The animal treatment was carried out according to Italian Legislative Decree No. 26 of 04 March 2014 and performed under the supervision of a local veterinary official. Using an operator-blinded design, the animals were randomly assigned to each experimental condition as previously described (Maqoud et al., 2020). The Organization for Animal Health of the University of Bari on 15 April 2019 (Prot. 34,091-x/10, 06 May 2019) and the Minister of Health, Rome (No. 673/2019-PR Prot. DGSAF0025334-P-04/10/2019, prog. 7307D.11), approved the protocol.

The experiments were performed on male Wistar rats (Charles River S. p.A., Calco, Lecco) with a mean body weight of 360 ± 14 g that were treated with minoxidil solutions and relative vehicles (number of rats = 10). A repeated escalating dose of 0.035% (0.777 mg/kg/day), 0.07% (1.554 mg/kg/day), and 3.5% (w/v-w) (77.7 mg/kg/day) of minoxidil (MXD) ethanol/glycol propylene solution (number of rats = 5) was topically applied to the rats; other rats were treated with vehicles (Maqoud et al., 2020). The cumulative dose of MXD at the end of the treatment period was 71.15 g/kg.

At the end of treatment, the animals were killed by the overdose of the anesthetic agent Zoletil 50/50 (Pfizer) i. p. 60 mg/kg, followed by cervical dislocation, and the tissue samples such as the heart, liver, kidneys, skin, and muscles, but in this work, we investigated the kidney, were immediately preserved in 10% buffered formalin. The EDTA blood and serum were processed. The histological examination was performed for each animal sample ≥ 48 h after fixation. The tissues and organs were embedded in paraffin, cut into 5 μm sections, and stained using standard techniques with hematoxylin and eosin. The histopathology and necropsy analysis included a macroscopic and microscopic examination of the tissue samples and an organ assessment of the preserved kidneys and other tissues. Finally, an examination of the cellular morphology, as well as the relationship between the nucleus, nucleolus, and cytosol was conducted. A histopathological scoring system, based on the semiquantitative ordinal system, was used to evaluate the severity of the observed lesions. The pathologists were blinded to the treatment of the analyzed samples. Images from 10 random fields were acquired for the 10 stained sections of each specimen using a D 4000 Leica DMLS microscope equipped with a camera and image analyzer, NIS-Elements BR (Nikon). The analysis of the sections was performed using Leica QWin software. Organ weight was not measured.

The autopsy of the various organs revealed the congestion of all the parenchyma. The kidneys on external examination showed no modifications, only in three subjects did the cut surface show small pale foci with a maximum size of 0.2 cm in diameter, and the remainder showed no significant alterations. The dissected kidneys were fixed in 10% buffered formalin, dehydrated, and subsequently embedded in paraffin, sectioned to 4 microns, and stained with hematoxylin and eosin (H&E), periodic Schiff acid (PAS), and tricolor by Masson for routine diagnostics. The diagnostic criteria used followed the guidelines published by the Society of Toxicologic Pathology (Hard and Seely, 2005; Hard and Seely, 2006), and the proliferative lesions of the tubules were reported by Hard and Seely (2005, 2006) and Hard et al. (2008).

We performed immunohistochemical staining of the kidney sections with the avidin–biotin streptavidin (LSAB) method using the OMNIS Immunohistochemistry Staining System, Agilent Technologies.

The slides were controlled by protocols using DakoLink software. Tissue sections were cut (4 μm thick), placed on poly-L-lysine-coated slides, xylene de-waxed, and dehydrated. IHC slides were automatically mounted, and coverslips were applied, after staining and dehydration. To detect the antigens, the sections were immersed in citrate buffer (0.1, pH 0.6) for 30 min with 0.3% hydrogen peroxide and then in methanol for 12 min to quench the peroxidase activity. After washing three times for 5 min each with phosphate-buffered saline (PBS), the sections were blocked by soaking them for 20 min at room temperature in PBS containing 1% bovine serum albumin. After rinsing with PBS, the blocked sections were incubated overnight at 48°C, and the sections were stained with polyclonal and/or monoclonal antibodies (Table 1).

