Association of Inherited Copy Number Variation in ADAM3A and ADAM5 Pseudogenes with Oropharynx Cancer Risk and Outcome

Inherited copy number variations (CNVs) can provide valuable information for cancer susceptibility and prognosis. However, their association with oropharynx squamous cell carcinoma (OPSCC) is still poorly studied. Using microarrays analysis, we identified three inherited CNVs associated with OPSCC risk, of which one was validated in 152 OPSCC patients and 155 controls and related to pseudogene-microRNA-mRNA interaction. Individuals with three or more copies of ADAM3A and ADAM5 pseudogenes (8p11.22 chromosome region) were under 6.49-fold increased risk of OPSCC. ADAM5 shared a highly homologous sequence with the ADAM9 3′-UTR, predicted to be a binding site for miR-122b-5p. Individuals carrying more than three copies of ADAM3A and ADAM5 presented higher ADAM9 expression levels. Moreover, patients with total deletion or one copy of pseudogenes and with higher expression of miR-122b-5p presented worse prognoses. Our data suggest, for the first time, that ADAM3A and ADAM5 pseudogene-inherited CNV could modulate OPSCC occurrence and prognosis, possibly through the interaction of ADAM5 pseudogene transcript, miR-122b-5p, and ADAM9.


Introduction
Oropharyngeal (OP) squamous cell carcinoma (SCC) is a subtype of head and neck (HN) SCC, and its main risk factors are tobacco and alcohol consumption [1]. Human papillomavirus (HPV) infection is also a consistent risk factor, and it seems to be a favorable prognostic factor for these patients [2]. Most OPSCC are diagnosed as an advanced disease [3][4][5], and cisplatin (CDDP), whether associated or not with radiotherapy (RT), has been used in HNSCC patients' treatment [6].
Copy number variations (CNVs) are defined as the loss or gain of a DNA segment of one kilobase (kb) or larger that is present at a variable copy number in comparison with a reference genome assembly [7]. CNVs have been broadly documented in tumor (somatic CNVs) [8] and germline (inherited CNVs) cells, and they represent a critical factor driving OPSCC development and outcome [9,10].
Inherited CNVs can provide valuable information for cancer risk prediction and prognosis evaluation, and they have been identified in several cancers [8,16,17]. In HNSCC, the nasopharyngeal SCC is the only subtype which has been studied so far, whereas the amplification of the 1q32 region containing the MAPKAPK2 oncogene [18]; the 11q14.3 region, containing the glutamatergic signaling GRM5 [19]; and the deletion of the 6p21.3 region, containing the MICA and HCP5 immune genes [20] were associated with nasopharyngeal SCC susceptibility. It is interesting to note that several genes affected by inherited CNVs have already been implicated in carcinogenesis themselves, indicating that these CNVs are related to cancer predisposition [8].
For a long time, pseudogenes were considered non-functional ("junk") regions of the DNA. However, thanks to advancements in genomics research, they are now recognized to regulate gene expression [21]. Pseudogene-derived transcripts, containing microRNA(miRNA)-binding elements, can exert the regulatory control of ancestral or other genes' expression levels by competing for miRNAs, acting as critical modulators in cancer development and progression [21,22].
Particularly in HNSCC, loss of copy number of the pseudogene PTENP1 (9p13.3 region) inhibited cell proliferation and tumor cell migration, possibly by acting as an miRNA competitor [23]. The increased expression of the pseudogene-derived long noncoding(lnc)-RNA FTH1P facilitated oral SCC progression by acting as a molecular sponge of miR-224-5p in order to modulate FZD5 oncogene expression [24] and by enhancing PI3K/Akt/GSK3b/Wnt/β-catenin signaling, which are involved in tumor cell migration and invasion [25]. The increased expression of FTH1P also promoted cell proliferation, migration, and invasion in laryngeal SCC tissues [26]. Moreover, the amplification of the 8p11.23-22 region, which contains the pseudogene ADAM5, was described in oral tumor samples [27]. The ADAM gene family encodes the proteins enrolled in cell adhesion, migration, and proliferation, and has been associated with carcinogenesis [28].
Since inherited CNVs in genes with importance in OPSCC are still poorly identified, we conducted a two-stage association study on patients with OPSCC and healthy controls from the southeastern region of Brazil and identified an inherited CNV in ADAM3A and ADAM5 pseudogenes-associated with risk of OPSCC and prognosis.

