Patterns of miRNA Expression in Primary and Metastatic Wilms Tumor

Introduction

WILMS TUMOR PROGRESSION AND PROGNOSIS:

Wilms tumor (WT; nephroblastoma) accounts for about 85% of pediatric renal tumors [1]. Molecular and histological similarities to the embryonic kidney are notable. Owing to a growing understanding of WT biology and improved treatment options, the 5-year survival rate now exceeds 90% in high-income countries. However, survival remains substantially lower in low- and middle-income regions (often below 50% in parts of Africa), where the incidence is among the highest worldwide [2].

In Europe, the International Society of Pediatric Oncology (SIOP) recommends preoperative chemotherapy followed by surgery, while an upfront nephrectomy is advocated in the United States by the Children’s Oncology Group [3]. Only about 17% of WT cases metastasize, resulting in poorer outcomes [4]. Metastasis mainly occurs in WT with unfavorable histology. Blastemal and diffuse anaplastic WTs are classified as high-risk tumors according to SIOP and are often metastasized [5]. Anaplasia is defined by the presence of polypoid, atypical, mitotic figures, an enlargement in nuclear size, and hyperchromasia of the tumor cells [6]. Lung metastasis is the most common, accounting for about 80% of cases, while metastases to the liver, bone, brain, and lymph nodes are less frequently observed [7]. A better understanding of the molecular background of WT metastasis is essential for accurate risk assessment and follow-up.

ROLE OF MIRCORNA EXPRESSION IN METASTASIS DEVELOPMENT:

MicroRNAs (miRNAs) are a key class of messenger RNA (mRNA) regulators, implicated in primary and metastatic cancer [8], which undergo a multi-step maturation process. They mostly bind to the 3′-untranslated (3′ UTR) region of targets to suppress their expression, but may also interact with other specific regions, including the 5′ UTRs of mRNAs or the promoters and coding sequences of target genes [9]. In this study, our objective was to gain a better insight into the biological background of WT metastasis formation through the analysis of miRNA expression patterns. Promising results were obtained with several miRNAs.

Few papers can be found on the subject and none of them appear to have taken a broad approach, involving only a limited number of miRNAs [10,11]. (These findings are discussed below in the section “WT Metastases and Their miRNA Signatures.”) To our knowledge, our paper is the first to report expression differences across a relatively wide range of miRNAs (84 studied in total, out of which 81 were successfully amplified) between primary and metastatic WTs, although the modest sample size, particularly for metastatic samples (n=6), constrains statistical power and positions the findings as exploratory.

Material and Methods

PATIENTS:

In cooperation with the departments of pathology of 3 Hungarian universities (Semmelweis University, University of Szeged, and University of Pécs), formalin-fixed, paraffin-embedded (FFPE) samples were collected from primary WTs (10), metastatic tumors (6), and healthy kidney tissue obtained from unaffected areas of WT and nephroblastomatosis surgical specimens (10). Overall, 12 patients were included in our study, and in 6 cases we obtained multiple samples. One patient contributed 2 metastatic samples (1 from each lung), while in another patient, 3 histologically different samples, 1 metastatic, and 2 primary tumor samples (1 with stromal and 1 with mixed histology) were studied and compared (Table 1). The 10 primary WT samples were selected on the basis of their frequency of occurrence and metastasis production (4 regressive, 3 mixed, 2 blastemal, and 1 stromal).

In accordance with the SIOP protocol, preoperative chemotherapy was administered using a regimen based on vincristine and actinomycin D, while those presenting with metastatic disease (stage IV) received a more intensive regimen including doxorubicin [4].

RNA EXTRACTION AND REVERSE TRANSCRIPTION:

We followed laboratory protocols as previously published by us and others [12–15]. Here, miRNA extraction was performed using miRNeasy FFPE Kits by Qiagen (Hilden, Germany). After the extraction, RNA concentration was measured by NanoDrop (Thermo Scientific NanoDrop Lite). Concentrations lower than 50 ng/μL were considered inadequate; otherwise, the step was followed by reverse transcription (RT) using miRCURY LNA RT Kits (Qiagen). Instead of 20 ng, as recommended by the manufacturer’s protocol, 200 ng RNA template was included per reaction mix to account for degraded RNA molecules in the size range of miRNA that are detectable in FFPE samples [16]. Total reaction volumes were 20 μL for each polymerase chain reaction (PCR) assay. Finally, cDNA was stored at −20°C until a RT-qPCR was performed.