Positive controls were inserted for each type of antibody during the immunohistochemical procedure using already tested tissue samples of organs as suggested by the datasheets. For the negative control, the primary antibody was omitted during immunohistochemical staining. The slides were controlled by protocols using the DakoLink software.

The antibodies were visualized using a biotinylated secondary antibody, an avidin–biotin–peroxidase complex, and 3-amino-9-ethyl carbazole as chromogens (Dako, Glostrup, Denmark). Nuclear counterstain was performed with Gill’s hematoxylin (Polysciences, Warrington, PA, United States), and then, the sections were dehydrated with ethanol and xylene before assembly. The sections were initially examined with a magnification of × 200 (i.e., objective × 20 and eyepiece lens × 10; 0.7386 mm² per field) and subsequently to × 400 × fields (i.e., objective × 40 and eyepiece lens × 10; 0.1885 mm² per field). Images of all sections were captured using an HD camera (Nikon Corporation, DS-Fi2 high-definition color camera, Tokyo, Japan) connected to an optical microscope (Nikon Corporation, Nikon Eclipse Ni-U, Tokyo, Japan) with 20x–40x objective; morphometric analyses were performed using an interactive image analysis system based on Arivis Vision4D, modular software for working with multi-channel 2D, 3D, and 4D (Nikon Corporation, NIS-Elements BR, Tokyo, Japan).

Appropriate positive and negative controls were included at each immunohistochiemical (IHC) run. Cancer cells were considered only positive when the appropriate staining patterns for the staining have been achieved. The extent of the immunohistochemical reaction of tumor cells was classified into levels, such as no positive tumor cells (0); positive cells < 10% (1+) corresponding to G1, positive cells 10%–50% (2+) = G2, and positive cells> 50% (3+) = G3 according to the guideline (Zizzo et al., 2019). A defined score from 0 to 1+ was given as negative and scores 2+ and 3+ as positive. Evaluation of the samples was performed independently by three of the authors (NZ, GP, and AT). Positive staining for a given marker was considered only when all the observers agreed on its specificity and distribution.

Additional kidney sections in rats were stained with an argyrophilic technique that assesses the NOR (region of the nucleolar organizer) (Dako, Glostrup, Denmark). Briefly, the sections were de-waxed and incubated with the AgNOR solution in a dark room for 45 min. After washing with distilled water, the sections were incubated in alcohol for 4 min and in xylene for 3 min and then mounted with a mounting medium. The number of nuclear points (AgNOR) from 100 cells was counted using a magnification of 1.000 x (oil immersion). The mean AgNOR score per cell was expressed as the AgNOR score.

A total specimen from 23 female dogs with spontaneous mammary neoplasia has been recruited into the archives of Pathology and Comparative Oncology to the Department of Medicine Veterinary (Bari-Italy) for diagnosis with the written consent of the animal owners and ethical approval from the University of Bari (Zizzo et al., 2019; Tricarico et al., 2022). The mammary glands were removed with radical mastectomy or partial, with regional lymph nodes. Surgical samples fixed in 10% buffered formalin and stained with hematoxylin–eosin were classified histologically following the Goldschmidt classification (Goldschmidt et al., 2011) and Peña for the tumor grading system (Patruno et al., 2009; Zizzo et al., 2010; Peña et al., 2013; Rasotto et al., 2017). The sections were cut (4 μm thick), placed on poly-L-lysine-coated glass slides, and subsequently, deparaffinized in xylene and dehydrated and immunohistochemically stained according to the labeled streptavidin avidin–biotin (LSAB) method (Ioachim, 2008) using antibodies in Table 2.

Analysis of the canine sections was performed as previously described and reported (Ioachim, 2008; Patruno et al., 2009; Zizzo et al., 2010; Goldschmidt et al., 2011; Rasotto et al., 2017; Tricarico et al., 2022).

