Characterization of type-specific HPV prevalence in a population of persistent cutaneous warts in Flanders, Belgium

Cutaneous warts are benign skin lesions caused by the human papillomavirus (HPV). Even though they are considered benign, they can have a considerable impact on the quality of life and cause serious illness in certain immunocompromised populations. Studies have shown that the efficacy of wart treatment is dependent on the causative HPV type. Therefore, in this article, we aim to determine the HPV genotype-specific prevalence in cutaneous warts of a Flemish population as part of the Omnivirol-Salycilic acid randomized controlled trial. Swab samples of cutaneous warts (n = 269) were collected during enrollment. The DNA extraction was performed on the automated NucliSENS® easyMAG® system (bioMérieux). The samples were analyzed with two separate in-house PCR assays capable of detecting the most prevalent cutaneous HPV types (i.e. wart-associated HPV qPCR) as well as the most relevant mucosal types (i.e. RIATOL qPCR assay). In total, the type-specific prevalence of 30 distinct HPV genotypes was determined. The beta-globin gene was used as a cellularity control and for viral load quantification. Data concerning wart persistence, previous treatment, wart type, and other relevant wart and patient characteristics was collected through a baseline questionnaire. The study population consisted mostly of persistent warts considering that 98% (n = 263) of the sampled skin lesions were older than six months and 92% (n = 247) had undergone previous treatment. The most prominent wart type was the mosaic verruca plantaris (42%, n = 113). The most prevalent HPV types were cutaneous HPV types 27 (73%, n = 195), 57 (63%, n = 169), and 2 (42%, n = 113). Only 2% (n = 6) of the lesions was HPV negative. The highest median viral loads were observed with HPV27 and 57 (i.e. 6.29E+04 and 7.47E+01 viral copies per cell respectively). The multivariate analysis found significant associations between wart persistence and certain wart types, the number of warts, and HPV genotypes. Based on these findings, persistent warts are more likely to: (1) be verruca vulgaris, verruca plantaris simple or mosaic, (2) to manifest as multiple warts, (3) and to be negative for HPV type 2 or 4. These characteristics can be useful in the clinical setting for future risk stratification when considering treatment triage and management. Trial registration: NCT05862441, 17/05/2023 (retrospectively registered).


Setting and population
Patients were included as part of the OVW-SA001 clinical trial conducted in a Belgian population between 2018 and 2019 29 .The study consisted of a double-blind, single-centered, randomized clinical trial that was conducted by the University of Antwerp, with Sonic Healthcare BeNeLux (Antwerp, Belgium) operating as the central laboratory.Ethical approval was obtained from the Ethical Review Board of the University of Antwerp (B300201734040).The study was conducted in accordance with applicable national regulations, Good Clinical Practice (2005/28/EC) 30 and the Declaration of Helsinki 31 .Informed consent was obtained from all participants prior to study entry.In case of minors, informed consent was obtained from legal guardians before study participation.Recruitment was done via flyers distributed via dermatologists, pharmacists and via own initiatives aiming to reach an immunocompetent population exhibiting one or more cutaneous warts, aged 12 years or older, and agreeing to sign informed consent written in Dutch.The study area comprised the Flemish region with enrichment in the Antwerp metropolitan area.

Study design
In total 269 patients were included, fulfilling inclusion criteria, hereby reaching pre-calculated statistical power.At inclusion, data concerning wart duration, previous treatment, wart type and other relevant wart and patient characteristics were collected through a structured baseline questionnaire (Additional File 1).Full details of clinical trial design and methodology are provided in Redzic et al. 28 .