Study Population
This study was conducted in two steps. Duriong step 1 (pilot-study), 49 patients and 49 controls were analyzed with the purpose of identifying CNVs with importance in OPSCC development, and in step 2, three CNVs were selected for data validation in 152 OPSCC patients and 155 controls.
More details about the dataset of the cohort of OPSCC patients and the controls can be found in Carron et al. [5]. The study was approved in keeping with the institutional review board guidelines (numbers: 424/2016 and 1,438,601) and according to the Declaration of Helsinki. All patients and controls provided written informed consent to participate in this study.
For survival analysis, we selected 139 patients; 13 out of 152 patients were sent to other services for treatment and follow-up, and no consistent clinical information could be retrieved. Patients were treated according to the institutional protocol, based on conventional procedures for CDDP, RT, and surgery [5]. The follow-up of patients was performed at 3-month intervals. The end of the follow-up period was November 2022. Ninetyone patients (67.0%) were submitted to chemotherapy (CT) and RT-combined treatment; 30 patients (22.0%) received CT, RT, and were submitted to surgery; 6 patients (4.4%) received RT and surgery; 4 patients (2.9%) received CT only; 2 patients (1.5%) received RT only, and 3 patients (2.2%) were submitted to surgery only. It was not possible to obtain consistent information about 3 patients' therapeutics enrolled in survival analysis (n = 139).

2.2.
Step 1: Screening of CNVs, Candidate DNA Region Choice, and CNVs Selection Ninety-eight individuals were studied in the first step. Genomic DNA from the peripheral blood of 49 base-of-tongue (BT) SCC patients and 49 controls was genotyped individually using DNA high-resolution microarrays containing 420,000 CNVs (Genome-Wide Human SNP Array 5.0, Affymetrix, Santa Clara, CA, USA) as detailed in Carron et al. [5].
The PennCNV software package was used to call CNVs based on the intensity data from the microarrays, including the log R ratio and the B allele frequency for each CNV [29]. The software BEDTools for genomic arithmetic integrated the CNV data from the sample analysis, merging the CNV regions that were common among them [30]. CNV calls were grouped into loci at least one kb in length. In order to select a CNV for further validation, a confidence score of 10 probes minimum was used as a threshold to classify reliable CNV calls [31], and, in order to restrict the analysis, at least 5 patients had to be identified with the CNV (copy number gain or loss). Next, the CNVs were selected by the presence of carcinogenesis genes in the CNV region, the process being performed using the Database for Annotation, Visualization, and Integrated Discovery (david.ncifcrf.gov/ accessed on 16 February 2021)) and the Kyoto Encyclopedia of Genes and Genomes pathway maps (https://www.genome.jp/kegg/ accessed on 16 February 2021).

Step 2: Validation of CNVs in OPSCC Risk
A total of 307 individuals were analyzed in Step 2, including the 98 individuals analyzed preliminary in Step 1. Genomic DNA from leukocytes of the peripheral blood of 152 OPSCC patients and 155 controls was analyzed through real-time polymerase chain reaction (PCR) using TaqMan Copy Number Assays (assay reference: Hs04875988_cn, Hs03274455_cn, and Hs04494662_cn for the 4q13.2, 8p11.22, and 20p13 chromosome regions, respectively; Applied Biosystems, Waltham, MA, USA), following the manufacturer's instructions. For the internal control, a predesigned TaqMan Copy Number Reference Assay RNase P (assay reference: 4403328; Applied Biosystems, USA) was used. Positive and negative controls were used in all reactions. The qPCR analysis was carried out using the 2 −∆∆Ct cycle threshold method [32], and the copy number was calculated using the Copy Caller™ Software v2.0 (Applied Biosystems, USA). A confidence level of 95% and a |z-score| value of less than 1.75 was applied in order to cull the CNVs. Copy number loss was defined as being equal or less than 1 copy per cell and copy number gain as equal or more than 3 copies per cell. Values of 20% of the samples were repeated in separate experiments with 100% agreement. As a result, ADAM3A and ADAM5 CNV were selected for further analysis.

Potential Pseudogene-miRNA-mRNA Interaction
After CNV validation, we searched for a potential pseudogene-miRNA-mRNA interaction in order to test the hypothesis that pseudogene-derived transcripts could compete for miRNA binding, modulating another type of gene expression [21]. First, we searched for any ADAM family genes sharing a highly homologous sequence to ADAM3A or ADAM5 pseudogenes, especially in the 3 -untranslated region (3 -UTR), using the Basic Local Align-ment Search Tool (blast.ncbi.nlm.nih.gov). As a result, the ADAM9 gene shared a highly homologous sequence with the ADAM5 pseudogene. Next, miRWalk 2.0 [33] was employed to identify potential miRNAs binding to the 3 -UTR of the ADAM9 gene. We selected the miRNAs that match at least seven nucleotides in the seed region (position 2-8 of miRNA). As a result, we selected the miR-122b-5p.

