Switch to raltegravir decreases soluble CD14 in virologically suppressed overweight women: the Women, Integrase and Fat Accumulation Trial

Authors


  • These data were presented in part at the 20th Conference on Retroviruses and Opportunistic Infections (Atlanta, GA, USA, 3–6 March 2013).

Abstract

Objectives

Soluble CD14 (sCD14) is a monocyte activation marker associated with increased mortality in HIV infection. We assessed 48-week changes in sCD14 and other inflammatory biomarkers in virologically suppressed, HIV-infected women switching to raltegravir (RAL) from a protease inhibitor (PI) or nonnucleoside reverse transcriptase inhibitor (NNRTI).

Methods

HIV-infected women with central adiposity and HIV-1 RNA < 50 HIV-1 RNA copies/mL continued their thymidine-sparing nucleoside reverse transcriptase inhibitor (NRTI) backbone and were randomized to switch to open-label RAL at week 0 (immediate) or 24 (delayed). In an exploratory analysis, inflammatory biomarkers were measured on stored fasting plasma.

Results

Of the 37 evaluable subjects, 78% were non-White; the median age was 43 years, the median body mass index (BMI) was 32 kg/m2 and the median CD4 count was 558 cells/μL. At baseline, biomarker values were similar between groups. After 24 weeks, median sCD14 significantly declined in subjects switching to RAL [−21% (P < 0.001) vs. PI/NNRTI −5% (P = 0.49); between-group P < 0.01]. After 48 weeks, immediate-switch subjects maintained this decline and delayed-switch subjects experienced a similar decline following the switch to RAL (−10%; within-group P < 0.01). Immediate-switch subjects also experienced an initial increase in tumour necrosis factor (TNF)-α that was neither maintained after 48 weeks nor seen in delayed-switch subjects. After adjustment for multiple testing, only declines in sCD14 remained significant.

Conclusions

In this randomized trial of women with central adiposity, a switch to RAL from a PI or NNRTI was associated with a statistically significant decline in sCD14. Further studies are needed to determine whether integrase inhibitors have improved monocyte activation profiles compared with PIs and/or NNRTIs, and whether measured differences between antiretroviral agents translate to demonstrable clinical benefit.

Introduction

HIV infection is characterized by a state of inflammation and immune activation that may not normalize with suppressive antiretroviral therapy (ART) [1-5], and may contribute to the development of end-organ disease in HIV-infected persons. Recently, circulating markers of inflammation [including interleukin-6 (IL-6), high-sensitivity C-reactive protein (hs-CRP) and soluble CD14 (sCD14)] have been shown to predict all-cause mortality in HIV infection [6-8], enhancing interest in biomarkers as predictors of morbidity and mortality in this patient population.

CD14 is a monocyte/macrophage surface marker that recognizes pathogen-associated molecular patterns and is a co-receptor for lipopolysaccharide (LPS) [9]. CD14 may be membrane bound or exist as sCD14 when shed or secreted from activated monocytes/macrophages or secreted by hepatic Kupffer cells [10, 11]. sCD14 is elevated in the setting of HIV infection and does not normalize with ART initiation [12-14]. Similarly, significant declines in sCD14 have not previously been documented in virologically suppressed patients switching or intensifying ART. The associations between higher sCD14 levels, increased all-cause mortality [7, 15, 16] and progression of HIV disease [15, 17] emphasize the need to both understand the mechanism of sCD14 elevation in HIV infection and determine whether interventions to normalize sCD14 levels/monocyte activation improve clinical outcomes.

Persistent immune activation in HIV-infected persons on ART may be the result of one or more stimuli such as concomitant infections and/or comorbidities, enterocyte damage leading to microbial translocation, or medication-specific toxicities. Determining how sCD14 changes with other markers of monocyte activation, microbial translocation and inflammation [including soluble CD163 (sCD163), intestinal-type fatty acid binding protein (I-FABP), tumour necrosis factor-α (TNF-α) and soluble TNF receptor II (sTNF-RII)] could help define the mechanism driving changes in sCD14 following ART initiation or switch.

This study describes changes in biomarkers of inflammation, immune activation and microbial translocation in a 48-week trial of virologically suppressed, HIV-infected women with central adiposity on protease inhibitor (PI)- or nonnucleoside reverse transcriptase inhibitor (NNRTI)-based ART who continued their thymidine-sparing nucleoside reverse transcriptase inhibitor (NRTI) backbone and were randomized to switch to raltegravir (RAL) immediately or after 24 weeks.