Study procedures
At inclusion, a photograph of the index wart (i.e. the largest wart) was taken, followed by sampling with a Floq swab (FLOQSwab Copan Diagnostics, Murietta, California, USA).Sample-taking procedures and processing details were done as described by Redzic et al. 32 .The sample was stored in Abbott multi-Collect medium www.nature.com/scientificreports/(Multi-Collect Specimen Collection Kit, Abbott Molecular Inc., Des Plaines, Illinois, USA) and kept at 4 °C prior to further processing.Briefly, DNA extraction was performed according to the optimized method for cutaneous samples on the automated NucliSens easyMAG platform (bioMérieux, Boxtel, The Netherlands).The samples were analyzed for the presence of HPV DNA by the in-house developed wart-associated cutaneous HPV assay, capable of detecting the above-mentioned 12 distinct cutaneous HPV types (i.e.HPV 1, 2, 3, 4, 7, 10, 27, 41, 57,  60, 63, 65)  33 , as well as the RIATOL HPV genotyping qPCR assay, capable of detecting mucosal HPV types 6,  11, 16, 18, 31, 33, 35, 39, 45, 50, 51, 52, 53, 56, 59, 66, 67 and 68 34 .The RIATOL qPCR assay is a fully validated HPV test, used to conduct primary HPV screening for cervical cancer 35,36 .This study used the assay without clinical cut-off at its full analytical potential.Both assays allow for quantitative measurement of viral load after calibration with synthetic gBlock gene fragments (Integrated DNA Technologies, Coralville, Iowa, USA).Sample adequacy (cellularity control) was assessed via amplification of beta-globin, a housekeeping gene, which was similarly applied as a measure of the number of cells present essential for subsequent viral load calculations.In summary, all samples were both qualitatively as well as quantitively tested for the presence of 30 distinct HPV types.A detailed description of assay design and validation can equally be found in Redzic et al. 33 .

Data management and analysis
Data management and analysis were conducted using STATA version 17 (Corporation, College Station, Texas, USA).In order to summarize data regarding patient-and wart-specific characteristics descriptive statistics such as numbers and percentages with 95% CIs were applied for categorical variables, and arithmetic mean, standard deviation, and median were used for numerical variables.The presence of any detectable viral load of a specific HPV type was considered diagnostic for infection with that specific type.
Kruskal-Wallis test was used to compare patients' age, number of warts, number of multiple infections and estimated viral loads between different HPV types.Student T-test was used to assesses the number of warts between the two genders and differences in viral load between single and multiple infections.Results were considered statistically significant at P-value ≤ 0.05.
Potential predictors of wart persistence were assessed via a multiple logistic regression model, in which wart persistence was defined as index wart duration of more than 6 months combined with resistance to previous treatment.Several variables were created for analysis.Age was dichotomized into two categories: ≥ 13 years and < 12 years, as patients younger than 12 years are known to have higher rates of wart clearance 37,38 .The number of HPV infections was treated as a categorical variable, with no HPV as the baseline category, one HPV and two or more HPV multiple infections as other categories.The number of warts was categorized as 1 (baseline) versus 2 or more warts.The type of warts variable was divided into four categories, with verruca plantaris mosaic as the baseline, verruca plantaris simple, verruca vulgaris, and "others" (i.e.verruca plana and filiformis).For the univariate analysis, logistic regression was fitted to measure the strength of the association of potential covariates.For model building, all predictors that had a P-value of less than 0.2 were considered, in addition to the potential confounders hypothesized to be of importance: the type of wart, the number of warts, the number of infections and specific cutaneous HPV types that had more than 10 observations (i.e.HPV 1, 2, 3, 4, 10, 27, 57, 63, and 65).
As no other predictors were identified, a multivariable logistic regression analysis was performed to simultaneously control for potential confounders.For variables that were not dichotomized, interaction terms were fitted, and linear assumptions were tested.Both the Akaike information criterion (AIC) and the Bayesian Information Criterion (BIC) were used as two different measures of model fit, along with the Variance Inflation Factor (VIF), to detect multicollinearity to ensure that yielded values were within acceptable ranges 39 .
To obviate problems related to the stability of parameter estimates that arose for the variable HPV 63 due to the flatness of the likelihood method, the Penalization likelihood method was used instead of the standard Maximum likelihood method 40 .

Ethics approval and consent to participate
Ethical approval obtained from ethics committee of Antwerp University Hospital, B300201734040.All participants gave written informed consent prior to entry into the study.

