Comparison of Real-Time Multiplex Human Papillomavirus (HPV) PCR Assays with INNO-LiPA HPV Genotyping Extra Assay

Abstract

Real-time type-specific multiplex human papillomavirus (HPV) PCR assays were developed to detect HPV DNA in samples collected for the efficacy determination of the quadrivalent HPV (type 6, 11, 16, and 18) L1 virus-like particle (VLP) vaccine (Gardasil). Additional multiplex (L1, E6, and E7 open reading frame [ORF]) or duplex (E6 and E7 ORF) HPV PCR assays were developed to detect high-risk HPV types, including HPV type 31 (HPV31), HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, and HPV59. Here, we evaluated clinical specimen concordance and compared the limits of detection (LODs) between multiplex HPV PCR assays and the INNO-LiPA HPV Genotyping Extra assay, which detects 28 types, for the 14 HPV types common to both of these methods. Overall HPV detection agreement rates were >90% for swabs and >95% for thin sections. Statistically significant differences in detection were observed for HPV6, HPV16, HPV18, HPV35, HPV39, HPV45, HPV56, HPV58, and HPV59 in swabs and for HPV45, HPV58, and HPV59 in thin sections. Where P was <0.05, discordance was due to detection of more HPV-positive samples by the multiplex HPV PCR assays. LODs were similar for eight HPV types, significantly lower in multiplex assays for five HPV types, and lower in INNO-LiPA for HPV6 only. LODs were under 50 copies for all HPV types, with the exception of HPV39, HPV58, and HPV59 in the INNO-LiPA assay. The overall percent agreement for detection of 14 HPV types between the type-specific multiplex HPV PCR and INNO-LiPA genotyping assays was good. The differences in positive sample detection favored multiplex HPV PCR, suggesting increased sensitivity of HPV DNA detection by type-specific multiplex HPV PCR assays.

INTRODUCTION

According to the World Health Organization (WHO), cervical cancer is the second most common cause of cancer death in women, with approximately 500,000 cases per year, and the presence of HPV infection has been implicated in more than 99% of cervical cancers (16). The accurate evaluation and genotyping of human papillomavirus (HPV) infections is of critical importance to establish and monitor HPV vaccine efficacy (13), to evaluate the epidemiology of HPV infections worldwide (4), and to accurately evaluate the oncogenic potential and disease association of high-risk HPV genotypes (7, 8, 15). In the last several years, advances in HPV detection methods have been achieved, and a wide variety of assays are now available. The majority of the commercially available protocols use degenerate and/or consensus primers, which allow the amplification of a large spectrum of HPV types by PCR and subsequent post-PCR detection of specific HPV groups or types. The most commonly used PCR assays amplify the L1 or the E6 and E7 region. It is important to understand the benefits and limitations of the different HPV detection methodologies in common use for appropriate interpretation of epidemiologic and clinical research on HPV and HPV-related disease (1, 2, 5, 9–11).

The Merck Research Laboratories (MRL) multiplex HPV PCR assays were developed for the low-level, type-specific detection of HPV for the efficacy determination of the quadrivalent HPV (type 6, 11, 16, and 18) L1 virus-like particle (VLP) vaccine (Gardasil) (8, 13, 14). The multiplex HPV PCR assays are based on the simultaneous detection of either three (L1, E6, and E7) or two (E6 and E7) open reading frames (ORFs). The HPV type- and ORF-specific amplicons generated in the PCR are detected simultaneously by the use of differently labeled type- and ORF-specific fluorescent probes. Fluorescence emission is then captured in real time during PCR cycling by using either the ABI 7700 sequence detection system (Foster City, CA) or the Stratagene Mx3005P real-time PCR system (La Jolla, CA).

The Innogenetics (Gent, Belgium) INNO-LiPA HPV Genotyping Extra assay is designed to identify 28 different HPV genotypes based on PCR amplification followed by reverse hybridization (5, 10, 12). Short PCR fragment (SPF10) biotinylated consensus primers are used to amplify a portion of the L1 ORF of numerous HPV types. The resulting biotinylated amplicons are hybridized to strips, onto which type-specific oligonucleotides have been immobilized, and then automated colorimetric detection is performed with the Auto-LiPA 48 instrument.

