Background
Renal cell carcinoma (RCC) is the third most prevalent urologic malignancy [
1], twice more common in men than in women [
2] disclosing rising incidence (2–4% per year) worldwide. RCC accounts for 2–3% of all malignant tumors in adults, displaying the highest mortality rate among urinary tract cancers [
1,
3]. Notwithstanding, renal cell tumors are morphologically and genetically heterogeneous [
4]. The three main subtypes of RCC - clear cell RCC (ccRCC), papillary RCC and chromophobe RCC [
1] - have distinct clinical behaviors, which should be considered for adequate patient management [
4].
CcRCC is simultaneously the most common and one of the most aggressive RCC subtypes, being prone to local invasion, metastization and death [
2,
5]. It comprises 70–75% of all RCC cases and is characterized by several distinct genetic and epigenetic alterations [
6]. About 25% of ccRCC patients present distant metastases at time of diagnosis, and in 20–50%, metastatic disease develops few years after diagnosis and surgical treatment of the primary tumor [
3,
5,
7,
8]. Furthermore, ccRCC is extremely resistant to radiation and to conventional chemotherapy [
3]. Therefore, biomarkers allowing for earlier diagnosis and accurate prognostication are required, improving current treatment and follow-up strategies [
9]. Indeed, metastatic dissemination is the most important prognostic factor in ccRCC [
10], highlighting the importance of accurately identifying patients at high risk of disease progression. Moreover, the identification of molecular biomarkers that might indicate risk of disease progression (recurrence, metastization) at the time of diagnosis might improve clinical management [
3,
7] effectively contributing to implementation of Precision Medicine [
10].
MicroRNAs (miRs) are small non-coding RNAs, 18–25 nucleotides long, that repress specific genes’ expression by targeting its 3′ untranslated region [
11‐
13]. MiRs’ deregulation has been shown to participate in tumorigenesis, affecting differentiation, invasion, migration and apoptosis [
10,
14] and has been implicated in urological tumors [
15]. Several studies have associated microRNAs (miRs) deregulation with ccRCC clinicopathological features, suggesting a role in tumor initiation and progression [
5,
7,
16,
17]. MiR-30a-5p, an intergenic miR (chromosome 6, 71,403,551–71,403,621 [− strand]), was suggested to play a role in cellular differentiation and development [
18], but its precise role remains largely unknown [
19,
20]. In ccRCC, an onco-suppressor function was proposed for miR-30a-5p, since its downregulation was associated with metastasis development [
5,
9]. Moreover, miR-30a-5p was found to inhibit autophagy, by targeting
BECN1, the gene encoding for beclin-1, a key protein for autophagosome formation [
3]. In addition, miR-30a-5p was shown to decrease tumor microvessel density, by targeting endothelial DLL4, which is enrolled in tumor angiogenesis [
5]. However, the mechanism underlying miR-30a-5p downregulation in ccRCC remains elusive [
21]. Similarly to protein coding genes, miRs’ downregulation might be associated with aberrant promoter methylation, a common feature of urological tumors [
22‐
25]. Thus, we sought to investigate for the first time, whether miR-30a-5p expression is regulated by promoter hypermethylation in ccRCC and evaluate its value as diagnostic and prognostic biomarker, both in tissue and urine samples.