ADAM9 Expression by qPCR
In order to assess the potential pseudogene-miRNA-mRNA interaction, we analyzed ADAM9 gene expression. The total RNA from leukocytes of the peripheral blood of 26 OP-SCC patients with a distinct copy number of the ADAM3A and ADAM5 pseudogenes (6 patients with copy number loss, 15 with two copies, and 5 with copy number gain) was extracted with TRIzol reagent (Life Technologies, Carlsbad, CA, USA), following the manufacturer's instructions. cDNA was generated using SuperScript III reagents (Life Technologies, USA). Experiments were performed using SYBR Green PCR Master Mix reagents (Applied Biosystems, USA) and specific primers for the ADAM9 gene (forward: 5 -GAATTCAGGAGCATCGGGTTC-3 , and reverse: 5 -CCCAGCGTCCACCAACTTAT-3 ), in triplicate per sample, and a negative control without template was included in each plate. The relative expression level of ADAM9 was normalized to the reference housekeeping gene actin β level (forward: 5 -AGGCCAACCGCGAGAAG-3 , and reverse: 5 -ACAGCCTGGATAGCAACGTACA-3 ) using the 2 −∆∆Ct cycle threshold method [32]. Values of 20% of the samples were repeated in separate experiments with 100% agreement. Results were expressed in arbitrary units (AUs).

Statistical Analysis
The association between disease statuses, BTSCC patients vs. controls, and CNVs for Step 1 was performed using the software ParseCNV, which uses a genome-wide segmentbased scoring approach [31]. Afterward, Fisher's exact test was used to compare the CNV frequency among cases and controls.
Receiver-operating characteristic (ROC) curve analysis was used to determine the optimal cut-off score for the ADAM3A and ADAM5 pseudogenes' copy number alterations (scores) to survival status using the 0, 1 criterion [34]. The sensitivity and specificity were plotted in order to generate the area under the ROC curve (AUC) against the survival status (alive vs. progression, relapse, or death by disease).
The differences between groups were analyzed by χ 2 or Fisher's exact test. Multivariate analysis using the logistic regression model served to obtain age-and tobacco status-adjusted crude odds ratios (ORs) with a 95% confidence interval (CI), and to assess the associations between CNVs and OPSCC in step 2.
Either Chi-square or Fisher's exact test was used to evaluate the possible associations between clinical characteristics, tumor aspects, and the selected CNVs. To evaluate the robustness of our results, the false discovery rate was computed to reflect the expected ratio of false-positive findings to the total number of significant findings.
Considering continuous variables, data sets were probed for normality using Shapiro-Wilk's test. For ADAM9 gene expression, the data set assumed normal distribution, and a t-test was used for analysis.
For survival analysis, the event-free survival (EFS) was calculated from the date of diagnosis to the date of progression, relapse, death caused by the disease, or last followup. OS was calculated from the date of diagnosis until the date of death, resulting in any cause, or last follow-up. EFS and OS times were calculated using Kaplan-Meier estimate probabilities, and the differences between survival curves were analyzed by the log-rank test. The prognostic impacts of clinicopathological aspects and each genotype in the survival of OPSCC patients were examined using Cox proportional hazard ratio (HR) regression. In a second step, all variables with p < 0.10 were included in a multivariate Cox regression (backward conditional step-wise selection). In the multivariate Cox regression, p values < 0.05 were considered statistically significant.
As a complementary survival analysis, we assessed the Kaplan-Meier Plotter analysis (kmplot.com/analysis/ (accessed on 22 March 2021), an online database established from the gene expression data and survival information of cancer patients from the Gene Expression Omnibus (GEO) database, in order to evaluate the prognostic value of ADAM3A and ADAM5 and miR-122b-5p in general HNSCC patients, since it was not possible to stratify for OPSCC [35]. A Kaplan-Meier survival plot and log-rank p were directly determined and displayed by the database.
For all statistical tests, the significance is two-sided and achieved when p values were < 0.05. All tests were conducted using the SPSS 21.0 software (SPSS Incorporation, Chicago, IL, USA).

