Reproducibility of p16INK4a Biomarker Levels, as Measured by ELISA, Among HIV-Positive Women in Western Kenya With Normal Cervical Exams During a 12-Month Follow-Up

42 Background: Cervical cancer disproportionately affects women in low- and middle-income countries (LMICs), and cytology-based programs are not feasible for use in these settings. This has prompted a search for alternative cervical-cancer screening or surveillance methods that can be adopted in LMICs to address the high disease burden. In this study, we seek to evaluate the performance and reliability of cervical p16INK4a biomarker as measured by an ELISA assay among a group of HIV-positive women in a low-resource setting who had normal screening results by gold standard at baseline and at 12-month follow-up.Methods: This prospective study took place at the Family AIDS Care and Education Services (FACES) clinic in Kisumu, Kenya. Participants underwent cervical cancer screening using VIA, colposcopy and collection of cervical HPV p16INK4a samples. Women with negative colposcopies were rescreened at 12-months, and cervical p16INK4a samples were collected. Among women with negative colposcopies at both baseline and follow-up, we investigated the reproducibility of p16INK4a biomarker levels within the same woman between the two visits. Biomarker levels were determined using an ELISA-based biochemical assay. We compared median p16INK4a levels between baseline and follow-up using the paired t-test. We also examined correlation using correlation coefficients and a Bland-Atman plot of differences versus average. A multivariate regression model was fit to identify demographic and clinical variables associated with absolute change in p16INK4a levels.Results: Among the ninety-three women who had normal cervical exams, mean p16INK4a levels increased significantly between baseline and follow-up period, at 20.2 U/ml vs 30.1 U/ml (p<0.01). The correlation coefficient between the values at the two time points was 0.61 (p<0.01), indicating a moderate but not perfect agreement between baseline and follow-up biomarker levels. A Bland-Altman plot of difference versus average of the two measures showed a mean difference of 10 units/ml, indicating poor agreement between the two measurements. In a multivariate regression model including age at screening, HPV status, hormonal contraception use, use of ART, VIA results, and CD4 count at baseline and follow-up, only age at screening and HPV status were significantly associated with greater absolute change in p16INK4a measurements (p< 0.01).Conclusion: Our results suggest that there is variability in levels of p16INK4a biomarker as measured by ELISA in HIV-positive women in low-resource settings with normal screening as determined by colposcopy over a 12-month follow-up period. Biomarker levels varied significantly more among older women and those who were HPV positive at baseline, despite normal cervical exams. These data suggest that elevations in p16INK4a biomarker levels may not be a reliable marker of dysplasia in this group of women. Further research in this population as well as replication of these results will be important.AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST: Chemtai Mungo No relationship to disclose Carol Conell No relationship to disclose May Maloba No relationship to disclose Elizabeth Bukusi No relationship to disclose Craig Cohen Consulting or Advisory Role: Symbiomix Inc. Megan Huchko No relationship to disclose