Human papillomavirus genotype detection from archival papanicolaou-stained cervical tests

Abstract

BACKGROUND:

Archival Papanicolaou (Pap)-stained cervical cytology tests may be the only source of a clinical sample for the evaluation of previous human papillomavirus (HPV) infection. Pap tests are ideal because the majority of women in countries with comprehensive screening programs would have had several collected and stored.

METHODS:

In the current study, HPV detection and genotyping were compared in samples collected from a conventionally fixed Pap test with those collected using an endocervical brush and collected in PreservCyt (liquid-based) in 87 women undergoing management for a high-grade Pap test abnormality. Cytology slides were scanned to create high-resolution digital images before the removal of cells because the DNA extraction process resulted in the destruction of the cells from the original sample.

RESULTS:

All previously identified high-grade abnormalities on the Pap tests were detectable on the digital images. β-globin was detected in all extracted Pap tests, indicating the presence of recoverable, amplifiable DNA. A total of 62 (71.3%) and 59 (67.8%) tests were found to have high-risk (HR) HPV detected on PreservCyt and fixed Pap test slides, respectively, with >87% concordance for the detection of HR HPV genotypes. Complete HPV genotyping concordance was observed in 62% and was partial in 26% of sample pairs, with very good agreement for HPV types 16 and 18 (κ = 0.850 and 0.903, respectively). Only 1 Pap test slide was found to be positive whereas the PreservCyt had no detectable HPV DNA, demonstrating a low false-positive rate (1%).

CONCLUSIONS:

The results of the current study confirm that imaging and subsequent HPV detection and genotyping in archival cervical smears can offer accuracy in HPV detection that is comparable to endocervical brush-collected PreservCyt samples. Cancer (Cancer Cytopathol) 2010;. © 2010 American Cancer Society.

The role of certain genotypes of human papillomavirus (HPV) as the primary causative factor for the development of anogenital cancers, especially invasive cervical cancer, is now well established.1-7 From the more than 100 HPV genotypes described, >40 are known to selectively infect the anogenital region.8 On the basis of epidemiologic data, those that are primarily associated with low-grade lesions are classified as “low-risk” (LR) genotypes, whereas those that have an association with high-grade lesions and cervical cancer are classified as “high-risk” (HR) genotypes. At least 13 genotypes are recognized as HR types (types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68).4, 9 However, 2 HPV types, HPV-16 and HPV-18, are more virulent, being consistently detected worldwide in approximately 70% of cervical cancer cases.10, 11 Consequently, 2 currently available prophylactic vaccines target these 2 types. Detection of HPV types in archival samples could be important in the postvaccination surveillance of women of screening age who may have had previous exposure to a particular HPV type. In addition, archival samples could also provide further information regarding the natural history of HPV infection.

The majority of women in Australia and several other countries with comprehensive Papanicolaou (Pap) test screening programs undergo routine cervical cancer screening, and these slides are usually stored for several years. Therefore, archival cytology slides can provide a readily available source of specimens for HPV detection and genotyping. Extraction of DNA from archival exfoliated cytology has previously demonstrated good yields of amplifiable DNA12-15 from tests of up to 10 years old. Detection of HPV DNA has also been performed successfully from archival cervical tests,16-18 demonstrating a high HPV positivity rate in those with high-grade squamous intraepithelial lesions (HSIL). A disadvantage of this method is the destruction of the cells on the original slide. Therefore, imaging of the slide is needed to enable future review of the cytopathology. To the best of our knowledge, assessment of the accuracy of HPV detection and genotyping from Pap tests has not been previously performed by direct comparison with matched fresh endocervical brush samples. Therefore, the objective of the current study was to evaluate the accuracy of HPV genotypes obtained in archival Pap-stained cervical cytology tests by direct comparison with routine endocervical brush samples collected in PreservCyt medium simultaneously. In addition, the imaging of the slides was evaluated to determine whether the quality of the image was comparable to that of the original cytology slide for diagnostic purposes.

MATERIALS AND METHODS

Study Population and Specimens

Specimens collected between May 2001 and December 2002 from a large study of 1679 women undergoing ablative treatment for a histologically confirmed cervical abnormality at the Royal Women's Hospital, Melbourne, Victoria, Australia were used.19 For the conventional Pap tests, immediately before undergoing ablative or excisional treatment for histological abnormalities of the cervix, cervical brush specimens were collected by Cervex brushes (Rovers Medical Devices BV, Oss, The Netherlands), placed onto glass slides, stained using Pap stain, and analyzed cytologically. Residual cells were rinsed into ThinPrep vials containing PreservCyt solution (Cytc Corporation, Bedford, Mass). Tissue excised under colposcopic guidance by loop electrosurgical excision procedure or cone biopsy was assessed histologically. DNA extracted from PreservCyt fluid as well as from conventional Pap tests was analyzed for the detection of HPV genotypes. Overall, 87 liquid-based cervical brush specimens and matching conventional Pap tests were selected for analysis for the current study. All Pap tests were stored 6 to 8 years before processing for this study. This study was approved by the Research and Ethics Committees of the Royal Women's Hospital, with all patients providing a written informed consent form.