Study Population
The sociodemographic data of 152 OPSCC and 155 controls are presented in Supplementary Table S1. The control individuals were younger than OPSCC patients (median age: 46 vs. 56 years, p < 0.001), and the number of tobacco and alcohol users was higher among patients than in controls (98.0% vs. 26.5%, p < 0.001; 92.8% vs. 63.9%, p < 0.001; respectively). The differences in the age and pattern of tobacco and alcohol use of individuals from each group were corrected in all comparisons of genotype frequencies by pertinent statistical analyses.
The frequencies of tumor characteristics of OPSCC patients are shown in Supplementary Table S2. HPV 16 was positive in only 1 out of 42 OPSCC patients analyzed in the study. Most patients presented tumor stage IV (67.1%), moderately differentiated tumors (78.3%), and tumors located in the BT (47.4%).

Step 1: Analysis of Screening and Selection of CNVs
After screening of CNVs (49 BTSCC patients vs. 49 controls), we observed 1666 copy number gain and 949 copy number loss having at least 1 kb in length associated with BTSCC risk (Supplementary Dataset), but only eight CNVs were identified in at least five patients (Supplementary Table S3). Among them, the GSTT1 gene, enrolled in carcinogen detoxification, is well known to play a role in HNSCC, so it was excluded from the validation selection [36]. Next, we selected three CNVs that harbor genes related to carcinogenesis pathways for validation. These CNVs regions included: the UGT2B17 gene (4q13. CNVs were dichotomized as total deletion, 1 or 2 copies vs. 3 or more copies to compare scores of CNVs, case and control, and clinical and pathological data. Analyzing the ROC curve, the cut-off value one for CNV (AUC = 0.335, p = 0.001, sensitivity = 64.9%, and specificity = 85.5%) was also included in the analyses.

Step 2: Analysis of Validation of CNVs
Three or more copies of the ADAM3A and ADAM5 pseudogenes were detected more frequently in OPSCC patients than in controls (16.4% vs. 4.5%, p = 0.002). Individuals carrying three or more copies of ADAM3A and ADAM5 were under 6.49-fold increased risk of OPSCC than carriers of two copies or fewer (Table 1). No associations between UGT2B17 and SIRPB1 CNVs were seen during the validation of these CNVs (Table 1). Subsequent analyses were performed only for ADAM3A and ADAM5 CNV due to the limited individual carriers of UGT2B17 and SIRPB1 with a copy number different than two.

Association between Clinical and Tumor Aspects with CNVs
No associations between ADAM3A and ADAM5 CNVs were seen in terms of clinical or tumor aspects ( Table 2). CNV: copy number variation; n: number of cases; others: tonsillar complex and soft palate. * The numbers of patients were not the same included in the study (n = 152), because no consistent information could be obtained from some individuals. ** p = 0.16 after false discovery rate adjustment.

Potential Pseudogene-miRNA-mRNA Interaction
We explored the potential regulatory mechanisms of pseudogenes ADAM3A and ADAM5 among which the hypothesis of the pseudogene-derived transcripts could compete with another ADAM-family gene for miRNA binding may be a representative approach. Among the ADAMs genes, ADAM9 and ADAM33 presented a highly homologous sequence at the 3 -UTR with the ADAM5 pseudogene ( Figure 1A). We selected ADAM9 for further validation and miRNAs search due to its potential association with ADAM5, as previously reported [26]. We identified the miR-122b-5p as a potential miRNA for ADAM9 and ADAM5 binding ( Figure 1A).

Potential Pseudogene-miRNA-mRNA Interaction
We explored the potential regulatory mechanisms of pseudogenes ADAM3A and ADAM5among which the hypothesis of the pseudogene-derived transcripts could compete with another ADAM-family gene for miRNA binding may be a representative approach. Among the ADAMs genes, ADAM9 and ADAM33 presented a highly homologous sequence at the 3′-UTR with the ADAM5 pseudogene ( Figure 1A). We selected ADAM9 for further validation and miRNAs search due to its potential association with ADAM5, as previously reported [26]. We identified the miR-122b-5p as a potential miRNA for ADAM9 and ADAM5 binding ( Figure 1A).  pseudogene-derived transcript and at ADAM9 3 -unstranslated region (7mer-m8 site). The miRNA "seed" region is presented in underlined font. The homologous sequence between ADAM5 and ADAM9 is represented in bold letters. (B) ADAM3A and ADAM5 pseudogenes copy number variation and ADAM9 expression in oropharyngeal squamous cell carcinoma (OPSCC) patients. The mean mRNA expression level was higher in OPSCC patients with copy number gain of the ADAM3A and ADAM5 pseudogenes (n = 5) (2.04 arbitrary units (AUs) ± standard deviation (SD): 0.89) than carriers of total deletion, one, or two copies (n = 21) (1.

