Clinical and pathologic features of 535 patients with congenital mesoblastic nephroma: a report from the Children’s Oncology Group

Abstract

Objective:

To explore clinicopathologic features of children with congenital mesoblastic nephroma (CMN) enrolled on Children‟s Oncology Group study AREN03B2 and a historical cohort of CMN patients.

Background:

CMN is a pediatric renal tumor of infancy, with histologic subtypes of cellular, mixed, and classic. Given its rarity, evidence-based clinical practice guidelines are unavailable. We collected clinicopathologic findings and outcomes data in 2 large cohorts of children with CMN.

Methods:

From 2004–2019, 6412 patients enrolled in AREN03B2 and underwent prospective central review of pathology materials, imaging studies, and operative reports. CMNs were identified and subclassified. Similar data was extracted from a historical cohort of CMNs collected by pathology reviewers between 1973–2001.

Results:

In total, 535 children were included (139 from AREN03B2; 396 from the historical cohort). In the ARE03B2 cohort, 137 had available follow-up data (median follow-up: 4.5 years). Ten children (7.2%) relapsed, and 4/10 children died of disease. Four of 55 (7.3%) children with local stage II cellular or mixed CMNs relapsed, and 6/37 (16.2%) children with local stage III cellular or mixed CMNs relapsed. No child with local stage I or classic CMNs (of any stage) relapsed. All relapses occurred within 1.5 years of diagnosis, and 4/10 relapses occurred within 3 months. In the historical cohort, 31 children (7.8%) relapsed; all relapses were local stage II or III cellular or mixed.

Conclusions:

While recurrences are uncommon, they are highly associated with cellular or mixed histologic subtypes and stage, providing key clinical information that may guide consideration of therapy and surveillance.

Introduction

Congenital mesoblastic nephroma (CMN) is a rare tumor constituting 3–5% of all pediatric renal neoplasms.^1–3 Most children with CMNs present in the first three months of life,^2–5 and over 90% are diagnosed by age six months.⁶ There are two histologically and genetically distinct patterns of CMN: 1) classic CMN, composed of uniform spindle cells arranged in fascicles with low cellularity and mitotic rate,⁷ and 2) cellular CMN, composed of densely packed, plump, blue spindle cells with a higher mitotic rate.^{8, 9} Most cellular CMNs are characterized by NTRK gene fusions, most commonly ETV6-NTRK3 as a result of t(12;15).^10–12 EML4-NTRK3, BRAF, and RET fusions have also been described.^13–17 Cases in which the classic and cellular patterns coexist are also recognized and are classified as mixed CMN.^{18, 19} Some mixed CMNs contain NTRK gene fusions, with a wide variation of NTRK fusion rates reported in the literature (0–83% of cases).^{10, 12, 20–23} EGFR alterations have been described in all CMN subtypes.^{24, 25} Given identical genetic alterations and histomorphology, similar ages at presentation, recurrence patterns, and outcomes, cellular CMNs have been postulated to represent intrarenal occurrences of infantile fibrosarcoma (IFS).^26–28

Overall, children with CMN have an excellent prognosis and are usually cured by surgical excision alone.⁵ Rarely, however, CMN can recur or metastasize.^27–30 Some of these children can be treated successfully for relapse, while others die of disease.³¹ Proposed risk factors for CMN relapse are cellular histology, incomplete surgical excision, and age at diagnosis over three months.^32–35 Prior studies have been limited by small sample sizes. To enhance understanding of this rare entity, pathologic and clinical features of children with CMN enrolled on the Children‟s Oncology Group (COG) AREN03B2 Renal Tumor Classification, Biology, and Banking study (AREN03B2) were explored and compiled with data extracted from a historical cohort of CMN patients presenting prior to AREN03B2.

Methods

AREN03B2 Patient Cohort

Children eligible for inclusion and enrolled on AREN03B2 with tumors classified as CMN on central pathology review from 2004 to 2019 were included in this study. Institutional Review Board approval was obtained before enrollment. Consent for participation in AREN03B2 was obtained from parents or legal guardians.

