In this prospective study of healthy volunteers, we found that 14 days of co-administration of AG did not significantly impact IDV PK. We also found that, although IDV acutely reduced insulin sensitivity by an average of 15%, AG did not change insulin sensitivity in IDV-treated healthy volunteers, providing evidence against its use in the treatment of PI-induced disorders of glucose metabolism.
Potential drug-herb interactions are an important safety concern for many clinicians and HIV-infected patients. Although herbal remedies are perceived as safe by many patients, some can inhibit and/or induce the CYP3A4 enzyme, the main metabolic pathway for most PIs and non-nucleoside reverse transcriptase inhibitors (NNRTIs), potentially leading to increased toxicity or therapeutic failure [15, 16]. The metabolic pathways of ginseng components are not well known. Ginsenosides, which are steroid-like molecules of the saponin class, are thought to be the active compounds of all ginseng species . There are more than 20 different types of ginsenosides in ginseng root and their relative concentrations varies depending on the species, the batch, and the part of the plant assayed (eg. root vs berry)[23, 24]. The ginsenosides also differ in their effects on metabolism. In in vitro models investigating the catalytic activity of c-DNA expressed CYP450 isoforms, the ginsenoside, Rd, has been shown to be a weak inhibitor of CYP3A4, while Rf increased CYP3A4 activity by 54% . In addition, metabolites of ginsenosides as well as non-ginsenoside components of ginseng have also been shown to inhibit CYP3A4 in experimental models [26, 27]. The clinical significance of these in vitro observations is not clear.
In previous human studies, Siberian ginseng (Eleutherococcus senticosus) has been shown to increase concentrations of nifedipine , a CYP3A4 substrate, but does not affect concentrations of midazolam , another CYP3A4 probe drug. It should be noted that Siberian ginseng is not a species of the genus Panax, contains no ginsenosides, and therefore the effects of Siberian ginseng are not generalizable to members of the Panax genus, such as AG. However, human studies in healthy volunteers using Panax ginseng (C.A. Meyer) have shown no interaction with midazolam  and no change in urinary 6-beta-OH-cortisol/cortisol ratio , both measures of CYP3A4 activity.
We also found that 3 days of IDV administration decreased insulin sensitivity by 15% in healthy volunteers. Our findings are similar to another healthy volunteer study, showing a 17% decrease in insulin sensitivity with IDV administration (800 mg q 8 hours), but smaller in magnitude when compared to another study using a higher dose (34% decrease with a single dose of 1200 mg) . In in vitro studies and animal models, PIs, such as IDV, impede glucose movement through the major glucose transporter in skeletal muscle, GLUT4, thereby inducing insulin resistance. The observations in humans, including ours, are consistent with this mechanism, although the extent to which it is clinically relevant in the pathogenesis of hyperglycemia among HIV-infected, HAART-treated patients remains unclear.
Although we observed that insulin sensitivity decreased with IDV administration, we did not find any change in insulin sensitivity with co-administration of AG. In previous studies, Vuksan, et al. showed that AG administration immediately prior to an oral glucose tolerance test was associated with 20% decrease in glucose AUC in both patients with type 2 diabetes mellitus and healthy volunteers , but this effect appears to be dependent on the batch of ginseng used . To rule out an inert batch of AG, we conducted a bioassay of our batch using an ob/ob mouse model and a hypoglycemic effect was observed, suggesting that our batch possessed some active components.
The mechanism underlying the previously observed effects of AG on glucose metabolism remains unknown and may be multifactorial. The modulation of digestion and enhancement of insulin secretion have been proposed based on findings in some animals models [36, 37], but ginseng administration has also been shown to improve insulin sensitivity in ob/ob mice by more than two-fold . It has been postulated that ginsenosides may intercalate into the cellular plasma membrane, thus modulating the cell signaling, electrolyte transport, and receptor binding . It is not known whether this effect could also modulate the effect of IDV on the glucose flux through the GLUT4 transporter.
Another factor that may have contributed to the lack of effect is the variability of IDV plasma concentrations between the two clamps, which has been previously observed . The effect of AG on IDV-induced insulin resistance would be best determined if IDV plasma concentrations were the same when it was given alone and when it was co-administered with AG. For this reason, we normalized the measure of insulin sensitivity for drug concentration in a post hoc analysis, by dividing M/I by the IDV concentration during the steady state portion of the clamp and found that this ratio was significantly higher in the IDV + AG condition.
Although the interpretation of ratios can be challenging  and should be done with caution, one explanation of this finding is that insulin sensitivity per unit of IDV concentration modestly improved with the administration of AG. Given the known effect of IDV on GLUT4 blockade, AG components may work directly at the plasma membrane to allow glucose to enter cells, either by improving glucose movement through GLUT4, increasing the concentration of GLUT4 in the plasma membrane, or facilitating glucose entry through alternate pathways. Another possible mechanism of AGs effect may be through the enhancement of local blood flow. Increased capillary recruitment mediated though nitric oxide is an important mechanism of increasing glucose and insulin delivery to skeletal muscle [40, 41]. IDV has been shown to cause endothelial dysfunction in healthy volunteers, likely by reducing nitric oxide production . Ginseng species have been shown in experimental models to increase nitric oxide production [43, 44] and therefore, may effect insulin sensitivity by this mechanism. This hypothesis requires further investigation.
Our study had additional limitations which may have implications for the generalizability of its findings. Although both men and women were eligible for participation, only men enrolled. In previous studies, inducibility of CYP3A4 by herbal compounds has shown an interaction by sex, whereby women showed a 74% increased effect of St. John's wort on CYP3A4 activity using a midazolam probe compared to men . In the same study, however, Panax ginseng showed no effect on the midazolam metabolism in either men or women. Further PK interaction studies in women may be necessary. In addition, our population was mostly African-American. Although there is evidence for genotypic differences in CYP3A4 between Caucasians and African-Americans, phenotypic differences in CYP3A4 activity have not been found . Finally, although we quantified the amount of common ginsenosides, there is substantial variability of composition even within species. As a result, as is the case in all botanical research using non-standardized complex compounds whose active component is unknown, the extent to which our findings are generalizable to other AG products is not clear.
In conclusion, this study in healthy volunteers found no evidence of a significant PK interaction between AG and IDV. Because CYP3A4 is the principal metabolic pathway of most HIV PIs and NNRTIs, significant drug-herb interactions with these antiretrovirals are unlikely. The metabolic effects of AG are more difficult to interpret. Although IDV significantly reduced insulin sensitivity, there was no change in insulin sensitivity with AG administration. However, after normalization for IDV concentrations, insulin sensitivity improved in the AG condition, leaving open the possibility of a modest effect on glucose metabolism. Further studies to clarify this question should attempt to reduce variability in IDV plasma concentrations, either by giving multiple doses of IDV at shorter intervals to maintain plasma IDV concentrations constant while insulin sensitivity is assessed, or using an IDV regimen boosted with ritonavir. Until these issues are resolved, there is no clear scientific basis to recommend AG for the treatment of glucose abnormalities in HIV-infected patients.