ADAM9 Expression by qPCR
Higher ADAM9 mRNA expression levels were seen in OPSCC patients with three or more copies of the ADAM3A and ADAM5 pseudogenes (n = 5) than carriers with total deletion or one or two copies (n = 21) (2.04 AUs ± standard deviation (SD): 0.89 vs. 1.12 AUs ± SD: 0.67, p = 0.02) ( Figure 1B).

Survival Analysis of OPSCC Patients
The median follow-up of the 139 patients enrolled in survival analysis was 39 months (range 4-200 months). The final status of patients was established in November 2022. At this time, 25 (18.0%) patients were alive without disease, 8 (5.7%) patients were alive with disease, 72 (51.8%) patients had died due to disease, and 34 (24.5%) patients had died of unrelated causes. The 5-year EFS and OS rates were 47.0% and 42.0%, respectively.
When we stratified the patients by ADAM3A and ADAM5 copy number (total deletion or one, or two, or three or more copies), we observed that CNVs status seems to have an association with EFS (p = 0.05) ( Figure 1C).
At 60 months of follow-up, lower EFS was observed in patients with advanced tumor stage (37.8% vs. 69.2%, p = 0.002) when compared with others, and in patients with total deletion or one copy of the ADAM3A and ADAM5 pseudogenes (39.7% vs. 50.7%, p = 0.04) when compared with carriers of two or more copies (K-M estimates) ( Figure 1D). The significance of the differences between groups remained the same in univariate Cox analysis. After multivariate Cox analysis, advanced tumor stage (HR: 2.34, 95% CI: 1.37-4.00, p = 0.002) and total deletion or one copy of the ADAM3A and ADAM5 pseudogenes (HR: 1.68, 95% CI: 1.07-2.66, p = 0.02) were found to be predictors of poor EFS (Table 3).
At 60 months of follow-up, lower OS was observed only in patients with advanced tumor stage (32.0% vs. 63.6%, p < 0.001) when compared with others (K-M estimates). ADAM3A and ADAM5 copy number loss or gain did not influence the OS of our 139 OPSCC patients (Table 3).
Additionally, the prognostic value of ADAM3A and ADAM5, and miR-122b-5p in 124 and 522 HNSCC patients, respectively, was further evaluated using the Kaplan-Meier Plotter online database. We found that lower expression of the pseudogenes ADAM3A and ADAM5 ( Figure 1E) indicated lower relapse-free survival (RFS) in general HNSCC patients (p = 0.002), and higher expression of miR-122b-5p was associated with lower OS in general HNSCC patients (p = 0.002) ( Figure 1F). Table 3. Clinical and pathological aspects and ADAM3A and ADAM5 copy number variation in the event-free survival and overall survival of 139 oropharyngeal squamous cell carcinoma patients.

