点击显示 收起
1 From the Departments of Medicine (AT-G, HNG, and LB) and Pediatrics (SH and RR) and the General Clinical Research Center (WK), Columbia University, New York, NY; the Department of Medicine (JA), St Luke's-Roosevelt Medical Center, Division of Infectious Disease (SR, RM, and WME-S), Harlem Hospital Center and Columbia University, New York, NY; the Department of Medicine, University of California Davis, Davis, CA (BO and LB); and the VA Northern California Health Care System (LB)
2 Supported by grants 65938 (principal investigator: LB) and 65954 (principal investigator: N Schachter) from the National Heart, Lung and Blood Institute; by the Columbia University General Clinical Research Center (RR00645); by the UC Davis Clinical Nutrition Research Unit (DK35747); and by the UC Davis General Clinical Research Center (RR019975). 3 Address reprint requests to L Berglund, Department of Medicine, University of California, Davis, UCD Medical Center, 2921 Stockton Boulevard, Suite 1400, Sacramento, CA 95817. E-mail: lars.berglund{at}ucdmc.ucdavis.edu.
ABSTRACT
Background: Features of the dyslipidemic pattern reported with the use of antiretroviral therapy predict enhanced postprandial lipemia, which is an emerging cardiovascular disease risk factor.
Objective: We evaluated the postprandial response to a physiologic, meal-based challenge in HIV-positive subjects without hyperlipidemia.
Design: We measured hourly lipid, lipoprotein, glucose, and insulin concentrations during a 13-h period in 25 nonwhite patients (13 women, 12 men): 13 receiving a protease inhibitor (PI)-based regimen (6 nelfinavir and 7 indinavir) and 12 receiving a nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimen (6 efavirenz and 6 nevirapine).
Results: Mean fasting HDL-cholesterol concentrations were lower in HIV patients than in healthy subjects without HIV infection matched for age, sex, and ethnicity (z score: –0.81 ± 0.9; P = 0.0001). Fasting triacylglycerol concentrations were not significantly different between HIV-infected patients and healthy subjects but were higher in PI-treated than in NNRTI-treated patients [median (interquartile range): 144 (110–191) and 89 (62–135) mg/dL; P = 0.007]. Average daylong triacylglycerol concentrations, but not incremental concentrations, were higher in the PI group than in the NNRTI group [205% (185–248%) and 125% (78–191%); P < 0.05]. For all HIV-positive patients, the fractional triacylglycerol increase was lower after breakfast than after lunch (20 ± 18% and 42 ± 40%, respectively; P < 0.04). Insulin concentrations were higher in PI-treated than in NNRTI-treated patients [22.6 (13.1–29.8) and 11.8 (7.1–19.1) µU/mL; P = 0.01] and increased in both groups in response to each meal, whereas glucose concentrations increased only after breakfast.
Conclusions: Despite baseline differences, incremental triacylglycerol and insulin responses to a physiologic caloric load among HIV-positive patients were not significantly affected by differences in the type of antiretroviral therapy.
Key Words: HIV • protease inhibitors • nonnucleoside reverse transcriptase inhibitor • NNRTI • antiretroviral treatment • ART • postprandial lipemia • insulin resistance • African Americans
INTRODUCTION
The introduction of potent antiretroviral therapy (ART) represents a major breakthrough in HIV treatment and has resulted in considerable improvements in both morbidity and mortality (1-6). However, it has become evident that the use of ART is associated with significant metabolic side effects, such as body fat redistribution, hyperlipidemia, and insulin resistance (7-11). Although variable lipid and lipoprotein changes have been reported in patients undergoing HIV ART (12-21), the most common lipoprotein pattern reported to date is characterized by decreased HDL-cholesterol concentrations, increased triacylglycerol concentrations, and a modest increase in LDL-cholesterol concentrations (9-11). This lipid pattern has similarities with the lipoprotein phenotype observed in patients with the metabolic syndrome (22, 23). In addition, the metabolic syndrome is characterized by the presence of insulin resistance, a pattern also frequently found during ART (9-11, 14). Studies have shown accelerated atherosclerosis and cardiovascular disease during ART of HIV-infected patients, which agrees with the notion of the metabolic syndrome as an atherogenic phenotype (22-27).
