Human Papillomavirus genotype concordance between Anyplex II HPV28 and Linear array HPV genotyping test in anogenital samples

Abstract

Anyplex II HPV-28 (HPV-28) can detect individually 28 HPV genotypes. We assessed the agreement between Linear array HPV genotyping (LA-HPV) and HPV-28 for detection of 27 HPV genotypes in 410 stored anogenital samples (75 anal samples, 335 physician-collected cervical samples) collected over 5 years from 410 individuals (13 men, 397 women), including 202 HIV-seropositive individuals. HPV DNA was detected in 393 (95.9%, 95% confidence interval (CI) 93.4–97.4) and 382 (93.2%, 95% CI 90.3–95.3) samples with HPV-28 and LA-HPV (p=0.13), respectively, for a good agreement of 96.3% (kappa=0.65). Of the 10,503 HPV typing results, 10,195 (780 positive, 9,577 negative) were concordant (agreement of 97.1%, 95% CI 96.7–97.4 (kappa=0.82, 95% CI 0.80–0.84). The mean type-specific concordance for 27 genotypes was 97.0%, 95% CI 95.8–98.5 (kappa = 0.86±0.07, 95% CI 0.83–0.88). Excellent agreement was obtained individually for all high-risk genotypes (kappa=0.81–0.97) and for most other genotypes except for types 42, 44, 54, 68 and 69. The mean number of types per sample in discordant samples detected with LA-HPV (3.0, 95% CI 2.7–3.4) was greater than in concordant samples (1.4, 95% CI 1.3–1.5; p<0.001). In conclusion, HPV-28 compared favorably with LA-HPV but was more frequently positive for HPV42 and 68.

Introduction

Infection by human papillomavirus (HPV) causes squamous intraepithelial lesions and invasive cancer of the anogenital tract. Accurate and sensitive genotyping of HPV isolates is essential for epidemiological studies, clinical trials and vaccine surveillance studies [1]. The Linear array HPV genotyping test (LA-HPV; Roche Diagnostic, Laval, Canada), a widely used HPV genotyping method, combines consensus PCR with reverse hybridization of amplicons [2]. LA-HPV is often considered as the reference test for genital HPV genotyping since it has been well characterized, validated for research and clinical use and detects 36 genital types [3–7]. Year to year LA-HPV reagents were standardized and produced under quality-controlled conditions, enabling comparison between studies. Unfortunately, the production of reagents for LA-HPV has ceased from Roche since 2019 due to unavailability of some reagents.

The Anyplex II HPV-28 (HPV-28; Seegene, Seoul, South Korea) test is a fully automated commercial multiplex real-time PCR assay for the individual identification of 28 genotypes. The assay combines dual priming oligonucleotide with tagging oligonucleotide cleavage and extension, identified by melting curve analysis in multiple fluorescent channels [8]. Because of automation, ease of use and objective readout, the Anyplex assay format simplifies HPV genotyping [8] and is now a widely used genotyping assay [9]. HPV-28 is thus a suitable alternative to LA-HPV for HPV genotyping. HPV-28 has been widely used for HPV genotyping in clinical and epidemiological studies [10–15] and has been evaluated in comparative studies against various genotyping assays [12, 16–21]. Anyplex HPV II HR detection assay has been validated using the international Valgent framework [8] and was shown to meet the international consensus guideline for HPV test requirements [22].

HPV-28 has been previously compared to LA-HPV using cervical samples [18–20], but to our knowledge these assays have not been compared using other specimens from the anogenital tract nor among people living with HIV. Given that LA-HPV is now unavailable, we conducted the present study to compare genotyping results of individual genotypes obtained with LA-HPV and HPV-28 using cervicovaginal samples collected in women living with and without HIV, and anal samples from men and women living with HIV. Samples from individuals living with HIV are prone to multiple type infections [23]. This information will be useful in view of the wider use of HPV genotyping by numerous research groups who will need to replace LA-HPV by another genotyping assay.

Material and methods

Clinical specimens.

