If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
1 Head and Neck Cancer Division, Oral & Maxillofacial Surgery Department, & Molecular Oncology Program, Medical Research Centre, Lagos State University College of Medicine, Ikeja, Lagos, Nigeria.
E.M. Okoturo
Footnotes
# these authors contributed equally to this manuscript 1 Head and Neck Cancer Division, Oral & Maxillofacial Surgery Department, & Molecular Oncology Program, Medical Research Centre, Lagos State University College of Medicine, Ikeja, Lagos, Nigeria.
Affiliations
Liverpool Head & Neck Centre, William Henry Duncan Building, University of Liverpool. L7 8TX. UK.
# these authors contributed equally to this manuscript 1 Head and Neck Cancer Division, Oral & Maxillofacial Surgery Department, & Molecular Oncology Program, Medical Research Centre, Lagos State University College of Medicine, Ikeja, Lagos, Nigeria. 2 Edge Hill University, Medical School, St Helens Road, Ormskirk, Lancashire, L39 4QP
Proliferative verrucous leukoplakia (PVL) is a rare form of oral leukoplakia with a relatively high transformation rate to oral squamous cell carcinoma (OSCC). Molecular analysis of this disorder at the genome level is limited and only identifies molecular similarities between PVL and OSCC. However, the clinical profile suggests that molecular differences may be more important.
Study Design
Whole exome sequencing of five PVL-associated OSCC and paired blood samples identified somatic mutations common to the tumours. Whole methylome analysis of four PVL-associated OSCC and three OSCC of non-PVL origin used Infinium MethylationEPIC “850k” BeadChip, and differential methylation was explored.
Results
In contrast to conventional OSCC, PVL-associated OSCC showed infrequent TP53 mutation and altered spectra of PIK3CA and NOTCH1 mutations. Unsupervised hierarchical clustering identified 63 probes that discriminated between PVL-associated OSCC and OSCC of non-PVL origin. Differences in methylation were most significant for divalent metal ion transport, particularly calcium movement.
Conclusion
Specific differences in mutation and methylation profiles between PVL-derived OSCC and OSCC of non-PVL origin suggest differences in their transformation pathways. Further studies of early PVL lesions may identify markers of transformation that are also applicable to more common oral premalignant disorders such as oral epithelial dysplasia.
Proliferative verrucous leukoplakia (PVL) is a rare form of oral leukoplakia first described in 1985 by Hansen et al1. The clinical course begins with an isolated oral white plaque that becomes multifocal and exophytic, often inexorably progressing over decades. It is persistent, irreversible and refractory to treatment, eventually undergoing malignant transformation to verrucous, then invasive, oral squamous cell carcinoma (OSCC) in more than 50% of cases1,2, although overall survival in PVL-OSCC is significantly better than for conventional OSCC3. The diagnosis of PVL is based on a number of clinical and histopathological criteria and is frequently established retrospectively after malignancy has been diagnosed1.
The molecular pathogenesis of PVL represents an intriguing and largely uncharted field of investigation. There are several critical clinical features that distinguish PVL from other head & neck mucosal lesions at risk of malignant transformation4. The very high rate of malignant transformation in PVL, well over 50% in most series5, contrasts with oral epithelial dysplasia (OED) where the risk is only around 10% for mild to moderate dysplasia and 24% for severe dysplasia and carcinoma in situ6. We have speculated that PVL would transform in all patients if given sufficient natural lifespan and observation7. Further, PVL generally occurs independently of known risk factors for OSCC, and is specifically unrelated to smoking, alcohol consumption or high risk HPV7. The clinical and pathological appearance and progression of PVL to OSCC is highly stereotyped, following a predictable stepwise continuum. As such, it is possible that PVL follows a distinct molecular origin and pathway to conventional OSCC or OED, with critical genetic or epigenetic determinants that may be unique to PVL, or perhaps simply much less commonly seen in comparable TCGA metadata relating to head and neck SCC. Given the inexorable progression to OSCC, it is also hypothesised that the critical event to malignant transformation is present even in early lesions, prior to development of epithelial dysplasia.
Recent systematic reviews of the molecular pathogenesis of PVL8,9 concluded that only aneuploidy showed promise as a putative marker of malignancy and highlighted the candidate gene approach of most previous studies. Evidently, these previous studies have been limited by their methodology to targeting aberrations commonly found in OSCC and only recently have there been reports of genome-wide studies of PVL that exploit the power of contemporary sequencing approaches10,11. These two, related, studies investigated differential gene expression and DNA methylation in a small cohort of PVL patients in comparison to normal oral mucosa. The authors then validated their results in silico against the expression data for 314 specimens and 31 adjacent normal samples from the TCGA OSCC sample set (2015) in which they demonstrated similar expression profiles10.
