CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

648 MR ELASTOGRAPHY DETECTS HIGHER LIVER FIBROSIS IN UNCONTROLLED HIV INFECTION

(68.2%) of cancer patients were women. For HIV-infected patients, median current CD4 cell count was 355 cells/μL (IQR 213 to 543) and 1189 (82.5%) were on ART at the time of their cancer diagnosis (median ART duration 4.8 years [IQR 1.2 to 8.8 years]). Of the 1075 with an available measurement (90.4%), 1028 (95.6%) had HIV viral load < 1000 copies/mL. Patients living with HIV were 3.3-fold more likely to develop cancer (95%CI 3.1 to 3.5) and excess risk was similar for men (SIR 3.5, 95%CI 3.2 to 3.9) and women (SIR 3.2 95%CI 3.0 to 3.4). With exception of breast and head and neck cancer, HIV was significantly associated with increased risk for all cancer types; estimated SIRs for individual cancers are shown in the Table. HIV infection accounted for an estimated 39.3% of cancer cases (population attributable fraction). Conclusion: With ART coverage exceeding 80%, more than a third of cancer cases are attributable to HIV infection in Botswana. Cancers in HIV-infected populations remain a large public health challenge and strategies to mitigate this burden are urgently needed.

Marianna K. Baum 1 , Sabrina S. Martinez 1 , Richard L. Ehman 2 , Kenneth E. Sherman 3 , Pedro J. Greer 1 , Jun Chen 2 , Robert Laird 1 , Angela Laird 1 , Leslie Seminario 1 , Angelique Johnson 1 , Alhanoof Alohaly 1 , Mukesh Mudgal 1 , Alicia Sneij 1 , Javier Tamargo 1 , Adriana Campa 1 1 Florida International University, Miami, FL, USA, 2 Mayo Clinic, Rochester, MN, USA, 3 University of Cincinnati, Cincinnati, OH, USA Background: Magnetic Resonance elastography (MRE) is a non-invasive, highly accurate method to assess hepatic fibrosis which is needed for therapeutic decisions or predicting disease outcomes. Studies to date have reported contradictory results regarding association between HIV mono- infection and liver fibrosis, possibly due to the difficulty in obtaining liver biopsies and using less accurate non-invasive indicators of liver disease in this population. The objective of this study was to examine the relationship between the extent of liver fibrosis in HIV mono-infected and non-infected adults. Methods: After consenting 176 HIV mono-infected and 113 HIV/HBV/ HCV un-infected adults from the Miami Adult Studies in HIV (MASH) cohort, demographics were collected. CD4 cell count and HIV viral load was obtained frommedical charts. Controlled HIV viral load was defined as having <400 copies/mL. MRE was conducted on a 3T Siemens MAGNETOM Prisma MRI and mean stiffness of the liver was calculated through an inversion algorithm that generates elastogram/stiffness maps of the tissue in kilopascals. Statistical analyses were completed using frequencies, Wilcoxon Two-Sample test and regression models. Results: The median age was 54 years (IQR: 49-59), 54.55%were male and 67.16%were African American. In those with HIV, 11% had detectable HIV viral load. Within the group infected with HIV, those who had undetectable HIV viral load had lower mean liver stiffness compared to those with detectable HIV viral load [2.266 kPa (IQR=2.480-1.970) vs. 2.237 kPa (IQR=2.450-1.950), P=0.025). Those who had detectable HIV viral load had a significantly higher mean liver stiffness compared to the uninfected group [2.691 kPa (IQR=3.080-2.180) vs. 2.237 kPa (IQR=2.450-1.950), P=0.024]. In a regression analysis, those with detectable HIV viral load had significantly higher mean liver stiffness (β=0.389, SE=0.180, P=0.033) after controlling for age, BMI, and CD4 cell count. Conclusion: MRE, an accurate, non-invasive method to detect liver fibrosis in a cohort of HIV infected and HIV/HBV/HCV uninfected adults, shows that detectable HIV viral load is associated with more advanced liver fibrosis than when HIV viral load is undetectable or when patients are not infected. These results indicate a deleterious effect of the HIV virus on the liver and confirm earlier reported analyses using FIB-4 to estimate liver fibrosis in similar cohorts. Scott Dryden-Peterson 1 , Neo Tapela 2 , Isaac Nkele 1 , Oaitse John 1 , Gita Suneja 3 , Surbhi Grover 4 , Memory Bvochora-Nsingo 5 , Malebogo Pusoentsi 2 , Joseph Makhema 1 , Mompati O. Mmalane 1 , Mukendi Kayembe 2 , Shahin Lockman 1 1 Botswana Harvard AIDS Institute Partnership, Gabarone, Botswana, 2 Botswana Ministry of Health, Gaborone, Botswana, 3 Duke University, Durham, NC, USA, 4 Botswana–UPenn Partnership, Gaborone, Botswana, 5 Life Gaborone Private Hospital, Gaborone, Botswana Background: Antiretroviral therapy (ART) has reduced incidence of Kaposi’s sarcoma and several types of lymphoma. However, the impact of ART on risk of developing other cancers is less certain, particularly in sub-Saharan Africa where the great majority of HIV-infected individuals live. Botswana has nearly achieved the UNAIDS 90-90-90 targets. We sought to estimate the excess risk for cancer among people living with HIV in Botswana in the setting of high ART coverage. Methods: We prospectively enrolled consenting patients presenting for specialized oncology care at the four principal treatment centers in Botswana (October 2010 to September 2017). Patients with unknown HIV status were tested. Cancer diagnoses were abstracted from clinical and pathologic records. After finding similar time to diagnosis and similar cancer stage, we assumed non-differential case capture by HIV status. We utilized population- representative data from the Botswana AIDS Incidence Survey (2013) to estimate age- and sex-standardized incidence ratios (SIR). Results: A total of 2700 patients with cancer were enrolled, including 1477 (54.7%) with HIV and 1140 (42.2%) without HIV. Eighty-three patients (3.1%) with unknown HIV status were excluded from further analyses. The majority