Processed cervicovaginal (n=335) and anal (n=75) specimens that had been collected between 2007 and 2013 from 410 participants (397 women, 13 men) in four cohort studies and one clinical trial were retrieved from processed samples stored at −80°C [24–28]. In each of these five studies, all participants provided written informed consent for HPV testing. Studies were approved by the local research ethics committees of participating institutions.

No more than one sample per participant was selected. Extracted DNA from 62 anal and 101 cervical samples were selected randomly from banked processed HPV-positive specimens collected in women living with HIV recruited in the EVVA study (Evaluation of Human Immunodeficiency Virus, Human PapillomaVirus, and Anal Intraepithelial Neoplasia in women) conducted to evaluate relationships between HPV and HIV infection, and cervical disease [24]. Cervical samples were obtained with a cytobrush in 1.5 mL of Preservcyt and anal samples were collected with Dacron swabs in 1.5 mL of Preservcyt. Extracted DNA from 136 cervico-vaginal samples were selected randomly from banked processed HPV-positive samples collected in women participating in HITCH (HPV Infection and Transmission among Couples through Heterosexual activity) cohort study [25, 26], a longitudinal study on HPV transmission among heterosexual couples. Cervico-vaginal samples were collected with Dacron swabs in 1.5 mL of Preservcyt.

The other samples were selected from our bank of extracted DNA with the aim of obtaining 1) a minimum of 10 positive samples for each of the genotypes detected by LA-HPV and 2) at least half of samples tested having multiple HPV genotypes. Thus, 72 processed cervical specimens were from women participating in the Canadian Cervical Cancer Screening Trial (CCCaST), a randomized controlled trial designed to compare HPV DNA testing and Pap testing for cervical cancer primary screening [27]. The later samples were collected with the cytobrush provided in the Digene cervical sampler kit in Specimen transport Medium (STM) and neutralized during sample processing before extraction with Master pure. Twenty-six processed cervical samples were from women living with HIV participating in an open labeled, multi-centered cohort study, to evaluate the safety, immunogenicity and predictors of seroresponse to the HPV vaccine [28]. These samples had been obtained with a Dacron Swab in 20 mM Tris buffer, pH 8.3. Thirteen processed anal samples were from men living with HIV participating in a cohort study on the natural history of anal intraepithelial neoplasia, the HIPVIRG study [23]. Anal samples were collected with a Dacron swab in 1.5 mL of Preservcyt.

In the parent studies, sample DNA was extracted with the Master Pure extraction kit (Epicentre, Madison, WI) [29] from an identical volume of cell suspension obtained with anal swabs, endocervical cytobrushes, or cervicovaginal swabs. The same DNA extract was tested in both assays.

LA-HPV.

LA-HPV had been done in the parent studies by one laboratory 0.5 to 10 years before the performance of HPV-28. Five μl of extracted DNA was tested in a final reaction volume of 100 μl with 50 μl of the kit working master mix containing MgCl₂, KCl, AmpliTaq Gold DNA polymerase, Uracil-N-Glycosylase, dATP, dCTP, dGTP, dUTP, dTTP, and biotinylated PGMY primers and ß-globin primers GH20 and PC04, as per the manufacturer recommendations [3]. Samples positive with the HPV52 cross-reactive probe were tested with a real-time PCR assay specific for type 52, as described previously [30]. Only samples reactive in the HPV52 real-time PCR assay were considered as HPV52-positive. LA-HPV detected 36 mucosal HPV genotypes (6, 11, 16, 18, 26, 31, 33, 34, 35, 39, 40, 42, 44, 45, 51, 52, 53, 54, 56, 58, 59, 61, 62, 66, 67, 68, 69, 70, 71, 72, 73, 81, 82 (two variants), 83, 84, and 89).

HPV-28 test.

