOTE. Colors indicating the risk of bias according to above categories: : low/: moderate/: high.

Abbreviations: CR, complete response; EFS, event-free survival; IR, incomplete response; LOH, loss of heterozygosity; NA, not available; NWTS, National Wilms Study Tumor; OS, overall survival; PCR, polymerase chain reaction; QUIPS, quality in prognosis studies; regimen DD4A, vincristine, dactinomycin, and doxorubicin; regimen EE4A, vincristine, dactinomycin; regimen M, vincristine, dactinomycin, and doxorubicin alternating with cyclophosphamide and etoposide.

^a PCR microsatellite as described by Grundy et al.¹⁶

^b Risk bias (QUIPS-tool): 1-study participation, 2-study attrition, 3-prognostic factor measurement, 4-outcome measurement, 5-study confounding, 6-statistical analysis and reporting.

DISCUSSION

Over the past 2 decades, a substantial number of studies have investigated potential relevant prognostic molecular biomarkers to be used for stratification of children with WT. This review shows significant heterogeneity in study design, patient populations, and the methodologic molecular testing approaches.

This considerable heterogeneity of methodologies used to assess molecular biomarkers is challenging for summarizing data across studies. The wide array of techniques that have been employed included conventional cytogenetics, PCR-based microsatellite assays, array-based CGH, MLPA, SNP arrays, and whole-genome and whole-exome sequencing (Tables 1-3 and Data Supplement, Tables S1-S3). These methods may differ in resolution, sensitivity, and ability to distinguish between types of genomic alterations (Fig 1). For instance, CN loss and LOH are distinct events, as LOH may occur with or without a net loss of genetic material; however, it is suggested that they can be correlated at a high frequency.⁴¹ Gratias et al²¹ reported that 96% of CN calls for both 1p and 16q were concordant with LOH calls, while also identifying subsets with copy-neutral LOH (2% for 1p and 3% for 16q) or CN loss without detectable LOH (1.7% for 1p and 1.3% for 16q). These observations underscore that CN loss and LOH are highly correlated but may also be complementary events, with observed discrepancies likely reflecting methodologic refinements in LOH detection. Structural chromosomal rearrangements can generate complex patterns, further complicating interpretation. Although these methodologic nuances are potentially important, the objective of this review was to describe the prognostic relevance of biomarkers in WT on the basis of the variety of methods used and to summarize the currently available data. Nonetheless, given various factors, a meta-analysis could not be performed. A limited sample size in some cohorts, which is unavoidable in the setting of a rare disease such as WT, in conjunction with a low event rate, could restrict the statistical power and potentially hinder the exploration of findings to broader populations. In addition, most results are derived from WT cohorts managed with the majority undergoing up-front nephrectomy wherein tumor tissue is derived from chemo-naïve tissues, which may not be directly translated into the stratification of WT patients treated with preoperative chemotherapy when tumor biomarkers are derived from chemotherapy treated tumor tissues.

Still, consistency in prognostic findings, despite variation in methods, reveals that chromosome 1q gain emerges as one of the most consistently reported prognostic biomarkers for relapse in WT. This is a biomarker with an incidence ranging 20%-30% in FHWT and even higher in stage IV.^21,24 The prognostic impact of 1q gain may be further highlighted by the findings that 70% of relapsed WT tumors have 1q gain, conceivably reflecting the positive selection of these cells.^42-44 Including this biomarker in de novo WT treatment stratification will hopefully reduce relapse and the inherent treatment burden associated with salvage therapy, as is being shown in the COG AREN2231 Trial (along with 1p LOH, 16q LOH, and 11p15 LOH/LOI; ClinicalTrials.gov identifier: NCT06401330) independent of whether up-front nephrectomy or preoperative chemotherapy is used consistent with the recent AREN0532 and AREN0533 studies, and in potential future SIOP trials employing preoperative chemotherapy. So far, only four studies included the prognostic value of this biomarker for OS in their multivariate analyses. Reasonably, EFS is the primary end point for risk stratification, with the main goal for WT therapy being to treat the patient the best way front-line, aiming to avoid very intensive relapse therapy.^11,45 So far, studies that included relevant histologic subtypes such as diffuse anaplasia or blastemal type (preoperative chemotherapy group only) are limited.

