Background
Antiretroviral treatment (ART) has dramatically increased the life expectancy of HIV-infected individuals. By 2015, more than one half of HIV-infected individuals are projected to be over 50 years old [
1]. This increased life expectancy has prompted questions about how aging with HIV-1 infection combined with HIV treatments might interact to impact
active life expectancy (e.g., having adequate mobility to function as a member of the community).
HIV-1 infection causes systemic immune system activation fostered by a complex array of insults [
2,
3]. The immune activation markers that are detected in HIV-1 infection are typically also elevated with increased age in the absence of HIV infection [
3‐
5]. Similar to aging, HIV infection is associated with the premature development of cardiovascular disease, thromboembolic disease, type 2 diabetes, cancer, neurocognitive decline, end-organ disease and frailty even when effective ART is implemented [
3,
5,
6].
In HIV infection, chronic stimulation by LPS and other substances emanating from a leaky intestinal barrier leads to the activation of innate immune cells, including monocytes and macrophages, which can be measured in sera based on the levels of the LPS-binding protein soluble CD14 [sCD14] [
7‐
9], and sCD14 levels remain elevated even with prolonged therapy [
10]. Notably, sCD14 has been implicated as a biomarker for the risk of non-AIDS mortality among HIV-infected subjects [
11]. Additional markers of innate immune activation, including interleukin-6 [IL-6] and the acute phase molecule C-reactive protein [CRP], are also predictive of non-AIDS HIV mortality [
12,
13]. Elevation of the coagulation factor D-dimer is independently associated with cardiovascular disease [
14] and mortality in HIV infection [
12]. The monocyte chemotactic protein osteopontin [OPN] is persistently elevated in HIV-1 infection [
15] and is further increased in HIV-associated dementia [
16]. Inter-cellular adhesion molecule-1 [ICAM-1] and vascular cell adhesion protein-1 [VCAM-1] are cleaved to their soluble forms following the activation of leukocytes or vascular endothelium. Both sICAM-1 and sVCAM-1 are elevated in HIV infection [
17] and associated with endothelial activation and/or an increased risk of cardiovascular disease [
18‐
21].
In addition to innate immune activation, adaptive immune activation is a hallmark of both aging and HIV-1 infection. Most notably, both advancing age and HIV-1 infection are associated with an increased frequency of memory CD4
+CD45RA
−CD45RO
+ T cells as well as expanded CD4
+CD57
+ and CD8
+CD57
+ T cell populations. CD57-bearing T cells possess a unique inflammatory senescent phenotype, whereby the cells are able to produce large amounts of inflammatory cytokines such as TNF but demonstrate impaired proliferation [
22,
23]. In both aging and HIV-1 infection, the direct cause of CD57
+ T cell accumulation is unclear, although chronic viral replication, for example human cytomegalovirus, may drive the expansion of CD57 cells through immune attrition [
24].
The proper structure (tissue composition) and function (strength & endurance) of the lower extremities is vital for maintaining mobility in late life. However, a paucity of research has investigated mobility functions and their determinants in HIV-infected older adults. Inflammation may provide a common link between HIV infection, aging and frailty. Both chronic HIV-1 infection (treated and untreated) and aging-associated frailty are characterized by elevated levels of pro-inflammatory cytokines such as TNF and IL-6 [
12,
25‐
27]. How the effects of HIV-1 infection, aging and inflammation combine to affect physical function and body composition remain mostly unknown.
Here, we determined the magnitude of HIV-associated inflammation while accounting for the traditional risk factors for age-related illness. A key objective was to examine the relationships between biomarkers of inflammation and physical function and to understand whether HIV-1 infection compounds the relationship between chronic inflammation and functional decline in older individuals. The cohort included older HIV-infected individuals (average age = 59.7 years, range 54–69 years) undergoing suppressive ART, compared to non-infected participants who were closely balanced for confounding variables including age, anthropometrics (body mass index), co-morbidities (e.g., diabetes and cardiovascular disease) and smoking status.