In this study, we evaluate the performance of the INNO-LiPA HPV genotyping PCR assay against that of the MRL multiplex HPV PCR assays. Only the 14 HPV genotypes common to both assays (HPV type 6 [HPV6], HPV11, HPV16, HPV18, HPV31, HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, and HPV59) were considered for comparison. The performance factors assessed were the assay concordance, relative sensitivity and specificity, and estimated limits of detection (LODs) for these 14 HPV types.

MATERIALS AND METHODS

Experimental design.

The concordance and relative sensitivity and specificity of each assay for each of the HPV types common to both were assessed through the testing of 360 human clinical swab samples and 278 selected human clinical thin-section samples by both multiplex HPV PCR and INNO-LiPA HPV genotyping assays. To compare assay sensitivities in a limit of detection analysis, a 2-fold dilution series of the HPV L1, HPV E6, and HPV E7 plasmids purified in a background of human genomic DNA (MRC5 cells) for HPV6, HPV11, HPV16, HPV18, HPV31, HPV45, HPV52, and HPV58, or that of previously quantified clinical specimens positive for HPV33, HPV35, HPV39, HPV51, HPV56, and HPV59, were tested in both assays.

Swab specimens.

Female genital swab specimens were collected as part of ongoing clinical trials to evaluate the efficacy of the quadrivalent HPV vaccine (13). As part of the clinical trials, institutional review boards at each center approved the protocols, and written informed consent was obtained from all subjects. Endo-Ecto cervical swabs and labial/vulvar/perineal/perianal swabs were collected from each patient, and each swab was placed in 1 ml of digene specimen transport medium (STM). Each specimen was divided into three pristine aliquots and frozen. Previous multiplex HPV PCR results were used as a screen to identify positive specimens from which to select retained aliquots and ensure adequate representation for each of the 14 HPV types evaluated in this study. Aliquots of 360 human clinical swab specimens which previously tested positive for one given HPV type (irrespective of positivity in other types) were selected to obtain a set of specimens that would include at least 24 positive specimens for each HPV type. Swab samples were processed through DNA isolation and were tested in both of the HPV assays.

Thin-section specimens.

Female biopsy specimens were collected as part of ongoing clinical trials to evaluate the efficacy of the quadrivalent HPV vaccine (13). As part of the clinical trials, institutional review boards at each center approved the protocols, and written informed consent was obtained from all subjects. Biopsy material was formalin fixed, paraffin embedded, and cut into 4-μm thin sections. Biopsy thin sections designated for PCR testing were placed into individual sterile tubes and shipped to the MRL laboratory for DNA extraction and multiplex HPV PCR testing. Due to the limited availability of archived thin-section specimens for this evaluation, 278 thin-section specimens with positive results for a particular HPV type, irrespective of the results for other types, were selected from the existing multiplex HPV PCR data. The previously isolated DNAs from the selected positive specimens were tested by the INNO-LiPA PCR assay.

Limit of detection specimens.

A 2-fold dilution series of HPV L1, HPV E6, and HPV E7 plasmids for HPV6, HPV11, HPV16, HPV18, HPV31, HPV45, HPV52, and HPV58, or that for previously quantified clinical specimens positive for HPV33, HPV35, HPV39, HPV51, HPV56, and HPV59, was repeated eight times on one plate. Quantification of clinical specimens was performed for duplex types through the identification and pooling of specimen aliquots with low threshold cycle values in previous multiplex HPV PCR results, suggesting the presence of high viral loads. A portion of the pooled specimens was subjected to DNA isolation and type-specific quantitative PCR using a standard curve of E6 or E7 type-specific plasmids of known copy numbers to estimate viral loads in the pool. The remaining pooled specimens were diluted appropriately based on quantitative PCR results to achieve input copy numbers across the 2-fold dilution series on each type-specific plate, ranging from 78 to 10,000 copies/purification well, representing either 1.56 to 200 copies/test for multiplex HPV PCR assays or 1.95 to 250 copies/test for the INNO-LiPA HPV assay. The data were obtained from a single assay run, with 8 replicates for each input copy number to assess the LODs.

DNA isolation.

