Background
Chronic HIV infection is characterized by progressive loss of CD4
+ T cells; suppression of viral replication with antiretroviral agents results in most subjects in rapid CD4 recovery [
1] and decreased T cell activation (e.g., CD38 expression [
2]). Defective early recovery has been demonstrated to be associated with increased morbidity [
3]; however, the extent of this recovery over time is difficult to predict, as it likely depends on multiple factors.
Baseline CD4+ T cell count remains the most relevant predictor of clinical progression and survival in subjects on antiretroviral therapy (ART) [
4‐
8], but by itself it has been shown to inadequately account for the variability in ART-mediated immune restoration, and "on treatment" assessment of CD4+ T cells retains a better prognostic value [
9]. Other factors positively associated with CD4+ T cell immune reconstitution include the presence of specific genotypes, such as Δ
32CCR5 [
10], antiretroviral regimen [
11] and, in some studies, pre-ART viral load [
12].
Immune activation of the T cell compartment (e.g., CD8
+ T cells), alterations of memory T cell subsets and depletion of innate immune subsets (e.g., NK and dendritic cells) are associated with advanced HIV infection [
1,
13‐
17]; however, while most of these cell subsets are at least partially recovered on ART, even though with different kinetics, their potential association with early CD4 recovery has not been explored.
In addition to viral and immunologic parameters, metabolic factors have been shown to be associated with disease progression, and are putative candidates to predict CD4 recovery: advanced HIV infection (i.e., low CD4 counts) is associated with chronic inflammation and increased immune activation, with alteration of metabolic parameters associated with lipid metabolism and increased atherogenic risk (as assessed by increased carotid intima-media thickness) in subjects of both sexes [
18,
19]. A number of studies have reported that subjects with advanced HIV infection have lower high-density lipoprotein (HDL) cholesterol, higher low-density lipoprotein (LDL) cholesterol and triglycerides [
20,
21], and CD4 counts appear to directly correlate with HDL cholesterol [
22,
23].
The existence of a relationship between metabolic markers, viremia and immune activation is also suggested by the observation that ART-mediated suppression of HIV replication results in a rapid normalization of a number of markers linked to cardiovascular risk [
24].
While these observations highlight the negative effects of HIV infection on lipid metabolism and overall atherogenic risk, it is of note that cohort-based observations indicate that high adiposity (which is normally associated with insulin resistance, dyslipidemia and atherogenesis) might be beneficial for HIV-infected individuals, contributing to lower steady state viral replication and slower disease progression [
25,
26]. Altogether, these observations suggest that adipose tissue accumulation and distribution may affect the immunological host/virus equilibrium in chronic HIV infection; however, the impact of adiposity on ART-mediated immune reconstitution remains undefined.
In a reported multivariate analysis, subject age, nadir and baseline CD4 count and initial viral load were found to be inversely associated with early CD4 response to suppressive ART [
12]; importantly, the predictive value of subject gender was ascribed to its effect on baseline CD4 measurements [
12,
27]. Predictive logistic regression models for incomplete CD4 response have been developed, based on subject age, baseline CD4
+ T cell count and early CD4 response [
28]; however, to our knowledge, there are at present no satisfactory models that adequately predict early (less than six months) CD4
+ T cell immune reconstitution. To our knowledge, adiposity-associated metabolic markers (e.g., BMI, serum lipid fractions, HOMA-2), have not used in these models, and their predictive role remains unclear.
Based on the reported association of viremia and CD4 counts with body mass index (BMI) and serum lipid levels, we sought to determine: (1) if adiposity and markers associated with lipid metabolism can affect the degree of early (six months [
3]) immune reconstitution after ART; and (2) if metabolic parameters could contribute to a predictive model for immune reconstitution that includes pre-ART viral, immune activation and CD4
+ T cell counts. The present study followed a cohort of ART-naïve, HIV-infected South African subjects. We demonstrate that metabolic parameters measured before ART have a significant effect on the degree of immune reconstitution attained after six months of continuous ART and contribute significantly to a predictive model of CD4
+ T cell immune reconstitution.
Discussion
We report that a multivariable model using pre-ART viral load, immunological parameters and metabolic variables predicts short-term CD4 recovery in subjects initiating ART to a substantially higher degree than previously reported models. The variability of the extent of immune reconstitution levels (i.e., CD4 gain) in response to ART-mediated viral suppression, confirmed in our cohort, suggests that a number of factors, in addition to successful viral suppression, might affect the extent of immune recovery. Pre-treatment CD4 counts, viral load and immune activation are recognized to play a role in determining the levels of immune recovery [
8,
10,
12,
34‐
36], but individually they have limited usefulness as predictors of early CD4 recovery [
9]. All individuals in our cohort received the same ART regimen, thus ruling out effects of post-ART CD4 recovery linked to differences in treatment regimens, as observed in other studies [
11].
Our results confirm that pre-ART VL, CD4 count and cellular activation (i.e., CD95 expression [
37,
38]), alone or in combination, have a significant, but limited value in predicting the CD4
+ T cell recovery outcome, explaining only 21% of its variability. The effect of baseline CD4 on ΔCD4 was negative, confirming a prior report [
39]; unlike earlier studies [
8], we did not assess the effect of baseline CD4
+ T cell levels on CD4 immune reconstitution, which was found to be positive, as we considered ΔCD4 (a measure incorporating CD4
BL) more relevant to assessing an immune reconstitution response. Prior studies have reported an effect of age and gender on CD4 outcomes of treatment [
12,
27]; while we failed to detect such associations in our cohort, the difference in outcome measured (ΔCD4 vs. CD4 count at endpoint) is likely responsible for this discrepancy.
