Pilot Study of a Commercialized Human Papillomavirus (HPV) Genotyping Assay: Comparison of HPV Risk Group to Cytology and Histology

We evaluated a commercialized PCR assay, Linear Array, thatdetects 37 human papillomavirus (HPV) genotypes, using a sampleof liquid cytology specimens (n = 534). We found a strong associationof an increasing level of HPV risk (HPV type 16 [HPV16] >HPV18 > other carcinogenic types > noncarcinogenic types> negative specimens) with increasing severities of cytologicinterpretations (P_Trend < 0.0005) and histologic diagnoses(P_Trend < 0.0005).

INTRODUCTION

Top

Introduction

Carcinogenic human papillomavirus (HPV) testing has now beenapproved in the United States as an adjunct to cytology fortriage of equivocal cytology at all ages and for general screeningfor women of $≥$ 30 years old (15). Several studies have now shownthat detection of specific carcinogenic HPV types, especiallyHPV type 16 (HPV16) and HPV18, may be useful in differentiatingcarcinogenic HPV-positive women at greater and lower risk ofhaving or developing precancer, cervical intraepithelial neoplasiagrade 3 (CIN3), or cancer (CIN3+) (1, 2, 8). Identifying womenwith persistent carcinogenic HPV infection may also be clinicallyuseful (9). Together, these data support a role for HPV genotypingin cervical cancer screening.

Commercial HPV genotyping assays are currently under development.We evaluated one assay, Linear Array (LA; Roche Molecular Systems,Alameda, CA), a PCR-based genotyping assay that detects 37 HPVgenotypes which is a commercialized version of the line blotassay (13).

MATERIALS AND METHODS

Top

Materials and Methods

Cervical specimens and data. We acquired blinded residual PreservCyt specimens, after cytologicinterpretation had been rendered, from 125 women with normalcytology, 125 women with atypical squamous cell (ASC) cytology,125 women with low-grade squamous intraepithelial lesion (LSIL)cytology, and 165 women with high-grade squamous intraepitheliallesion (HSIL) cytology or worse ( $≥$ HSIL) from the Medical Universityof South Carolina. The institutional review board (IRB) of theMedical University of South Carolina approved the study, andthe use of these specimens was deemed exempt from review bythe NCI IRB. We subsequently excluded 12 specimens called $≥$ HSILbecause of conditions unrelated to cervical abnormalities (e.g.,endometrial carcinoma histopathology).

In addition to the original histologic diagnosis, each caseunderwent a pathology review to ascertain the histologic diagnosis.Importantly, six cases originally called CIN2-3 and one casecalled CIN3 were reclassified as CIN2, and eight cases originallycalled CIN2-3 and one case called CIN2 were reclassified asCIN3. Also, two cases called CIN2 and two cases called CIN1-2were reclassified as CIN1. We were not able to retrieve histologyresults for four women.

To supplement the number of specimens from women with severedisease, the University of Arizona supplied 12 blinded specimensfrom women attending colposcopy, 10 of whom were diagnosed withCIN3 (including carcinoma in situ) and 2 of whom were diagnosedwith cancer. Specimens were collected under an IRB-approvedprotocol, and their use was deemed exempt from review by theNCI IRB.

HPV testing. Aliquots were tested using the commercially available LA HPVgenotyping test according to the manufacturer's instructions.Briefly, DNAs were extracted from clinical specimen aliquotsby using a QIAamp MinElute Media kit (QIAGEN, Inc., Valencia,CA), and target DNAs were amplified by PCR. LA utilizes thePGMY09/11 L1 consensus primer system and includes coamplificationof a human cellular target, ß-globin (6), as an internalcontrol. Detection and HPV genotyping are achieved by usinga reverse line blot method (7, 12), and the test includes probesto genotype 37 anogenital HPV types (6, 11, 16, 18, 26, 31,33, 35, 39, 40, 42, 45, 51 to 59, 61, 62, 64, 66 to 73, 81,82 subtype IS39, 82 subtype W13b, 83, 84, and 89). The onlydeviation from the LA product insert protocol was to implementautomated sample preparation for extraction of up to 96 specimensat a time on a QIAGEN MDx platform (using a MinElute Media MDxkit according to the manufacturer's instructions) rather thanprocessing 24 specimens per batch by the manual vacuum method.Results for six specimens were excluded because of a failureto amplify the ß-globin internal control, indicatingpoor DNA recovery and an invalid sample.

Some women (n = 86) had also been tested previously by HybridCapture 2 (HC2; Digene Corporation, Gaithersburg, MD), a DNAtest that targets 13 carcinogenic HPV types (HPV16, -18, -31,-33, -35, -39, -45, -51, -52, -56, -58, -59, and -68). Thisassay has been demonstrated to cross-react with HPV66 (3, 13),an HPV type that was recently recognized as carcinogenic (5).These 14 types were considered the carcinogenic HPV types forthis study.

Statistics. HPV testing results were categorized hierarchically accordingto cancer risk (HPV risk category) (HPV16 > HPV18 > othercarcinogenic HPV types > noncarcinogenic HPV types > PCR-negativesamples). We tested the association of HPV risk groups withthe severities of cytologic interpretation (normal < ASC< LSIL < $≥$ HSIL) (n = 522) and histologic diagnosis (normal[which included women with no biopsy taken based on a colposcopicimpression of normality] < CIN1 < CIN2 < CIN3 <cancer) (n = 530), using a Pearson ${chi}$ ² test and the Mantel-Haenzelextension test for trend. We compared the agreement for detectionof carcinogenic HPV by LA and HC2 by calculating kappa statisticsand crude agreement with 95% confidence intervals (95% CI) andtested for statistical differences in test positivity usingMcNemar's ${chi}$ ² test.

