By Harvey S. Bartnof, MD, Staff Physician at AVERI, AIDS Virus Education and Research Institute in San Francisco, California

PART 3:
     Drug Levels Affect Viral Outcome GART Study
     Avoid St. John's Wort with Nevirapine (Viramune)
     Tenofovir DF Levels Increased with Food

Higher Drug Concentrations Associated with Better Virologic Response in "GART" Study

• Greater viral load reductions seen with increased number of "active" drugs on baseline "genotype" resistance tests.
• "Inhibitory Quotients" incorporated drug levels and baseline "phenotype" drug resistance tests.

Previously, the GART (Genotypic Antiviral Resistance Testing) Study (CPCRA 046) had shown that among patients having a detectable viral load with a PI (protease inhibitor) drug combination regimen, baseline genotype resistance testing with "expert" advice about a new regimen led to a significantly greater viral load reduction after 8 and 12 weeks. The additional viral load reduction was approximately 0.5 log copies per milliliter in the GART arm, when compared to the no-GART arm and was highly significantly different. Now, GART researchers have reported that the number of active drugs in the new regimen and the blood levels of those drugs, in addition to a higher "inhibitory quotient" were all significantly associated with greater viral load reductions. The report was presented by Richard M. W. Hoetelmans, PharmD of Virco, and the lead author was Douglas Mayers, MD of Henry Ford Hospital in Detroit, Michigan.

A total of 153 patients were enrolled in GART. Random (untimed) drug levels in blood plasma (no cells) were available for 147 of them (96%) at 4, 8, and 12 weeks. Drug levels at week 12 were classified as greater than or less than the median (average) reference level of the group using measurements at weeks 4, 8, and 12. (Therefore drug levels were not reported as "therapeutic" ["optimal"] or "sub-therapeutic," using a standardized, laboratory reference level.) The inhibitory quotient ("IQ") for each drug was used at week 12 to relate the random drug level concentration to the "fold-change in IC50" (concentration to inhibit 50% of the patient's HIV isolate at baseline, compared to wild-type, i.e., "phenotypic" drug resistance). (The IQ incorporates the ratio of the measured drug levels in patients, usually the minimum or lowest concentration to baseline drug resistance information from the same patient's HIV isolate.) Drug levels were measured using the "Virco PK" test. Phenotype drug resistance was measured using Virco's Antivirogram test. Genotype drug resistance was measured by mutations found with "ABI" testing and expert virologist interpretation. Therefore, the number of "active" drugs was determined by the expert virologist, with active referring to a baseline genotype that was "sensitive" or "possibly resistant" to each drug.

The baseline median CD4 count was approximately 230 cells per microliter with a viral load of 4.4 log (28,085) copies per milliliter. Approximately half had previously been on only one anti-HIV drug regimen with 1 PI (protease inhibitor) and 2 NRTI (nucleoside analog) drugs. A detectable HIV RNA at baseline was defined as at least a 3-fold increase from the previous baseline, for at least 16 weeks. Patients had been randomized to GART (78 patients) or no-GART (75 patients). Prior to randomization, the PI drug was indinavir (Crixivan) for 54%, nelfinavir (Viracept) for 33%, ritonavir (Norvir) for 7% and saquinavir (Fortovase/Invirase) for 6%. Only 10% previously had experience with an NNRTI (non-nucleoside) drug. At the time of randomization, 109 patients (71%) received an NNRTI drug as a part of their new regimen.

The results were as follows. The first statistical "multivariate" analysis used baseline genotype resistance information and blood levels of drugs at week 12. After adjusting for baseline viral load, a significantly greater viral load reduction (-0.5 log copies per milliliter) occurred for each active drug that was measured at a level greater than the median (p=0.0001). Interestingly, the viral load reduction for each active drug with a level less than the median also was significant  (-0.23 log copies per milliliter, p=0.03). Yet, when analyzed by drug class, much of those viral reductions appeared to be due to an NNRTI drug in the regimen. After adjusting for baseline viral load and other drug classes in the regimen, the mean viral load decrease was significantly greater for those whose regimen included an active NNRTI drug with a level greater than the median (-0.63 log, p=0.001) than those whose regimen included an active PI drug with a level greater than the median (-0.19 log), p=0.26).

When the results were presented by IQ (see above), there was a significantly greater and proportional viral load reduction for each drug that had a 12-week IQ greater than the median (p=0.01). Specifically, for only one drug with an IQ greater than the median, the viral load reduction at 12 weeks was 0.6 log, 2 drugs with an IQ greater than the median (-1.0 log), 3 drugs (-0.9 log), and 4 or 5 drugs (-1.4 log), see table below.

Average Change in Viral Load at Week 12 According to Number of Drugs with Inhibitory Quotient (IQ) * Greater than the Median

*Inhibitory quotient (IQ) in the current study was defined as a random, untimed blood level of each anti-HIV drug at week 12 divided by the "fold-change" (phenotype resistance) of the patient's HIV isolate ("IC50") at baseline (see report for additional explanation). A p-value less than 0.05 is statistically significant.