Methods

Study design

Complete methods for the parent Women, Integrase and Fat Accumulation Trial have previously been published [18]. Briefly, HIV-infected women with central adiposity (defined as waist circumference > 94 cm or waist-to-hip ratio > 0.88) and HIV-1 RNA < 50 HIV-1 RNA copies/mL on a regimen of tenofovir or abacavir and emtricitabine or lamivudine plus a PI or NNRTI were randomized 1:1 to substitute the PI or NNRTI for RAL 400 mg by mouth twice a day (bid) at week 0 (immediate-switch) or week 24 (delayed-switch). Subjects randomized to delayed-switch provided an internal control group of subjects on continued PI/NNRTI therapy for the first 24 weeks. During weeks 24–48, all subjects received RAL. The study was not blinded, as randomization required switching to RAL vs. continued standard of care.

Subjects were recruited from five centres in North America between September 2008 and July 2010. Inclusion criteria were: age ≥ 18 years, documented HIV-1 infection, central adiposity, continuous virological suppression since ART initiation and current HIV-1 RNA < 50 copies/mL, current ART with a compatible NRTI backbone plus a PI or NNRTI (as above), no change in ART for ≥ 12 weeks prior to screening and ability and willingness to provide informed consent.

The parent study hypothesized that, in women experiencing central fat gain on a PI/NNRTI, switch to a more metabolically neutral agent (RAL) might prevent ongoing fat gain or allow partial reversal of lipohypertrophy. As such, the study was powered to observe a ≥ 10% difference in computed tomography-quantified visceral fat between RAL- and PI/NNRTI-treated subjects over 24 weeks. While anticipated reductions in total and low-density lipoprotein (LDL) cholesterol were observed in RAL-treated subjects, only a 5.4% between-group difference in visceral fat was observed (RAL −3.6% visceral fat; PI/NNRTI +1.9%) [18].

A protocol-defined, exploratory analysis of changes in inflammatory biomarkers was performed on stored plasma samples. The institutional review boards/ethics committees of the participating institutions approved all study documents and procedures, and all subjects provided written informed consent prior to initiation of study procedures.

Assessments

Biomarker assessments

Complete parent study assessments have previously been published [18]. For this analysis, blood for plasma isolation was obtained in EDTA tubes at weeks 0, 24 and 48 and centrifuged for 15 min at 2000 rpm and 20–22°C within 30 min of collection. Samples were stored at the sites in 1-mL aliquots at −80°C until the end of the study, when they were sent to the University of California, Los Angeles for sorting and cataloging prior to shipment to the Laboratory for Clinical Biochemistry Research at the University of Vermont, where all assays were performed under the supervision of one of the authors (RT).

sCD14, sCD163, IL-6, sTNF-RII and soluble vascular cell adhesion molecule-1 (sVCAM-1) were measured using the Human Quantikine® (R&D Systems, Minneapolis, MN, USA) enzyme-linked immunosorbent assay (ELISA), TNF-α using the Millipore Human Adipokine Panel B multiplex assay (EMD Millipore Corporation, Billerica, MA, USA), I-FABP using the Human FABP-2 DuoSet® (R&D Systems) ELISA, d-dimer using the StagoSTA®-Liatest® assay (Diagnostica Stago S.A.S., Asnières-sur-Seine, France), C-telopeptide (CTP) using the Immunodiagnostic Systems (IDS) UniQTM ICTP ELISA and pro-collagen type 1 N-terminal pro-peptide (P1NP) using the IDS UniQTM P1NP radioimmunoassay (IDS Limited, Boldon, UK). All assays had coefficients of variation of 10% or less.

Statistical analyses

Baseline characteristics were compared between treatment groups using the Mann−Whitney U-test for continuous variables and Fisher's exact test for categorical variables. Median values and interquartile ranges (IQRs) are reported for continuous variables, and percentages for categorical data.

Median, between-group, 24-week change scores for all biomarkers were compared using the Wilcoxon sign-rank test. Additionally, 48-week change scores were calculated for the immediate-switch group, and a pooled analysis of biomarker changes in the 24 weeks following the switch to RAL was performed for all subjects. Spearman or Kendall tau rank correlation coefficients were calculated to assess relationships between (1) changes in biomarkers and (2) changes in biomarkers and clinical parameters. All analyses were as-treated, excluding subjects who did not remain on the study regimen and/or did not have an observed primary endpoint. A supplemental intent-to-treat analysis and analyses of log-transformed mean values were also performed and produced similar results (data not shown).

