The global epidemiology of perinatally HIV-infected adolescents: a Collaborative Initiative for Paediatric HIV Education & Research (CIPHER) Global Cohort Collaboration analysis Amy Slogrove, Ali Judd, Valeriane Leroy for the CIPHER Global Cohort Collaboration Adolescent Project Team Good afternoon ladies & gentlemen. I am very pleased to share this analysis of the epidemiology of perinatally HIV-infected adolescents on behalf of the CIPHER global cohort collaboration adolescent project team AIDS 2016, Durban, South Africa

Background Global access to antiretroviral therapy (ART) is expanding Emerging population of perinatally HIV-infected adolescents (PHA) across all regions HIV-infected adolescents face barriers compared to adults Access to ART Remaining in care Access to ART is expanding across the globe, resulting in an emerging population of perinatally HIV-infected adolescents across all regions of the world. HIV-infected adolescents face major barriers compared to HIV-infected adults, specifically in access to ART and remaining in care

Objective & Definitions Primary Objective Describe the global epidemiology and geographic trends of characteristics and outcomes (mortality, transfer out, loss to follow-up) of PHA surviving beyond age 10 years Definitions PHA – entered care before age 10 years, with no known non-vertical route of HIV-infection and were followed beyond age 10 years Lost to follow-up (LTFU) – last visit >365 days prior to database closure; censored 365 days after last visit The primary objective of this study is to describe the global epidemiology and geographic trends of characteristics and outcomes, specifically mortality, transfer and loss to follow-up, of perinatally HIV-infected adolescents surviving beyond age 10 years. PHA were defined as HIV-infected children who entered care before age 10 years, as a proxy for perinatal HIV-infection, in addition had no known non-vertical route of HIV-infection and had at least one visit beyond age 10 years. Adolescents were considered as lost to follow-up if their last visit was more than 365 days prior to database closure, and were censored at 365 days after last visit.

Methods CIPHER Global Cohort Collaboration Individual retrospective data from 12 cohort networks were pooled, representing 50 countries South America & Caribbean: CCASAnet North America: IMPAACT & PHACS Europe & Central Asia: EPPICC South & Southeast Asia: EPPICC, IeDEA - Asia-Pacific, MSF Sub-Saharan Africa: BIPAI, IeDEA - Central Africa, IeDEA - East Africa, IeDEA - Southern Africa, IeDEA - West Africa, MSF, Optimal Models (ICAP) Through the CIPHER Global Cohort Collaboration, individual retrospective data from 12 cohort networks across 5 regions of the world and representing 50 countries were pooled.

Methods Describe characteristics at first visit, ART start, age 10 years and last visit compared by region Cumulative incidence for outcomes calculated by competing risks analysis (mortality, transfer out, loss to follow-up) Mortality hazard ratios - Cox proportional hazards regression This introductory analysis describes characteristics at first visit, ART start, age 10 years and last visit by region. Outcomes were compared as cumulative incidence functions calculated by competing risks analysis to be able to estimate mortality in the presence of high rates of transfer and loss to follow up. Mortality was compared across regions by hazard ratios from Cox proportional hazards regression. Adjusted hazards ratios were calculated using only complete cases without missing data as well as using all cases following multiple imputation for missing CD4 and height measurements.

South America & Caribbean Perinatally HIV-infected Adolescents Total N = 38 187 Female 50.6% North America 1 032 (3%) Female 50.1% Europe & Central Asia 3 054 (8%) Female 51.7% South & SE Asia 2 902 (8%) Female 50.1% In total 38187 adolescents were included, 79% from sub-Saharan Africa, 8% each from Europe & Central Asia and South & SE Asia, with smaller contributions from North America and South America & Caribbean. Just over half of the adolescents in each region were female South America & Caribbean 903 (2%) Female 53.8% Sub-Saharan Africa 30 296 (79%) Female 50.5%