A large amount of genomic data concerning pathological and polymorphic variants in various tissues in both humans and animals, phenotypic features, bibliographic documents, alternative transcripts, gene expression, and other data derived from omics studies are available through databases implemented on the NCBI Entrez system (https://www.ncbi.nlm.nih.gov/). Currently, we have focused our preliminary research on systematic and integrated navigation of the various bioinformatic resources starting from the Gene database (Brown et al., 2015). Here, detailed analyses were reported for data regarding gene expression and variation.

Moreover, starting from PubMed data reported in the Gene database and through this, in OMIM, we have selected publications where the terms cancer and tumor have been mined.

A protocol allowing the retrieval of information available in the literature concerning the four genes ABCC8, ABCC9, KCNJ11, and KCNJ8 here was described taking as an example the ABCC8 gene. Starting from the NCBI site and browsing the available databases (35 at the time of the research) for searching data regarding ABCC8, non-null results are available in 29 databases. The protocol has been performed to start from the PubMed dataset available through Bibliography/Related articles in PubMed/see all articles (q1 = #1) and from the PubMed OMIM dataset (q1 = #7) available through the Related data from the gene entry. The terms considered are “cancer” and “tumour,” both searched in all fields (queries #2,#3,#8, and #9) and in MesSH Terms (queries #4, #5, #10, and #11). The application of the OR operator for #2, #3, #4, and #5 and for #8, #9, #10, and #11 has produced the lists of our interest derived from q1 (i.e., q6 associated to 21 papers) and q7 (i.e., q12 associated to 22 papers). Finally, applying the OR operator to q6 and q12 has produced the list q13 associated with 38 papers (that were further analyzed). The same protocol was reproduced for the other three genes, ABCC9, ABCC9, KCNJ11, and KCNJ8 genes.

We have focused on the sections Variation, Bibliography, and Expression. The section Bibliography reports links to Related articles in PubMed (the list of papers selected by the curators of the Gene database) and GenRif (short notes reported as communication by registered users). Thus, by clicking on “Bibliography/Related articles in PubMed/see all articles,” the user is automatically migrated from Gene to PubMed where a list of 274 papers is available. However, because the need is to choose papers reporting data regarding oncological studies, the advanced query procedure is activated based on the selection of the term “cancer” as free text or as a mesh term combined with list 1; according to the usage of logical operators, a list of 25 papers has been obtained.

The analysis of gene expression data produced through both microarrays and RNA-seq technology has been performed thanks to the availability at the NCBI of the Human Protein Atlas (HPA) database (Fagerberg et al., 2014) (BioProject PRJEB4337, DOI: 10.1074/mcp.M113.035600) annotating RNA-seq data of 27 different tissues samples from 95 human individuals to determine the tissue specificity of all protein-coding genes. The four genes’ analysis has been also performed by browsing the Expression Atlas database (Papatheodorou et al., 2020) (https://academic.oup.com/nar/article/48/D1/D77/5609521), where data regarding samples recruited within the Pan Cancer project (https://www.embl.de/campaigns/pancancer/) and whose expression has been estimated are annotated and available at the EBI (https://www.ebi.ac.uk).