Sample size was determined by the parent study (n = 37). All biomarker analyses were exploratory. However, 37 subjects provided 80% power to see a minimum between-group effect size of: sCD14, 453.0 ng/mL; sCD163, 372.0 ng/mL; I-FABP, 1501.0 pg/mL; IL-6, 13.0 pg/mL; d-dimer, 0.4 μg/mL; TNF-α, 2.1 pg/mL; sTNF-RII, 842.0 pg/mL; sVCAM-1, 536.0 ng/mL; CTP, 2.3 μg/L, and P1NP, 43.0 μg/L. All statistical tests were two-sided with a nominal alpha level of 0.05. Analyses were exploratory and were performed with and without adjustment for multiple testing. Data analysis and management were performed using sas 9.2 or 9.3 (SAS Institute, Inc., Cary, NC).

Results

Patient population

Sixty-one subjects were screened and 39 were enrolled in the study. Eighteen subjects were randomized to an immediate-switch, and 21 to a delayed-switch. Thirty-seven subjects completed the week 24 primary endpoint, and 36 completed the week 48 endpoint. No study withdrawals were RAL-related. Complete demographic and baseline clinical characteristics of the 37 participants included in the as-treated analysis are provided in Table 1. At baseline, randomization groups were well balanced, although the delayed-switch group had a higher rate of current tobacco use (24% in the immediate-switch group vs. 60% in the delayed-switch group). The median age was 43 years, the median body mass index (BMI) was 32 kg/m2 and 75% of subjects self-identified as Black or Hispanic. Sixty-two percent of subjects were on a PI at entry (vs. 38% on an NNRTI), and the most commonly reported NRTIs were tenofovir (78%) and emtricitabine (68%).

Table 1. Baseline demographic and clinical characteristics
 ImmediateDelayedOverall
  1. Values in the table are percentage or median with interquartile range. The Mann−Whitney U-test and Fisher's exact test were used to determine statistical significance for continuous and categorical variables, respectively.
  2. BMI, body mass index; ART, antiretroviral therapy; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; LDL, low-density lipoprotein; HDL, high-density lipoprotein; hs-CRP, high-sensitivity C-reactive protein.
  3. aP = 0.05. Otherwise, no statistically significant between-arm differences.
  4. bDefined as self-reported diagnosis or on-therapy at baseline.
Ethnicityn = 17 (100%)n = 20 (100%)n = 37 (100%)
African American53%65%59%
Hispanic23%10%16%
White18%25%22%
Asian6%0%3%
Age (years)41 (39, 47)46 (36, 51)43 (37, 49)
BMI (kg/m2)34.7 (28.8, 37.6)30.4 (27.7, 35.4)32.0 (28.0, 36.5)
Tobacco use (current)a24%60%43%
CD4 count (cells/μL)563 (447, 747)554 (354, 770)558 (422, 747)
Time on ART (years)5.1 (3.1, 7.1)2.7 (1.6, 6.3)3.7 (2.4, 7.1)
PIn = 11 (65%)n = 12 (60%)n = 23 (62%)
Atazanavir/ritonavir35%30%32%
Atazanavir6%15%11%
Fosamprenavir/ritonavir0%5%3%
Fosamprenavir0%5%3%
Lopinavir/ritonavir18%5%11%
Nelfinavir6%0%3%
NNRTIn = 6 (35%)n = 8 (40%)n = 14 (38%)
Efavirenz18%30%24%
Etravirine6%0%3%
Nevirapine12%10%11%
NRTIn = 17 (100%)n = 20 (100%)n = 37 (100%)
Abacavir18%25%22%
Lamivudine29%35%32%
Emtricitabine71%65%68%
Tenofovir82%75%78%
Waist circumference (cm)106.0 (102.0, 121.0)102.4 (99.2, 113.0)105.5 (99.5, 118.0)
Hip circumference (cm)117.5 (102.1, 127.0)106.5 (102.2, 124.4)115.5 (102.1, 127.0)
Waist:hip ratio0.96 (0.90, 0.99)0.97 (0.93, 1.02)0.96 (0.92, 1.00)
Glucose (mg/dL)84.0 (78.0, 93.0)88.5 (80.0, 97.5)87.0 (78.0, 94.0)
Total cholesterol (mg/dL)179.0 (162.0, 206.0)199.0 (164.5, 221.5)188.0 (162.0, 214.0)
Triglycerides (mg/dL)116.0 (85.0, 144.0)129.0 (101.0, 176.0)118.0 (92.0, 152.0)
LDL cholesterol (mg/dL)113.0 (103.0, 123.0)116 (89.0, 138.1)115.8 (93.0, 128.0)
HDL cholesterol (mg/dL)47.6 (40.2, 57.0)49.1 (39.0, 55.0)49.0 (40.0, 57.0)
hs-CRP (mg/dL)2.7 (0.6, 6.0)4.7 (0.8, 7.5)3.2 (0.6, 6.5)
Diabetesb0%0%0%
Hyperlipidaemiab18%25%22%

Baseline biomarker characteristics

At baseline, no significant differences in median sCD14, sCD163, I-FABP, IL-6, d-dimer, TNF-α, sTNF-RII, sVCAM-1, CTP or P1NP were observed between subjects randomized to the immediate- vs. delayed-switch arms (Table 2).