Required submissions for central pathology review included radiologic imaging, operative reports, and a full set of H&E slides. Histologic subtype and local stage were determined by central review pathologists (EJP, LNP, ALT, DAH, and MMC). CMNs were subclassified as cellular, classic, or mixed as follows: tumors with >80% cellular component were classified as cellular, those with 5–80% cellular pattern were classified as mixed, and all others were classified as classic. Additional data collected included age at diagnosis, sex, type of surgical procedure, tumor size, tumor weight, and pathologic staging parameters. The central reviewers documented the COG local stage through central pathologic and central surgical review.³⁶ All cases were reviewed prospectively without knowledge of clinical outcome.

Referring centers were not consistently required to have pulmonary imaging to be eligible for AREN03B2 as some CMNs are considered „benign‟, however, all patients were required to submit an abdominal CT. When available, chest CTs were centrally reviewed for evidence of pulmonary metastatic disease for patients on AREN03B2 (other imaging modalities were not documented).

During the study period, clinical reporting forms were required for five years after initial diagnosis, and information on treatment regimens (if given), site and timing of relapse, and patient deaths were extracted from these forms.

Historical CMN Cohort

Between 1973 and 2001 (prior to AREN03B2), pediatric renal tumors were often sent to the National Wilms Tumor Study (NWTS)/COG pathology central reviewers, either as consults from the referring pathologist or as part of central review for NWTS studies. The NWTS/COG central review pathologists collected and recorded the clinicopathologic characteristics of a large series of tumors from patients with CMNs during this time. These files were searched for a diagnosis of CMN. In this database, 396 cases of CMN were identified. Of the 396 identified CMN patients, 135 were registered in NWTS-5 but had no follow-up data. The historical database documented relapses when reported by the referring institution, along with relapse site, time to relapse, and any patient deaths.

Results

AREN03B2 Patient Cohort

Patients

Between 2004 and 2019, 6,412 total patients were registered on AREN03B2, and 139 had a confirmed central review diagnosis of CMN. Clinicopathologic features of these 139 AREN03B2 patients/tumors are summarized in Table 1.

Table 1:

Clinicopathologic features of CMNs on AREN03B2 and the Historical (NWTS or Consult) Cohorts

Characteristic	AREN03B2, N = 1391	NWTS or Consult, N = 3961	p-value2
Age (Months)			0.7
(Mean, Median)	(2.6, 1.1)	(4.2, 2.0)
(Range)	(0.0, 25.0)	(0.0, 48.0)
Unknown	0	10
Pattern			0.3
Cellular	83 (61%)	239 (60%)
Classic	30 (22%)	69 (17%)
Mixed	23 (17%)	88 (22%)
Unknown	3	0
Disease Stage			0.4
Stage I	15 (16%)	58 (24%)
Stage II	47 (51%)	112 (47%)
Stage III	29 (32%)	65 (27%)
Stage IV	1 (1.1%)	2 (0.8%)
Unknown	47	159
Tumor Diameter (cm)			0.3
(Mean, Median)	(7.9, 7.9)	(7.5, 7.0)
(Range)	(2.2, 15.0)	(0.6, 17.0)
Unknown	10	187
Tumor Weight (g)			0.3
(Mean, Median)	(256, 167)	(275, 203)
(Range)	(4, 1,160)	(19, 2,006)
Unknown	27	127
Recurrence	10 (7.2%)	31 (7.8%)	0.8
Recurrence Type			0.2
Both	0 (0%)	2 (6.5%)
Local	4 (40%)	20 (65%)
Metastatic	6 (60%)	9 (29%)
Site of metastatic recurrence			0.3
Bone	1 (17%)	0 (0%)
Bone and Lung	0 (0%)	1 (7.7%)
Brain	0 (0%)	3 (23%)
Lung	5 (83%)	9 (69%)

¹n (%)

