HPTN International Scholar (2018-19) Factors associated with HIV infection in serodiscordant couples in an African setting: HPTN 052 study sub-analysis Wilfred T. Gurupira MPH HPTN International Scholar (2018-19) University of Zimbabwe College of Health Sciences – Clinical Trials Research Centre Innovation in HIV Prevention Research Workshop, 22 August 2019 Thank you for the lovely introduction. I have always been fascinated by the complexities of human behaviour especially when part of a group or as a couple. Fortunately for me, my introduction into HIV prevention clinical trials was through the 052 serodiscordant couples study. This is a study which many people in this room took part in. And it’s a study which I hope everyone knows and still remembers. To refresh your memory,

HPTN 052: Immediate vs Delayed ART for HIV Prevention in Serodiscordant Couples HIV-infected, sexually active serodiscordant couples; CD4+ cell count of the infected partner: 350-550 cells/mm3 (N = 1763 couples) Immediate HAART Initiate HAART at CD4+ cell count 350-550 cells/mm3 (n = 886 couples) Delayed HAART Initiate HAART at CD4+ cell count ≤ 250 cells/mm3* (n = 877 couples) *Based on 2 consecutive values ≤ 250 cells/mm3. Primary efficacy endpoint: virologically linked HIV transmission Primary clinical endpoints: WHO stage 4 events, pulmonary TB, severe bacterial infection and/or death Couples received intensive counseling on risk reduction and use of condoms The study sought to examine the effect of ART on preventing HIV transmission to uninfected partners in serodiscordant relationships.

HPTN 052 was a randomized clinical trial conducted across 4 continents at 13 sites in 9 countries. The results were striking.

The study found that ART had a 96% reduction in the risk of transmission to these uninfected partners. 052 firmly established HIV treatment as prevention and has been recognised as being a landmark study.

Research gap: disproportionate number of seroconversions in Africa Number of couples enrolled Number of seroconversions (linked) Africa = 954 (54%) Asia = 531 (30%) Americas = 278 (16%) TOTAL = 1763 Africa = 23 (82%) Asia = 5 Americas = (18%) TOTAL = 28 (as at interim analysis 2011) Another result that struck me was the disproportionate number of seroconversions at African sites. A little over half of the 1763 couples enrolled into the study were from African sites. Yet 82% of partner-linked seroconversions in the study were observed at these African sites. This made me wonder why?

Primary objective To identify and characterize risk factors associated with HIV infection in serodiscordant couples enrolled into the HPTN 052 study at African sites compared to non-African sites Secondary objectives: To compare serodiscordant couples across regions To compare patterns of seroconversions by region, study arm and linkage To determine factors associated with HIV infection at African sites compared to non-African sites To answer this question we conducted a secondary analysis of 052 data. The main objective was to identify and characterize risk factors associated with HIV infection in 052 serodiscordant couples at African sites as compared to non-African sites. There were three secondary objectives. First, we compared couples across regions. Second, we compared patterns of seroconversions by region, study arm and linkage. Finally we determined factors associated with HIV infection at African sites compared to non-African sites. So what did we find: When comparing couples across regions at baseline

Results: 1. Comparison of couples Similarities Age group Education level (primary and secondary) Unprotected sex (no sex or 100% condom use) Marital status (married, cohabiting) Plasma RNA viral load (>400 copies/ml) Randomization arm (immediate vs deferred) Differences (at African sites) Gender: HIV-infected females Education level (no school and post-secondary): less with no school or post-secondary school Unprotected sex (<100% condom use): reports of <less than 100% condom use # of sexual partners in past week: more reports of 3-4 encounters Plasma RNA viral load (<400 copies/ml): more participants with <400 copies/ml Type of serodiscordancy: more HIV+female, HIV-males couples we found similarities in age, education at the primary and secondary levels, reports of no sex or 100% condom use, marital status, plasma viral load greater than 400 copies per ml, and randomization arm. Differences were observed in the following: Gender – There were more HIV infected females at African sites. African sites had less participants with no school or post-secondary school There were more reports of less than 100% condom use and reports of more sexual partners in the past week at African sites. African sites had more participants with viral load less than 400 copies per ml. Finally, Due to the gender differences I mentioned earlier, there were more HIV-infected female HIV-uninfected male couples at African sites.