In addition to fasting lipid concentrations, postprandial lipemia has been indicated as an emerging cardiovascular disease risk factor (22, 28). Although fasting lipid concentrations determine hypolipidemic treatment decisions, postprandial lipemia is used primarily as an indicator of risk rather than as a guide to therapy. So far, although there is a wealth of information on fasting lipid concentrations, relatively few studies have focused on postprandial lipemia (28-30), and this condition remains virtually unexplored in HIV ART. Interestingly, several components of the lipid profile that are associated with both the metabolic syndrome and the use of ART in HIV-infected patients, such as decreased HDL cholesterol and increased triacylglycerol concentrations, have been suggested as predictors of the postprandial lipid response (28-32). Furthermore, most of the postprandial studies undertaken to date have used conditions involving a metabolic stress, such as a high-fat load (28-30). Although the postprandial state following physiologic meal intakes represents the norm during daytime, investigations addressing the response to physiologic caloric loads are scarce.
To reflect a physiologic state, we evaluated the response to defined, standardized meals representing a natural food consumption pattern in a cohort of normolipidemic African American and Hispanic HIV-infected patients receiving ART. Because differences in fasting lipid and lipoprotein concentrations have been noted in response to different ART regimens, such as protease inhibitor (PI)-based compared with nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimens (33, 34), we hypothesized that such baseline differences in lipid concentrations would translate into a different postprandial response in patients receiving a PI-based or an NNRTI-based antiretroviral regimen.
SUBJECTS AND METHODS
Patients
HIV-positive African American and Hispanic patients were recruited from outpatient HIV clinics at Harlem Hospital Center in New York. Eligibility for enrollment in the study was based on the presence of documented HIV infection, ongoing stable antiretroviral regimen for >6 mo, and the absence of hyperlipidemia. Patients were excluded if they had a history of diabetes mellitus, had been receiving ART for <6 mo, had consumed appetite stimulants or anabolic agents during the 30 d before enrollment, had hemoglobin concentrations <11 mg/dL, had a history of cirrhosis of the liver, had any opportunistic infections that required treatment in the past 30 d, were pregnant, had fasting cholesterol or triacylglycerol concentrations >200 mg/dL, were receiving hypolipidemic therapy, had Cushing syndrome or untreated hypogonadism, or were being treated with growth hormone, anabolic steroids (including dehydroepiandrosterone), ketoconazole, or corticosteroids. The study was approved by the Institutional Review Boards at Columbia University, Harlem Hospital Center, St Luke's-Roosevelt Medical Center, VA Northern California Health Care System, and University of California, Davis, and informed consent was obtained from all participants.
Overall, 25 patients were recruited for the study: 12 men and 13 women. Twenty-three patients were African American and 2 were Hispanic; 13 patients were undergoing PI-based ART (6 on nelfinavir and 7 on indinavir), and 12 patients were undergoing NNRTI-based ART (6 each on nevirapine and efavirenz). The type of antiretroviral regimen was defined as PI-based for patients taking 2 or more nucleoside reverse transcriptase inhibitors (NRTIs) in combination with at least 1 PI or a regimen containing ritonavir and saquinavir in combination with 1 NRTI and no NNRTIs. NNRTI-based ART regimens were combinations of
The subjects were admitted to the Columbia University General Clinical Research Center (GCRC) in the evening. They continued their respective standard ART regimen throughout the study. After admittance, the patients fasted until the morning breakfast at 0900 the following day. At 0700, an indwelling catheter was placed and kept patent by a saline infusion. The first blood draw was carried out at 0800, which was followed by hourly blood samples until 2000. At meal times, the blood draw was obtained before serving the meal. Meals were served at the following times: breakfast at 0900, lunch at 1200, and dinner at 1700. All meals were prepared by the GCRC Bionutrition Unit as described below. After the blood sample taken at 2000, the catheter was removed and the patients discharged. Because the first meal was given after the 1-h blood sample, we defined the baseline concentrations as the average of the 0-h and 1-h time points.