In contrast to the two studies that compared PVL with normal oral mucosa, our aim was to interrogate the genome-wide and epigenome-wide characteristics of PVL-associated OSCC and identify where the pattern of DNA methylation or somatic mutations contrasted with conventional OSCC. We hypothesised that this may give novel insight into the molecular mechanism of malignant transformation in OSCC.
Materials and Methods
Patients
Patients with PVL-associated OSCC (PVL-OSCC) were identified from cases under observation at the Oral Dysplasia Clinic of the Liverpool University Dental Hospital over the period 2008-2017 (Table 1). A diagnosis of PVL for this study was defined as a progressive clinical course based on Hansen's histological grading1 with numerous biopsy episodes, post excisional recurrence and a propensity to develop into OSCC. Patients were included in this REC approved study (EC 47.01 & H/10/1002) after giving written, informed consent. The histopathology of the biopsied material was re-examined and cases with uncertainty about a diagnosis of PVL were excluded from this research, leaving a total of 22 patients.
Control cases were taken from a previously used cohort of OSCCs (REC number 07/Q1505/1512) (Table 1). Archival, formalin-fixed paraffin embedded (FFPE) tissue was preferred for single gene, confirmatory assays because this was comparable to the format of the majority of the PVL material: it was available for 20 cases. These 20 plus an additional 5 cases were available as fresh tissue. These cases will be referred to as conventional OSCCs (cOSSCs).
Samples
Following surgery, 5mm3 tissue were obtained from within the tumour of PVL-OSCC and cOSCC, outwith any necrotic area or tumour margins. Tissue with histopathologically low Hansen grade (Grade 1-3) was obtained from the same or separate resections undertaken on the same day and termed low-grade PVL (LG-PVL). Tissue was snap frozen and stored at -80°C. Paired blood samples were taken as controls for whole exome analysis. Genomic DNA from all fresh frozen samples was extracted using a DNEasy blood and tissue kit (Qiagen) according to manufacturer's protocol.
Archival, formalin-fixed paraffin embedded (FFPE) tissue from PVL and cOSCC cases was obtained from the local Pathology department. LG-PVL were taken from areas with Hansen's grade 1-3 within PVL-OSCC tumour specimens. Genomic DNA was extracted using a DNA FFPE Tissue kit (Qiagen) according to the manufacturer's protocol.
Whole Exome Sequencing (WES)
Genomic DNA from 5 fresh, frozen PVL-OSCC cases (Table 1) with paired blood samples were subject to Whole Exome sequencing at the Centre for Genomic Research (GRC), University of Liverpool. Briefly, target capture and library construction were undertaken using SureSelect Human All Exon v7 (Agilent). The libraries were sequenced on an Illumina Hiseq 4000 platform.
All reads were aligned to the reference genome (WG38) using BWA (Burrows-Wheeler Alignment tool)13. Somatic variants were identified by subtracting variants also detected in the paired blood sample genome sequence using Strelka214 utilising a base call quality score of 90% and a minor allele frequency of ≤1.0.
Whole Genome Methylation Profiling (WGMP)
Genomic DNA was isolated from fresh frozen samples obtained from 4 PVL-OSCC, 3 cOSCC and 4 premalignant PVL samples, 2 of which were obtained from tissue adjacent to the PVL-OSCC samples (Table 1). DNA samples were transferred to Edinburgh Genomics for bisulphite treatment and genome-wide microarray hybridisation using Infinium MethylationEPIC “850k” BeadChip. Duplicate hybridisation was undertaken for 3/4 PVL-OSCC, 1/3 cOSCC and 2/4 LG-PVL samples. Arrays were scanned on Illumina's iScan technology and the data transferred to the Computational Biological Facility (CBF), University of Liverpool.
Preprocessing and normalisation
EPIC arrays were analysed using functionality within the minfi package15. Briefly, samples were checked for global hybridisation quality based on an average detection p value threshold of < 0.05 for all probes. Individual CpG probes were discarded from the array if their detection p value was > 0.01, indicating a failed position. Probes were also discarded if they were located on chromosomes X & Y, within 2 base pairs of a single nucleotide polymorphism (SNP) with a minor allele frequency greater than 0.05 or if the probe was previously found to cross hybridise to multiple genomic locations16,17. Sample wise normalisation was performed using the regression of correlated probes (RCP) algorithm to correct for biases between type I and type II probe distributions18.