Methods
Subject inclusion/exclusion and enrollment
Twenty-one HIV-infected participants were recruited from the Gainesville Veterans Administration HIV clinic and the community at large. All of the participants provided written informed consent based on documents approved by the University of Florida Institutional Review Board. The protocols at Malcom Randall VA Medical Center are approved concurrently through the University of Florida IRB. The inclusion criteria were as follows: age ≥ 54 years old, well-controlled HIV on ART for at least 12 months, CD4 > 100 cells/ul and plasma HIV RNA < 5,000 copies. Combination regimens were either NRTI-based (Efavirenz + FTC and tenofovir; one case was treated with Efavirenz + abacavir) or protease inhibitor (PI)-based (atazanavir, ritonavir, fosamprenavir, saquinavir or darunavir + FTC and tenofovir). At the time of the study, 20 HIV-infected subjects had plasma HIV RNA below 50 copies/ml (the cutoff for the assay); one subject had an HIV RNA level of 208 copies per ml. The CD4
+ T cell counts in the HIV-infected subjects were typically above 500/mm
3 (mean = 560, SD = 243). Subjects were excluded from the study for any of the following criteria: active AIDS-defining illness, taking stavudine or zidovudine, hepatitis B or C infection, severe arthritis, uncontrolled hypertension, unstable angina, severe congestive heart failure, low body mass index (<20 kg/m
2), poorly controlled diabetes, treatment for cancer in the previous 6 months, peripheral vascular disease, Parkinson’s disease, multiple sclerosis, amyotrophic lateral sclerosis, renal failure, use of anabolic steroids, cognitive impairment identified as having a Mini-Mental State Exam Score < 24, or inflammatory disease (rheumatoid arthritis, inflammatory bowel disease, among others). It was expected that there would be increased inflammation in the HIV-infected group, and elevated sCD14 levels have been widely reported as a common feature of HIV-related inflammation [
11]. With the goal of determining how inflammation relates to physical composition/function, we powered the study on plasma sCD14 levels. Ten HIV-uninfected control participants were recruited after screening the HIV-infected subjects for plasma levels of sCD14. The mean sCD14 level of 21 HIV-infected subjects was 1,892 ng/ml (SD = 394). Using α = 0.05,
n = 9 provides 80 % power to detect a 25 % difference in sCD14 levels between the groups. HIV-infected subjects frequently develop age-related co-morbidities such as cardiovascular disease and respiratory disease, so a ‘healthy’ non-HIV infected control group would not be suitable for comparison. A self-report questionnaire was used to assess co-morbidities including cardiovascular conditions (controlled hypertension, previous hospitalization for myocardial infarction, pacemaker, stroke or abnormal heart rhythm) or respiratory conditions (shortness of breath, asthma or recent chest congestion). Control participants were recruited from the community, enrolled after testing negative for HIV, and balanced to the HIV cases based on average age, body mass index, and smoking status, as well as the frequency of active diabetes, cardiovascular conditions and pulmonary conditions (Table
1). Following the balancing approach, there were no significant differences between the groups in terms of age, BMI, chronic diseases or the use of non-HIV medications related to chronic disease.
Table 1
Cohort characteristics
Age* years | 62.5 (58–69) | 59.7 (54–69) | 0.100a
|
BMI* kg/m2
| 29.3 (24.8–39.5) | 30.1 (20.7–71.8) | 0.640a
|
Chronic disease | | | |
Respiratory | 30 % | 33 % | 1.000b
|
Cardiovascular | 90 % | 62 % | 0.205b
|
Diabetes | 30 % | 38 % | 1.000b
|
Current Smoker | 20 % | 24 % | 1.000b
|
Non-HIV medications | | | |
Aspirin regimen | 50 % | 43 % | 1.000b
|
Hypertension | 50 % | 76 % | 0.222b
|
Cholesterol | 90 % | 71 % | 0.379b
|
Glucose control | 30 % | 28 % | 1.000b
|
HIV medications | | | |
NNRTI | N/A | 57 % | N/A |
NRTI | N/A | 62 % | N/A |
PI | N/A | 33 % | N/A |
Measurement of plasma biomarkers
Whole blood samples were collected in sterile Vacutainer™ (Becton Dickinson, Franklin Lakes, NJ) acid citrate dextrose tubes and processed within 12 h. The PBMC and plasma samples were stored at −180 °C in liquid nitrogen and −80 °C, respectively, in non-pyrogenic polypropylene cryovials (Nunc Cryotubes™) [
28]. LPS levels were quantified using the Limulus Amebocyte Lysate [LAL] chromogenic assay (Lonza Inc., Allendale, NJ) as previously described [
28]. The plasma samples were diluted 1:4 in 0.15 M NaCl prior to analysis, and the lower limit of detection was 0.1 endotoxin units [EU] per milliliter. The following soluble markers of immune and endothelial activation were measured by ELISA: sCD14, osteopontin [OPN], C-reactive protein [CRP], soluble ICAM-1 [sICAM-1], sVCAM-1 (R&D Systems Inc., Minneapolis, MN), and IL-6 (BD Biosciences, San Diego, CA). The coagulation marker D-dimer was measured by ELISA (American Diagnostica GmbH, Stamford, CT).