Materials and methods
Patients and sample collection
Independent patient cohorts, two retrospective and one prospective, were selected for this study. Cohort #1 comprises 235 ccRCC patients, consecutively diagnosed and treated with nephrectomy, at Portuguese Oncology Institute of Porto (IPO Porto) between 2000 and 2017. For control purposes, normal kidney tissue from 25 patients subjected to nephrectomy due to upper urinary tract urothelial carcinoma was obtained. Tissue samples from primary tumors and normal kidney were collected immediately after surgery and promptly frozen at − 80 °C. Frozen tissue samples were cut in a cryostat and tumor cell content over 70% was confirmed in two hematoxylin and eosin stained slides taken before and after frozen section collection for nucleic acid extraction. A second cohort composed of 53 ccRCC patients, primarily diagnosed from 2007 to 2013 at IPO Porto, voluntarily provided 50 mL of voided urine samples (Cohort #2 - Testing). For control purposes, urine samples were collected from 57 healthy donors at IPO Porto (2009 to 2010). After collection, urine samples were centrifuged at 4000 rpm for 20 min at 4 °C and washed in PBS 1x. Lastly, pellets were frozen at − 80 °C. A third cohort (Cohort #3 – Validation) comprised 171 ccRCC patients, primarily diagnosed from 2015 to 2018 at Homburg University Hospital (Germany) provided, after informed consent, voided urine samples. For control purposes, urine samples were collected from 85 healthy donors at IPO Porto (2015–2017). After collection, 4 mL of whole urine was transferred into a tube and frozen at − 80 °C, until further usage.
Relevant clinical data was retrieved from clinical charts (Table
1). All procedures performed in studies involving human participants were performed in accordance with the ethical standards of the institutional ethics committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all participants, according institutional regulations. This study was approved by the Institutional Review Board (Comissão de Ética para a Saúde) of IPO Porto, Portugal (CES-518/2010) and Jena University Hospital IRB.
Table 1
Clinicopathological data of tissue and urine samples used in this study
Number of Patients, n | 235 | 25 | 53 | 57 | 171 | 85 |
Median age, years (range) | 65 (32–86) | 71 (52–89) | 61 (38–81) | 49 (41–64 | 66 (36–87) | 55 (45–65 |
ccRCC, n (%) |
ccRCCm | 6 (2.55) | n.a. | 15 (28.3) | n.a. | 11 (6.4) | n.a. |
Non-ccRCCm | 229 (97.45) | n.a. | 38 (71.6) | n.a. | 160 (93.6) | n.a. |
Stage, n (%) |
I | 127 (54.0) | n.a. | 26 (49.1) | n.a. | 121 (70.8) | n.a. |
II | 33 (14.0) | n.a. | 4 (7.5) | n.a. | 8 (4.7) | n.a. |
III | 69 (29.4) | n.a. | 18 (33.9) | n.a. | 31 (18.1) | n.a. |
IV | 6 (2.6) | n.a. | 5 (9.5) | n.a. | 11 (6.4) | n.a. |
Fuhrman Grade, n (%) |
1 | 7 (3.0) | n.a. | 2 (3.8) | n.a. | 8 (4.7) | n.a. |
2 | 99 (42.1) | n.a. | 26 (49.1) | n.a. | 46 (26.9) | n.a. |
3 | 104 (44.3) | n.a. | 18 (33.9) | n.a. | 6 (3.5) | n.a. |
4 | 25 (10.6) | n.a. | 7 (13.2) | n.a. | 2 (1.2) | n.a. |
k.a | n.a. | n.a. | n.a. | n.a. | 109 (63.7) | n.a. |
Follow up |
Median, months (range) | 61 (0–194) | n.a. | 58.00 (2.00–91.00) | n.a. | n.a | n.a. |
Patients without remission (%) | 2 (0.85) | n.a. | 5 (9.4) | n.a. | n.a | n.a. |
Recurrence (%) | 43 (18.3) | n.a. | 10 (18.9) | n.a. | n.a | n.a. |
Death due to ccRCC | 39 (16.6) | n.a. | 10 (18.9) | n.a. | n.a | n.a. |
TCGA data analysis in ccRCC patients
Data on miR-30a-5p expression and methylation from ccRCC tumors and matched normal tissue samples was retrieved from The Cancer Genome Atlas (TCGA) database. MicroRNA-30a-5p expression data from samples hybridized by the University of North Carolina, Lineberger Comprehensive Cancer Center, using Illumina HiSeq 2000 Sequencing system, were downloaded from data matrix including 516 ccRCC samples (
http://tcga-data.nci.nih.gov/tcga/tcgaDownload.jsp). DNA methylation data from miR-30a locus was evaluated using Illumina Infinium Human DNA Methylation 450 array and includes the methylation levels of 319 ccRCC samples. The provided value was pre-processed and normalized according to “level 3” specifications of TCGA (TCGA FPKM-UQ value; see
http://cancergenome.nih.gov/ for details). This data is available for download through the NCI GDC data portal (
https://portal.gdc.cancer.gov/).