2. Patterns of seroconversions by region, arm and linkage Looking at the patterns of seroconversions

Taking a closer look at seroconversions at African sites alone, Incidence of any HIV-1 seroconversion (all sites): disproportionate number of seroconversions in Africa   Africa non-Africa Randomization Arm # Partner infections Incidence Rate (per 100 person years) 95% CI  All partner infections  Immediate ART therapy 16 0.75 [0.43, 1.22] 3 0.14 [0.03, 0.40] Delayed ART therapy 51 2.50 [1.86, 3.29] 8 0.37 [0.16, 0.74] Total 67 1.61 [1.24, 2.04] 11 0.25 [0.13, 0.46] Linked partner infections 2 0.09 [0.01, 0.34] 1 0.05 [0.00, 0.26] 37 1.81 [1.28, 2.50] 6 0.28 [0.10, 0.61] 39 0.93 [0.66, 1.28] 7 0.16 [0.07, 0.33] We broke down the big picture of HIV incidence to look at distribution by region, randomization arm, and we finally concentrated on the numbers indicating the differences in incidence of all seroconversions and partner-linked seroconversions. Taking a closer look at seroconversions at African sites alone,

Number of any and partner-linked HIV-1 seroconversion by site (African sites only) We observed that seroconversions were not evenly distributed among the sites. In fact two sites, Liliongwe in Malawi and Kisimu in Kenya had by far the largest number of seroconversions compared to other African sites. If we excluded data from these 2 sites, HIV incidence rates between African and non-African sites would be comparable.

3. Risk of partner infection Finally we looked at some baseline variables to determine association with HIV infection.

Cumulative hazard:

Cumulative hazard:

4. Factors associated with seroconversion Finally we looked at some baseline variables to determine association with HIV infection.

Baseline variables associated with HIV infection We found 2 that were associated with infection. The first was education status. Primary and secondary schooling was associated with HIV infection at African sites. Second, was randomization arm. Being in the deferred arm was associated with HIV infection. None of the baseline variables we looked at had any association with HIV infection at non-African sites. Bringing it all together,

Discussion Factors associated with HIV infection: Baseline CD4 count, randomization arm (delayed) and education status (primary & secondary school) Regional and in country incidence variations (Kisumu and Lilongwe; South Africa and Malawi sites) Unexplored factors: desire for children, number of children, duration of relationship, relationship dynamics To-be-explored: time varying behavioural variables e.g. outside sexual partners Two sites drove the incidence of HIV infections at African sites. Two baseline variables were associated with HIV infection. There were some factors which we could not explore, a limitation which arose as this was a secondary analysis. We would have wanted to explore desire for children, or the number of children, or duration of the serodiscordant relationship or my personal favourite, relationship dynamics Data on all these variables was not available for analysis. We are yet to explore the association between time-varying factors and HIV infection. Based on the results, and the limitations we found, we have concluded that

Conclusion Need to explore factors associated with HIV infection among African serodiscordant couples there is need to explore factors associated with HIV infection among African serodiscordant couples. It have been an interesting journey to get here.

And like Sir Isaac Newton I see further because I stand on the shoulders of giants.

SCHARP and SCHARP statisticians My CTU leadership This work is supported by the HPTN Scholars Programme which is part of the HIV Prevention Trials Network (HPTN) Mentors: Profs Taha E. Taha and James Hakim SCHARP statisticians: Jing Wang and Amber Guo UZ CHS – CTRC: Z.M. Chirenje, J. Brown, A. Markowitz HPTN International Scholars Programme team: Erica Hamilton and Gabriela Salinas-Jimenez HPTN 052 Study team All HPTN Scholars The HPTN is funded by the National Institute of Allergy and Infectious Diseases (UM1AI068619, UM1AI068613, UM1AI1068617), with co-funding from the National Institute of Mental Health, and the National Institute on Drug Abuse, all components of the U.S. National Institutes of Health. These are my giants. My mentors SCHARP and SCHARP statisticians My CTU leadership The HPTN Scholars programme team especially Erica and Gabriela The 052 study team And my fellow Scholars past and present As I stand tall, with my head firmly in the clouds I long for the day when someone in this room at meeting just like this will announce that we have overcome the challenges of HIV prevention in serodiscordant couples. Thank you for listening!

HPTN Scholars and Mentors (June 2019) www.hptn.org/research/scholars https://twitter.com/HIVptn https://www.facebook.com/HIVptn