Flow cytometry analysis of cell surface phenotypes
The following flow cytometry antibodies were purchased from BD Biosciences (San Jose, CA): anti-CD3-PE Cy7, anti-CD4-Alexa488, anti-CD8-PacBlu, anti-CD28-PE, anti-CD57-APC, anti-CD45RO-Alexa 700, anti-HLA-DR-APC, anti-CD14-PacBlu, anti-CD11a-FITC, anti-CD16-PE-Cy7, anti-CD163-PE, anti-CD62L-APC, and anti-CD86-Alexa 700. Two multi-color panels were used for T cells, and two panels were used for monocytes. Data were collected using a BD LSRII flow cytometer and analyzed with FCS Express software (DeNovo Software, Los Angeles, CA).
400 m rapid walk
Participants were asked to walk 400 m (20 m per lap) at a rapid pace as described elsewhere [
29]. Walking speed was calculated as the distance walked divided by the time elapsed. At the end of each lap, the participants were asked about their physical exertion on a 0 (none) to 10 (highest) scale [
30]. Lap variability may indicate fatigue and was calculated as the standard deviation in split times for each lap.
The SPPB test is a common measure of physical performance in older adults and is described elsewhere [
31]. Briefly, the test consists of timed measures of standing balance in three positions (side by side position, semi tandem position, and tandem position), walking speed over 4 m, and time to stand up and sit down 5 times in a chair as quickly as possible. Each of the 3 performance measures was assigned a score ranging from 0 to 4 according to normative data published elsewhere [
31], with 4 indicating the highest level of performance, and 0 representing an inability to complete the test. A summary score was created by adding each performance score; the summary score therefore ranges from 0 to 12. Excluding the balance test, values were reported for the speed (or time) to complete each task and score.
Lower extremity tissue composition
T1-weighted 3D-magnetic resonance imaging (MRI) was used to quantify the tissue volumes of the right leg using a Phillips 3.0 Tesla magnet (Philips Medical Systems, Bothell, WA) as described previously [
32]. Muscle, subcutaneous adipose tissue (SAT), and inter-muscular adipose tissue (IMAT) were measured volumetrically over 20 contiguous axial slices (10 in the mid-thigh and 10 in the mid-calf region) as previously described by our group [
33]. Values are expressed as the absolute volume in centimeters cubed (cm
3) and as a percent of the total volume. MRIs were collected for 18 HIV cases and 10 non-infected controls.
Lower extremity muscle strength and fatigue
Maximal knee extension and flexion isokinetic peak torque were measured using a Biodex isokinetic dynamometer (Shirley, NY). Participants were asked to complete 50 concentric contractions at 90°/s with their right leg. The peak torque (in Newton-meters) and total work (in joules) achieved during the trials was used for the data analyses. A fatigue index was calculated as the change in muscle work (in joules) during the first 16 repetitions (the 1st third) compared to the last 16 repetitions (the last 3rd). A negative value indicated a decrease in muscle work capacity in the final 16 repetitions.
Statistical analyses
For all parameters, outliers were detected using the Grubb’s extreme studentized deviate [ESD] method with an alpha value set at 0.01. After the outliers were removed, each parameter was tested for a normal distribution using the D’Agostino & Pearson omnibus normality test. Normally distributed parameters were compared using unpaired t-tests, and non-normally distributed parameters were compared by Mann–Whitney U-tests. A total of 5 outlier values were detected and removed; when the outliers were re-introduced and the analyses were re-run, no qualitative differences in outcomes or new significant differences were detected. All of the reported data excluded outliers. Pearson’s correlation and simple linear regression were used to determine the relationships between two variables.