Differential methylation
Univariate statistical analysis for each CpG was performed using the linear models for microarray data (limma) package19. The duplicateCorrelation function was used to estimate the consensus correlation between arrays originating from the same biological donor and was subsequently used as a parameter for lmFit. To implement the linear models, a contrast matrix was built with coefficients for ‘PVL-OSCC and cOSCC’ and ‘PVL-OSCC and LG-PVL’. Significance was determined based on a Benjamini Hochberg (BH) adjusted FDR < 0.05
Differential methylation was assessed in the context of regions using DMRcate20. Regions were defined as blocks of 1000 nucleotides for which a Gaussian kernel smoothed function is fit. A region was considered to be differentially methylated if its BH corrected FDR was < 0.1 and had an absolute mean beta fold change > 0.1.
Analysis
All principal component analysis (PCA) were performed in R (Version 4.0.2). M values were centred and scaled prior to matrix decomposition using the prcomp function from the stats package21. Heatmaps were generated using the pheatmap function within the pheatmap package22. All hierarchical clustering was performed using the ward.D2 method. Statistical analysis to test association between principal components and variables in the data was undertaken using one-way ANOVAs. If there was a significant difference for the main effect (p value < 0.05) a Tukey's Honest Significant Difference (TukeyHSD) post-hoc comparison was applied. Significance for post-hoc testing was accepted as an adjusted p value < 0.05.
Single gene assays
Somatic variants
Pyrosequencing assays for selected, PVL-OSCC- and cOSCC-specific, somatic changes were designed using Oligo 7.0 software (MBI). Oligonucleotides were synthesized by Eurofins Genomics, Ebersberg, Germany (Supplemental Table 1). Pyrosequencing was undertaken on 3 DNA samples from the WES cohort plus an additional 12 PVL-OSCC, 13 LG-PVL and 20 cOSCC FFPE specimens (Table 1) using Pyromark Gold reagents (Qiagen) on a Pyromark Q96 workstation (Qiagen) and the resulting pyrograms analysed for the presence of variants at the given site.
Promoter methylation
DNA was converted with sodium bisulphite (EZ DNA methylation-Gold, Zymo Research). Quantitative methylation specific PCR (qMSP) and pyrosequencing methylation (PMA) assays were designed using Oligo 7.0 (MBI) and PyroMark® Assay Design SW 2.0 software (Qiagen), respectively. Oligonucleotides were synthesized by Eurofins Genomics, Ebersberg, Germany (Supplemental Table 1). Lymphocyte DNA artificially methylated to 20%, 10%, 5% and 2.5% were used as standards as previously described23.
For qMSP of the CDKN2A (p16) promoter, a previously published method was used23. DNA from 12 PVL-OSCC and 11 cOSCC were analysed. Assays were multiplexed with ACTB as an internal normalisation control and run in duplicate. The mean Ct value from the duplicates was used to calculate relative quantitation (RQ) values using the ΔΔCt method with the unmethylated lymphocyte DNA being used as the reference and RQ was compared to the standards to obtain a semi-quantitation of DNA methylation.
For PMA of the MGMT promoter, PCR products from 20 PVL-OSCC and 20 cOSCC FFPE specimens were subject to pyrosequencing using Pyromark Gold reagents (Qiagen) on a Pyromark Q96 workstation (Qiagen) and the resulting pyrograms analysed for the presence of methylation as previously described27.
Statistical analysis
A Students t test was used for comparisons between PVL-OSCC and cOSCC variant frequencies
Results
Whole exome sequencing
Seven genes were identified as having PVL-OSCC specific mutations (present in at least 3/5 samples: Table 2). Two of these are also commonly mutated in conventional OSCC (NOTCH1 and PIK3CA). An interrogation of the WES data for alterations in other genes mutated at more than 15% in conventional OSCC revealed changes in PVL-OSCC samples at lower frequencies (Table 2).