Table 2. Baseline biomarker distributions
 ImmediateDelayedOverallBetween-group
P-value
  1. Median baseline values are shown with interquartile range. The Wilcoxon rank sum test was used for determination of statistical significance, with two-sided α = 0.05.
  2. sCD14, soluble CD14; sCD163, soluble CD163; I-FABP, intestinal-type fatty acid binding protein; IL-6, interleukin-6; TNF-α, tumour necrosis factor-α; sTNF-RII, soluble tumour necrosis factor receptor II; sVCAM-1, soluble vascular cell adhesion molecule-1; CTP, C-telopeptide; P1NP, pro-collagen type 1 N-terminal pro-peptide.
n172037 
sCD14 (ng/mL)2175.7 (1940.1, 2403.9)2170.9 (1958.7, 2444.8)2175.7 (1948.1, 2432.6)0.62
sCD163 (ng/mL)629.0 (405.1, 723.4)606.1 (514.7, 753.0)613.2 (480.3, 749.6)0.49
I-FABP (pg/mL)1840.0 (1224.0, 2163.9)1755.7 (1288.7, 2245.7)1793.7 (1224.0, 2195.1)0.87
IL-6 (pg/mL)3.8 (2.3, 6.1)3.8 (3.1, 6.9)3.8 (2.6, 6.6)0.81
d-dimer (μg/mL)0.3 (0.1, 0.3)0.2 (0.1, 0.4)0.3 (0.1, 0.4)0.68
TNF-α (pg/mL)4.3 (3.6, 5.5)5.2 (4.1, 7.4)4.7 (3.7, 6.2)0.12
sTNF-RII (pg/mL)2862.2 (2543.0, 3669.7)3149.4 (2739.5, 3432.7)3067.6 (2690.3, 3542.0)0.34
sVCAM-1 (ng/mL)870.2 (644.8, 938.6)859.4 (751.8, 962.9)870.2 (686.9, 938.6)0.57
CTP (μg/L)3.2 (3.1, 3.4)3.7 (2.8, 4.9)3.2 (2.9, 3.8)0.20
P1NP (μg/L)48.6 (37.4, 72.0)55.6 (42.5, 83.6)53.1 (39.5, 75.8)0.28

Changes in biomarkers between weeks 0 and 24

Changes in biomarkers for both randomization groups are presented in Table 3. After 24 weeks, a significant median decline in sCD14 was observed in RAL-treated subjects (−461.9 ng/mL; −21%; IQR −704.0, −253.7 ng/mL; P < 0.001) compared with subjects remaining on a PI or NNRTI (−102.6 ng/mL; −5%; IQR −277.4, 107.6 ng/mL; P = 0.28; between-group P < 0.01). This decline in sCD14 occurred regardless of whether subjects switched from a PI or NNRTI, and was accompanied by an increase in TNF-α (RAL: 0.3 pg/mL; 7%; IQR −0.2, 0.6 pg/mL; P = 0.05; PI/NNRTI: −0.1 pg/mL; −2%; IQR −0.9, 0.3 pg/mL; P = 0.28; between-group P = 0.05). Subjects experiencing sCD14 declines below the median drove the increase in TNF-α among RAL-treated subjects. An insignificant increase in sTNF-RII (16.5 pg/mL; 0.6%; IQR −76.4, 236.1 pg/mL; P = 0.55) that was statistically different from the change seen in PI-/NNRTI-treated subjects (−195.7 pg/mL; −6%; IQR −333.6, −47.9 pg/mL; within-group P < 0.001; between-group P < 0.01) was also observed. sTNF-RII did not change significantly in any sCD14 subgroup. Changes in sCD14, TNF-α and sTNF-RII are illustrated in Figure 1. No statistically significant within- or between-group changes in other biomarkers were observed between weeks 0 and 24.

Figure 1.

Forty-eight-week changes in soluble CD14 (sCD14), tumour necrosis factor-α (TNF-α) and soluble tumor necrosis factor receptor II (sTNR-RII). RAL, raltegravir.