Discussion
Both HIV infection and normal aging in the absence of HIV-1 infection are associated with chronic inflammation that negatively impacts overall health. Frailty develops earlier in HIV-infected individuals than uninfected individuals [
37,
38], and a suspected source of increased frailty is chronic inflammation which develops early in HIV-1 infection [
39]. In non-HIV-1-infected older persons, there is a well-established correlation between the biomarkers IL-6, TNF and CRP and frailty, as demonstrated in the Newcastle 85+ study [
40], although no association between immunosenescence and frailty was detected. In HIV-1 infection, the VACS index, a multi-parameter score that combines biomarkers for HIV-1 disease and organ system injury but not inflammation, predicts frailty [
41]. Here, we investigated how HIV infection augments the inflammation that occurs normally in older individuals with the goal of understanding whether potentially additive inflammatory effects of HIV-1 and advancing age accelerate the development of frailty. Our understanding of the effects of HIV-1 and advancing age on systemic inflammation may be confounded by additional factors including obesity, age-related co-morbidities, behaviors (e.g., smoking) and non-HIV medications. We carefully balanced HIV-1-infected and un-infected subjects for these factors and found that HIV-1 infection, even with suppressive antiretroviral therapy, is associated with a number of unique inflammatory phenotypes compared to a control group with similar co-morbidity and medication profiles. As a population, the HIV-1-infected group had significantly elevated levels of plasma sCD14, CRP and IL-6 compared to the uninfected controls. Although they failed to reach statistical significance, modestly elevated levels of sICAM1 and sVCAM1, markers of endothelial activation, were observed in the HIV-infected subjects. Our previous studies in infected and uninfected young adults revealed similar differences in sVCAM levels [
42].
Based on our previous work and that of others [
7,
9,
28,
43‐
47], we hypothesized that plasma LPS levels would be elevated in older HIV-infected subjects compared to uninfected subjects, however, LPS levels were similar in both groups. Our previous studies of microbial translocation focused on much younger individuals (infants and children), among whom there were clear differences in LPS levels in HIV-1-infected children compared to uninfected children [
28]. The similar levels of plasma LPS among the uninfected and HIV-1-infected older individuals was surprising, but we can speculate that there may be age-related differences in intestinal permeability and microbial translocation (even in the absence of HIV-1 infection) that mask the biomarkers of gut pathology that are easily detected in younger cohorts. Alternatively, co-morbidities such as diabetes, heart disease or kidney disease, which were prevalent among both the HIV-infected and uninfected groups, may be associated with microbial translocation. For example, endotoxemia is associated with atherosclerosis in non-HIV-infected subjects [
48].
Elevated levels of plasma sCD14 are clearly associated with poorer health status and health outcomes in HIV patients [
7,
11,
49]. While the source of sCD14 is attributed to systemic monocyte/macrophage activation, there is little evidence to implicate peripheral blood monocytes versus tissue macrophages in the elevated production of sCD14. One recent study in younger subjects (median age = 41 years) found a positive correlation between plasma levels of IL-6, D-dimer, CRP, or sCD163 and multiple phenotypic alterations in peripheral blood monocytes, including the frequency of CD16
+ monocytes [
50]. We found no evidence of monocyte phenotypic alterations in HIV-infected subjects despite markedly elevated plasma levels of sCD14. We did find, however, that HIV-infected subjects had a reduced frequency of peripheral blood monocytes compared to uninfected subjects. Whether the association between high sCD14 and a low frequency of circulating monocytes is due to an increased rate of monocyte extravasation or apoptosis or alternatively to reduced production from bone marrow precursors remains unknown. Our previous study showed that human peripheral blood monocytes produce relatively higher levels of sCD14 in comparison to macrophages, yet only macrophages responded to LPS by releasing more sCD14 [
51]. Together, these findings suggest that elevated sCD14 in older individuals may originate from tissue macrophages rather than monocytes.