DNA was extracted from 200 μl of each specimen using the Qiagen Spin blood kit in a 96-well format, in accordance with the manufacturer's protocol for preparations of swabs and plasmids. For thin sections, an overnight digestion of the paraffin-embedded tissues using Qiagen proteinase K in ATL buffer in a background of 0.1 μg/ml Escherichia coli DNA at 56°C was performed in lieu of the proteinase digestion step in the Spin blood kit. DNA isolation from digested thin-section specimens then proceeded according to the manufacturer's protocol. HPV-negative MRC5 cells (500,000 cells/well) and HPV16-positive SiHa cells (2,500 cells/well) were included in the processing of each plate to serve as positive controls for DNA isolation and HPV detection. Negative-control wells included in plate processing contained 200 μl of STM buffer for swab batches, E. coli DNA for plasmid batches, and overnight digest buffer for thin-section batches. All isolated DNAs were eluted into 200 μl of AE buffer.

MRL HPV type-specific multiplex PCR.

Multiplex HPV PCR assays are type-specific assays designed to simultaneously amplify and detect the L1, E6, and E7 ORFs of HPV6, HPV11, HPV16, HPV18, HPV31, HPV45, HPV52, and HPV58 or the E6 and E7 ORFs of HPV33, HPV35, HPV39, HPV51, HPV56, and HPV59. Annealing and subsequent amplification with HPV type- and gene-specific primer sets with simultaneous annealing of type- and ORF-specific fluorescently labeled oligonucleotide probes and the capture of fluorescence emissions during each amplification cycle were performed on either the ABI 7700 sequence detection system, as previously described (5), or the Stratagene Mx3005P real-time PCR instrument. Assays were reoptimized and revalidated for use on the Stratagene Mx3005P instrument, with Cal Fluor Orange and Quasar 670 as the E6 and E7 ORF reporters used, respectively, for all 14 type-specific HPV assays. The multiplex assays used on each instrument platform had similar performance sensitivities in bridging validation studies and are considered equivalent for clinical testing use. For each sample, in order to be considered positive for any of the HPV types tested, the sample must be positive for at least two of the type-specific ORFs (two of L1, E6, and E7 for multiplex assays and both E6 and E7 for duplex assays). To be considered HPV negative, the specimen must have a positive result in a separate human β-globin PCR assay to verify the quality of the specimen and DNA isolation. Under MRL's standard clinical testing rules, samples that result in a single-ORF-positive result would be retested from a fresh aliquot of the specimen for verification. In this analysis, single-ORF-positive results were accepted and not independently verified due to limited sample availability.

HPV multiplex PCR was performed in either a 25-μl (Stratagene instrument) or 50-μl (ABI instrument) total reaction mixture containing 12.5 μl or 25 μl Qiagen QuantiTect PCR master mix (low ROX formulation for Stratagene instrument), 0.25 μl or 0.5 μl each of uracil-DNA glycosylase, 100× combined primer stock, and 100× stock of each fluorescently labeled probes, and 7.25 μl or 18.5 μl of diethyl pyrocarbonate (DEPC)-treated water to bring the master mix volume to 21 μl or 46 μl before 4 μl of sample DNA is added.

INNO-LiPA HPV genotyping.

The INNO-LiPA HPV Genotyping Extra assay utilizes a cocktail of biotinylated consensus primers (SPF10) to amplify a 65-bp region within the L1 ORF of multiple HPV types. The PCR mixture (50 μl) consisted of a master mix containing 37.7 μl of HPV Genotyping Extra Amp, 2.3 μl of HPV Genotyping Extra ENZ, and 5 μl of DEPC-treated water before 5 μl of sample DNA was added.

The biotinylated PCR products are then genotyped by hybridization to HPV type-specific oligonucleotide probes bound to nitrocellulose membranes and detected by an alkaline phosphatase-streptavidin conjugate and colorimetric detection. Post-PCR hybridization and colorimetric detection were performed using the Auto-LiPA 48 instrument in accordance with the manufacturer's recommendations. The preprogrammed test method controlled the hybridization/stringent wash temperatures, timing, reagent addition, and aspiration. At the completion of color development, the strips were scanned and evaluated by the Line Reader and Analysis software (LiRAS) to determine if defined bands for a particular genotype probe were visible. All were confirmed by visual inspection by trained operators. A sample was considered positive if at least one of the defined type-specific banding patterns or one of the HPV control lines is positive.

