The performance of two commercially available detection systems for high-risk HPV (hrHPV), Hybrid Capture 2 (HC2) and in situ hybridization (ISH), were compared on cervical scrapings.
The performance of two commercially available detection systems for high-risk HPV (hrHPV), Hybrid Capture 2 (HC2) and in situ hybridization (ISH), were compared on cervical scrapings.
Using general primer (GP)-mediated GP5+/6+-polymerase chain reaction (PCR)-enzyme immunoassay and reverse line blot genotyping, 76 liquid-based cervical samples were identified with ≥ 1 of the 12 hrHPV types present in the probes of the HC2 and ISH assays. The positivity rate of the assays and the HC2 viral load were determined and related to cytologic findings (n = 76 samples) and histologic findings (n = 43 samples).
Overall, HC2 scored significantly more samples positive compared with ISH (P < 0.01). Seventy-four of 76 samples (97%) were positive according to HC2. Forty-six of 76 samples (61%) were positive according to ISH, including 80% and 70% of samples that were classified cytologically as moderate dysplasia and severe dysplasia, respectively. All women with underlying cervical intraepithelial neoplasia (CIN) lesions and 67% of women without CIN had positive HC2 samples. ISH scored 33%, 66%, 88%, and 73% of samples positive of women with no CIN, Grade 1 CIN (CIN 1), CIN 2, and CIN 3, respectively. The HC2 viral load was significantly higher in women who had a cytologic diagnosis of dysplasia (P < 0.01) and in women who had an underlying diagnosis of CIN (P < 0.01) compared with women who had neither. In addition, the viral load was significantly higher in ISH positive samples compared with ISH negative samples (P < 0.01).
An increased HC2 viral load was associated with an increased chance of underlying high-grade CIN disease in women who tested hrHPV GP5+/6+-PCR positive. Moreover, although positive ISH results were associated with an increased overall viral load in the sample, the analytic sensitivity of ISH was too low to detect all women with prevalent high-grade CIN. Cancer (Cancer Cytopathol) 2004;102:11–8. © 2003 American Cancer Society.
It has been well established that infection with high-risk human papillomavirus (hrHPV) is the main causative factor of cervical carcinoma.1, 2 Consequently, adding an hrHPV test to the classic Papanicolaou (Pap) smear may improve the efficacy of cervical carcinoma screening programs, the triage of women with ambiguous or borderline cervical smears, and the monitoring of women who have been treated for high-grade cervical intraepithelial neoplasia (CIN).3–5 Several studies have revealed that HPV testing yields a superior negative predictive value (i.e., the proportion of women without disease among those with negative test results), approaching 100%, for Grade 3 CIN (CIN 3) lesions and cervical carcinomas (lesions ≥ CIN 3).6–10 However, the clinical specificity (i.e., the proportion of women with negative test results among those without lesions ≥ CIN 3) is subject to improvement, because a substantial subset of women who do not have or who develop lesions ≥ CIN 3 test positive for hrHPV.4, 6, 8 Recent data based on real-time polymerase chain reaction (PCR) and semiquantitative methods indicate that an increased viral load within a cervical smear is associated with an increased relative risk of both prevalent and incident lesions ≥ CIN 3,11–15 although differential effects between different viral genotypes have been reported.16, 17
Currently, two commercial hrHPV testing systems are available that are not based on target DNA amplification. In theory, these methods have a lower analytic sensitivity for detecting hrHPV infections compared with most PCR assays. Consequently, these assays may detect women who have an increased risk of lesions ≥ CIN 3 with greater specificity than PCR. Among these, the United States Food and Drug Administration-approved Hybrid Capture 2 (HC2) system (Digene Corporation, Gaithersburg, MD) uses a liquid hybridization format followed by signal amplification to detect 13 hrHPV types (i.e., HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, and HPV-68) using an RNA cocktail probe with a sensitivity of approximately 100,000 viral copies per milliliter of a cervical sample suspension.18 The other system involves a DNA in situ hybridization (ISH) assay (Ventana Medical Systems, Tucson, AZ) using a cocktail probe that represents 13 hrHPV types (i.e., HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, HPV-68, and HPV-70). HC2 and ISH share 12 common HPV types, with each methodology also containing 1 unique type: HPV-39 (HC2) and HPV-70 (ISH). ISH has an analytic sensitivity of approximately 40 HPV copies per nucleus of a given cell. Both methods can be applied to samples collected for liquid cytology.