Two aliquots of 5 μl of extracted DNA were tested in each of the two 20 μl reaction mixtures with primer set A and B, according to the manufacturer’s instructions on the CFX96 real-time PCR system (Bio-Rad Laboratories). Data analysis of the test results was generated automatically using the Anyplex software. Results were obtained without knowledge of previous LA-HPV results. The 28 HPV genotypes detected include types 6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 43, 44, 45, 51, 52, 53, 54, 56, 58, 59, 61, 66 68, 69, 70, 73 and 82).

Quality controls in PCR assays.

In both assays, ß-globin gene coamplification screened for the presence of inhibitors, degraded or inadequate quantities of human DNA. Samples negative for ß-globin were considered inadequate for PCR analysis. Weak (10 HPV18 DNA copies), and strong positive HPV controls, were included in each amplification run as controls in LA-HPV. Negative and positive controls provided in the HPV-28 kit were included in each HPV-28 run, as suggested by the manufacturer. Precautions to avoid contamination were effective at all times. Our laboratory participates successfully to the WHO Global HPV LabNet DNA Genotyping Proficiency for HPV detection and typing [9].

Statistical methods.

The crude percent agreement between both detection methods was the percentage of samples with identical results using both methods. Agreement for overall positivity (HPV DNA positive), for positivity for high-risk HPV types (HR) as a group and for positivity for each type was calculated. High-risk types included types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 66 and were detected with both assays [31]. Nine genotypes (HPV34, 62, 67, 71, 72, 81, 83, 84 and 89) could be detected only with LA-HPV while HPV43 could be detected solely by HPV-28. Thus, 27 genotypes were detected with both assays (6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 44, 45, 51, 52, 53, 54, 56, 58, 59, 61, 66, 68, 69, 70, 73, 82), and were considered in the comparison between HPV-28 and LA-HPV. Samples with discordant results were not retested. The unweighted kappa statistic was calculated to adjust for chance agreement between HPV detection methods [32]. In general, a kappa value above 0.75 indicates excellent agreement, between 0.40 and 0.75 indicates fair to good agreement, and below 0.40 represents poor agreement beyond chance. 95% confidence intervals (95% CI) were calculated with the modified Wald interval. The number of types detected per sample by each genotyping assay was compared using the Sign test, since the frequency of types per sample was not normally distributed. Proportions were compared with the z statistic test. Odds Ratios were calculated with logistic regression with 95% confidence intervals (95% CI). A p value <0.05 was considered statistically significant. Statistical analyses were performed with Statistica 6.0 (StatSoft).

Results

A total of 410 processed anogenital samples previously tested with LA-HPV were analyzed with HPV-28. All samples yielded a positive ß-globin test in both assays. As assessed by LA-HPV for 36 genotypes, 17 were negative, 169 samples contained only one type, and 224 contained multiple HPV types (2 types=79, 3=46, 4=29, 5=25, 6=8, 7=10, 8=9, 9=7, 10=5 and >10=6; median=2.0; range=0–16). The distribution of types shown in Supplementary Table 1 indicates that despite our wide collection of processed samples, less than 10 samples could be retrieved for HPV26 (n=8), HPV34 (n=3), HPV69 (n=6) and HPV71 (n=6).

Results from both assays were first compared for the detection of HPV DNA considering all types detected by each assay. As shown in Table 1, HPV DNA was detected in 393 (95.9%, 95% CI 93.4–97.4) and 382 (93.2%, 95% CI 90.3–95.3) of 410 samples with LA-HPV and HPV-28, respectively (p=0.13). A good agreement for overall HPV DNA positivity was obtained considering all genotypes detected by each assay or only those detected with both assays (Table 1). Considering high-risk genotypes, 279 samples were positive and 107 were negative with both assays, 10 were positive only with LA-HPV and 14 were positive only with HPV-28, for an agreement of 94.2% (95% CI 91.4 – 96.1) and an excellent kappa of 0.86 (95% CI 0.79 – 0.91).

Table 1.

Comparison of HPV-28 and LA-HPV assays for detection of HPV DNA in 410 anogenital samples.