Imaging of Pap Tests

Before extraction of the cellular material from Pap tests, slides were digitally scanned using the ScanScope T3 virtual microscopy slide scanner (Aperio, Vista, Calif) and ScanScope Console software (version 7.00.08.1020) provided the user interface. After all the slides were scanned, the digital images were analyzed and confirmed for quality at ×400 magnification using the ImageScope software package (Aperio). Detection of HSIL on the image was evaluated by selecting 23 slides (including 13 confirmed HSIL cases) from routine Pap screening that were heavily weighted with cases that had previously been reported as being high grade on manual screening. This was evaluated in a blinded fashion to ensure that high-grade abnormalities were detectable on the imaged slides.

DNA Extraction

DNA from PreservCyt-extracted samples using the MagNA Pure LC system was used.20 Extracted DNA was stored in −30°C before testing. Pap-stained slides were processed using a modification of a previously described method.18 Briefly, each Pap test slide was placed in xylene in separate disposable 50-mL tubes for 2 to 7 days until the coverslips could be easily removed. The test material was collected with a new razor blade and transferred into a 1.5-mL centrifuge tube with 1 mL of fresh xylene. After a 45-minute incubation to remove the remaining mounting solution, the material was pelleted by centrifugation, washed twice in 95% ethanol, and air dried at room temperature. The tissue pellet was then digested at 60°C for at least 1 hour with 200 μL of tissue lysis buffer (Roche Molecular System, Alameda, Calif) and 4 μL of proteinase K solution (MagNA Pure LC DNA 1 kit; Roche Molecular System). Lysed material was extracted using automated MagNAPure LC (Roche Molecular System) and the associated DNA Isolation Kit I in accordance with the manufacturer's protocol.

HPV Detection and Genotyping

HPV genotyping of DNA extracts was performed using the LINEAR ARRAY (LA) HPV Genotyping Test (Roche Molecular System) according to the manufacturer's instructions, with minor modifications.21, 22 The LA can simultaneously detect up to 37 mucosal HPV genotypes, including 13 HR types (types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68) and 24 LR types (6, 11, 26, 40, 42, 53, 54, 55, 61, 62, 64, 66, 67, 69, 70, 71, 72, 73, 81, 82, 83, 84, IS39 [82 variant], and CP6108).23, 24 The test also incorporates the amplification of a region within the β-globin gene as an internal control. Polymerase chain reaction (PCR) was performed in a 100-μL volume using 50 μL of extracted DNA.21, 22 PCR products were denatured, hybridized, and detected using the recommended LA protocol, with the modification of using a shaker and dry-air incubator at 53°C in place of the shaking waterbath.21 Genotyping strips were interpreted using the HPV reference guide and classified as: “HR HPV positive,” (detection of ≥1 of the 13 HR HPV genotypes), “LR HPV” (detection of ≥1 of the 24 LR HPV genotypes), or “HPV negative,” (no HPV genotypes detected).

A limitation of the HPV LA test is the cross-reactivity of the HPV 52 probe with types 33, 35, and 58 amplicons. Samples that were positive for the HPV-52 probe alone were classified as HPV-52–positive, whereas those that were reactive with this probe and ≥1 of the HPV types 33, 35, and/or 58 probes were tested further using a real-time PCR assay and HPV-52–specific hydrolysis probe.25 This assay specifically detects HPV-52 DNA in the presence or absence of other genotypes.

Results of the LA were also compared with previously evaluated Digene Hybrid Capture (Qiagen, Valencia, Calif) and Roche Amplicor HPV (Roche Molecular System) assays.26

Specimen contamination and carryover were prevented by frequent glove changing, using barrier tips, prior aliquoting of all reagents, and performing pre-PCR and post-PCR steps in different rooms specifically allocated for PCR.

Statistical Analysis

We examined agreement between the screening of conventional Pap slides and the image by estimating the kappa statistic, with 95% confidence intervals (95% CIs). HPV results were cross-tabulated according to the HPV type that was detected by the Pap test and endocervical brush collected in PreservCyt. The frequencies of HPV detection and genotypes were compared between Pap tests and PreservCyt using a chi-square or Fisher exact test when appropriate. A 2-tailed P value of <.05 was used for all tests.