²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test

Pathology

Of the 139 children with centrally confirmed CMN, 131 were initially enrolled on AREN03B2 with an institutional pathologic diagnosis of CMN, three with a diagnosis of Wilms tumor (WT), one with a diagnosis of clear cell sarcoma of the kidney (CCSK), one with a differential diagnosis of “CMN versus CCSK”, and three cases came to central review with no institutional diagnosis. Of the 139 centrally confirmed CMN cases, 136 could be further subclassified as classic, cellular, or mixed (detailed clinicopathologic features by histologic subtype are summarized in Table 2). Of the three cases that were not further subclassified, one had insufficient slides for classification and two had unusual histologic findings that precluded definitive subclassification. In 94 cases, lymph nodes (LNs) were sampled with nephrectomy. No nodes were found to be positive for tumor, excepting one patient with a mixed CMN involving a lymph node by direct extension (this is included as one case with a “positive LN” in Table 2).

Table 2.

AREN03B2 Cohort by Pattern - Demographics and Tumor Characteristics

Characteristic	Cellular, N = 831	Classic, N = 301	Mixed, N = 231	p-value2
Age (Months)				<0.001
(Mean, Median)	(3.60, 2.16)	(0.83, 0.48)	(1.29, 0.25)
(Range)	(0.02, 25.03)	(−0.01, 4.79)	(0.02, 7.61)
Sex				0.5
Female	34 (41%)	15 (50%)	12 (52%)
Male	49 (59%)	15 (50%)	11 (48%)
Race				0.079
Asian	6 (8.6%)	1 (3.7%)	0 (0%)
Black	4 (5.7%)	3 (11%)	4 (25%)
Multiple Races	0 (0%)	1 (3.7%)	0 (0%)
Native American	0 (0%)	1 (3.7%)	0 (0%)
White	60 (86%)	21 (78%)	12 (75%)
Unknown	13	3	7
Local Stage				0.2
Stage 1	14 (17%)	3 (10%)	0 (0%)
Stage 2	39 (47%)	17 (57%)	16 (70%)
Stage 3	30 (36%)	10 (33%)	7 (30%)
Disease Stage				0.4
Stage I	11 (17%)	2 (13%)	0 (0%)
Stage II	30 (48%)	9 (56%)	7 (70%)
Stage III	22 (35%)	4 (25%)	3 (30%)
Stage IV	0 (0%)	1 (6.3%)	0 (0%)
Unknown	20	14	13
Rupture/Spill	22 (27%)	0 (0%)	2 (8.7%)	<0.001
Rupture/Spill Timing				0.10
Indeterminate	4 (18%)	0 (NA%)	2 (100%)
Intraoperative Spill	16 (73%)	0 (NA%)	0 (0%)
Preoperative Rupture	2 (9.1%)	0 (NA%)	0 (0%)
Tumor Weight (g)				<0.001
(Mean, Median)	(355, 298)	(81, 75)	(198, 128)
(Range)	(29, 1,160)	(4, 243)	(23, 730)
Unknown	20	4	3
Tumor Diameter (cm)				<0.001
(Mean, Median)	(9.18, 9.10)	(5.46, 5.00)	(7.02, 7.20)
(Range)	(3.70, 15.00)	(3.10, 10.00)	(2.20, 13.00)
Unknown	9	1	0
LN Sampling	64 (77%)	13 (43%)	17 (74%)	0.002
LN Status				0.3
Negative	64 (100%)	13 (100%)	16 (94%)
Positive	0 (0%)	0 (0%)	1 (5.9%)

¹n (%)

²Kruskal-Wallis rank sum test; Pearson’s Chi-squared test; Fisher’s exact test

Radiology

AREN03B2 patients with and without a chest CT are summarized in Supplemental Digital Content Table 1, http://links.lww.com/SLA/F556. Evaluation of pulmonary metastatic disease was not possible in 47 of 139 (34%) CMN cases (CT not submitted or technically uninterpretable), thus, a significant proportion of patients had unknown overall disease stage (Table 2). Of the 92 patients with evaluable pulmonary CTs, only one with classic CMN was classified at time of diagnosis as stage IV. Although this patient is included in this report as stage IV, on review of follow-up information, we confirmed that the lung lesions were not biopsied or resected, no further therapy was given, and there was no further documented relapse or progression. This prompted a repeat (blinded for reason) expert radiology review, which favored atelectasis.

