Methylation changes and INS-IGF2 expression predict progression in early-stage Wilms tumor

Abstract

Wilms tumor, the most common pediatric kidney cancer, accounts for 5% of childhood cancers and is classified by stage and histological subtype. Despite high survival rates (80–85%), approximately 15% of patients experience relapse, reducing survival to around 50%. Epigenetic changes, particularly DNA methylation, play a critical role in Wilms tumor pathogenesis. This study investigates the prognostic potential of DNA methylation in stage I and II patients with favorable histology, aiming to identify early relapse biomarkers. Genome-wide methylation was assessed using methylation microarrays in tumor tissues from relapsed patients (n = 9) and those with complete responses (n = 9), alongside normal tissues (n = 3 each). Differentially methylated probes and regions were analyzed, with additional ROC and survival analyses. Real-time PCR was used to measure IGF2 and INS-IGF2 gene expression. The analysis revealed hypomethylation in intergenic regions in remission patients, identifying 14 differentially methylated positions as potential biomarkers. Increased INS-IGF2 expression was associated with relapse, suggesting its role in disease progression. While the study concentrated on stages I and II patients, where relapse rates are lower, this focus inherently led to a smaller sample size. Despite this, the findings provide valuable insights into the potential role of DNA methylation markers for monitoring disease progression and guiding personalized treatment in Wilms tumor patients.

Graphical abstract

Genome methylation analysis of WT tumor and normal tissues from complete remission and relapse patients revealed 14 differentially methylated probes (DMPs) and three differentially methylated regions (DMRs) in tumor samples between both groups. Most DMPs demonstrated strong predictive performance for overall and event-free survival. RNA expression analysis showed elevated INS-IGF2 levels in relapse tumor tissue, highlighting its role in WT progression.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-024-01775-y.

Background

Wilms tumor (WT), also known as nephroblastoma, is the most commonly observed renal tumor within the genitourinary tract of children and constitutes 5% of all childhood malignancies [1]. WTs are usually classified by location, spread and number of kidneys involved to stages I–V, and by histological subtype to favorable (FHWT) or unfavorable histology (UFHWT) [2]. As per the Children’s Oncology Group (COG) protocol, fully resectable stages I, II and III WT are treated with upfront nephrectomy followed by chemo-radiotherapy. For inoperable stage III, as well as stages IV, and V, preoperative chemotherapy is required to shrink the tumor before surgery. Despite high overall survival (OS) rates of 80–85% [3, 4], even reaching 90% in high-income countries, approximately 15% of treated patients experience relapse. Recurrence of the disease significantly lowers overall survival to around 50% and complicates patient outcomes [5].

Similar to other cancers, requiring multiple hits, several genetic and epigenetic changes were identified to be involved in WT etiology [6]. Underlying predisposition syndromes and genetic changes were identified in up to 15% of the WT cases, serving as the first hit for cancer development. One of the first genes identified to predispose to WT development is the Wilms tumor 1 gene, WT1, a transcriptional factor crucial for normal kidney development [7]. Developmental syndromes such as WAGR (Wilms’ tumor, aniridia, genital anomalies and retardation) and DDS (Denys–Drash syndrome) are associated with WT1 pathogenic variations and were shown to have an increased risk of WT development (30–75%) and lower mean ages of WT diagnosis [8, 9]. Another major predisposition factor for WT development is the loss of imprinting (LOI) at imprinting control regions 1 and 2, (IC1 and IC2) on chromosome 11, which causes an activated expression of IGF2, a cellular proliferation factor, and reduced expression of H19, a long noncoding RNA that regulates cellular growth [10]. Syndromes with LOI abnormalities, such as BWS (Beckwith–Wiedemann syndrome) subtype characterized by IC1 gain of methylation, have an increased risk for WT (~ 25%) [10, 11]. Other predisposition factors include CTR9 mutations, DICER1 syndrome and TRIM28 mutation, which mostly affect chromatin or methylation levels at the chromatin level [12].

