SEARCH

SEARCH BY CITATION

Keywords:

  • diagnosis;
  • drug susceptibility assay;
  • microscopy;
  • observational;
  • tuberculosis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Background and objective

While commercial liquid culture techniques have emerged over 20 years ago, Ziehl–Neelsen (ZN) smear microscopy remains the primary method for diagnosis of tuberculosis (TB) in China because of cost considerations. The microscopic observation drug susceptibility (MODS) assay has been evaluated in different parts of the world to determine whether it can give comparable result to commercial liquid techniques. However, most reports detail evaluation of sputum specimens. This study evaluated the performance of MODS assay for detection of Mycobacterium tuberculosis in extrapulmonary specimens in a Chinese population.

Methods

A total of 173 samples, including pleural fluid (n = 112) and cerebrospinal fluid (CSF, n = 61) samples, were collected from patients suspected to have extrapulmonary TB and tested by ZN smear microscopy, Lowenstein–Jensen (LJ) culture and the MODS assay. Discordant results among MODS assay and the other two methods were resolved by 90-day follow-up evaluation for all suspected patients.

Results

The sensitivity of the MODS assay on pleural fluid and CSF samples was 20.5% and 37.5%, respectively, while the specificity of MODS assay on both types of samples approximated 100%. The median time to culture results for the MODS and LJ methods was 14 days, 32 days for pleural fluid, and 9 days and 31 days for CSF samples, respectively.

Conclusions

MODS assay is useful to diagnose extrapulmonary TB and may be an effective and affordable method in resource-limited countries.


Abbreviations
CSF

cerebrospinal fluid

LJ

Lowenstein–Jensen

MODS

microscopic observation drug susceptibility

TB

tuberculosis

TBM

tuberculous meningitis

TBP

tuberculous pleurisy

ZN

Ziehl–Neelsen

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Timely and accurate diagnosis is one of the key steps in the process of controlling tuberculosis (TB). Along with the development of techniques for detecting TB, various kinds of new methods for the diagnosis of TB,[1] including nucleic acid amplification tests and serological antibody tests, are presently emerging. However, there are still some lingering concerns regarding the application of these methods.[2-4] Microbiological diagnosis remains the gold standard for TB diagnosis.

The conventional microbiological methods for TB diagnosis are primarily Ziehl–Neelsen (ZN) smear microscopy and Mycobacterium tuberculosis culture. ZN smear is simple and rapid but has poor sensitivity. As the Lowenstein–Jensen (LJ) culture takes an average of 4–5 weeks to yield results, this method can hardly meet clinical demand. In comparison, commercial automated liquid culture methods are rapid and sensitive.[5] However, they are overly expensive for routine use. In China and other major developing countries, ZN smear microscopy remains a primary method for diagnosis of TB. Developing rapid, sensitive and affordable methods for diagnosing TB is still one of the major problems that need to be resolved in developing countries.

The microscopic observation drug susceptibility (MODS) assay, originally described by Caviedes et al. in 2000,[6] is a novel liquid culture method for the detection of M. tuberculosis. This technique uses Middlebrook 7H9 liquid medium and relies on direct microscopic observation to detect cord formation, which is characteristic of M. tuberculosis growth. Published research has demonstrated that the MODS assay is faster, more sensitive and inexpensive compared with other microbiological assays for M. tuberculosis detection, which make it a feasible mean of diagnosing TB in developing countries.[7-10] Although it has been extensively evaluated, most reports deal with the evaluation on sputum specimens. In this study, we undertook a prospective evaluation of the MODS assay to determine its performance characteristics for detection of M. tuberculosis in pleural fluid and cerebrospinal fluid (CSF) samples in a Chinese population in a region with the second highest prevalence of TB in the world.[11, 12]

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Patient and sample collection

The study was conducted in the First Affiliated Hospital of Nanchang University and the Chest Hospital of Jiangxi Province, China, from April 2010 to January 2011 in two target populations, with consecutive recruitment. These two target populations consisted of patients suspected of tuberculous pleurisy (TBP) and tuberculous meningitis (TBM), who provided pleural fluid and CSF samples, respectively. Study exclusion criteria for all target populations were as follows: patients less than 16 years of age, patients already receiving TB therapy for more than 7 days, and inability or unwillingness to give written informed consent.