Table 3. Changes in biomarkers
 Week 0-24 changesBetween-group PWeek 0-48 changesWeek 24-48 changesPooled 24-week changes
ImmediateDelayedImmediateWithin-group PDelayedWithin-group PImmediate 0–24 weeksWithin-group P
Delayed 24-48 weeks
  1. Median within-person change scores are shown with interquartile ranges. The Wilcoxon rank sum test was used for determination of statistical significance, with two-sided α = 0.05. Significant P-values are shown in bold.
  2. sCD14, soluble CD14; sCD163, soluble CD163; I-FABP, intestinal-type fatty acid binding protein; IL-6, interleukin-6; TNF-α, tumour necrosis factor-α; sTNF-RII, soluble tumour necrosis factor receptor II; sVCAM-1, soluble vascular cell adhesion molecule-1; CTP, C-telopeptide; P1NP, pro-collagen type 1 N-terminal pro-peptide.
  3. *Within-group P < 0.05.
n1720 17     
sCD14 (ng/mL)−461.9 (−704.0, −253.7)*−102.6 (−277.4, 107.6)0.003−494.1 (−764.8, −269.4)< 0.0001−217.6 (−498.8, 14.4)0.006−308.9 (−704.0, −97.0)< 0.0001
sCD163 (ng/mL)29.4 (−71.6, 98.0)−27.5 (−44.8, 36.8)0.343.0 (−84.2, 94.0)1.0070.6 (−7.0, 165.7)0.0549.8 (−26.7, 125.4)0.05
I-FABP (pg/mL)−150.1 (−380.4, 173.3)250.2 (−379.7, 948.9)0.16−93.5 (−347.9, 227.3)0.75261.2 (−110.6, 1341.2)0.08149.7 (−380.4, 885.9)0.37
IL-6 (pg/mL)0.5 (−0.9, 1.3)−0.5 (−2.0, 0.3)0.16−0.2 (−0.4, 0.7)0.670.0 (−0.6, 1.0)0.770.3 (−0.9, 1.1)0.48
d-dimer (μg/mL)−0.1 (−0.2, 0.0)0.0 (−0.1, 0.1)0.290.0 (−0.1, 0.1)0.870.1 (−0.1, 0.2)0.310.0 (−0.1, 0.1)0.80
TNF-α (pg/mL)0.3 (−0.2, 0.6)*−0.1 (−0.9, 0.3)0.050.1 (−0.9, 0.9)0.810.2 (−0.1, 0.9)0.150.3 (−0.2, 0.8)0.01
sTNF-RII (pg/mL)16.5 (−76.4, 236.1)−195.7 (−333.6, −47.9)*0.00512.8 (−415.8, 195.4)0.4673.9 (−197.8, 207.2)1.0059.2 (−172.4, 218.4)0.66
sVCAM-1 (ng/mL)−1.5 (−170.2, 30.8)−46.0 (−173.2, 76.9)0.99−2.7 (−169.9, 39.1)0.16−1.9 (−41.0, 61.1)1.00−1.8 (−95.0, 59.7)0.38
CTP (μg/L)0.1 (−0.6, 0.8)0.1 (−0.3, 0.4)0.84−0.0 (−0.7, 0.6)0.910.3 (−0.5, 0.7)0.440.2 (−0.6, 0.7)0.47
P1NP (μg/L)4.0 (−7.6, 12.9)−4.0 (−15.1, 4.8)0.31−2.5 (−18,9, 1.8)0.080.0 (−23.8, 4.5)0.460.6 (−13.9, 9.4)0.90

Changes in biomarkers between weeks 24 and 48

Changes in biomarkers for both randomization groups are presented in Table 3. After 48 weeks, subjects randomized to an immediate-switch maintained a reduction in sCD14 (total 48-week change −494.1 ng/mL; −23%; IQR −764.8, −269.4 ng/mL; P < 0.0001). Subjects randomized to a delayed-switch saw a significant decline in sCD14 following the switch to RAL at week 24 (−217.6 ng/mL; −10%; IQR −498.8, 14.35 ng/mL; P < 0.01; Figure 2). Following the switch to RAL, both groups achieved similar sCD14 declines (week 48 between-group P = 0.48).

Figure 2.

Individual-level changes in soluble CD14 (sCD14) over 48 weeks.

In the delayed-switch group only, the switch to RAL was also associated with an increase in sCD163 (70.6 ng/mL; 12%; IQR −7.0, 165.7 ng/mL; P = 0.05). No other statistically significant changes in biomarkers were observed in either randomization group after 48 weeks. Of note, upon the switch to RAL, no significant increase in TNF-α or sTNF-RII was observed in subjects in the delayed-switch arm. Additionally, at week 48, the small increases in TNF-α and sTNF-RII initially observed in immediate-switch subjects no longer retained statistical significance.