DNA was extracted from all clinical samples using phenol-chloroform method. Bisulfite modification was performed using EZ DNA Methylation-Gold™ Kit (Zymo Research, Orange, CA, USA), that integrates DNA denaturation and bisulfite conversion processes into one-step, according to recommended protocol. For urine samples from cohort #3, a pre-amplification step was performed prior to the quantitative methylation-specific PCR. SsoAdvanced™ PreAmp Supermix (Bio-Rad Laboratories Inc., Hercules, CA, USA) was used, following manufacturer recommendations. In brief, 8 μL of DNA template was added to 12 μL of nuclease-free water, 5 μL of Preamplification assay pool and, 25 μL of SsoAdvanced PreAmp Supermix (2x) and pre-amplified for 12 cycles.
Quantitative Methylation-specific PCR (qMSP) assays were carried out in triplicates using Xpert Fast SYBR (Grisp, Porto, Portugal), according to recommended protocol. Sequence-specific primers used in this study were designed to include the two CpGs tested in TCGA database and synthesized by Sigma Aldrich (Sigma-Aldrich, St. Louis, MO, USA) (Supplementary Table
2). Furthermore, the primer’s coverage sites within the methylated gene are available in Supplementary Table
3. For each sample, miR-30a-5p
me status was normalized to the endogenous control β-Actin.
Samples were suspended in TRIzol® reagent (Invitrogen, Carlsbad, CA, USA) and chloroform (Merk Milipore, Burlington, MA, USA) was added after cells were lysed. RNA concentrations and purity ratios were determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). RNA samples were stored at − 80 °C until further usage.
MicroRNAs expression assay
Reverse transcription (RT) was performed using TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) according to manufacturer’s instructions. Quantitative Real-Time PCR (RT-qPCR) was performed in triplicates using TaqMan Small RNA Assays for miR-30a-5p (Assay ID 000417, Thermo Fisher Scientific, Waltham, MA, USA) and Xpert Fast Probe (Grisp, Porto, Portugal), according to recommended protocol. For each sample, miR expression was normalized to endogenous control RNU48 (Assay ID: 001006, Thermo Fisher Scientific, Waltham, MA, USA).
Statistical analysis
Differences in methylation and expression levels and relationships between clinical variables were assessed using Kruskal-Wallis and Mann-Whitney U non-parametric tests for multiple groups (more than two) and pairwise comparisons, respectively. In multiple comparisons, Bonferroni’s correction was applied for pairwise comparisons, dividing the original P-value by the number of groups. P-values were considered statistically significant if inferior to 0.05 for comparisons between two groups.
For miR-30a-5pme, receiver operator characteristics (ROC) curves were constructed by plotting the true positive (sensitivity) against the false-positive (1-specificity) rate, and area under the curve (AUC) was calculated. Specificity, sensitivity, and accuracy were determined. For this, the empirical cut-off obtained by ROC curve analysis [sensitivity + (1-specificity)] was established. This cut-off value combines the maximum sensitivity and specificity, ensuring perfect categorization of the samples as positive and negative for methylation test.
Disease-free survival, disease-specific survival, and metastasis-free survival curves (Kaplan-Meier with log rank test) were constructed considering clinicopathological variables (stage and nuclear grade) and categorized miR-30a-5pme or expression status. A Cox-regression model (multivariable model) was computed considering all significant clinical variables, to assess the relative contribution of each variable to the follow-up status.
Statistical analysis was performed using SPSS 25.0 for Windows (SPSS Inc., Chicago, IL, USA) and graphs were built using GraphPad Prism 6.0 software for Windows (GraphPad Software, San Diego, CA, USA).
