Inequality in outcomes for adolescents living with perinatally-acquired HIV in sub-Saharan Africa: a Collaborative Initiative for Paediatric HIV Education and Research (CIPHER) cohort collaboration analysis I’m very pleased to present this work on behalf of the CIPHER Cohort Collaboration Adolescent Project Team. Amy Slogrove & Marcel Yotebieng for the CIPHER Cohort Collaboration Adolescent Project Team IAS Conference 2017

The authors have no conflicts to declare Disclosures The authors have no conflicts to declare

World AIDS 2016 Global analysis of 38,187 adolescents living with perinatally-acquired HIV (APH) 30,296 (79%) sub-Saharan Africa 2-4 x greater mortality hazard At World AIDS 2016, we shared the first results from the CIPHER Cohort Collaboration Adolescent project. This was a multi-regional analysis of >38,000 adolescents living with perinatally-acquired HIV, [KEY] 79% of whom were living in SSA. Even after accounting for marked baseline differences, and in the context of probable under-ascertained mortality in SSA, adolescents in Africa had a [KEY] 2-4 fold greater hazard of mortality compared to adolescents in Europe, North America and South & Southeast Asia

Sub-Saharan Africa Home to 80% of the 1.8 million adolescents living with HIV 14/15 countries with the highest burden of adolescent HIV Progress in diagnostic and treatment interventions not uniform across the continent Younger adolescent AIDS-related deaths starting to decline in some countries, but continue to rise in others Sub-Saharan Africa (SSA) is a complex region marked by diversity and inequality. SSA is also home to 80% of the 1.8 million adolescents age 10 to 19 years living with perinatally- or horizontally-acquired HIV and 14 of the 15 countries with the highest burden of adolescent HIV. Progress in scaling up HIV diagnostic and treatment interventions has not been uniform across the continent. Furthermore, although AIDS-related deaths in younger adolescents have started to decline in a number of high-burden countries, they continue to rise in others. In pursuit of the Sustainable Development Goals, particularly achieving health for all at all ages and reducing inequality within and between countries, we wanted to understand how outcomes differ for adolescents living with perinatally-acquired HIV within Africa.

Objective & Definitions Primary Objective Compare characteristics and outcomes (mortality, transfer out, loss to follow-up) of APH by country income group (CIG) in sub-Saharan Africa Definitions APH – entered care before age 10 years, with no known non-vertical route of HIV-infection and were followed beyond age 10 years (survived beyond age 10 years) Lost to follow-up (LTFU) – last contact >365 days prior to database closure; censored 365 days after last visit Thus, the primary objective for this analysis was to compare the characteristics and outcomes, specifically mortality, transfer and loss to follow-up, of APH by country income group in SSA. APH were defined as HIV-infected children who entered care before age 10 years, as a proxy for perinatally acquired HIV, and had at least one visit beyond age 10 years. This analysis includes only children that had survived to at least 10 years of age. Adolescents were considered as lost to follow-up if their last contact was more than 365 days prior to database closure.

Methods CIPHER Global Cohort Collaboration Pooled individual retrospective data from 12 cohort networks This sub-Saharan Africa analysis: 25 countries represented by 7 networks Baylor International Pediatric AIDS Initiative (BIPAI) International Epidemiology Databases to Evaluate AIDS (IeDEA) IeDEA - Central Africa IeDEA - East Africa IeDEA - Southern Africa IeDEA - West Africa Médecins Sans Frontières (MSF) Pediatric Cohorts Identifying Optimal Models for Care in Africa (Optimal Models ICAP) Through the CIPHER Global Cohort Collaboration, individual retrospective data from 12 cohort networks across 5 regions of the world were originally pooled. This analysis includes 7 networks representing 25 countries.

Methods Characteristics compared by Country Income Group (CIG) at first visit, ART start, age 10 years and last visit World Bank CIG designation for median year of first visit low, lower-middle, upper-middle income Cumulative incidence for outcomes calculated by competing risks analysis (mortality, transfer out, loss to follow-up) Mortality hazard ratios - Cox proportional hazards models Characteristics were compared by CIG at key time-points being first visit, ART start, age 10 years and last visit. Countries were classified according to World Bank CIG designation for their median year of first visit. Outcomes were compared as cumulative incidence functions calculated by competing risks analysis to estimate mortality in the presence of high rates of transfer and loss to follow up. Mortality was further compared across CIG using Cox proportional hazards models.

