To the Editor:
Lipid level elevations are among the most frequently described adverse events in patients receiving lopinavir/ritonavir (LPV/r), especially in antiretroviral-experienced populations where up to 40% of patients treated with this drug have been found to develop grade 3 or higher lipid level elevations. 1 The exact mechanism responsible for dyslipidemia associated with LPV/r is unknown. A number of studies have found a relationship between LPV plasma levels and lipid abnormalities during salvage therapy with LPV/r. 2–5 In a prospective study of treatment-experienced patients who started therapy with LPV/r, we 3 found an association between trough LPV plasma concentrations and lipid changes, suggesting that patients with higher LPV plasma concentrations may be at greater risk of dyslipidemia. In contrast to previous studies, Torti et al 6 failed to demonstrate an association between lipid levels and LPV plasma concentrations in an analysis of patients who had started LPV/r treatment. They suggest that the influence of LPV on lipid variations is caused by an intrinsic pharmacodynamic drug effect independent of pharmacokinetic variability. A number of important differences between the study by Torti et al 6 and previous studies should be considered before any conclusion can be drawn.
First, the study populations may be different. Previous studies showing an association between LPV plasma levels and lipid abnormalities usually included antiretroviral-experienced patients. 2–5 We 3 studied a population of heavily pre-treated patients with advanced HIV disease who were enrolled in an LPV/r expanded access program. More than one half of the subjects had clinically evident lipodystrophy and had lipid abnormalities at baseline. During therapy with LPV/r, 36% and 16% of patients developed grade 2 and grade 3 or higher lipid level elevations, respectively, and one third had to start therapy with lipid-lowering drugs. Unfortunately, information concerning baseline characteristics of the subjects studied is lacking in the report by Torti et al 6; however, it is remarkable that only 11% of their patients developed grade 2 hyperlipidemia, and apparently no cases of grade 3 or higher lipid level elevations were observed. The incidence of lipid abnormalities found in that study was, therefore, much lower than that found in other studies evaluating the relationship between LPV plasma levels and lipid changes 2–5 and that reported across phase 2/3 clinical trials with LPV/r. 1
Second, our investigation 3 was specifically designed to address the hypothesis that patients who develop metabolic abnormalities during treatment with LPV/r might be those with higher plasma exposure to LPV. Therefore, we undertook a 48-week prospective study with systematic clinical, laboratory, and pharmacokinetic evaluations and predefined strict criteria. Because protease inhibitors exhibit a wide interpatient and intrapatient pharmacokinetic variability, to have a more reliable measure of overall drug exposure, LPV plasma concentrations (and simultaneously lipid levels) were measured several times throughout the study period (weeks 4, 8, 12, 24, 36, and 48 [6 times in total]). In contrast, the report by Torti et al 6 provides results from an analysis of selected patients who had started LPV/r–containing regimens and had “triglyceride and total cholesterol levels available at baseline and at 2 points of the follow-up up to week 12.” Trough LPV plasma concentrations were determined 1, 4, and 12 weeks after initiation of LPV/r treatment.
Third, the statistical analysis was different. In our study, 3 LPV plasma concentrations for each patient were summarized by computation of the means of the plasma levels measured in the same patient at different time points of the study. We used the Pearson correlation coefficient to evaluate the correlations between pharmacokinetic and lipid parameters at different time points during the study. In the study by Torti et al, 6 however, to evaluate the possible relationships between trough LPV levels and lipid concentrations, area-under-the-curve values were used as synthesis measures of the trough levels measured in the same patient, even though very few measurements might have been present in some cases according to the selection criteria described. Although the area under the curve could be a good indicator of drug exposure when several measurements are used, the analysis could be flawed for 2 related reasons: the few measurements taken and the lack of statistical power. Estimating overall drug exposure through ≤3 trough plasma concentrations might have been inaccurate in some patients and may give a high level of variability. Furthermore, triglyceride and cholesterol levels were available only at baseline and at any 2 points of the 12-week observation period. Altogether, the variability in LPV exposure measurement, the short follow-up period to detect the effect, and the additional variability in triglyceride and cholesterol measurements diminish the statistical power of the study by Torti et al 6 to identify a correlation between lipid abnormalities and LPV plasma concentrations.