Table 2Most frequent genetic changes in PVL-OSCC and conventional OSCC
The genetic changes observed in the TTN, LRP2, LRP5, JMJD1C and EPG5 genes in PVL-OSCC were classified as of moderate-low impact because they produced amino acid changes with limited structural effect or no amino acid change (Supplemental Table 2). However, the changes observed in NOTCH1 and PIK3CA were classified as of high impact, producing amino acid changes that altered protein secondary/tertiary structure or inserted a stop codon. Analysis of mutation profiles demonstrated the presence of mostly novel genetic alterations in the PVL-OSCC samples and absence of more commonly observed mutations (Columns 1 & 3, Table 3). The only two mutations observed in both PVL- and conventional OSCC were at PIK3CA p.E545K, a mutation found in around 9% of all cancers annotated in TGCA, and NOTCH1 p.R353C, a mutation found at low frequency in both PVL- and conventional OSCC. Pyrosequencing assays were designed to cover PIK3CA p.E545K and p.E542K mutations, the PIK3CA p.H1047R mutation, and NOTCH1 p.G326D and p.C387X/p.C387Y mutations. Samples for pyrosequencing included DNA from specimens used to generate the primary WES data, LG-PVL, to assess early/late stage appearance and conventional OSCC samples. In this larger cohort, the frequency of mutation for PIK3CA p.E545K fell to levels comparable to those observed in conventional OSCC (P>0.05)(Table 3). Notably, 2/3 of the LG-PVL samples demonstrating PIK3CA p.E545K mutations also showed that mutation in the corresponding PVL-OSCC tissue.
Principal component analysis (PCA) was used to assess variance based on methylation profiles between PVL-OSCC, cOSCC and LG-PVL tissue. PCA based on unbiased, genome-wide methylation profiles of ∼785,000 probes demonstrated significant differences in variance within principal component 1 (PC1) between tissue from LG-PVL and tissue from cOSCC (p = 0.026) and PVL-OSCC (p = 0.005) (Figure 1). However, no significant differences in variance were observed between cOSCC and PVL-OSCC in PC1 or later components, indicating similar global methylation patterns in the two tissues.
Figure 1Unbiased analysis of genome-wide methylation differences between cOSCC, LG-PVL and PVL-OSCC. A: principal component analysis; B: box & whisker plot of data from principal component 1 (PC1).
A total of 63 differentially methylated probes between PVL-OSCC and cOSCC were identified using a univariate model in limma with a Benjamini Hochberg FDR < 0.05. (Supplementary Table 3). Proportional to the distribution of probes tested on the EPIC array, the 63 differentially methylated probes contain 1.7 fold enrichment for probes located within CpG islands with 42/63 probes located within a coding region of a gene. A further 19 regions were identified based on defined blocks of 1000 nucleotides within the 63 CpGs to be differentially methylated (DMRs) based on an FDR of 10%. (Supplementary Table 4). Interestingly, the promoters of genes that are commonly methylated in cOSCC, such as CDKN2A, were absent from both of these lists.
Targeted multivariate clustering and PCA
Multivariate unsupervised hierarchical clustering and principal component analysis showed that all 3 tissues; LG-PVL, PVL-OSCC and cOSCC could be discriminated by the methylation profiles of the 63 differentially methylated loci (Figure 2A). Interestingly, this analysis grouped LG-PVL more closely with cOSCC than with PVL-OSCC. Moreover, using only the 63 probes identified from the univariate model, PC1 and PC2 were shown to capture ∼90% variance and to be significantly associated with tissue type (Figure 2B), with PC1 separating PVL-OSCC from cOSCC and PC2 separating LG-PVL from both OSCC tissue types. LG-PVL was grouped more closely to cOSCC than PVL-OSCC in PC1. However, 16 of the 63 differentially methylated loci showed little or no difference in methylation levels between PVL-OSCC and LG-PVL and may be early events in the development of PVL-OSCC (Supplementary Table 5). Seven of the 16 DMPs are located in a CpG island and 4 of these are associated with known genes: MACF1, SFRP2, LMX1A and ME3.
Figure 2Multivariate clustering using 63 differentially methylated probes. A: Unsupervised hierarchical clustering based on scaled M values of methylation at 63 differentially methylated probes; B: Principal component analysis of the 63 differentially methylated probes.
The 63 differentially methylated probes were analysed for overlapping biological processes. Differences in methylation were most significant for divalent metal ion transport, particularly calcium movement (Supplementary table 3). Interestingly, these gene promoters were generally less methylated in PVL-OSCC compared with cOSCC.
Analysis of gene promoters commonly methylated in OSCC
None of the gene promoters that are commonly hypermethylated in cOSCC were identified in the list of probes that were differentially methylated between PVL-OSCC and cOSCC (Supplementary Tables 3 & 4). This implied that the methylation levels at these promoters are similar in PVL-OSCC and cOSCC. In order to investigate this further, multivariate unsupervised hierarchical clustering was undertaken using 315 probes representing 24 genes shown to have significantly altered promoter methylation in OSCC25,26 (Supplementary Table 6). This confirmed the separation of LG-PVL tissue from all cancer tissue and suggested that, although there were many similarities in the methylation profiles of PVL-OSCC and cOSCC, some differences were observable (Figure 3).
