HPV testing for clarifying borderline cervical smear results

Two large studies in the United States have provided strong evidence that testing for human papillomavirus DNA is a useful tool for managing women with borderline results on cervical smear tests.¹,² In response to those results and subsequent cost analyses, integrated managed care organisations, including Kaiser Permanente of California, have recently implemented human papillomavirus testing for managing women with atypical squamous cells of undetermined significance (ASCUS) on cervical smear tests. In the past year many health insurance plans have done the same. Consequently, a large majority of insured women in the United States now have cover for human papillomavirus testing in response to atypical squamous cells of undetermined significance. Large cytology laboratories, providing over half of US cervical smear results, encourage reflexive human papillomavirus testing for such cases. Despite this, a study by Rebello et al in this issue urges “caution in the clinical use of testing for human papillomavirus testing” (p 893).³ How do we interpret these somewhat conflicting messages about such an important issue in cervical cancer prevention?

Methodological differences may account for some of the inconsistencies between study results. The recent US reports (the Kaiser Permanente Borderline Pap Study¹ and baseline data from the ASCUS-LSIL triage study or ALTS²) have involved women with a report of atypical squamous cells of undetermined significance on cervical screening. However, Rebello et al studied women with “persistent mild dyskaryosis or borderline nuclear change,” which may represent persistent atypical squamous cells of undetermined significance or low grade squamous intraepithelial lesions (LSIL). Rebello et al's reported prevalence of human papillomavirus of 61% is not unexpected for a mixture of atypical squamous cells of undetermined significance and low grade squamous intraepithelial lesions, based on results from previous studies with the Hybrid Capture (HC II) test for high risk human papillomavirus types (Digene, Inc).¹,²,⁴

However, Rebello et al's 35% prevalence of histologically defined high grade cervical intraepithelial neoplasia is much higher than the 7-16% reported previously for women with cytological findings of atypical squamous cells of undetermined significance or a low grade squamous intraepithelial lesion.¹,²,⁵ As suggested by Rebello et al, the requirement for persistent cytological abnormalities in their study population may account for higher disease prevalence. However, the disease prevalence may also be inflated by two additional differences in study methods.

Firstly, the Kaiser Permanente and the ALTS investigations obtained final diagnoses by colposcopically directed biopsy or endocervical curettage, while Rebello et al performed large loop excision of the transformation zone on all patients. Perhaps lesions that require a large loop specimen for detection contribute to the additional disease prevalence in the current study. Secondly, the Kaiser Permanente and the ALTS investigators defined high grade cervical intraepithelial neoplasia by consensus review by expert pathologists, while Rebello et al used a single pathologist to make the diagnosis. The issue at stake in each of these investigations is the diagnostic performance of the HC II human papillomavirus test in identifying women with histologically defined high grade cervical intraepithelial neoplasia. Thus, the accurate classification of this histological endpoint is paramount to defining test sensitivity, disease prevalence, and predictive value.

The Kaiser Permanente study of cases of atypical squamous cells of undetermined significance showed that human papillomavirus testing identified 89.2% of women (mean age 38 years) with underlying high grade disease; and in women aged under 30 the sensitivity and negative predictive value were 100%. Similarly in the ALTS trial the HC II human papillomavirus test in cases of atypical squamous cells of undetermined significance (mean age 29 years) was 95.9% sensitive in identifying women with underlying high grade cervical intraepithelial neoplasia, and the predictive value of a negative test was 98.9%. In the women studied by Rebello et al (mean age 32 years), HC II human papillomavirus testing had a sensitivity of 93% for high grade cervical intraepithelial neoplasia. The authors emphasise the modest negative predictive value (89%) in women aged under 30. This low negative predictive value, and the high disease prevalence, are inconsistent not only with the ALTS and Kaiser Permanente reports but also with previous human papillomavirus triage studies.⁶,⁷

We should be intrigued by high grade cervical intraepithelial lesions that are apparently not associated with high risk human papillomavirus. They may be transient lesions associated with low risk human papillomavirus types, and thus not issues in terms of cancer risk. In addition, these lesions may represent likely candidates for reclassification on pathology review. Perhaps they are more accurately described as mild to moderate cervical intraepithelial neoplasia, as opposed to high grade lesions.

Alternatively, these may be bona fide high grade lesions with potential to progress. The cases were women with persistent mild abnormalities on cervical smear tests. This criterion may have selected women with lesions that were anatomically difficult to sample or identify, whether by cytology, human papillomavirus testing, or colposcopy. If that is the case, the study's implications may be more a treatise on the failings of colposcopy and biopsy than on human papillomavirus testing to identify occult lesions.

It would be informative to investigate further the specimens collected by Rebello et al, perhaps restricting the analysis to women who had atypical squamous cells of undetermined significance. Consensus review of the histology and subsequent recalculation of the HC II diagnostic performance would be useful. Comprehensive (including low risk types) human papillomavirus testing of confirmed high grade cases of cervical intraepithelial neoplasia that were negative for high risk human papillomavirus might solve another piece of the mystery. (Testing could be conducted on the lesion tissues if the original human papillomavirus specimens are not informative.) This venture would complement the forthcoming prospective data from the ALTS trial,⁸which will address whether dangerous precancerous lesions are lurking in women who were considered at low risk on the basis of a negative HC II human papillomavirus test.

Papers p 893