Statistical analysis.

The swab and thin-section clinical sample HPV results obtained from use of the two methodologies were merged and analyzed by 2-by-3 or 2-by-2 tables, as appropriate. The agreement rate, proportion of positive agreement (P_pos), which is calculated as twice the number of agreed positives/(total number of specimens + number of agreed positives − number of agreed negatives), proportion of negative agreement (P_neg), which is calculated as twice the number of agreed negatives/(total number of specimens − number of agreed positives + number of agreed negatives), and P values by McNemar's test or the Wilcoxon signed-rank test (a measure of the imbalance in the distribution of discordant pairs) were calculated. The relative sensitivity and specificity of the assays were calculated.

To compare the estimated assay sensitivities, limits of detection (LODs) were calculated as the lowest HPV type-specific copy numbers for which the logistic model-predicted positivity rate exceeds 97.5%. The comparison of observed LODs was assessed via simulation to determine the probability and the 95% confidence interval (CI) of an extreme difference in the observed LOD ratios, assuming no difference between methods. Statistical significance was demonstrated if observed LOD ratios fall outside the 95% confidence interval range of the simulated LOD ratio.

RESULTS

Concordance, sensitivity, and specificity between the HPV multiplex PCR assay and the INNO-LiPA HPV genotyping assay in clinical specimens. (i) Swab specimens.

Infection status classification agreement results were organized in a 2-by-3 cross-tabulation for each HPV type, classifying each sample result as positive, negative, or a retest (single-gene positives) for the multiplex HPV PCR assay and classifying each sample result as positive or negative for the INNO-LiPA PCR assay.

As shown in Table 1, the agreement rates between the multiplex HPV PCR and INNO-LiPA assays were at least 90.1% with single-ORF-positive results considered and at least 90.7% without single-ORF-positive result consideration. The overall HPV agreement rates were 93.9% and 94.2% for swab samples, including and excluding single-ORF-positive samples, respectively. For only four HPV types, HPV11, HPV31, HPV33, and HPV52, there was no statistical evidence of an imbalance in the samples between the two assays, as determined by the Wilcoxon signed-rank test when single-ORF-positive results were considered. Without consideration of single-ORF-positive results, HPV51 detection also was not statistically different with either assay. The proportion of positive agreements ranged from 0.233 for HPV59 to 0.941 for HPV11 and was 0.737 for the 14 types combined. The proportion of negative agreements by type ranged from 0.945 for HPV16 to 0.996 for HPV11 and was 0.968 for the 14 types combined. For all HPVs except HPV52, which had balanced discordance, the multiplex HPV PCR assay detected more positive swab samples than the INNO-LiPA PCR assay.

Table 1.

HPV multiplex PCR and INNO-LiPA HPV assay concordance for swab specimens

HPV type	No. of specimens with multiplex PCR/INNO-LiPA result of:						Analyzed with single-ORF-positive results		Analyzed without single-ORF-positive results
	+/+	+/−	−/+	−/−	Retest/+	Retest/−	Agreement rate (%)	Exact P valuea	Agreement rate (%)	Ppos	Pneg	Exact P valueb
6	26	17	3	313	0	1	94.3	0.001	94.4	0.722	0.969	0.001
11	24	2	1	333	0	0	99.2	1.000	99.2	0.941	0.996	1.000
16	52	30	1	268	1	8	90.1	<0.001	91.2	0.770	0.945	<0.001
18	29	13	2	313	1	2	95.4	0.004	95.8	0.795	0.977	0.010
31	39	16	9	294	1	1	92.8	0.199	93	0.757	0.959	0.230
33	24	8	8	319	0	1	95.4	1.000	95.5	0.750	0.976	0.803
35	16	15	0	329	0	0	95.8	<0.001	95.8	0.681	0.978	0.000
39	24	25	3	305	3	0	91.8	<0.001	92.2	0.632	0.956	<0.001
45	18	15	0	327	0	0	95.8	<0.001	95.8	0.706	0.978	0.000
51	60	8	2	286	0	4	96.7	0.026	97.2	0.923	0.983	0.114
52	33	12	12	301	1	1	93.1	1.000	93.3	0.733	0.962	0.838
56	30	24	1	297	1	7	91.9	<0.001	92.9	0.706	0.960	<0.001
58	23	27	1	309	0	0	92.2	<0.001	92.2	0.622	0.957	<0.001
59	5	33	0	317	0	5	90.1	<0.001	90.7	0.233	0.951	<0.001
Overall	403	245	43	4,311	8	30	93.9	<0.001	94.2	0.737	0.968	<0.001

^aP value determined by the Wilcoxon signed-rank test.