Figure 3Multivariate clustering using 315 probes associated with differentially methylated genes identified from the literature. Unsupervised hierarchical clustering based on scaled M values of methylation
In addition, the methylation status of the CDKN2A (p16) promoter was investigated by RT-qMSP23 and the promoter of MGMT by PMA27 in DNA prepared from FFPE tissue. None of the PVL-OSCC tumours demonstrated CDKN2A promoter methylation compared with 5/9 (56%) cOSCC tumours (P=0.013). Similarly, only 3/18 (17%) PVL-OSCC showed MGMT promoter methylation compared with 5/18 (28%) cOSCC, but this was not statistically significant (P=0.33).
Discussion
PVL transforms into OSCC at a frequency of more than 50% in comparison with 10% for mild to moderate OED and 24% for severe OED and carcinoma in situ5,6. Historically, molecular studies have attempted to show that the molecular profile of PVL-OSCC is similar to cOSCC. However, the current study aimed to investigate the differences between these two entities in an effort to explain the differences in their transformation rates.
Whole exome sequencing identified 7 genes presenting with mutations in ≥60% of samples, only 2 of which (PIK3CA and NOTCH1) had previously been identified as commonly mutated in cOSCC24. In contrast, TP53 mutations were less common in this cohort. Of course, these data may reflect the small sample size used, but the TP53 data reflect previous observations in PVL28. The PI3CA mutations identified in PVL-OSCC tumours were different to those most commonly observed in cOSCC, but have been previously shown to have pathogenic potential in other cancers, most notably breast and endometrium29. The p.T1025A alteration located in the PI3K/PI4K domain of the protein has been shown to increase transforming ability in cell lines in culture and is predicted to be a gain of function mutation30. The p.E545K alteration is located in the helical domain, is comonly observed in a number of cancers and results in an amino acid substitution of opposite charge that uncouples the catalytic and regulatory subunits of the protein31, as does the p.E542K mutation that is more commonly observed in head and neck cancer. Interestingly, the p.E545K mutation has also been shown to alter fatty acid metabolism, including the upregulation of arachadonic acid that is needed for the synthesis of prostaglandins, a pro-inflammatory molecule implicated in cancer development32 and may thus be more important in the pathogenesis of transformation. The p.E545K mutations of PI3CA were also observed in early (LG-PVL) lesions, which suggests that this oncogenic driver is an early occurring event. However, there is no support in the literature for a role for early mutation of PI3CA in OED. NOTCH1 alterations can be oncogenic or tumour suppressive in HNSCC33. The mutations observed in this study are all located in the extracellular subunit and may affect receptor binding, with 2 of them changing cysteine residues and potentially affecting secondary and tertiary structure that is reliant on disulphide bonding.
Methylation profiling comparing PVL-OSCC with non-malignant LG-PVL identified 157,963 differentially methylated probes and 26,498 differentially methylated regions. If the list is restricted to promoters (located -1500 to -200bp upstream from the start site and within the 5′UTR) then 4,871 islands are differentially methylated. This is of a similar scale to that described by Herreros-Pomares et al who identified 4,647 DMRs10 in a series of 10 PVL when compared with 5 healthy tissues. However, there were no common targets in the top 50 DMRs from the two datasets. Herreros-Pomares then showed similarities in profiles between PVL data and OSCC on TCGA for 13 differentially methylated, differentially expressed probes. In contrast, the data presented in the current paper identifies differences between the methylation profiles of PVL-OSCC and cOSCC. Unsurprisingly, none of the 63 DMPs that are differentially methylated were identified by Herreros-Pomares as having similar methylation profiles in PVL and OSCC.
It may be postulated that DMPs that are differentially expressed in PVL-OSCC compared with cOSCC but that show similar methylation levels in a comparison of LG-PVL with PVL-OSCC indicate early events in the transformation process. Of the four known genes that fell into this group, only MACF1 showed a methylation pattern consistent with tumour development. This microtubule associated protein is implicated in proliferation, migration and cell signalling in several cancers34 and showed reduced methylation in PVL-OSCC. In contrast, reduced expression of SFRP2 and LMX1A have been previously associated with colorectal and bladder cancer, respectively35,36, but both showed decreased methylation in PVL samples in our study. Similarly, ME3 shows increased expression in pancreatic cancer37, but was characterised by increased methylation in our PVL samples. It is well known that methylation status does not always negatively correspond with expression and it should be remembered that the relative levels of methylation reported in the current study are in comparison with cOSCC and not with normal tissue. Nevertheless, the implication is that MACF1 is a gene worthy of further study in the development of PVL-OSCC.
