**4. Discussion**

This study evaluated the HPV positivity rates and the HPV genotypes distribution before and after the switch from a nested-based PCR to a real-time PCR on a three-year observation window. Multivariate analysis showed that the HPV prevalence was not a ffected by the introduction of the new real-time assay. The sensitivity of a nested PCR compared to a real-time amplification method was analyzed by several authors, but the conclusions were not unique [12–14].

On the other hand, some differences in the distribution of HPV genotypes were detected. In particular, the pairwise comparison showed some statistically significant differences in the distribution of HPV-45, 68, 40, 42, and 43. The analysis also highlighted that these differences were due to the QIAcube-Real Time PCR group. Logistic regression analysis only confirmed that the Real Time introduction was associated with an increase of the prevalence of HPV-42 (aOR: 4.08, 95%CI: 1.71–9.73).

Currently, more than 100 distinct molecular tests are available on the global market for the detection of HPV DNA [15]. However, despite the approving of several tests for clinical use in USA and Europe, some tests detect a pool of 12 HPV genotypes (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59) while other tests also include HPV 66 and 68 [15]. On the other hand, some commercially available tests allow specific HPV genotyping [16]. Some studies reported that genotyping techniques might be useful to improve the triage and follow up of HPV infected women [17]. In fact, some authors suggested that the genotypes 16/18 are related to a more elevated progressive risk [18,19].

An important source of variations is that the detection of specific HPV genotypes may be a ffected by the di fferent sensitivities of the currently available genotyping methods.

In a study of Del Pino et al. [16] several genotyping tests (AnyplexTM II HPV 28 Detection System, Linear Array HPV genotyping test, Gp5+/6+ PCR-EIA-RH, CLART HPV2 Assay) were compared with Hybrid Capture 2 and a concordance about 80% or higher was reported. In particular, the comparison of the genotype distribution between AnyplexTM and Linear Array, Gp5+/6+ and CLART2, respectively, showed completely di fferent genotypes in five (4.0%), two (2.3%), and three (2.9%) samples.

In a study of Lim et al. [20], AnyplexTM was compared with MolecuTech REBA HPV-ID and HPV DNAChip and the percentage of HPV genotype agreemen<sup>t</sup> ranged from 93.7% to 100.0%. However, relatively high rates of discordance between the three assays were reported for HPV 31, 42, and 44. The authors also examined the performance of the assays in the detection of five common HPV genotypes (HPV 16, 18, 45, 52, and 58) and reported that the sensitivity rates varied according to the detection method. In particular, MolecuTech showed a low sensitivity for HPV 52 (42.9%) while AnyplexTM and HPV DNAChip had low sensitivities for HPV 45 (25.0%).

Comparison of AnyplexTM with Euroarray on 150 samples by Latsuzbaia et al. [21] showed a Kappa concordance below 0.7 for HPV 40, 68, 73, and above 0.95 for HPV 11, 16, 18, 33, and 59. Interestingly, a statistically significant di fference between assays was detected for HPV 42 genotype since it was more frequently detected by AnyplexTM assay.

Marcuccilli et al. [22] reported that AnyplexTM, compared with HPV Sign Genotyping Test, detected more high risk and low risk genotypes. In particular, better agreements were reported for HPV 16, 18, 35, and 70. Statistically significant differences were reported for HPV 31, 51, 52, 53, 56, 58, 59, 66, 73, 6, 42, 44, 54, and 61. These genotypes, except HPV 73, were more frequently detected by AnyplexTM assay.

Di fferences in HPV genotype detection were also reported by Estrade et al. [23] after comparison of AnyplexTM with PGMY-CHUV assay. In particular, HPV 40, 42, 54, and 68 were significantly more frequently detected by AnyplexTM. On the contrary, HPV 51 was more frequently detected by the PGMY-CHUV assay.

The sequential design does not permit to infer the di fferent sensitivities or specificities of the two diagnostic tests. However, the quasi-experimental approach of the study may be advantageous to acquire some preliminary information to better design subsequent studies or to speculate regarding the clinical usefulness of a new diagnostic assay in a relatively cheap manner. In particular, the latter option may be of some interest in a lack of funds context. To our knowledge, this is the first time that a quasi-experimental before-and-after approach has been applied to the evaluation of a new assay for the HPV evaluation. In particular, the data sugges<sup>t</sup> that the introduction of Real-Time PCR did not affect the HPV prevalence, but it was associated with modification of the distribution of HPV-42. Other studies will be needed to clarify the reasons for these variations. However, other limits of this study must be considered. In particular, the temporal window is quite short and the influence of pluriannual trends cannot be excluded. Moreover, the composition of the population analyzed is not known and the di fferences in the distribution of the HPV genotypes may reflect di fferences in the behaviors or risk factors of the analyzed patients. However, the presence of background trends was also excluded by the logistic regression analysis and this makes it possible to assume that the behavioral and social parameters of the analyzed population have not changed over time.

Despite these limits, such study suggests to carefully evaluate the introduction of a new type test for HPV also regarding the impact on the distribution of HPV genotypes. In particular, it may be considered as a potential confounding factor for evaluation of public health programs to control HPV spreading in the populations. At the same time, our study highlights the importance to carefully evaluate temporal dynamics with multivariate methods. Related to this, the accumulation of more rich surveillance data should be encouraged to ensure that demographic shifts in infection patterns could be accurately monitored.
