Detection of tuberculosis genes associated with drug-resistance in paraffin-embedded tissue specimens using next generation sequencing technology

Abstract

Objective: To evaluate the use of multiplex PCR amplicon sequencing (mPCR-NGS) technology in detecting gene mutations related to drug resistance of Mycobacterium tuberculosis (MTB) in formalin-fixed paraffin-embedded tissue specimens, and to explore its clinical value in the diagnosis of drug-resistant tuberculosis. Methods: Fifty clinical MTB strains isolated in the Changping District Tuberculosis Control Institute of Beijing from April 2013 to October 2015 with drug susceptibility test (DST) results of rifampicin, isoniazid, ethambutol, streptomycin, ofloxacin, capreomycin, kanamycin and amikacin available were recovered, including 42 drug-resistant strains and 8 drug-sensitive strains. The mPCR-NGS test was established to detect genes related to the 8 anti-tuberculosis drugs according to the previously published studies and databases. Fifty-five paraffin-embedded tissue specimens from drug-resistant tuberculosis patients were collected in the Department of Pathology, Beijing Chest Hospital, Capital Medical University during November 2017 to September 2018. All the specimens showed no less than one mutation in the gene regions related to drug resistance of any of the 4 drugs (rifampicin, isoniazid, ethambutol or fluoroquinolones) by probe melting curve assay. The effectiveness of mPCR-NGS test was evaluated on clinical MTB isolates using phenotypic DST as the reference. Clinical evaluation of mPCR-NGS test on formalin-fixed paraffin-embedded specimens from TB patients was performed using probe melting curve assay as the reference. The sensitivity, specificity and coincidence of mPCR-NGS were analyzed. Results: Using phenotypic DST as the reference, the sensitivities of the mPCR-NGS for detecting drug-resistance of rifampicin, isoniazid, streptomycin, and ethambutol were 95% (38/40), 93% (27/29), 93% (27/29), and 72% (13/18), respectively; and the specificities were 100% (10/10), 95% (20/21), 100% (21/21), and 94% (30/32), respectively. The sensitivities for capreomycin, kanamycin and amikacin were all 100% (2/2, 3/3, 3/3), and the specificities were 98% (47/48), 100% (33/33) and 100% (47/47), respectively. The sensitivity and specificity of ofloxacin were 70% (7/10) and 100% (40/40), respectively. The total coincidence rate for the 8vdrugs was 94%, and the Kappa value was 0.87. The 55 paraffin-embedded tissue specimens included in this study were all tested by probe melting curve assays. Among them 28 were resistant to rifampicin, 37 resistant to isoniazid, 13 resistant to ethambutol, and 17 resistant to fluoroquinolones. Using the probe melting curve assay as the reference, the sensitivities of the mPCR-NGS for detecting resistant to rifampicin, isoniazid, ethambutol, and fluoroquinolones were 100% (28/28), 95% (35/37), 100%, and 100%, respectively; and the specificities were all 100% (42/42, 38/38). The total coincidence rate of the two methods was 99%, and the Kappa value was 0.98. Conclusions: mPCR-NGS showed good sensitivities and specificities in detecting drug-resistant gene mutations both in clinical MTB isolates and paraffin-embedded tissue specimens. mPCR-NGS has the potential to be an accurate and rapid molecular pathological technology for diagnosis of drug-resistant tuberculosis.