^bP value determined by McNemar's test.

(ii) Thin-section specimens.

Table 2 shows the cross-tabulation of infection status classification agreement for the selected thin-section samples and classification of each sample result as positive or negative for both the multiplex HPV PCR and INNO-LiPA PCR assays. Because the thin-section specimens were preselected based upon already existing multiplex HPV PCR data, no single-ORF-positive or “retest” results were present in this data set.

Table 2.

HPV multiplex PCR and INNO-LiPA HPV assay concordance for thin-section specimens

HPV type	No. of specimens with multiplex PCR/INNO-LiPA result of:				Agreement rate (%)	Ppos	Pneg	Exact P valuea
	+/+	+/−	−/+	−/−
6	30	0	1	247	99.6	0.984	0.998	1
11	16	1	0	261	99.6	0.970	0.998	1
16	19	5	6	248	96	0.776	0.978	1
18	25	0	0	253	100	1.000	1.000	1
31	22	3	1	252	98.6	0.917	0.992	0.625
33	20	5	0	253	98.2	0.889	0.990	0.063
35	20	5	0	253	98.2	0.889	0.990	0.063
39	28	2	0	248	99.3	0.966	0.996	0.5
45	12	9	0	257	96.8	0.727	0.983	0.004
51	29	1	6	242	97.5	0.892	0.986	0.125
52	30	3	8	237	96	0.845	0.977	0.227
56	17	6	3	252	96.8	0.791	0.982	0.508
58	18	8	0	252	97.1	0.818	0.984	0.008
59	9	12	0	257	95.7	0.600	0.977	<0.001
Overall	295	60	25	3,512	97.8	0.874	0.988	<0.001

^aP value determined by McNemar's test.

The HPV type-specific agreement rates between the multiplex HPV PCR assay and INNO-LiPA assay were at least 95.7% for thin-section samples. The overall HPV agreement rate was 97.8%. For HPV45, HPV58, and HPV59, the multiplex HPV PCR assay detects significantly more positive samples than the INNO-LiPA assay. For the other 11 HPV types evaluated, there was no statistical evidence of an imbalance in the discordant samples between the two assays. The proportion of positive agreements ranged from 0.600 for HPV59 to 1.000 for HPV18 and was 0.874 for the 14 types combined. The proportion of negative agreements ranged from 0.977 for HPV52 and HPV59 to 1.000 for HPV18, with 0.988 for the 14 types combined.

The swab samples, excluding single-ORF-positive and thin-section samples, were used to asses the sensitivity and specificity of the multiplex HPV PCR assay relative to the INNO-LiPA PCR assay and those of the INNO-LiPA PCR assay relative to the multiplex HPV PCR assay. Table 3 shows that the overall sensitivity for the 14 HPV types in the multiplex assay relative to the INNO-LiPA assay was 0.90 for swabs and 0.92 for thin sections, with overall specificities of 0.95 and 0.98, respectively. The overall sensitivity of the INNO-LiPA assay relative to the multiplex HPV PCR assays was 0.62 for swabs and 0.83 for thin sections, with overall specificities of 0.99 for both swabs and thin sections. Sensitivity was individually lower in both specimen types in the INNO-LiPA assay for HPV45, HPV58, and HPV59 and, in swabs only, was additionally lower for HPV6, HPV16, HPV18, HPV35, and HPV56. The specificities of the two assays were similar with all specimens tested for all HPV types.

Table 3.