The expansion of CD57-expressing T cells is typical of both HIV infection and advancing age [
34]. In CD57
+ cells specific for HIV antigens, replicative senescence arises from chronic antigenic exposure; this effect is similar to that observed in chronic infection with cytomegalovirus (CMV) associated with an increased frequency of both CD4
+ and CD8
+T cells expressing CD57 [
52]. In healthy populations, the proportion of CD8
+CD57
+T cells expands with increasing age [
53]. The combined effects of age and CMV infection result in an accumulation of CD8
+ CD57
+ T cells lacking the co-stimulatory molecule CD28 [
54]. We observed a significant increase in the proportion of CD8
+ CD28
− CD57
+ but not CD8
+ CD28
+ CD57
+ T cells in the HIV cohort. The CMV status of both HIV-infected and control groups was unknown, so the cause of expanded senescent CD8
+ T cells requires further investigation.
Similar to CD8
+ T cells, the expansion of CD4
+ CD57
+ CD28
− T cells is associated with chronic viral infections [
55‐
57]. We found an increase in the frequency of CD4
+ CD57
+ CD28
+ T cells in HIV-infected subjects but no difference in CD4
+ CD57
+ CD28
− T cells. Expansion of CD57-expressing T cells occurred independently of the total CD4 T cell number in HIV-infected subjects. Instead, the increased frequency of memory CD4
+ CD45RO
+ T cells was associated with decreased CD4
+ T cell numbers. This finding is in accordance with another study showing that increasing age and the concomitant reduced production of naïve CD4
+ T cells, rather than the expansion of senescent CD57
+ T cells, underlies the deficit in CD4 T cell reconstitution observed in younger, treated HIV patients [
24].
We anticipated that HIV disease would result in some degree of frailty and/or reduced physical performance compared to similarly-aged uninfected control subjects. Frailty is a common trait of HIV-1 infection and has been linked to numerous causes including chronic inflammation, polypharmacy, and coagulopathy [
39,
58]. Here, the control and HIV groups demonstrated nearly identical physical characteristics with similar leg tissue volume and adiposity. Strength, fitness and fatigue measures were also highly similar between the groups. Frailty in HIV-1 infection is strongly correlated with a high viral load and CD4
+ T cell decline [
38,
41], whereas lipodystrophy syndrome and fat redistribution are associated with combination antiretroviral therapy [
59]. The patients in our study had well-controlled viremia and stable CD4
+ T cell counts, which may in part explain the similarities in leg musculature between the uninfected and infected groups. It is well established that antiretroviral therapy can cause redistribution of limb fat to the trunk [
59] or increased visceral fat within the limbs [
60], but no change in leg adiposity was observed in our HIV-1-infected group. It is possible that the variability in measurable lipodystrophy in the HIV-1 infected individuals (estimates range from 13 to 70 % [
59]) is too great to observe changes on our relatively small cohort. In addition, exclusion of subjects taking the NRTI drugs stavudine or zidovudine may explain the similarities in adiposity, because HIV-associated lipoatrophy has been related to exposure to these drugs [
61].
Despite strong evidence for increased inflammation in the HIV-1 group, the infected and uninfected groups had similar muscle and physical performance characteristics. Thus, although inflammation is a characteristic of aging-related frailty and HIV-1 infection, our study suggests that increased inflammation in older HIV-1-infected individuals causes no greater frailty than what is found in similarly aged uninfected individuals. Ours is the first study that we know of where older uninfected and HIV-1-infected individuals were carefully balanced for age-related co-morbidities and medications to specifically assess the role of HIV-associated inflammation in frailty. A major limitation of our conclusion is that the study size is small; this work should be repeated in larger studies. Nonetheless, inflammation was profoundly elevated in the HIV-1-infected group, so we accept this as convincing evidence that inflammation alone does not always lead to frailty.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MAW carried out immunoassays, analyzed data and wrote the manuscript. TB, AMJ, MS, CL, MP and TM enrolled the study subjects, performed performance/tissue composition analyses, and contributed to writing the manuscript. JWS and MMG oversaw the study, performed data analysis/interpretation and contributed to writing the manuscript. All authors read and approved the final manuscript.