A late breaker oral presentation from the HPTN-052 study was widely reported from IAS this year, with the simple conclusion that it added to the evidence supporting clinical benefits of earlier treatment.1
These findings are reassuring, and certainly take away nothing away from the more important primary prevention endpoint results that treatment dramatically reduces the risk of sexual transmission.2 However, the primary clinical endpoint of benefits to the HIV positive partners, were driven by a few very specific events.
The study randomised over 1700 HIV positive people with CD4 counts 350-550, who were in serodifferent heterosexual relationships, to either start treatment immediately or wait until their CD4 count dropped to 250 cells/mm3. Patients were recruited at sites in Africa (54%), Asia (34%) and South America (16%).
HPTN-052 presented first results following a DSMB recommendation to discontinue the study four years earlier than expected. This year, data from a final 855 patient years of follow up (total 4038 years) and a broader range of clinical events was presented by Beatriz Grinsztejn for the study team.
Median (IQR) CD4 and viral load at baseline was approximately 430 cells/mm3 (360-520) and 4.4 log copies/mL (3.8-4.9). Median follow-up was 2.1 years (1.6-2.9) across both arms. A total of 213 patients (24%) in the deferred arm started treatment at a median CD4 count and viral load of 230 cells/mm3 (197-249) and 4.4 log copies/mL (3.9 – 4.9). Time to ART was a median 3.8 years (IQR 3.5-4.4) with approximately 1 year on ART compared to 2 years in the immediate treatment group.
134 people experienced at least one primary clinical event (77 vs 57; 9% vs 6%) including 26 deaths (15 vs 11) and 21 non-AIDS events (9 vs 12), all in the delayed versus immediate arms respectively.
Compared to immediate ART, delayed ART was associated with higher rates of AIDS-defining events (p=0.03) and tuberculosis (p=0.02); and higher incidence (per/100 years) of tuberculosis (1.8, 95% CI=[1.3, 2.6] vs 0.8, [0.5, 1.3], p=0.009) and all targeted clinical events (29.0, [26.3, 31.9] vs 24.7, [22.3, 27.3]; p=0.02).
However there was only a trend to shorter time to first primary clinical event [HR 1.37 (0.97 - 1.93), p=0.07] with no differences seen in the Kaplan-Meier plots over the first year.
The number of secondary endpoints was numerically similar between both groups (n=317 vs 298; 36% vs 34%), although incidence rates became significant (incidence 494 vs 427/100 PY; p=0.05). Herpes and candida were more likely with delayed ART and dyslipidaemia was more likely with early ART.
Although the study reported clinical benefits from earlier treatment this was not consistent across all events and all analyses. While earlier treatment may have advantages, this study should not clearly translate to a conclsion for treatment at higher CD4 counts.
Firstly, the significantly higher rate of extrapulmonary tuberculosis in the delayed arm (17 vs 3 cases; p< 0.002) was driven by higher reports in two sites in India, rather than reflecting a trend from the study as a a whole.3 Secondly, serious non-AIDS events, rather than being reduced by earlier treatment, occured more frequently with immediate treatment.
Another detail, from the supplementary information made available with publication in the New England Journal of Medicine last year, is that the three suicides in HPTN-052 were all in the immediate arm in patients taking efavirienz: a caution that should temper any recommendation to widely prescribe treatment at higher CD4 counts. These indivudals may have died many years before they were likely to have a demonstrable clinical benefit from treatment.