Cervical screening

The introduction of human papillomavirus (HPV) testing has the potential to improve the accuracy of cervical cancer screening, but it has led to a multiplicity of possible protocols. So the decision to establish guidelines based on risk thresholds (benchmarks) to allow equal management for equal risk, irrespective of the screening test, is highly appropriate [1]. In the United States, for example, the benchmark that allows for a ‘routine’ recall after 3 years is a risk of biopsy-confirmed cervical high-grade squamous intraepithelial neoplasia or worse (bHSIL+) of 0.69%, based on the 3-year risk associated with a negative cytology. The benchmarks for 6–12-month return and for immediate colposcopy are 3-year risk of 8.8 and 14.4%, respectively [2]. The gold-standard approach to prevent an excess of bHSIL+ and cervical cancer in women living with HIV (WLHIV) has been controversial [3]. Despite early evidence of the effectiveness of HPV testing in screening of WLHIV [4], high HPV prevalence and the accompanying low specificity have been of great concern [5]. However, the work-up of a large proportion of HPV-positive WLHIV may be an unavoidable price to pay if, following good stratification into risk groups, levels of bHSIL+ risk within these groups can be shown to be similar to those in the general population. On the other hand, the prompt use of combined antiretroviral therapy (cART) can avoid immune impairment and make too frequent screening of WLHIV unnecessary or even deleterious. The proposal by Robbins et al.[6] in this issue of AIDS to revisit screening guidelines in WLHIV is, therefore, very timely. To compare bHSIL+ risk in WLHIV and the general population, Robbins et al.[6] focused on the management of screening results by type of primary test (cytology and HPV testing) and previous history of cytology screening using post-2000 findings from the Women's Interagency HIV Study, https://statepi.jhsph.edu/wihs/wordpress/. Analyses that included HPV testing were appropriately restricted to women aged 30–64 years, among whom cervical cancer prevention is most effective and the use of HPV testing is best validated [2]. The report by Robbins et al.[6] shows that when both cytology and HPV testing are negative or after repeatedly negative cytology, time-to-benchmarks for WLHIV are similar to those in the general population (2-year or 3-year return). Likewise, the recommendation of 6–12-month return in cytology-negative/HPV-positive WLHIV (bHSIL+ risk at 3 years ∼5%) and of immediate colposcopy in HPV-positive WLHIV with atypical squamous cells of undetermined significance (bHSIL+ risk at 3 years ∼15%) are also similar [2]. Partial genotyping for HPV16/18 is also recommended for risk stratification of HIV-uninfected HPV-positive women in the framework of cotesting with cytology [2] or in primary HPV screening [7], but no guidelines currently exist for its use in WLHIV. Although previous findings from sub-Saharan Africa [8] as well as from the Women's Interagency HIV Study [9] provided support to the utility of HPV16 genotyping to predict bHSIL+ in WLHIV, data were not deemed sufficient to draw conclusions on partial HPV genotyping [6]. In the meantime, a meta-analysis of ours [10] has suggested that the differences in HPV carcinogenicity that have supported partial genotyping algorithms in the HIV-uninfected population, namely the predominance of HPV16 and 18 worldwide, followed by HPV45 for Africa [11], also hold true in WLHIV [10]. Partial genotyping may therefore reduce referral rates and overtreatment in WLHIV who undergo regular screening in high-income countries and help prioritize diagnostic work-up of WLHIV in low-infrastructure settings, where cytology is not available. Robbins et al.[6] are also proposing for the first time to use CD4+ cell count for risk stratification in WLHIV. Although the risk of bHSIL and cervical cancer is certainly related to immunity [12,13], we think that this approach can be problematic. The most appropriate CD4+ cell count threshold is unclear. Robbins et al.[6] mainly used CD4+ cell count 500 cells/μl as a threshold showing some risk discrimination ability. However, they also used in some instances CD4+ cell count 350 cells/μl and mention that a group of WLHIV with CD4+ cell count less than 200 cells/μl had a particularly elevated bHSIL+ risk. This circumstance was rare in their study [6], but is frequent in women in whom HIV infection is detected late, that is in low/middle-income countries and in a nonnegligible fraction of women who do not suspect to be HIV-infected in high-income countries. These women with severe immunosuppression need of course to start cART immediately and may merit a particularly careful cervical work-up at that time. However, a woman with CD4+ cell count more than 500 cells/μl may also have had a much lower CD4+ cell count (nadir) before starting cART. In fact, indirect evidence points to the influence of nadir CD4+ cell count for cervical cancer risk [13,14] and suggests the nonreversibility of certain preinvasive stages in HPV carcinogenesis [13]. Reliance on CD4+ cell count at a screening visit may be insufficient and the collection of the entire CD4+ history would complicate real-life screening programs whose priority, to prevent cervical cancer, remains to reach as many WLHIV as possible in different health settings [12,15]. If, as this benchmark approach suggests, a similar risk stratification by HPV and/or cytology can work irrespective of HIV status, it opens the possibility of screening and managing WLHIV in ways that are conceptually consistent with those recommended in the general female population. This would greatly facilitate the task of caregivers and overcome situations in which WLHIV are either underscreened or overscreened. Indeed, in low/middle-income countries, WHO currently recommends the use of HPV-based screening irrespective of HIV status [16]. Conflicts of interest There are no conflicts of interest.