ANALYZING MIRNA EXPRESSION LEVELS WITH RT-QPCR ARRAYS:

RT-qPCR experiments were performed using miRCURY LNA miRNA Focus PCR Panels (Qiagen) and miRCURY LNA SYBR Green PCR Kits (Qiagen). On the PCR array, 84 individual primer pairs against specific miRNA targets were located in separate wells. These miRNAs are known to play various roles in WTs and in other genitourinary tumors [15]. Primers for small nucleolar RNA, C/D box 44 (SNORD44), small nucleolar RNA, C/D box 38B (SNORD38B), small nucleolar RNA, C/D box 49A (SNORD49A), and U6 small nuclear RNA (U6 snRNA) were also present in the arrays as endogenous controls. Technical replicates were not performed, due to the limited availability of metastatic WT samples and financial constraints associated with the array platform.

Complementary DNAs from control, primary, and metastatic samples were amplified with a Roche LightCycler 96 PCR system (Roche, Basel, Switzerland). The total reaction volume (20 μL) from the RT was used for the experiment. PCR steps included a 2-min heat activation at 95°C followed by a 2-step cycling consisting of a denaturation for 10 s at 95°C and a combined annealing/extension for 60 s at 56°C. The experiment was continued for 45 cycles. Finally, a melting curve analysis (95°C for 60 seconds, 40°C for 60 seconds, then 65°C to 97°C with a ramp of 0.07°C/second) was performed. Cycle threshold (Ct) values over 40 (in all cases) and those between 35 and 40 (in case of a dubious melting curve) were considered negative (unamplified). Data were evaluated with the LightCycler 96 software (Roche, version 1.1).

ΔCt (delta Ct) values were calculated relative to control Ct values defined as the mean of the 4 endogenous controls included in the PCR arrays by the manufacturer, shown to be stable in the tissue context in our previous work [15]. ΔCt values were determined separately in healthy tissue, primary WT, and metastatic samples for each miRNA, and compared using statistical methods as described below.

NETWORK ANALYSIS:

A network analysis was performed for the targets of 5 miRNAs (miR-17-3p, miR-34b-3p, miR-375-3p, miR-616-3p, and miR-34c-5p) using the miRNet web tool (https://www.mirnet.ca/), which integrates predicted and experimentally validated interactions. Results were obtained for all 5 miRNAs except miR-375-3p, for which no interactions were identified; therefore, it was not included in the final network. Network analysis was followed by a functional gene enrichment analysis (KEGG pathways and Reactome database).

The miRNet platform was used for network construction and functional enrichment analysis using its default parameters. Pathway enrichment analysis in miRNet is performed using multiple-testing correction based on the Benjamini-Hochberg false discovery rate. The background gene sets correspond to the genes annotated in the selected pathway databases (Reactome and KEGG) within the miRNet framework.

STATISTICAL ANALYSIS:

Statistical analyses were performed using MedCalc software (version 23.0.9). We compared ΔCt values derived from metastatic samples with the expression results of both the healthy samples and the primary WT samples by 2 separate t tests, followed by a Kruskal-Wallis test to compare all 3 groups. (Due to the relatively small sample size, formal assessments of normality and variance homogeneity have limited statistical power. Therefore, we evaluated results using parametric and nonparametric tests, and prioritized findings that consistently yielded low P values under both approaches.) As this study was designed as an exploratory, hypothesis-generating analysis rather than a confirmatory study with predefined endpoints, no fixed significance threshold was applied and multiple-testing corrections were not used as a decision criterion. Instead, exact P values are reported for all miRNAs to allow transparent interpretation of data.