Discussion
Over the last decade, the frequency of incidentally detected RCC has significantly increased, mostly due to the widespread use of imaging techniques. CcRCC, the most prevalent RCC, carries worse prognosis than other common RCC subtypes, as approximately 20–40% of cases develop distant metastases [
7], which are the main cause of RCC-related mortality, the highest among urologic cancers [
26,
27]. Thus, biomarkers capable of accurately identifying ccRCC cases prone to metastasize, mostly among early stage tumors at diagnosis, would be a major clinical breakthrough. MiR-30a-5p expression downregulation has been reported in ccRCC, associated with metastatic disease and adverse prognosis. This being said, we aimed to determine whether (a) miR-30a-5p expression silencing was due to aberrant promoter methylation and (b) miR-30a-5p
me levels might not only accurately detect ccRCC in tissue and urine samples, but also identify patients at increased risk to develop metastatic disease independently of standard clinicopathological parameters.
Firstly, TCGA dataset was surveyed and two CpG loci at miR-30a promoter were identified as putative regulators of its expression in ccRCC. Further, miR-30a-5p
me inversely correlated with miR-30a-5p expression levels in ccRCC. Moreover, these results were mostly corroborated in IPO Porto ccRCC cohort #1, also confirming previous reports [
28,
29], and providing compelling evidence that miR-30a-5p downregulation in ccRCC might be caused by aberrant promoter methylation. Thus, our results add miR-30a-5p to the growing list of epigenetically-deregulated microRNAs in urologic malignancies [
22‐
25], reinforcing the contribution of epigenetic alterations to renal carcinogenesis.
Notwithstanding the mechanism underlying miR-30a-5p downregulation in ccRCC, our study firstly demonstrated that miR-30a-5p
me levels might be a specific biomarker for this cancer type. Indeed, since high methylation levels are cancer-specific, they may be used as a tool for ccRCC identification, both in tissue (e.g., as an ancillary tool for histopathological or cytopathological workup of renal mass) and urine samples, providing, in the latter case, a non-invasive tool for early disease detection in high-risk populations [
30,
31] (e.g., patients with end-stage chronic renal disease undergoing haemodialysis). Although other hypermethylated miRs (miR-9, miR-124-3) have been proposed as molecular biomarkers for ccRCC [
32,
33], their performance in urine samples has not been assessed, yet. Thus, to the best of our knowledge, this is the first miR methylation-based urine biomarker to be proposed for ccRCC. Importantly, it should be emphasized that miR
me assessment has several advantages, including higher stability, reduced amount of clinical material requirements and methodological celerity compared to RNA expression assays. Thus, methylation analysis is more robust, enabling the development of tests for use in clinical practice [
34,
35].
The rising incidence of incidentally detected RCC presents a significant clinical challenge owing to the risk of overtreatment. Thus, perfecting prognostic models through the inclusion of molecular biomarkers might contribute to reduce that risk. Remarkably, we demonstrated that higher miR-30a-5pme levels assessed in tissue samples independently predicted shorter time to relapse, showing promise as biomarker for risk-stratification among ccRCC and more accurate identification of the high-risk patient subset which may require alternative therapeutic interventions.
Although miR-30a-5pme biomarker performance in Cohort #3 was not impressive, it should be highlighted that miR-30a-5pme levels were able to identify six out of each ten ccRCC in urine samples and, notably, correctly classified seven out of each ten suspects. This simple and cost-effective method is likely to increase patient compliance while reducing the risk of mistreatment. Moreover, since metastatic patients disclosed significantly higher miR-30a-5pme levels than non-metastatic ccRCC, this non-invasive test might also provide relevant information concerning patient monitoring after curative-intent surgery.
The main limitation of our study is the relatively small number of urine samples tested. Moreover, accuracy might be improved by adding additional markers to the panel, as we previously demonstrated for other urologic cancers [
22‐
25]. Nonetheless, the novelty of using a miRNA methylation marker with diagnostic and prognostic value, amenable for non-invasive detection, constitutes, in our view, a relevant contribution to the field and, hopefully, will stimulate the design of validation studies in larger and independent series.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.