Dr. Hoetelmans concluded that the short-term virologic response in GART was determined by "both the number of active drugs and the drug levels for each drug in the new regimen." Also, an improved virologic response was associated with "increased levels of random, untimed drug concentrations in [blood] plasma." The association held whether genotype drug resistance or phenotype drug resistance tests were used. In addition, a "maximal virologic response was obtained among patients taking multiple 'active' drugs with levels in the upper 50% of the reference distribution." Lastly, a significantly greater viral load reduction at week 12 occurred "if one or more drugs were used with an IQ greater than the median," where IQ is defined as a random "drug concentration [divided by the] fold-change in IC50 at baseline."

There were some limitations to the study's results and conclusions.

First, 25% of patients did not receive any active drugs. Second, the NNRTI drug analysis did not incorporate "protein binding," and this class of anti-HIV drugs is highly protein bound (only "unbound" drug is active and free to move into cells to block HIV growth). Third, only 71% received an NNRTI drug. Fourth, the results would not apply to the NRTI drugs, whose active "triple phosphate" component must be measured inside blood immune cells. Fifth, the results did not show a significant association between antiviral efficacy and levels of PI drugs, only NNRTI drugs. Sixth, the "random" (untimed) nature of the blood tests might limit the potential applicability to other populations (most other studies have examined the minimum or lowest blood concentration). Seventh, 12 weeks of follow-up is relatively short, although this is an accepted limitation of the GART study. Eighth, this is a retrospective (after-the-fact) analysis and not prospective (the previously published genotype testing/viral load results part was prospective). Last, this reviewer would have preferred to see IQ results using drug levels obtained at 4 weeks, from the perspective of potential clinical applicability and intervention in other patient populations. A 12-week measurement of anti-HIV drugs in blood might be rather late to institute a dosing increase, if levels were found to be low. Ideally, one would want to intervene no later than 4 weeks to potentially increase drug dosing if levels were low, before new resistance could develop after an additional 8 weeks. However, this study was neither designed to show nor did it prove that intervening with a dose modification if levels were low might increase blood levels of drugs and therefore virologic response. Also, the results might have been similar if 4-week drug levels alone had been used.

Nonetheless, the new findings about drug levels in the GART study add to the number of retrospective studies that find a significant association between blood levels of anti-HIV drugs and virologic response. Also, the new findings make GART somewhat similar to the VIRADAPT study in which greater viral load reductions at 6 months were significantly associated with "optimal" PI drug trough (lowest level) concentrations and incorporating baseline genotypic resistance testing as part of devising a new drug regimen.

References
Baxter JD and others. A randomized study of antiretroviral management based on plasma genotypic antiretroviral resistance testing in patients failing therapy. CPCRA 046 Study Team for the Terry Beirn Community Programs for Clinical Research (CPCRA) on AIDS. AIDS 2000 June 16;14(9):F83-93

Clevenbergh P and others. Persisting long-term benefit of genotype-guided treatment for HIV-infected patients failing HAART. The Viradapt Study: week 48 follow-up. Antiviral Therapy 2000 March;5(1):65-70.

Durant J and others. Importance of protease inhibitor plasma levels in HIV-infected patients treated with genotypic-guided therapy: pharmacological data from the Viradapt Study.AIDS. 2000 July 7;14(10):1333-9.

Mayers D, Hoetelmans RMW and others. Both antiretroviral drug levels and drug resistance are associated with short-term virologic responses to subsequent drug regimens in CPCRA 046. Abstract and oral presentation 5.2 at the 2nd International Workshop of Clinical Pharmacology of HIV Therapy; April 2-4, 2001; Noordwijk, the Netherlands.

St. John's Wort (SJW) for Depression Significantly Decreases Nevirapine Levels in Blood

• Researchers recommend against combining the two drugs.
• Caution might be needed about combining SJW with any NNRTI or PI drug.

At last year's 1st International Workshop on Clinical Pharmacology of HIV Therapy, Stephen C. Piscitelli, PharmD, then at the NIH, reported that St. John's wort significantly decreased blood levels of indinavir (Crixivan, PI drug), most likely to "induction" of the liver's "P450 CYP3A4 isoform" (enzyme). St. John's wort (herpericum perforatum) is an over-the-counter herbal product that has benefits against clinical depression. It has also been shown in HIV negative persons to decrease blood levels of oral contraceptives ("birth control pills"), theophylline (for asthma), and warfarin (Coumadin blood "thinner").