Test results	No. of samples positive by each assaya		Total
	LA-HPV Positive	LA-HPV Negative
HPV-28 Positive	380 (378)	2 (4) 382	(93.2)
HPV-28 Negative	13 (7)	15 (21)	28 (6.8)
Total	393 (95.9)	17 (4.1)	410 (100)

^aStraight numbers are considering all genotypes detected by each assay (LA-HPV could detect 36 and HPV-28 could detect 28 genotypes) for an agreement of 96.3% (95% CI 94.0–97.8) and a good kappa value of 0.65 (95% CI 0.46 – 0.73). Numbers in parenthesis were obtained by considering only the 27 genotypes detected in both assays for an agreement of 97.3% (95% CI 95.2–98.6) and an excellent kappa value of 0.78 (95% CI 0.61 – 0.88). LA-HPV: Linear array HPV genotyping assay; HPV-28: Anyplex II HPV-28.

We then compared individual typing results for the 27 genotypes detected with both assays on the 389 positive specimens that tested positive for HPV with at least one of the PCR tests, for a total of 10,503 type-specific results. Overall, 780 positive results for the same type and 9,577 negative results were obtained in both PCR tests, while 92 positive typing results were obtained only with LA-HPV and 216 only with HPV-28. Overall agreement was obtained for 10,195 of 10,503 typing results (97.1%, 95% CI 96-7-97.4) for an excellent kappa value of 0.82 (95% CI 0.80–0.84).

The genotype-specific agreement between HPV-28 and LA-HPV assays for the 27 genotypes detected by both assays is presented in Table 2. The mean agreement between assays reached 97%±2.5% (95% CI 95.8% – 98.5%) and ranged from 93.2% to 100% for a mean kappa value of 0.86±0.07 (95% CI 0.83 – 0.88). The level of agreement between tests scored above 97% for the majority of genotypes except for HPV51 and HPV61, which were detected more frequently by LA-HPV only than by HPV-28 only, and for HPV42, 44, 53, 54, and 68 which were detected more often by HPV-28 only than by LA-HPV only (Table 2). Agreements less than excellent with kappa values below 0.75 were seen for genotypes 42, 44, 54, 68 and 69. Results obtained for HPV69 should be interpreted with caution since only seven samples were positive for HPV69.

Table 2.

Genotype specific comparison of HPV-28 and LA-HPV for 27 genotypes on 410 anogenital samples

	No. type-sample combinationsb
Typea	+/+	LA+/HPV28−	LA-/HPV28+	−/−	% agreement (95%CI)	Kappa (95%CI)
6	35	0	7	368	98.3 (96.5–99.2)	0.90 (0.80–0.90)
11	10	0	4	396	99.0 (97.4–99.7)	0.83 (0.57–0.83)
16	72	3	1	334	99.0 (97.4–99.7)	0.97 (0.91–0.98)
18	34	2	0	374	99.5 (98.1–100)	0.97 (0.87–0.98)
26	6	2	0	402	99.5 (98.1–100)	0.86 (0.48–0.87)
31	38	1	3	368	99.0 (97.4–99.7)	0.95 (0.85–0.97)
33	20	1	1	388	99.5 (98.1–100)	0.95 (0.79–1.00)
35	21	1	2	386	99.3 (97.8–99.9)	0.93 (0.77–0.97)
39	29	3	1	377	99.0 (97.4–99.7)	0.93 (0.81–0.96)
40	15	0	5	390	98.8 (97.1–99.6)	0.85 (0.66–0.85)
42	25	0	29	356	92.9 (90.0–95.1)	0.60 (0.49–0.60)
44	8	23	35	344	85.9 (82.1–88.9)	0.14 (0.02–0.30)
45	32	6	1	371	98.3 (96.5–99.2)	0.89 (0.77–0.92)
51	35	10	4	361	96.6 (94.3–98.0)	0.81 (0.69–0.88)
52	38	3	9	360	97.1 (94.9–98.4)	0.85 (0.73–0.90)
53	49	4	10	347	96.6 (94.3–98.0)	0.86 (0.76–0.91)
54	34	6	18	352	94.2 (91.4–96.1)	0.71 (0.58–0.79)
56	31	1	9	369	97.6 (95.5–98.7)	0.85 (0.73–0.88)
58	41	2	1	366	99.3 (97.8–99.9)	0.96 (0.87–0.99)
59	28	4	3	375	98.3 (96.5–99.2)	0.88 (0.74–0.95)
61	26	12	3	369	96.3 (94.0–97.8)	0.76 (0.61–0.83)
66	35	1	8	366	97.8 (95.8–98.9)	0.87 (0.76–0.90)
68	30	0	37	343	91.0 (87.8–93.4)	0.58 (0.48–0.58)
69	4	2	1	403	99.3 (97.8–99.9)	0.72 (0.29–0.90)
70	31	4	6	369	97.6 (95.5–98.7)	0.85 (0.72–0.92)
73	39	0	11	360	97.3 (95.2–98.6)	0.86 (0.77–0.87)
82	14	1	7	388	98.1 (96.1–99.1)	0.77 (0.56–0.82)