Patients were defined as ‘confirmed TBP’ if they had symptoms suggestive of TBP and satisfied one of the following conditions: (i) either a positive ZN smear or a LJ culture for M. tuberculosis from the sputum, pleural fluid or pleural biopsy samples; and (ii) the presence of typical caseating granulomas in the pleural biopsy. Patients with clinical manifestations of TBP but failing to meet either of these two conditions were defined as ‘clinically suspected TBP’ and received a 90-day follow up after the initial sample collection. Patients undergoing follow up were defined as ‘confirmed TBP’ if they (i) satisfied one of the two above conditions; (ii) had a good clinical response to TB therapy; or (iii) died from TB. Patients were defined as ‘TBP unlikely’ if they satisfied none of the confirmed TBP conditions, (i) recovered in the absence of TB treatment and (ii) had no clinical response to TB therapy.

Patients with symptoms suggestive of TBM were defined as ‘confirmed TBM’ or ‘clinically suspected TBM’ using the methods of TBP definition. Patients defined as ‘clinically suspected TBM’ underwent a 90-day follow up and were defined as ‘confirmed TBM’ or ‘TBM unlikely’ according to the same procedures of clinically suspected TBP patients definition.

The TB diagnostic criteria and treatment procedures followed were in accordance with ‘Guideline of clinical diagnosis and treatment: TB section’[13] of China.

The study was approved by the Ethics Committee of the First Affiliated Hospital of Nanchang University and was carried out in compliance with the Helsinki Declaration. All the samples were taken as part of routine clinical care, and this study did not alter the standard of care. Written informed consent was obtained from all subjects prior to study.

Sample processing

Pleural fluid and CSF samples were centrifuged at 3000 g for 15 min, the supernatant was removed and the pellets were resuspended in a final volume of 2 mL of Middlebrook 7H9 broth containing oleic-albumin-dextrose-catalase (Becton Dickinson, Sparks, MD, USA).

ZN smear

ZN smears were performed on centrifuged sample pellets.[14] For each new batch of ZN smear detection, M. tuberculosis H37Rv (ATCC 27294) and Escherichia coli (ATCC 25922) were used as positive or negative control for acid-fast stain. A sample was deemed positive if two or more acid-fast bacilli were identified among 300 fields in a smear.

LJ culture

Each resuspended pellet (200 μL) was inoculated on two slants of LJ medium (Becton Dickinson). The slants were incubated at 37°C and examined daily for 8 weeks.

For each new batch of culture, two slants inoculated with 200 μL of Middlebrook 7H9 broth containing no M. tuberculosis or 50 CFU of M. tuberculosis H37Rv were included to serve as negative or positive control. All the cultures were examined by ZN smear microscopy, and a sample was defined as positive if at least one of two culture positive slants was confirmed by ZN smear. A colloidal gold immunochromatography kit (Hangzhou Genesis Biodetection & Biocontrol Ltd, Hangzhou, China) was then used to distinguish M. tuberculosis from non-tuberculous mycobacteria for all acid-fast bacteria.

A culture was considered to be contaminated if the bacterial growth was smear negative for acid-fast bacilli; the contaminated culture was subsequently discarded, and the original sample was cultured again after being decontaminated once more.

