Evaluating different samples & techniques for hr-HPV DNA genotyping to improve the efficiency of risk profiling for oral & cervical cancers in Sikkim, India

Study design and sampling technique

This is a cross-sectional study conducted from November 2018 to February 2022 (39 months) at the Central Referral Hospital, Gangtok, Sikkim. A convenient (non-probability) approach was used for sampling. Sample size (n=400) was calculated using Epi info, version 7.2.1.0, assuming 95 per cent confidence interval (CI) and five per cent acceptable margin of error (α=0.05).

Study population

A total of 1200 samples of three different types; (cervical, oral gargle and urine from 200 women and oral gargle and urine from 200 men) were collected from consenting participants between 18-65 yr of age following clearance from the Institutional Ethics Committee. The enrolled participants were volunteers and patients from outpatient departments (OPD) of Otorhinolaryngology (ENT) and Obstetrics and Gynaecology (OBG) during routine cytological screening.

Inclusion and exclusion criteria

Men and women of 18 yr of age and older were included for urine and oral samples. Women of 18 yr of age or above, who were married/sexually active and those clinically indicated as standard of care determined by the treating physician, were included for cervical sample collection. Pregnant women, non-consenting cases, immunocompromised individuals, women with history of total hysterectomy, radical trachelectomy and patient on radiotherapy were excluded.

Sample collection and processing

Paired urine and oral gargle samples from 200 men and women, and physician-collected cervical smear for HPV-DNA assay and cytology were collected from women participants (n=200). The cervical cells used for cytology and digene specimen transport medium (STM) vials were stored at -20°C, till hybridization with HC2. First voided morning urine, and gargle samples were collected before brushing, in 50 ml sterile containers and stored at 4-6°C for 4-6 hrs before processing. The urine and gargle samples were stored in 1X Phosphate Buffer Saline (PBS) at -20°C for batch-wise gDNA extraction.

DNA extraction

gDNA was extracted from all the 800 urine and oral gargle samples by standard Proteinase K digestion followed by phenol-chloroform method with modifications as reported by earlier, which the investigators replicated^9,10. The extracted gDNA was quantified, and amplified using standard protocols and the study algorithm is elaborated in Figure 1.

Close

Fig. 1.

Study algorithm. dept, department; M, male; F, female; Pap, papanicolaou test; qPCR, quantitative real time PCR.

Export to PPT

High-risk HPV detection for cervical samples with HC2

The cervical samples were tested for 13 hr-HPV genotypes (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68) by qualitative HC2 (digene, Qiagen)¹¹. A relative light unit per Cut-off (RLU/Co) ≥2 was taken as threshold for HC2 positivity.

High-risk HPV detection from urine and oral gargle samples of men and women using qPCR

hr-HPV DNA from both urine and oral gargle samples were amplified by multiplex real time qPCR for14 hr-HPV genotypes (16, 18, 31, 39, 45, 59, 33, 35, 56, 68, 51, 52, 58, 66). A Cycle threshold (Ct) value of 30 was taken as a positive cut-off value for qPCR and housekeeping gene (β-globin) was used as internal amplification control (Amplisen, Biorad CFX 96 thermocycler¹². Two sets of positive calibrators, one set of negative calibrators, and two positive control samples from cancer cell lines (SiHa and HeLa procured from the Amity University, Noida) were used as controls.

Data analysis

HPV in different specimen(s) were reported in terms of frequency and percentages. The performance characteristics were compared and analyzed. Sensitivity, specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV) and accuracy were calculated for different samples screened in qPCR to compare and validate one test vs. another, and one sample vs. another. The degree of agreement was evaluated with Cohen’s kappa constant (κ) for paired samples as follows: very good (0.81-1); good (0.61-0.81); moderate (0.41-0.6); fair (0.21-0.4); and poor (<0.2). A multi-layered comparison model was used to compare hr-HPV testing by HC2 with Paps, the only available gold standard or cervical cancer (CC) risk screening (Fig. 2). For technique comparison the qPCR of urine sample was then compared with HC2 of cervical samples, and qPCR for oral sample was then compared to qPCR for urine for the same individual. The area under the curve (AUC) for Receiver Operating Characteristic (ROC) with 95 per cent confidence interval (CI) was used for measuring the performance accuracy of the above algorithm. AUC with a value of one or near one was considered as an ideal and good measure of the screening model, whereas AUC<0.5 reflected poor measure of the study model. Data was recorded in excel and the statistical analysis was performed in SPSS, version 27.

Close

Fig. 2.