Using the MedCalc software, log2 fold change (log2FC) values of distinct primary and metastatic samples from the same patients (patients 9 and 12; Table 1) were compared and visualized using multiple line graphs. To evaluate the potential impact of including multiple samples originating from the same patients, we assessed intrapatient similarity using permutation analysis and performed parallel statistical analyses with and without 1 of the 2 same-patient metastatic samples. This approach was used to test the robustness of the observed miRNA expression differences to potential intrapatient correlation (see “Intrapatient Variability Between Primary and Metastatic WT Samples” in the Results).

Results

ALTERATIONS OF MIRNA EXPRESSION:

While 81 miRNAs were successfully amplified in most samples (Tables 2–4), 3 out of the 84 that were present on the array (hsa-miR-3163, hsa-miR-3662, and hsa-miR-3666) were excluded from further analysis due to a lack of amplification in any of the metastatic samples, as well as in most healthy tissue and primary tumor samples. Five miRNAs (miR-17-3p, miR-34b-3p, miR-34c-5p, miR-375-3p, and miR-616-3p) produced low P values both in the 2 t tests (ranging between P<0.0001 and 0.0258) and in the Kruskal-Wallis test (between P=0.0061 and P=0.0322) with all 6 metastatic samples included (Table 5). Interestingly, expression levels of all 5 miRNAs were higher in metastatic samples than in primary WTs and control samples (Table 3; Figures 1, 2). The identified miRNA deregulations (Figure 3) are explored further in the Discussion.

NETWORK ANALYSIS:

The results of our network analyses are shown in Figure 4. We identified 43 genes that are regulated by at least 2 of the studied miRNAs. Among these, polypeptide N-acetylgalactosaminyltransferase 7 (GALNT7), methylenetetrahydrofolate dehydrogenase (NADP+-dependent) 1 (MTHFD1), Scm-related gene containing four Mbt domains 2 (SFMBT2) and cyclin-dependent kinase 4 (CDK4) seem the most relevant based on degree and betweenness values. On the same basis, the role of miR-34b-3p seems to be central within the network.

We note that our network analysis was based on predicted interactions. However, GALNT7 has been experimentally validated as a target of miR-17-3p and miR-34c-5p [17,18]. The interaction of CDK4 with miR-34b-3p and miR-34c-5p has also been validated [19].

A functional gene enrichment analysis was also performed based on identified miRNA-mRNA interactions. According to the Reactome database, these miRNAs regulate the expression of several proteins involved in the regulation of cell death by influencing TP53 and NOTCH signaling as well as the cell cycle. KEGG pathway analysis showed strong enrichment in “Pathways in cancer” and in specific cancer types, including genitourinary neoplasms (Figure 4).

INTRAPATIENT VARIABILITY BETWEEN PRIMARY AND METASTATIC WT SAMPLES:

In cases in which multiple metastatic or primary samples were available from the same patient, individual miRNA profiles were arranged from left to right in order of increasing similarity of expression (Figure 5). The 2 metastatic samples from patient 9 showed remarkably similar deregulation patterns, with some miRNAs differing in expression levels but maintaining the same direction of change. Primary samples of different histological types in patient 12 displayed greater variation, both between each other and in comparison with the metastatic sample.

To assess whether the 10 primary tumor samples can be treated as independent observations or whether same-patient origin should be accounted for, we performed a permutation analysis based on pairwise differences in ΔCt values. For each miRNA, we computed the absolute difference in ΔCt for all possible pairs of samples. The permutation P value was defined as the probability that a randomly selected pair would exhibit an absolute ΔCt difference less than or equal to that observed between the 2 primary tumor samples from patient 12 (all permutation P values are reported in Table 2). The median permutation P value across miRNAs was 0.47, indicating that the 2 samples from patient 12 were not more similar to each other than expected by chance relative to other sample pairs.