MODS culture

The methods have been described in previous reports.[8, 15] Briefly, the MODS procedure was carried out using sterile 24-well tissue culture plates. Eight hundred microlitres of Middlebrook 7H9 broth containing 10% oleic-albumin-dextrose-catalase enrichment and PANTA antibiotic supplement (Becton Dickinson) was added to each of the two test wells, and 200 μL of the decontaminated samples was subsequently inoculated into each well. In each plate, a positive control well of M. tuberculosis H37Rv was included, in which a 1/50 dilution of 0.5 McFarland standard equivalent in broth was added, a well containing only medium was included to serve as negative control. Plates were sealed with Scotch polyethylene tape (Fisher, Springfield, NJ, USA) and placed within a gas-permeable plastic bag. Plates were then incubated at 37°C and examined daily from day 5 to day 15 and twice per week from day 16 to day 40 with an inverted light microscope. Any cord formation observed in samples was considered possible positive. Each possible positive culture was confirmed by ZN smear and distinguished by colloidal gold immunochromatography assay. A sample was considered negative if there was no evidence of growth in both test wells. Fungal or bacterial contamination was recognized by rapid overgrowth and clouding in wells within 5 days of inoculation. Contaminated cultures were discarded, and additional decontamination and culturing were undertaken using a stored portion of the original sample.

With the exception of centrifugation, sample processing and the performance of inoculation, smear preparation and colloidal gold immunochromatography assay were done in class II biosafety cabinet. Before centrifugation, all samples were sealed in centrifuge tube with screw cap.

Statistical analysis

Data were analysed with SPSS version 13.0 software (SPSS Inc, Chicago, IL, USA). The comparison of the time to detection of culture positivity was performed by the Wilcoxon signed-rank test. McNemar's chi-square test was used to compare the sensitivities and specificities of detection of the three methods. Fisher's exact tests were employed for categorical variables, and non-parametric Mann–Whitney tests were used for continuous variables. All reported confidence intervals are two-sided 95% confidence intervals, and P-values <0.05 were regarded as statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study patients and samples

Two hundred and twenty-three patients suspected of having extrapulmonary TB were referred to this study. Fifty patients were not recruited due to exclusion criteria or sample conditions. A total of 173 samples were collected, including 112 pleural fluid samples from patients with suspected TBP and 61 CSF samples from patients with suspected TBM. Recruitment of patients and assignment are shown in Figures 1 and 2. The demographic and clinical characteristics of the study population are shown in Table 1.

figure

Figure 1. Diagram for assignment of study patients suspected of tuberculous pleurisy (TBP). LJ, Lowenstein–Jensen; MODS, microscopic observation drug susceptibility.

Download figure to PowerPoint

figure

Figure 2. Diagram for assignment of study patients suspected of tuberculous meningitis (TBM). LJ, Lowenstein–Jensen; MODS, microscopic observation drug susceptibility.

Download figure to PowerPoint

Table 1. Demographic and clinical characteristics of patients
CharacteristicPatients with suspected TBP (%)Patients with suspected TBM (%)
Clinical TBPTBP unlikelyClinical TBMTBM unlikely
  1. Summary measure is n (%) for all categorical characteristics.

  2. HIV, human immunodeficiency virus; IQR, interquartile range; PTB, pulmonary TB; TB, tuberculosis; TBM, tuberculous meningitis; TBP, tuberculous pleurisy.

Total78343229
Male gender55 (70.5)21 (61.8)15 (46.9)17 (58.6)
Age (IQR)40 (27–54)64 (56–72)39 (21–49)45 (29–52)
TB historyYes10 (12.8)5 (14.7)4 (12.5)4 (13.8)
 No64 (82.1)29 (85.3)25 (78.1)24 (82.8)
 Unknown4 (5.1)03 (9.4)1 (3.4)
TB contactYes4 (5.1)1 (2.9)2 (6.3)0
 No62 (79.5)26 (76.5)25 (78.1)26 (89.7)
 Unknown12 (15.4)7 (20.6)5 (15.6)3 (10.3)
HIV statusYes1 (1.3)000
 No72 (92.3)31 (91.2)27 (84.4)28 (96.6)
 Unknown5 (6.4)3 (8.8)5 (15.6)1 (3.4)
SymptomsCough65 (83.3)29 (85.3)14 (43.8)9 (31.0)
 Fever53 (67.9)21 (61.8)30 (93.8)26 (89.7)
 Night sweats40 (51.3)11 (32.4)8 (25.0)5 (13.8)
 Weight loss37 (47.4)19 (55.9)10 (31.3)5 (13.8)
 Chest pain56 (71.8)27 (79.4)4 (12.5)3 (10.3)