One of the earliest identified epigenetic hallmarks of cancer is the hypomethylation of repetitive regions across the entire genome. Paradoxically, it was later identified to be associated with regions of hypermethylation in promoters of tumor suppressor genes, leading to their silencing. Aberrant DNA methylation is an early event in tumorigenesis, occurring not only in cancerous tissues but also in abnormal non-neoplastic tissues [13, 14]. With a focus on focal hypermethylation of tumor suppressor gene promoters, DNA demethylating agents have been approved as cancer chemotherapeutic agents either alone or in combination, in the treatment of myelodysplastic syndromes (MDS), acute myeloid leukemia and chronic myelogenous leukemia [15, 16]. This, however, has to be approached with caution, because induced DNA hypomethylation has been proven to cause carcinogenesis or tumor progression in many models [17, 18].

The early onset of DNA methylation changes make detection of these an attractive diagnostic and prognostic biomarker. Several studies link DNA methylation changes with patient survival and prognosis in several cancers such as brain, breast, lung and head and neck cancers [19–23]. Because of the strong association of aberrant DNA methylation and WT development, several studies investigated its association with WT progression, prognosis and patient survival. Focal methylation changes, such as reduced P73 promoter methylation, was found to be associated with poor prognosis [24]. Other studies have linked overall methylation patterns with risk of disease progression and have formed stratification criteria based on methylation patterns [25, 26]. These studies have helped define specific methylation markers to identify high-risk groups, but still little is known about the causes of progression in a subset of the otherwise considered low-risk WTs.

In this study, we aimed to identify biomarkers for relapse in patients with stages I and II FHWT. We used microarray technology to explore methylation patterns in both WT and adjacent healthy tissues from excised kidneys prior to any chemotherapy treatment. We report several differentially methylated regions (DMRs) that show potential as prognostic biomarkers for disease relapse and overall patient survival. Finally, to explain the mechanistic role of DNA methylation changes in WT relapse, we explore the genes controlled by the differentially methylated IGF2 region and show elevated expression of INS-IGF2 transcript in tumor tissue of patients who have relapsed after treatment.

Methods

Ethical approval and sample collection

Informed consents, approved by the Institutional Research Ethics Board (IRB) at CCHE 57357, were collected from guardians of patients with stage I or II WT with favorable histology. Tumor and normal FFPE tissue samples were collected from the Pathology Department from two cohorts of patients: nine who achieved complete remission (CR) and were followed up for at least 6 years post-treatment (mean follow-up period: 93 months), and nine who experienced recurrent disease within 5 years of treatment (mean relapse time: 16 months). The clinicopathological characteristics of the patients recruited are listed in Table 1.

Table 1.

Clinicopathological characteristics of enrolled WT patients

	Complete remission group (n = 9)	Relapse group (n = 9)	Complete remission normal samples (n = 3)	Relapse group normal samples (n = 3)
Age (years) (Mean ± SD)	2.75 ± 1.68	4.07 ± 3.17	2.99 ± 1.72	6.47 ± 4.4
Gender (Female, Male)	6, 3	3, 6	2, 1	0, 3
Stage (I, II)	4, 5	5, 4	1, 2	2, 1
5-year overall survival (% survival)	100%	33%	100%	0%

Sample preparation

Ten sections of 8 μm thickness were obtained from formalin-fixed, paraffin-embedded (FFPE) tissue blocks of the initially resected kidney samples (prior to any chemotherapy), from eighteen tumor tissues (nine from complete remission patients and nine from relapse patients) and six adjacent normal tissues (three from complete remission patients and three from relapse patients). Genomic DNA was extracted from the FFPE sections using QIAamp DNA FFPE tissue kit (Qiagen) according to the manufacturer’s instructions. DNA was quantified using the Denovix Fluorometer (dsDNA High Sensitivity). The quality of the extracted FFPE DNA samples was assessed by Illumina FFPE QC kit (Illumina Inc.). Bisulfite conversion of extracted DNA was performed using the EZ DNA methylation kit (D5002, Zymo Research) according to the manufacturer’s instructions using the alternative incubation conditions recommended for the Illumina Infinium methylation arrays. Bisulfite-converted FFPE DNA was then restored with Infinium HD FFPE DNA Restore Kit (WG-321-1002, Illumina Inc.). Restored bisulfite-modified DNA samples were hybridized to the Illumina Infinium Human Methylation EPIC bead chips and scanned using the Illumina iScan microarray scanner (Illumina Inc.) according to the manufacturer’s recommendations.