The availability of data regarding both normal and disease-associated samples allows for estimating the differential expression of the same gene. At this aim, the TNMplot web tools, a web tool for comparing gene expression in normal, tumor, and metastatic tissues, were considered. The TNMplot.com web tools are based on data generated from gene arrays from the Gene Expression Omnibus of the National Center for Biotechnology Information (NCBI-GEO) or RNA-seq from The Cancer Genome Atlas (TCGA), Therapeutically Applicable Research to Generate Effective Treatments (TARGET), and the Genotype-Tissue Expression (GTEx) repositories. Statistical significance was calculated using Mann–Whitney or Kruskal–Wallis tests. The entire database contains 56.938 samples, including 33.520 samples from 3.180 genetic chip-based studies (453 metastatic, 29.376 tumor, and 3.691 normal samples), 11.010 samples from TCGA (394 metastatic, 9.886 tumors, and 730 normal), 1.193 samples from TARGET (1 metastatic, 1.180 tumor and 12 normal), and 11.215 normal samples from GTEx. The tumor tissues of gene expression data that are considered in TNMPlot are adrenal cancer, acute myeloid leukemia (AML), bladder cancer, breast cancer, colon cancer, esophageal carcinoma, liver cancer, lung adenocarcinoma, lung squamous cell carcinoma, ovarian serous cystadenocarcinoma, pancreatic adenocarcinoma, prostate adenocarcinoma, rectum adenocarcinoma, renal clear cell carcinoma, kidney chromophobe, kidney renal papillary cell carcinoma, skin cutaneous melanoma, testicular germ cell tumors, thyroid carcinoma, uterine carcinosarcoma and uterine corpus, and endometrial carcinoma. The statistical significance of the differential expression is calculated by applying a Mann–Whitney U test.

Pan Cancer’s RNA-seq data were analyzed by selecting the section gene expression comparison on paired tumor and adjacent normal tissues. The research was carried out for every single gene in the different tissues through the use of the TNMplot.com web tools.

The TNMplot.com web tools, a web tool for comparing gene expression in normal, tumor, and metastatic tissues, are used. The TNMplot.com web tools are based on data generated from both gene arrays from the NCBI-GEO or RNA-seq from TCGA, TARGET, and the GTEx repositories. The entire database contains 56.938 samples, including 33.520 samples from 3.180 genetic chip-based studies (453 metastatic, 29.376 tumor, and 3.691 normal samples), 11.010 samples from TCGA (394 metastatic, 9,886 tumors, and 730 normal), 1.193 samples from TARGET (1 metastatic, 1.180 tumor and 12 normal), and 11.215 normal samples from GTEx. The research was carried out for every single gene in the different tissues (Le Gallo et al., 2012; Soucek et al., 2015; Cheon et al., 2022; Zhang et al., 2022).

Instead, via kmplot.com/webtools, considering the Pan Cancer projects, the dataset of RNA-seq samples with follow-up comprised 9,663 specimens from 26 distinct tumor types. Across the entire database, the median follow-up for overall survival (OS) was 24.3 months, and for relapse-free survival (RFS), it was 23.8 months. Correlations between gene expression and survival were calculated using Cox proportional hazards regression. The correlation search was performed on the www.kmplot.com database, using the gene name, selecting trichotomization (Q41vs. Q4: lower quartile vs. upper quartile), dividing the search by gender, using the race restriction, and selecting only whites (Győrffy, 2021).

Statistical significance was calculated using Mann–Whitney or Kruskal–Wallis tests.

The correlation between gene expression and overall survival (OS) was made using Cox proportional hazard regression analysis. Package R “survival” v2.38 (http://CRAN.R-project.org/package=survival/) was used to calculate log-rank p-values, hazard ratios (HRs), and confidence intervals at 95% (CI). Furthermore, survival differences were visualized by generating Kaplan–Meier survival plots.

Grouping and hierarchical clustering were applied to visualize the characteristic cancer genes associated with survival in different cancer types using the Genesis software.

All coding preferred terms per S.O.C. were screened in the EudraVigilance database (http://www.adrreports.eu/), and those potentially associated with KATP channel actions were reported and subject to analysis. The data were extracted from EudraVigilance, collected, and analyzed on Excel software (Microsoft 10.00). The data were collected per age and sex. The duplication of a specific report per system organ class (S.O.C.) was evaluated manually and excluded from the analysis (Mele et al., 2014).

The data were reported as an average ± E.S. unless otherwise specified. The significance between data pairs was calculated by paired Student’s t-test for p < 0.05. One-way ANOVA was used to evaluate significance within and between data with variance ratio F > 1 at significant levels of p < 0.05.

