The impact of HPV infection on risk of progression and overall mortality in vulvar squamous cell carcinoma: a retrospective single-center analysis

Aims: To investigate the impact of Human Papillomavirus (HPV) infection on clinic and histopathologic characteristics, and prognostic factors of patients affected by vulvar squamous cell carcinoma (VSCC). Methods: Fifty-six patients diagnosed with VSCC at the IRCCS Fondazione Policlinico San Matteo, Pavia, Italy, from March 2001 to February 2016, were enrolled in a retrospective analysis. HPV DNA was detected by the INNO-LiPA HPV genotyping assay, version EXTRA II, on corresponding pathological specimens. Clinic and histopathologic characteristics were compared through Fisher's exact test. Kaplan–Meier survival curves and Cox regression models were used to analyze


Introduction
Vulvar cancer represents less than 5% of all gynaecological malignancies, with an incidence of 2-3 per 100000 women per year in developed countries [1]. More than 90% of vulvar tumours are squamous cell carcinomas. Two different pathogenetic pathways have been historically described: one related to Human Papillomavirus (HPV) infection and another HPV-independent, that usually develops in women affected by chronic vulvar dermatosis [2,3]. The carcinogenesis related to HPV-persistent infection accounts for 43-60% of vulvar squamous cell carcinomas (VSCCs) and is especially caused by high-risk (HR) HPV genotypes, in particular HPV 16, but also HPV 18, 31, 33 and 45 [4].
These two kinds of VSCC are characterized by different clinic and histologic features. Nevertheless, there are still poor and controversial data about the effect of HR HPV infection on prognosis, probably due to the low incidence of the disease and the heterogeneous distribution of patients regarding stage of disease and type of treatment [5,6]. Several authors have already investigated prognostic factors for relapse and survival in patients affected by VSCC, that seem to be first related to nodal involvement [7,8]. Some authors did not prove a significant difference of survival in patients with HPV-positive and HPV-negative vulvar cancers, whereas others highlighted a better survival rate in HPVrelated carcinomas [9,10]. A better prognosis has been previously reported in HPV-positive oropharyngeal and anal cancers and has been related to an historically-documented better response to chemo-and radiotherapy [11][12][13]. Nowadays, these evidences have not still been confirmed in HPV-related VSCCs.
Therefore, the aim of this study was to investigate clinic and histopathologic characteristics and prognostic factors of patients affected by VSCC and treated in a single Institute. In addition, we analysed the impact of HPV infection on characteristics of tumour, risk of relapse and survival rate, with a particular attention to HR HPV genotypes and multiple infections.

Population
All patients diagnosed with vulvar cancer, who underwent surgical treatment at the Department of Obstetrics and Gynaecology of the IRCCS Fondazione Policlinico San Matteo, University of Pavia, Italy, from March 2001 to February 2016, were retrieved from our file archives and enrolled in a retrospective analysis. The Institutional Review Board of our hospital approved the study and all patients signed an informed consent for the use of sensitive data for health research before surgery. Patients were included if the following criteria were met: (i) age at diagnosis of 18 years or older; (ii) histologic confirmation of VSCC; (iii) available data about follow-up longer than 3 months. Patients were excluded in case of non-squamous vulvar cancers.
Data regarding principal socio-demographic and clinic characteristic of patients were recorded. Macroscopic and histologic characteristics of tumour were recovered from surgical and pathological reports. Tumour stage was assigned according to the 2009 International Federation of Obstetrics and Gynecologists (FIGO) [14]. Primary treatment modality, adjuvant chemo-and/or radiotherapy as well as type, time and treatment of any recurrence were registered. Follow-up was routinely scheduled every 3-4 months during the first 2 years and thereafter every 6 months at the dedicated Vulvar Pathology Clinic of our Department. A dedicated database was prospectively filled in at each follow-up visit. To improve accuracy of survival data, telephonic interviews were allowed in case of lost to follow-up patients.

Statistical analysis
The results have been reported as medians, absolute frequencies and percentages. Fisher's exact test was used for comparisons between nominal variables. Age was considered either as ordinal or as nominal variable with a cut-off at 75 years. Disease-free survival (DFS) was defined as the time from treatment to local and/or regional recurrence or metastasis. Overall survival (OS) was defined as the time from the date of treatment to the date of death of disease or to the last follow-up visit. For inferential analysis, the follow-up of patients was truncated at 5 years. Deaths from other causes not related to vulvar cancer and without documented progression or relapse were censored at the date of death. Outcome data, including "no evidence of disease" (NED) versus progression/recurrence of disease and alive patient versus dead patient, were described by the Kaplan-Meier survival curves, that were compared by use of the log-rank test.
Univariate Cox models were used to analyze the factors studied as prognostic factors for DFS and OS using the risk estimation as hazard ratio (HR) and their 95% confidence interval (95% CI). Multivariate regression models (Cox proportional hazard model) were applied to adjust group comparisons to possible confounders regarding progression/recurrence and mortality. Multivariate regression models included all statistically significant variables based on univariate Cox models and HPV-positive status. Data were analyzed with the R freeware statistical software (R version 3.3.0). P-value of .05 or less was considered statistically significant.