Algorithm for performance characteristics of different techniques for HPV DNA Assay and its matrix for understanding the comparison between samples and between techniques. ASCUS, atypical squamous cells of undetermined significance; LSIL, low grade squamous intraepithelial lesion; HSIL, high grade squamous intraepithelial lesion; U, urine; O, oral, Ct, cycle threshold, RLU/Co, relative light unit per cut off. ^*Pap positive was any sample with cytology reports including ASCUS, LSIL and HSIL, ¹Pap used as reference gold standard for comparing the performance of HC2. ²HC2 taken as reference gold standard for comparing the performance of qPCR(U). ³qPCR(U) taken as reference point for comparing the performance of qPCR(O). Note: ^*Pap is the only gold standard technique used as single reference point in ROC to assess the performance of HC2, qPCR(U) and qPCR(O).

Export to PPT

High-risk HPV detection for physician-collected cervical samples using Pap and HC2 among women

Abnormal cervical cytology was reported in six per cent (12/200):33.3 per cent (4/12) were atypical squamous cells of undetermined significance (ASC-US), 8.3 per cent (1/12) was low grade squamous intraepithelial lesion (LSIL), and 58.3 per cent (7/12) were high grade squamous intraepithelial lesion (HSIL). hr-HPVs were detected in 17.5 per cent (35/200) by HC2. Eleven of the twelve women with abnormal cervical cytology tested positive for hr-HPV by HC2 (Table I).

High-risk HPV detection for urine and oral gargle sample using qPCR:

1. hr-HPV infection status in women: 25.5 per cent (51/200) urine samples were positive for hr-HPV and seven per cent (14/200) oral gargle samples by qPCR. Of the 51 positive samples, 62.7 per cent (32/51) were infected with only a single hr-HPV genotype, and the rest were co-infections (more than two genotypes). Among the positives in oral gargle, 85.7 per cent (12/14) were positive for single hr-HPV type and the remaining exhibited infection with multiple hr-HPV genotypes (Supplementary Fig.1). Distribution of multiple hr-HPV infection were as follows: for urine 13 (68.4%) samples had two genotypes and six samples (31.6%) had three genotypes, and for oral samples two and three genotypes were equally distributed (50% each) (Supplementary Table).

2. hr-HPV infection status in men: hr-HPV were detected in eight per cent (16/200) urine and five per cent (10/200) oral gargle by qPCR (Table I). Among those with HPV-positive urine samples, 56 per cent (9/16) were infected with a single hr-HPV genotype and the rest had multiple genotypes. In oral gargle samples, 20 per cent (2/10) were positive for single hr-HPV type, and the rest had multiple genotypes. Among those with multiple genotypes in urine, 28.6 per cent (2/7) had two genotypes, 57.1 per cent (4/7) had three genotypes, and 14.3 per cent (1/7) were with four genotypes. Similarly, the two and three genotypes were equally distributed: 37.5 per cent (3/8) for each oral sample(1/8) followed by 12.5 per cent each with four and five strains (Supplementary Table & Supplementary Fig.1).

High-risk HPV genotype profile in urine and oral gargle sample using qPCR:

1. hr-HPV detection in women's urine and oral gargle samples: HPV 16 was the most frequent genotype detected (33.3%) in urine and in 42.9 per cent of oral gargle samples (Fig. 2). Variations of genotypes were observed for urine and oral. In urine, HPV 39 and 45 were 19.6 per cent each followed by HPV 68 at 17.6 per cent, HPV 35 and 58 at 11.8 per cent each, the other HPVs 18, 59, 33, 52, 56, 31, 51, 66 were distributed in descending order (Supplementary Fig. 2). Similarly for oral gargle samples, the other strains detected in descending order were HPV 68 (28.6%), HPV 39 & 45 (14.3% each) and HPV 59, 35 and 58 (7.1% each).

2. hr-HPV detection in men's urine and oral gargle samples: The most frequent genotypes detected were HPV 35 and 33 (37.5% each) in urine, and HPV 33 (80%) in oral gargle samples. In urine, the frequency of occurrence in descending order were HPV 68 (31.25%), HPV 16 (25%), HPV 39 (12.5%) and HPV 31, 45, 56, 59 (6.25% each). For oral gargle samples, the highest infection rate with HPV 33 was followed by HPV 35 and 68 (40% each), HPV 45 (30%) and HPV 56 (20%) and HPV 31 and 59 (10 per cent each). hr-HPV 18, 51, 52, 58 and 66 strains were not detected in any sample.