South America & Caribbean Median (IQR) year of birth Total Year 2000 (1998; 2002) North America Year 1994 (1992; 1996) Europe & Central Asia Year 1995 (1991; 1999) South & SE Asia Year 2001 (1999; 2002) The median year of birth for this adolescent cohort was the year 2000, ranging from a median year of 1994 in North America to 2001 in South & Southeast Asia South America & Caribbean Year 1998 (1995; 2000) Sub-Saharan Africa Year 2000 (1999; 2002)

Age: Median (IQR) Age in Years First Visit ART Start Last Visit 15 6.7 5 10 15 Age in Years 6.7 (4.4; 8.4) 7.5 (5.2; 9.2) 12.4 (11.1; 14.4) This slide shows the age of adolescents at the 3 key time points, first visit, ART start and last visit. To orientate you to the graphics, the bar to the extreme left of each group represents the total cohort, with each region then represented individually to the right, and the horizontal red line indicates age 10 years. The median age at first visit for the total cohort was 6.7 years, ranging from a median of 8 months in North America to 7 years in sub-Saharan Africa. At ART start the median age for the total cohort was 7.5 years ranging from a median of 1 year in North America to 8 years in sub-Saharan Africa. Adolescents were followed to a median age of 12.4 years ranging from just over 12 years in Africa to 16.5 years in Europe. This indicates that for most regions we are only describing very early adolescence. First Visit ART Start Last Visit

CD4 Percent: Median (IQR) 10 20 30 40 CD4 percent (%) Corresponding with the variation in age at first visit and ART start is also wide variation in CD4% at these 2 time points, ranging from a median CD4 of 10% in S&SE Asia to 30% in North America. By age 10 years and last visit there was far less heterogeneity across regions with median CD4% in all regions greater than 25%. 16% (9; 25) 28% (20; 34) 14% (8; 20) (21; 35) First Visit ART Start Age 10 Years Last Visit

CD4 Percent: Median (IQR) 10 20 30 40 On ART 29% (21; 35) CD4 percent (%) Not on ART 25% (19; 32) Focussing on the far right panel, this now shows CD4% at last visit by adolescents on ART or not at their last visit, and as expected CD4% is somewhat lower at a medina of 25% in those not on ART compared to median 29% in those on ART at their last visit. First Visit ART Start Age 10 Years Last Visit

Proportion on ART % on ART First Visit Age 10 Years Last Visit 88% (33 514) Ever received ART 50 10 30 20 40 60 70 80 90 80% (30 072) 67% (26 953) % on ART This slide represents the proportion of children on ART at the 3 time points. 88% of children received ART at some stage. At age 10 years 2/3 of adolescents were on ART and 80% were on ART at their last visit. Of those not on ART at their last visit, 62% were ART naïve and 38% were ART experienced, but no longer on ART at last visit. 12% (4037) of those ever receiving ART started > 10 years of age. 8% (3 091) First Visit Age 10 Years Last Visit

Virologic Suppression HIV viral load available in 39% of adolescents on ART 75% (6 627/8 825) 72% (9 388/13 114) 50 10 30 20 40 60 70 80 % Suppressed Only 39% of adolescents that received ART had an HIV viral load measurement. Within this sub-group 75% at age 10 years and 72% at their last visit had an HIV viral load below 400 copies/ml or below the threshold of detection for the test at the time. Age 10 Years Last Visit

WHO Height-for-Age Z-score (HAZ) 1 -1 -2 -3 Median (IQR) HAZ Growth, measured by WHO height-for-age Z-score was severely impaired and in all regions HAZ at first visit and ART start fell well below the WHO Z-score of zero, indicated by the red horizontal line. By age 10 years and last visit some recovery was seen in North America and Europe, but substantial impairment remained particularly in S&SE Asia and SSA. -1.92 (-2.91; -0.97) -1.95 (-2.91; -1.02) -1.53 (-2.35; -0.72) -1.59 (-2.45; -0.72) First Visit ART Start Age 10 years Last Visit

WHO Height-for-Age Z-score (HAZ) 1 -1 -2 -3 On ART -1.58 (-2.43; -0.72) Median (IQR) HAZ Not on ART -1.62 (-2.50; -0.72) Focussing again on the far right panel, this now shows height-for-age Z-score by those on ART or not at their last visit, with very little difference between these two groups; median z-score of -1.58 in those on ART compared to -1.62 in those not on ART at last visit. -1.59 (-2.45; -0.72) First Visit ART Start Age 10 years Last Visit