Pooled changes in biomarkers for all subjects following switch to RAL

When 24-week post-switch data for all subjects (weeks 0–24 for the immediate-switch; weeks 24–48 for the delayed-switch) were pooled to improve power, the median sCD14 decline remained significant (−308.9 ng/mL; −14%; IQR −704.0, −97.0 ng/mL; P < 0.0001). Pooled analysis also detected significant increases in sCD163 (previously observed in both groups but only significant in the delayed-switch group; median 49.8 ng/mL; 8%; IQR -26.7, 125.4 ng/mL; P = 0.05) and TNF-α (previously observed in both groups but only significant in the immediate-switch group; median 0.3 pg/mL; 6%; IQR −0.15, 0.79 pg/mL; P = 0.01). No other statistically significant changes in biomarkers were observed in the pooled analysis, including sTNF-RII.

Adjustment for multiple testing

After adjustment for multiple testing, significance for biomarker change scores was defined as P < 0.001. While the decline in sCD14 in individual study arms approached but did not reach statistical significance (immediate-switch weeks 0–24, P = 0.003; delayed-switch weeks 24–48, P = 0.006), declines in sCD14 were significant for the 48-week change in the immediate-switch group and in the pooled 24-week analysis (both P < 0.0001).

Correlations between changes in biomarkers and clinical parameters

At baseline, sCD14 correlated with sCD163 (r = 0.40; P = 0.01) and I-FABP (r = 0.34; P = 0.04), and sCD163 correlated strongly with sVCAM-1 (r = 0.82; P < 0.0001), TNF-α (r = 0.68; P < 0.0001), sTNF-RII (r = 0.74; P < 0.0001) and LDL cholesterol (r = −0.41; P = 0.01). I-FABP correlated positively with sTNF-RII (r = 0.38; P = 0.02), visceral fat volume (r = 0.50; P < 0.01) and high-density lipoprotein (HDL) cholesterol (r = 0.43; P < 0.01), and negatively with current CD4 T-cell count (r = −0.36; P = 0.03).

In the immediate-switch group, 24-week changes in sCD14 correlated only with changes in hs-CRP (although no significant change in hs-CRP was observed; data previously published [18]; r = 0.55; P = 0.03). Correlations between changes in sCD163 and visceral fat (r = 0.56; P = 0.05), I-FABP and d-dimer (r = −0.56; P = 0.02) and TNF-α and CD4 T-cell count (r = −0.53; P = 0.03) were also present. In the delayed-switch group, significant correlations were observed between 24-week changes in sCD14 and d-dimer (r = 0.48; P = 0.03); TNF-α and sTNF-RII (r = 0.59; P < 0.01), CD4 T-cell count (r = −0.44; P = 0.05) and sVCAM-1 (r = 0.47; P = 0.04); and sTNF-RII and sVCAM-1 (r = 0.59; P < 0.01).

In the analysis of pooled 24-week changes following the switch to RAL, changes in CTP correlated with changes in sCD163 (r = 0.51; P = 0.001) and waist circumference (r = −0.44; P = 0.01), and changes in I-FABP correlated with changes in BMI (r = −0.35; P = 0.04).

Discussion

In this randomized trial in HIV-infected women with central adiposity, a switch to RAL was associated with statistically significant within- and between-group declines in sCD14 compared with subjects remaining on a PI or NNRTI. While RAL was associated with greater declines in sCD14 than NNRTI-based regimens in a small study of treatment-naïve subjects [19], to our knowledge a decline in sCD14 in virologically suppressed patients switching ART has not previously been described. This finding may have important clinical implications, as sCD14 has been associated with all-cause mortality in HIV infection [7, 15, 16].

In the SMART study, a gradient effect of sCD14 quartile on mortality was observed, with an odds ratio (OR) for mortality of 2.3 per increase in sCD14 IQR [7]. Setting the SMART overall mortality rate (1.55%) as the median mortality rate and using the per sCD14 IQR increase in OR for mortality (2.3) as a basis to calculate the OR for a one quartile change, it can be hypothesized that a one quartile increase in sCD14 might translate to a 52% increase in mortality among SMART subjects. The limitations of extrapolating these data to different patient populations are significant, and include the fact that an intervention to lower sCD14 may not have the same mortality benefit as initiating ART with a lower baseline sCD14 level; however, baseline sCD14 values in our study were similar to those in SMART, and, if the SMART data can be generalized to other patient populations, it is possible that the 21% decline in sCD14 we observed over 24 weeks in women switching to RAL might translate to an estimated 44% reduction in mortality. Or, for a similar mean follow-up time (16 months), approximately 200 subjects would need to switch to RAL to save one life.

Additionally, higher circulating levels of sCD14 and other markers of monocyte activation and/or microbial translocation have been associated with end-organ diseases including cardiovascular disease (sCD14 [20-22], sCD163 [23] and LPS [21, 24]), neurocognitive decline (sCD14 [25], sCD163 [26] and LPS [27]), and non-alcoholic steatohepatitis [11], suggesting that, if a true beneficial effect of RAL on sCD14 exists, its long-term use could be associated with a smaller burden of comorbid disease than other antiretroviral agents.