^aOnly the 27 genotypes shared by both assays were considered for analysis.

^b−/−, negative with both assays; −/+ HPV-28 positive and LA-HPV negative, +/−, HPV-28 negative and LA-HPV positive, +/+, positive with both assays.

LA-HPV or LA: Linear array HPV genotyping assay; HPV-28 or HPV28: Anyplex II HPV-28.

We then investigated if the level of agreement between assays depended on the HIV status of participants or on site of sampling (anus versus cervix). Levels of agreement between the assays below 97% were obtained for individuals living and not living with HIV (see supplement table 2) for HPV types 42, 44, 54 and 68, as was reported for all participants in Table 2. The difference of agreement reached statistical significance for HPV44 (supplement table 2). However, for HPV types 51, 53 and 61, an agreement below 97% was encountered only in specimens from individuals living with HIV. When the site of sampling was considered (see supplement table 3) a level of agreement below 97% was obtained with anal specimens and not cervical specimens for HPV types 51, 53, 61, but for both sites for HPV types 42, 44, 54 and 68. The difference between anal and cervical samples in percent agreement between assays was statistically significant for HPV types 44, 51, 61 and 68. Results were similar when only specimens from individuals living with HIV were considered (data not shown).

Overall, 226 (55.1%, 95% CI 50.3–60.0) of 410 samples yielded fully concordant genotype-specific results between assays and 144 (35.1%, 95% CI 30.7–40.0) were either partially concordant (n=127) or fully discordant (n=17). Of 226 concordant samples, 21 were negative for HPV DNA, 140 were positive for a single genotype and 65 contained multiple genotypes (2 genotypes, n=40; 3 to 4 types, n=20; 5 to 7 types, n=5). Overall, 140 of 146 (95.9%, 95% CI 91.1 – 98.3) single type infections yielded concordant results as opposed to 65 of 243 (26.8%, 95% CI 21.6 – 32.7) multiple type infections (p=0.001). A greater concordance between assays was obtained on cervical samples (206 of 335, 61.5%, 95% CI 56.2 – 66.6) than anal samples (20 of 75, 26.7%, 95% CI 17.9 – 37.7) irrespective of HIV status, on cervical samples (52 of 101, 51.5%, 95% CI 41.9 – 61.0) than anal samples (16 of 62, 25.8%, 95% CI 16.5 – 38.0) in women living with HIV, and in samples from women not living with HIV (136 of 208, 65.4%, 95% CI 58.7 – 71.5) compared to HIV-seropositive participants (90 of 202, 44.6%, 95% CI 37.9 – 51.5). Discordant samples had at least one to 5 (median=1) different genotypes between assays. One or more additional genotypes were detected by HPV-28 and LA-HPV in 143 and 63 samples, respectively. Discordant samples had a significantly higher number of genotypes per sample detected with the LA-HPV (mean of 3.02±2.39, 95% CI 2.67 – 3.37, range 0 – 12) compared to fully concordant samples (mean of 1.4±1.1 types, 95% CI 1.3–1.5, range=0–7 types; p<0.001). Moreover, the proportion of samples with concordant genotyping results containing one type per sample (140/160 samples, 87.5%, 95% CI 81.4–91.8) was significantly greater than for concordant samples with ≥2 types (81/222, 36.5% (95% CI 30.4–43.0). The number of genotypes per sample was significantly associated with complete concordance between assays (OR 2.41, 95% CI 1.53–3.29).

