NNRTI and PI drug plasma levels are as critical as drug sensitivity in achieving treatment success in antiretroviral experienced patients

Simon Collins, HIV i-Base

The importance of achieving therapeutic levels of PIs and NNRTIs was raised in several studies. Previous presentations from key resistance studies GART and VIRADAPT showed in 1999 and 2000 that both resistance and drug levels were essential to optimise treatment success. [1, 2]

An updated analysis of the role of each factor from the GART study was presented at this resistance meeting and the results again conclusively showed the critical importance of both factors. [3]

In the GART study,153 patients with > 3-fold rise in HIV RNA on a PI + 2 RTI combination for over 16 weeks were randomised to receive genotype report plus recommendations or no genotype results. The PK study included measuring plasma drug levels from random (untimed) samples for 147 patients (using VircoPK assay) for all drugs, including RTIs, at baseline and at weeks 4, 8 and 12. These levels were used for the reference distribution and drug levels at week 12 were used to determine individual patients as either >median or < median. Inhibitory quotients (IQ) were calculated at ‘drug level at week 12 / fold change of IC50 at baseline’ and were also classified as < or > median for the group.

Genotype and phenotype results were also used to define the number of active and inactive drugs for each patient. Results calculated as change from baseline to week 12 in multivariate analysis and relationship to drug levels (DL) are shown below.

Table1: multivariate analysis by genotype and drug level

Average change (log) p value
Active drug + DL > median -0.48 <0.002
Active drug + DL < median -0.22 0.04
Inactive drug + DL> median -0.27 0.07
Inactive  drug + DL < median -0.02 0.87

Table 2: multivariate analysis by phenotype and drug level

Average change (log) p value
Active drug + DL > median -0.46 <0.0001
Active drug + DL < median -0.15 0.13
Inactive drug + DL > median -0.25 0.23
Inactive drug + DL < median -0.05 0.78

Sub-optimal levels of active drugs produce a reduced antiviral effect than higher concentrations of inactive drugs – reaching statistical significance in the genotypic analysis. Looking at individual drug classes, produced more significant results for NNRTIs than PIs (p=0.004 vs 0.08 respectively). As would be expected with nucleosides, there was no predictive value from drug levels and active drugs with antiviral activity (p=0.54).

A clear relationship was also found to the average change in viral load at week 12 according to the number of drugs with mean IQ > median. For the whole cohort, the additional viral load change is -0.28 log (p<0.0001) for each drug with median IQ >median (compared to IQ < median ) .

Table 3: IQ > median and viral load change

No of drugs in regimenwith IQ>median n Average change in viral load
0 29 -0.18 log
1 40 -0.58
2 35 -1.06
3 24 -0.85 (p<0.05)
4 or 5 9 -1.44

This was a complicated and thorough analysis. It showed once again the importance of adequate drug levels in order to generate an optimal treatment response. Even though the differences produced may appear small, they can be particularly significant in a salvage setting where any contribution to viral suppression is likely to contribute to duration of response.

In a separate open label study in 32 NNRTI-naïve, PI- and RTI-experienced subjects, Piketty and colleagues looked at drugs levels and resistance when treated with a salvage treatment of SQV (1000mg) / RTV (100mg) BID with efavirenz and two recycled RTIs. [3]

Pre-dose trough levels were measured for efavirenz, ritonavir and saquinavir and three-hour post dose peak levels for saquinavir . As with all TDM studies, interpatient variability for these levels shown below varied considerably, with lowest levels of saquinavir barely detectable and upper levels often 10-fold higher than the median. Variation was less marked for efavirenz but nevertheless ranged from <30% – >300% of the optimal levels of 2.5mg/l.

See Table 4 below for results.

In a multivariate analysis plasma exposure, and baseline phenotypic resistance, were the best predictors for virological response at week 48.

Table 4: saquinavir results

Odds ratio [CI 95] p value
Weight 1.13 [1.01 – 1.26] 0.04
Cmax 9.30 [0.72 – 120.05] 0.08
Log SQV resistance index 9.87 [1.19 – 82.10] 0.03

These results suggested further study for the utility of TDM, especially in patients with higher body weight.


  1. VIRADAPT – Therapeutic Drug Monitoring (TDM) of PIs and optimal response in naive and experienced patients – DrFax 76, October 1999
  2. GART – Added value of TDM – DrFax 83 Feb 2000.
  3. Mayers D et al – Both drug resistance and antiretroviral drug levels are associated with short term virological responses to subsequent drug regimens in CPCRA 046 (GART Study). Abstract 87. Antiviral Therapy 2001; 6 (Supplement1):67
  4. Piketty C et al – Plasma drug levels and resistance are predictors of efficacy for the combination including saquinavir, ritonavir and efavirenz. Abstract 120. Antiviral Therapy 2001; 6 (Supplement1):89

Links to other websites are current at date of posting but not maintained.