 |
WEPE094 - Poster Exhibition
Factors associated with elevated D-dimer levels in HIV-positive individuals
A. Borges1, J. O'Connor2, A. Phillips2, J. Baker3, M. Vjecha4, M. Losso5, H. Klinker6, G. Lopardo7, I. Williams2, J. Lundgren1, for the INSIGHT SMART, ESPRIT and SILCAAT Study Groups
1University of Copenhagen, Rigshospitalet, Copenhagen HIV Programme, Copenhagen, Denmark, 2University College London Medical School, London, United Kingdom, 3University of Minnesota, Department of Medicine, Minneapolis, United States, 4Veterans Affairs Medical Center, Washington, United States, 5Hospital JM Ramos Mejia, Buenos Aires, Argentina, 6University of Wuerzburg Medical Center, Würzburg, Germany, 7Fundación Centro de Estudios Infectológicos, Buenos Aires, Argentina
Background: In the SMART and ESPRIT studies, elevated D-dimer levels were strong predictors of all-cause mortality. The factors associated with elevated D-dimer levels during HIV infection, however, remain poorly understood. Methods: In this cross-sectional study, all SMART, ESPRIT and SILCAAT participants with available D-dimer measurements at baseline were included (N= 9,837). Factors independently associated with elevated D-dimer levels were identified by linear and logistic regression models. The following predictors were investigated: age, gender, race, body mass index, markers of inflammation (CRP and IL-6), past and current exposure to different ART classes. For linear regression, log10-transformed D-dimer levels were used as the dependent variable. For logistic regression, participants were dichotomized into two groups: low and elevated D-dimer levels; the latter defined as levels greater than 0.377 µg/mL (4th quartile). Regression coefficients and OR with 95% CI were calculated to assess the contribution of predictors. Results: Factors independently associated with higher D-dimer levels by linear regression were: female sex (+0.141, p< 0.0001), black race (+0.055, p< 0.0001), older age (+0.061 per 10 years older, p< 0.0001), being off ART at baseline (+0.075, p< 0.0001), higher viral load (+0.037 per log10, p< 0.0001), CRP (+0.009 per µg/mL, p< 0.0001) and IL-6 (+0.001 per pg/mL, p< 0.0001). In contrast, higher CD4+ count (-0.019 per 100 cells higher, p< 0.0001) and use of PI-containing ART regimens were associated with lower D-dimer levels (Tables). Logistic regression yielded consistent results, with female sex (OR=2.36, p< 0.0001) and black race (1.82, p< 0.0001) being important predictors of elevated D-dimer levels.
| Factors | Number of subjects (% of total) | D-dimer levels (µg/mL)
median (IQR) | Multivariable linear regression: adjusted regression coefficients
(95% CI) | | Demographics | | | | | Age (per 10 years) | 9,837 (100) | - | 0.061 (0.053; 0.068) | | Race | | | | | Black | 1,912 (19.4) | 0.30 (0.18-0.53) | 0.055 (0.038; 0.073) | | White | 6,643 (67.5) | 0.22 (0.15-0.34) | Ref | | Gender | | | | | Female | 1,961(19.9) | 0.30 (0.20-0.51) | 0.141(0.125; 0.157) | | Male | 7,876(80.1) | 0.22 (0.15-0.34) | Ref | | Body mass index (kg/m2) | 9,837 (100) | - | 0.001 (-0.000; 0.002) |
[Linear regression parameter estimates (N=9,837)]
| Factors | Number of subjects
(% of total) | D-dimer levels (µg/mL)median (IQR) | Multivariable linear regression: adjusted regression coefficients (95% CI) | | CRP (per µg/mL) | 9,837 (100) | - | 0.009 (0.008; 0.010) | | IL-6 (per pg/mL) | 9,837 (100) | - | 0.001 (0.001; 0.001) |
[Linear regression (inflammatory markers)]
| Factors | Number of subjects
(% of total) | D-dimer (µg/mL)
median (IQR) | Multivariable linear regression: adjusted regression coefficients
(95% CI) | | CD4+ count per 100 (cells/mm3) | 9,837 (100) | - | -0.019 (-0.021; -0.016) | | HIV viral load, per log10 (copies/ml) | 9,837(100) | - | 0.037 (0.029; 0.045) | | Baseline ART use | | | | | Off ART | 821 (8.4) | 0.34 (0.22-0.62) | 0.075 (0.037; 0.114) | | PI based | 3,690 (37.5) | 0.24 (0.16-0.36) | -0.040 (-0.070; -0.011) | | NNRTI based | 3,584 (36.4) | 0.22 (0.15-0.35) | -0.027 (-0.057; 0.004) | | PI and NNRTI | 783 (8.0) | 0.22 (0.15-0.33) | -0.045 (-0.082; -0.009) | | NRTI only | 959 (9.8) | 0.26 (0.17-0.39) | Ref |
[Linear regression (HIV-specific variables)]
Conclusions: Black race, female sex, older age, viral replication and inflammatory markers are independent predictors of elevated D-dimer levels during HIV infection. Baseline ART use, PI-containing regimens in particular, is associated with lower D-dimer levels, a finding possibly attributable to lower viral loads and higher CD4+ counts in those on ART.
Download the e-Poster
Back to the Programme-at-a-Glance
|
|