Discussion
In our retrospective analysis, among 56 women affected by VSSC, diagnosed and treated in a single Institute, HPV-DNA was detected in about 52% of corresponding pathological specimens, through a widely validated HPV-genotyping assay, that shows excellent sensitivity and accuracy on FFPE tissues [16]. The most prevalent genotype in our population was HPV 16, as already reported by several Authors [17,18]. However, the other more frequent genotypes were HPV 52 and HPV 44, that were never related before to VSCC [4,19]. This could be the result of changes of HPV genotypes distribution, consequent both to the introduction of new HPV genotypes, due to immigration, and different patterns of exposure, related to social and behavioural variables, as already shown by Dal Bello et al. in Italian women affected by cervical intraepithelial neoplasia (CIN) [20]. We compared two groups of VSCC patients, HPV-dependent and HPV-independent, characterized by a similar sample. In literature, several rates of HPV-positive VSCC have been reported, between 15% and 79%, and this wide range is probably related to different HPV-DNA detection techniques, histologic type of vulvar cancer and geographical areas [6,[21][22][23]. HPV-positive VSCCs usually affect younger patients, with a higher prevalence of immunosuppression, and are significantly associated to uVIN [2,24,25]. All these evidences were confirmed in our population study, also when considering HR HPV and multiple HPV genotypes infections, as well as in HPV-related CIN [26]. Our study could be considered as an original analysis because it explored disease-and treatment-specific characteristics of VSCC not only in relation to HPV infection but also to HPV class-risk. According to our analysis, HR HPV-positive VSCCs were significantly associated to lower rate of ulceration and stromal invasion and appeared as FIGO stage I in more than 80% of cases, therefore not requiring lymph node dissection in about 57% of tumours. Other common but not statistically significant features of this group of VSCCs included tumour size < 20 mm, unilateral lesion and lower incidence of lymph node metastasis, with subsequent lower need of adjuvant radiotherapy. This could explain the significantly higher rate of mortality for any cause in HPV-negative VSCC patients, that were usually older at the time of diagnosis and showed more aggressive disease-specific characteristics, especially a higher prevalence of lymph node metastasis and FIGO stage > I. Instead, a higher but not statistically significant incidence of progression/relapse was found in HPVdependent VSCC, probably related to higher rate of positive surgical margins. Nevertheless, HPV-positive VSCC patients were characterized by lower overall mortality, probably because of less aggressive disease-specific features and subsequent more successful treatment of relapse. In our cohort it was not possible to investigate any correlation between better prognosis of HPV-dependent VSCCs and relevance of radiotherapy for control of the disease, as shown for HPV-positive head and neck tumours, since all relapses of HPV-positive tumours were successfully treated with surgical excision [11,12]. On the contrary, in 2016 Lee et al. demonstrated that the presence of HPV was significantly related to better 5-year PFS and OS and lower 5-year rate of in-radiation field relapse, but this retrospective analysis included only patients with VSCC who received radiotherapy with or without surgical resection [19]. In agreement to what reported by previous studies, both DFS and OS were significantly decreased in case of lymph node metastasis, and especially of lymph node extracapsular spread. In addition, positive surgical margins affected the risk of progression/relapse of the disease, whereas age ≥ 75 years, FIGO stage > I and adjuvant radiotherapy increased the risk of mortality for any cause. All these results were statistically significant both at univariate and multivariate Cox regression analysis and confirmed which are the well-known prognostic factors of VSCC [7,27]. Several Authors have already investigated the existing relationship between HPV infection and prognosis of VSCC, showing opposing results. In 2011, Alonso et al. highlighted that HPVpositivity could not be considered as an independent prognostic factor, as evidenced by our study [6]. Other studies have recently shown a better prognosis related to HPV-positive VSCCs, but in most cases they did not supply the method applied for HPV detection [28]. A retrospective chart review of patients with VSCC stage I showed that HPV-positive and p16-positive patients were less likely to recur and had no disease-specific deaths, thus suggesting a less aggressive type of VSCC related to HPV infection. However, the sample size of available biomarker data was limited, p values were all large (p≥0.25) and multivariate analyses to control for confounding factors could not be applied [28]. Also, Tringler et al. reported a better prognosis related to p16 INK4a -positive tumours and van de Nieuwenhof et al. found an increased survival in patients showing uVIN next to VSCC [21,29]. Although p16 INK4a expression and uVIN are notoriously associated to HPV-positivity, they could not be considered as conclusive markers of HPV infection. More recently, Hinten et al. showed that Disease-specific survival (DSS), DFS and OS were significantly better in HPV-related VSCC patients in a large cohort of patients [30]. It is noteworthy that VSCC was considered as HPVrelated VSCC in case of >25% p16 INK4a expression and HPV presence and even if >25% p16 INK4a expression, HPV absence, but uVIN presence. Moreover, only 17% of VSSCs were HPV-related, that is a lower percentage than reported by previous studies [2].
In the end, we could not find a significant association between HPV infection, especially HR HPV and multiple genotypes infection, and risk of progression/relapse of disease and mortality for any cause. However, in our cohort, multivariate regression analysis showed a relevant protective effect of HPV-positivity on risk of disease progression/relapse, in case of concomitant lymph node metastasis and positive surgical margins, and overall mortality in two bivariate analysis, adjusting HPV for lymph node extracapsular spread and FIGO stage > I, respectively. The main weaknesses of our study include bias related to its retrospective single-Institute nature and the small sample size. However, VSCC represents a rare neoplasia of the female genital tract, and the number of enrolled subjects is similar to other single-center studies [31]. In conclusion, no significant association has been proven between HPV infection and risk of progression and overall mortality, even though several clinic and histologic characteristics of VSCC have been related to HR HPV infection. Especially, risk of progression is reduced in HPV-positive patients with node metastasis and positive surgical margins. In addition, HPV infection has a protective effect on overall mortality in case of node extracapsular spread or FIGO stage > I. Nevertheless, multicentre studies with broad sample size are needed to confirm this evidence that could lead to a tailored management of HPVrelated VSCCs.