Comparison of hr-HPV detection in different samples by different techniques using Cohen’s kappa:

1. hr-HPV detection in urine and oral gargle samples of women by qPCR: The overall concordance of hr-HPV detection between urine and oral gargle was moderate (κ=0.50). Furthermore, for comparison between HPV genotypes, between samples, the quality of agreement was moderate for HPV 16 (κ=0.45). However, no agreement could be established for the other type-specific hr-HPV infection status between the paired urine and gargle sample (κ value <0.20; Table I).

2. hr-HPV detection in urine and oral gargle samples of men by qPCR: Overall, the concordance of hr-HPV detection between urine and oral gargle samples was good (κ=0.754). While comparing the specific genotypes between samples, quality of agreement between the two sample types were moderate for HPV 33 (κ=0.556) and fair for two genotypes HPV 16 (κ=0.324) and HPV 35 (k=0.385). No agreement could be established for the remaining hr-HPV genotypes between the paired samples (κ value<0.20; Table I).

3. The Cohen kappa scores between the two different techniques (Pap and HC2 for HPV detection) in paired cervical samples: Moderate agreement was observed for hr-HPV detection in cervical samples by HC2 as compared to Pap (κ=0.42; Table II).

4. The Cohen kappa scores between the different techniques (qPCR in reference to HC2 and Pap for HPV detection) with two different biological samples of Women: No agreement could be established for hr-HPV detection in urine by qPCR as compared to HC2 used for HPV detection in cervical samples (κ=0.06). However, fair agreement could be observed when HPV detection in urine by qPCR was compared to that of Pap (κ=0.28; Table II).

5. The Cohen kappa scores between urine and oral gargle samples of men and women by qPCR: The quality of agreement for hr-HPV detection in urine and oral gargle samples by qPCR was moderate (κ=0.50) in women and good (κ=0.75) in men (Table II).

Performance characteristics of different techniques

Table II and Figure 2 summarize the performance characteristics of Pap, HC2 and qPCR for hr-HPV detection in paired cervical, urine and oral gargle samples of women (n=200), and urine, and oral gargle samples of men (n=200). The comparison model within biological samples (oral, with urine and urine with cervical), and within techniques (Pap, HC2, and qPCR) were assessed for their performance. Pap was considered as the ‘gold standard’ in this study.

1. Performance of HC2 with reference to Pap (gold standard): The sensitivity, specificity, PPV, NPV and accuracy of HC2 as a screening tool were 91.7, 87.2, 31.4, 99.4 and 87.5 per cent respectively.

2. Performance of qPCR with reference to HC2, and Pap (gold standard): The sensitivity, specificity, PPV, NPV and accuracy of qPCR for HPV detection in urine were 85.7, 87.3, 96.6 and 79.5 per cent, respectively, in comparison to HC2. In reference to Pap-Cytology, the sensitivity, specificity, PPV, NPV and accuracy of qPCR for urine was 91.7, 78.7, 21.6, 99.3and 81.5 per cent, respectively.

3. Performance of qPCR for different samples (urine and oral gargle) of men and women as prospective candidates for viable biological samples for screening of hr-HPV infection: The sensitivity, specificity, PPV, NPV and accuracy of qPCR used for hr-HPV detection in oral gargle samples when compared with urine were 27.5, 100, 100, and 81.5 per cent respectively in females and 62.5, 100, 100 and 97 per cent in men. When the performance of qPCR was compared for HPV detection in oral gargle and urine samples in both men and women, the sensitivity, specificity, PPV, NPV and accuracy were 35.8, 100, 100, 88.6 and 89.3 per cent, respectively (Table II).

4. Performance accuracy of the multi-layered comparison algorithm: The ROC and AUC was evaluated for each assay (HC2 and qPCR) compared to Pap (Fig. 3). The AU-ROC was 0.895 (95 per cent CI 0.8, 0.989) for HC2 for hr-HPV detection in cervical, 0.852 (95 per cent CI 0.753, 0.951) for qPCR for hr-HPV detection in urine, and 0.596 (95 per cent CI 0.411, 0.78) for qPCR for hr-HPV detection in oral gargle samples. The AUC for both HC2 and qPCR urine was close to 1, establishing a good measure for hr-HPV detection. The AUC was lower for qPCR for oral samples (0.59).

Close

Fig. 3.

Comparison of receiver operating characteristic (ROC) curves for Hybrid Capture 2 (HC2) assays for cervical (C) and quantitative real time PCR (qPCR) for urine (U) and oral gargle sample (O) with reference to Pap as gold standard technique. AUC, area under curve.

Export to PPT