Evaluation of gene chips for the rapid diagnosis of drug-resistance genes in Mycobacterium tuberculosis

Objective To evaluate the performance of gene chips in the detection of Mycobacterium tuberculosis (MTB) and determination of drug resistance. Methods Morning sputum specimens were collected from 933 suspected pulmonary tuberculosis patients at the First Affiliated Hospital of Xiamen University from June 2017 to June 2018. The sensitivity and specificity of gene chips was compared with that of GeneXpert MTB/RIF (GeneXpert) and BACTEC MGIT 960 liquid culture, and the Roche proportional drug sensitivity testing method (MGIT 960 culture+proportional method DST). Results Of 933 cases of sputum specimens, 484 sputum specimens were determined to be MTB positive using gene chips, 462 were detected as positive using GeneXpert, and 411 were detected as positive using MGIT 960. Three hundred and ninety-five samples were detected as MTB positive by all three methods, 232 cases were then selected randomly for evaluation of drug resistance, results from the Roche proportion method being used as a reference standard. For isoniazid resistance, gene chips had a sensitivity of 86.67% (26/30), and a specificity of 96.53% (195/202). Their consistency with the Roche proportion method was 95.26% (221/232), with a Kappa value of 0.798 (0.682-0.914). For rifampicin resistance, the sensitivity of gene chips was 93.75% (3032), specificity was 97.00% (194/200), consistency was 96.55% (224/232), and the Kappa value was 0.862 (0.768-0.956). For GeneXpert, the sensitivity of rifampicin resistance was 84.38% (27/32), specificity was 97.00% (194/200), consistency was 95.26% (221/232), and the Kappa value was 0.803 (0.691-0.915). The Kappa values obtained were all higher than 0.75 indicating that all three methods had good consistency. Conclusion We did not detect any significant difference between the gene chip, GeneXpert, and Roche proportional methods for the detection of rifampicin and isoniazid resistance in Mycobacterium tuberculosis.