Background:
Tuberculosis (TB) is an important cause of morbidity and mortality in people with HIV (PWH). A prediction test that accurately identified those at risk of active TB would allow targeted chemoprophylaxis. The monocyte to lymphocyte ratio (MLR) and hemoglobin level collected routinely in HIV care both display an ability to predict active TB development.
Methods:
We previously identified an MLR threshold ≥ 0.23 optimally predicted incident TB after ART initiation. In this study, we used ACTG A5175 trial data as a validation cohort. We assessed the utility of baseline MLR and anemia severity, alone and in combination, for predicting incident TB in PWH in the first year after ART initiation. Participants starting ART were included in this analysis if they had no active TB at study entry or the 12 months before enrollment. Cox regression was used to assess associations of MLR and anemia severity with incident TB. Harrell’s C index was used to describe single model discrimination and model prediction was compared using log-likelihood based methods.
Results:
Total of 1,455 participants were included. Median (IQR) age was 34(29,41) years; baseline CD4 was 174(92,234) cells/mm3; 1,246(86%) participants were from high TB burden countries and 48% were women. Fifty-four participants were diagnosed with TB within 1 year of ART initiation. Median time from ART start to TB diagnosis was 4.1 (IQR 1.3,8.4) months. The hazard ratio (HR) for incident TB was 1.77[ 95% confidence interval (CI); 1.01-3.07]; p = 0.04 for those with MLR ≥0.23 versus MLR < 0.23. Compared to non anemic participants, the HR for mild/moderate anemia was 3.35[95%CI; 1.78-6.29; p < 0.001] and 18.16[95%CI; 5.17-63.77; p < 0.001] for severe anemia. After combining parameters, there were small increases in adjusted HR (aHR) for MLR ≥0.23 to 1.83[95%CI; 1.05-3.18], and increasing degrees of anemia severity (aHR 3.38[95%CI; 1.80-6.35] for mild/moderate anemia and 19.09[95%CI; 5.43-67.12] for severe anemia, respectively). C indices (95%CI) were 0.57(0.51–0.63), 0.66(0.60–0.72) and 0.69(0.62-0.76) for MLR, anemia severity and both factors combined, respectively. The model AIC decreased from 762.34 for anemia severity alone to 759.56 after addition of MLR (P=0.03).
Conclusions:
Addition of MLR to anemia severity improved prediction of incident TB. Routinely measured MLR and hemoglobin levels should be accessed at ART initiation to help identify those who would benefit from TB preventive interventions.
The Cox proportional hazard model and C index of MLR and anemia severity for incident tuberculosis among participants