Bioinformatics analysis

The bioinformatics workflow is summarized in Fig. 1. R software (v4.3.1) and minfi R package (v1.44.0) [27] were used to load the Illumina Human Methylation EPIC intensity data files (IDAT) and check their quality. To overcome technical variations between probe types I and II, functional normalization [28] was employed. Probes with detection p values less than 0.01 were removed. EPIC Manifest (v. 1.0.0B5) was used for probe annotation, with the exclusion of probes with polymorphism in their CG sites with minor allele frequency (MAF) larger than 0.01 (i.e., non-rare polymorphism), probes present on single-nucleotide polymorphism sites (SNPs) [29] or sex chromosomes and probes matching more than one genomic site (i.e., cross-reactive probes) [30, 31]. Only preprocessed probes with annotation data were selected, and 547,055 probes were ultimately chosen for further analysis.

Fig. 1.

Methylation analysis bioinformatics workflow. The workflow includes data processing milestones (normalization, preprocessing, M-value calculations, followed by differential expression, ROC curves and survival analysis). The diagram was created using draw.io software

M-values were calculated, and differential analysis was performed using the limma R package (v3.54.2). Our comparisons included relapsed tumor samples (RT) versus complete remission tumor samples (CRT), as well as tumor (RT and CRT) versus normal tissue samples (CRN and RN). Differentially methylated probes (DMPs) from both comparisons were determined based on an absolute log2 fold change (abs(log2FC) ≥ 0.5), with a false discovery rate (FDR) of < 0.05. Volcano plots were generated from the normalized M-values for DMPs, and a heatmap was plotted for visualization.

Further investigations were performed to compare CRT and RT samples. To identify overall differences, UMAP analysis was performed using UMAP R library (v0.2.10.0.) with number of neighbors = 8, and using the first two components. To test differences for each probe subtype, average M-value for each probe subtype was calculated for each sample group, statistical significance was analyzed using Mann–Whitney U-test, and was considered significant if p value < 0.05. To probe the utility of the identified DMPs as biomarkers, logistic regression was performed using the Scikit-learn (v1.3.2) library in Python, and the resulting model was used to plot receiver operating characteristic (ROC) curves. ROC curves were also plotted using IBM SPSS Statistics v22.0 to identify M-value cut-offs showing the highest sensitivity and specificity for each probe. Survival analysis was conducted using the survival (v3.5.7) and survminer (v0.4.9) R packages, based on the cut-offs for each probe obtained earlier. To test the relationship between methylation levels (M-values) in each of the fourteen DMPs and time to relapse, linear regression was performed, with time to last follow-up used in complete remission patients. To further investigate DMRs, we utilized the dmrcate function from the DMRcate [32] package (v2.12.0), using a lambda of 1000 and C of 2, while consecutive probes were set to true.

RNA expression

Further sections were obtained from the same FFPE samples as previous. RNA was extracted from the FFPE sections using AmoyDx FFPE RNA extraction kit (AmoyDx, Xiamen, China) according to manufacturer’s instructions. RNA was quantified using the Qubit Fluorometer (RNA broad range). The quality of the RNA was assessed using Agilent RNA 6000 Nano (Agilent technologies, USA). cDNA was generated using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, USA). RT-PCR was used to analyze relative expression levels using Maxima SYBR Green qPCR Master Mix (Thermo Fisher Scientific, USA). Ct values were analyzed relative to GAPDH Ct values. Primers used for GAPDH are FP: TGCCCTCAACGACCACTTTG and RP: CCACCACCCTGTTGCTGTAG, for IGF2; FP: TGGCATCGTTGAGGAGTGCTGT and RP: ACGGGGTATCTGGGGAAGTTGT, and for INS-IGF2; FP: ATCATCGTCCAGGCAGTTTCGG and RP: ACACAAGCTCGGTGGTGACTCT. ΔCt was calculated as Ct (IGF2 or INS-IGF2) − Ct (GAPDH) and then converted to 2^−ΔCt. Expression is reported relative to adjacent normal tissue for each group separately; CRT relative to CRN and RT relative to RN. Statistical significance was analyzed using Mann–Whitney U-test and was considered significant if p value < 0.05.