Relative sensitivities and specificities for multiplex PCR and INNO-LiPA assays

HPV type	Multiplex PCR relative to INNO-LiPA				INNO-LiPA relative to multiplex PCR
	Sensitivity		Specificity		Sensitivity		Specificity
	Swab	TSa	Swab	TS	Swab	TS	Swab	TS
6	0.90	0.97	0.95	1.00	0.60	1.00	0.99	1.00
11	0.96	1.00	0.99	1.00	0.92	0.94	1.00	1.00
16	0.98	0.76	0.90	0.98	0.63	0.79	1.00	0.98
18	0.94	1.00	0.96	1.00	0.69	1.00	0.99	1.00
31	0.81	0.96	0.95	0.99	0.71	0.88	0.97	1.00
33	0.75	1.00	0.98	0.98	0.75	0.80	0.98	1.00
35	1.00	1.00	0.96	0.98	0.52	0.80	1.00	1.00
39	0.89	1.00	0.92	0.99	0.49	0.93	0.99	1.00
45	1.00	1.00	0.96	0.97	0.55	0.57	1.00	1.00
51	0.97	0.83	0.97	1.00	0.88	0.97	0.99	0.98
52	0.73	0.79	0.96	0.99	0.73	0.91	0.96	0.97
56	0.97	0.85	0.93	0.98	0.56	0.74	1.00	0.99
58	0.96	1.00	0.92	0.97	0.46	0.69	1.00	1.00
59	1.00	1.00	0.91	0.96	0.13	0.43	1.00	1.00
Overall	0.90	0.92	0.95	0.98	0.62	0.83	0.99	0.99

^aTS = thin section.

Comparison of the limits of detection of the HPV multiplex PCR assays and INNO-LiPA HPV genotyping assay.

The assay sensitivity or limit of detection (LOD) is the lowest copy number that can be reliably classified as HPV type-specific positive in the PCR assay. A single run of eight replicates of plasmid or quantified clinical samples in a 2-fold dilution series from 1.56 to 200 copies/test and 1.95 to 250 copies/test were used to assess LODs for the multiplex HPV PCR and INNO-LiPA PCR assays, respectively. Due to the availability of only a single run of data to assess LODs, the difference in the estimated LODs between assay methods was assessed via simulation. Table 4 shows the observed LODs for the multiplex HPV PCR and INNO-LiPA PCR assays and the observed versus simulated LOD ratios with corresponding 95% confidence intervals. HPV6, HPV35, HPV39, HPV45, HPV58, and HPV59 demonstrated differences between the observed and simulated LOD ratios falling outside the 95% CI, indicating that the assays do not perform equally for these HPV types. Only HPV6 was deemed more sensitive in the INNO-LiPA assay in the LOD analysis.

Table 4.

Multiplex HPV PCR and INNO-LiPA observed versus simulated LOD ratios with corresponding 95% CIs^c

HPV type	Observed no. of LOD copies/test		Observed LOD ratio	95% CI of simulated LOD ratio
	Multiplex PCR	INNO-LiPA
6	12.6	<1.95a	6.46	3.34
11	2.8	4	0.7	5.81
16	8.3	15.6	0.53	3.51
18	9.4	6.9	1.35	3.95
31	6.4	12.2	0.52	2.95
33	6.5	4.2	1.55	2.2
35	3.7	23	0.16	3.9
39	4.2	94.4	0.04	3.2
45	3.7	42.9	0.09	3.88
51	4.9	4.3	1.16	3.36
52	13.3	47.8	0.28	5.58
56	22.3	11.7	1.91	5.07
58	3.2	>250b	0.01	4.19
59	12.6	>250b	0.05	3.84

^aThe minimum input copy number/test for INNO-LiPA is 1.95. The LOD result for HPV6 is inconsistent with the sensitivity assessment for HPV6 using clinical samples.

^bThe maximum input copy number/test for INNO-LiPA is 250.

^cHPV6, HPV35, HPV39, HPV45, HPV58, and HPV59 (in boldface) demonstrated differences between the observed and simulated LOD ratios falling outside the 95% CI, indicating that the assays do not perform equally for these HPV types.