Socio-demographics and wart-characteristics
Figure 1A provides an overview of the geographical locations were all the study samples originated from, with most samples (89%) being collected from the Antwerp metropolitan region.A summary of the population-and wart-specific characteristics can be found in Table 1.Median age of participants was 39 (7-76) years for males and 38 (12-82) years for females, and 60.2% of the population was female.At study enrollment, 66 patients (24.5%) had one wart, 156 (58.0%) had 2-10 warts and 47 (17.5%) had more than 10 warts.
All subjects are represented in Fig. 2 by the number of warts at the time of enrollment per age group and gender.There was no significant difference in the number of warts between the two genders (Student T-test P > 0.05).However, a significantly larger number of warts was found in the age groups ≤ 15 and 16-25 (Kruskal Wallis P < 0.005), suggesting that adolescents and young adults on average, exhibit a higher number of warts.The highest median number of warts was detected in male patients ≤ 15 years old (i.e.13.5 warts per subject).No significant difference in the number of HPV multiple infections between different age groups was found (Kruskal-Wallis P > 0.05).
Viral loads ranged from 1.11E−03 to 9.31E+06 copies per cell and showed a type-specific manifestation (Table 3).As for cutaneous HPV types, highest median viral loads were observed for HPV 27 (6.29E+04copies per cell) and 57 (7.47E+01 copies per cell), and these differed significantly from viral loads of other types (Kruskal-Wallis P < 0.000001).Other genotypes displayed median viral loads ranging from 6.61E−02 to 2.39E+00 copies per cell for cutaneous types, and from 1.98E+01 to 6.09E+04 for mucosal types.The type-specific viral loads found in single infections did not significantly differ from those found in multiple infections within the same HPV type (Student T-test P > 0.05).

Predictors of wart persistence
After adjusting for all other covariates in the multivariate model, there were four significant associations with the persistence of the index wart (Table 4).Compared to the verruca plantaris mosaic, warts from the "others" category had 86% (95% CI 0.03-0.83)lesser odds of persistence, although the sample size was small (n = 7).Furthermore, as predicted, having two or more warts was significantly associated with the outcome, with patients harboring two or more warts having 2.65 higher odds (95% CI 1.07-6.55) of persistence than patients with one  As regards to warts containing more than two distinct HPV types 50/135 (37%) were verruca plantaris mosaic, 45/135 (33%) verruca vulgaris, 37/135 (27%) verruca plantaris simple, and 3/135 (2%) verruca plana.(B) HPV type-specific prevalence in cutaneous warts.A total of 694 HPV infections was detected in the study population.Cutaneous HPV types 27 (28%, 195/694), 57 (24%, 169/694), and 2 (16%, 113/694) were the most commonly found types, with HPV type 7 being the exception, as it was not detected in this population.Certain mucosal HPV types (i.e.HPV 6, 31, 39, 51, 52, 53, 59, 66 and 68) were detected in low percentages (< 2%) and are depicted in group (red).The HPV type-specific distribution according to number of multiple infections is displayed as well with the number of subjects in each group between brackets.

