Diagnostic accuracy and clinical value of polymerase chain reaction tests for Mycobacterium tuberculosis in peritoneal dialysis effluent: A 20-year single-centre retrospective study

Abstract

Mycobacterium tuberculosis (MTB)-related peritonitis is a rare but serious complication in patients receiving maintenance peritoneal dialysis (PD). Early diagnosis is difficult due to the low sensitivity and delayed results of conventional microscopy and culture methods. MTB polymerase chain reaction (PCR) testing in PD effluent is recommended as a diagnostic adjunct, but real-world data remain limited. We conducted a 20-year single-centre retrospective study in a tuberculosis-endemic region to evaluate the diagnostic accuracy and clinical utility of MTB-PCR in PD effluent. Among 372 tests, MTB-PCR demonstrated sensitivity 50%, specificity 100%, negative-predictive value 94.6% and positive-predictive value 100%, using diagnoses based on a composite of clinical and laboratory criteria as the reference standard. Sensitivity showed a numerical trend of improvement from 33.3% with earlier assays to 50–85.7% with newer assays. Of 72 patients with culture-confirmed MTB-PD peritonitis, 13 (18.1%) were diagnosed via MTB-PCR. Compared to those diagnosed by non-PCR methods, MTB-PCR-diagnosed patients had shorter time to anti-tuberculosis treatment initiation (median 8 vs. 22 days, p ≤ 0.001) and shorter hospital stay from presentation to treatment (median 8 vs. 17 days, p = 0.008). They also had a numerically lower rate of PD catheter removal prior to treatment initiation [0/13 (0%) vs. 9/53 patients (17.0%), p = 0.186]. Rates of permanent transfer to haemodialysis and all-cause mortality at 1 year were similar among the two groups. These findings suggest a role for early MTB-PCR testing in suspected MTB-PD peritonitis. Further studies are needed to confirm the findings and optimize diagnostic strategies.