Event-Free Survival
Overall Survival

Discussion
We investigated herein whether inherited CNVs alter the risk and prognosis of OPSCC patients. We initially observed that CNVs in the 4q13.2 region, containing the UGT2B17 gene; the 8p11.22 region, containing the ADAM3A and ADAM5 pseudogenes; and the 20p13 region, containing the SIRPB1 gene, were identified in BTSCC patients in a pilot study (Step 1). However, after validation in a large cohort of OPSCC patients (Step 2), the 4q13.2 and 20p13 regions did not confirm their association with OPSCC risk.
In HNSCC, only one study with the UGT2B17 gene was found [37]. Researchers suggested that UGT2B17 somatic deletion could modify the effects of smoking on TP53mutations, increasing the relapse rates among HPV-positive HNSCC patients [37]. However, we did not find any relationship between the 4q13.2 region and OPSCC risk. A possible explanation for our distinct result may be the prevalence of HPV-negative patients in our cohort and that Mafune et al. [37] analyzed tumor cells for somatic mutation while in our study, we searched for inherited CNVs in patients' blood samples.
A complex pattern of amplification and deletion of the 20p13 region was previously associated with oral SCC risk [27]. This is the only study focusing on the potential role of SIRPB1 in HNSCC, although the researchers did not find an association between SIRPB1 and oral SCC risk after validation in tumor samples [27], as was the case in our study.
We observed that the gain of more than three copies of the 8p11.22 region, which contains the ADAM3A and ADAM5 pseudogenes, was associated with OPSCC risk. Moreover, Vincent-Chong et al. [27] detected, in tumor samples, amplification in the ADAM5 region (8p11.23-p11.22) associated with oral SCC risk [27]. In contrast, two studies identified the loss of the somatic copy number of the 8p11 region to be associated with oral SCC risk in a male-prevalent cohort [38] and in oral and laryngeal SCC cell lines [39]. To the best of our knowledge, there are no studies focusing on the role of the referred CNV in the development or progression of pharyngeal SCC, neither somatic nor in an inherited way. In addition to somatic CNVs in tumor cells, identifying inherited cancer susceptibility in an individual may be used clinically to facilitate diagnosis and to recognize high-risk patients that should receive a closer follow-up.
ADAM3A and ADAM5 are classified as pseudogenes once they do not seem to produce a specific protein [40]. The prevalent existence of pseudogenes indicates that they may play a key role in basic physiology and disease progression, including that of cancer [22,41]. The ways in which ADAM3A and ADAM5 may mediate tumorigenesis in OPSCC are still unclear, but pseudogenes can interact with parental genes or other gene loci, altering their sequences and/or transcriptional activities [41].
In order to support the hypothesis that pseudogene-derived transcripts could compete for miRNA binding, we searched for a potential pseudogene-miRNA-mRNA interaction. We found that the ADAM5 pseudogene shared a highly homologous sequence at 3 -UTR of the ADAM9 gene, and this sequence was predicted to be a binding site for miR-122b-5p. In fact, we observed that OPSCC patients with more than three copies of the ADAM3A and ADAM5 pseudogenes presented higher ADAM9 mRNA expression levels, possibly as a result of the competition between the ADAM5 pseudogene and the ADAM9 gene for the miR-122b-5p binding site. ADAM9 is known to be involved in tumor formation and progression, especially in HNSCC [42][43][44][45]. In addition, in salivary samples, lower expression of miR-122b-5p and higher expression of ADAM9 were found to be a useful biomarker in the screening and diagnosis of oropharynx and oral SCC, respectively [44,46]. Moreover, miR-122b-5p has been found to interact with DRAIC lncRNA, transcribed from the 15q23 region, in nasopharyngeal SCC [47]. Our data suggest that gain of copies (more than three copies) of ADAM3A and ADAM5 could lead to an increase of pseudogene-derived transcript that competes for miR-122b-5p, allowing for enhance of ADAM9 expression and increasing OPSCC risk [42][43][44][45].
Furthermore, we found that OPSCC patients with total deletion or one copy of the ADAM3A and ADAM5 pseudogenes presented lower EFS when compared with carriers of two or more copies. In the literature, total deletion of ADAM3A was associated with poor prognosis in early-stage non-small cell lung cancer patients [48]. We also found, using the Kaplan-Meier Plotter online database, that lower expression of the pseudogenes ADAM3A and ADAM5 indicated worse RFS, and that higher expression of miR-122b-5p indicated lower OS in mixed HNSCC patients [35].
The contradictory results of ADAM3A and ADAM5 CNV in OPSCC risk and prognosis could be explained by miR-122b-5p and therapy interaction. In agreement with our pseudogene-miRNA-mRNA hypothesis, the copy number loss of ADAM5 leads to more available miR-122b-5p circulating, which could interact and silence other genes, including those involved in therapy modulation [21]. In lung cancer cells, miR-122-5p was found to be required for radiation-induced TGF-β activation and the epithelial-mesenchymal transition process, facilitating tumor cell invasion and migration [49]. It is worth noting that 95% of our patients cohort received irradiation treatment, thus, it is possible that miR-122-5p could be modulating patients' therapy leading to a worse prognosis. In addition, high expression of miR-122b was identified in exosomes derived from HNSCC cells [50].
In summary, we identified for the first time that inherited abnormality CNV in the pseudogenes ADAM3A and ADAM5 modulates the OPSCC occurrence and prognosis possible by ADAM9 and miR-122b-5p interaction. We are aware of the study limitation, especially as the results were obtained in a relatively small number of patients from a single country and the lack of functional analysis to prove our pseudogene-miRNA-mRNA hypothesis. We highlight that this hypothesis must be validated with larger sample sizes and by functional analysis.
Nevertheless, our results can help researchers to identify new biomarkers for OP-SCC diagnosis and to predict prognosis. Our study also contributes to target therapy development for patients' treatment, especially for those that do not respond to conventional therapy.