Comparison with the DELTA Study
Results from HIV-infected patients enrolled in the present study were compared with a previously conducted diet study in HIV-negative subjects, the DELTA (Dietary Effects on Lipoproteins and Thrombogenic Activities) Study (35, 36). As part of that study, 2 separate feeding protocols were carried out. In the second DELTA study protocol, DELTA 2, subjects aged 21–68 y of different ethnicity, including African American and Hispanic, and having characteristics of the metabolic syndrome (low HDL-cholesterol, high triacylglycerol, or high insulin concentrations) were recruited to participate in a multicenter, randomized, double-blind study with a 3-period crossover design. None of the participants took any medications known to affect plasma lipids or hemostatic factors. Eligibility was based on meeting one or more of the following requirements: 1) HDL cholesterol
The diet composition and the distribution of calories over the meals used in the present study are summarized in Table 1. The Bionutrition Unit in the Columbia University GCRC developed menus for 3 different caloric levels (2000, 2200, and 2500 kcal) with the use of standardized recipes. For each meal, the food was weighed before and after it was consumed, and the amounts of calories consumed were calculated. As seen in Table 1, about 90% of the calories provided were consumed, which resulted in a mean caloric intake of 1957 kcal/d. The caloric distribution was designed to provide 25–28% of total energy for breakfast, 35% for lunch, and 37–40% for dinner. The mean relative distribution of calories consumed during the day was 23.1% (breakfast), 39.7% (lunch), and 37.2% (dinner) for the PI group, and 27.3%, 35.8%, and 37.5%, respectively, for the NNRTI group.
View this table:
TABLE 1. Diet composition
The distribution of caloric content was based on the National Cholesterol Education Program (NCEP) Step I recommendations, with 55% of calories from carbohydrates, 15% from protein, and 30% from fat (37). The diets provided an average of 24 g fructose and 18 g fiber per day. The energy requirement of each individual participating in the study was calculated by using the Harris-Benedict equation for basal energy expenditure with an activity factor of 1.3 (38). The food intake of each subject and the menus provided (Table 2) were calculated by using the University of Minnesota (Minneapolis, MN) NUTRITION DATA SYSTEMS software program (version 4.02/30).
View this table:
TABLE 2. Dietary menu components
Laboratory analyses
Immediately after each blood draw, plasma and serum were separated by centrifugation at 30 000 x g for 20 min at 4 °C. The plasma and serum samples were aliquotted and immediately transferred to a –80 °C freezer where they were stored until analyzed. Plasma total, LDL-, and HDL-cholesterol; triacylglycerol; and glucose concentrations were measured by using standard enzymatic techniques as previously described (39). The CVs for these assays were 3–5%. The laboratory participates in the lipid and lipoprotein standardization program of the Centers for Disease Control and Prevention. Serum concentrations of apolipoproteins (apo) A-I and B were determined by immunonephelometric procedures on a Beckman Array 360 nephelometer with commercially available reagents (Beckman, Brea, CA) (39). The interassay CVs for these measurements were 3–5%. Plasma insulin concentrations were measured by using commercially available reagents without cross-reactivity with proinsulin concentrations (Linco Research, St Charles, MO). The interassay CV for this measurement was 6–8% (36).
Statistics
Insulin concentrations were square root transformed and concentrations of triacylglycerols were log transformed to achieve normal distributions before statistical analysis. Normally distributed variables were described as means ± SDs and nonnormally distributed data as medians with interquartile ranges. To compare study subjects with population norms, z scores were calculated for triacylglycerols and HDL cholesterol with adjustment for age, sex, BMI, and race/ethnicity with the use of data from the National Health and Nutrition Examination Survey (NHANES). Means between groups were compared by use of Student's t test or analysis of variance. All statistical analyses were done by using SAS software, version 9.1 (SAS Institute, Cary, NC). Statistical significance was set at P < 0.05.
RESULTS
To explore whether differences in baseline lipoprotein, glucose, and insulin concentrations predicted the postprandial triacylglycerol response in normolipidemic HIV-positive patients, we analyzed these concentrations in relation to data for an HIV-negative population and then evaluated the relation between baseline concentrations and the triacylglycerol excursion level during the day in the HIV-positive subjects. Initially, we compared the fasting, baseline lipid, glucose, and insulin concentrations of the HIV groups with those of a healthy population. We calculated z scores for triacylglycerols and HDL cholesterol with adjustment for age, sex, BMI, and race/ethnicity by using data from NHANES. As shown in Figure 1, the z score for HDL cholesterol was significantly lower for the HIV subjects as a group (P = 0.0001), whereas no significant difference in z score for triacylglycerols was found for the HIV-positive group.