Poster Abstracts

650 SOUTH AFRICAN HIV CANCER MATCH STUDY: A PILOT STUDY TOWARDS PRECISION PUBLIC HEALTH Mazvita Sengayi 1 , Wenlong C. Chen 1 , Adrian Spoerri 2 , Elvira Singh 1 , Matthias Egger 2 , Julia Bohlius 2 1 National Health Laboratory Service, Johannesburg, South Africa, 2 Institute of Social and Preventive Medicine, Bern, Switzerland Background: Precision Public Health capitalizes on large population-based data to provide targeted interventions for well-defined populations. We use machine learning to create a national cohort of HIV-positive people with cancer outcomes in South Africa. Methods: For this pilot study we retrieved laboratory test results for HIV-ELISA, CD4 cell counts and HIV-RNA from one South African province (Northern Cape) stored at the central data warehouse of the National Health Laboratory Service (NHLS) for the period 2004-2013. We used machine learning (dedupe; https:// github.com/dedupeio/dedupe) to identify records belonging to the same person and created a cohort of unique HIV-positive persons. This cohort was then linked to the National Cancer Registry. Linkage variables included first names, surnames, date of birth and sex. Persons with ≥2 CD4 cell count measurements were included for analysis. We calculated cancer incidence rates per 100,000 person-years from the date of first CD4 cell count to cancer diagnosis or last CD4 cell count date, whichever came first. We present cancer incidence rates stratified by sex. Results: We retrieved 653,071 laboratory records and created a cohort of 212,746 unique HIV-positive persons; 80,204 were included in incidence analyses. 66%were female, median age was 32 years (IQR 25-40) and median first CD4 cell count was 310 cells/µL (IQR 169-493). We identified 1,410 cancers, 63% (n=889) were incident. Overall cancer incidence per 100,000 person-years was 518 in men and 367 in women. The five cancers with the highest incidence rates were Kaposi sarcoma, cervical cancer, breast cancer, basal cell carcinoma of the skin and Non Hodgkin Lymphoma (Figure). After completion of the pilot study, we will use 49 million NHLS laboratory test results covering all South African provinces to create the national South African HIV Cancer Match Study. Conclusion: Machine learning allows to create a national HIV cohort with cancer outcomes based on large laboratory and cancer registry data. The spectrum of cancers seen in this pilot study is similar to those in previous South

649 HIV AND CANCER RISK IN CONTEXT OF HIGH ART COVERAGE IN BOTSWANA

CROI 2018 241