On 27 May 2015, early results from the international Strategic Timing of AntiRetroviral Treatment (START) study were announced by Dr Anthony Fauci, head of the US National Institute of Allergy and Immune Diseases (NIAID). The findings are expected to change HIV treatment guidelines globally.1
The results were not anticipated and will have implications for everyone: doctors, researchers, guidelines committees, policy makers and funders -- and HIV positive people, whether or not they are yet on treatment.
Timeline for DSMB Recommendations
Although only limited findings have been released from START, they show surprises that nobody predicted.
The top-line results are based on data collected up to 13 March 2015. A more detailed analysis will be available in time for the International AIDS Conference being held in Vancouver in July, and will include data to the end of May 2015.
The first surprise is that the results are at least 18 months earlier than expected. This is due to a recommendation from the independent Data and Safety Monitoring Board (DSMB) for the study, who were tracking unblinded results. On 15 May 2015, this small group of experts decided that the primary research question had been definitively answered. Although the overall rate of serious events was low, and involved fewer endpoints than expected, the difference between the two arms of the study was highly significant. Based on pre-defined rules for changing the study, the DSMB recommended that all participants be offered treatment and that follow-up should continue as planned. The continued follow up is important: the START study is therefore still ongoing.
START is a large randomised international study and the size, randomisation and global network involved are all key to the importance of the results. The primary research question was whether the benefits and risks of starting antiretroviral treatment (ART) at high CD4 counts outweighed the benefits and risks of waiting until later. The definition of "early" vs "late' was having a CD4 count above 500 cells/mm3 vs waiting until 350 cells/mm3. Until now, observational cohort studies had reported conflicting results on whether there were clinical advantages of starting earlier and randomised studies had deferring treatment until even lower CD4 cut-offs.
The two groups in START were compared based on the different rates of serious clinical events. These events included important AIDS-related and non-AIDS related illnesses together with deaths from all causes. Numerous sub-studies were included to look at the impact of both HIV and ART on other key areas of health in early infection -- including on bone heath, neurological function, cardiovascular risks, lung function and quality of life. Although information about the sub studies has not been released, these unblinded results were available to the DSMB when making their recommendations.
Enrolment and Patient Characteristics
From April 2009 to December 2013, START enrolled 4685 HIV positive people at 211 sites in 35 countries. By the time of the data cut in March for the 2015 DSMB review, the mean follow up time was about 3 years, contributing to 7000 patient years of follow-up.
Baseline characteristics for the main study and related sub-studies have been well described in the annual DSMB open reports and in a recent open access supplement to HIV Medicine.2,3
Age at enrolment ranged from 18 to 81 years, with a median age of 36 (IQR: 29 to 41). Just over half of participants were gay men and more than a quarter were women. Most people were enrolled within a year of their HIV diagnosis (median estimated time 1.0 (IQR: 0.4 to 3.0) years).
START is a global study; with 33% of participants in Europe, 25% in South America or Mexico, 21% in Africa, 11% in the US, 8% in Asia and 2% in Australia.
Given that entry criteria included having a CD4 count >500 cells/mm3, the study enrolled many people at CD4 counts that were significantly higher than expected, with 20% off participants starting above 800 cells/mm3. The median CD4 count was 651 cells/mm3 (IQR: 584-765; range 503 to 2296). Viral load at baseline also showed that this was a group in early infection. Median viral load was about 12,000 copies/mL (IQR: 3,000 to 40,000) and 8% had viral load <400 copies/mL.
The number of other medical complications within this diverse group is important. At baseline, almost one-third were current smokers, half had at least one cardiovascular risk based on the Framingham calculator, almost one in five either had hypertension or were on hypertensive treatment and 8% either had high blood lipids or were on lipid-lowering drugs. Just over 3% had diabetes, were using diabetes treatment, or had high fasting glucose. Prevalence of viral hepatitis was 2.9% and 3.7% for coinfection with hepatitis B and C, respectively. Other important health issues included about that 3% had documented alcohol or substance use issues and 6% had a psychiatric diagnosis (including depression, bipolar and other conditions).
These baseline data suggested that cardiovascular disease (CVD) and non-AIDS cancers would be the most common serious events. START was designed as an endpoint-driven study and modeling projected that 213 serious events would be needed to show that earlier treatment would reduce the risk of events by about one third. The study was expected to run until the end of 2016.
Top Line Results: ART Has Greatest Impact on HIV-Related Events
In the press conference, results were released for three separate endpoints, see Table 1 below. The number of events for each endpoint and the relative differences between the early and late treatment groups were also provided. These figures are based on data from March 2015 and so will change slightly when the final dataset become available.
The differences were highly significant for endpoints 1 and 2, but not for endpoint 3.
- The combined endpoint of AIDS, serious non-AIDS or death was reduced by 53% in the early treatment group (HR 0.47; 95%CI 0.32 to 0.68). This was based on 41 vs 86 events in the early vs deferred groups, with rates of 0.60 vs 1.25 per 100 person years (PY).
- The combined endpoint of AIDS or death was reduced by 70%, (HR 0.30; 95%CI 0.17 to 0.55). This was based on 14 vs 46 events in the early vs deferred group, with rates of 0.20 vs 0.66 per 100 PY.
- The combined endpoint of serious non-AIDS events or non-AIDS related deaths was reduced by 33% (HR 0.67; 95%CI 0.42 to 1.09). This was based on 28 vs 41 events in the early vs deferred groups with rates of 0.41 vs 0.59 per 100 PY. This difference is not statistically significant because the 95% confidence intervals cross 1.0.
These preliminary results were not expected. They show that early treatment had a greater impact on HIV-related illnesses than on non-AIDS events. To understand the importance of this finding, the clinical concern behind some treatment guidelines already recommending ART at CD4 counts above 500 cells/mm3 was driven by risk of non-AIDS illnesses. Non-AIDS events include serious heart, liver, kidney disease and non-AIDS cancers.
|Table 1. Primary Endpoint and Its Components in Open DSMB Report (15 May 2015)|
||Arm A/B (95% CI)
|AIDS, serious non-AIDS,
or death (primary)
||0.47 (0.32 to 0.68)
|AIDS or AIDS death
||0.30 (0.17 to 0.55)
or non-AIDS death
||0.67 (0.42 to 1.09) NS**
* PY = patient years, ** NS = non significant