The 2 metastatic samples from patient 9 exhibited highly concordant miRNA deregulation patterns across the panel, making additional similarity testing uninformative. To address potential intra-patient correlation, we therefore repeated all statistical analyses using a patient-level dataset in which each patient contributed a single metastatic lesion (n=5), omitting the chronologically second metastatic sample from patient 9. The direction and overall pattern of miRNA deregulations were unchanged, and median P values across the 81 miRNAs were lower in the patient-level analysis than in the sample-level analysis (Table 5). These findings indicate that the observed differences are not driven by the inclusion of 2 metastatic lesions from a single patient. Given that the sample-level analysis (n=6 metastatic lesions) yielded higher P values despite the larger sample size, it was considered the more statistically conservative approach and was retained as the primary analysis underlying the reported findings, while patient-level results are presented in parallel for completeness.

Discussion

WT METASTASES AND THEIR MIRNA SIGNATURES:

At diagnosis, approximately 17% of patients have indications of hematogenous metastasis, and most of them involve the lungs. Less typical locations include liver and bones. High-risk tumors and relapses of metastatic disease have a poorer prognosis [20]. With a combination of chemotherapy (adjuvant and/or neoadjuvant), surgery, and radiation therapy, an overall survival exceeding 90% has been achieved for localized WT. Overall survival for metastasized WT is about 75%, while in the case of recurrence, it is 50% [21]. The molecular background distinguishing primary and metastatic tumors could reveal deregulated processes that may help clinicians identify biologically aggressive cases even before a metastasis develops. A well-established miRNA profile may prove to be an important piece to that puzzle.

Expression of miRNAs has already been reported in progressing and metastatic WT, but evidence is scarce. Peréz-Linares et al found differential expressions for several miRNAs applying TaqMan low-density array followed by validation cohort trials (RT-qPCR and in situ hybridization). miR-29a had low expression in both FHWT (favorable histology Wilms’ tumor) and metastatic disease. Expression differences between the metastatic and control groups were more significant. They also reported a lower expression of miR-200b-3p in metastasis compared with the control group [11]. In the present research, this miRNA was found to be similarly expressed in metastatic and primary tumors (Tables 2, 3), but results reported by Peréz-Linares et al may be more informative given their larger sample size (n=21 metastatic samples compared with our n=6).

Zhang et al noted the downregulation of miR-203-5p in WT tissue [22]. We found the same alteration in our own dataset in primary but not in metastatic WTs (Tables 2–4). However, another group suggested a close correlation between miR-203-5p levels and the presence of lymphatic metastases, concluding that miR-203a-5p may function as a tumor-suppressor in the development of WT [10].

OUR FINDINGS IN LITERATURE CONTEXT:

Among the 81 miRNAs profiled in this study, several exhibited distinct expression differences between primary and metastatic samples (Tables 2, 3). In the following discussion, we focus on miRNAs with the strongest statistical support: miR-17-3p, miR-34b-3p, miR-34c-5p, and miR-375-3p. We also include miR-616-3p, which yielded low P values in parametric and nonparametric tests (Figure 3); however, interpretation is limited by the low number of successful amplifications, as explained below (Tables 2–4). As this study is cross-sectional in design, causal relationships with metastasis cannot be inferred, but observed expression differences may still serve as a basis for hypothesis generation.

The role of miR-17-3p in invasion, tumor proliferation, and metastasis was reported in several cancer types. Liu et al found an overexpression of miR-17-3p in WT tissues compared with healthy adult kidney samples [23]. miR-17-3p is a member of OncomiR-1, an oncogenic cluster of miRNAs. Kort et al profiled the expressions of several Oncomir-1 members (miR-17-3p was not included, but miR-17-5p was), and found them to be overexpressed in WT compared with other kidney tumor types [24]. miR-17 was also significantly deregulated in the blood samples of patients with WT before therapy compared with normal controls [25]. In prostate cancer development, the circSMARCA5-miR-181b-5p/miR-17-3p-tissue inhibitor of metalloproteinase 3 (TIMP3) axis is crucial. Both miRNAs may attenuate circSMARCA5 function, affecting tumor migration and invasion [26]. miR-17-3p may also promote tumor growth in prostate cancer by repressing TIMP3 [27]. In our dataset, consistent with findings in other cancer types, miR-17 appears to be highly expressed in metastatic WT (displaying a log2FC value of 3.1 between metastatic and control, and 2.64 between metastatic and primary tumor samples in our study).