APH in sub-Saharan Africa by Country Income Group Total APH N=30,296 78,619 person-years of adolescent follow-up Low Income N=22,925 (75.7%) 20/25 countries Lower-Middle Income N=1,386 (4.6%) 3/25 countries Over 30,000 adolescents, from 25 countries were included in this analysis, 20/25 countries were designated as low income countries (shown in orange) and home to 75% of the included adolescents. Lower middle income countries included Cameroon, Lesotho and Swaziland (in blue) and Upper Middle Income Countries included Botswana and South Africa (in green). This cohort accumulated a total of 78,619 person-years of follow-up between 10 and 19 years of age. Upper-Middle Income N=5,985 (19.8%) 2/25 countries

Birth Cohort All APH 64% born >2000 Range 1990-2005 Low Income Lower-Middle Income 65% born > 2000 Range 1996-2004 Almost 2/3 of adolescents were born in the year 2000 or later, but birth year ranged from as early as 1990 to as recently as 2005. Upper-Middle Income 57% born >2000 Range 1990-2005

Age in Years Median (IQR) To orientate you to this series of graphics, each cluster represents a time point, first visit on the left, ART start in the middle and last visit on the right. Within each cluster the median and IQR for the total cohort is shown for low income in orange, lower-middle income in blue and upper-middle income in green. In this slide the red-dashed horizontal line indicates age 10 years for ease of reference. The median age at first visit was just over 7 years and 7.9 years at ART start, with a slightly lower age at ART start in upper-middle compared to lower-middle and low income countries. Median age at last visit was 12.1 years, with little difference between CIG, indicating that this is largely a young adolescent cohort. L 7.3 (5.5; 8.7) 8.1 (6.3; 9.5) 12.0 (10.9; 13.7) LM 7.2 (5.7; 8.6) 7.8 (6.2; 9.3) 12.1 (10.9; 13.8) UM 6.6 (4.3; 8.4) 7.3 (5.2; 8.9) 12.4 (11.0; 14.3)

CD4 Count in cells/µl Median (IQR) Mean (95% CI) CD4 change ART Start – Last Visit: Low: 295 (286; 303) cells/µl Lower Middle: 463 (440; 486) cells/µl Upper Middle: 353 (338; 367) cells/µl CD4 cells/µl The median CD4 count at ART start, in the 2nd cluster, was close to 300 for all CIG; [KEY] however the mean CD4 count change between ART start and last visit was greatest in lower-middle Income countries at 463 cells/ul and lowest in low income countries at 295 cells/ul. L 418 (211; 721) 310 (165; 520) 652 (414; 947) 668 (434; 945) LM 391 (221; 616) 292 (174; 417) 707 (479; 972) 735 (532; 985) UM 361 (172; 662) 318 (162; 558) 719 (475; 1006) 729 (523; 971)

WHO Height-for-Age Z-score Median (IQR) Mean (95% CI) HAZ change ART Start – Last Visit: Low: 0.16 (0.14; 0.18) Lower Middle: 0.04 (-0.10; 0.02) Upper Middle: 0.44 (0.40; 0.49) Z-score WHO HAZ was similar across CIG at first visit and ART start, with the median Z-score below -2 in all CIG, indicating that the majority of children were already stunted at ART start. [KEY] There was some improvement in height following ART start in all CIG. Adolescents in upper-middle income countries experienced the largest improvement in height, a mean of 0.44 Z-scores. L -1.98 (-2.96; -1.03) -2.01 (-2.97; -1.08) -1.66 (-2.46; -0.90) -1.77 (-2.60; -0.95) LM -1.91 (-2.72; -1.08) -2.08 (-2.95; -1.33) -2.03 (-2.77; -1.30) -2.02 (-2.77; -1.30) UM -1.97 (-2.88; -1.10) -2.02 (-2.86; -1.17) -1.55 (-2.29; -0.87) -1.54 (-2.31; -0.77)

Outcomes Cumulative Incidence (95% CI) By Country Income Group Low Income Lower Middle Income Upper Middle Income Cumulative Incidence % Transfer LTFU Mortality Age in Years The cumulative incidence of mortality (shown in blue) was similar in low and lower-middle income countries at 3.5 and 3.9% respectively, and lowest at 1% in upper-middle income countries on the far right. However, LTFU (in green) was highest in upper-middle income countries at 14%. Low Income Lower Middle Income Upper Middle Income Mortality (%) 3.5 (3.1; 3.8) 3.9 (2.7; 5.4) 1.1 (0.8; 1.4) Transfer out (%) 17.5 (16.8; 18.3) 27.5 (24.2; 31.0) 23.7 (22.4; 25.1) LTFU (%) 13.1 (12.4; 13.8) 8.3 (6.3; 10.6) 14.1 (12.9; 15.3)

