In this prospective study, we sought to determine whether previously-described abnormalities in CBC-derived ratios for patients with new symptomatic active TB could be detected at the earlier stage of new asymptomatic TB infection, as identified by IGRA conversion. Among Tanzanian adolescents enrolled in the DAR-PIAT trial, baseline RBC measures (specifically RBC count, hemoglobin, and hematocrit) were modestly but significantly decreased among subsequent IGRA converters relative to those who remained IGRA-negative throughout the study period. Of note, no significant differences were observed between IGRA converters and controls with respect to WBC measures, platelet count, or CBC-derived ratios. Analysis of IGRA-converter subgroups demonstrated that persistent converters had the greatest number of CBC components that were statistically different from controls at baseline, followed by late converters, transient converters, and early converters. Hematocrit was the only CBC component to be significantly reduced among all IGRA converter sub-groups evaluated. With respect to CBC-derived ratios, MLR was modestly but significantly decreased in transient converters relative to controls, while NLR and PLR were not significantly different in any sub-group analyses.
CBCs collected at the time of IGRA conversion revealed little in the way of differences between IGRA converters and time-matched controls. Specifically, there were no significant differences observed with respect to any CBC components or CBC-derived ratios between either early converters or late converters and time-matched controls, while in the comparison of transient and persistent converters, only absolute monocyte count differed between groups (decreased in the persistent converter group). Notably, in the comparison of early converters versus late converters, late converters had a significantly increased hemoglobin and hematocrit, although a similar time-dependent trend was observed among IGRA-negative controls as well. Finally, with respect to the effect of treatment on CBC components or CBC-derived ratios, no significant differences were observed among participants (including both IGRA converters and controls) who received vaccine versus placebo. In sub-group analysis, however, a modest but significant difference in RDW was observed between late converters who received the DAR-901 vaccine versus those who received saline placebo.
A number of prior studies have evaluated CBC-derived ratios (MLR, NLR, and/or PLR) in the context of symptomatic TB disease, specifically related to either predicting subsequent TB disease development, diagnosing TB disease, or monitoring disease response to antimicrobial therapy [5,6,7,8,9,10,11,12, 15,16,17,18,19,20,21,22,23,24,25]. These studies were heterogeneous with respect to the geographic region from which participants originated, presence of comorbidities (such as HIV), age, concurrent use of TB therapy, and choice of control groups, making comparisons challenging.
MLR is arguably the best studied of the CBC-derived ratios evaluated here, with prior studies demonstrating that: 1) MLR differs between TB cases and controls at the time of symptomatic disease [8,9,10,11,12], 2) MLR normalizes in response to TB therapy [8, 10, 12], and 3) baseline differences in MLR predict subsequent development of TB disease [5,6,7]. Both increased and decreased MLR values appear to be associated with risk of subsequent TB development  or current TB disease [10, 11], although the biological explanation for this phenomenon is unclear. Similarly, NLR has previously been associated with the subsequent development of TB disease  and the presence of active TB disease (including both pulmonary and extrapulmonary sites) [18,19,20,21,22, 25], as well as the need for TB retreatment in previously-treated individuals , and development of acute respiratory distress syndrome (ARDS) in subjects with miliary tuberculosis . Most prior studies have reported that concurrent NLR is elevated in TB disease subjects relative to controls, but that NLR may be even further elevated in the setting of other acute respiratory infections . Finally, an elevated PLR has been associated with symptomatic TB disease at both pulmonary and extrapulmonary sites [22, 23, 25].
To the best of our knowledge, the present study is the first to evaluate CBC ratios associated with asymptomatic TB infection. Baseline MLR, NLR, and PLR did not differ significantly between subsequent IGRA converters and IGRA-negative controls, although sub-group analysis revealed a modest but significant decrease in MLR among transient converters relative to controls. We suspect that the discrepancy between our findings and those of prior studies that have evaluated CBC-derived ratios in the setting of TB disease likely results from the evaluation of immunologically distinct points in the natural history of tuberculosis (i.e., asymptomatic TB infection versus active TB disease). For example, an elevated NLR could result from an elevated neutrophil count, as is commonly observed in the setting of acute infectious processes, and/or a reduced lymphocyte count, as has been described previously in the setting of severe TB disease . Similarly, an elevated PLR may result from reactive thrombocytosis (e.g., due to inflammation) and/or a reduced lymphocyte count. None of these laboratory abnormalities are classically associated with asymptomatic TB infection, but are well-characterized in the setting of active infections, such as TB disease.
A novel observation of the present study is that certain baseline red blood cell measures (namely RBC count, hemoglobin, and hematocrit), were modestly but significantly reduced between subsequent IGRA converters compared with IGRA negative controls. Furthermore, hematocrit was significantly reduced in all IGRA converter sub-groups evaluated relative to controls. Importantly, this was not a marker of overall undernutrition, a known risk factor for tuberculosis , since baseline BMIs did not differ between IGRA converters and IGRA negatives (p = 0.91, data not shown). The reduced red cell indices observed could either represent a marker of susceptibility or a consequence of new infection prior to IGRA positivity. Anemia at the time of TB disease diagnosis is a well-documented phenomenon in the literature, most likely representing a form of anemia of chronic disease (with or without concomitant iron-deficiency anemia) . Iron dysregulation has been shown to increase susceptibility to the development of TB disease among HIV-positive individuals , although this has not been studied in the broader population and comparable data are not available for asymptomatic TB infection. Based on prior findings in the context of TB disease as well as our own findings in asymptomatic TB infection, we speculate that iron dysregulation may increase susceptibility to the development of asymptomatic TB infection, potentially account for our finding of decreased RBC count, hemoglobin, and hematocrit in subsequent IGRA converters.
There are three limitations of the present study that should be acknowledged. First, the diagnosis of TB infection is an inherently challenging task, and IGRA itself represents an imperfect diagnostic assay. Thus, the “gold standard” against which we are evaluating the performance of MLR, NLR, and PLR, as well as other CBC measures, may have misclassified a subset of IGRA converters and/or controls. The “true” performance of these laboratory values with respect to differentiating between these two groups is therefore impossible to determine with our current diagnostic capabilities. Secondly, only a small number of study participants (n = 4) in the DAR-PIAT trial developed active TB disease during the follow-up period, and we consequently do not have the statistical power to compare CBC measures or CBC-derived ratios between TB-infected participants who developed TB disease versus those who did not. The study may have additionally been underpowered to detect differences in MLR, NLR, and PLR between IGRA-negative and IGRA-positive individuals given the relatively small number of participants who converted during the follow-up period. Third, because MLR, NLR, and PLR are hypothesized to represent non-specific markers of inflammation, inclusion of other controls groups, such as individuals with other chronic infections (e.g., HIV) or chronic inflammatory conditions, would have been potentially beneficial. However, our study was designed primarily as a TB vaccine trial and therefore did not include such groups, although others have considered similar controls in prior studies [18,19,20, 23, 24].