Surgery

Two patients had initial biopsy; one received neoadjuvant EE4A and then underwent delayed nephrectomy, and one had biopsy followed by nephrectomy and received no chemotherapy. Neither of these patients relapsed. All other patients underwent upfront nephrectomy.

LN sampling was performed in 70% of children overall (77%, 74%, and 43% for those with cellular, mixed, and classic histology, respectively) (Table 2).

Thirty-four percent of patients (30–36% across histology subtypes) were stage III due to preoperative rupture, intraoperative spill, or positive surgical margin. Rupture/spill was noted in 27% of patients with cellular histology (73% intraoperative spill, 9.1% preoperative rupture, and 18% indeterminate). No rupture/spill was noted in any patient with classic histology and in only two patients (8.7%) with mixed histology.

Chemotherapy

Twelve children, 10 with cellular CMN and 2 with mixed CMN, were reported to have received adjuvant/neoadjuvant systemic chemotherapy after initial diagnosis. The official responses in the AREN03B2 Case Reporting Forms (CRFs) regarding chemotherapy are reported and statistically analyzed in Supplemental Digital Content Table 2, http://links.lww.com/SLA/F556. These included 6 children receiving EE4A (vincristine and dactinomycin for 19 weeks) and an additional 6 children reported as “other” or “indeterminate.” Further individual review of narrative comments in AREN03B2 records (by JJD and EAM) clarified that for these patients, 1 additional child received EE4A, 3 received VAC (vincristine, dactinomycin, and cyclophosphamide), 1 received vincristine, doxorubicin, dactinomycin, and etoposide, and 1 received vincristine, daunomycin, and cyclophosphamide. For all other patients, institutions reported no chemotherapy given.

Outcomes

Outcome data were available for 137 of the AREN03B2 CMN patients, with a median follow-up of 4.5 years. Ten relapsed (7.2%), with a time to relapse ranging from 1.5 to 18 months (Table 3). Of the relapse cases, four children had local stage II tumors and six had local stage III tumors at time of diagnosis. For stage III cases, four were due to positive margin and two were due to rupture or spill. In six of the relapses (60%), LNs were removed at time of nephrectomy. In four of those cases, the LNs were sent for central pathology review and were negative for metastatic tumor. Relapse sites included four cases of local recurrence in the tumor bed and six cases of distant metastatic disease. Four children with relapse died within one year following relapse (three with local recurrence and one with metastatic disease to the lung). All were reported to have died of progressive disease.

Table 3.

Description of Relapses for Patients on AREN03B2 (N = 137)

Local (abdominal) Stage	Reason	Chemo	Relapse Site	Relapse Months	Followed Years	Vital Status
Stage 2	Sinus	None	Tumor Bed	1.54	0.28	Dead
Stage 2	Sinus	None	Bone	14.16	5.96	Alive
Stage 2	Capsule	None	Lung	2.46	0.97	Dead
Stage 2	Sinus	None	Lung	3.68	4.05	Alive
Stage 3	Preoperative Rupture	None	Lung	6.64	6.59	Alive
Stage 3	Margin	None	Tumor Bed	3.29	5.14	Alive
Stage 3	Margin	None	Lung	2.46	5.44	Alive
Stage 3	Margin	Other	Tumor Bed	2.07	0.20	Dead
Stage 3	Margin	None	Lung	17.94	2.03	Alive
Stage 3	Rupture and/or spill (indeterminate)	None	Tumor Bed	10.61	0.93	Dead

All relapses occurred in children with tumors classified at diagnosis as cellular or mixed histology. Four of 55 (7.3%) patients with local stage II cellular or mixed CMNs relapsed, and six of 37 (16.2%) with local stage III cellular or mixed CMNs relapsed. Table 4 compares characteristics of CMN patients that relapsed versus those that did not. Median tumor weight in children that relapsed was larger than in children who did not relapse (430g vs. 154g; p = 0.005) (Table 4). Tumor weight also correlated with histology (Table 2). Tumor diameter correlated with relapses and histology in a similar manner (Tables 2, 3, and 4).

