Cardiovascular Journal of Africa: Vol 29 No 3 (May/Junel 2018)

29 / 76

CARDIOVASCULAR JOURNAL OF AFRICA • Volume 29, No 3, May/June 2018 AFRICA 159
participants (p < 0.01), while ICAM-1 and E-selectin did not
differ by HIV status, as shown in Fig. 1. Similarly, sCD163, IL6
and cIMT did not differ by HIV status. The reported p -value for
IL-6 includes five HIV-infected outliers; exclusion of these did
not alter the direction of the association.
In adjusted analysis, controlling for demographic factors and
traditional CVD risk factors, HIV infection was associated with
increased log VCAM-1 and ICAM-1 but not with E-selectin
levels (p = 0.01, p = 0.03 and p = 0.75, respectively), as shown
in Table 2. More traditional risk factors were more significantly
associated with VCAM-1 than with ICAM-1 or E-selectin
(significant p-values in bold in Table 2).
In unadjusted analysis, higher log sCD163 was associated
with decreased ICAM-1 and E-selectin levels (p < 0.01 and p =
0.01) but not log VCAM-1 (p = 0.07), as shown in Table 3. These
associations were linear based on graphical plot. Inclusion of a
quadratic term for log sCD163 in each of the ICAM-1, VCAM-
1 and E-selectin models did not alter the reported association
(data not shown). In preliminary diagnostics prior to performing
adjusted analysis, there was a statistically significant negative
correlation between sCD163 and IL-6 (r = –0.27, p < 0.01);
however, the estimates were similar between the univariate and
adjusted models so both were retained in the final models.
Final multivariate models for each marker of endothelial
dysfunction demonstrated sCD163 to be associated with
decreased ICAM-1 (–61.5, 95% CI: – 105.4, 17.6, p < 0.01), and
E-selectin levels (–8.56, 95% CI: –15.30, –1.82, p = 0.01) but not
log VCAM-1 (0.08, 95% CI: –0.02, 0.18, p = 0.13), as shown in
Table 3. These associations were linear. Neither IL-6 nor cIMT
was associated with any biomarker of endothelial function.
Discussion
We found that HIV-infected adults in SSA who had attained viral
suppression on ART had biomarkers consistent with greater
endothelial dysfunction when compared to HIV-uninfected
controls, even after adjusting for traditional CVD risk factors.
In this small cross-sectional study of adults between 30 and 50
years, endothelial dysfunction was associated with sCD163 but
not IL-6 or cIMT.
This evidence of ongoing endothelial dysfunction despite
chronic viral suppression among HIV-infected patients when
compared to HIV-uninfected controls indicates that endothelial
dysfunction, first observed prior to ART initiation in this
setting,2 persists following ART initiation, similar to what
has been observed in other settings.5-7 Specifically, persistently
elevated ICAM-1 levels following viral suppression were
observed among black South Africans,28 Kenyan women29 and
a predominantly male Danish cohort 7 following mean ART
duration of approximately three, one and 12 years, respectively.
This aligns well with our findings following a mean of nine years
of ART exposure.
Furthermore, our finding of elevated VCAM-1 levels
following prolonged ART was previously observed in the same
black South African cohort studied by Fourie et al.,28 but not in
the Danish cohort.7 To the best of our knowledge, our findings
support those of Fourie et al. in a mixed male and female cohort
and expand on these findings by providing ascertainment of
date of ART initiation.28 Taken together, our findings in relation
to ICAM-1 and VCAM-1 build on those of Fourie et al.28 and
Graham et al.1 in SSA and also highlight the strong effect of HIV
disease on endothelial dysfunction, even after prolonged viral
suppression and controlling for multiple traditional CVD risk
factors and the use of statin therapy.
We did not observe an association between HIV status and
elevated E-selectin level, in keeping with a similar report in a
Table 2. Association between VCAM-1, ICAM-1, E-selectin levels and HIV-status in multivariate models (among HIV-infected participants only)
CVD risk factor
Log VCAM-1 (ng/ml) ICAM-1 (ng/ml) #E-selectin (ng/ml)
Estimate (95% CI) p-value Estimate (95% CI) p-value Estimate (95% CI) p-value
HIV status (ref = controls) 0.14 (0.04, 0.25) 0.01 55.17 (5.87, 104.46) 0.03 1.19 (–6.15, 8.52) 0.75
Age (per 5-year increase) 0.08 (0.03, 0.13) < 0.01 9.03 (–14.62, 32.68) 0.45 –1.59 (–5.07, 1.89) 0.37
Gender (ref = female) –0.18 (–0.30, –0.06) < 0.01 –18.62 (–72.85, 35.61) 0.50 2.49 (–5.72, 10.70) 0.55
Body mass indexa (kg/m2 ) –0.02 (–0.03, –0.01) < 0.01 1.81 (–3.52, 7.14) 0.50 0.22 (–0.58, 1.01) 0.59
Mean arterial pressureb 0.002 (–0.002, 0.006) 0.44 –0.08 (–1.99, 1.83) 0.93 0.20 (–0.09, 0.49) 0.17
Cholesterol (mmol/dl) –0.001 (–0.003, 0) 0.01 –0.15 (–0.68, 0.38) 0.58 0.01 (–0.08, 0.10) 0.82
Glycosylated haemoglobin (%) 0.01 (–0.06, 0.09) 0.75 20.30 (–14.29, 54.88) 0.25 6.16 (1.43, 10.10) 0.01
Previous smoker (not current) vs never –0.001 (–0.123, 0.121) 0.99 25.02 (–33.28, 83.33) 0.40 4.31 (–4.2, 12.93) 0.33
Current smoker vs never 0.07 (–0.10, 0.23) 0.45 123.93 (44.47, 203.40) < 0.01 10.72 (–0.47, 21.90) 0.06
Statin use (ref = on statin) –0.21 (–0.47, 0.05) 0.12 5.25 (–119.28, 129.78) 0.93 –5.85 (–26.56, 14.85) 0.58
VCAM-1, vascular cell adhesion molecule; ICAM-1, intercellular adhesion molecule.
#Significant interaction not reported because of high rates of missing data.
aBody mass index calculated as body weight (kg) divided by height2 (m).
bMean arterial pressure calculated as one-third systolic blood pressure (mmHg) plus two-thirds diastolic pressure (mmHg).
Table 3. Association between sCD163, IL-6 levels, cIMT and biomarkers
of endothelial dysfunction among HIV-infected participants models
Univariate models Multivariable models
Estimate p-value Estimate (95% CI) p-value
Log VCAM-1* Adj R2 = 0.02
IL-6 –0.0008 0.14 –0.0006 (–0.002, 0.0005) 0.27
Log sCD163* 0.08 0.07 0.08 (–0.02, 0.18) 0.13
cIMT –0.23 0.54 –0.33 (–1.07, 0.40) 0.37
ICAM-1 Adj R2 = 0.05
IL-6 0.19 0.40 0.02 (–0.43, 0.47) 0.93
Log sCD163* –61.7 < 0.01 –61.5 (–105.4, –17.6) < 0.01
cIMT –45.1 0.78 1.27 (–309.8, 312.4) 0.99
E-selectin Adj R2 = 0.05
IL-6 0.23 0.66 0.15 (–0.86, 1.15) 0.77
Log sCD163* –7.8 0.01 –8.56 (–15.30, –1.82) 0.01
cIMT –2.5 0.93 5.64 (–51.2, 62.5) 0.84
VCAM-1, vascular cell adhesion molecule; ICAM-1, intercellular adhesion
molecule
*Data log transformed for univariate linear regression analysis for VCAM-1
and sCD163.