DISCUSSION

The results of this HPV assay comparison between the MRL multiplex HPV PCR assays and the INNO-LiPA HPV Genotyping Extra assay for evaluation of 14 HPV types (HPV6, HPV11, HPV16, HPV18, HPV31, HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, and HPV59) show the general comparability of the two assays for most of the HPV types evaluated. A previous study (5) comparing the same two HPV genotyping assays was limited because of the use of separate DNA isolation systems and dilution of the specimens tested in multiplex PCR. There, the multiplex and INNO-LiPA HPV assays were comparable for most HPV types, even though the multiplex assay was executed under suboptimal conditions. The current study approached the assay comparison in a more stringent manner by testing specimen sets with identical processing and DNA isolation in both assay formats. DNA was extracted from clinical swab specimens, thin-section specimens, and quantified HPV plasmids/positive specimens, followed by testing with each assay per the assay's standard protocol. It is important to note that the standard protocols for each assay differ in the volume of template DNA used in the amplification reaction, as follows: 10 μl for INNO-LiPA and 4 μl for multiplex HPV PCR. Here, specimen DNA used in INNO-LiPA was diluted by half (5 μl DNA plus 5 μl DEPC-treated water) to more closely match the input to that of the multiplex HPV PCR assay.

Clinical specimen comparisons were performed, evaluating the assays relative to one another with samples of unknown status. Concordance in swab specimens is shown in Table 1, with analyses performed both including and excluding multiplex HPV PCR single-ORF-positive results. Single-ORF-positive specimens are potential low-copy-number-positive specimens which do not immediately meet the minimum two-ORF-positive criteria for positivity in the multiplex HPV PCR assays and would be repeated in MRL standard HPV clinical testing for independent verification of the HPV status from a fresh aliquot of the same specimen. Due to limited specimen availability for this evaluation, no retesting was performed to resolve single-ORF-positive results to a definitive positive or negative HPV status, so assessments were performed both with and without single-ORF-positive consideration. Regardless of the inclusion of specimens with single-ORF-positive results, statistically significant differences in HPV detection were observed in swab specimens for all types, excluding HPV11, HPV31, HPV33, and HPV52. When single-ORF-positive results are not considered, HPV51 is also excluded from having significantly different detection in the two assays. For all of the remaining HPV types, multiplex HPV PCR was able to detect more positive specimens. Even though overall agreement rates for the detection of 14 HPV types are >90%, it is clear that this agreement is driven by the agreement of HPV-negative specimens, as shown through the high proportion of negative agreement rates. The observed lower proportion of positive agreement rates indicate that there are discrepancies in the two assays' abilities to detect type-specific HPV positives, suggesting differences in assay sensitivity.

Paraffin-embedded biopsy tissue specimens pose an additional challenge for DNA extraction and subsequent HPV detection. Biopsy tissue which is formalin fixed, paraffin embedded, and sectioned often has a low yield in DNA extractions and potentially damaged or modified DNA (3). In the case of DNA damage, PCR assays with smaller amplicons reduce the chances of DNA damage impacting assay performance. In the case of a low yield of DNA from biopsy thin-section specimens, high assay sensitivity is a must due to the potentially low cell numbers and/or copy numbers of target DNA recovered from the tissue. In the thin-section specimens evaluated for this study, Table 2 shows the agreement of HPV detection by type for the multiplex and INNO-LiPA assays. Compared with those of swab specimens, HPV assay agreement rates of thin-section specimens are slightly higher, above 95.7%, and the overall agreement for all 14 types evaluated was 97.8%. This may be due in part to the preselection of specimens determined to be positive by multiplex HPV PCR that was necessary for thin sections due to limited specimen availability. In the thin-section specimens evaluated, less discordance was observed overall, as evidenced by comparatively higher proportion of positive agreement values, and the discordance varied with swab results for certain HPV types. Numerically, more HPV6, HPV16, HPV51, and HPV52 positives were detected by the INNO-LiPA assay, though the discordance was not statistically significant. The only statistically significant discordances (P < 0.05) in thin-section specimens were observed for HPV45, HPV58, and HPV59 and for these types, more specimens were positive by multiplex HPV PCR than INNO-LiPA.

In this comparison study, neither assay was definitively considered a “gold standard” in HPV detection. As such, sensitivity and specificity in both types of clinical specimens evaluated were calculated relative to one another, as shown in Table 3. The assessment of the multiplex assays relative to INNO-LiPA shows a high degree, >0.90, of both sensitivity and specificity. However, the sensitivity rates are lower, 0.62 for swabs and 0.83 for thin sections, when evaluating INNO-LiPA relative to multiplex PCR assays, while specificity remains high for both at 0.99. No clinical diagnoses were available in this study for any of the specimens tested to enable an evaluation of relative clinical sensitivity and specificity for HPV detection via multiplex HPV PCR and INNO-LiPA.