Table 4.

Non-Relapsed vs Relapsed AREN03B2 CMNs

Characteristic	Non-Relapsed CMN, N = 1291	Relapsed CMN, N = 101	p-value2
Age (Months)			0.2
(Mean, Median)	(2.36, 1.07)	(6.06, 4.39)
(Range)	(−0.01, 15.59)	(0.05, 25.03)
Tumor Diameter			0.016
(Mean, Median)	(7.68, 7.20)	(9.87, 10.25)
(Range)	(2.20, 15.00)	(3.50, 13.00)
Unknown	10	0
Tumor Weight			0.005
(Mean, Median)	(242, 154)	(470, 430)
(Range)	(4, 1,160)	(195, 670)
Unknown	24	3
Disease Stage			0.7
Stage I	15 (17%)	0 (0%)
Stage II	44 (51%)	3 (50%)
Stage III	26 (30%)	3 (50%)
Stage IV	1 (1.2%)	0 (0%)
Unknown	43	4

¹n (%)

²Wilcoxon rank sum test; Fisher’s exact test

Six of the 10 children who relapsed had full staging with chest CT imaging submitted for central review, and all six demonstrated no evidence of pulmonary metastases at time of diagnosis. The remaining four children did not have evaluable upfront pulmonary imaging, resulting in incomplete upfront disease staging.

Of the 10 children who relapsed, one received initial adjuvant chemotherapy (VAC, Supplemental Digital Content Table 2, http://links.lww.com/SLA/F556). Three relapsed patients were given chemotherapy after a diagnosis of relapse. One patient received DD4A and ICE, one received DD4A only, and one received vincristine and dactinomycin. No child received radiation therapy.

Historical CMN Patient Cohort

There were 396 children with confirmed historical CMNs and are summarized alongside AREN03B2 patients in Table 1. Additional details by histologic pattern are presented in Table 5 and Supplemental Digital Content Table 3, http://links.lww.com/SLA/F556. Overall disease stage and details on chemotherapy received, surgical approach, and intraoperative events were not available.

Table 5.

NWTS/OC Cohort by Pattern - Demographics, Tumor, Relapse Sites

Characteristic	Cellular, N = 2391	Classic, N = 691	Mixed, N = 881	p-value2
Age (Months)				<0.001
(Mean, Median)	(5.6, 3.0)	(0.9, 0.0)	(2.7, 0.0)
(Range)	(0.0, 48.0)	(0.0, 24.0)	(0.0, 20.0)
Unknown	4	2	4
Stage				0.009
Stage I	45 (25%)	9 (18%)	4 (6.2%)
Stage II	63 (35%)	22 (45%)	27 (42%)
Stage II vs III	26 (14%)	10 (20%)	21 (32%)
Stage III	44 (24%)	8 (16%)	13 (20%)
Stage IV	2 (1.1%)	0 (0%)	0 (0%)
Unknown	59	20	23
Tumor Weight (g)				<0.001
(Mean, Median)	(356, 290)	(102, 93)	(203, 150)
(Range)	(23, 2,006)	(19, 380)	(27, 745)
Unknown	80	20	27
Tumor Diameter (cm)				<0.001
(Mean, Median)	(8.4, 8.0)	(5.0, 4.5)	(7.6, 7.0)
(Range)	(0.6, 17.0)	(1.6, 8.5)	(3.0, 15.0)
Unknown	111	28	48
Fusion Transcript				0.9
Negative	4 (44%)	0 (NA%)	3 (30%)
Positive	5 (56%)	0 (NA%)	7 (70%)
Unknown	230	69	78
Internal-Renal VI				<0.001
No	177 (82%)	63 (100%)	73 (95%)
Possible	11 (5.1%)	0 (0%)	2 (2.6%)
Yes	29 (13%)	0 (0%)	2 (2.6%)
Unknown	22	6	11
Vascular Pattern				0.030
Angiocentric Necrosis	89 (44%)	0 (0%)	11 (22%)
No Vascular Pattern	51 (25%)	1 (100%)	19 (38%)
Vascular Pattern	64 (31%)	0 (0%)	20 (40%)
Unknown	35	68	38
Recurrence	21 (8.8%)	0 (0%)	10 (11%)	0.021
Recurrence Type				0.8
Local	14 (67%)	0 (NA%)	6 (60%)
Metastatic	6 (29%)	0 (NA%)	3 (30%)
Both	1 (4.8%)	0 (NA%)	1 (10%)
Distant Recurrence				0.3
Bone and Lung	1 (11%)	0 (NA%)	0 (0%)
Brain	1 (11%)	0 (NA%)	2 (50%)
Lung	7 (78%)	0 (NA%)	2 (50%)

