**3. Results**

From 1 January 2012 to 31 December 2014, 1742 samples (1605 cervical-vaginal and 137 urethral swabs) from 1433 patients (1328 female and 105 male patients, Female to Male ratio = 12.64) were analyzed.

During the observation period, the number of analyzed samples increased with time from a minimum value of 306 in 2012 to an intermediate value of 666 and to a maximum value of 770 in 2014. Moreover, after the introduction of the real-time assay an increase of analyzed cervico-vaginal swabs (87.66% vs. 94.12%) and a percentage decrease of urethral swabs (12.34% vs. 5.88%) was observed.

From January 2012 to the end of June 2013, 535 samples were extracted by QIAcube System (Qiagen, Hilden, Germany) followed by a nested-PCR technique. From July 2013 to the end of December 2013, 437 specimens were extracted by QIAcube followed by mRT-PCR. From January to December 2014, 770 samples were extracted by MagNa Pure 96 system (Roche) and amplified by mRT-PCR. The HPV prevalence in the three groups was 48.97% (262), 49.65% (217), and 51.16% (394), respectively (Chi-Squared test *p*-value = 0.719).

After stratification for the three extraction-amplification combinations, significant di fferences in HPV types prevalence rates were observed for HPV-45, 68, 40, 42, and 43. In particular, the lowest prevalence for HPV-45 was observed in the Magna Pure-Real Time PCR group (0.00%) while HPV-68, 40, 42, and 43 were less observed in the Qiagen-Real Time PCR group (0.37%, 0.00%, 2.06%, and 0.00%, respectively) (Table 1).


**Table 1.** Human papillomavirus (HPV) genotype prevalence in the three di fferent combination groups of extraction and amplification methods.

NS: Non-Significant; S: Significant. *p*-values were calculated on a 2 × 3 matrix by Fisher's test and then corrected by Benjamini and Hochberg's correction (False Discovery Rate (FDR) <1%).

In particular, after pairwise comparison, QIAcube-real time combination resulted in higher HPV-45 prevalence compared to Magna Pure-Real Time PCR (2.99% vs. 0.00%), in lower HPV-68, HPV-40, HPV-42 prevalence rates compared to both Magna Pure-Real Time PCR and QIAcube-Nested PCR (HPV-68: 0.37% vs. 3.20% and 0.37% vs. 3.90%, respectively; HPV-40: 0.00% vs. 1.83% and 0.00% vs. 2.08%, respectively; HPV-42: 2.06% vs. 10.53% and 2.06% vs. 13.25%, respectively) and in lower HPV-40 prevalence compared to QIAcube-Nested PCR (0.00% vs. 1.82%).

On logistic regression analysis, the introduction of real-time assay was associated with an increase of HPV-42 (aOR: 4.08, 95% CI: 1.71–9.73) (Table 2). Moreover, the logistic regression analysis did not reveal the presence of background trends. The comparison of the estimates of the odds ratios of the models built on the samples dataset and the patients dataset revealed absolute di fferences <1 in the majority of cases (Table 2).


**Table 2.** Evaluation by logistic regression models of introduction of real time assay on HPV and HPV genotypes detection.

Bolded: Statistically significant after BH's correction. Each variable in the first column was the dependent variable (Absence vs. Presence) of the model while the independent variables were the Age (linear variable), the Sex (Male vs. Female), the Year of testing (linear variable), the DNA Extraction method (Qiagen vs. Magna Pure) and the Test Type (Real-Time PCR assay vs. Nested assay). Each regression analysis was performed on both the samples and the patient datasets.