FIGURE 1.. Differences in z scores (±SD) for triacylglycerols (TG) and HDL cholesterol (chol) between healthy control subjects and all HIV-positive patients (n = 25; ), the patients receiving a protease inhibitor (PI)-based antiretroviral therapy regimen (n = 13; ), and the patients receiving a nonnucleoside reverse transcriptase inhibitor–based antiretroviral therapy regimen (n = 12; ). Data for the healthy population, which was matched for age, sex, BMI, and ethnicity, were from the National Health and Nutrition Examination Survey. *Significantly different from zero, P = 0.0001. **Significantly different from the patients receiving PI-based antiretroviral therapy, P = 0.016.
Comparisons of baseline lipid values with those for the healthy population (z score calculations) were also done separately for the PI and NNRTI groups. As shown in Figure 1, although the PI-treated group had a numerically lower z score for HDL cholesterol than did the NNRTI group, the difference between the 2 groups was not significant. For triacylglycerols, there was a significant difference in z scores between PI-treated patients and the NNRTI-treated patients (P = 0.016).
We then evaluated whether baseline and postprandial variables differed between PI-treated and NNRTI-treated HIV-positive patients. The clinical characteristics and fasting baseline lipid concentrations of the 2 groups of subjects are given in Table 3 and are compared with values for HIV-negative subjects with features of the metabolic syndrome from the DELTA Study. As shown in the table, PI-treated patients had a significantly higher mean BMI and fasting insulin and triacylglycerol concentrations than did the NNRTI-treated group. In contrast, total and LDL-cholesterol concentrations and glucose concentrations did not differ significantly at baseline between the 2 groups. Fasting apo A-I and HDL-cholesterol concentrations were higher in the NNRTI group. The mean BMI of 25.8 for the NNRTI group was at the borderline between values defining normal weight and overweight. In contrast, the PI patients had a mean BMI of 29.7, which is at the borderline between values defining overweight and obesity. The PI-treated group had metabolic laboratory values reminiscent of the metabolic syndrome, with triacylglycerol and glucose concentrations approaching the upper limit of the normal distribution and an HDL-cholesterol concentration below the level associated with cardiovascular disease risk (<40 mg/dL). Compared with the DELTA subjects, both HIV-positive groups were older and had lower apo B concentrations. Additionally, whereas the NNRTI group had lower triacylglycerol concentrations and higher HDL-cholesterol concentrations, the PI group had lower apo A-I concentrations and higher glucose and insulin concentrations than did the DELTA subjects (Table 3).
View this table:
TABLE 3. Clinical characteristics of the patients and control subjects1
We related baseline variables to the postprandial triacylglycerol response during the day, expressed as the incremental area under the curve (AUC). As shown in Table 4, among HIV-infected patients there was a significant positive correlation between baseline triacylglycerol, LDL-cholesterol, and apo B concentrations and the triacylglycerol incremental AUC response over the day, whereas HDL-cholesterol concentrations correlated negatively. When the incremental AUC was dichotomized for the postbreakfast (1–4 h) and postlunch (4–9 h) periods, there was no significant association between any of the baseline lipid variables and the postbreakfast response, whereas all lipid variables except for apo A-I were significantly associated with the postlunch incremental AUC response. Baseline glucose and insulin concentrations were not significantly associated with the postprandial triacylglycerol response.