In this study, we tested the performance of the HC2 and Ventana ISH assays on liquid-based samples of cervical scrapings that, on the basis of general primer (GP)-mediated GP5+/6+-PCR and subsequent reverse line blot typing,19 were found to contain ≥ 1 of the 12 hrHPV types represented in the probe cocktails of both assays. The specific objectives of this study were to determine 1) whether application of either or both assays would increase the clinical specificity of detecting prevalent lesions ≥ CIN 3 without loosing sensitivity and 2) whether ISH positivity of at least one cell reflects an increased overall viral load in a cervical sample.
A total of 115 cervical samples used for the purpose of this study were collected from 115 women during a 3-month period in a routine gynecologic setting at the Department of Obstetrics and Gynecology, VU University Medical Center. Indications for visiting a gynecologist included having an abnormal cervical smear in the population-based screening program and monitoring after treatment for CIN 3. Of each sample, a classic Pap smear was obtained for cytomorphologic analysis using the CISOE-A classification system, which is used currently in the Netherlands.20 According to this system, smears are classified as normal, borderline dysplasia, mild dysplasia, moderate dysplasia, severe dysplasia, suspected carcinoma in situ, or suspected carcinoma. The brush with the remaining material was placed in a container with 20 mL PreservCyt medium (Cytyc Corporation, Boxborough, MA). This suspension was used to make a liquid-based preparation on coated slides using a T2000 device (Cytyc Corporation). Of the material left in the PreservCyt medium, 2 mL were centrifuged for 10 minutes at × 10,000 g; and the pellet was resuspended in 500 μL Tris-HCl, pH 7.4, for subsequent DNA extraction using the High Pure PCR Template Preparation kit (Roche, Penzberg, Germany) for HPV PCR purposes. A second 4-mL sample was prepared for HC2 analysis. In patients who underwent a colposcopy-directed biopsy within 6 months after the cervical smear, the histology data were correlated with HC2 and ISH results.
This study followed the local ethical guidelines of the Institutional Review Board of the VU University Medical Center, Amsterdam. Consent was obtained from all patients after the nature of procedure had been explained.
Extracted DNA was analyzed first by β-globin gene PCR21 to determine whether it was suitable for HPV PCR analysis. All 115 cervical samples were positive according to β-globin PCR and subsequently were subjected to the GP5+/6+-PCR enzyme immunoassay followed by reverse line blot typing, as described previously.19 Of 101 samples that were identified as GP5+/6+-PCR positive, 86 samples contained, as a single infection or as multiple infections, ≥ 1 of the 12 hrHPV types (i.e., HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, and HPV-68) represented in the cocktail probes of both the HC2 assay and the ISH assay. In 10 samples, insufficient residual material remained for testing with HC2, leaving a total of 76 cervical samples that were used ultimately in this study. The 76 samples were from women with a mean age of 35 years (range, 19–63 years). These included 23 samples that were classified as normal (mean age, 33 years; age range, 22–55 years), 14 samples that were classified as borderline dysplasia (mean age, 31 years; age range, 22–58 years), 14 samples that were classified as mild dysplasia (mean age, 40 years; age range, 19–63 years), 15 samples that were classified as moderate dysplasia (mean age, 36 years; age range, 25–57 years), and 10 samples that were classified as severe dysplasia (mean age, 37 years; age range, 25–54 years).
Sample material collected in PreservCyt medium was made suitable for HC2 using a sample conversion kit (Digene Corporation). In brief, 400 μL of conversion buffer were added to 4 mL of sample material, mixed vigorously, and centrifuged at × 3000 g for 15 minutes. The pellet was resuspended in a mixture of 50 μL denaturation reagent containing indicator dye and 100 μL specimen transfer medium (STM) and was incubated for 15 minutes at 65 °C. The suspension was mixed vigorously and incubated for another 30 minutes at 65 °C. Seventy-five microliters of the sample suspension were used for HC2 testing with the hrHPV cocktail probe specific for 13 hrHPV types (i.e., HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, and HPV-68) according to recommendations of the manufacturer.22 Positive controls containing 1 pg/mL of cloned HPV-16 DNA and negative controls (provided by the manufacturer) were included in each assay. Positive controls tested positive throughout, whereas negative controls were negative consistently in the HC2 assay.