Performance of MODS in TBP diagnosis

All acid-fast bacteria from LJ culture and MODS culture in this study were demonstrated to be M. tuberculosis by colloidal gold immunochromatography. Of the 112 total TBP suspects, 78 patients received a final diagnosis of TBP by laboratory test (pleural biopsy, ZN smear and LJ culture) or clinical follow up (Fig. 2). Among these 78 TBP patients, 16 were positive for M. tuberculosis by MODS assay, 5 were positive for M. tuberculosis by LJ culture and 1 was positive for M. tuberculosis by ZN smear. Using clinical diagnosis as reference, the sensitivity of the MODS assay, LJ culture and ZN smear in TBP diagnosis was 20.5% (16/78), 6.4% (5/78) and 1.3% (1/78), respectively (P < 0.001), and the specificity values of all these three assay was 100% (Table 2).

Table 2. Sensitivity, specificity and predictive values for the detection of M. tuberculosis in three types of sample using clinical diagnosis as the gold standard
CharacteristicPatients with suspected TBPPatients with suspected TBM
Percentage95% CIPercentage95% CI
  1. CI, confidence interval; LJ, Lowenstein–Jensen; MODS, microscopic observation drug susceptibility; NPV, negative predictive value; PPV, positive predictive value; TBM, tuberculous meningitis; TBP, tuberculous pleurisy; ZN, Ziehl–Neelsen.

MODS assaySensitivity20.5 (16/78)12.2–31.237.5 (12/32)21.1–56.3
 Specificity100 (34/34)89.7–100100 (29/29)88.1–100
 PPV10079.4–10010073.5–100
 NPV35.425.8–45.059.245.4–73.0
LJ cultureSensitivity6.4 (5/78)2.1–14.318.8 (6/32)7.2–36.4
 Specificity100 (34/34)89.7–100100 (29/29)88.1–100
 PPV10047.8–10010054.1–100
 NPV31.823.1–41.552.739.5–65.9
ZN smearSensitivity1.3 (1/78)0.0–6.93.1 (1/32)0.1–16.2
 Specificity100 (34/34)89.7–100100 (29/29)88.1–100
 PPV1002.5–1001002.5–100
 NPV30.620.2–39.248.335.7–60.9

Performance of MODS in TBM diagnosis

Of the 61 total TBM suspects, 32 were confirmed to have TBM by microbiological test (ZN smear and LJ culture) or clinical follow up. Among these 32 TBM patients, the MODS assay, LJ culture and ZN smear identified 12, 6 and 1 positive samples, respectively. The sensitivity was 37.5% (12/32) for the MODS assay, 18.8% (6/32) for LJ culture and 3.1% (1/32) for ZN smear. The specificity values of all of the three assay was 100% (Table 2).

Time to culture results

For pleural fluid samples, the median time before the MODS assay-yielded results was significantly shorter than required by the LJ culture (14 days vs 32 days, P = 0.042). In CSF samples, the median time to culture results was 9 days and 31 days for MODS and LJ cultures, respectively (P = 0.024).

Contamination and costs

The proportion of contaminated pleural fluid cultures was 3.6% (4/112) for MODS culture and 5.4% (6/112) for LJ culture. No contamination was observed in MODS and LJ cultures of CSF samples. All contaminated cultures were re-inoculated using stored decontaminated samples, and no contamination was observed in these re-inoculated cultures.