The proportional reporting ratio (P.R.R.) was calculated using the equation a/(a+c)/b/(b + d), where a is the reaction of interest to a given drug of interest, b is the reaction of interest for all other drugs in the class, c are all other reactions to a given drug of interest, and d represents all other reactions to all other drugs in the class. Signal definition: P.R.R. ≥ 2, a minimum of three ratios/cases for the reaction of interest, X² ≥ 4. No signal was identified if P.R.R. is = 1.

The IC₅₀ data of different KATP channel blockers were pooled from the literature and were obtained from concentration–response curves of transmembrane KATP channel currents vs. at least five to seven different drug concentrations. The IC₅₀ values were derived from the standard formula and reflect the drug concentration that would inhibit 50% of the KATP channel current when measured in a drug-free solution and expressed as fractional current. The data were obtained in a variety of cell lines and sources for Sur and KATP channel proteins to obtain the IC₅₀ values from different laboratories as previously reported (Abdelmoneim et al., 2012; Maqoud et al., 2016).

Correlations between gene expression and survival were calculated using Cox proportional hazard regression and plotting Kaplan–Meier survival plots. The correlation search was performed on the www.kmplot.com database, using the gene name, selecting trichotomization (Q41 vs. Q4: lower quartile vs. upper quartile), dividing the search by gender, using the race restriction, and selecting only whites. RNA-seq data were reported in RPKM (reads per kilobase million) or FPKM (fragments per kilobase million) and TPM (transcripts per kilobase million).

The long-term topical treatment with high dosing of the minoxidil ethanolic solution induces cell proliferation and cancer in male rats. All five of the rats treated with the MXD ethanol/glycol propylene solution showed lesions in all the organs sampled. In the repeated escalating doses of the MXD (0.035%–3.5% w/v-w) experiment, the rats treated once daily with either of the MXD at concentrations of 0.035% (0.0777 g/kg/day), 0.07% (0.1555 g/kg/day), and 3.5% (w/v-w) (7.777 g/kg/day) for 2 months experienced mild and insignificant reductions in their arterial blood pressure with no change in HR, shown by telemetry monitoring in free-moving rats (Maqoud et al., 2020). The cumulative dose of MXD that the experimental groups received using this protocol was 71.15 g/kg. Renal cancer was not fatal in these rats, and no reduction in body weight or other clinical signs were observed.

The animals were killed after treatment. The postmortem renal biopsy from the MXD-treated animals showed degenerated proximal and distal convoluted tubules, thickened basement membrane, and a clear nucleus with karyomegaly, and prominent or absent nucleoli were indicators of rat tumor renal section. Also, the tumor apical portion of the pyramid showed tubules and papillary ducts with cells arranged randomly, cell pleomorphism, high nucleus/cytoplasm ratio, prominent nucleoli, and discrete mitosis in the renal section following H.E. staining from three rats (Figures 1A, B). All cases are reported in Table 3. On the other hand, for AgNor staining, two to four positive regions of the nucleolar organizer (NOR) granules were identified in the nuclei of the cells of the distal tubules and the ascending flap of the loop of Henle of the kidney in subjects with neoplasia (Figure 1C). Immunohistochemical staining of the kidney section with Ki67 was performed to evaluate the proliferative status of the cells, and we found that 85.2% ± 1% of the cells in the analyzed sections were positive for the antibody reaction following treatment with the MXD ethanol/glycol propylene solution in the higher dosing group. These sections also showed an elevated reaction to CD117 (Figures 2A–C), E-cadherin (Figures 2D–F), and cytokeratin (Figures 2G–I) cells which showed 10%–50% positive cells vs. positive controls. The sections of the tissues were all positives to vimentin, calponin, CD10, CD44, Ki67, and p53 (Figures 3A–F) vs. non-cancerous renal tissues.