RESULTS

Imaging of Pap Tests

Detection of HPV DNA from Pap-stained cervical tests required removal of cells from Pap-stained slides and hence eliminated the possibility of subsequent review of cytology. To have the Pap test cytopathology available for subsequent review, the slides were scanned and imaged (Fig. 1).

Figure 1. Scanned image of a Papanicolaou (Pap) test before removal of cells from the Pap test slide is shown. Arrows indicate high-power examination of the abnormal areas.

A blind review of 23 slides (including 13 HSIL and 10 from routine cytology screening) demonstrated the detection of all presumptive high-grade abnormalities, with the remainder being normal cytologically (Table 1) (κ = 0.92 [95% CI, 0.75-1.0]).

Table 1. Comparison of Detection of High-Grade Abnormality on Slide Image and Slide^†

	Original Pap Slide
Scanned Pap Image	Negative	LSIL	HSIL	Total
Pap indicates Papanicolaou; LSIL, low-grade squamous intraepithelial lesion; HSIL, high-grade squamous intraepithelial lesion. ^a Kappa = 0.92 (95% confidence interval, 0.75-1.0).
Negative	9	1	0	10
LSIL	0	0	0	0
Possible HSIL	0	0	2	2
HSIL	0	0	11	11
Total	9	1	13	23

HPV Detection and Genotyping

The recovery of amplifiable DNA from archival smears, as measured by amplification of human β-globin, demonstrated the presence of an amplicon for all 87 smears evaluated. The signal for β-globin was found to be lower from conventionally prepared Pap-stained smears, with the majority only displaying the high copy β-globin band (Fig. 2). Representative samples are shown in Figure 2, demonstrating close similarity of detectable genotypes. Comparison of HPV genotypes demonstrated no significant difference between the 2 sample types (Table 2), although more HPV genotypes were detected in liquid-based specimens compared with those extracted from conventional slides. Concordance of the detection of HPV-16 and HPV-18 was assessed as very good (κ = 0.85 [95% CI, 0.73-0.97] and 0.90 [95% CI, 0.71-1.00], respectively) (Table 2).

Figure 2. Comparison of human papillomavirus genotypes obtained from PreservCyt media and respective Papanicolaou (Pap) tests is shown. Linear Array strips from each pair (right strip, Pap test; left strip, PreservCyt) are shown with respective genotypes. Strip pairs indicating different genotypes are highlighted in bold and underlined.

Table 2. Comparison of HPV Detection and Genotypes Obtained From Conventionally Collected and Processed Pap Tests and Respective Liquid-Based Prepared PreservCyt Media

Result	PreservCyt (%)	Pap Test (%)	P
HPV indicates human papillomavirus; Pap, Papanicolaou; LR, low-risk; HR, high-risk; HPV-16, HPV type 16; HPV-18, HPV type 18.
Negative	16 (18.4)	20 (23.0)	.57
Single HPV	30 (34.5)	32 (36.8)	.87
Multiple HPV	42 (48.3)	36 (41.4)	.45
LR HPV	9 (10.3)	8 (9.2)	.99
HR HPV	62 (71.3)	59 (67.8)	.74
HPV-16	30 (34.5)	32 (36.8)	.87
HPV-18	6 (6.9)	5 (5.7)	.99

Overall, 54 (62%) conventional Pap tests were found to have genotypes that were identical to those detected from PreservCyt, whereas an additional 23 (26%) were found to have fewer types detected than were present in PreservCyt (Fig. 3). Ten PreservCyt samples (11.5%) had additional or discordant HPV genotypes detected, with only 1 Pap test sample having HPV detected (genotypes 31 and 45) whereas the matching PreservCyt sample was HPV negative (Fig. 3).

Figure 3. Detailed genotype profile of all samples comparing Papanicolau (Pap) tests with PreservCyt is shown. Numbers depict the number of sample pairs in each category. Those not highlighted were a perfect match for the human papillomavirus (HPV) genotype. Those that are lightly highlighted had fewer types of HPV detected by the Pap test and those with darker hashed lines are those that had more or different types of HPV detected on Pap test slides.

A comparison of the HR genotypes detected on Pap tests with cytological diagnosis demonstrated that all 28 HSIL smears studied were HR HPV positive (Table 3). The conventional Pap test slides were originally reported as having no abnormality detected in 22 smears, 10 of which were HR HPV positive. Histology on available biopsies at the time of treatment indicated that 6 of these 10 HR HPV-positive smears had cervical intraepithelial neoplasia of type 2 and greater, with an additional 3 being positive for HPV DNA by other assays such as Amplicor and/or Hybrid Capture (data not shown), thus suggesting underlying HPV in the absence of morphological changes.