Global hypomethylation coupled with targeted promoter hypermethylation is a key feature of OSCC38,39. Subtle differences between OSCC populations have been documented40, so it is not surprising that we have identified differences between the methylation patterns in PVL-OSCC and cOSCC. None of these differences in methylation were in gene promoters that are commonly methylated in cOSCC, so we hypothesized that these were critical, and common, to the transformation process in all oral cancer. However, multivariate unsupervised hierarchical clustering of 315 probes representing 24 genes shown to have significantly altered promoter methylation, either in our previous study25 or in several candidate gene studies26 did show differences between PVL-OSCC and cOSCC methylation patterns. Interestingly, comparatively lower levels of methylation were observed in PVL-OSCC compared with cOSCC for a number of these probes, and we validated this in the CDKN2A gene promoter. This suggests that different mechanisms are involved in the development of PVL-OSCC compared with cOSCC.
Gene ontology analysis of the 63 differentially methylated probes identified divalent metal ion transport, particularly calcium movement, and other signalling pathways as being affected, although the gene promoters were generally less methylated in PVL-OSCC compared with cOSCC, suggesting increased expression of these pathways. However, the pattern was by no means uniform, with some pathway members being hypermethylated in cOSCC and others in PVL-OSCC. Calcium movement is associated with intracellular signalling which is often activated in cancer, but this is not the only role of this cation. Increased calcium in the mitochondria can stimulate the calcium-sensitive dehydrogenases of the Kreb's cycle leading to increased ATP production and increased release of superoxides41, which in turn may stimulate proliferation and DNA damage, respectively.
In conclusion, we have identified specific differences in mutation and methylation profiles between PVL-associated OSCC and conventional OSCC, suggesting differences in the transformation pathway in these two entities. However, it was not possible to identify a single genetic or epigenetic pathway / gene unique to PVL. Although the data is based on few samples, these were carefully curated for a clinical and pathological diagnosis of PVL and augmented with samples from premalignant, low grade PVL. Further studies should concentrate on analysis of these earlier lesions in order to identify markers of transformation that may also be applicable to more common oral premalignant disorders such as oral epithelial dysplasia42.
Funding
TO was supported by: Academic Staff Training and Development Programme (TETFund), Lagos State University, Ojo, Nigeria (LASU/VC/TETF/AST&D/001); the British Association of Oral and Maxillofacial Surgeons (BAOMS); The University of Liverpool PhD bursary; University of Liverpool Technology Directorate Voucher (40428830). The funding sources had no role in the study design, collection and analysis of data, writing of the manuscript or decision to publish.
Statement of clinical relevance
PVL, an oral potentially premalignant disorder with a high rate of transformation, shows a novel spectrum of mutations and differences in genome methylation compared with conventional oral SCC which may provide insight into its discrete aetiology.
References
[1] Hansen LS, Olson JA, Silverman SJ. Proliferative verrucous leukoplakia. A long-term study of thirty patients. Oral Surgery, Oral Medicine, and Oral Pathology. 1985;60:285-98.
[2] Silverman SJ, Gorsky M. Proliferative verrucous leukoplakia: a follow-up study of 54 cases. Oral Surgery,Oral Medicine, Oral Pathology, Oral Radiology and Endodontology. 1997;84:154-7.
[3] Gonzáles-Moles MÁ, Warnakulasuriya S, Ramos-García P. Prognosis parameters of oral carcinomas developed in proliferative verrucous leukoplakia: A systematic review and meta-analysis. Cancers (Basel). 2021;13:4843.
[4] Akrish S, Ben-Izhak O, Sabo E, Rachmiel A. Oral squamous cell carcinoma associated with proliferative verrucous leukoplakia compared with conventional squamous cell carcinoma - a clinical, histologic and immunohistochemical study. Oral Surgery, Oral Medicine, and Oral Pathology and Oral Radiology. 2015;119:318-25.
[5] Ramos-García P, Gonzáles-Moles MÁ, Mello FW, Bagan JV, Warnakulasuriya S. Malignant transfromation of oral proliferative verrucous leukoplakia: A systematc review and meta-analysis. Oral Dis. 2021;27:1896-907.
