Clinical validation of methylation biomarkers for optimal detection of high‐grade vulvar intraepithelial neoplasia

The precursor lesions of vulvar squamous cell carcinoma (VSCC) include human papillomavirus (HPV)‐associated and HPV‐independent squamous neoplasia with a varying cancer risk. Our study aimed to validate the accuracy of previously identified DNA methylation markers for detection of such high‐grade vulvar intraepithelial neoplasia (VIN). A large clinical series of 751 vulvar lesions, originally diagnosed as high‐grade VIN, were reassessed and categorized into HPV‐associated or HPV‐independent vulvar disease categories. Together with 113 healthy vulvar controls, all samples were tested for 12 methylation markers with quantitative multiplex methylation‐specific PCR (qMSP). Performance of individual markers and selection of an optimal marker panel for detection of high‐grade VIN was determined by logistic regression analysis. SST was the best‐performing individual marker (AUC 0.90), detecting 80% of high‐grade VIN cases, with excellent detection of HPV‐independent VIN (95%), known to have the highest cancer risk. Merely 2% of controls tested methylation positive for SST. Selection of a marker panel, including ZNF582, SST and miR124‐2, resulted in a comparably high accuracy for detection of high‐grade VIN (AUC 0.89). In conclusion, we clinically validated the accuracy of 12 DNA methylation markers for detection of high‐grade VIN. SST, as a sole marker or in a panel, provides an optimal diagnostic tool to distinguish high‐grade VIN in need of treatment, particularly HPV‐independent VIN, from low‐grade or reactive vulvar lesions. These findings warrant further prognostic validation of methylation biomarkers for cancer risk stratification of patients with VIN.

independent VIN, from low-grade or reactive vulvar lesions. These findings warrant further prognostic validation of methylation biomarkers for cancer risk stratification of patients with VIN.

K E Y W O R D S biomarker, DNA methylation, vulvar intraepithelial neoplasia
What's new?
Only a minority of vulvar intraepithelial neoplasia progress to cancer, and objective molecular biomarkers that reflect the cancer risk in patients are currently needed. Our study on more than 800 specimens validates the accuracy of 12 previously identified DNA methylation markers for detection of high-grade vulvar intraepithelial neoplasia. SST, as a sole marker or in a panel, provides an optimal diagnostic tool to distinguish high-grade vulvar intraepithelial neoplasia in need of treatment from low-grade or reactive vulvar lesions. Identification of vulvar lesions with a high cancer risk by methylation markers can contribute to personalized patient management and prevent mutilating overtreatment.

| INTRODUCTION
High-grade vulvar intraepithelial neoplasia (VIN) is the precursor lesion of vulvar squamous cell carcinoma (VSCC) and is categorized as human papillomavirus (HPV)-associated vulvar high-grade squamous intraepithelial lesion (vHSIL) or HPV-independent VIN, of which the most renowned subtype is differentiated VIN (dVIN). 1,2 VHSIL accounts for the majority of high-grade VIN lesions and is diagnosed in relatively younger patients. HPV-independent VIN usually arises in a background of lichen sclerosus (LS) or lichen planus (LP), chronic dermatoses of the anogenital area and is mostly diagnosed in postmenopausal women. 1,3 The incidence of both VSCC and its precursor lesions is rising. [4][5][6] Patients with vHSIL and HPV-independent VIN have a varying risk of developing cancer, with a 10-year cumulative VSCC risk of 10% vs 50%, respectively. 6,7 Due to this risk, patients with vHSIL and HPV-independent VIN frequently undergo surgical interventions resulting in physical and psychosexual morbidity. 8 However, as only a minority of vHSIL patients develop cancer, prognostic biomarkers reflecting the cancer risk could reduce the necessity for mutilating overtreatment. In contrast, HPV-independent VIN has a high cancer risk, yet is a challenging diagnosis. With regards to clinical presentation, HPV-independent VIN can resemble non-neoplastic lesions such as LS or LP. Histological characteristics can be subtle and have overlapping features mimicking reactive or inflammatory conditions such as LS, squamous hyperplasia or vulvar low-grade squamous intraepithelial lesions (vLSIL). 2,9,10 As a consequence, misdiagnosis of HPVindependent VIN may occur, leading to undertreatment of patients with such potentially aggressive precursor lesions. Therefore, cancerpredicting biomarkers can potentially aid in the identification of HPVindependent VIN during diagnostics.
Host cell DNA methylation is known as a hallmark in HPV-induced carcinogenesis which can lead to silencing of tumor suppressor genes. 11 In fact, DNA methylation is already appreciated as a biomarker for detection of lesions with a high cancer progression risk in cervical and anal neoplasia. [12][13][14][15] Previously we have shown that several genes showed higher methylation levels with increasing severity of disease in a well-defined cross-sectional series of vulvar lesions, including healthy vulvar tissue, high-grade VIN, high-grade VIN adjacent to VSCC and VSCC cases. 16 The purpose of the current study was to validate the accuracy of previously identified DNA methylation markers for detection of high-grade VIN in a large, revised clinical series of vulvar lesions.

| Study samples and ethics
Patients diagnosed with high-grade VIN without prior or concurrent VSCC were selected from a population-based historical cohort provided by the Dutch Nationwide Pathology Databank (PALGA), as described previously. 6,17 This historical cohort contains all vulvar pathology reports of patients diagnosed with vulvar LS, VIN and/or VSCC in the provinces Noord-Holland and Flevoland between 1991 and 2011. 18 A total of 894 patients with high-grade VIN were identified. 6 Of each patient, archived formalin-fixed paraffin-embedded (FFPE) tissue blocks of the incident high-grade VIN lesion (ie, first biopsy with high-grade VIN diagnosis) were requested. In addition, healthy vulvar tissue samples were included as a control group. These samples were collected between 2018 and 2021 from healthy patients during esthetic or reconstructive genital procedures in V-Klinieken, Leiden and Amsterdam UMC, location VUmc, respectively.