¹n (%)

²Kruskal-Wallis rank sum test; Pearson’s Chi-squared test

Outcomes

Outcomes from the historical cohort are summarized in Table 6. There were 31 children (7.8%) with reported relapse, with a time to relapse ranging from two to 11.5 months post-nephrectomy. Of the relapse cases, ten died of disease (32%). Relapse was more likely to occur in local stage III disease (Table 7). As in AREN03B2, all relapses were cellular or mixed tumors.

Table 6.

Description of Relapses for patients in the Historical Cohort (N = 31)

ID	Disease_Stage	Relapse_Site	Relapse_Months	Vital
1	Stage II	Bone and Lung	5.0	Died of Disease
2	Stage II	Brain	7.0	Died of Disease
3	Stage II	Brain	2.0	Died of Disease
4	Stage II	Lung	11.5	Alive
5	Stage II	Lung	10.0	Alive
6	Stage II	Lung	3.0	Alive
7	Stage II	Local and Lung	2.0	Alive
8	Stage II	Local	7.0	Died other Causes
9	Stage II vs III	Local	6.0	Died of Disease
10	Stage II vs III	Local	3.0	Alive
11	Stage II vs III	Local	4.0	Alive
12	Stage II vs III	Local	4.0	Died of Disease
13	Stage III	Local and Lung	2.0	Died of Disease
14	Stage III	Lung	9.0	Died of Disease
15	Stage III	Lung	3.0	Alive
16	Stage III	Local	2.0	Alive
17	Stage III	Local	2.0	Alive
18	Stage III	Local	3.0	Alive
19	Stage III	Local	2.5	Alive
20	Stage III	Local	2.0	Alive
21	Stage III	Local	9.0	Alive
22	Stage III	Local	2.0	Alive
23	Stage III	Local	5.0	Alive
24	Stage III	Local	7.0	Alive
25	Stage III	Local	7.5	Died of Disease
26	Stage III	Local	3.0	Died of Disease
27	Stage III	Local	11.0	Alive
28	Stage III	Local	11.0	Alive
29	Stage III	Local	4.0	Died of Disease
30	Stage III	Local	6.0	Alive
31	Unknown	Lung	6.0	Alive

Table 7.

Non-Relapsed vs Relapsed Historical CMNs

Characteristic	Non-Relapsed CMN, N = 3651	Relapsed CMN, N = 311	p-value2
Age (Months)			0.2
(Mean, Median)	(4.0, 2.0)	(5.7, 3.0)
(Range)	(0.0, 48.0)	(0.0, 25.0)
Unknown	10	0
Tumor Diameter			0.051
(Mean, Median)	(7.4, 7.0)	(9.3, 9.0)
(Range)	(0.6, 17.0)	(4.0, 15.0)
Unknown	169	18
Tumor Weight			0.004
(Mean, Median)	(263, 190)	(400, 350)
(Range)	(19, 2,006)	(31, 1,130)
Unknown	120	7
Disease Stage			<0.001
Stage I	58 (22%)	0 (0%)
Stage II	104 (39%)	8 (27%)
Stage II vs III	53 (20%)	4 (13%)
Stage III	47 (18%)	18 (60%)
Stage IV	2 (0.8%)	0 (0%)
Unknown	101	1
Mitoses			0.2
1	33 (9.7%)	1 (3.8%)
2	251 (74%)	19 (73%)
3	51 (15%)	4 (15%)
4	6 (1.8%)	2 (7.7%)
Unknown	24	5
Necrosis			0.054
Extensive	191 (56%)	19 (73%)
No	98 (29%)	2 (7.7%)
Yes	52 (15%)	5 (19%)
Unknown	24	5
Internal-Renal VI3			0.002
No	298 (89%)	15 (65%)
Possible	12 (3.6%)	1 (4.3%)
Yes	24 (7.2%)	7 (30%)
Unknown	31	8
Sinus VI			0.021
No	281 (87%)	14 (67%)
Possible	8 (2.5%)	0 (0%)
Yes	35 (11%)	7 (33%)
Unknown	41	10