View this table:
TABLE 4. Correlation coefficients between baseline variables and postprandial triacylglycerol (TG) response1
We then compared the difference in triacylglycerol concentrations at baseline and 4 and 8 h postprandially for HIV-positive patients and HIV-negative subjects with features of the metabolic syndrome as described in the Subjects and Methods. Overall, the triacylglycerol increase over 8 h was not significantly different for the 2 groups (Figure 2). However, the distribution of the increase over the day differed substantially. Compared with HIV-negative subjects, HIV-positive subjects had a lower fractional triacylglycerol increase after breakfast (20 ± 18% compared with 70 ± 42%, respectively; P < 0.0001) but a higher fractional triacylglycerol increase after lunch (42 ± 40% compared with –9 ± 27%, respectively; P < 0.0001). These results suggest a later triacylglycerol response in the ART-treated HIV-positive subjects than in the HIV-negative subjects. To assess possible effects of age, we performed a subgroup analysis by only including subjects from both studies within an overlapping age range (30–60 y); the mean ages of the subgroups were 39 ± 7 y (n = 42) for the DELTA subjects and 41 ± 6 y (n = 23) for the HIV-positive patients. The P values for the triacylglycerol AUC for the time periods 0–4 and 0–8 h remained <0.0001, as they were for the full groups.
FIGURE 2.. Mean (±SD) changes in triacylglycerol concentrations after breakfast and lunch in HIV-positive patients (n = 25; ) and in HIV-negative subjects with features of the metabolic syndrome (n = 68; ). The HIV-negative subjects were part of the DELTA (Dietary Effects on Lipoproteins and Thrombogenic Activities) Study. Meals were provided after 1 h (breakfast), 4 h (lunch), and 9 h (dinner). Compared with HIV-negative subjects, HIV-positive subjects had a lower fractional triacylglycerol increase after breakfast (20 ± 18% and 70 ± 42%, respectively; P < 0.0001) but a higher fractional triacylglycerol increase after lunch (42 ± 40% and –9 ± 27%, respectively; P < 0.0001).
The HIV-positive subjects had significantly higher insulin concentrations at baseline than did the HIV-negative subjects with features of the metabolic syndrome [median (interquartile range): 21.8 (8.6–28.7) and 10.4 (7.6–16.4) µU/mL; P = 0.0009], and the difference compared with the HIV-negative subjects remained at 4 and 8 h (P = 0.009 and P = 0.0008, respectively; data not shown). No significant differences in glucose concentrations were noted by HIV status. Notably, within the HIV group, the fractional triacylglycerol increase differed significantly after breakfast and after lunch (20 ± 18% and 42 ± 40%; P < 0.04).
The daylong results for triacylglycerol, insulin, and glucose for the PI and NNRTI groups separately are presented in Figure 3. Although baseline concentrations differed, as described above, the excursion pattern for triacylglycerols was similar for the PI- and NNRTI-treated groups (Figure 3A). Thus, there was virtually no increase in triacylglycerol concentrations after the breakfast meal for either group, whereas triacylglycerol concentrations started to increase after the lunch meal, with a broad peak 3 h after lunch. Thereafter, triacylglycerol concentrations decreased and this declining pattern continued through the postdinner measurements. Although the PI group had higher triacylglycerol concentrations than did the NNRTI group, and a significantly higher AUC, the incremental AUC for triacylglycerols for the 2 groups did not differ significantly (Table 5). For both groups, the incremental AUC for the 4–8-h period was significantly higher than the corresponding AUC for the 0–4-h period (Table 5).
FIGURE 3.. Mean (±SD) hourly concentrations of triacylglycerols (TG), insulin, and glucose in patients receiving a protease inhibitor–based antiretroviral therapy regimen (n = 13; ) and in patients receiving a nonnucleoside reverse transcriptase inhibitor–based antiretroviral therapy regimen (n = 12; ). Triacylglycerol and insulin concentrations were transformed by using logarithm and square root (Sqrt), respectively. Meals were provided after 1 h (breakfast), 4 h (lunch), and 9 h (dinner).
View this table:
TABLE 5. Postprandial triacylglycerol excursions1
As shown in Table 3, fasting insulin concentrations were significantly higher among PI subjects than in the NNRTI group (median concentrations: 22.6 and 11.8 µU/mL; P = 0.01), and insulin concentrations remained higher throughout the day for the PI group, with a trend for a higher AUC (61.7 and 37.7 µU/mL; P = 0.07; Table 6). However, the incremental AUC for insulin did not differ significantly between the PI and NNRTI groups. In both groups, insulin concentrations fluctuated during the day, with a well-defined peak shortly after each meal (Figure 3B). Glucose concentrations peaked after breakfast for both groups, being higher among the PI subjects, whereas glucose only increased modestly from baseline after lunch and after dinner (Figure 3C). The daylong average concentration for glucose was higher for the PI group than for the NNRTI group (111 ± 9 versus 103 ± 10 mg/dL), although the difference was not significant (P = 0.06). No significant difference in the incremental postprandial daylong average for glucose was found.