In this study, the cost per pleural fluid and CSF sample was approximately $US0.5 for the MODS assay and $US 1.0 for LJ culture. Costs were calculated based upon locally purchased reagents and did not include equipment costs.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

There are several published reports on the performance characteristics of MODS in relation to the detection of M. tuberculosis. The performance characteristics of MODS assay on extrapulmonary specimen have rarely been evaluated. In this study, 112 pleural fluid samples from patients with suspected TBP and 61 CSF samples from patients with suspected TBM were simultaneously tested with ZN smear microscopy, LJ culture and the MODS assay. Results showed that MODS was the most sensitive assay for both pleural fluid and CSF samples.

Compared with LJ culture, the sensitivity of the MODS assay was more than three times higher for pleural fluid samples and two times higher for CSF samples. Meticulous resolution of discordant results is essential when an investigated assay is more sensitive than reference assays. Therefore, we performed a 90-day follow-up evaluation for all of the suspected patients. To ensure the reliability of our results, staff members without the knowledge of the MODS assay results carried out all of the follow-up evaluations. Using clinical diagnosis supplemented with clinical follow up as the gold standard, we demonstrated that the MODS assay is a reliable test for TBP and TBM diagnosis, with excellent specificity and significantly higher sensitivity than both ZN smear and LJ culture. To our knowledge, this is the first study regarding the application of the MODS assay on TBM diagnosis. Notably, the sensitivity of MODS on CSF samples in this study was relatively lower than that from the only published research on TBM diagnosis with the MODS assay from Caws and co-workers, which was 64.9% based on clinical diagnostic and microbiological criteria.[16] However, the sensitivity of ZN smear and LJ culture from that particular study was also significantly higher than most of the other reports on extrapulmonary TB diagnosis,[17-20] which could be due to the differences in their clinical diagnostic criteria and the significantly higher TB prevalence in Vietnam.[21]

Besides the excellent sensitivity of the MODS assay, other outstanding advantages are easy performance and rapid detection. As reported by others, our results show that MODS provides results significantly faster than LJ culture. The median time to detection with LJ culture of the pleural fluid sample was 32 days, while the MODS results were generally obtained within 2 weeks. The MODS assay accurately identified patients with TBM in less than one-third the time required for detection by LJ culture.

Generally, increased sensitivity brings the risk of increased contamination. However, this study demonstrated that the frequency of contamination is relatively lower with the MODS assay than LJ culture (3.6% vs 5.4% in pleural fluid samples; no contamination was observed in CSF samples cultured by MODS). This is mainly due to the use of PANTA antibiotic supplement in MODS liquid medium.

Besides performance characteristics, cost is another important factor for the feasibility and application of any test, particularly in resource-constrained settings. The non-proprietary nature of MODS substantially limits costs, and it requires relatively little infrastructure. In this study, the cost per pleural fluid and CSF sample was approximately $US0.5 for the MODS assay and $US1.0 for LJ culture. Similar findings regarding the cost of MODS were reported in previous studies.[12] Due to limited funds, we did not perform detection with commercial automated liquid culture techniques, such as MGIT 960, the automated mycobacteria testing system accommodate mycobacteria growth indicator tube from Becton Dickinson. From published data, we can conclude that the cost of commercial automated liquid culture techniques is far higher ($US8–13 per sample) than the MODS assay; however, both the sensitivity and specificity of MODS are comparable with commercial automated liquid culture techniques.[12]

Although there is a biosafety concern on the MODS assay because of the usage of liquid medium and the need of several readings, this risk is limited considerably by handling MODS plates in ziplock bags. In fact, World Health Organization has approved MODS as a moderate-risk method and guided that class I or class II biological safety cabinets are adequate for handling MODS assay.[22]

In conclusion, MODS assay performs well in detection of M. tuberculosis in both pleural fluid and CSF samples. This makes it an effective and affordable method for diagnosis of extrapulmonary TB in resource-limited countries.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This study was supported by grants from the National Natural Science Foundation of China (No. 81060001) and the Foundation of Sci & Tech Research Project of Jiangxi province (No. 2009BSB11219).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References