Adherence of 80-95% not good enough for long-term treatment success in British Columbia HIV patients

Adherence of less than 95% is associated with a substantially lower chance of a good response to treatment, even using relatively relaxed definitions of virologic and immune success, according to a new analysis reported in AIDS. The study, by investigators in British Columbia, also found that people who took less than 80% of their medication (people likely to be missing doses on a weekly basis) had only a 10 – 15% probability of achieving and maintaining a good response to treatment during the four-year follow-up period, while those taking between 80% and 95% of all doses had no more than a 41% probability of a good response to treatment.

Successful antiretroviral treatment (ART) is dependent on sustaining high rates of adherence, but there has been recent conflicting evidence over just how critical it is for a person to maintain adherence above 95% (the level associated with the lowest risk of treatment failure in several past studies). For example, a recent Spanish study found only a very small difference in rates of viral rebound over one year between people who missed no more than one in ten doses (90% adherence) and people who missed no more than one in five doses (80% adherence).

However, the British Columbia study, following patients for four years but using a more relaxed definition of treatment success which permitted patients to spend up to a third of the follow-up period with detectable viral load, found that even this modest goal was substantially less likely to be met if adherence fell into the 80% to 95% range, when compared to adherence above 95%.

The study was conducted using patient data from the British Columbia Centre for Excellence in HIV/AIDS, which has distributed no-cost ART to eligible HIV-positive individuals in that Canadian province since 1992. The population in this study included 878 adult patients (over 18 years of age) who began ART between January 2000 and November 2004, and had at least two years of follow-up data.

The median age was 40 years, 80% were men and 28% had a history of injecting drug use. The cohort had reasonably advanced disease at baseline, with a median CD4 cell count of 165 cells/mm³ and median viral load of 100,000 copies/ml; 17% had been diagnosed with AIDS.

Patients were followed until November 2005. Over the time of follow-up (median 3.7 years), the median CD4 cell count increase was 145 cells/mm³; "immunologic success" was defined as having a CD4 cell increase of 145 cells/mm³ or greater. Virologic outcome was defined as the percentage of follow-up time with a plasma viral load below 50 copies/ml. The median virologic outcome was viral suppression (below 50 copies/ml) for 65% of follow-up time; "virologic failure" was defined as maintaining suppression for less than 65% of follow-up time. Adherence was simply based on prescription refills.

Outcomes were grouped into three categories: "best" (immunologic and virologic success), "worst" (immunologic and virologic failure), and "incomplete" (mixed response). Of the 878 cohort members, 394 (45%) had "best" responses, 350 (40%) "incomplete" and 134 (15%) "worst" responses.

Better outcomes were dependent on better adherence. Several other factors also predicted better response in an unadjusted analysis, including male sex, no history of injecting drug use (IDU), no use of unboosted protease inhibitors (PIs), and age.

A final analysis that adjusted for age, injecting drug use, baseline CD4 cell count and viral load, AIDS diagnosis, physician experience and follow-up time found that the effects of adherence varied between types of ART regimen. Unboosted protease inhibitors were associated with poorer outcomes at all adherence levels. Participants were more likely to have best responses on boosted-PI or NNRTI-based regimens at adherence rates of at least 95%. Adherence rates between 80 and 95% on NNRTI-based regimens were likely to result in either best or incomplete responses.

The probability of having the best response to treatment was between 60% and 70% for patients taking NNRTIs or boosted protease inhibitors at 95% adherence, but the probability of the best response to treatment was only 35 to 41% when taking either class of drugs at 80 to 95% adherence. At an adherence level below 80% the probability of the best response to treatment was somewhere between zero and 15%.

The worst, as compared with the best responders, had much higher risks of emergence of drug resistance (odds ratio [OR]: 10.56; 95% confidence interval [CI]: 5.93–18.81) and of death (OR: 6.09; 95% CI: 2.57–14.42), even after adjusting for adherence and ART regimen. Those with adherence below 40% were at sixfold greater risk of death during the four year follow-up period compared to those with adherence of at least 95%.

The authors say that their definition of treatment success is more stringent than the measure of viral rebound used in other studies looking at adherence, because it rules out patients who may be experiencing viral `blips` just above the limits of detection of the viral load test. In other respects however, the definition could be viewed as generous, since it permitted patients who spent a third of the follow-up period with detectable viremia to be counted as treatment "successes" - a debatable classification.

The crude assessment of adherence - pharmacy refills - limits the validity of the data, say the authors. Nevertheless, the investigators concluded that - as in many other trials - "patients not attaining the best virologic and immunologic responses are at a high risk for emergence of drug resistance and mortality, and these responses are highly dependent on the adherence level and initial HAART regimen. Patients on protease inhibitor-single did worse no matter the adherence level."