Results

Global hypomethylation and focal hypermethylation are hallmarks of WT pathogenesis

To observe the overall methylation changes that occur during tumorigenesis in WT, we performed methylation analysis to compare the 18 tumor tissue samples and the six adjacent normal tissue. As previously shown for WT and other cancers, there was a significant difference in methylation levels between normal and cancerous tissues (Fig. 2a). A total of 53,423 probes were significantly differentially methylated between tumor and adjacent normal tissues, and the top 100 are shown in Fig. 2b. The majority of the hypomethylated probes were far from CpG sites, in the open sea regions, and within gene bodies (Fig. 2c and d). Hypermethylated probes, however, were more abundant at promoter-associated regions (Fig. 2c and d).

Fig. 2.

Global differences in methylation levels between WT and adjacent normal tissues. a Volcano plot of methylation differences between WT and adjacent normal tissues, showing a total of 53,423 DMPs, 32,984 hypomethylated and 20,439 hypermethylated. b Heatmap showing the top 100 DMPs, annotated for relation to CpG island, and genes. c–d Stacked bar plot showing localization of hyper- and hypo-methylated CpG sites in WT tissue relative to CpG islands (c) and to their adjacent genes (d)

Methylation differences can predict WT relapse

In an attempt to discover methylation differences that can help predict relapse in stages I and II FHWT patients, we compared tumor DNA methylation levels on upfront nephrectomy samples between nine patients who remained in complete remission for at least 5 years post-treatment, and nine who had relapsed within 5 years. When analyzing all probes together, we could not see distinct clustering between relapse tumor samples (RT) and complete remission tumor samples (CRT) using UMAP clustering (Fig. 3a). When the probes were separated based on their location relative to genes and CpG islands, we observed clustering between RT and CRT samples only in the open sea intergenic regions (Fig. 3a and Supplementary Fig. 1), where CRT samples showed significantly lower methylation levels (Fig. 3b, Mann–Whitney U-test, p value < 0.05).

Fig. 3.

Methylation differences between relapse and complete remission patients. a UMAP clustering of methylation probes for CRT and RT samples for all probes, and for probes in the open sea intergenic regions. b Bar plot of average M-values for each probe subtype and sample group, statistical significance is tested using Mann–Whitney U-test, one asterisk for p value < 0.05. c Volcano plot of methylation differences between CR and R patients showing 14 DMPs; 13 hypomethylated and 1 hypermethylated. d Heatmap showing M-values across different samples in all four groups, annotated for relation to genes and to CpG islands

Fourteen DMPs between relapse (R) and complete remission (CR) patients were identified (Fig. 3c and d), with almost all of them hypermethylated in relapse patients. When analyzed in the context of adjacent normal tissue, these DMPs can be divided into two major groups. The first group of probes shows pronounced hypomethylation in complete remission patients, where the majority of which are in open sea, intergenic and gene body regions. This includes probes in the gene body of C12orf42 and AGXT2L1, and in the proximity of RP11-65D13.1 and the lncRNA AP000233.4. The second group of probes shows hypermethylation in relapse patients. This includes probes in the promoter-associated region of the inflammatory gene prostaglandin endoperoxide synthase 2, PTGS2. A notable probe is cg02807948, which is the only probe that is significantly hypomethylated in relapse patients, and it resides within the imprinted IGF2:Ex9-DMR (DMR2).

Specific methylation foci represent promising biomarkers of the disease prognosis

The potential of utilizing these probes as biomarkers for disease prognosis was explored through logistic regression models, followed by the generation of receiving operator characteristic (ROC) curves. The area under curve (AUC) with p values and cut-off M-values showing the best sensitivity and specificity are reported in Table 2. Most of the probes showed AUC > 0.8 and p value < 0.05, with the best performing probe, cg21987076, showing an AUC = 1. Forming a panel with all 14 probes as a prognostic marker showed AUC of 1. By the cut-off value that was determined by the ROC curve, patients were divided into two groups, those with M-values above and below the cut-off. Kaplan–Meier survival analysis and log-rank test were used to compare the difference in event-free survival (EFS) and overall survival (OS) times between both groups. Probes with their log-rank test p values < 0.05 for both EFS and OS are shown in Fig. 4. Once again, the best performing probe is cg21987076, showing the greatest difference in EFS and OS rates between both groups. Altogether, for all probes except cg02807948, a higher M-value above the cut-off is associated with poor prognosis.