Discussion
In this study, a vast prevalence of HPV was found in cutaneous warts, with only a limited number of warts identified as negative for the analyzed HPV genotypes (2%).From a pre-analytical perspective, our findings confirm the efficient collection of test samples via a non-invasive, swab-based technique.The efficiency of painless collection via surface swabs has been described rendering swab samples a reliable tool to test for viral presence in skin lesions 32,42,43 .It can be expected that by circumventing the need for invasive sample collection, the potential of HPV genotyping in cutaneous lesions will be more easily exploited as patient's reluctance to undergo testing will be substantially reduced as well as the need for highly experienced professionals to apply the technique 43 .In concordance, the observed high HPV positivity rate also confirms efficiency at the analytical level.The cutaneous wart-associated HPV genotyping assay was carefully designed to ensure maximum sensitivity 33 , simultaneously allowing viral load determination.Due to this high sensitivity, HPV types were detected with high accuracy, resulting in the identification of at least one cutaneous HPV type in 97% of all samples, with HPV 27 (73%), 57 (63%) and 2 (42%) being the most prevalent types.This high cutaneous HPV type-specific  www.nature.com/scientificreports/prevalence further confirms that the most relevant HPV genotypes were selected and included in the cutaneous wart-associated HPV assay.Likewise, the RIATOL qPCR HPV assay, capable of detecting mucosal HPV types, has been exploited at full analytical potential without the use of the previously described clinical cut-off for primary cervical cancer 35 .This assay is highly sensitive and allows viral load determination even at low cell counts.The high sensitivity of the employed assays correspondingly explains the high prevalence of multiple infections (76.2%), which is in contrast with current literature reporting only 4-46% multiple infections in immunocompetent subjects using similar sampling techniques 27,42,44,45 .However, Schmitt et al. did report the highest prevalence of multiple infections in swab samples currently published (46%), and confirmed that use of a more sensitive method leads to a superior ability to detect multiple HPV infections in the same swab sample.
Thus far, the largest study regarding HPV prevalence in cutaneous warts was conducted by Bruggink et al. and comprised 744 warts of 246 immunocompetent patients 27 .The lower prevalence of multiple infections found, can be explained by the fact that in their study individual warts are considered as the unit of analysis instead of patients i.e. multiple warts of the same patients were sampled for further HPV analysis, propagating therefore repetitive results in the population.Furthermore, exclusively 'new' warts were included in their study, defined as warts without any prior treatment from a general practitioner or dermatologist 27 .Considering that in the current study, 92% of subjects were already subjected to prior treatment, the high percentage of multiple infections could be an indicator of persistence.In addition, Bruggink et al. excluded mosaic warts with a diameter of ≥ 1 cm from their study, while this type of warts compromised 20% (n = 55) of our population 27 .
To the best of our knowledge, the most comprehensive study exploring HPV genotyping and including a viral load component was the study conducted by Skubic et al. 46 .In contrast, the latter study focused on histological specimens, hereby generating strong evidence on genotype distribution and investigating the role of HPV viral load in cutaneous warts 46 .Our findings, obtained by the collection of cutaneous swabs, largely confirm the findings by Skubic et al., showing the highest prevalence of HPV 27 and 57, with comparable levels of viral load distribution as well, despite essentially different starting material.Based on observations by de Koning et al. as well as Garcia-oreja et al., high concordance between cutaneous swabs and biopsies can be expected, thus strengthening findings from this study 42,43 .
In this population, HPV 27 and 57 exhibited significantly higher viral loads in comparison to other types, correlating with high viral shedding and associated highest infectious potential, which is translated in highest prevalence in the study population 42,47,48 .
The multivariate analysis found significant associations between wart persistence and certain wart types, number of warts and HPV genotypes.Having verruca plana or filiformis significantly decreased the risk of persistence in comparison to verruca plantaris mosaic.The mosaic wart is already recognized as the most persistent wart type and was as well enriched in this study population (42%) 37 .Number of warts being a risk factor for persistence is to be expected, knowing that persistent warts are resistant to treatment and, by definition, have a duration longer than 6 months, which increases the risk of autoinoculation and transmission of the HPV infection from one body site to another.As regards to the role of HPV genotyping, seeing that the high viral load of HPV 27 www.nature.com/scientificreports/and 57 results in a subsequent high infectious potential, these types appear to be omnipresent in the population, not only in persistent warts but also in warts that are defined as not persistent 3,27,46 .This finding makes them an inadequate predictor of risk.However, the logistical regression did show that some HPV types are in fact, less likely to cause persistent warts i.e.HPV 2 and 4 (OR 0.38 and 0.31 respectively).HPV 4 has previously been identified by Bruggink et al. as having the most favorable natural course in plantar warts (cure rate 94% (95% CI 73-99%)) 25 .HPV 2, however, did not have a promising natural course in their analysis (cure rate 0% (95% CI 0-26) in plantar warts and 3% (95% CI 0-16) in common warts).Claims about cure rates of other HPV types were also not possible, due to a lack of sufficient numbers per HPV type in different treatment groups.Despite certain discrepancies, our findings together with Bruggink et al. do consolidate the future applicability of HPV genotyping in clinical management of cutaneous warts.