View this table:
TABLE 6. Daylong triacylglycerol, glucose, and insulin concentrations1
DISCUSSION
In the present study, we evaluated whether baseline concentrations predicted postprandial responses to a physiologic caloric load in 25 HIV-infected African American and Hispanic subjects without hyperlipidemia (total cholesterol and triacylglycerol concentrations < 200 mg/dL) and receiving stable antiretroviral regimens and whether the postprandial response was modulated by PI-based or NNRTI-based ART. We also compared baseline variables for the HIV-positive patients with those of a normal, healthy population and the postprandial triacylglycerol response to results obtained previously in HIV-negative subjects with signs of insulin resistance.
The results showed that although the HIV population investigated had no signs of hyperlipidemia at baseline, the HIV-infected patients had lower HDL-cholesterol concentrations than did healthy, population-based control subjects matched for age, sex, ethnicity, and BMI. Baseline total and LDL-cholesterol, triacylglycerol, and apo B concentrations were significant, positive predictors of the postlunch postprandial incremental triacylglycerol response, whereas HDL-cholesterol concentrations negatively predicted the same response. Within the HIV group, fasting triacylglycerol and insulin concentrations and the postprandial triacylglycerol AUC were higher among PI-treated patients than in NNRTI-treated patients. However, the incremental triacylglycerol response was similar between the 2 ART groups.
Most studies of postprandial responses to a test meal to date have measured the response to a standardized, high-caloric meal or fat-rich shake, which is usually given in the morning (28-30). Although this approach is useful in addressing clearing mechanisms of postprandial lipoproteins and the possible accumulation of remnant lipoproteins, it does not reflect the common eating pattern. The diet intake pattern used in the present study, however, reflects the recommended eating pattern for free-living individuals. The response to the physiologic caloric load in the HIV-positive subjects in the present study therefore represents a closer image of the daily occurring postprandial state. Although there is a growing interest in evaluating daylong lipid measurements in the assessment of postprandial lipemia (40, 41), the overall experience from such studies is limited. On the basis of the results of traditional, high-fat load studies, it is well established that HDL-cholesterol and triacylglycerol concentrations are important predictors of the postprandial response (28-30). Both of these lipid fractions are variables used to define the metabolic syndrome, and persons with this syndrome are expected to have an increased potential for postprandial lipemia (22, 23). The HIV-positive subjects receiving ART who were evaluated in the present study had significantly lower HDL-cholesterol concentrations but similar triacylglycerol concentrations under fasting conditions as did control subjects matched for age, sex, BMI, and race/ethnicity. It is in this context of interest to note that the combination of low HDL-cholesterol and high triacylglycerol concentrations was reported in HIV-infected subjects before widespread use of ART (42). Furthermore, although mean fasting glucose concentrations were <110 mg/dL, below the cutoff included in the definition of the metabolic syndrome (23), insulin concentrations were high, which is compatible with some degree of insulin resistance. Thus, although frank hyperlipidemia was an exclusion criterion in the present study, the HIV subjects assessed had several of the characteristics associated with the metabolic syndrome.
To assess predictors of the postprandial triacylglycerol response in the HIV-positive patients under physiologic conditions, we tested for correlations between baseline lipid, apolipoprotein, glucose, and insulin concentrations and the incremental triacylglycerol AUC. Several lipid indexes, including triacylglycerol, HDL-cholesterol, LDL-cholesterol, and apo B concentrations, showed a significant correlation with the incremental triacylglycerol response postlunch. This response was not related to baseline glucose or insulin concentrations. Notably, several of the lipid indexes characterizing the dyslipidemia seen in HIV-infected patients receiving ART were among these predictors, which suggests that the degree of dyslipidemia observed in patients with HIV who are receiving ART may influence postprandial lipemia during physiologic conditions. Interestingly, the response was predicted by LDL-cholesterol or apo B concentrations, which under more pronounced stress-test conditions have not uniformly predicted postprandial lipemia (28, 31).
