Human papillomavirus infection of the uterine cervix. Tissue sampling and laboratory methods affect correlations between infection rates and dysplasia.

Two common tissue sampling techniques--colposcopic biopsy and cervical scrape--and two common human papillomavirus (HPV) detection techniques--Southern blot and dot blot (SB and ViraPap [VP])--were compared to determine whether differences in these techniques alter correlations between "oncogenic" HPVs and cervical neoplasia. In 87 women with persistently abnormal Papanicolaou (Pap) smears, concurrent biopsy and scrape specimens contained HPV in 21 (24%) and contained no HPV in 26 (30%); 30 scrape specimens (34.5%) tested positive when the biopsy tested negative and 10 (11.5%) scrape specimens tested negative when the biopsy tested positive (overall concordance, 54%). Concordance for the most prevalent HPVs (16/18) was 59%. In carcinoma in situ, HPV was found in biopsy samples significantly more frequently than in scrape specimens: 17 of 23 (75%) biopsy samples versus 9 of 23 (39%) scrape specimens (P = 0.018). Conversely, in mild or no dysplasia, 0 of 42 biopsy samples tested positive for HPV 16/18 compared with 12 of 42 scrape specimens (29%; P = 0.0001). Of 229 specimens analyzed by SB and VP, 43 (19%) tested positive and 148 (65%) tested negative for HPV by both methods (concordance, 84%). Corroborative results indicated that 29 of 35 (83%) VP-positive SB-negative results were truly positive compared with none of three SB-positive VP-negative results. Both the cervical sampling technique and the method for HPV detection can significantly affect statistical correlations between cervical dysplasia and HPV type.