The results of the HC2 assay were expressed as relative light units (RLU), representing the ratio between the emission from a sample to the average of three positive controls. Samples with an RLU ≥ 1 were scored HC2 positive. Moreover, RLU values, including those from samples with values < 1 RLU, were used to assess the viral load.
Liquid-based preparations were subjected to ISH using the Ventana HPV lipopolysaccharide-binding protein (LBP) high-risk cocktail probe (specific for HPV-16, HPV-18, HPVv31, HPV-33, HPV-35, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, HPV-68, and HPV-70), the automatic BenchMark device, and the standardized BenchMark ISH View Blue Detection Kit for HPV (Ventana Medical Systems), as recommended by the manufacturer. Briefly, slides were incubated first with the HPV LBP high-risk probe and, subsequently, with a biotinylated antifluorescein antibody to bind the probe. The bound antibody was visualized with an indirect biotin streptavidin system coupled to alkaline phosphatase. Samples were scored positive in the Ventana ISH assay when at least one cell in the slide showed clear nuclear staining. Two individuals independently scored slides. In four samples with discrepant findings, revision was performed by a third individual for definitive scoring. Liquid-based preparations of CaSki cells (containing ≈ 500 copies of HPV-16) served as positive controls in all assays. CaSki cells also were tested with the ISH negative control probe (Ventana Medical Systems). In addition, slides with SiHa cells (containing 1–10 copies of HPV-16) were included in the assays. CaSki cells were consistently ISH positive, whereas SiHa cells did not stain in any of the assays. CaSki cells did not stain after hybridization with the ISH HPV negative probe.
For a mutual comparison of the positive rates of the HC2 assay and the ISH assay and their relation to cytology and histology, a two-sided sign test for paired scores was used. The positive rates of each test separately were compared using the chi-square test or the Fisher exact test for small group sizes. For comparison of the viral load between the different histopathologic categories and between ISH positive versus ISH negative samples, the mean RLU values were compared using the Mann–Whitney U test. P values < 0.05 were considered significant for α.
Fifty-three of 76 HPV GP5+/6+-PCR positive cervical samples used in this study contained a single HPV type, and 23 samples contained multiple HPV types. The latter included 16 samples that also contained HPV types that differed from the 12 hrHPV types that the cocktail probes of the HC2 and ISH assays had in common (Table 1).
|HPV type||Normal||Borderline dysplasia||Mild dysplasia||Moderate dysplasia||Severe dysplasia||Total|
Multiple infections were most common in normal smears (35%), borderline dysplastic smears (36%), and mild dysplastic smears (43%); and their frequency decreased to 13% and 20% in smears classified as moderate dysplasia and severe dysplasia, respectively (Table 1).
Among the 12 hrHPV types that were selected specifically for this study, HPV-16 was most prevalent and was found in 33 samples as single infection (n = 22 samples) or as multiple infections (n = 11 samples). This was followed by HPV-56 (n = 14 samples); HPV-31 (n = 10 samples); HPV-33, HPV-51, and HPV-52 (n = 5 samples each); HPV-18 and HPV-45 (n = 4 samples each); HPV-35 and HPV-58 (n = 3 samples each); and HPV-59 and HPV-68 (n = 1 sample each) (Table 1).