[6] Mehenna HM, Rattay T, Smith J, McConkey CC. Treatment and follow-up of oral dysplasia - a systematic review and meta-analysis. Head & Neck. 2009;31:1600-9.
[7] Borgna SC, Clarke PT, Schache AG, Lowe D, Ho MW, McCarthy CE, et al. Management of proliferative verrucous leukoplakia: justification for a conservative approach. Head & Neck. 2017;39:1997-2003.
[8] Okoturo EM, Risk JM, Schache AG, Shaw RJ, Boyd MT. Molecular pathogenesis of proliferative verrucous leukoplakia: a systemic review. British Journal of Oral and Maxillofacial Surgery. 2018;56:780-5.
[9] Rintala M, Vahlberg T, Salo T, Rautava J. Proliferative verrucous leukoplakia and its tumor markers: Systematic review and meta-analysis. Head & Neck. 2019;41:1499-507.
[10] Herreros-Pomares A, Llorens C, Soriano B, Bagan L, Moreno A, Calabuig-Fariñas S, et al. Differentially methylated genes in proliferative verrucous leukoplakia reveal potential malignant biomarkers for oral squamous cell carcinoma. Oral Oncol. 2021;116:105191.
[11] Llorens C, Soriano B, Trilla-Fuertes L, Bagan L, Ramos-Ruiz R, Gamez-Pozo A, et al. Immune expression profile identification in a group of proliferative verrucous leukoplakia patients: a pre-cancer niche for oral squamous cell carcinoma development. Clinical Oral Investigation. 2021;25:2645-57.
[12] Shaw RJ, Hobkirk AJ, Nikolaidis G, Woolgar JW, Triantafyllou A, Brown JS, et al. Molecular staging of surgical margins in oral squamous cell carcinoma using promoter methylation of p16INK4A, cytoglobin, E-cadherin, and TMEFF2. Annals of Surgical Oncology. 2013;20:2796-802.
[13] Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics. 2009;25:1754-60.
[14] Kim S, Scheffler K, Halpern AL, Bekritsky MA, Noh E, Källberg M, et al. Strelka2: fast and accurate calling of germline and somatic variants. Nature Methods. 2018;15:591-4.
[15] Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, et al. Minfi: A flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30:1363-9.
[16] Chen YA, Lemire M, Choufani S, Butcher DT, Grafodatskaya D, Zanke BW, et al. Discovery of cross-reactive probes and polymorphic CpGs in the Illumina Infinium HumanMethylation450 microarray. Epigenetics. 2013;8:203-9.
[17] Pidsley R, Zotenko E, Peters TJ, Lawrence MG, Risbridger GP, Molloy P, et al. Critical evaluation of the Illumina MethylationEPIC BeadChip microarray for whole-genome DNA methylation profiling. Genome Biol. 2016;17:208.
[18] Niu L, Taylor JA. RCP: A novel probe design bias correction method for Illumina Methylation BeadChip. Bioinformatics. 2016;32:2659-63.
[19] Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analysis for RNA-sequencing and microarray studies. Nuc Acids Res. 2015;43:e47.
[20] Peters TJ, Buckley MJ, Statham AL, Pidsley R, Samaras K, Lord RV, et al. De novo identification of differentially methylated regions in the human genome. Epigenetics & Chromatin. 2015;8:6.
[21] RCoreTeam. R: A language environment for statistical computing.: R Foundation for Statistical Computing, Vienna, Austria; 2021.
[22] Kolde R. pheatmap: Pretty heatmaps. R package version 1.0.12. 2019.
[23] Nikolaidis G, Raji O, Markopoulou S, Gosney J, Bryan J, Warburton C, et al. DNA methylation biomarkers offer improved diagnostic efficiency in lung cancer. Cancer Res. 2012;72:5692-701.
[24] TCGA. NCI GDC Data Portal. Data Release 29.0 ed.
[25] Jithesh PV, Risk JM, Schache AG, Dhanda J, Lane B, Liloglou T, et al. The epigenetic landscape of oral squamous cell carcinoma. Br J Cancer. 2013;108:370-9.
[26] Mascolo M, Siano M, Ilardi G, Russo D, Merolla F, De Rosa G, et al. Epigenetic disregulation in oral cancer. International Journal of Molecular Sciences. 2012;13:2331-52.