Furthermore, unlike EFS, OS is confounded by such salvage therapy, and we are already seeing evidence that intensification may overcome the prognostic impact of such biomarkers (Tables 4 and 5).^37,38 That said, despite strong evidence in favor of its prognostic relevance, additional validation in the setting of preoperative chemotherapy, especially when biomarkers are derived from chemotherapy-treated tumor tissue, is needed, particularly as this approach is being implemented in ongoing clinical trials. The ongoing UMBRELLA study, involving 161 centers from 32 countries on three continents, aims to address this prospectively.⁴ Pending these results, SIOP has not adopted any of these biomarkers into risk stratification guidelines, and outside of formal COG trials wherein tumor tissue is obtained before chemotherapy, prospective validation of the role of 1q gain and demonstrating direct benefit from therapeutic modulation for such patients is desired before widespread nonresearch clinical use and/or guideline-based care.

Regarding other biomarkers, 1p LOH present in 8%-11% and 16q LOH in 20% of WT,^{16,19,20,22-26,46,47} were first described in 1992 and extensively studied since.⁴⁸ In up-front nephrectomy series, 1p LOH has been linked to higher relapse risk, especially in localized disease,^16,29 while 16q LOH shows a stronger and independent association with reduced EFS and OS.¹⁶ Recently, the COG identified stage III patients with 1p LOH or 16q LOH plus LN+ as a subgroup with significantly worse survival outcomes^27,28 (Table 3). The combination of 1p and 16q LOH (5%-6%) seems to characterize a subset of patients with a stronger adverse prognostic signature in patients with FHWT. These biomarkers have successfully guided treatment intensification, primarily aimed at preventing relapses, a strategy that was first described in COG AREN0532/0533 studies by Dix et al. A critical finding in this work was that the adverse prognostic impact of such biomarkers can be overcome through modest therapy intensification, thereby reducing the burden of therapy in those who relapse, who otherwise would require very intensive salvage therapy^37,38 (Tables 4 and 5).

Evidence of prognostic value of additional CNVs, such as chromosome 12 gain, 14q loss, and chromosome 22 loss was explored,^15,22,26 but further validation in large well-characterized WT cohorts in needed to define their prognostic contribution. Among other chromosomal anomalies, 11p15 LOH/LOI³⁰ has shown potential prognostic relevance from UVAs, particularly because of its association with an increased risk of relapse in the VLRWT treated with surgery only on NWTS-5 and AREN0532 COG trials. Hence, this alteration has only been studied primarily in the context of up-front nephrectomy and is found in a relatively small subset of patients, limiting the overall impact of this biomarker so far.³⁰

Studies that have only been able to pursue UVAs suggest that mutations in specific genes also have a potential prognostic impact in a subset of WTs. TP53 is a tumor suppressor gene that codes for the cellular tumor antigen p53 and is located on the short arm of chromosome 17. Often referred to as the guardian of the genome, this multifunctional transcription factor binds to its target DNA sequence to induce cell cycle arrest, DNA repair, or apoptosis. Alterations in TP53, including mutations and/or 17p loss, have been traditionally described to be associated with DA,31,49 being present in 50%-77% of DAWT cases. Mascchieto et al49 investigated whether the association between TP53 alterations and DA conferred additional prognostic value. Data from 40 patients with DAWT, treated by both the COG and SIOP approach, were included. The study found that DA patients with wild-type TP53 had significantly better OS and EFS than their mutated counterparts. In recent years, some studies have attempted to replicate these findings, suggesting a significant association, independent of histologic subtype, between TP53 alterations, but have not yet been able to determine whether the association between TP53 alterations and DA can be regarded as an independent prognostic biomarker. Further validation in large, well-characterized WT cohorts is needed to clarify the contribution of TP53 alterations beyond histologic subtype. TP53 mutations appear to represent late, progressive events that can affect a subset of tumor cells and carry clinical relevance, particularly in advanced stages; more than half of the patients with stage III and IV harboring TP53 alterations experience relapse or death.^31,50

