Individual ARV pharmacokinetics may be affected by variability in drug absorption, protein binding, CYP450 metabolism, and efflux pump (e.g., P-glycoprotein) activity. [56
] Drug interactions in HIV-infected patients commonly occur when cART regimens are selected and the patient is already taking other medications. [58
] In addition, many patients with active substance use experience drug interactions from SRD treatment with methadone or buprenorphine. [31
] Furthermore, active substance is often an important consideration that may influence NNRTI and PI pharmacokinetics during periods of irregular adherence. Lastly, additional medications for co-morbidities such as hyperlipidemia and depression may lead to drug interactions resulting in an unpredictable pharmacokinetic profile. [39
Our findings suggest the use of TDM to determine net pharmacokinetic effects in patients with SRD. In addition, our data indicate that substance users can be monitored during a TDM program given adequate adherence counseling and primary care follow up. Furthermore, HIV clinicians should consider complex drug interactions and co-morbidities as interactive pharmacologic factors within an individual patient that, when considered together, are important aspects of delineating individual antiviral exposure during chronic ARV dosing. [6
] Although these factors are logical and are likely a part of clinical care to varying degrees, some consideration should be given to examining alternative approaches to managing HIV infection that include the use of TDM to identify patients who may benefit from more extensive pharmacokinetic evaluation [33
Another aspect of using TDM to evaluate NNRTI and PI pharmacokinetics is the relationship between adherence and NNRTI and PI plasma concentration measurements. The integration of adherence measurements with pharmacokinetics to arrive at conclusions about drug exposure has been previously described. [60
] In our study, patients were able to demonstrate short-term adherence to their regimens in a manner that allowed TDM to be used to gain estimates of drug exposure. Although we excluded some patients in this analysis who were not able to have their samples drawn within the desired range for sample collection times, we anticipate that all of the data will be used in conjunction with pharmacokinetic modeling approaches that will be employed to generate AUC values and predicted trough concentrations.
In the present study, the observed concentration distribution of atazanavir, lopinavir and efavirenz among patients without SRD was consistent with previously reported data. The mean ATV trough concentration of 0.956 µg/mL was higher than that noted in the British TDM study, 0.774 µg/mL [62
], but similar to that reported by European Medicine Agency, 0.862 µg/mL. The possible reasons for the higher than expected ATV trough concentrations may include the diverse patient population in our study: 40% African Americans and 34% Hispanics, and our attempts to promote adherence. To our knowledge, this is the first report indicating significantly lower ATV trough concentrations (~3 fold) among patients with active SRD. For LPV, the median (IQR) of 5.30 µg/mL (3.25–7.19 µg/mL) was similar to that observed in a recent ACTG A5073 study, 5.60 µg/mL (3.30–8.20 µg/mL) [63
]. Active SRD resulted in a ~30% decrease in LPV mean trough concentrations. For efavirenz, the median (IQR) EFV trough concentration of 2.371 µg/mL (IQR 1.690–3.634 µg/mL) among patients without active SRD was consistent to that reported 2.011µg/mL (IQR 1.539–2.556 µg/mL) [64
]. However, the impact of active SRD was minimal as the EFV trough concentrations were similar between SRD and non-SRD arms.
The observations in this study indicated low PI concentrations in patients with active substance use. In particular, there were statistically significant differences in ATV troughs among patients with and without active SRD. There are a number of potential causes of lower PI concentrations in substance users in our study. First, self-report may overestimate adherence in patients with HIV [65
]. Therefore, it is possible that incomplete adherence led to the differences in trough concentrations in these groups and active substance users in our study may have been less adherent than patients without SRD. However, if adherence fully explained the differences in PI trough concentrations between groups, we might have seen a disparity in the EFV trough concentrations. Second, the relatively small number of patients in each group limits the statistical power of the observations. A trend toward lower median trough concentrations was noted in patients with SRD taking LPV (3.75 vs.
5.30 µg/mL, p=0.111). The same trend is true for EFV (1.917 vs.
2.371 µg/mL, p=0.211).
A significantly lower proportion of antiretroviral-treated patients with active SRD failed to suppress HIV (HIV RNA ≤ 75 copies/ml) thus leading to a greater potential for treatment failure. This finding is consistent with previous results from retrospective studies. [18
] In addition, patients with SRD had higher rates of inadequate PI and/or efavirenz exposure than those without SRD, possibly increasing the emergence of resistant HIV in this population. TDM of PIs and efavirenz represents a research tool that provides a “pharmacokinetic phenotype” that reflects individual differences in pathophysiology, genetics, and other factors that make it challenging to predict ARV pharmacokinetics particular among HIV patients with SRD.
Our data also highlight the number of CNS acting drugs that are prescribed in combination with antiretrovirals. Although we have grouped these drugs together for the purpose of summarizing them, it is interesting to note the extensive use of anti-anxiety agents, antidepressants and antipsychotics. In particular, the striking number of CNS acting drugs suggests that more rigorous attention to the initiation of antiretrovirals, particularly efavirenz, is an important consideration in substance users.
There are a few limitations of the present study. First, the relatively small sample size has weakened the statistical power of these observations. Second, the large inter-patient variability in the point evaluation of pharmacokinetics, i.e. trough concentrations could have underestimated the impact of substance use. In this case, the development of a pharmacokinetic model will allow samples to be collected during a range of time points, with subsequent model-derived trough concentrations then used to individualize dosing. These types of analyses will allow our dataset to be examined for the importance of maximum, minimum or time-averaged PI or EFV exposure as predictors of antiviral response or long-term toxicity. In addition, a detailed analysis of possible substance-drug and drug-drug interactions and their effects on antiretroviral therapy should be performed at individual substance and agent level.
In summary, TDM provides a clinical research tool for determining the net pharmacokinetic effects of contributing factors that are present in patients with HIV and active substance use. Despite the wide interindividual variability of trough concentrations, a significant association between substance use and low ATV trough concentrations was observed. These pharmacokinetic differences may also influence viral load suppression during long-term ARV treatment. These findings require additional investigation with more intensive pharmacokinetic assessment to identify individual factors that are contributing to suboptimal ARV exposure in patients with SRDs.