The comparative test performance of dot filter hybridization (Viratype) and conventional morphologic analysis to detect human papillomavirus.

To investigate the test performance of a commercially available detection kit for human papillomavirus (HPV), the relationship between the detection of HPV by dot filter hybridization (DFH) and by standard morphologic methods was studied. Four hundred two cervical samples taken from 381 patients referred to a colposcopy clinic were examined. Human papillomavirus DNA sequences were identified and typed using commercially available anti-sense RNA probes. Simultaneous cytologic smears were obtained in 289 patients, directed biopsy samples in 284, and both smears and biopsy samples in 171 samples. Human papillomavirus DNA was detected in 164 specimens (41%), of which 24 (15%) were type 6/11, 74 (45%) were type 16/18, 39 (24%) were type 31/33/35, and 27 (16%) were untyped due to the presence of multiple positive signals. Viral types 16/18 and 31/33/35 were eight and six times more frequent in cervical intraepithelial neoplasia (CIN) II/CIN III lesions than in condyloma/CIN I, respectively. When the cytologic diagnosis was considered the standard of reference, the results of DFH for the detection of HPV were concordant in 167 (56%) paired samples. The sensitivity of DFH was 48% and the specificity was 77%. The distribution of the morphologic diagnoses in the group of false-negative results and true-positive results was similar. When the histologic diagnosis was considered the standard of reference, the efficiency of DFH was 62%, the sensitivity was 59%, and the specificity was 79%. In the subgroup of 118 samples with simultaneous smear and biopsy and at least one positive examination, 42 (36%) were positive by all three methods, 42 (36%) by two, and 34 (29%) by one, including 6 (5%) by DFH alone. Fifteen cases more were detected by the complementary use of DFH and cytology than with cytology alone. The results demonstrated that the sets of patients positive for HPV when detected by DFH or by morphologic methods were not identical but rather overlapped. The detection of HPV may be slightly improved by using DFH in addition to conventional examinations. A significant number of HPV-positive patients without a morphologic lesion and patients with low-grade lesions had HPV 16/18 or 31/33/35, suggesting a possible role for typing in establishing a risk profile. However, given uncertainties in understanding the biology of HPV-associated lesions, the role, if any, of clinical testing for HPV by DFH remains to be defined.