*2.4. Statistical Analysis*

All of the statistical analyses were performed using SPSS statistics software Version 21.0 (Chicago, IL, USA). Testing the difference in frequencies of attributive features was performed using the Chi-square (χ 2 ) test of independence and quality of the match. The Student's t-test was used to compare values between the two age groups, a numerical characteristic. A one-way analysis of variance (ANOVA) and the Bonferroni post-hoc test were applied to compare values between three or more data groups. Frequencies were used to present the analysis of the oncogenic activity of multiple HR HPV infections. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and their 95% confidence intervals (CIs) of HR HPVs DNA and E6/E7 mRNA HPVs detection and cytology test were calculated. To quantify the diagnostic capabilities of the selected test and evaluate its significance, the receiver operating characteristics (ROC) curve was used, which enables testing the significance of differences in the discriminating potential of different variables for the same binary outcome. It is based on a graphical presentation of pairs of sensitivity and specificity that can be obtained by estimating the threshold value for all values of discontinuous variables of the sample. Univariate and multivariate logistic regression were used to analyze the connection between two or more features, generating adequate statistical models. Multivariate logistic regression analysis was applied to all of the analyzed factors to construct a predictive model and named the most relevant predictors for the development of HSIL. A *p*-value of less than 0.05 defined as statistically significance.

#### **3. Results** used, which enables testing the significance of differences in the discriminating potential

#### *3.1. Cervical Cytology* of different variables for the same binary outcome. It is based on a graphical presentation

A total of 365 specimens obtained from women in the north part of the Republic of Serbia (Vojvodina) were classified based on the Bethesda System 2014 into four categories by cytological criteria. of pairs of sensitivity and specificity that can be obtained by estimating the threshold value for all values of discontinuous variables of the sample. Univariate and multivariate logistic regression were used to analyze the connection between two or more features,

generating adequate statistical models. Multivariate logistic regression analysis was ap‐

#### *3.2. HR HPV DNA in Cervical Samples* plied to all of the analyzed factors to construct a predictive model and named the most

The cervical samples were analyzed for 12 HR HPVs, where 246 out of 365 (67.4%) had HPV-DNA-positive results, which indicates that the overall prevalence of HPV in the study population was 67.4%. All of the HPV genotypes covered by the HPV DNA test were identified (*n* = 274) in the study population (246 HR-HPV-positive cervical samples). The most prevalent HPV genotype is HPV 16 which makes up 38.3% (105/274) of the total HPV-detected genotypes in 42.7% (105/246) of HP- DNA-positive samples. HPV 31 takes second place with 17.2% (47/274) of total HPV-detected genotypes in 19.1% (47/246) of HPV-DNA-positive samples. Equally represented are HPV 33 and HPV 51, each with 8.8% (24/274) of total HPV-detected genotypes in 9.8% (24/246) of HPV-DNA-positive samples (Figure 1). The results show that those HR HPVs make up 73% (200/274) of the detected genotypes, including multiple infections, which determined those cervical samples (*n* = 172) for further examination of oncogenic activity, according to the study's aim. Multiple HPV infections were found in 15.7%. The most common co-infections were those with HPV 16 and 31, found in 7.6% (13/172) of cases with multiple infections (Table 2). relevant predictors for the development of HSIL. A *p*‐value of less than 0.05 defined as statistically significance. **3. Results** *3.1. Cervical Cytology* A total of 365 specimens obtained from women in the north part of the Republic of Serbia (Vojvodina) were classified based on the Bethesda System 2014 into four categories by cytological criteria. *3.2. HR HPV DNA in Cervical Samples* The cervical samples were analyzed for 12 HR HPVs, where 246 out of 365 (67.4%) had HPV‐DNA‐positive results, which indicates that the overall prevalence of HPV in the study population was 67.4%. All of the HPV genotypes covered by the HPV DNA test were identified (*n* = 274) in the study population (246 HR‐HPV‐positive cervical samples).