Conclusions
Given the high prevalence of HPV types in cutaneous lesions, and their relationship with the clinical manifestation of warts, HPV genotyping with viral load determination can be of added value in daily clinical practice.
Combined with the knowledge that the persistence of a lesion is linked to both the HPV genotype, wart type and number of warts, risk assessment can be performed, and therapeutical options can be suggested.In case that the above-defined risk factors are present in a subject, a more aggressive treatment approach can be applied in order to inhibit not only the spread of the infection in the general population but also further autoinoculation.Previous studies are in line with these findings.However, further research is needed to compose a comprehensive tool to guide clinicians in wart treatment and/or follow-up.For a more extensive assessment of HPV-dependent wart persistence, upcoming studies must entail large clinical trials comprising not only different treatment groups (e.g.salicylic acid, cryotherapy and wait-and-see) but also ensuring a sufficient HPV type-specific allocation in each group.Only then, a head-to-head comparison of different genotypes is possible, creating the possibility to make distinct claims about treatment response and natural course based on HPV type.Nevertheless, the present study already provides certain insights in possible predictors of wart persistence and encourages a more patientcentered and directed approach to wart risk and treatment stratification.

Figure 1 .
Figure 1.(A) Origin of samples collected during the OVW-SA001 clinical trial.Darker blue colors represent higher density of collections per geographical area.(B) Index wart treatment history.The Veen-diagram depicts the number of patients that applied different types of treatments on their index warts.In summary, 22 index warts did not have any previous treatment, while cryotherapy was the primary treatment for most index warts (n = 201), followed by salicylic acid (n = 154) and various other types of therapy (n = 119).The most common treatment combination was cryotherapy combined with salicylic acid (n = 65).

Figure 2 .
Figure 2. Number of warts per age group in the male and female population.Each study participant is depicted by a colored dot.Male patients are shown in blue, female patients in red.The number of patients per age group was respectively: 33 [≤ 15], 46 [16-25], 35 [26-35], 55 [36-45], 39 [46-55], 37 [56-65], and 24 [≥ 66].The median number of warts per each age group is depicted by a colored line, with the highest median number of warts (13.5) found among patients aged ≤ 15 years.The highest number of warts per patient was found in age group ≤ 15 years and equaled to 54 warts in total.

Table 1 .
Distribution of patient-and wart-specific characteristics among the study population.
Variable Number of patients (n = 269) Percentage (%) Vol.:(0123456789) Scientific Reports | (2023) 13:17492 | https://doi.org/10.1038/s41598-023-44154-ywww.nature.com/scientificreports/HPV type-specific prevalence As regards to sample quality, 261 (97.0%) samples were considered valid as demonstrated by a sufficient B-globin amplification.Moreover, patients showed a high HPV positivity rate where only 6 (2.2%) lesions were HPV negative, while 58 (21.6%) were positive for a single HPV genotype, and 205 (76.2%) contained multiple HPV infections.A high level of multiple infections was observed in 135 (50.2%) cases harboring three or more HPV infections.The maximum number of multiple infections in one patient was seven.Figure 3A provides an illustration off all subjects based on number of multiple infections per specific index wart type.The HPV multiple infection status is not dependent on wart type, seeing that similar distributions of number of infections can be found in all wart types identified.The most prevalent HPV types were cutaneous HPV types 27 (195/269; 72.5%), 57 (169/269; 62.5%) and 2 (113/269; 42.0%) (Fig.
. Concerning HPV genotypes, only HPV 2, and HPV 4 were found to be statistically associated with the outcome, with their presence resulting in a respective decrease of 66% and 70% odds of index wart persistence.

Table 2 .
Most prevalent multiple infection trends in HPV-positive index warts.

Table 3 .
Viral load quantification of identified HPV types.The table depicts the number of index warts (n) positive for each HPV type and the median viral load per specific type.Viral load is expressed as number of viral copies per cell.Viral load inter quartile range (IQR) per type is likewise described.

Table 4 .
Association of several covariates with wart persistence: multivariable logistic regression analysis.