An important finding from our study 3 was that those patients with grade 3 or higher lipid level elevations had higher trough LPV concentrations than did subjects who did not develop them. Indeed, most patients developing significant dyslipidemia had trough LPV concentrations of >8 μg/mL. In contrast, none of the patients studied by Torti et al 6 developed grade 3 hyperlipidemia, despite the fact that apparently 73% of subjects had trough LPV concentrations of >8 μg/mL. According to their interpretation, this finding supports the lack of association between LPV plasma levels above a predefined cutoff and elevation of lipid levels. A note of caution should be cast when comparing results of plasma concentrations measured by nonstandardized procedures in different laboratories and at different time points. All plasma concentrations of LPV in our study were determined at steady state. It should be noted that 40% of values of >8 μg/mL found by Torti et al 6 occurred at week 1, a time point at which plasma levels may have not reached the steady state in all patients. 7 Therefore, values obtained at week 1 may be misleading and should be interpreted with caution. Furthermore, even if the results of LPV plasma concentrations were comparable, LPV plasma levels of >8 μg/mL would not mean that hyperlipidemia had necessarily developed. Available data indicate only that patients with higher trough LPV concentrations may be at greater risk of dyslipidemia. The association found in previous studies does not imply that the elevation of lipid levels is just a direct effect of LPV/r. Dyslipidemia probably results from a complex interaction between different factors, including drug exposure. In fact, some of the patients who had high trough LPV concentrations in our study 3 did not develop hyperlipidemia.
The relationship between drug levels and toxicity has always been complex and difficult to prove. However, the inability to find a clear-cut association in observational studies should not be considered as evidence of the lack of such a relationship. Methodological issues and confounding factors should be taken into consideration. Only appropriate, well-designed studies can prove the plausible hypothesis of the association between drug exposure and toxicity linked to antiretroviral drugs. Assuming that a relationship between LPV drug exposure and lipid abnormalities probably exists, it may not be easy to unveil in an observational study. Several factors, including the magnitude of changes in lipid levels and the accuracy of the measures of drug exposure used to minimize intrapatient variability, may influence the ability to disclose an association. The reduced incidence of hyperlipidemia among the population analyzed by Torti et al 6 and the limited measurements of plasma drug concentrations and lipid levels during the observation period may have been potential caveats of that study.
Finally, we share the concern expressed by Torti et al 6 on LPV/r dose reduction based on plasma drug concentrations to prevent or treat abnormal lipid level elevations. Because most patients who developed significant dyslipidemia in our study 3 had high trough LPV concentrations, we suggested that those patients may benefit from therapeutic drug monitoring, but we did not recommend a straightforward LPV/r dose reduction. Moreover, in our report, 3 we pointed out the need for further studies to assess whether individualization of the LPV dose regimen may result in a decrease in the percentage of patients developing metabolic abnormalities associated with LPV/r–containing regimens with no loss of efficacy.
*Infectious Diseases Unit Internal Medicine Department Hospital General Universitario de Elche Alicante, Spain, †Pharmacy Service Hospital General Universitario de Elche Alicante, Spain, ‡Public Health Department University Miguel Hernández Alicante, Spain
1. Bernstein B, Moseley J, King M, et al. Safety of Kaletra: data from phase II and phase III clinical trials. 1st International AIDS Society Conference on HIV Pathogenesis and Treatment, Buenos Aires, Argentina, July 2001.
2. González de Requena D, Blanco F, García Benayas T, et al. Correlation between lopinavir plasma levels and lipid abnormalities in patients taking lopinavir/ritonavir. AIDs Patient Care STDS
3. Gutiérrez F, Padilla S, Navarro A, et al. Lopinavir plasma concentrations and changes in lipid levels during salvage therapy with lopinavir/ritonavir-containing regimens. J Acquir Immune Defic Syndr
4. Tseng A, Loutfy M, Phillips E, et al. The relationship between lipid elevations and lopinavir concentrations in HIV-infected patients on LPV/r-containing salvage regimens [abstract H-1916]. 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, CA, September 2002.
5. Calboreanu A, Dimet J, Treluyer JM, et al. Relation between lopinavir plasma blood concentrations and hypertriglyceridemia in naive and protease inhibitors-experienced patients. 10th Conference on Retroviruses and Opportunistic Infections, Boston, MA, February 2003.
6. Torti C, Quiros-Roldan E, Regazzi-Bonora M, et al. Lipid abnormalities in HIV-infected patients are not correlated with lopinavir plasma concentrations. J Acquir Immune Defic Syndr
7. Corbett AH, Lim ML, Kashuba DM. Kaletra (lopinavir/ritonavir). Ann Pharmacother