[27] Shaw RJ, Hall GL, Lowe D, Bowers NL, Liloglou T, Field JK, et al. CpG island methylation phenotype (CIMP) in oral cancer: Associated with a marked inflammatory response and less aggressive tumour biology. Oral Oncol 2007;43:878-886
[28] Gopalakrishnan R, Weghorst CM, Lehman TA, Calvert RJ, Bijur G, Sabourin CL, et al. Mutated and wild-type p53 expression and HPV integration in proliferative verrucous leukoplakia and oral squamous cell carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endodontics. 1997;83:471-7.
[29] Tate JG, Bamford S, Jubb HC, Sondka Z, Beare DM, Bindal N, et al. COSMIC: the Catalogue Of Somatic Mutations In Cancer. Nuc Acids Res. 2019;47:D941–D7.
[30] Ng PK, Li J, Jeong KJ, Shao S, Chen H, Tsang YH, et al. Systemic functionnal annotation of somatic mutations in cancer. Cancer Cell. 2018;33:450-62.
[31] Leontiadou H, Galdadas I, Athanasiou C, Cournia Z. Insights into the mechanism of the PI3CA E545K activating mutation using MD simulations. Scientific Reports. 2018;8:15544.
[32] Koundouros N, Karali E, Tripp A, Valle A, Inglese P, Perry NJS, et al. Metabolic fingerprinting links oncogenic PIK3CA with enhanced arachidonic acid-derived eicosanoids. Cell. 2020;181:1596-611.
[33] Fukusuki T, Califano J. The NOTCH Pathway in Head and Neck Squamous Cell Carcinoma. J Dent Res. 2018;97:645-53.
[34] Miao Z, Ali A, Hu L, Zhao F, Yin C, Chen C, et al. Microtubule actin cross-linking factor 1, a novel potential target in cancer. Cancer Sci. 2017;108:1953-8.
[35] Zhang H, Qi J, Wu Y-Q, Zhang P, Jiang J, Wang Q-X, et al. Accuracy of early detection of colorectal tumours by stool methylation markers: A meta-analysis. World J Gastroenterol. 2014;20:14040-50.
[36] Zhao Y, Guo S, Sun J, Huang Z, Zhu T, Zhang H, et al. Methylcap-Seq reveals novel DNA methylation markers for the diagnosis and recurrence prediction of bladder cancer in a Chinese population. PLoS ONE. 2012:7:e35175.
[37] Zhang Q, Li J, Tan XP, Zhao Q. Effects of ME3 on the proliferation, invasion and metastasis of pancreatic cancer cells through epithelial-mesenchymal transition. Neoplasma 2019;66:896-907.
[38] Foy J-P, Pickering CR, Papadimitrakopoulou V, Jelinek J, Lin SH, William Jr WN, et al. New DNA methylation markers and global DNA hypomethylation are associated with oral cancer development. Cancer Prev Res. 2015;8:1027-35.
[39] Chatterjee R, Das S, Chandra A, Basu B. Epigenome-Wide DNA Methylation Profiles in Oral Cancer. In: Wei LK, editor. Computational Epigenetics and Diseases: Academic Press; 2019. p. 213-31.
[40] Basu B, Chakraborty J, Chandra A, Katarkar A, Baldevbhai JRK, Chowdhury DD, et al. Genome-wide DNA methylation profile identified a unique set of differentially methylated immune genes in oral squamous cell carcinoma patients in India. Clinical Epigenetics. 2017;9:13.
[41] McCormack JG, Halestrap AP, Denton RM. Role of calcium ions in regulation of mammalian intramitochondrial metabolism. Physiology Reviews. 1990;70:391-425.
[42] Morandi L, Gissi D, Tarsitano A, Asioli S, Gabusi A, Marchetti C, et al. CpG location and methylation level are crucial factors for the early detection of oral squamous cell carcinoma in brushing samples using bisulfite sequencing of a 13-gene panel. Clinical Epigenetics. 2017;9:85.
CRediT authorship contribution statement
E.M. Okoturo: Data curation, Formal analysis, Funding acquisition, Investigation, Writing – original draft, Writing – review & editing. D. Green: Data curation, Formal analysis, Visualization, Writing – original draft, Writing – review & editing. K. Clarke: Data curation, Visualization, Writing – review & editing. T. Liloglou: Data curation, Formal analysis, Writing – review & editing. M.T. Boyd: Conceptualization, Writing – review & editing. R.J. Shaw: Conceptualization, Funding acquisition, Resources, Writing – original draft, Writing – review & editing. J.M. Risk: Conceptualization, Data curation, Funding acquisition, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing.
00081-0&id=gr1.jpg){kind=link}
00081-0&id=gr2.jpg){kind=link}
00081-0&id=gr3.jpg){kind=link}