The immunohistochemical reaction against the Sur2A subunit (Figure 4C) was markedly elevated in the cytosolic compartment in CD117, E-cadherin, and cytokeratin cells showing 10%–50% positive cells vs. positive controls (Figures 4C, E, F), and a less marked immunohistochemical reaction is observed in the other subunits in the cancerous cells (Figures 4A, B, D). The immunohistochemical panel used allowed us to highlight a different percentage of immunohistochemical reactivity in the various structures of the renal parenchyma in subjects with neoplastic and non-neoplastic lesions (Table 1).

Therefore, high-dose administration of minoxidil induced cell proliferation in renal Ki67-positive cells and the upregulation of the Sur2A located in the cytosolic compartment in renal cancer cells in male WT rats (Figure 4C) vs. controls (Figure 4E).

In a study conducted on samples of spontaneous mammary neoplasia in purebred and mixed-breed dogs aged between 4 and 15 years (average 9.5), not subjected to chemotherapy treatments, from 2015 to 2019 in various veterinary clinics in Bari, Italy, biopsies were collected after tumor removal surgically from female dogs by radical mastectomy or regional mastectomy, with inguinal lymph node removal (Zizzo et al., 2019). Immunohistochemical reaction investigations showed that the Sur2A subunit was markedly elevated in the cytosolic compartment in G3 cells (Figures 5A–C) vs. positive controls, and no changes were observed for G1 or G2 cells; in contrast to the Sur2A subunit for the other subunits, no marked reaction was observed in cancerous cells. The tumor samples (N cases = 23) were classified as the carcinoma samples: tubular, tubule-papillary, and cystic-papillary as previously reported (Zizzo et al., 2019).

Therefore, the cytosolic immunohistochemical reaction against Sur2A h was found in two different cancers in distinct species supporting the involvement of Sur2A in these cancer types. We, therefore, tested the role of the KATP channels in the pharmacovigilance database and omics databases.

Pharmacovigilance data were analyzed to assess the possible correlation between KATP channels and adverse cancer reactions induced by drugs targeting KATP channels. Several cases of cancers were observed with KATP channel blockers used in the treatment of type II diabetes. The most commonly reported were pancreatic carcinoma, malignant neoplasm, and myelodysplastic syndrome followed by bladder cancer and some cases of hepatocellular, breast, prostate, colon, and renal cancers (Table 4).

The sulfonylureas and glinides targeting the Kir6.2-Sur1 subunits showed an elevated case report number for pancreatic carcinoma and bladder cancers, but a low number of reports for common cancers.

The Kir6.1/2-Sur2A/B antagonist zoledronic acid showed instead higher case report numbers for breast > malignant neoplasm > prostate > lung neoplasm malignant > myelodysplastic syndrome > bladder > renal > ovarian > cancers and lower number of cancer cases for other cancer types including pancreatic carcinoma (Table 4).

We, therefore, evaluated the possible relationship between the P.R.R. of the drugs for different cancer types and for hypoglycemia and their capability to block the KATP channel subunits. The P.R.R. for pancreatic cancers of glipizide, gliclazide, and nateglinide appears to be inversely correlated with their IC₅₀ to block the recombinant Kir6.2-Sur1 subunits with the glipizide, gliclazide, and nateglinide showing the highest cancer risk and the lowest hypoglycemia risk (Figure 6) within the KATP channel blockers including zoledronic acid. On the opposite, repaglinide, glibenclamide, and glimepiride showed the lowest P.R.R. for pancreatic cancer and the highest hypoglycemia risk (Figure 6).

Zoledronic acid, a non-selective KATP channel blocker (Maqoud et al., 2021), showed no risk for pancreatic cancer, moderate cancer risk with P.P.R. values ≥ 2 for several cancer types including bladder and renal cancers, but high cancer risk with P.P.R. values > 10 for breast and prostate cancers (Figure 6).

