Interlaboratory agreement of different human papillomavirus DNA detection and typing assays in cervical scrapes

More than 90% of high-grade cervical intraepithelial neoplasia (CIN 2/3) and cervical cancers are associated with high-risk (HR) human papillomavirus (HPV) types. HPV tests applicable for population screening have to detect all HR HPV types in a simple manner. It is likely that HPV testing will augment cytological investigations in screening programs (Cuzick et al., 1995; Meijer et al., 1992; Reid et al., 1991; Schneider et al., 1996). In addition, women with the cytologic diagnosis of atypical squamous cells of undetermined significance (ASCUS) can be triaged on the basis of HPV positivity (Cox et al., 1995; Hatch et al., 1995). Persistent HR HPV infections are a marker for progressive CIN disease (Remmink et al., 1995), indicating the importance of HPV detection methods. In view of such broad applications, it is essential that HPV detection assays are standardized with respect to sensitivity and specificity. We designed a study to assess inter-method variations according to HR HPV detection and HR HPV typing in cervical scrapes. Three different polymerase chain reaction (PCR)-based methods using 2 different PCR primer pairs to generate either a 150-bp or a 450-bp PCR fragment were used in 5 laboratories. Cervical scrapes were obtained from women attending the outpatient clinics of the University Hospital Vrije Universiteit (Amsterdam, The Netherlands). The cervical scrapes were collected in 5 ml phosphate-buffered saline (PBS), centrifuged and the cell pellets were resuspended in 1 ml 10 mM Tris-HCl, pH 7.5, and frozen at −70°C. One hundred microliters were taken, boiled for 10 min at 100°C, cooled on ice and centrifuged for 1 min at 3,000 g. Ten microliters of these crude cell suspensions were used for PCR. In the reference laboratory (laboratory 1) crude cell suspensions were analyzed by general primers GP5+ and bioGP6+ PCR (de Roda Husman et al., 1995b; Jacobs et al., 1997). Fifteen HPV-positive and 5 HPV-negative cervical scrapes were encoded and distributed to the 4 participating laboratories (laboratories 2–5) for HPV testing. All 20 cervical scrapes were positive by β-globin PCR according to de Roda Husman et al. (1995a) and thus contained DNA of at least 509 bp that could be amplified. In addition, laboratories 2–5 received 10 μl of 5 aliquots representing a range of 10-fold DNA dilutions of the cervical carcinoma cell line SiHa (10 ng/μl, 1 ng/μl, 100 pg/μl, 10 pg/μl and 1 pg/μl). Laboratories 1 and 2: For HPV analysis, the general primers GP5+ and bioGP6+, which span a region of 140–150 bp from the L1 open reading frame of a broad spectrum of HPV genotypes, were used in general primer (GP)-PCR (de Roda Husman et al., 1995b), with the exception that the GP6+ primer was biotinylated (bioGP6+). The GP-PCR amplicons were analyzed by an enzyme immunoassay (EIA) with 14 different HPV type-specific oligonucleotides to detect HR HPV types individually (Jacobs et al., 1997). Laboratory 3: General primers GP5+/GP6+ were used and PCR amplicons were separated electrophoretically. Positive samples were further analyzed by a type specific (TS)-PCR. The TS-PCR was performed using combinations of HPV 6/11, 16, and 18 specific cloning site-flanking primers as described by van den Brule et al. (1990). The PCR amplicons of samples positive by GP-PCR and negative by TS-PCR were sequenced for HPV typing. Laboratories 4 and 5: The presence of HPV DNA was analyzed by PCR using L1 degenerate primer MY09/MY11 for a broad spectrum of mucosal HPV types. For HPV typing, PCR fragments were analyzed by a restriction fragment length polymorphism (RFLP) and hybridization analysis (Meyer et al., 1995). Hybridization signals were detected using a chemiluminescence-based system (ECL, Amersham, Ayelsbury, UK). All samples were tested in a blinded manner. Of 20 patients, 5 cervical scrapes were HPV negative, 6 contained 1 HPV type, 8 contained 2 and 1 contained 3 different HPV types. All 5 HPV negative scrapes were found to be HPV negative by all laboratories. Thus, false-positive HPV results due to contamination did not occur. Laboratories 1 and 2 detected 4 dilutions of SiHa DNA (100 ng to 100 pg), laboratories 3 and 5 detected 3 dilutions (100 ng to 1 ng SiHa DNA). For laboratory 4, data concerning this dilution series were not available. The different sensitivity for HPV 16 DNA detection in SiHa cells may be explained by the higher sensitivity of the EIA used in laboratories 1 and 2 and appears not to be due to the different primer pairs GP5+/GP6+ and MY09/MY11. The overall agreement rate for the detection of HR HPVs was between 75% and 100% (κ values 0.50 to 1.0) (Table I). Lower agreement rates, between 54% and 92%, were found for HPV typing, especially in cervical scrapes with multiple infections (Table I). Direct sequencing used by laboratory 3 appears to detect only the overrepresented HPV type in cervical scrapes, because not more than 1 HPV type was detected in 9 multiple infections. By the MY09/MY11 PCR RFLP detection method in 2 cervical scrapes (samples 13 and 29, see Table I) a different HPV type was detected in laboratories 4 and 5. This finding may be due to different specificity of both primer pairs or more probably false interpretation of the analyzed PCR fragment by RFLP for HPV typing. Excellent agreement rates (90% to 100%) for HR HPV detection were found between laboratories 1, 2 and 3 using the GP5+/GP6+ primer pair, and excellent to good agreement rates were obtained with the MY09/MY11 primers (laboratories 4 and 5) with reference laboratory 1 (overall agreement rates of 95% and 75%). For HPV typing, the results varied more strongly (laboratories 3–5), independently of the 2 primer combinations used. All these methods amplify a fragment in the same region of L1. Because these fragments also represent the target for subsequent typing it is likely that irregularities in this step are responsible for the discordant results. Overall, good agreement rates for HR HPV detection were found irrespective of the protocol used and excellent agreement rates were found by consensus L1 primers in laboratories using standardized methods. The slight differences of the HR HPV results may have been due to interpretation of the weakly positive signals, the L1 primers used (consensus or degenerate), or the failure to amplify very small amounts of HPV DNA. The observed differences for HPV typing imply that quality controls need to be improved to obtain excellent final agreement between laboratories. Yours sincerely, Ingo Nindl [email protected]*, Marcel Jacobs , Jan M.M. Walboomers , Chris J.L.M. Meijer , Herbert Pfister , Ulrike Wieland , Thomas Meyer?, Eggert Stockfleth?, Ruediger Klaes**, Magnus von Knebel Doeberitz**, Achim Schneider*, Matthias Duerst*