We found a meaningful negative association between LDL/HDL ratio and CD4
+ T cell recovery. While this finding is novel, associations of lipid levels and viral replication have been reported [
40‐
43], suggesting the possibility that the observed relationship between LDL/HDL ratios and immune recovery may result in part from direct effects on viral function. A number of studies have demonstrated the effects of membrane cholesterol and lipid rafts on viral penetration and/or budding [
44‐
46]. Moreover, apolipoprotein A1, a component of HDL, has been shown to directly affect the viral life cycle at the viral entry and syncytium formation stages [
47‐
49]). A recent study indicated an association of hypocholesterolemia with a reduced response to ART [
50], and studies with cholesterol-lowering agents have shown mixed results [
51‐
56].
Adiposity has generally been associated with better viral control and slower disease progression in ART-naïve, HIV-positive subjects [
25,
26,
57,
58]. While in our cohort, BMI did not predict ΔCD4 in response to ART, in keeping with a prior report that did not detect a lack of response to ART in obese subjects [
59], we did observe a negative association between waist/hip ratio and CD4 gain, indicating that subjects with low waist to hip ratios (i.e., with low central adiposity) are likely to have better immunologic recovery. One possible hypothesis to explain the disconnect between BMI and waist/hip ratio predictive values is that antiretroviral drugs may be metabolized differently or be less bio-available in subjects with higher central adiposity (i.e., high waist/hip ratio). It is also possible that abdominal adipose tissue, particularly the visceral depot, secretes factors that may modulate the effects of the ART or directly interfere with immune reconstitution [
60].
While we did not evidence significant differences in time to viral suppression to < 50 c/ml between normal, overweight and obese subjects (Figure
1), we cannot exclude that metabolic events may be associated with residual levels of viral replication, affecting in turn short-term CD4 recovery. Importantly, the overall HDL- cholesterol values in our cohort were low, with 61% of the subjects being classified as dyslipidemic [
33], in keeping with prior reports in HIV-infected African populations [
61,
62], and there was a high prevalence of overweight/obesity [
63] (79% of women and 48% of men had BMI > 25 kg/m
2). Based on these observations, as well as the present contribution, further studies in larger cohorts will be necessary to determine if metabolic parameters play the same role in low-central adiposity individuals, and to further explore the relationship between lipids and viral control.
Altogether our data indicate that metabolic parameters contribute to predicting the degree of immune reconstitution achieved upon viral suppression. While our study does not address the pathophysiologic mechanisms underlying this relationship, prior reports indicate that fat accumulation promotes low-level inflammation, which, in turn, has been shown to be associated with lack of immunologic reconstitution [
38], suggesting a possible biological pathway.
By including pre-ART metabolic parameters in conjunction with baseline CD4, viral load and immune activation, our final model accounts for 44% of the variability in CD4+ T cell gain in response to viral suppression, representing, to our knowledge, the best predictive model on immune reconstitution to date, and represents a marked improvement over more conventional assessments (e.g., baseline CD4+ T cell counts alone or with viral load).
While not designed to support clinical interventions, our results, if supported by validation in a larger cohort, suggest the testable hypothesis that clinical and behavioural interventions aimed at reducing weight in subjects with central adiposity, as well as pharmacological intervention aimed at improving LDL/HDL ratios (e.g., statins), might improve the immunological outcomes or ART, at least in the short term.
As with all modeling techniques, there are limitations to our findings. In the first place, we modeled the effect of the assessed variables on the change in CD4 between baseline and six months on ART: it remains to be determined if incorporating multiple early CD4 measurements would improve the predictivity of the model. Moreover, the predictive value of the model will have to be validated in a larger independent cohort.
In addition, due to the relatively small size of the study, we did not assess the effect of clinical conditions that could affect some of the parameters studies here (e.g., hypertension, diabetes).
As we gain a more accurate estimate of response to ART, it remains to be determined, through further studies, how each variable impacts CD4 recovery mechanistically and whether additional predictors may improve the reliability of the prediction.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
LA was responsible for study design, data management, data analysis, and manuscript and illustration preparation. ASF supervised the statistical analysis, and contributed to data discussion and manuscript preparation. CF was responsible for clinical coordination and patient interaction, and contributed to data discussion and manuscript revision. XY was responsible for statistical analysis, and contributed to data discussion and manuscript revision. NJC was responsible for lipid assessment, and contributed to critical analysis, data discussion and manuscript preparation. DG was responsible for flow cytometry supervision, and contributed to data discussion and manuscript revision. DL was responsible for flow cytometry analysis and CD4 assessment, and contributed to manuscript revision. WS was responsible for clinical laboratory supervision, and contributed to data discussion and manuscript preparation. EP contributed to data discussion and manuscript revision. IS was responsible for supervising clinical coordination and patient interaction, and contributed to data discussion and manuscript preparation. LJM was responsible for supervising immunology laboratory assessments, and contributed to study design, critical analysis and manuscript preparation.
All authors read and approved the final manuscript.