Now, Monique M. R. de Maat, PharmD, of Slotervaart Hospital in Amsterdam, the Netherlands has reported that St. John's wort significantly increases the "clearance" of nevirapine (Viramune, NNRTI drug) by 35%, thereby decreasing blood "exposure" (concentration) of the drug. The "full PK" (continuous pharmacokinetic, blood concentration) curves of nevirapine from 13 patients plus 758 single time-point nevirapine concentrations led to a database of 1,330 concentration measurements. Five HIV patients taking nevirapine had at least one blood test each while they were and were not taking concomitant St. John's wort. Statistical modeling was used to construct the "PK" (pharmacokinetics) analyses. The authors concluded, "Because a low exposure to nevirapine may enhance the development of viral resistance and treatment failure, the concomitant use of nevirapine and St. John's wort should be avoided."

Also, until proven otherwise, it would be reasonable not to take St. John's wort (SJW) while taking anti-HIV therapy with either an NNRTI or PI drug. It is quite possible and somewhat likely that similar negative interactions would occur with other drugs in those drug classes. Ideally, the respective pharmaceutical companies of each of the other PI and other NNRTI drugs would perform studies or provide funding to determine whether similar negative interactions occur when SJW is combined with their respective NNRTI or PI drug(s).

Interestingly, a potent prescription antidepressant medication (venlafaxine, Effexor) was shown to significantly decrease indinavir (Crixivan, PI drug) blood levels at the 1999 ICAAC (Interscience Conference on Antimicrobial Agents and Chemotherapy). Venlafaxine has effects by modifying levels of several naturally occurring "neurotransmitters" in the brain. All of these findings underscore the potential for negative drug interactions when combining prescription and even some over-the-counter medications with anti-HIV medications.

References
De Maat MMR and others. A population pharmacokinetic model of nevirapine reveals drug interaction with St. John's wort. Abstract and oral presentation 1.2 at the 2nd International Workshop of Clinical Pharmacology of HIV Therapy; April 2-4, 2001; Noordwijk, the Netherlands.

Levin GM and others. Venlafaxine [Effexor] and indinavir: results of pharmacokinetic interaction study. Abstract 661 at the 39th Interscience Conference On Antimicrobial Agents and Chemotherapy (ICAAC). September 26-29, 1999, San Francisco, California.

Moore LB and others. St. John's wort induces hepatic drug metabolism through activation of the pregnane X receptor. Proceedings of the National Academy of Science U S A. 2000 June 20;97(13):7500-2.

Piscitelli SC and others. Indinavir concentrations and St. John's wort. Lancet 2000;355:547-548.

(no author listed). Patients being treated for HIV should avoid St. John's Wort: NIH. Journal of the American Dental Association. 2000 April;131(4):439, 442.

(no author listed). St. John's Wort and HAART. AIDS Patient Care STDS. 2000 May;14(5):281.

Tenofovir DF Absorption Increased with Fat Meal
The pharmacokinetics (drug concentrations) of tenofovir DF (PMPA) in healthy volunteers were reported by Dr. B. P. Kearney of Gilead Sciences. Tenofovir is an experimental nucleotide analog drug that has activity against HIV and HBV (hepatitis B virus), with one-daily dosing. A total of 40 volunteers (50% women, 17% non-Caucasian) took 300 mg tenofovir DF after an overnight fast or with a meal that was 50% fat by calories. The results showed that food significantly increased the total drug exposure ("area-under-the-curve") by 40%, while the maximal concentration insignificantly increased by 14%. The minimal concentration also was increased somewhat with food on the graph, but the exact percentage was not reported. The 300 mg tablet was shown to be equivalent to four 75 mg tablets. The findings confirm that which has been found in HIV positive patients. Gilead recently opened an Expanded Access program for tenofovir DF for advanced HIV patients with limited treatment options. Physicians may call 1-800-GILEAD-5.

A recent study of tenofovir DF 300 mg or placebo added to an existing anti-HIV regimen in 552 highly treatment-experienced patients with a detectable viral load at baseline showed significant virologic benefits after 24 weeks (Study 907). At baseline, there was a mean 5 years of prior anti-HIV drug experience with a very common finding of multi-drug resistance. The mean baseline CD4 count was 427 cells per microliter, with a viral load of 3.4 log (2,290) copies per milliliter. The results showed that after 24 weeks, 45% of the tenofovir arm achieved an undetectable load (limit 400 copies per milliliter) compared to 13% in the placebo, a significant difference. Also, 22% of the tenofovir arm achieved less than 50 copies per milliliter, compared to 1% in the placebo arm, a significant difference. The mean viral load reduction for tenofovir patients was -0.6 log copies per milliliter. The CD4 count increase was not reported. Adverse events were reported to be similar in both treatment arms, and the discontinuation rate was approximately 6% in both arms.

References
Chamberlain LJ. Tenofovir decreases HIV RNA significantly. Infectious Disease News 2001 March;14(3):23.

Kearney BP and others. Effect of formulation and food on the pharmacokinetics of tenofovir DF. Abstract and poster 3.18 at the 2nd International Workshop of Clinical Pharmacology of HIV Therapy; April 2-4, 2001; Noordwijk, the Netherlands.