Although the mechanism of sCD14 decline in subjects switching to RAL is unknown, one possibility is that increased RAL penetration into the gut (vs. PI/NNRTI) promotes local control of viral replication and inflammation and decreased microbial translocation. In a small study of HIV-uninfected men, Patterson et al. reported rapid penetration of RAL into gastrointestinal tissue, with levels throughout the colon 160–650-fold greater than plasma levels. Additionally, RAL achieved higher levels in gastrointestinal tissue than other antiretroviral agents [28].

A similar potential mechanism is reduced viraemia and/or viral replication in areas other than the gut. However, prior RAL switch and intensification studies have not consistently demonstrated improved residual viraemia or low-level viral replication [defined as decreased HIV-1 viral load via ultra-sensitive assay or increased 2-long-term repeat (2-LTR) circles] with RAL initiation [29-34]. Additionally, studies demonstrating increased 2-LTR circles with RAL intensification saw effects predominately in PI-treated subjects [29, 30]. While measurement of 2-LTR circles and HIV-1 viral load via ultra-sensitive assay was beyond the scope of this study, sCD14 decline following a switch to RAL was not restricted to PI-treated subjects. RAL intensification has also demonstrated inconsistent improvements in T-cell activation [29, 30, 35, 36], and improved d-dimer [29] and LPS [33] but not sCD14 levels [33, 35, 36].

Finally, the observed decline in sCD14 might be attributable to known, beneficial effects of RAL on lipid levels [18, 37]. For example, reduction in circulating lipid levels could lead to reduced hepatic inflammation and steatosis (leading to decreased sCD14 secretion from the liver), as has been observed with statin use [38]. Although we did not detect correlations between changes in sCD14 and lipids or directly measure oxidized lipid levels in our study, oxidized LDL cholesterol stimulates CD14 expression on circulating monocytes [39], and oxidized HDL cholesterol activates monocytes in vitro [40]. Thus, it is reasonable to hypothesize that oxidized lipids may be a mediator of monocyte activation in HIV-infected patients.

It is important to note that, although we did not observe statistically significant changes in I-FABP or sCD163 following the switch to RAL (vs. continued PI/NNRTI), we were not powered for these endpoints, and decreased microbial translocation and/or monocyte activation could contribute to the observed decline in sCD14 levels. Additionally, although I-FABP is a known marker of enterocyte damage [41], its utility as a marker of microbial translocation in virologically suppressed, HIV-infected patients has recently been challenged [42]. Similarly, the lack of statistically significant changes in markers of vascular function (sVCAM-1) and bone metabolism (CTP and P1NP) was probably heavily influenced by both our lack of power to observe these exploratory endpoints and the large observed physiological variability. These results should therefore be interpreted as neutral rather than indicating a lack of an effect of RAL on vascular function and/or bone turnover.

Although physiological variability was large, a significant 24-week increase in TNF-α was observed in the immediate-switch group. sTNF-RII also increased in the immediate-switch group (although not significantly). Both increases were statistically different from the lack of change in TNF-α and the decrease in sTNF-RII values observed in subjects remaining on PI or NNRTI; however, after 48 weeks the increase in TNF-α was no longer significant in the immediate-switch group, and no significant changes in TNF-α or sTNF-RII were observed in delayed-switch subjects following the switch to RAL. The increase in sTNF-RII also was not significant in the pooled analysis. Additionally, the observed changes in TNF-α and sTNF-RII were small in magnitude compared with the change in sCD14 (7% for TNF-α, 0.6% for sTNF-RII and -21% for sCD14), are of unknown clinical significance and did not vary significantly by entry regimen. This latter finding is in contrast to the SPIRAL study, in which subjects switching from PI to RAL experienced significant declines in TNF-α [43]. Most importantly, only changes in sCD14 retained significance after adjustment for multiple testing. Further studies are needed to assess whether these findings can be replicated in larger cohorts, and to determine the mechanism of sCD14 decline in patients switching to RAL.

Finally, the study of sex differences in markers of immune activation is critically important, and documenting changes in biomarkers in HIV-infected women who are virologically suppressed is needed. For example, recent studies demonstrating associations between HIV infection and increased sCD14 and sCD163 were not designed to assess sex differences [12, 44, 45]. Complicating this is the observation that healthy HIV-infected women may have lower sCD14 and higher sCD163 levels than age-matched men [45]. The contribution of age to HIV and sex effects is also important: although sCD163 levels increase with age, Martin et al. recently reported sCD163 levels in HIV-infected women (87% on ART) that were similar to those in HIV-uninfected women 14.5 years older [44], a finding previously described in HIV-infected men [46]. Thus, understanding the contribution of sex to immune activation is necessary in order to optimize care for women living with HIV.

