Evaluation of Microscopic Observation Drug Susceptibility (MODS) and the string test for rapid diagnosis of pulmonary tuberculosis in HIV/AIDS patients in Bolivia

Abstract

Background: Tuberculosis (TB) is the most common opportunistic infection and the leading cause of death in HIV-positive people worldwide. Diagnosing TB is difficult, and is more challenging in resource-scarce settings where culture-based diagnostic methods rely on poorly sensitive smear microscopy by Ziehl-Neelsen stain (ZN).

Methods: We performed a cross-sectional study examining the diagnostic utility of Microscopic Observation Drug Susceptibility liquid culture (MODS) versus traditional Ziehl-Neelsen staining (ZN) and Lowenstein Jensen culture (LJ) of pulmonary tuberculosis (TB) and multidrug-resistant tuberculosis (MDRTB) in HIV-infected patients in Bolivia. For sputum scarce individuals we assessed the value of the string test and induced sputum for TB diagnosis. The presence of Mycobacterium tuberculosis (Mtb) in the sputum of 107 HIV-positive patients was evaluated by ZN, LJ, and MODS. Gastric secretion samples obtained by the string test were evaluated by MODS in 102 patients.

Results: The TB-HIV co-infection rate of HIV patients with respiratory symptoms by sputum sample was 45 % (48/107); 46/48 (96 %) were positive by MODS, 38/48 (79 %) by LJ, and 30/48 (63 %) by ZN. The rate of MDRTB was 9 % (4/48). Median time to positive culture was 10 days by MODS versus 34 days by LJ (p < 0.0001). In smear-negative patients, MODS detected TB in 17/18 patients, compared to 11/18 by LJ (94.4 % vs 61.0 %, p = 0.03 %). In patients unable to produce a sputum sample without induction, the string test cultured by MODS yielded Mtb in of 9/11 (82 %) TB positive patients compared to 11/11 (100 %) with induced sputum. Of the 10 patients unable to produce a sputum sample, 4 were TB-positive by string test.

Conclusion: MODS was faster and had a higher Mtb detection yield compared to LJ, with a greater difference in yield between the two in smear-negative patients. The string test is a valuable diagnostic technique for HIV sputum-scarce or sputum-absent patients, and should be considered as an alternative test to induced sputum to obtain sample for Mtb in resource-limited settings. Nine percent of our TB+ patients had MDRTB, which reinforces the need for rapid detection with direct drug susceptibility testing in HIV patients in Bolivia.

Fig. 1

Recruitment of patients, samples obtained, and resulting groups of patients for sample analysis. 134 patients were enrolled in the study. 7 patients were unable to produce at least one sample (either string or sputum) and were excluded. 5 patients were excluded as their cultures were contaminated. 5 samples had long delays in processing by lab (>12 days); these patients were excluded. 107 subjects submitted a sputum sample sufficient for evaluation by ZN, LJ, and MODS that that were included in the analysis. Additionally, 24 of these patients submitted both induced and spontaneous sputum sample for comparison. Of the 107 patients who submitted an adequate sputum sample, 92 also submitted a string sample. 10 additional patients submitted string sample but did not submit sputum sample. Of these102 patients that submitted a string test sample, 70 patients (68.6 %) were able to produce sputum spontaneously (“sputum productive”). 22 patients (21.5 %) were only able to produce a sputum sample after induction (“sputum-scarce”). Ten patients (9.8 %) were unable to produce any sputum (“sputum-absent”). While we note the results of these 10 string samples, these ten patients were not included in our larger analysis of TB diagnostic methods as there was no sputum for comparison

Fig. 2

Cumulative percentages of the time to culture positivity for culture-positive samples according to culture method (a), according to smear status of sputa (b and c), and according to sample type in sputum-productive and sputum-scarce patients (d and e). A. Median time to sputum culture positivity cultured by MODS vs LJ (10 days vs 34 days, p <0.0001). b. Within MODS culture of sputa, effect of smear-positive vs smear-negative on time to culture positivity (median 8 vs 13 days, p <0.001). c. Within LJ of sputa, effect of smear-positive vs smear-negative on time to culture positivity (median 30 vs 40 days, p = 0.1367) d. Within the sputum-productive cohort, time to culture positivity of MODS sputum vs MODS string test (median 9 vs 13 days, p = 0.0058). e Within the sputum-scarce cohort, time to culture positivity of MODS sputum vs MODS string test (median 11 vs 14 days, p = 0.3159). String samples were not neutralized prior to storage and this may have affected both diagnostic yield and time to positivity. Median times to processing with interquartile ranges for each group of samples are below: All MODS Sputum Samples: Median 3 days (IQR 1–5). MODS Smear-positive Sputum Samples: Median 4 days (IQR 2–5). MODS Smear-negative Sputum samples: Median 2 days (IQR 2–4.75). LJ Sputum samples: Median 5 days (IQR 2–6). All MODS String Samples: Median 3 days (IQR 1–5). MODS Sputum-Productive Sputum Samples: Median 3 days (IQR 2–5). MODS Sputum-Productive String Samples (Median 3 days (IQR 2–5). MODS Sputum-Scarce Sputum Samples: Median 4 days (IQR 2–6). MODS Sputum-Scarce String Samples: Median 4 days (IQR 2–6)