Seventy-four of 76 samples (97%) were scored positive by the HC2 assay. The two negative HC2 samples included 1 normal sample that contained HPV-16 and HPV-81 and 1 mildly dysplastic sample that contained HPV-56 (Tables 1, 2A).
|Result||No. of patients (%)||P valuea||RLU HC2|
|Normal||10/23 (43)||22/23 (96)||< 0.01||17||3–85|
|Borderline dysplasia||8/14 (57)||14/14 (100)||0.03||34||7–221|
|Mild dysplasia||9/14 (64)||13/14 (93)||0.22||143||22–834|
|Moderate dysplasia||12/15 (80)||15/15 (100)||0.25||629||310–1068|
|Severe dysplasia||7/10 (70)||10/10 (100)||0.25||94||24–480|
|Total||46/76 (61)||74/76 (97)||< 0.01||—||—|
|No CIN||1/3 (33)||2/3 (67)||1.00||3||1–5|
|CIN 1||6/10 (60)||10/10 (100)||0.13||101||22–430|
|CIN 2||7/8 (88)||8/8 (100)||1.00||447||70–933|
|CIN 3/ACIS||16/22 (73)||22/22 (100)||0.03||346||83–825|
|Total||30/43 (70)||42/43 (98)||< 0.01||—||—|
The ISH assay was positive in 46 samples (61%). The number of positive cells varied from 1 to > 100 cells per slide, but most positive samples exhibited between 5 and 15 positive nuclei. Representative examples of ISH staining are shown in Figure 1. Women who had any degree of dysplasia showed a higher ISH positive rate (36 of 53 women; 68%) compared with women who had normal smears (10 of 23 women; 43%; P = 0.05). ISH positivity increased from 43% in normal smears to 80% in moderately dysplastic smears but declined to 70% in severely dysplastic smears (Table 2, section A). However, the differences in ISH positivity between the various cytologic categories were not statistically significant.
Nondetection of HPV with the ISH assay was not related specifically to the lack of certain HPV types as probes or to a particular HPV type in the specimens. The samples that remained negative in the ISH assay included smears containing single infections of HPV-16 (n = 8 smears), HPV-18 (n = 1 smear), HPV-31 (n = 2 smears), HPV-33 (n = 2 smears), HPV-35 (n = 1 smear), HPV-45 (n = 1 smear), HPV-51 (n = 1 smear), HPV-52 (n = 1 smear), HPV-56 (n = 8 smears), HP-V59 (n = 1 smear), and several multiple infections. HPV-56 was missed most often by ISH. ISH was negative in 8 of 10 scrapings that contained HPV-56 as a single infection (Table 1).
Overall, the difference in the positivity rate between the HC2 assay and the ISH assay was statistically significant (P < 0.01). Although HC2 scored more samples positive in all cytologic categories (Table 2, section A), the difference in the positivity rate between the methods was statistically significant only for normal smears (P < 0.01) and for borderline dysplastic smears (P = 0.03; Table 2, section A). However, when smears were subgrouped into 3 categories (i.e., normal, borderline and mild dysplasia [BMD], and moderate and severe dysplasia [> BMD]), differences in the positivity rate between both assays appeared statistically significant for all categories (normal, P < 0.01; BMD, P = 0.01; > BMD, P = 0.03).
Colposcopy-directed biopsies were undergone by 44 women, and 33 women did not undergo biopsy. Histopathologically, the biopsies comprised no CIN (n = 3 women; mean age, 44 years; age range, 37–52 years), CIN 1 (n = 11 women; mean age, 42 years; age range, 30–63 years), CIN 2 (n = 8 women; mean age, 37 years; age range, 26–57 years), and CIN 3/adenocarcinoma in situ (n = 22 women; mean age, 36 years; age range, 25–54 years).
The HC2 positivity rate was 67% (i.e., 2 of 3 samples) for women with no CIN and 100% for the 40 women with any CIN lesion (Table 2, section B). ISH results were positive in 33% of women (1 of 3 samples) without CIN and in 73% of women (29 of 40 samples) with any CIN lesion. These differences were not statistically significant, possibly due to the small sample size. Among women with CIN, the ISH results were positive in 60% of women with CIN 1, in 88% of women with CIN 2, and in 73% of women with CIN 3; again, these differences were not statistically significant.