Outcomes: Total Cohort Transferred out 15.6% (95% CI 15.1; 16.0) Cumulative Incidence Function LTFU 11.3% (95% CI 10.9; 11.8) For the total cohort, between 10 and 15 years of age, cumulative incidence for observed mortality (in blue) was 2.6%, LTFU (in green) was 11.3% and transfer out (red) was 15.6% Mortality 2.6% (95% CI 2.4; 2.8%) Time in years from age 10

South America & Caribbean Outcomes: by Region All PHA North America South America & Caribbean Transferred Out Cumulative Incidence Function LTFU Mortality Europe South & Southeast Asia Sub-Saharan Africa This slide shows the cumulative incidence curve for each region with the curve for all adolescents in the upper left panel for comparison, the red dashed horizontal line indicating 5%. Mortality between 10 and 15 years of age was below 5% in all regions, with LTFU and transfers out close to 5% in most regions except SSA, in the bottom far right panel, where transfers (in red) reached almost 20% by age 15 years and LTFU (in green) approximately 13%. Time in years from age 10

Mortality Hazard Ratios Unadjusted HR (95% CI) Adjusted HR* Total N 38 187 Europe & Central Asia 1.00 North America 1.70 (0.87; 3.31) 2.18 (1.11; 4.31) South & Southeast Asia 3.21 (2.03; 5.07) 1.75 (1.06; 2.89) South America & Caribbean 6.07 (3.88; 9.50) 4.52 (2.85; 7.18) Sub-Saharan Africa 4.35 (3.02; 6.28) 2.53 (1.65; 3.88) The unadjusted hazards for mortality was significantly elevated in South & Southeast Asia, South America & Caribbean & Sub-Saharan Africa relative to Europe

Mortality Hazard Ratios Unadjusted HR (95% CI) Adjusted HR* Total N 38 187 Europe & Central Asia 1.00 North America 1.70 (0.87; 3.31) 2.18 (1.11; 4.31) South & Southeast Asia 3.21 (2.03; 5.07) 1.75 (1.06; 2.89) South America & Caribbean 6.07 (3.88; 9.50) 4.52 (2.85; 7.18) Sub-Saharan Africa 4.35 (3.02; 6.28) 2.53 (1.65; 3.88) The adjusted model, presented in the last column, takes into consideration the marked differences by region in age, CD4 and height at the first visit, as well as differences in calendar period of birth across regions, all factors determining mortality. I would like to preface the presentation of this adjusted model with the fact that this is still a preliminary analysis and there are a number of regional differences, particularly related to changes over time that we are still finding solutions to dealing with in the analysis. That said, in this adjusted model there is an increase in the HR for North America, but for the remaining 3 regions there are reductions in the mortality hazards after adjusting for baseline differences. Nevertheless, adolescents in South & Southeast Asia, South America & the Caribbean and Sub-Saharan Africa remain at a substantially elevated hazard of mortality compared to European adolescents, despite surviving to 10 years of age. * Preliminary multivariable model: adjusted for gender, calendar period of birth, age at first visit, CD4% at first visit, HAZ at first visit