Survival Analysis Hazard Ratio (95% CI) Low Income Lower Middle Income Upper Middle Income 1. Unadjusted Hazard Ratio (HR) [N=30,296] 3.05 (2.27; 4.09) 3.57 (2.30; 5.54) Reference 2. Adjusted* HR – complete cases only [N=13,985] 3.75 (2.02; 6.95) 3.74 (1.80; 7.78) 3. Adjusted* HR – multiple imputation for missing CD4, WAZ, HAZ [N=30,296] 2.50 (1.85; 3.37) 2.96 (1.90; 4.61) 4. Adjusted HR# – multiple imputation for missing CD4, WAZ, HAZ & restricted to those ever on ART [N=26,018] 2.67 (1.94; 3.67) 3.07 (1.91; 4.95) On to the survival analysis...In reference to upper-middle income, adolescents in low and lower-middle income countries had a 3 fold elevated hazard for mortality. Adjusting for baseline characteristics the mortality hazards increased somewhat when including only complete cases (shown in model 2) and decreased somewhat when including all cases with multiple imputation for missing CD4, weight and height (shown in model 3). Restricting only to adolescents that did ever receive ART, made little difference to the relative mortality hazard across CIG as seen in model 4, and adolescents in low and lower-middle income countries that did receive ART still had a 2.5-3 fold greater hazard for mortality than adolescents in upper-middle income countries that did receive ART. * Adjusted for baseline characteristics – gender; age, CD4 count-, WAZ-, HAZ- at first visit; birth cohort; on ART ever # Adjusted for baseline characteristics – gender; age, CD4 count-, WAZ-, HAZ- at first visit; birth cohort

Conclusions The current generation of APH in SSA largely experienced improvement in immune status and growth despite starting ART at an advanced age Even when receiving ART, inferior growth improvement and higher mortality was seen in Low & Lower Middle Income compared to Upper Middle Income Countries Limitation: differences in outcomes by CIG may represent differential mortality ascertainment and the type of cohorts in each CIG (routine care vs. centres of excellence) rather than inequality according to CIG In conclusion, the current generation of adolescents surviving with perinatally-acquired HIV in sub-Saharan Africa largely experienced improvement in immune status and growth despite starting ART well into childhood. Even when receiving ART, inferior growth improvement and higher mortality was observed in APH from LIC and LMIC compared to UMIC signalling the role of factors beyond the ART program in determining the health and well-being of APH. Without broader national capacity development in LIC and LMIC in SSA, and measurable progress towards reducing inequality within and among countries, outcomes for APH in LIC and LMIC will continue to lag behind those of their peers in UMIC in this region. We do recognize however, that differences in outcomes by CIG may represent differential mortality ascertainment or the type of cohorts represented in each country income group rather than true inequity in outcomes by country income groups.

Next Steps Detailed growth analysis planned A better understanding of outcomes in APH that are not retained in care is needed to appropriately compare and interpret estimates of mortality Value in ongoing cohorts of APH to understand the changing population of adolescents living with perinatally-acquired HIV Within this CIPHER adolescent cohort, a detailed growth analysis is planned to further interrogate determinants of impaired growth in APH. A better understanding of outcomes in APH that are not retained in care is needed to appropriately compare and interpret estimates of mortality in this population. There is certainly value in continuing to follow cohorts of HIV-infected children aging in to adolescence, as the current generation of APH may be substantially different to future generations that will likely have started ART in infancy and may be impacted by different issues.

Acknowledgements: Adolescent Project Team Data center: Mary-Ann Davies, Michael Schomaker, Amy Slogrove Data managers: Sebastian Wanless, Charlotte Duff Members: BIPAI: Nancy Calles CCASAnet: Jorge Pinto EPPICC: Josiane Warszawski, Ali Judd IeDEA Asia-Pacific: Kulkanya Chokephaibulkit IeDEA Central Africa: Marcel Yotebieng IeDEA East Africa: Kara Wools-Kaloustian IeDEA Southern Africa: Nicky Maxwell IeDEA West-Africa: François Eboua, Valériane Leroy IMPAACT: Paige Williams MSF: Jihane Ben-Farhat Optimal Models (ICAP): Chloe Teasdale PHACS: George Seage I would like to acknowledge the adolescent project team

Acknowledgements International AIDS Society CIPHER - Steering Committee, Cohort Collaboration Oversight Group, Manager (Marissa Vicari) UCT CIDER Data Centre - Mary-Ann Davies, Michael Schomaker, Dolphina Cogill, Nicky Maxwell Contributing networks and their study participants Funders: IAS-CIPHER is made possible through funding from CIPHER Founding Sponsor ViiV Healthcare and Janssen. Individual networks contributing to the CIPHER Cohort Collaboration have received the following financial support: The International Epidemiology Databases to Evaluate AIDS (IeDEA) is supported by the U.S. National Institutes of Health’s National Institute of Allergy and Infectious Diseases, the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Cancer Institute, the National Institute of Mental Health, and the National Institute on Drug Abuse: Central Africa, U01AI096299; East Africa, U01AI069911; Southern Africa, U01AI069924; West Africa, U01AI069919. The Optimal Models project was supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention under the terms of Cooperative Agreement Number 5U62PS223540 and 5U2GPS001537. This work is solely the responsibility of the authors and does not necessarily represent the official views of any of the governments or institutions mentioned above. The IAS, CIPHER, The UCT Data Centre and all of the very generous contributing networks, their participants and funders