Within the KATP channel openers, minoxidil showed a higher number of A.D.Rs. within the class of breast cancer, skin cancer, and malignant neoplasms while diazoxide did not (Table 4). Due to the low number of events, the P.R.R. of the drug class was calculated aggregated only for the S.O.C. (neoplasm benign, malignant, and unspecified disorders) of the KATP channel openers. The P.R.R. values were 0.42, 1.45, and 1.65 for nicorandil, diazoxide, and minoxidil, respectively. None of these drugs induced a statistically significant cancer risk within the class.

Therefore, the minoxidil-induced upregulation of the KATP channels composed of the Kir6.1/2-Sur2A/B subunits can be associated with high cancer risk in some common cancer including breast cancer in humans but not pancreatic cancer in line with the lack of expression of these subunits in pancreatic beta cells. While the sulfonylureas and the glinides blocking the Kir6.2-Sur1 subunits highly expressed in this tissue show high pancreatic cancer risk inversely related to the hypoglycemia risk. In line with this observation, the Kir6.2-Sur1 opener diazoxide shows no cancer risk including pancreatic cancer.

We, therefore, investigated the available omics data related to KATP channel genes and their involvement in cancers.

The structured databases available at the NCBI report four human genes whose official name is ATP-sensitive potassium channel subunit (KATP)ʼ ABBC8 (Sur1), ABCC9 (Sur2 A/B), KCNJ11 (Kir6.2), and KCNJ8 (Kir6.1) genes. In detail, we focused our attention on data concerning mutations, literature, BioProjects, and gene expression where high-throughput data could open new scenarios regarding the role of these genes in cancer.

The BioProject database collects complete and in-progress large-scale data regarding projects about genome sequencing, expression, metagenomic, annotation, and mapping of bio-data. Hence, BioProject has a central role to reach molecular and literature databases. However, searching for data correlating the gene name and the term cancer yielded data only for the ABCC8 gene. More in one BioProject (PRJNA541505) regarding ABCC8 is in progress, but at present, no relevant data are available. In PRJNA187483: GEO: GSE43807, BioProject aimed to investigate associations between ABC transporter expression and outcome of breast cancer patients by qPCR in post-treatment and non-neoplastic tumor samples from 68 breast cancer patients treated with neoadjuvant chemotherapy. The data produced underlined the significant associations of ABCC1 and ABCC8 levels in tumors with the grade and expression of hormone receptors (Hlaváč et al., 2013).

As far as the Variation section available in the Gene database is concerned, we have selected variants annotated in ClinVar (Landrum et al., 2018) as single-nucleotide missense mutations and validated them as pathogenic/likely pathogenic by multiple submitters. While no missense mutations are reported in KCNJ8, eight, twenty-seven, and five missense mutations related to KCNJ11, ABCC8, and ABCC9, respectively, have been observed in phenotypes prevalently related to well-known diseases such as diabetes mellitus, hyperinsulinemic hypoglycemia, and Brugada syndrome and the recently identified C.S. (Supplementary Table S1). However, no evidence regarding cancers is reported. Hence, we have moved toward an approach of data mining in literature data.

Starting from the scientific publications available in PubMed and selected by the Gene and OMIM (Amberger and Hamosh, 2017) database curators, a mining approach based on the Entrez retrieval system, as described in “Methods—Omics analysis,” has highlighted that besides the well-known role of the channels in diabetes and hyperinsulinemia, differential expression and mutations affecting the four genes are also correlated with some cancer types (Supplementary Table S2).

The differential expression of genes (DEGs) may be the cause or can contribute to cancer development.

Here, we report the expression values of the four genes of our interest by browsing the HPA and Pan Cancer databases. Moreover, a detailed analysis of the four genes of our interest was performed through the Kaplan–Meyer plotter platform (https://kmplot.com/analysis/).

The section “Expression” available through the “Gene database” reports the expression of the gene as measured within the HPA project (BioProject: PRJEB4337, PMID 24309898), where RNA-seq results regarding 27 different tissues from 95 human healthy individuals were produced to estimate tissue specificity of all protein-coding genes. The expression value was quantified in the number of RPKM placed.