Table 2.

Metrics from logistic regression and ROC curve analysis

Probe ID	AUC	p value	Cut-off (M-value)	Sensitivity (%)	Specificity (%)
cg21846220	0.988	4.87 × 10–4	> − 1.276	100	88.9
cg15822093	0.988	4.87 × 10–4	> − 0.873	100	88.9
cg09455265	0.951	1.27 × 10–3	> 0.159	100	88.9
cg16101346	0.877	7.08 × 10–3	> − 1.26	66.7	100
cg07086381	0.988	4.87 × 10–4	> 0.367	100	88.9
cg21987076	1.000	3.49 × 10–4	> 0.705	100	100
cg02807948	0.642	0.31	< − 3.27	64.2	31.0
cg07025752	0.988	4.87 × 10–4	> − 1.913	100	88.9
cg25147026	0.901	4.11 × 10–3	> 0.159	100	88.9
cg08690112	0.938	1.72 × 10–3	> − 0.399	77.8	100
cg07746081	0.889	5.41 × 10–3	> 1.326	88.9	88.9
cg04558713	0.975	6.75 × 10–4	> − 0.164	88.9	100
cg13787671	0.778	0.047	> − 1.924	77.8	88.9
cg01545223	0.926	2.32 × 10–3	> − 2.694	88.9	88.9

Area under the ROC curve (AUC) for all fourteen methylation probes, the cut-off M-values and their associated sensitivity and specificity

Fig. 4.

DMPs can be used as prognostic biomarkers for predicting event-free (EFS) and overall survival (OS). Kaplan–Meier survival analysis was performed after dividing samples based on cut-offs obtained from ROC curves. a Event-free survival differences are plotted for the 14 probes, with p values calculated using the log-rank test. b Overall survival differences are plotted for the 14 probes with significant p values (< 0.05) from the log-rank test

To identify if the level of methylation affects the time for disease relapse, we performed linear regression analysis for the fourteen DMPs. We identified three probes that correlate with time to disease relapse, cg15822093 (R² = 0.83, p value = 1.6 × 10^–7), cg07086381 (R² = 0.765, p value = 2.03 × 10^–6) and cg21987076 (R² = 0.717, p value = 9.23 × 10^–6). Altogether, these findings suggest that these probes, especially cg21987076, hold significant promise as prognostic biomarkers, with potential clinical implications for predicting survival outcomes and disease relapse.

Hypomethylation in IGF2:Ex9-DMR causes elevated expression of INS-IGF2 contributing to resistance to chemotherapy

We investigated the methylation differences across consecutive probes, defining regions of differential methylation (DMRs). When comparing CRT vs RT samples; we identified three significant DMRs (Table 3) on chromosomes 13, 11 and 1. These DMRs overlap with several genes, including ncRNAs genes; small nucleolar RNA (SNORD38), long non-coding RNA (LINC0559) and microRNA (MIR483); pseudogenes PEX12P1, FAR1P1, KRT8P27 and RNA5SP34 and proliferatorive genes such as insulin growth factor 2 (IGF2) and BMP/Retinoic Acid Inducible Neural Specific 3 (BRINP3). To further characterize methylation changes relative to surrounding normal tissue, we compared RT vs RN samples and CRT vs CRN samples (Supplementary Table 2). The RT vs RN comparison yielded 1543 DMRs, while CRT vs CRN yielded 296 DMRs, suggesting more extensive methylation changes in relapse samples.

Table 3.