We then investigated if the use of HPV-28 and LA-HPV would estimate similar genotype-specific rates of detection. As shown in Figures 1a and 1b, for all but HPV42 and 68, the rate of genotype-specific detection was not significantly different between the two assays. HPV42 and HPV68 were detected significantly more frequently with HPV-28 than with LA-HPV. Considering only the 27 HPV genotypes shared by both genotyping systems, a greater number of types per sample was detected with HPV-28 (mean of 2.43 types per sample (95% CI 2.22–2.64); median of 2 types per sample) compared to LA-HPV (mean of 2.14 types per sample (95% CI 1.95–2.34); median of 1 type per sample; p<0.01). Nevertheless, there was a good correlation between assays for the number of types detected per sample, as shown in Figure 2.

Figure 1. Rate of detection of individual HPV genotypes in 410 anogenital samples, by assay.

LAHPV : Linear array HPV genotyping assay; HPV-28: Anyplex II HPV-28. The 27 HPV genotypes detected with both assays were included. The difference between detection rates were statistically significant (p=0.001) for HPV42 and HPV68 (z-test of significance between proportions). The classification of HPV genotypes into high, low and unknown risk is defined in the methods section.

Figure 2.

Correlation between the number of genotypes detected per sample on 410 anogenital samples with LA-HPV and HPV-28 considering only the 27 genotypes detected with both assays.

Figure 1. The dashed lines represent the 95% CI of the regression line. R²=0.82, p<0.01.

Discussion

The current comparative study was undertaken on 410 anogenital samples to assess the agreement for identification of individual genotypes between LA-HPV and HPV-28 on previously-characterised anal and cervical specimens. All samples were adequate for PCR analysis. Samples were selected to cover the diversity of genital HPV types aiming to test at least 10 samples per type, which proved impossible for types 26 and 69. To our knowledge, this is the first study to compare HPV-28 with LA-HPV in anal samples and in samples from people living with HIV. Inclusion The inclusion of samples collected from the anal canal or from individuals infected with HIV also increased the likelihood of testing samples with multiple type infections. Multiple type infections represent a greater challenge for genotyping systems based on PCR and increase the likelihood of obtaining discordant results possibly due to false-negative results due to competition during amplification for some types [3, 17, 19].

Agreement between HPV-28 and LA-HPV for HPV DNA detection was excellent using a variety of anogenital specimens, as reported by others using only cervical specimens [18, 19]. We also demonstrated a very good to excellent agreement between assays of 93.2% to 100% considering individually each of the 27 genotypes detected with both assays. As reported by others, agreement was lower for types 42, 54 and 68 [17–19]. HPV-28 proved to be more often positive than LA-HPV for these types as well as for HPV44. A lower agreement between LA-HPV and HPV-28 for genotypes 11, 40, 53 and 61, was reported in one study but was not confirmed by another one [18–20]. We also found a lower agreement for types 53 and 61. The percentage of fully concordant samples at the genotypic level of 55% between HPV-28 and the LA-HPV was lower than reported by other teams, but our study included a greater proportion of multiple type infections [17, 18]. A lower agreement was demonstrated in specimens collected from HIV-seropositive participants and from the anus. However, agreement was excellent for high-risk genotypes

Concordance was higher for specimens containing less genotypes as reported by others [19]. This suggests that LA-HPV is more susceptible to reagent competition than HPV-28. Competition due to coamplification of ß-globin and HPV DNA or due to the presence of multiple HPV genotypes in multiplex PCR was shown to hamper the sensitivity of consensus PCR tests [3, 33, 34]. The multiplexing strategy using DPO primers may generate less competition. Others have also described a higher rate of discordant results between LA-HPV and HPV-28 in multiple type infections [17, 19].

