Predictive value of human papillomavirus type for histological diagnosis of women with cervical cytological abnormalitiesBMJ 1995; 310 doi: https://doi.org/10.1136/bmj.310.6972.94 (Published 14 January 1995) Cite this as: BMJ 1995;310:94
- M P M Burger, gynaecologista,
- H Hollema, pathologista,
- W J L M Pieters, pathologistb,
- W G V Quint, molecular biologistc
- a University Hospital, 9713 EZ Groningen, Netherlands
- b Laboratory of Pathology SSZOG, 9675 HJ Winschoten, Netherlands
- c Diagnostic Centre SSDZ, 2625 AD Delft, Netherlands
- Correspondence and requests for reprints to: Dr Matthe P M Burger, Department of Obstetrics and Gynaecology, Section Oncological Gynaecology, University Hospital, Oostersingel 59, NL-9713 EZ Groningen, Netherlands.
- Accepted 9 November 1994
In the past 10 years ablative treatment modalities for cervicalintraepithelial neoplasia have been replaced by electrosurgical loop excision of the transformation zone in many centres. Gynaecologists who master the electrosurgical technique are inclined to treat without biopsy. Concern has been expressed that this leads to overtreatment, as cytological abnormalities may result from benign reactive changes.
The broad spectrum of different human papillomavirus types has recently been grouped on the basis of a phylogenetic or evolutionary tree. The main branches of this tree lead to specific groups of human papillomavirus types with similar tissue tropism and oncogenic potential. One of the main branches comprises type 16 and related viruses such as types 31 and 33. Human papillomavirus type 18 belongs to a separate main branch.1
We investigated whether analysis of human papillomavirus in cervical scrapes might help in the better selection of patients for loop electrosection at the first visit.
From 1 September 1988 to 1 September 1993 we surveyed patients who had either one cervical smear showing severe dyskaryosis (n=163) or two cervical smears showing mild or moderate dyskaryosis (n=157). The interval between two abnormal smears was a maximum of one year. The mean age of the patients was 34.8 (SD 8.0) years. Cervical scrapes were analysed for human papillomavirus by a general primer mediated polymerase chain reaction. Positive samples were examined for types 6, 11, 16, 18, 31, and 33 separately by means of type specific primers.2 If none of these types could be detected the type remained unknown. Four weeks after the cervix had been scraped we took representative colposcopically directed biopsy samples of atypical epithelium. If cervical intraepithelial neoplasia of any grade was diagnosed we performed loop electrosection or cold knife conisation. Cervical neoplasia was classified according to the most severe grade found histologically.
In the group of 163 women with a severely dyskaryotic smear the prior probability of cervical intraepithelial neoplasia grade II or worse was151/163 (92.6%). For our purpose this high prior probability signified that analysis of human papillomavirus was less useful in this patient category.
Results, comment, and conclusion
The table shows the human papillomavirus types detected in relation to the histological diagnoses in the 157 patients with two mildly or moderately dyskaryotic cervical smears. In this group the prior probability of cervical intraepithelial neoplasia grade II or worse was 91/157 (58.0%). When human papillomavirus type 16 was detected the (posterior) probability of cervical intraepithelial neoplasia grade II or worse was 36/45 (80.0%). Other investigators reported a similar finding.3 When infections with viruses of the type 16 related phylogenetic branch were considered together, with the exclusion of mixed infections with types 6/11 or 18, the probability of cervical intraepithelial neoplasia grade II or worse was 51/62 (82.3%). Three of the 12 human papillomavirus positive patients without neoplasia in biopsy samples probably had a false negative diagnosis because their cervical smears remained abnormal. They were subsequently diagnosed with cervical intraepithelial neoplasia grade II or III. These three patients harboured human papillomavirus type 16 or 33. When we corrected the calculations for these apparent misdiagnoses the probability of cervical intraepithelial neoplasia grade II or worse was estimated to be 54/62 (87.1%) if human papillomavirus type 16, 31, or 33 was detected.
When human papillomavirus type 18 was detected3 the probability of cervical intraepithelial neoplasia grade II or worse was 4/11 (36.4%). Cervical intraepithelial neoplasia grade II or worse was found significantly less often in human papillomavirus type 18 positive women than in type 16 positive women (4/11 versus 36/45; P=0.008 (Fisher's exact test)). Other investigators reported that human papillomavirus type 18 was equally distributed over low grade and high grade intraepithelial neoplastic lesions.4 In a study on the morphology of lesions and observer variability type 18 associated lesions were preferentially classified as low grade.5
We conclude that our findings endorse the appropriateness of the phylogenetic classification. In women with two mildly or moderately dyskaryotic cervical smears the detection of human papillomavirus type 16, 31, or 33 (but not type 18) will help to reduce overtreatment if it is hospital policy to perform loop electrosection of the transformation zone without biopsy.
This study was supported by grant OG91–053 from the Fund forInvestigative Medicine of the National Health Insurance Council.