Molecular data were compared with the cytological results and age categories of patients. The distribution of cytological groups was analyzed within the most prevalent HR-HPV-DNA-positive samples (172 cervical samples), including multiple infections. A minority of women, 16.9% (*n* = 29), had normal results, whereas 83.1% (*n* = 143) showed different abnormalities. A total of 26.7% (*n* = 46) of the examined women had ASCUS; in 25.6% (*n* = 44), LSILs were found, whereas HSILs were detected in 30.8% (*n* = 53). The mean age of the patients was 36.7 years. Among the specimens, the number of Serbian women who were ≤30 years, 31–44 years, and ≥45 years old accounted for 36.5% (*n* = 68), 36.0% (*n* = 62), and 24.4% (*n* = 42) of the samples, respectively (Table 3). The most prevalent HPV genotype is HPV 16 which makes up 38.3% (105/274) of the total HPV‐detected genotypes in 42.7% (105/246) of HP‐ DNA‐positive samples. HPV 31 takes second place with 17.2% (47/274) of total HPV‐detected genotypes in 19.1% (47/246) of HPV‐DNA‐positive samples. Equally represented are HPV 33 and HPV 51, each with 8.8% (24/274) of total HPV‐detected genotypes in 9.8% (24/246) of HPV‐DNA‐positive samples (Figure 1). The results show that those HR HPVs make up 73% (200/274) of the detected genotypes, including multiple infections, which determined those cervical samples (*n* = 172) for further examination of oncogenic activity, according to the study's aim. Multiple HPV infections were found in 15.7%. The most common co‐infections were those with HPV 16 and 31, found in 7.6% (13/172) of cases with multiple infections (Table 2).

**Figure 1. Figure 1.** Genotype Genotype-specific distribution of HR HPVs. ‐specific distribution of HR HPVs.

The distribution of the most frequently detected HPVs concerning cytology is shown in Table 4. The prevalence rates of HR HPV 16 ranged from 44.8% in the group of NILM cytology to 75.5% in the HSIL group. Contrarily, the prevalence of HR HPV 31 is similar in the groups of NILM (37.9%), ASCUS (34.8%), and LSIL (31.8%), while it is lower in the group of HSIL (11.3%). HPV genotypes 33 and 51 are present in all of the cytological

groups in less than 21%. The statistically significant difference in the prevalence between the number of positive findings of HPV 16 (χ 2 test; *p* = 0.035) and HPV 31 (χ 2 test; *p* = 0.017) was determined, depending on the degree of severity of the cytological findings, which was not determined for HPV 33 and 51 (χ 2 test; *p* = 0.706, *p* = 0.790, respectively) (Table 4).

**HPV Infection HR HPV DNA** *n* **(%)** *n* **(%)** Single 16 83 (48.3) 145 (84.3) 31 28 (16.3) 33 18 (10.5) 51 16 (9.3) Multiple 16, 31 13 (7.6) 27 (15.7) 16, 51 6 (3.5) 31, 33 3 (1.7) 16, 33 2 (1.2) 31, 51 2 (1.2) 16, 31, 33 1 (0.6) Total: 172 (100) 172 (100)

**Table 2.** Distribution of the most prevalent HR HPVs (HR HPV 16, 31, 33, and 51) in single and multiple infections.

**Table 3.** Cervical cytology and age of female patients diagnosed with the most prevalent HR HPVs.


SD—standard deviation; NILM—negative for an intraepithelial lesion or malignancy; ASCUS—atypical squamous cells of unknown significance; LSIL—low-grade squamous intraepithelial lesions; HSIL—high-grade squamous intraepithelial lesions.

A statistically significant difference was found in the number of female patients concerning the cytological findings and the age of the patients (χ 2 test; χ <sup>2</sup> = 29.500; *p* = 0.000) (Table 5). The statistically significant difference was determined regarding the cytological findings and the age of the patients, where the female patients diagnosed with HSIL were significantly older compared to the other groups (ANOVA; F = 9.321; *p* < 0.001). The Bonferroni post-hoc test determined that the female patients with HSIL are statistically significantly older than those with ASCUS (*p* < 0.001), NILM (*p* < 0.001), and LSIL (*p* = 0.012) (Figure 2, Table 5). Female patients with confirmed HPV 31 are statistically significantly younger (33 years) than the other HR-HPV-DNA-positive patients (t = 2.317; *p* = 0.022). The average age of the female patients with confirmed HPV DNA 16 was 36.9 years; with HPV DNA 33, it was 34.1 years, while the average age of patients with HPV DNA 51 was 40.5 years. The statistical analyses show that the proportion of HPV 16 positivity maintained at the same level as age. Conversely, the proportion of HPV 31

positivity decreases with age. The detection of HR HPV 33 genotypes decreases with age, while HR HPV 51 increases (Table 5).


**Table 4.** Distribution of the most frequently detected HR HPVs according to cytology.

\* *p* < 0.05. NILM—negative for an intraepithelial lesion or malignancy; ASCUS—atypical squamous cells of unknown significance; LSIL—low-grade squamous intraepithelial lesions; HSIL—high-grade squamous intraepithelial lesions.

**Table 5.** Age-specific distribution of female patients with different cytological groups and genotypes.


SD—Standard Deviation; #—ANOVA; §—t test; \* *p* < 0.05; \*\*\* *p* < 0.001.