Differentially methylated regions and overlapping genes between CR and R patients

DMR	Chromosome	Start position	End position	Width	Number of CpGs	Min. smoothed FDR	Overlapping genes
1	chr13	90,488,675	90,883,434	394,760	8	1.54 × 10–44	SNORD38, PEX12P1, FAR1P1, KRT18P27, LINC00559, RNA5SP34
2	chr11	2,154,432	2,156,282	1851	4	1.31 × 10–40	IGF2, INS-IGF2, MIR483
3	chr1	190,232,637	190,323,079	90,443	3	5.36 × 10–40	RP11-547I7.1, BRINP3

Due to its established association with WT, we examined chromosome 11 in more detail. Figure 5a shows the M-values across this region for all three comparisons, highlighting three known imprinting-associated DMRs (iDMRs); H19/IGF2:IG-DMR (ICR1) and IGF2:Ex9-DMR (DMR2) and IGF2:alt-TSS-DMR (DMR0). Individual sample M-values for each group across this region are presented in Supplementary Table 1. As shown in Table 3, Supplementary Table 2 and Fig. 5a, both RT and CRT exhibit gain of methylation in H19/IGF2:IG-DMR (IC1), whereas the other two secondary iDMRs, IGF2:Ex9-DMR (DMR2) and IGF2:alt-TSS-DMR (DMR0), show variable changes. Specifically, IGF2:Ex9-DMR (DMR2) showed loss of methylation in both CRT and RT as compared to their respective normal tissues with a greater effect in RT samples (Supplementary Table 2, Fig. 5a). IGF2:alt-TSS-DMR (DMR0) exhibits loss of methylation only in RT samples (Supplementary Table 2 and Fig. 5a). To assess the functional impact of these methylation changes on RNA expression, we examined IGF2 and INS-IGF2 expression via real-time PCR. Ct values of IGF2 and INS-IGF2 normalized to GAPDH, were used to calculate ΔCt values, and relative expression was derived by dividing 2^−ΔCt for tumor vs adjacent normal tissue within each group. Our results show a 20-fold increase in INS-IGF2 expression levels in tumor tissues of relapse patients than that of complete remission patients (p value < 0.001), whereas IGF2 expression differs by less than twofold between these groups (Fig. 5b). Altogether, this indicates that while there is a similar gain of methylation at IC1, the loss of methylation in DMR2 and DMR0 correlates with higher INS-IGF2 levels, potentially contributing to chemotherapy resistance and tumor relapse (Fig. 5c).

Fig. 5.

Loss of methylation in IGF2:Ex9-DMR causes elevated expression of INS-IGF2 contributing to resistance to chemotherapy a Visualization of methylation levels and differentially methylated regions in each of the four subgroups across three imprinting-associated DMRs on chromosome 11. Loss of methylation is observed in tumor samples at IC1 and DMR2, whereas DMR0 shows loss of methylation in tumor samples of relapse patients only. Coordinates for each DMR (GRCh37) are as follows: H19/IGF2 (IC1): chr11:2018812-2024740; IGF2:Ex9-DMR (DMR2): chr11:2153991-2155112 and IGF2:alt-TSS-DMR (DMR0): chr11:2168333-2169768. b RNA expression levels of IGF2 and INS-IGF2 by real-time PCR. ΔCt values are calculated as (mean Ct _{IGF2/INS-IGF2} − Ct _GAPDH) and converted to 2^−ΔCt. For tumor tissue from each group, relapse and complete remission patients, 2^−ΔCt values reported relative to normal adjacent tissue. Data are presented as mean ± standard error of mean (SEM) relative expression: **p < 0.01; ***p < 0.005 (by Mann–Whitney U-test). c Illustrative model describing how loss of methylation in IGF2:Ex9-DMR leads to poor response to therapy by increasing INS-IGF2 expression levels

Discussion

Although current WT therapies achieve survival rates of over 85%, long-term survivors of WT are at increased risk of treatment-related morbidity and mortality [1, 33]. Treatment complications include cardiotoxicity from anthracycline treatment, musculoskeletal effects and the development of secondary malignant neoplasms [34]. Survival rates also significantly decrease to around 30% in cases of tumor relapse [1, 2, 35]. Current efforts focus on the early detection of prognostic markers to effectively adjust treatment approaches, minimize treatment doses for low-risk patients and enable early identification of relapse potential to optimize treatment strategies.

In our study, we used methylation microarrays to explore global methylation levels in both tumor tissues and adjacent normal tissues from initially resected kidneys of stages I and II FHWT patients. We report both regions of hypomethylation and regions of focal hypermethylation in tumor tissue compared to adjacent normal tissues. The majority of the hypomethylation is in open sea regions, as well as in intergenic and gene body associated regions, whereas hypermethylation was mostly observed in CpG islands and promoter-associated regions. This is consistent with other studies that report hypomethylation as a hallmark of cancer, particularly in intergenic and open sea regions, leading to the activation of gene expression from repeat regions [36], as well as focal hypermethylation [37], inhibiting the expression of tumor suppressor genes.

