TABLE 2.
Cox regression models of hepcidin, iron homeostasis, and inflammation in men and women at HIV diagnosis and associations with all-cause mortality1
| HR (95% CI)3 |
|||||
| Plasma biomarker and tertile at HIV diagnosis | n (%) | Clinical cutoff or tertile limits2 | Univariate model | Adjusted model4 | Adjusted model plus ACT5 |
| Hepcidin, μg/L | |||||
| Lowest | 65 (33) | ≤7.8 | Reference | Reference | Reference |
| Intermediate | 66 (33) | >7.8 to <57.6 | 1.95 (1.22, 3.10) | 0.96 (0.56, 1.63) | 0.97 (0.56, 1.67) |
| Highest | 65 (33) | ≥57.6 | 3.02 (1.91, 4.78) | 1.07 (0.61, 1.87) | 1.11 (0.59, 2.08) |
| Ferritin, μg/L | |||||
| Lower than normal | 23 (12) | <12 | 0.62 (0.31, 1.21) | 0.61 (0.29, 1.25) | 0.58 (0.28, 1.21) |
| Normal | 98 (50) | —6 | Reference | Reference | Reference |
| Elevated | 45 (23) | —6 | 2.10 (1.37, 3.23) | 1.57 (0.97, 2.51) | 1.90 (1.14, 3.18) |
| Very elevated | 30 (15) | >1000 | 3.91 (2.44, 6.28) | 2.09 (1.19, 3.67) | 2.78 (1.49, 5.17) |
| Transferrin, g/L | |||||
| Highest | 65 (33) | ≥1.89 | Reference | Reference | Reference |
| Intermediate | 66 (33) | >1.47 to <1.89 | 2.42 (1.37, 4.24) | 0.78 (0.39, 1.57) | 0.79 (0.40, 1.58) |
| Lowest | 65 (33) | ≤1.47 | 4.36 (2.85, 6.66) | 1.92 (1.12, 3.31) | 2.13 (1.21, 3.75) |
| Lower than normal | 133 (68) | <2.0 | 2.81 (1.81, 4.37) | 1.02 (0.57, 1.84) | 1.03 (0.57, 1.88) |
| Normal7 | 63 (32) | 2.0–3.6 | Reference | Reference | Reference |
| Iron, μmol/L | |||||
| Lower than normal | 92 (47) | <20 | 1.24 (0.87, 1.77) | 1.02 (0.68, 1.53) | 1.02 (0.68, 1.54) |
| Normal8 | 100 (51) | 20–55 | Reference | Reference (n = 168) | Reference (n = 168) |
| Elevated | 4 (2) | >55 | 0.93 (0.22, 3.82) | 0.91 (0.60, 1.37) | 0.78 (0.18, 3.27) |
| sTfR, nmol/L | |||||
| Lower than normal | 6 (4) | <10.6 | 0.87 (0.35, 2.16) | 0.52 (0.2, 1.35) | 0.52 (0.20, 1.35) |
| Normal9 | 123 (63) | 10.6–29.99 | Reference | Reference | Reference |
| Elevated | 67 (34) | >29.9 | 1.05 (0.73, 1.53) | 0.91 (0.60, 1.38) | 0.90 (0.58, 1.38) |
| Hemoglobin, g/L | |||||
| Anemic | 111 (79) | —10 | 3.26 (1.75, 6.07) | 2.75 (1.31, 5.76) | 2.72 (1.29, 5.72) |
| Nonanemic | 29 (21) | —10 | Reference | Reference | Reference |
| ACT, g/L | |||||
| Lower than normal | 2 (1) | <0.2 | 0.74 (0.10, 5.33) | 0.95 (0.12, 7.38) | NA |
| Normal11 | 151 (77) | 0.2–0.6 | Reference | Reference | NA |
| Elevated | 43 (22) | >0.6 | 1.89 (1.27, 2.80) | 1.00 (0.60, 1.66) | NA |
All assays were performed using plasma. Unadjusted model sample size, n = 196, except for hemoglobin models (n = 140); adjusted models, n = 168, except for hemoglobin models (n = 119) due to missing data. ACT, α1-antichymotrypsin; NA, not applicable; sTfR, soluble transferrin receptor.
There is no established clinical reference range for hepcidin, and distribution of raw transferrin data indicated tertile classification was informative and therefore both tertiles and clinical cutoffs were included for transferrin.
The category with the lowest risk of mortality served as the reference category.
Adjusted for HIV type (HIV-1 plus HIV-dual or HIV-2), age, gender, BMI, and absolute CD4 cell count (>500, 200–500, or <200 cells/μL) at HIV diagnosis. Gender was not included in regression models in which gender was used to establish clinical cutoffs (anemia, ferritin); age was not included when age was used to classify ferritin categories.
Adjusted for ACT (continuous) plus all variables in footnote 4.
Ferritin normal reference ranges: age 18–44 y: men, 12–200 μg/L; women, 12–150 μg/L; age ≥45 y: men, 12–300 μg/L; women, 12 to 200 μg/L (32).
Transferrin normal reference range: 2.0–3.6 g/L (33); no participants had above-normal transferrin concentrations.
Iron normal reference range: 20–55 μmol/L (33).
sTfR normal reference range: 10.6–29.9 nmol/L for living at low altitude and black (34).
Anemia: hemoglobin <13 g/L for men and <12 g/L for women (30).
ACT normal reference range: 0.2–0.6 g/L (33). A normal reference range was not provided by the assay manufacturer, and given assay method sensitivity and inconsistencies regarding the existence of a “normal” reference range for ACT, the normal limits should be interpreted with these considerations (28).