Limitations

This study has several limitations. First, the sample size was small, biomarker measurements were exploratory in nature and physiological variability was high. While the possibility of types I and II error exist in this exploratory analysis, the magnitude of sCD14 improvement, its reproducibility across treatment arms and its significance after adjustment for multiple testing lead us to believe that the observed improvement in sCD14 represents a true finding. The fact that observed correlations between biomarkers (for example, positive correlations between baseline sCD14, sCD163 and I-FABP, and the negative correlation between change in sTNF-RII and CD4 count) were in keeping with physiological expectations supports this conclusion.

Next, the high prevalence of generalized obesity in this cohort (median BMI 32 kg/m2) probably confounds any effect of RAL on biomarkers of inflammation. For example, sCD163 may be elevated in obese subjects [47], and Koethe et al. described the loss of incremental BMI effect on sCD14 in obese HIV-infected subjects [48]. Finally, we are unable to determine the mechanism of sCD14 decline in women switching to RAL, including whether the decrease arises from the switch to RAL or the switch away from the PI or NNRTI. Thus, a larger study designed to provide mechanistic insight and powered to detect clinically significant effect sizes for changes in biomarkers is needed.

Conclusions

In this randomized trial of virologically suppressed, HIV-infected women with central adiposity, a switch to RAL from a PI or NNRTI was associated with a statistically significant decline in sCD14. This is the first study to demonstrate significant changes in sCD14 following ART switch in subjects well controlled on ART, and may have important implications for mortality and/or the development of comorbidities in treated HIV-infected patients. Further studies are needed to assess whether this finding can be replicated in larger cohorts and to determine the mechanism of this decline.

Acknowledgements

This work was supported by funding from the Merck and Co. Investigator-Initiated Studies Program to JSC, funding from Merck Canada to SLW, career support from the Ontario HIV Treatment Network to SLW, and grants from the National Institutes of Health (M01-RR000865, K24 AI56933 to JSC, P30-AG028748 and T32 MH080634). The investigators would like to thank the study staff and subjects for their participation in this project, the staff at The Laboratory for Clinical Biochemistry Research (University of Vermont Department of Pathology, Colchester, VT) for their assistance with inflammatory biomarker assay performance and data interpretation, and Dr Heather McCreath, Dr Chi-Hong Tseng and Ms Diana Liao at the University of California, Los Angeles for their assistance with data analysis. We would also like to thank Drs Michael Dube, Marshall Glesby, and Kathleen Squires for their service on the Data Safety Monitoring Board.

Conflicts of interest: JEL has provided consulting services to Merck and Co. GAM has served as a scientific advisor or speaker for Bristol Myers Squibb, GlaxoSmithKline, Abbott, Tibotec, Merck and Gilead Sciences, has received research grants from Bristol Myers Squibb, GlaxoSmithKline, Abbott, Merck and Gilead Sciences, and is currently serving as the DSMB Chair for a Pfizer-sponsored study. TH has received a research grant from Merck and Co. CAW has received grant funding from GlaxoSmithKline and Theratechnologies, and served as an event adjudicator for a Pfizer study. AM was the Medical Director for HIV/Endocrinology at EMD Serono, Inc., at the time this work was performed, but participated independently of this position through her affiliation with Tufts University. SLW has provided consulting services to Merck and Co., and received a research grant from Merck Canada to help support this work. She has also served as an advisor and speaker for AbbVie, Tibotec, Bristol Myers Squibb, ViiV Healthcare and Gilead Sciences. SAS has no conflicts of interest to report. RT has served as a scientific advisor or speaker for AbbVie and Merck, and performed work supported by the grant from Merck to JSC for the conduct of this study. JSC received a research grant for the conduct of this study from the Merck and Co. Investigator-Initiated Studies Program.

Contributions to authorship: JEL was the primary author, served as Co-Principal Investigator for the protocol, aided in protocol revisions and contributed to study oversight and data analysis. GAM developed the original study design and protocol with JSC, served as Co-Principal Investigator for the protocol and contributed to the analytic plan and manuscript preparation. TH, CAW, AM and SLW were Co-Investigators, enrolled participants and contributed to manuscript preparation and review. SAS assisted with data analysis and contributed to manuscript preparation. RT was responsible for biomarker assay oversight, provided scientific consultation and assisted with manuscript preparation. JSC developed the original study design and protocol with GAM, was Co-Principal Investigator of the protocol and contributed to manuscript development.

Ancillary