The potential of DNA methylation as a prognostic marker has been investigated in numerous studies, particularly due to the early onset of these changes. Research on WT has examined methylation markers across various WT samples [24–26] and showed that most methylation differences are influenced by histopathological subtype and tumor stage. In this study, we specifically focus on low-risk stages I and II WT patients, to identify the earliest methylation biomarkers for relapse. Interestingly, we observe a higher hypomethylation pattern particularly in intergenic open sea regions, as a good prognostic indicator for complete remission following kidney resection and chemotherapy. This is in contrast with several studies that report the association of hypomethylation with more progressive disease [36, 38, 39]. Furthermore, we demonstrate the effectiveness of fourteen identified DMPs as discriminative biomarkers. By using cut-offs from ROC curves, we stratified the samples and found correlations with overall survival and event-free survival. Moreover, our analysis indicates that methylation levels may not only predict the occurrence of disease relapse but also the timing of relapse. Through linear regression, we identified three probes—cg15822093, cg07086381 and cg21987076—that significantly correlate with time to relapse, highlighting their potential as prognostic markers for disease progression. Several studies have shown that DNA methylation can be detected in cell-free circulating DNA (cfDNA) [40–42], highlighting its potential as a valuable biomarker in various cancers, while minimizing the need for invasive tissue biopsies. This suggests that the biomarkers identified in this study could be further evaluated as non-invasive blood-based indicators, enabling real-time monitoring of tumor dynamics and treatment response.

Several of the DMPs identified were located around genes that have reported involvement in carcinogenesis and overall survival. Higher expression of RP11-65D13.1 is observed in squamous cell carcinoma [43], whereas lower expression of AGXT2L1 is observed in digestive cancers [44]. Higher expression of the long non-coding RNA AP000233.4, C12orf42 and PTGS2 are associated with better survival [45–47]. However, since methylation changes across one or two isolated probes may not reflect on overall gene expression levels, we focused our attention on the DMR involving IGF2 gene.

The methylation status of the chromosome 11 region, particularly the IC1 located between the IGF2 and H19 genes, has been extensively studied. In addition to IC1, two other iDMRs are present at the IGF2 locus, IGF2:alt-TSS (DMR0) and IGF2:Ex9-DMR (DMR2). IC1 is imprinted under normal conditions, ensuring the mono-allelic expression of H19 and IGF2 with DMR2 also participating in this regulation. Specifically, DMR2 is believed to facilitate paternal expression of IGF2, through interacting in its methylated paternal form in cis with paternal methylated IC1 activating the expression of IGF2, by placing its promoters in close vicinity to the enhancers downstream of H19 [48]. Loss of imprinting during embryogenesis is observed in syndromic disorders such as BWS and Silver–Russell syndrome (SRS), resulting in over-growth or under-growth disorders, respectively. Loss of imprinting at IC1 is also seen in several cancers such as hepatocellular carcinoma and nephroblastoma, underlying the high prevalence of WT in BWS patients. While IC1 shows gain of methylation in WT of both syndromic and nonsyndromic origins, the status of DMR0 and DMR2 follows a different pattern. DMR0 was shown to be hypomethylated in WT patients, but hypermethylated in BWS patients [49], whereas DMR2 shows gain of methylation in BWS patients with maternal origin gain of methylation at IC1 [50].