¹n (%)

²Wilcoxon rank sum test; Fisher’s exact test

³VI: vascular invasion

Comparisons Between the AREN03B2 and Historical Cohort

The overall distribution of histologic CMN subtypes was similar between those enrolled on AREN03B2 and the historical cohort (Table 1). Age at diagnosis (p = 0.7), tumor diameter (p = 0.2), and tumor weight (p = 0.3) were also similar.

Discussion

In this study, we present an analysis of pathologic and clinical data from children with CMN enrolled on the AREN03B2 Renal Tumor Biology and Classification Study, as well as data from a large patient cohort of retrospectively collected CMNs. Of note, the historical collection is not included with the AREN03B2 analysis for three major reasons: 1) these cases were subject to selection bias as most were referrals for consult, 2) overall disease stage was unknown for all cases in this group, and 3) clinical treatment, surgical approach, and outcome data were limited as most cases were not part of a formal clinical protocol and follow-up was not collected. Despite the limitations outlined above, the AREN03B2 and historical cohorts are remarkably similar (Table 1). We initially aimed to study the impact of individual prognostic factors on outcome through multivariable modeling, with a goal to guide clinical decisions. However, given the rarity of the primary diagnosis of CMN compounded by its overall low relapse rate, the number of children with events are too few in this cohort to reasonably fit a multivariable Cox regression model.

Our data support that risk of relapse is restricted to children with stage II and III mixed and cellular subtypes. Tumor weight and diameter correlated with histology and risk of relapse, with larger tumors more likely to be mixed and cellular CMN subtypes. Tumor rupture/spill notably occurred largely in patients with cellular histology (27% of cellular CMNs had tumor spill/rupture, and the majority of these [73%] occurred as intraoperative spill).

Despite the frequency of rupture and spill in cellular CMNs, the limited number of patient relapse events did not allow us to draw firm conclusions regarding a correlation of rupture/spill and risk of relapse. However, the high rate of intraoperative spill highlights a need for continued investigation into non-invasive methods, such as circulating tumor DNA, to identify an NTRK fusion. There is growing evidence on the use of neoadjuvant TRK inhibitors in NTRK fusion-positive tumors resulting in remarkable reduction in tumor size and potentially improving safe surgical resectability.³⁷ Of note, establishing histology and molecular data through tumor biopsy is not an advocated approach for patients with renal tumors that are considered resectable in the COG treatment paradigm because biopsy and gross residual disease after the diagnostic procedure are criteria for stage III disease. For malignant tumors other than CMN, this would result in additional therapy. We continue to advocate for upfront radical nephrectomy, acknowledging that this can be challenging in cellular CMN given the tendency towards tumor infiltration medially along the renal hilar vessels, as well as common cystic changes. Additionally, given the young age of many of these infants, these renal tumors are generally quite large relative to the body cavity size. These tumor characteristics pose distinct technical challenges.

Interestingly, no CMNs of any subtype in this cohort were reported to metastasize to LNs. However, LN sampling was performed in only 77% and 74% of patients with cellular and mixed histology, respectively, and in only 43% of patients with classic histology. The COG Renal Tumors Committee strongly recommends routine LN sampling for all pediatric renal tumors, as this is critical for staging and therapy assignment, including avoidance of any subsequent chemotherapy whatsoever for very low risk Wilms tumor. Tumor histology is often not known at the time of surgery.