The MYCN protein, encoded by MYCN gene, located on chromosome 2p24.3 codes for a transcription factor of the MYC family, which is involved in the regulation of multiple cell functions, such as cell proliferation, differentiation, and apoptosis.³⁴ This gene has also been extensively studied in other pediatric tumors such as neuroblastoma, where its amplification confers a worse outcome.⁵¹ In the context of WT, MYCN alterations seem relevant molecular events, with potential implications on adverse outcome; that is, initial studies identified MYCN gain in approximately 10%-18.5% and MYCN mutations (P44L) in 3% of preoperative chemotherapy setting,^24,33,34,52 with a higher prevalence in high-risk histologic subtypes.^34,53,54 These findings were extended to a larger preoperative chemotherapy cohort assessing chemotherapy-treated tumor tissues (n = 810),³⁴ which identified MYCN mutations predominantly in blastemal-type tumors and absent in those with focal or DA; also, a correlation with local recurrence was suggested. Our review shows that MYCN gain and mutations may have prognostic value in WT and that the identification of MYCN alterations across multiple nonanaplastic histologic subtypes reinforces its potential as a biomarker for stratifying patients beyond current morphologic criteria.

FBXW7, an ubiquitin ligase component that acts as a tumor suppressor gene, is involved in the degradation of proto-oncogenes such as MYCN, and its disruption may contribute to WT progression through dysregulation of shared regulatory pathways.⁵³ This complex interplay, where both genes are linked, potentially leads to additive effects on tumor aggressiveness and progression. Alteration in FBXW7 has been suggested as an adverse prognostic factor for outcome in one small study²⁴; confirmation in larger cohorts is needed.

Mutations in SIX1/SIX2, DROSHA, and DGCR8 have been implicated in the dysregulation of developmental transcription and miRNA biogenesis, contributing to impaired mesenchymal-to-epithelial transition and tumor progression in WT. Although neither mutation alone was associated with adverse outcomes, their co-occurrence defines a distinct molecular subset of FHWT with increased relapse and mortality, particularly in association with blastemal histology.^36,55 Notably, recent analyses of paired primary and relapsed WT samples have shown that these co-occurring mutations are positively selected in the recurrent disease, supporting their role as potential oncogenic drivers.⁵⁶ Despite the limited number of cases studied, this recurrence-associated enrichment highlights their possible prognostic relevance and suggests that they may contribute to treatment resistance and disease progression, warranting further investigation in larger cohorts.

This review summarizes available studies on the prognostic significance of specific genomic alterations in WT and identified 1q gain and 1p and/or 16q LOH/loss as recurrent molecular biomarkers for relapse risk. Their relevance in the context of the preoperative chemotherapy setting, especially when tumor tissue is derived after chemotherapy, remains less clear. Emerging evidence also suggests a potential prognostic role for 11p15 LOH/LOI in very specific subsets, on the basis of up-front nephrectomy data, as well as TP53 and MYCN alterations, but current results are based on UVA only. Selective molecular-driven up-front therapy intensification studies for WTs for patients with certain adverse biomarkers, such as combined 1p and 16q LOH, have already demonstrated that higher relapse rates associated with such biomarkers can be overcome.

Taken together, these findings contribute to advancing risk-adapted therapeutic strategies and refining prognostic assessment in WT. Currently, stratification is mainly based on histology and stage, but in the past years, COG has already included biomarkers to stratify treatment. Obviously, histology and molecular biomarkers are associated (such as TP53 and DA). Further unraveling the underlying biologic mechanisms remains essential to fully understand the clinical relevance of specific genomic alterations. Importantly, there is a pressing need for prospective, adequately powered studies to validate and/or address these biomarkers across different treatment contexts, an objective currently being addressed by ongoing efforts such as the COG AREN2231 and UMBRELLA SIOP-RTSG 2016 protocols.

AUTHOR CONTRIBUTIONS

Conception and design: Agustina Oller, Patrick Kemmeren, Daniela Perotti, Harm van Tinteren, Andrew L. Hong, Manfred Gessler, James I. Geller, Marry M. van den Heuvel-Eibrink

Collection and assembly of data: All authors

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).