Here, we report, as expected, the gain of methylation at the IC1 region in tumor samples from both groups compared to the adjacent normal tissues. In tumor tissues of relapse patients, we also observed loss of methylation of DMR2, which was significantly greater than that seen in complete remission patients, in addition to a loss of methylation at DMR0, which was not observed in complete remission patients. These findings suggest a new role for DMR0 and DMR2 in chemoresistance in WT patients. We investigated the possible transcription effects of these methylation changes, on IGF2 and the typically unexpressed INS-IGF2 fusion transcript. We observed a 20-fold higher expression level of INS-IGF2 transcript in tumor tissues of relapse patients when compared to complete remission patients, in contrast with a modest twofold higher expression level of IGF2. The overexpression of INS-IGF2 has been reported in cancers such as pheochromocytomas [51] and was shown to promote cellular proliferation and migration in lung cancer [52]. The proposed mechanism of action for the INS-IGF2 transcript in promoting cancer progression involves its action as a cis-acting regulation of the IGF2 gene, resulting in its overexpression. However, our findings indicating only modest changes in IGF2 expression suggest that INS-IGF2 might exert its proliferative effects through alternative, unexplored mechanisms. The loss of methylation at DMR0 and DMR2 is a likely reason for the increased expression of the INS-IGF2 fusion transcript. DMR0 is located between the INS and IGF2 genes, and its demethylation could potentially lead to run-off transcription, producing the fusion transcript. Loss of methylation at DMR2, while IC1 remains methylated—unlike the typical pattern in paternal IGF2 expression—may induce abnormal chromatin changes that promote the expression of the INS-IGF2 fusion. Additionally, DMR2 overlaps enhancer regions, and several studies report conflicting theories on the relationship between DNA methylation and enhancer activity or the recognition of enhancers as promoter elements [53–55].

Naturally, this study has inherent limitations that must be acknowledged. Due to the focus on identifying early biomarkers of WT disease progression, the study concentrated primarily on stages I and II FHWT patients. This choice limited the sample size, given the rarity of WT and the low relapse rates among these stages. Additionally, the observed methylation differences may be specific to Egyptian pediatric patients. We also noted significant discrepancies in age and gender distribution between the complete remission and relapse groups, which may impact the interpretation of our findings. Future validation studies, requiring multi-center collaboration, are essential to obtain a larger sample size, address potential confounding factors such as age and gender and confirm the utility of DNA methylation as a prognostic marker. Altogether, despite the small sample size and these limitations, this study provides valuable preliminary insights and lays the groundwork for future research with larger cohorts.

Conclusions

The analysis of DNA methylation patterns and INS-IGF2 expression in early-stage Wilms tumor patients with favorable histology reveals significant potential for these biomarkers in predicting relapse and disease progression. The identification of methylation changes at the chromosome 11 DMR2 and the association with increased INS-IGF2 expression provides mechanistic insights into the molecular mechanisms underlying disease progression. Further validation and exploration of these biomarkers could lead to more targeted and personalized treatment strategies, ultimately improving long-term outcomes for Wilms tumor patients.

Abbreviations

WT

Wilms tumor

DMP

Differentially methylated probe

COG

Children’s Oncology Group

OS

Overall survival

WAGR

Wilms’ tumor, aniridia, genital anomalies and retardation

DDS

Denys–Drash syndrome

LOI

Loss of imprinting

IC1, IC2

Imprinting control regions 1 and 2

BWS

Beckwith–Wiedemann syndrome

MDS

Myelodysplastic syndromes

IRB

Institutional Research Ethics Board

IDAT

Intensity data files

MAF

Minor allele frequency

SNPs

Single-nucleotide polymorphism sites

RT

Relapsed patients-tumor tissue

CRT

Complete remission patients-tumor tissue

RN

Relapse patients-normal tissue

CRN

Complete remission patients-normal tissue

FDR

False discovery rate

ROC

Receiver operating characteristic

DMR

Differentially methylated regions

Ct

Cycle threshold

AUC

Area under curve

EFS

Event-free survival

iDMRs

Imprinting-associated DMRs

SEM

Standard error of mean

SRS

Silver–Russell syndrome

Author contributions

A.A.S. and W.Z. conceived and designed the study. N.E. and A.S.A. revised and provided the samples. D.J. performed the experiments. D.J. and M.Y.A. analyzed the data and interpreted the results. D.J. wrote the main manuscript text. D.J. and M.Y.A. prepared the figures. All authors contributed to manuscript revisions. All authors reviewed and approved the final manuscript.

Funding

Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). This study was funded by the Association of Friends of the National Cancer Institute (AFCNI).

Data availability

All data generated during this study are included in this article and its supplementary files. Raw methylation data were deposited at the Gene Expression Omnibus (GEO) under accession number GSE269241.

Declarations

Competing interests

The authors declare no competing interests.