The risks and benefits of upfront and surveillance imaging merit careful consideration in patients with cellular or mixed CMNs. Patients with local stage II cellular or mixed CMNs had an overall relapse rate of 7.3%, and local stage III cellular or mixed CMNs had an overall relapse rate of 16.2% in the AREN03B2 cohort. An important limitation of this study is that not all patients had available or evaluable pulmonary CT scans at diagnosis. The data presented here suggest that children with classic CMN of any local stage, or local stage I patients with any histology, may not benefit from diagnostic lung imaging because there were no recurrences in these groups. For children with cellular or mixed histology and local stage II or III tumors, diagnostic lung imaging is recommended. This can be performed post-operatively, when histology and local stage is established, however, appropriateness of pre- versus post-surgical chest imaging should consider the age of the infant, as it relates to the likelihood of histologic diagnosis. Infants are more likely to have CMN than WT in the first month, but after 2 months, the probability of WT increases significantly⁴ if WT is the more likely histology, pre-operative CT should be considered as this staging will be required and may be complicated post-operatively by atelectasis and effusions. For these same children considered at risk for relapse, post-surgical short-interval surveillance of lungs and the local tumor site at least every three months is also recommended as relapses occurred quickly (4/10 relapses occurred within three months). The use of CT to detect relapse for patients with WT was shown not to provide a survival advantage over chest x-ray and ultrasound;³⁷ we do not know if this would also apply to CMNs. However, with the low risk of recurrence and the concern of radiation exposure for young infants, the use of chest x-ray and ultrasound are reasonable considerations for surveillance in children with CMN. Recurrences were not reported beyond 1.5 years from diagnosis, so longer-term follow-up is likely unnecessary.

It has long been recognized that the overall risk of recurrence for infants with CMN is low. On AREN03B2, 92.7% of patients with stage II mixed or cellular CMNs did not relapse, and 83.8% with stage III mixed or cellular CMNs did not relapse. However, 30–40% of children with relapse died of disease, underlining the gravity of the occurrence of relapse and need for surveillance, particularly in the identified higher risk sub-group of children with stage II- III tumors with any cellular histology (mixed or cellular subtypes.) Factors preventing relapse for those children with higher risk tumors are not known. Only one of the ten children with relapsed CMN on AREN03B2 was reported to have received adjuvant chemotherapy, but there is insufficient evidence to conclude that adjuvant therapy prevented relapse in others. Of the 92 children with higher risk for relapse (stage II/III cellular/mixed CMNs), 12 (13%) received adjuvant chemotherapy, and only one relapsed (8%). Of the 80 higher-risk children that were not reported to have received adjuvant chemotherapy, nine relapsed (11%). To our knowledge, the use of newer targeted treatments using NTRK inhibitors were not available to any of the patients in either cohort, but are an area of needed future study for this population, given multiple reports of demonstrable clinical activity in both adjuvant and neo-adjuvant settings.^39–43

In summary, this analysis of a combined cohort of 535 children represents the largest reported cohort of centrally reviewed patients with CMN. Our review has established the following conclusions:

• Stage II and III mixed and cellular subtypes have the highest risk for relapse, and there is a low risk of relapse for all other children with CMN.

• No CMNs of any subtype were reported to metastasize to LNs; higher local stage (III) was due to positive margins and intraoperative spill in many cases, emphasizing the challenges of surgery in young infants with large, cellular tumors.

• Tumor weight and diameter correlated with histology (mixed and cellular CMNs were larger) and with risk of relapse.

• Tumor rupture/spill notably occurred almost exclusively in patients with cellular histology.

• No clear role for adjuvant therapy has been demonstrated. It is likely unnecessary for patients with classic histology CMNs and stage I CMNs of any subtype. It may be unnecessary for all patients prior to relapse.

• Historic relapse therapy is shown to be inadequate, and the poor outcomes after relapse highlight the need for prospective study of the role of NTRK inhibitors for children with CMN.