Serum biomarkers for monitoring response to tuberculosis treatment: an assessment of the effect of different covariates among slow and fast treatment responders

The long duration of tuberculosis treatment, as well as the 2-year post-treatment follow-up period often required for predicting relapse, present a hindrance to drug development and treatment monitoring efforts. Therefore, there is need for treatment response biomarkers to inform treatment time shortening, clinical decision-making, and inform clinical trials. To assess the abilities of serum host biomarkers to predict treatment response among active PTB patients. Active pulmonary TB patients (n = 53) as confirmed by sputum MGIT culture were enrolled at a TB treatment centre in Kampala, Uganda. We evaluated concentrations of 27 serum host biomarkers at baseline, month 2, and month 6 following the initiation of anti-tuberculosis treatment using the luminex platform for their ability to predict sputum culture status at month-2 post treatment initiation. There were significant differences in concentrations of IL1ra, IL1β, IL6, IP10, MCP-1, and IFNγ during treatment. A bio-signature comprising TTP, TNFα, PDGF-BB, IL9, and GCSF best predicted month 2 culture conversion with sensitivity and specificity of 82% (95% CI; 66 -92% and 57 -96% respectively). Slow anti-TB treatment responders had higher pro-inflammatory marker levels during treatment. The strongest correlation was observed between VEGF and IL12p70 (0.94), IL17A and basic FGF (0.92), basic FGF, and IL2 (0.88), and IL10 with IL17A (0.87). We identified host biomarkers that predicted early response to PTB treatment, which may be valuable in future clinical trials and treatment monitoring. Similarly, strong correlations between biomarkers provide options for biomarkers substitutions during the development of treatment response monitoring tools or point of care tests.

结核病治疗周期漫长，且预测复发通常需要长达2年的治疗后随访周期，这两大因素均对药物研发与治疗监测工作造成阻碍。因此，亟需能够指导治疗周期缩短、辅助临床决策并助力临床试验的治疗应答生物标志物。本研究旨在评估血清宿主生物标志物对活动性肺结核（Pulmonary Tuberculosis, PTB）患者的治疗应答预测能力。本研究于乌干达坎帕拉的一家结核病治疗中心招募了53例经痰液MGIT培养确诊的活动性肺结核患者。我们采用Luminex平台，在抗结核治疗启动后的基线、第2个月及第6个月时，对27种血清宿主生物标志物的浓度进行检测，以评估其预测治疗启动后第2个月痰液培养状态的能力。治疗期间，IL1ra、IL1β、IL6、IP10、MCP-1及IFNγ的浓度存在显著差异。由TTP、TNFα、PDGF-BB、IL9及GCSF组成的生物特征标签，对第2个月痰液培养转阴的预测效果最优，其灵敏度与特异度分别为82%（95%置信区间：66%~92%与57%~96%）。抗结核治疗应答缓慢的患者，其治疗期间的促炎标志物水平更高。相关性最强的生物标志物组合为VEGF与IL12p70（相关系数0.94）、IL17A与碱性成纤维细胞生长因子（basic FGF，0.92）、basic FGF与IL2（0.88）以及IL10与IL17A（0.87）。本研究筛选出了可预测肺结核治疗早期应答的宿主生物标志物，这类标志物有望在未来的临床试验与治疗监测中发挥应用价值。此外，生物标志物间的强相关性，也为治疗应答监测工具或即时检验（point of care tests, POCT）的开发提供了生物标志物替代选择方案。