The average number of types per sample was higher with HPV-28 than LA-HPV. However, there was a strong correlation between assays for the number of types detected per specimen. In several of the WHO proficiency panel reports, PGMY-based assays were less sensitive for low copy number samples than HPV-28 [5, 6, 9, 17, 35]. Despite these results, the rate of detection of common HPV genotypes was similar between these assays, except for types 68 and 42. A higher rate of detection of HPV42 with HPV-28 compared to PGMY-based systems has been reported by others [17, 19, 20]. Samples positive for HPV42 in HPV-28 yet negative with LA-HPV did not contain other genotypes systematically, suggesting the increased sensitivity of HPV-28 was not related to cross-reactivity (data not shown). HPV68 discordant samples can be explained by mismatches in the PGMY09 primer to HPV68a variant [5, 17, 19]. Adding HPV68a-specific primers to PGMY primers improved HPV68 detection [17]. Others have reported a higher rate of detection using HPV-28 for types 39, 40, 54 and 61, although inconsistently [17, 20]. In contrast, Cornall et al. reported a higher rate of detection with LA-HPV of HPV52 and 61 [19]. Our study found a higher detection rate of HPV54 with HPV-28 and of HPV61 with LA-HPV, but without reaching statistical significance. These differences between studies could be due to differences in population tested in terms of HPV prevalence and HPV polymorphism. HPV51 polymorphism primers and probe utilized in HPV-28 explained discordant results between HPV-28 and LA-HPV in one study. Although more samples were positive for HPV51 in our study with LA-HPV, this difference was not statistically significant. The polymorphism responsible the reduced sensitivity of HPV-28 for HPV51 may be less frequent in our population. The sensitivity of the confirmatory assay for HPV52 to confirm results with the LA-HPV cross-reactive probe could also have played a role in discrepancies between studies. Discordant results between assays could also be related to genotype-specific differences in limits of detection of each assay [19].

One limitation of the current study was the use of HPV-positive specimens to increase the sample size for each type. Random selection of samples originally positive with LA-HPV could have introduced a selection bias in favor of LA-HPV and also explains the small number of HPV-negative samples. Since an important proportion of samples contained multiple HPV types and the majority of samples came from individuals living with HIV, the test panel is not representative of the prevalence of HPV in a population of asymptomatic individuals. However, the high prevalence of HPV infection and the presence of multiple type infections in our population allowed high precision to measure inter-assay agreements for several types. Unlike our study, previous comparative studies on LA-HPV and HPV-28 reported comparisons for several types detected infrequently [17, 19, 20] Our decision to use these selected samples afforded the opportunity to scrutinize assay performance under the more extreme conditions seen with specimens containing several HPV types.

Another consideration may be that specimens analyzed were stored for several years between assays and there could be some specimen degradation over time. Both tested the same preparation of extracted DNA that was stored at −80°C and all proved to be adequate for PCR analysis. We chose to compare HPV-28 and LA-HPV without adding a third technique. In recent years, next generation sequencing (NGS) has been utilized for HPV genotyping and has been compared to HPV-28 [21]. Although promising and allowing to detect many more genotypes than commercial assays, NGS failed to detect genotypes detected with HPV-28 in a significant number of samples.

In conclusion, HPV-28 performed comparably to LA-HPV for most of the 27 genotypes that could be detected with both assays. However, HPV-28 detected significantly more genotypes especially context of in multiple type infections compared to LA-HPV. HPV-28 will be of great value for epidemiological studies and clinical trials to monitor HPV infection at the level of genotypes in genital samples.

Data Availability Statement:

The study does not share the data although tables provides the raw data obtained in this comparison study.