Table 3. Comparison of Detection of HR HPV Genotype on Pap Tests Compared With Cytological Diagnosis

Pap Test HR HPV Detection	Cytology
	Negative	LSIL	Possible HSIL	HSIL	Unsatisfactory/ Not Available	Total
HR HPV indicates high-risk human papillomavirus; Pap, Papanicolaou; LSIL, low-grade squamous intraepithelial lesion; HSIL, high-grade squamous intraepithelial lesion.
Negative	12	6	4	0	6	28
Positive	10	11	8	28	2	59
Total	22	17	12	28	8	87

DISCUSSION

Detection of HPV genotypes in archival samples could be important in postvaccination surveillance in women to assess previous exposure to HPV genotypes, as well as to obtain further information on the natural history of HPV infection. Because the majority of women in countries with comprehensive Pap screening programs undergo routine cervical cancer screening, archival cytology slides can provide a readily available source of specimens for HPV detection and genotyping. The Australian accreditation standards for cytology laboratories require the retention of Pap test slides for a designated period of time (in Australia, cervical cytology smears are required to be stored for 10 years27) to enable future evaluations or a review of slides. With advances in imaging and data storage, it is now possible to digitally scan the entire surface of the slide and store the digital data for future reference. Such an imaging system is currently being used for telecytology, automated screening of Pap test slides, and quality assurance, as well as training and education.28 In the current study, the blind evaluation of 23 slides (including oversampling for 13 HSIL slides), indicated that all cases with high-grade abnormalities were detectable on the evaluation of scanned images of slides at ×400 magnification. Studies with more cases would need to be performed before the implementation of this method by each laboratory. However, it is important that patients provide informed consent for this additional technique because the image only captures a single plane of the 3-dimensional smear. Thus, it is possible that abnormal cells are not easily observed, particularly in areas in which >1 layer of cells are aggregated, because of the inability to adjust the depth of the image in the scanned slides. In comparison, it is possible to easily adjust the focal plane while examining tests by conventional microscopy. In implementing this digital image storage process, all tests obviously would need to have a complete and thorough descriptive cytological diagnosis before the recovery of cellular material. The additional high-resolution digital scanning of these tests before this process allows for the visualization, archiving, and easy access of the majority, if not all, of the abnormalities on the test.

Previous studies have shown that good-quality nucleic acid can be recovered12-14 in cervical tests after as much as 10 years of storage. HPV detection has also been described for Pap tests up to 9 years old, with 98% recoverability.17 In the current study, adequacy of amplification for slides stored for up to 6 years was noted when all tests were evaluated. Comparison of HPV genotypes was performed using LA, which targets a 450-base pair (bp) region of L1 gene. Although it has been reported previously that targeting amplicons >200 bp could result in lower detection levels,17 the current study used methodology identical to that routinely used on PreservCyt samples. Thus, direct comparison of HPV genotypes was possible and this demonstrated a very good overall concordance, with 62% of specimens demonstrating full genotype concordance and 26% demonstrating partial concordance. In cases of partial concordance, fewer genotypes were detected on Pap tests compared with PreservCyt medium. It may be possible to achieve better concordance using PCR directed toward a shorter L1 region; however, this would not allow direct comparison with the methodology used for PreservCyt samples previously extracted and tested by LA. Furthermore, the observed differences may be attributed to sampling differences or temporal degradation of lower copy number genotypes. The high correlation noted between the detection of HPV types 16 and 18 is an important finding, because reliably demonstrating the presence or absence of these genotypes in tests is particularly relevant for patients who may have previously received treatment or vaccination. In this patient population, excellent concordance between HR HPV genotypes and underlying abnormality was achieved.

The current study evaluated the recoverability of HPV DNA from conventional Pap tests; however, similar concordance could be observed with archival liquid-based Pap tests, although this requires further validation prior to implementation. Thus, the results of the current study demonstrate excellent validation of the accuracy of HPV genotypes from cells recovered from archival Pap tests when compared with HPV genotypes present in PreservCyt samples. The current study was evaluated in a selected sample of women with high-grade cytological abnormalities and normal Pap test findings. Similar studies using samples that are more representative of cytological abnormalities observed in screening populations may be required to further validate this method.

Acknowledgements

We thank the women who participated in this study.

CONFLICT OF INTEREST DISCLOSURES

Funding for this study was obtained from a National Health and Medical Research Council project grant (509002).