Overall, in women who underwent biopsy, the HC2 positivity rate was significantly higher compared with the ISH positivity rate (P < 0.01). After stratification into the different histologic categories, no significant differences in the positivity rate between the two assays were found for women with no CIN, CIN 1, and CIN 2. However, the HC2 assay scored significantly more samples positive in women with CIN 3 (P = 0.03). When women were subgrouped according to histology into 2 categories (i.e., ≤ CIN 1 and ≥ CIN 2), differences in positivity rate between both assays appeared statistically significant only for the category ≥ CIN 2 (P = 0.02). Data on the ISH positive and HC2 positive results in relation to both cytology and histology are summarized in Table 3.
|Cytology results||No. of positive histology results (%)|
|No biopsy||No CIN||CIN 1||CIN 2||CIN 3/ACIS|
|Normal||8/20 (40)||19/20 (95)||0/1 (0)||1/1 (100)||1/1 (100)||1/1 (100)||1/1 (100)||1/1 (100)||—||—|
|Borderline dysplasia||5/7 (71)||7/7 (100)||—||—||1/3 (33)||3/3 (100)||—||—||2/4 (50)||4/4 (100)|
|Mild dysplasia||1/3 (33)||3/3 (100)||1/2 (50)||1/2 (50)||2/4 (50)||4/4 (100)||2/2 (100)||2/2 (100)||3/3 (100)||3/3 (100)|
|Moderate dysplasia||2/3 (67)||3/3 (100)||—||—||2/2 (100)||2/2 (100)||3/4 (75)||4/4 (100)||5/6 (83)||6/6 (100)|
|Severe dysplasia||—||—||—||—||—||—||1/1 (100)||1/1 (100)||6/9 (67)||9/9 (100)|
|Total||16/33 (48)||32/33 (97)||1/3 (33)||2/3 (67)||6/10 (60)||10/10 (100)||7/8 (88)||8/8 (100)||16/22 (73)||22/22 (100)|
Considering the cytologic categories, HC2 viral load levels were significantly higher in women who had any degree of dysplasia (median RLU, 182; interquartile range [IQR], 21–726) compared with women who had normal smears (median RLU, 17; IQR, 3–85; P < 0.01). Among women with dysplasia, the highest viral load levels were found in those with moderate dysplasia (median RLU, 629; IQR, 310–1068) (Table 2, section A). These levels, however, did not differ significantly from those in mildly dysplastic samples (median RLU, 143; IQR, 22–834) and severely dysplastic samples (median RLU, 94; IQR, 24–480) but were significantly higher compared with the viral load in normal and borderline dysplastic smears (borderline: median RLU, 34; IQR, 7–221; P = 0.01). After stratification of samples into normal, BMD, and > BMD, the viral load levels appeared significantly higher in > BMD smears (n = 25 smears; median RLU, 321; IQR, 94–985) compared with BMD smears (n = 28 smears; median RLU, 80; IQR, 10–654) and normal smears (n = 23 smears; median RLU, 17; IQR 3–85; P < 0.01). There were no meaningful differences in these values when only smears that contained single infections were included in the analysis.
With respect to histology, women who had CIN (n = 40) had higher viral loads (median RLU, 301; IQR, 55–771) (Table 2B) compared with women who did not have CIN (n = 3 samples; median RLU, 3; IQR, 1–5; P < 0.01). Among women with CIN lesions, the median RLU values were 101 (IQR, 22–430) for CIN 1, 447 (IQR, 70–933) for CIN 2, and 346 (IQR, 83–825) for CIN 3, with none of these values differing significantly. After the categorization of women into groups with ≤ CIN 1 and ≥ CIN 2, median viral load levels appeared significantly higher in samples from women who had ≥ CIN 2 (median RLU, 346; IQR, 83–825) compared with women who had ≤ CIN 1 (RLU, 28; IQR, 7–235; P = 0.04).
Moreover, the HC2 viral load was significantly higher in ISH positive samples (n = 46 samples; median RLU, 346; IQR, 87–970) compared with ISH negative samples (n = 30 samples; median RLU, 10; IQR, 4–143; P < 0.01). In addition, when only samples with single infections were taken into account, the viral load remained significantly higher in ISH positive samples (RLU, 373; 95% confidence interval, 138–502) compared with ISH negative samples (RLU, 8; 95% confidence interval, 9–69; P < 0.01).