The HIV-positive subjects had a temporal postprandial triacylglycerol pattern in which the greatest increase occurred late in the day with a muted early postprandial triacylglycerol increase. In contrast, the glucose response after a meal was greater in the morning. Because the degree of infectious burden during HIV may affect key metabolic pathways (9, 43-45), it is possible that this late triacylglycerol response may be influenced by HIV status. To address this possibility, we performed an exploratory analysis using historical data from HIV-negative control subjects who participated in the DELTA Study. Although we did not carry out a direct parallel study in HIV-positive and HIV-negative subjects, the 2 groups (HIV-positive and HIV-negative subjects) shared key metabolic features, which provided an opportunity to compare results across HIV status. In addition, the HIV-positive and HIV-negative subjects were given the same type of diet, which was based on the AHA Step I diet. As shown in Figure 2, the triacylglycerol response was significantly more delayed for HIV-positive than for HIV-negative subjects, irrespective of ART regimen. This may suggest that HIV status modulates the degree of postprandial lipid exposure. However, further studies are needed to explore this possibility.
In previous studies, higher HDL-cholesterol concentrations were reported for persons receiving NNRTI-based ART than for those receiving PI-based ART (35). Although we did not find any such differences in the present study, which was likely due to the relatively small number of subjects enrolled and the exclusion of subjects with hyperlipidemia, PI-treated subjects had higher fasting triacylglycerol concentrations than did NNRTI-treated subjects. However, we noted no significant differences between the 2 HIV-positive groups with regard to postprandial triacylglycerol, insulin, or glucose responses. These findings indicate that after a physiologic caloric load based on the AHA Step I diet, postprandial concentrations of these key metabolic markers are not affected by the choice of ART regimen in subjects without fasting hyperlipidemia. Therefore, our findings may suggest that the activity of the lipid clearance pathways might be sufficient to accommodate the repeated physiologic meal challenges in normolipidemic ART-treated subjects, irrespective of PI- or NNRTI-based therapy. Although the incremental postprandial increase did not differ significantly between the 2 treatment groups, PI-treated subjects were exposed to a higher postprandial triacylglycerol concentration. Further studies are needed to explore whether such exposure could confer an increased cardiovascular disease risk.
We recognize that our study has some limitations. The number of participants was limited, and because all HIV-infected patients were receiving antiretroviral regimens, it is difficult to differentiate the effects of HIV status and those of ART. In addition, the recruitment of normolipidemic subjects limited our ability to address postprandial lipid concentrations in HIV-infected patients with fasting hyperlipidemia, a group likely prone to an enhanced postprandial response. Furthermore, we recruited African American and Hispanic patients. We recognize that our results need to be interpreted with these limitations in mind. However, the study also offers some strengths. We evaluated the detailed response to a postprandial challenge in HIV-infected patients, an understudied area, which allowed us to identify predictors of this response. Furthermore, this is one of the first studies addressing the effects of a physiologic caloric load in this group of patients.
In summary, we showed that baseline lipid variables predicted triacylglycerol response to a physiologic caloric load in HIV-positive subjects without hyperlipidemia who were receiving ART with either PI- or NNRTI-containing regimens. Furthermore, the response was similar in patients treated with PI-based or NNRTI-based ART. This finding suggests that in a physiologic setting and under normolipidemic conditions, choice of ART regimen may not be a major modulator of postprandial lipid metabolism; however, larger studies are needed to confirm this result. The slow postprandial increase in triacylglycerol concentrations for both groups suggests that HIV disease or its treatment may modulate postprandial triacylglycerol metabolism. To further assess whether postprandial lipemia would differ depending on antiretroviral regimen in response to a more pronounced dietary challenge, studies using a standardized, high-calorie challenge are warranted.
ACKNOWLEDGMENTS
All authors contributed to the study design. The following authors contributed to the data collection (AT-G, SR, RM, BO, WK, JA, and LB), data analysis (SH, RR, HNG, WME-S, and LB), and manuscript preparation (AT-G, SH, RR, JA, HNG, WME-S, and LB). None of the authors had a conflict of interest to disclose.
REFERENCES