As shown in Table 5, the data regarding healthy tissues where the RPKM is higher than 2.0 are reported for the four genes. Brain tissue is the only tissue in healthy subjects that shows significant expression values for all four genes of interest. Tumor samples TPM from Pan Cancer expression values were also reported (Table 5).

The screening of significant DEGs in a specific disease status is of great support to locating genes whose role may support the definition of a therapeutic program. To this aim, the four genes have been analyzed through the TNMPlot web tool (Bartha and Győrffy, 2021).

The ABCC8 genes show a significant downregulation vs. non-cancerous tissues in almost all the common cancers, ABCC9 is also prevalently downregulated except for the kidney renal clear cell carcinoma and pancreatic cancer which showed an upregulation vs. the non-cancerous tissues (red arrow in Figure 7). The KCNJ8 gene was also downregulated vs. controls except for testicular germ cell\tumors and kidney renal clear cell carcinoma that showed an upregulation of the gene expression.

The KCNJ11 gene was markedly upregulated in kidney chromophobe and prostate adenocarcinoma, kidney renal papillary cell carcinoma, breast cancer, uterine carcinosarcoma, and uterine corpus endometrial carcinoma (Figure 7).

The pathogenic and clinical significance of the observed gene expression changes was evaluated. We found that the upregulation of the ABCC8 gene was correlated with overall survival in the lung and pancreatic ductal adenocarcinoma with a low risk of progression in white male populations (H.R. values = < 1). Also, the ABCC9 gene expression was correlated with a low risk of progression in pancreatic ductal adenocarcinoma in white females (H.R. values = < 1) (Table 6).

An opposite trend (H.R. values = > 1) of the ABCC8 gene was observed in the stomach and lung squamous adenocarcinoma in white males and females, respectively. Similarly, the ABCC9 and KCNJ8 genes’ upregulation was associated with a reduced probability of survival in lung squamous cell carcinoma in males, and the ABCC9 gene also in bladder and breast cancers in white female patients. The KCNJ11 gene was significantly associated with negative prognosis in Asian males, black men, and black females in liver hepatocellular carcinoma, kidney renal papillary cell carcinoma, and uterine corpus endometrial carcinoma, respectively (Table 6).

We further tested for a possible association between known gene variants and cancers. The previous meta-analysis indeed shows that KCNJ11 rs5219 is a common variant for type II diabetes (T2D) risk in the world population and has a similar effect on the susceptibility risk to T2D in the Europeans and East Asian, the Japanese (Kuruma et al., 2014). We tested for the known common variants associated with (T2D), cardiovascular diseases, and race in the general European population (rs5219), the French (rs1799859), the Turks (rs1799854 and rs1799859), and the Iranians (rs757110) that is, however, associated with a decreased T2D risk in the British. In non-Caucasian populations, the KATP variants are associated with increased T2D risk in the Japanese (rs5219 and rs757110), the Mongolians (rs1799858 and rs2074308), the Indians (chr11:17417205C > T and rs5210), and the Nigerians (rs1799854) rather than the Punjab population of India (rs1799854 and rs1801261) and the sub-Saharan Africans (rs5219) in Ghana and the Nigerians. The KATP variants are also associated with T2D-related stroke in the Poles (e.g., rs1799854), coronary atherosclerotic heart disease (CAD) in the British (rs757110 and rs61688134), heart failure (HF) in the Ukrainians (rs5210, rs5219, and rs757110) and the Americans (rs5219), and atrial fibrillation (AF) in the Americans (rs72554071) (Liu et al., 2022).

KCNJ11 rs5219 showed no association with pancreatic cancer in the Japanese population (Kuruma et al., 2014) and a weak association with CRC cancer risk (Cheng et al., 2011). ABCC8 rs757110 has been associated with longevity in the Korean population (Park et al., 2009).

However, we currently failed to evidence a significant association of known variants reported in the databases with cancers.