In this study, two commercially available hrHPV detection methods were applied to liquid-based samples of 76 cervical scrapings that were positive for ≥ 1 of 12 hrHPV types according to GP5+/6+-PCR analysis. HC2 scored significantly more samples positive compared with ISH and, compared with PCR, only two samples were HC2 negative, and neither of those two women had CIN. This difference in the positivity rate can be explained at least in part by the fact that HC2 is a signal-amplification technology, whereas the ISH test does not use an amplification step.
We found that ISH positivity was correlated strongly with increased HC2 viral loads and, consequently, reflected an overall increased viral load in a cervical sample. Given the relation between viral load and CIN disease, it may be assumed that this method allows differentiation between women with and without CIN disease. However, the ISH assay missed a substantial subset of women (i.e., 23%; 7 of 30 samples) with high-grade CIN lesions, indicating that its analytic sensitivity of ≈ 40 HPV copies per cell is too low for use in clinical practice or cervical cancer screening programs. Whether an increase in sensitivity of this method would result in a better performance in terms of clinical sensitivity and specificity remains a subject for future studies.
In this series of samples, the HC2 assay could not differentiate between women with normal cytology and women with dysplasia or between women without and without CIN when the cut-off point of 1 pg/mL was used. Consequently, when using this cut-off point, the HC2 assay is unlikely to increase markedly the clinical specificity of lesions ≥ CIN 3 compared with GP5+/6+-PCR. However, HC2 viral load levels were significantly higher in women who had cytologic and/or histologic abnormalities compared with their normal counterparts. Moreover, the viral load was significantly higher in > BMD smears compared with ≤ BMD smears and in smears from women that revealed ≥ CIN 2 compared with smears that revealed ≤ CIN 1. These values did not alter substantially after an analysis of samples with single infections only, indicating that these differences were not due to potential differences in the frequency of multiple HPV infections or bias due to a cumulative level of viral loads in samples that were infected by multiple HPV types.
The data collected are in line with the concept that increased viral loads are associated with an increased chance of CIN, with a sustained high load or even an increased load implying an increased risk of progression to high-grade CIN or cervical carcinoma.23 Whether a low viral load in women with CIN is predictive of future viral clearance and regression of the lesion could not be addressed in this cross-sectional study, which lacked follow-up.
The HC2 viral load data collected in this study were in line with data from recent studies17, 24 showing that viral load analysis, to a certain degree, can differentiate women with normal histology or cytology from women with any level of cytologic or histologic abnormality, respectively. However, compared with those studies, median viral load levels tended to be higher in our study, in line with the finding that HC2 with the 1 pg/mL cut-off point was positive for almost all GP5+/6+-PCR positive samples. This seems to contradict findings by others showing that PCR-based methods score more samples positive than HC2.17 However, it is worth noting that this study involved a selected group of women who visited a gynecologic outpatient clinic, and data collected from this hospital population cannot be extrapolated simply to a normal screening population. In addition, due to the conversion protocol for samples collected in PreservCyt Solution, as used in this study, an approximately two-fold concentrated sample was tested ultimately in the HC2 compared with samples that would have been collected in 1 mL STM. Therefore, the sensitivity of this assay may have been slightly higher than that in samples collected in STM.
A potential differential effect of viral load by different HPV types, as reported by others,16, 17 could not be addressed in this study because of the small sample numbers. Finally, it is noteworthy that no definitive conclusions can be drawn about the clinical sensitivity and specificity of HC2 and ISH compared with PCR, because the study involved a selected group that did not include GP5+/6+-PCR negative samples or samples that contained HPV types that differed from the set studied here.
The current data suggest that not HC2 positivity, per sé, but an increased HC2 viral load is associated with an increased chance of underlying, high-grade CIN disease in women who have positive hrHPV GP5+/6+-PCR results. Moreover, although positive ISH results are linked tightly to increased viral loads in cervical samples, the sensitivity of ISH is too low to detect all women with high-grade CIN disease.
The authors thank Prof. C. J. L. M. Meijer for his continuous support. They are grateful to Mr. B. Bakker, Mrs. T. Tadema, Mrs. N. Fransen Daalmeijer, and Mrs. P. van Diemen for excellent technical assistance. Coated slides for liquid-based cytology were kindly provided by Ventana Medical Systems. Sample Conversion Kits and Hybrid Capture 2 testing kits were kindly provided by Digene Corporation.