| Sample processing
Retrieved FFPE tissue blocks were sectioned using the sandwich method, in which the first and last sections were stained with hematoxylin-eosin (H&E). In-between sections were cut for immunohistochemistry (IHC) staining. Additional in-between sections were used for DNA isolation and subsequent HPV genotyping and DNA methylation analysis. Healthy vulvar tissues were only subjected to DNA methylation analysis.

| HPV genotyping methods
High-risk (hr-)HPV DNA testing was performed using the QIAscreen HPV PCR test (QIAgen, Hilden, Germany), according to manufacturer's instructions. The assay is directed against the HPV E7 region and detects hr-HPV type 16 and 18, as well as 13 other hr-HPV genotypes (ie, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 67 and 68) as a pool. 19 Low-risk (lr-)HPV DNA testing was performed for selected cases upon histological reassessment when deemed necessary to reach a final diagnosis, such as hr-HPV negative cases which had a viral morphology, HPV-independent VIN cases with a p53 wild-type pattern and inconclusive cases. To detect the presence of lr-HPV, the L1-region-based GP5+/6+ PCR was used followed by enzymatic immunoassay (EIA). 20 Samples with a positive EIA result were subjected to a reverse line blot genotyping assay. Samples with a β-actin threshold cycle (Ct) value of >32 were regarded as invalid, as this indicated poor sample quality due to insufficient DNA or inadequate bisulfite conversion. 21 Methylation levels were normalized to the reference gene (β-actin) and calibrator of each multiplex using the comparative Ct method (2 ÀΔΔCt Â 100) to obtain ΔΔCt ratios. 22

| Statistical analysis
Methylation levels were visualized for each individual marker by constructing boxplots of the log2 transformed ΔΔCt ratios for each disease category. VSCC and inconclusive cases were excluded from this analysis due to low sample sizes. Significant differences between categories were assessed for each marker by the Kruskal-Wallis test, followed by pairwise post hoc Mann-Whitney U testing with T A B L E 1 Baseline characteristics of the study population.  F I G U R E 1 DNA methylation levels of 12 genes across five vulvar disease categories. Number of valid cases per disease category is presented for each multiplex in the legend. Differences between categories were tested by the Kruskal-Wallis test, followed by pairwise post hoc Mann-Whitney U testing with Bonferroni correction for multiple comparisons; ns, not significant; *P < .05; **P < .01; ***P < .001; ****P < .0001.

| Study population and disease categories
Of the historical cohort of 894 patients with high-grade VIN, FFPE tissue blocks were retrieved from 791 patients (88%). Subsequently, 40 cases were excluded due to insufficient residual tissue for any further processing. In addition, 113 healthy vulvar samples were collected. This resulted in a total series of 864 vulvar tissues which were included in the study.
Upon histological reassessment of the originally diagnosed highgrade VIN lesions (n = 751), the majority of cases were categorized as HPV-associated vHSIL (n = 575) or vLSIL (n = 87) (  A high variety in methylation levels was found in vHSIL, whereas methylation levels in dVIN were more consistently high. Although there was a slight increase in methylation levels from controls towards the nondysplastic and the vLSIL groups, these differences were mostly not statistically significant.

| Performance individual markers
All markers showed a moderate to good individual performance to dis-   In accordance with our earlier findings, we found a strong correlation between methylation levels and severity of vulvar diagnosis in both HPV-associated and HPV-independent lesions. 16 Particularly for challenging HPV-independent precursor lesions that mimic reactive or inflammatory nondysplastic lesions, the use of these methylation assays can be very useful as an adjunct diagnostic tool.
Most markers showed significantly higher methylation levels in dVIN compared to vHSIL, which is consistent with the higher cancer risk in patients with dVIN. 6,7 Interestingly, a more heterogeneous methylation pattern was observed within the vHSIL group, likely reflecting a varying cancer risk. A previous study by our research group also identified SST as one of the best-performing methylation markers for cancer risk stratification in HPV-induced anal lesions. 23 Longitudinal studies are needed to further explore the prognostic value of methylation markers for cancer risk stratification in vulvar lesions as well.
Even though we tested 12 DNA methylation markers primarily known to be associated with HPV-induced carcinogenesis, 14,15 we acknowledge that the most common etiopathogenic pathway of development towards VSCC is independent of HPV. While to a certain extent different epigenetic alterations may be expected in HPVassociated vs HPV-independent vulvar premalignant lesions, a recent genome-wide methylation study on vulvar cancers also showed overlapping DNA methylation events between these two pathogenic pathways. 24 In that study, Dasgupta et al identified 199 genes which were differentially methylated in VSCC compared to healthy vulvar tissue.
These included 5 out of the 12 markers that were analyzed in the present study, including markers SST and ZNF582 from our marker panel.
Since the study by Dasgupta et al included more HPV-independent VSCC than HPV-associated VSCC (15 vs 3 cases, respectively), this further supports the potential of these markers for the detection of HPV-independent vulvar lesions. 24 In addition, some of our markers tested are known as pan-cancer markers, including GHSR and SST. 25,26 These markers are known to be hypermethylated in a variety of cancers, and therefore a potential biomarker for lesions arising from different pathogenic pathways. 26,27 Consistently, SST was also the best-performing individual marker in our previous study to distinguish high-grade VIN without adjacent VSCC from healthy vulvar controls (AUC 0.93), irrespective of HPV status. 16 SST encodes a growth-regulatory peptide hormone (somatostatin)