In the case of miR-34b-3p, minimal or no expression occurred in control and primary samples, while metastatic samples showed a moderate expression (log2FC: 4.17 metastatic vs control; 4.36 metastatic vs primary). miR-34c-5p showed a similar pattern (log2FC: 4.34 metastatic vs control; 3.58 metastatic vs primary). miR-34b-3p engages in the network of the TP53 tumor suppressor and may sabotage the p53-DAPK axis by dysregulating essential proteins such as BCL2 apoptosis regulator or cellular MYC [28]. Huang et al found it to be downregulated in the serum of patients with renal cell carcinoma, while upregulation was observed in the renal cell carcinoma tissue [28,29]. Still, the miR-34 family is predominantly recognized as a tumor suppressor in malignancies of the genitourinary system (and other tissues), with oncogenic roles barely reported [30]. Although our data are cross-sectional, the observed expression differences raise the possibility that such a role may exist in the context of WT metastases. In addition to miR-34b-3p and miR-34c-5p discussed above, miR-34a expression levels were also higher in our dataset. However, this pattern was observed similarly in primary and metastatic disease, and the corresponding P value (0.23) does not provide strong statistical support for the observation. Nevertheless, the consistency of our data across 3 different members of the miRNA family, all pointing toward the same interpretation, is intriguing. While miR-34 family members have rarely been studied in WT before, Liu et al investigated the expression of miR-34a-5p and, like us, reported an upregulation (with a fold change of 3.23 compared with adult normal kidney and 4.34 compared with fetal kidney samples) [23]. Similarly, Ludwig et al observed an increase in the expression of miR-34a (fold change: 5.47) and miR-34b (fold change: 2.26) in WT tissue [31]. In light of our findings within the context of existing literature, we propose that the miR-34 family may function differently in WT than in most other cancer types. We encourage further research to explore this possibility in greater depth.

In our sample pool, miR-375-3p showed a consistently higher expression in metastatic samples compared with control and primary tumor samples (log2FC: 4.2 metastatic vs control; 4.73 metastatic vs primary). To our knowledge, no other authors have studied this miRNA in WT yet, but in colorectal cancer, its expression has been reported to both increase [32] and decrease [33] in association with metastasis. This dual role may be explained by the miRNA’s ability to inhibit autophagy [34]. Autophagy is known to promote most cancers, but blocking it may play a paradoxical role in metastasis, facilitating the dissemination of tumor cells into large, advanced macrometastases [35]. In advanced cancer, an increase in miR-375 expression may also contribute to the tumor’s escape from immune surveillance [32].

The most uncertain results were obtained with miR-616-3p due to overall high Ct values and expression being detectable in only 3 out of 6 metastatic samples (Tables 2–4). Acknowledging the limitations of interpretation, we tentatively suggest that its expression may be higher in metastasis than in both the control and primary samples (log2FC: 2.63 metastatic vs control; 3.69 metastatic vs primary). The role of this miRNA has been reported in several tumor types but not in WT. An upregulation of miR-616-3p has been observed in prostatic and hepatocellular carcinoma. miR-616-3p overexpression was also observed in human gastric cancer, with a suggested role in angiogenesis and epithelial-mesenchymal transition [36]. miR-616 acts as an oncogene in glioma, non-small cell lung cancer, and breast cancer. Promotion of migration, invasion, and proliferation in breast cancer by directly targeting TIMP2 has been demonstrated by functional in vitro experiments [37].

Conclusions

To gain a better understanding of WT pathogenesis and metastasis development, an analysis of miRNA deregulation may offer a relatively accessible key as miRNAs can be studied using FFPE samples (typically available in large numbers) and play important roles in suppressing gene expression. Such contributions may be integrated into the broader understanding of disease mechanisms.

In the context of limited available literature, our findings highlight a set of miRNAs that may be of interest in metastatic WT. These observations are exploratory and warrant further investigation to clarify their relevance and possible involvement in WT pathogenesis.