Pre-Treatment Neutrophil Count as a Predictor of Antituberculosis Therapy Outcomes: A Multicenter Prospective Cohort Study

Abstract

Background: Neutrophils have been associated with lung tissue damage in many diseases, including tuberculosis (TB). Whether neutrophil count can serve as a predictor of adverse treatment outcomes is unknown.

Methods: We prospectively assessed 936 patients (172 HIV-seropositive) with culture-confirmed pulmonary TB, enrolled in a multicenter prospective cohort study from different regions in Brazil, from June 2015 to June 2019, and were followed up to two years. TB patients had a baseline visit before treatment (month 0) and visits at month 2 and 6 (or at the end of TB treatment). Smear microscopy, and culture for Mycobacterium tuberculosis (MTB) were performed at TB diagnosis and during follow-up. Complete blood counts were measured at baseline. Treatment outcome was defined as either unfavorable (death, treatment failure or TB recurrence) or favorable (cure or treatment completion). We performed multivariable logistic regression, with propensity score regression adjustment, to estimate the association between neutrophil count with MTB culture result at month 2 and unfavorable treatment outcome. We used a propensity score adjustment instead of a fully adjusted regression model due to the relatively low number of outcomes.

Results: Among 682 patients who had MTB culture results at month 2, 40 (5.9%) had a positive result. After regression with propensity score adjustment, no significant association between baseline neutrophil count (10³/mm³) and positive MTB culture at month 2 was found among either HIV-seronegative (OR = 1.06, 95% CI = [0.95;1.19] or HIV-seropositive patients (OR = 0.77, 95% CI = [0.51; 1.20]). Of 691 TB patients followed up for at least 18 months and up to 24 months, 635 (91.9%) were either cured or completed treatment, and 56 (8.1%) had an unfavorable treatment outcome. A multivariable regression with propensity score adjustment found an association between higher neutrophil count (10³/mm³) at baseline and unfavorable outcome among HIV-seronegative patients [OR= 1.17 (95% CI= [1.06;1.30]). In addition, adjusted Cox regression found that higher baseline neutrophil count (10³/mm³) was associated with unfavorable treatment outcomes overall and among HIV-seronegative patients (HR= 1.16 (95% CI = [1.05;1.27]).

Conclusion: Increased neutrophil count prior to anti-TB treatment initiation was associated with unfavorable treatment outcomes, particularly among HIV-seronegative patients. Further prospective studies evaluating neutrophil count in response to drug treatment and association with TB treatment outcomes are warranted.

Keywords: biomarker; neutrophil count; neutrophils; treatment outcome; tuberculosis.

Figure 1

Neutrophil count at baseline by treatment outcome. (A) Comparison of neutrophil count at baseline by treatment outcome (favorable/unfavorable, among N=691 patients); (B) comparisons of neutrophil count by treatment outcome, stratified by HIV status (577 seronegative patients, with 34 unfavorable outcome, and 114 seropositive patients, with 22 unfavorable outcome). Favorable treatment: cure or treatment completion. Unfavorable treatment: death, failure, recurrence. P-value computed via Wald test.

Figure 2

Association between neutrophil count at baseline with time until unfavorable treatment outcome. Kaplan-Meier curves comparing the impact of higher (>7500/mm³) and lower (<7500/mm³) neutrophil count at baseline on the probability of facing a favorable outcome, considering (A) the total population (919 patients), (B) stratified by HIV status (749 seronegative and 170 seropositive patients). The red line corresponds to low neutrophil count and the blue line corresponds to high neutrophil count, with their respective 95% confidence intervals. Favorable treatment: cure or treatment completion. Unfavorable treatment: death, failure, recurrence. Log rank p-values: 0.002 (A), <0.001 (B, left), 0.1 (B, right).