From July 15th, 2009 to January 10th, 2011, 1,252 women underwent CC screening at the Department of Gynecology and Obstetrics of China-Japan Friendship Hospital (Beijing, China). Written informed consent was obtained from all the patients. The study was approved by the ethics committee of China-Japan Friendship Hospital. The patient mean age was 40.8±9.5 years (range, 19–71); 90% of patients were aged ≥30 years and 86% of them were 30–65 years. Specimens for the liquid-based cytology test (LCT), HC II test and real-time fluorescent PCR assay were collected from the women during their visit. Colposcopy was performed when ≥1 of the results of these three tests was abnormal, and a histological sample was taken when a lesion was identified. Women exhibiting acute genital inflammation were excluded from this study.

The women were asked to avoid vaginal douching, vaginal drug administration 3 days prior to their visit to the hospital as well as sexual activity 24 h prior to their visit. During the collection of cervical cells, we firstly cleaned and polished the cervix with cotton pledgets. Secondly, cervical samples were obtained with cervix brushes rotated for three times on any days of the menstrual cycle, except those during menstrual bleeding. Thirdly, we removed samples from the brushes and placed them into sample tubes containing preserving fluid and marked the name of the patient as well as the date. Specimens were preserved for <24 h at 2–8°C or for 30 days at −20°C. Repeated specimen thawing was avoided.

Smears were stained using the AutoCyte Prep staining machine and observed by professional gynecology-pathology doctors with cytological diagnosis aptitude. Cytologic diagnosis was performed according to the Cervical Cytology Criteria of the 2001 Bethesda System of the International Cancer Society. According to this system, abnormalities of cervical squamous epithelium include atypical squamous cells (ASC), low- (LSIL) and high-grade squamous intraepithelial lesions (HSIL), as well as squamous cell carcinomas (SCC).

HR HPV DNA testing was performed using the automated HC II test (Qiagen, Gaithersburg, MD, USA). HC II test is a sandwich capture molecular hybridization assay that utilizes chemiluminescent detection to provide a semi-quantitative result. The test was calibrated to detect ∼5,000 genome/ml equivalents of target HPV, represented by a Relative Light Unit (RLU) measurement greater than or equal to the cut-off value calculated in each run by a series of standards. A measurement less than the cut-off value was scored as negative. The samples were analyzed for the presence of HR HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68. Three positive and 3 negative controls (provided by the manufacturer) were included in each run.

This assay included: i) DNA extraction, involving the release of HPV and cellular DNA by lysing cervical specimens under denaturing conditions at elevated temperatures in the presence of protease. The assay did not require DNA purification. ii) PCR amplification and hybridization reaction, involving the use of biotinylated primers in order to define a sequence of ∼150 bp in length within the polymorphic L1 region of the HPV genome. The primers, pooled in the same PCR Master mix, were designed to amplify viral DNA from the same 13 HR HPV types included in the HC II test. Capture probes representing regions internal to the amplified sequences were used to identify HPV of human DNA. PCR amplification and hybridization reaction were performed using the Roche LightCycler^®. Thirty cytological samples were detected simultaneously, and each time negative and positive controls were established. Following the reaction, Roche LightCycler^® computer was used to calculate the cycle threshold (CT) of each sample. CT was the cycle number of each reaction flourescence signal that reached the setting threshold. iii) Result interpretation involved the baseline and threshold exceeding culmination of the amplication curve of the negative control, while the CT value did not indicate detection (blank). Evaluation of response effectiveness depended on the blank negative control and a CT of ≤36 in the positive control. When the CT value of the samples was ≤40, samples were evaluated as positive, while samples were considered negative with a CT >40. According to the manufacturer’s instructions, this test was able to detect the genotypes of the 13 HR HPV types at 5,000 copies/ml.

Colposcopic examination of the cervix was performed in women with either HR HPV-positive results or cytological abnormalities, or both. Negative cytological results and the two HR HPV DNA tests used predicted a low risk of cervical neoplasia. A histological sample was obtained when a lesion was identified, and 4 histological samples were obtained at 3°, 6°, 9° and 12° of the normal transformation zone. Endocervical canal curettage was performed when colposcopic examination was unsatisfactory. The patients with high-grade lesions underwent loop electrosurgical excision procedure (LEEP), cold knife conization (CKC) or hysterectomy for further diagnosis and treatment. The most severe diagnosis was determined as the final histopathological diagnosis. The diagnostic criteria of high-grade lesions were defined as equal to or more severe than CINII (CINII+).