In addition to a head-to-head comparison of HPV detection in clinical specimens, we also performed an estimation of the LOD copy number for each HPV type in both of the assays. This assessment was performed using known copy numbers of plasmids for HPV types run as three-ORF L1/E6/E7 multiplex assays or using previously quantified clinical specimens for two-ORF E6/E7 duplex assays because L1 type-specific plasmids were not available. Since INNO-LiPA is solely L1 based, the use of serial dilutions of quantified clinical specimens for these types was necessary to be able to estimate a comparative LOD for the duplex assay HPV types. HPV assay concordance and positivity increased with the increasing HPV copy number input, as expected. Estimates of LOD by assay were similar for 8 of the 14 HPV types evaluated. The multiplex HPV PCR assays were shown to be significantly more sensitive in the LOD for five types, HPV35, HPV39, HPV45, HPV58, and HPV59, as demonstrated by LOD estimates outside the 95% confidence interval range of the simulated LOD ratio. Only HPV6 was shown to have a significantly lower LOD in INNO-LiPA than in multiplex PCR, but this result is inconsistent with the observed sensitivity assessments in clinical specimens. An additional limitation of the current study is that a more accurate determination of assay LODs would be obtained with more than the single data run or with more replicates (n = 8) for each copy number than were used. This may explain the inconsistent HPV6 INNO-LiPA LOD result observed. A recent study by Klug et al. (6) evaluated the analytical sensitivity of INNO-LiPA SPF10 assays compared to those of four other HPV detection methods for a limited number of HPV types. They demonstrated sensitivity as low as 1,000 viral genome equivalents of HPV6, HPV16, HPV18, HPV31, and HPV33, even in the presence of other HPV types. Our analysis demonstrated sensitivity at below 25 copies per test for all 14 types in the multiplex HPV PCR assays. A total of 9 of 14 types were below 25 copies per test in INNO-LiPA, and 3 additional types were below 100 copies per test. The INNO-LiPA HPV58 and HPV59 LODs were above the highest input copy number for that assay, 250 copies per test. Wide variations in sensitivities across all HPV types evaluated are a concern in consensus primer-based PCR assays and can result in missed or inconsistent HPV type-specific data reporting.

Understanding the performance characteristics of HPV genotyping assays is critically important for accurate interpretation of data from a wide variety of studies, including those evaluating HPV prevalence, HPV disease burden, and HPV vaccine efficacy. One of the benefits of multiplex HPV PCR assays is the “twice positive” algorithm to ensure accuracy. The detection of multiple open reading frames for each HPV type decreases the chances for false-negative results due to the presence of genetic variants of an HPV type or due to integration of viral DNA into the host genome. Requiring multiple type-specific ORFs to be positive also reduces the risk of false-positive results. In standard clinical testing practices, if a single-ORF result occurs in a specimen tested in multiplex HPV PCR, it is retested from a fresh aliquot and not considered positive unless the same (or additional) ORF is reproducible for the same HPV type, thus providing an additional independent verification of positivity status. Here, INNO-LiPA was performed as an independent test, resulting in a single HPV positivity determination for only the L1 ORF for each specimen, as designed. The use of a two-test process of Hybrid Capture II followed by INNO-LiPA genotyping was evaluated by Iftner et al. (5). There, a post hoc analysis of multiplex HPV PCR compared to INNO-LiPA as a stand-alone test suggested that the INNO-LiPA assay is more sensitive than the Hybrid Capture II assay screen for HPV positives.

In conclusion, detection of most of the 14 HPV types common to the multiplex HPV PCR and INNO-LiPA HPV genotyping assays was similar in clinical swab specimens, biopsy thin-section specimens, and estimates of LODs. However, detection of HPV39, HPV58, and HPV59 by INNO-LiPA was consistently less sensitive in all analyses performed. Although there are different applications of HPV detection assays in regard to analytical versus clinical sensitivity, analytically sensitive and type-specific measures of HPV infections are necessary for adequate epidemiological evaluations and are critical for continued monitoring of vaccine efficacy. This study adds to our understanding of the performance characteristics, utility, and comparability of the MRL multiplex HPV PCR and INNO-LiPA HPV genotyping assays.