RESULTS

Top

Results

We found a strong trend of increasing severity of cytologicinterpretation with increasing likelihoods of testing positiveby LA for any HPV type (P_Trend < 0.0005) and of testing positivefor a carcinogenic HPV type (P_Trend < 0.0005) (Table 1).Increasing severity of cytologic interpretation was associatedwith more-carcinogenic HPV risk categories (P_Trend < 0.0005).Among women with ASC, 38% tested negative, 11% tested positivefor noncarcinogenic HPV types, 36% tested positive for othercarcinogenic HPV types, 2% tested positive for HPV18, and 13%tested positive for HPV16. In comparison, among women with $≥$ HSIL,6% tested negative, 5% tested positive for noncarcinogenic HPVtypes, 34% tested positive for other carcinogenic HPV types,8% tested positive for HPV18, and 48% tested positive for HPV16.

View this table:
[in this window]
[in a new window]
TABLE 1. Relationship of HPV risk group and severity of cytologic interpretation

Likewise, we found a strong trend of increasing severity ofhistologic diagnosis with increasing likelihoods of testingpositive for any HPV type (P_Trend < 0.0005) and of testingpositive for a carcinogenic HPV type by LA (P_Trend < 0.0005)(Table 2). Increasing severity of histologic diagnosis was associatedwith more-carcinogenic HPV risk categories (P_Trend < 0.0005).Among women with CIN1, 7% tested negative, 20% tested positivefor noncarcinogenic HPV types, 45% tested positive for othercarcinogenic HPV types, 3% tested positive for HPV18, and 25%tested positive for HPV16. In comparison, among women with CIN3,4% tested negative, 0% tested positive for noncarcinogenic HPVtypes, 23% tested positive for other carcinogenic HPV types,9% tested positive for HPV18, and 64% tested positive for HPV16.

View this table:
[in this window]
[in a new window]
TABLE 2. Relationship of HPV risk group and severity of histologic diagnosis

For a subset of 86 women, we had test results for HC2 to comparewith the LA results. The kappa value between Hybrid Capture2 and LA for detection of carcinogenic HPV was 0.74 (95% CI,0.60 to 0.89), and the percent agreement was 88% (95% CI, 80%to 94%). There were five women who tested positive by HC2 andnegative by LA for carcinogenic HPV and an equal number of HC2-negative,LA-positive results, and thus there was no statistical differencein the number of positive tests (P = 1.0). We also comparedthe detection of individual HPV types by LA and the LA resultsfor HPV risk groups to the HC2 results (Table 3). A high percentageof infections by any HPV type detected by LA tested positiveby HC2, which targets only 13 carcinogenic types but is cross-reactivewith another newly designated carcinogenic HPV type (HPV66)(3, 5) and some noncarcinogenic (untargeted) HPV types (3).However, when we categorized HPV infections detected by LA accordingto cancer risk, we found that most of the untargeted types testedpositive because there was a coinfection with one or more targeted(carcinogenic) HPV types. Among the 10 single-type HPV infectionsby untargeted HPV types (excluding HPV66), only 3 tested positiveby HC2. In those three cases, LA detected HPV42, HPV53, andHPV67, the last two of which have been reported previously aspotential cross-reactive types (3). By comparison, 19 of 26(73%) women with single-type HPV infections by HPV66, as detectedby LA, tested positive by HC2.

View this table:
[in this window]
[in a new window]
TABLE 3. Comparison of HPV type and HPV risk group to Hybrid Capture 2 (HC2) test results for 86 women

DISCUSSION

Top

Discussion

We found that genotyping using LA demonstrated the expectedassociations of HPV risk categories with severities of cytologicinterpretation and histologic diagnosis. A 12% prevalence ofcarcinogenic HPV in women is fairly typical compared to a worldwideseries (4) and is consistent with that for other U.S. populations(14). Virtually all cases of CIN3 and all five cases of cancerwere carcinogenic HPV positive, with roughly 60% of all casesof CIN3+ testing positive for HPV16, as expected (11). We alsonoted an absence of HPV18 in CIN3, but two of five (40%) cancerswere HPV18 positive, a pattern that has been observed previously(10). The identification of women with carcinogenic HPV wasin good agreement with the results of HC2, a pooled-probe testfor carcinogenic HPV types.

In this study, we implemented an automated sample prep optionthat increased the batch size from 24 to 96 specimens per day.Further automation, such as strip detection and interpretation,is in development and may increase the efficiency of HPV genotypingfor clinical labs. LA is currently the only validated, commerciallyavailable HPV test that is manufactured under good manufacturingpractices so that the reagents are standardized. LA is currentlycommercially available in the United States for research useonly and is a CE mark in vitro diagnostic in Europe, but itis currently in clinical trials for FDA approval as an in vitrodiagnostic.

We conclude that LA may be useful for HPV genotyping. Here wedemonstrated the expected correlation of HPV risk groups withthe severity of cervical lesions. Compared to tests that detecta pool of carcinogenic HPV types but do not distinguish whichtype is present, HPV genotyping will likely be better for monitoringpersistent carcinogenic HPV infection, which is a prerequisitefor progression to cervical precancer.

FOOTNOTES

* Corresponding author. Mailing address: Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., Room 5123, EPS MSC 7234, Bethesda, MD 20892-7234. Phone: (301) 435-3976. Fax: (301) 402-0916. E-mail: castlep{at}mail.nih.gov.

Published ahead of print on 13 September 2006.

REFERENCES

Top