Havumaki Joshua, Hillemann Doris, Ismail Nazir, Omar Shaheed Vally, Georghiou Sophia B, Schumacher Samuel G, Boehme Catharina, Denkinger Claudia M
Foundation for Innovative New Diagnostics, Geneva, Switzerland.
National Reference Laboratory for Mycobacteria, Forschungszentrum Borstel, Borstel, Germany.
PLoS One. 2017 Mar 24;12(3):e0173804. doi: 10.1371/journal.pone.0173804. eCollection 2017.
Despite recent diagnostic advances, the majority of multidrug-resistant tuberculosis (MDR-TB) cases remain undiagnosed. Line probes assays (LiPAs) hold great promise to curb the spread of MDR-TB as they can rapidly detect MDR-TB even when laboratory infrastructure is limited, yet few of these assays are currently widely available or supported by World Health Organization (WHO) policy.
The aim of this prospective, blinded, non-inferiority study was to compare the performance of YD Diagnostics REBA MTB MDR LiPA (YD) to the WHO-endorsed Hain MTBDRplus V1 LiPA (Hain V1) for the detection of rifampicin and isoniazid resistance. In phase 1, YD and Hain V1 diagnostic performance was assessed with selected culture isolates and results were compared to phenotypic drug susceptibility testing (DST) results and targeted sequencing data. In phase 2, both assays were tested on processed sputum samples and results were compared to phenotypic DST results.
In phase 1, YD did not achieve non-inferiority to Hain V1. For isoniazid resistance detection, Hain V1 had a sensitivity of 89% (95%CI 83.8-93%) and specificity of 99.4% (95%CI 96.9-100%). While YD had a similar sensitivity of 92% (95%CI 87.3-95.4%), the specificity was inferior at 92.6% (95%CI 87.6-96%). For rifampicin resistance detection, Hain V1 had a sensitivity of 90.2% (95%CI 84.8-94.2%) and specificity of 98.5% (95%CI 95.7-99.7%) while YD had an inferior sensitivity of 72.4% (95%CI 65.1-78.9%) and a comparable specificity of 98% (95%CI 95-99.5%). Similar results were observed in phase 2. For MDR-TB detection, the sensitivity and specificity of Hain V1 was 93.4% (95%CI 88.2-96.2%) and 96.2% (95%CI 88.2-96.8%), respectively, compared to 75.7% (95%CI 68-82.2%) and 92% (95%CI 88.2-94.9%) for YD.
YD did not achieve non-inferiority with Hain V1. Further improvements and repeat evaluation of YD is necessary prior to recommending its use for clinical settings.
尽管近期诊断技术有所进步，但大多数耐多药结核病（MDR-TB）病例仍未得到诊断。线性探针分析（LiPAs）有望遏制MDR-TB的传播，因为即使实验室基础设施有限，它们也能快速检测出MDR-TB，但目前这些分析方法中很少有能广泛应用或得到世界卫生组织（WHO）政策支持的。
这项前瞻性、盲法、非劣效性研究的目的是比较YD诊断公司的REBA MTB MDR LiPA（YD）与WHO认可的海恩MTBDRplus V1 LiPA（海恩V1）检测利福平及异烟肼耐药性的性能。在第一阶段，用选定的培养分离株评估YD和海恩V1的诊断性能，并将结果与表型药物敏感性试验（DST）结果及靶向测序数据进行比较。在第二阶段，对处理后的痰液样本进行两种分析方法的检测，并将结果与表型DST结果进行比较。
在第一阶段，YD未达到非劣效于海恩V1的标准。对于异烟肼耐药性检测，海恩V1的敏感性为89%（95%CI 83.8 - 93%），特异性为99.4%（95%CI 96.9 - 100%）。虽然YD的敏感性与之相似，为92%（95%CI 87.3 - 95.4%），但其特异性较差，为92.6%（95%CI 87.6 - 96%）。对于利福平耐药性检测，海恩V1的敏感性为90.2%（95%CI 84.8 - 94.2%），特异性为98.5%（95%CI 95.7 - 99.7%），而YD的敏感性较差，为72.4%（95%CI 65.1 - 78.9%），特异性与之相当，为98%（95%CI 95 - 99.5%）。在第二阶段观察到了类似结果。对于MDR-TB检测，海恩V1的敏感性和特异性分别为93.4%（95%CI 88.2 - 96.2%）和96.2%（95%CI 88.2 - 96.8%），而YD分别为75.7%（95%CI 68 - 82.2%）和92%（95%CI 88.2 - 94.9%）。
YD未达到非劣效于海恩V1的标准。在推荐将其用于临床之前，有必要对YD进行进一步改进和重复评估。