Sensitivity, specificity and accuracy of the real-time fluorescent PCR assay, the HC II test and the LCT were determined against the presumed HPV status based on the combination of the two HPV tests used and histological pathological results. The 95% confidence intervals (CI) were calculated using Fisher’s exact test. Agreement between the two HPV tests used was assessed by Cohen’s κ, with values of 0.00–0.40 indicating poor, 0.40–0.75 fair and 0.75–1.00 excellent agreement. The χ² test was performed for group comparisons. All the tests were two-sided, and P<0.05 was considered to indicate a statistically significant difference.

The comparison between HC II test and real-time fluorescent PCR assay of 1,252 specimens is shown in Table I. Among the 1,252 women studied, the two tests gave similar results for 1,155 samples, with an overall level of agreement of 92.25% (Cohen’s κ=0.814) indicating excellent agreement. However, the examination of 97 samples using the two tests gave discordant results. HPV genotyping of these samples are to be performed in future studies. Assuming that samples with positive results examined using the two tests were infected by HR HPV, and that samples with negative results were HR HPV-negative or infected by LR HPV, ‘conditional’ sensitivity and specificity can be evaluated for the two tests used.

There were 477 patients who underwent colposcopic examination and biopsy. Histopathological results were considered to be the ‘gold standard’. Sherman et al (10) found that the cumulative incidence rate was only 0.79%, when CIN or CC and negative HPV cases were not detected using cervical cytological examination, while CIN and cancer cases were not identified in these histopathology samples. On these grounds, women with normal LCT results and negative results from the two HR HPV DNA tests used were classified into the ‘negative’ group (women without CIN or CC). As shown in Fig. 1, 1,015 specimens were histopathologically negative (normal or inflammation). Among these samples, 182 (17.93%) were positive using the HC II test, while 162 (15.96%) were positive using the real-time fluorescent PCR assay. Additionally, 56 samples were diagnosed as CINI using histopathological examination of which 27 (48.21%) were positive using the HC II test and 29 (51.79%) were positive using the real-time fluorescent PCR assay. Fifty-five samples were diagnosed as CINII using histopathological examination of which 50 (90.90%) were positive using the HC II test, while 49 (89.09%) were positive using the real-time fluorescent PCR assay. A further 104 samples were diagnosed as CINIII using histopathological examination of which 100 (96.15%) were positive using the HC II test, while 99 (95.19%) were positive using the real-time fluorescent PCR assay. Furthermore, 22 samples were diagnosed as CC using histopathological examination of which 21 (95.45%) were positive using the HC II test, while 20 (90.90%) were positive using the real-time fluorescent PCR assay. We compared the negative and positive histopathological groups, where the HR HPV infection rate was 19.51 and 94.48%, respectively, when samples were examined using the HC II test. A statistically significant difference between two groups was observed (χ² = 411.57; P<0.05).

Due to the fact that CINII and CINIII patients did not undergo the CINI morphology stage, the difficulty to distinguish HPV infection from the changes of CINI at cytopathological appearance under a colposcope, and the significant proportion of samples that would undergo spontaneous regression, we classified histopathological results of CINII, CINIII and CC into the positive group in this study, and normal or inflammation and CINI samples into the negative group. Sensitivity and specificity were also calculated at this point. The correlation between the two HPV tests used and histopathological outcomes is shown in Table II.

In this study, there were 181 patients with CINII+, 94.48% of whom were positive using the HC II test and 92.82% of whom were positive using the real-time fluorescent PCR assay. Out of the 1,071 samples of the negative histopathology group, there were 862 (80.49%) negative samples using the HC II test and 880 (82.17%) negative samples using the real-time fluorescent PCR assay. Out of the 872 negative samples using the HC II test, there were 862 patients without CINII+, indicating a negative predictive value of 98.85%. Out of the 893 negative samples using the real-time fluorescent PCR assay, there were 880 patients without CINII+, indicating a negative predictive value of 98.54%.