Discussion Survivor cohort that beat the odds Limited to PHA that presented before 10 years of age At first visit and ART start, age & CD4% differed substantially by region By 10 years and at last visit, CD4% , proportion on ART and proportion virologically suppressed were similar across regions Height growth remained severely impaired in most regions even in those receiving ART Observed mortality < 3% during adolescence in PHA surviving to 10 years of age should be considered as a conservative estimate Mortality during adolescence is substantially elevated for perinatally HIV-infected adolescents in South America & Africa relative to Europe In interpreting these findings it needs to be born in mind that this is a survivor cohort, and that in most regions these are children that beat the odds for their generation and made it to adolescence. Also important to note is that this analysis was limited to adolescents that did present before age 10 years and excludes the important group of perinatally HIV-infected adolescents that are still only being identified and diagnosed after age 10 years. In this context we see that there was substantial variation by region in age and CD4% at first visit and ART start. However by age 10 years and last visit, CD4%, proportion on ART and proportion virologically suppressed were similar across regions. Height growth remained severely impaired even in adolescents receiving ART. Loss to follow-up was high in SSA, and observed mortality of less than 3% during adolescence should be considered as a conservative estimate in light of probable under-ascertainment of mortality as well as the relatively short duration of follow-up only capturing early adolescence for most PHA. Despite probable under-ascertainment of mortality in Africa, perinatally HIV-infected adolescents in Africa remain at a substantially elevated hazard of mortality during adolescence relative to their European peers as do adolescents in South America & the Caribbean.

Adolescent Project Team Co-chairs: Ali Judd (EPPICC) and Valeriane Leroy (IeDEA-West-Africa) Data center: Mary-Ann Davies, Michael Schomaker, Amy Slogrove Data managers: Sebastian Wanless, Charlotte Duff Members: BIPAI (Baylor): Nancy Calles CCASAnet: Jorge Pinto EPPICC: Josiane Warszawski 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 IMPAACT: Paige Williams MSF: Jihane Ben-Farhat Optimal Models (ICAP): Chloe Teasdale PHACS: George Seage

Acknowledgements International AIDS Society CIPHER Steering Committee CIPHER Global Cohort Collaboration Oversight Group Marissa Vicari – Manager, CIPHER UCT CIDER Data Centre Mary-Ann Davies, Michael Schomaker, Dolphina Cogill, Nicky Maxwell Networks BIPAI, CCASAnet, EPPICC, IeDEA-Asia-Pacific, IeDEA-Central Africa, IeDEA-East Africa, IeDEA-Southern Africa, IeDEA-West Africa, IMPAACT, MSF, PHACS, Optimal Models

LTFU-Mortality Scenarios Reference = Europe Original HR estimate A 100% of all LTFU is mortality B 50% of all LTFU is mortality C 20% of all LTFU is mortality D 1 5 10

LTFU-Mortality Scenarios Reference = Europe Original HR estimate A 50% of LTFU in Africa & 5% of LTFU rest B 20% of LTFU in Africa & 5% of LTFU rest C 5% of LTFU in Africa & 20% of LTFU rest D 1 5 10

Mortality Hazard Ratios Unadjusted HR (95% CI) Complete Cases Only Adjusted HR* Multiple Imputation Total N 38 187 10 302 Europe & Central Asia 1.00 North America 1.70 (0.87; 3.31) 2.63 (0.67; 10.41) 2.18 (1.11; 4.31) S&SE Asia 3.21 (2.03; 5.07) 1.77 (0.57; 5.47) 1.75 (1.06; 2.89) South America & Carib 6.07 (3.88; 9.50) 3.75 (1.14; 12.36) 4.52 (2.85; 7.18) Sub-Saharan Africa 4.35 (3.02; 6.28) 4.10 (1.51; 11.13) 2.53 (1.65; 3.88) * Adjusted for gender, calendar period of birth, age at first visit, CD4% at first visit, HAZ at first visit

Mortality Hazard Ratios Unadjusted HR (95% CI) Complete Cases Only Adjusted HR* Multiple Imputation Total N 38 187 10 302 Sub-Saharan Africa 1.00 Europe & Central Asia 0.23 (0.16; 0.33) 0.24 (0.09; 0.66) 0.40 (0.26; 0.61) North America 0.39 (0.22; 0.69) 0.64 (0.18; 2.33) 0.86 (0.44; 1.68) S&SE Asia 0.74 (0.55; 1.00) 0.43 (0.23; 0.81) 0.69 (0.51; 0.94) South America & Carib 1.39 (1.03; 1.88) 0.92 (0.36; 2.34) 1.79 (1.28; 2.50) HR with Africa as reference – models the same as in main presentation * Adjusted for gender, calendar period of birth, age at first visit, CD4% at first visit, HAZ at first visit