Tuberculosis is a major public health problem in the world and India accounts for 27% of the total 10.0 million new cases reported worldwide. In developing countries like India, Extrapulmonary tuberculosis (EPTB) is an emerging clinical problem due to increasing immunocompromised conditions.  EPTB accounts for 15% to 20% among all cases of tuberculosis. Due to the non specific clinical presentation and paucibacillary nature of extrapulmonary tuberculosis specimens, clinical and laboratory diagnosis is made difficult.  In India, Ziehl Neelsen (ZN) smear microscopy is widely used for sputum screening but they are less sensitive for EPTB specimens. Conventional culture on LJ is better compared to smear microscopy, but poor in sensitivity compared to BACTEC Mycobacterium Growth Indicator Tube (MGIT960), which is an automated liquid culture system. In LJ M.tuberculosis takes 2 to 6 weeks to grow but to declare a negative report it takes 8 weeks. Automated systems like MGIT consumes less time when compared to LJ. Hence this study was undertaken to study the performance of culture from EPTB specimens in automated MGIT960 system and to compare it with Lowenstein-Jensen (LJ) culture.
This prospective research study was approved by the Institutional Human Ethical Committee and was carried out in compliance with the principles of the Declaration of Helsinki. Written informed consent was obtained from patients before collection of clinical samples. Patients with a clinical suspicion of extrapulmonary tuberculosis were included in the study. Patients with pulmonary tuberculosis, sputum positive extrapulmonary tuberculosis and those treated with ATT (Anti-tuberculous treatment) drugs were excluded.
A total of 565 single clinical samples collected from EPTB patients were processed. These included body fluids like pleural fluids, ascitic fluids, adrenal fluid, cerebrospinal fluids, synovial fluids, peritoneal fluids, pericardial fluids, urine samples and other specimens like pus, lymph node aspirates, cold abscess, gastric aspirates, biopsies, TB spine, and bone marrow aspirate. With sterile precautions, these samples were processed for culture on MGIT960 and LJ media.
Decontamination and Processing:
Processing of Sterile body fluids:
Sterile body fluids like C.S.F, pleural fluid, ascitic fluid, synovial fluid, peritoneal fluid, pericardial fluid, adrenal fluid and bone marrow aspirates were transferred to sterile screw capped falcon tubes in a Type 2B Bio safety cabinet and centrifuged at 3500 rpm for 20 minutes. The supernatant was discarded and the deposit was used for smear and inoculation of LJ and MGIT960 without decontamination.
Processing of Pus/Urine/Gastric aspirates:
For specimens like pus and gastric aspirates which are likely to have contaminating bacteria, smears were prepared directly without concentration. For urine samples, early morning fully voided urine was collected for three consecutive days in sterile wide mouth 250 ml containers. The samples were transferred to a 50ml falcon tube, centrifuged and smears were made from the deposit. After preparing the smears, the deposits were further digested and decontaminated for culture.
Digestion and Decontamination procedure:
Specimens were decontaminated using MycoPrep (Becton Dickinson; BBL MycoPrep) specimen digestion/decontamination kit. The decontamination procedure was carried out as per the instructions of the manufacturer. Briefly, Equal volume of specimen and activated MycoPrep reagent (NALC-NaOH solution) were added into a sterile falcon centrifuge tube and capped. (Final concentration of NaOH is 1%). Vortex mixed and allowed to stand for 15 minutes at room temperature .The tubes were centrifuged at 3500 rpm for 15-20 minutes. The supernatant was discarded and the sediment resuspended in 2ml of sterile phosphate buffer and used for inoculation of LJ and MGIT tubes.
Culture by conventional method on LJ:
LJ media (Himedia / MicroXpress / Becton Dickinson, Heidelberg, Germany) were used for inoculation of decontaminated specimens. LJ slant were inoculated with 0.5 ml of processed sample/sterile specimen and incubated in the slanting position for up to one week and kept in upright position afterwards. Growth of Acid Fast Bacilli (AFB) was confirmed by Ziehl-Neelsen staining and bacterial contamination was detected using Gram staining.
Culture using fluorometric automated liquid culture system:
The BD BBL Mycobacteria Growth Indicator Tube:
MGIT tube contains 110 µL of fluorescent indicator (Tris 4, 7-diphenyl-1, 10-phenanthroline ruthenium chloride pentahydrate in a silicone rubber base) and 7mL of modified Middlebrook broth base flushed with 10% CO2 and capped tightly using polypropylene caps. The broth base has additionally casein peptone. The broth base should be supplemented with OADC enrichment and PANTA antibiotic mixture to make it as a complete medium before inoculation of specimens. Growth supplement contains 15 mL Middlebrook OADC enrichment made up of Bovine albumin, Dextrose, polyoxyethylene stearate, Catalase and Oleic acid. PANTA vial contains a mixture of antibiotics which include Polymyxin B 6″,000 units, Amphotericin B 600 µg, Nalidixic acid 2″,400 µg, Trimethoprim 600 µg and Azlocillin 600µg approximately in a lyophilized vial.
Lyophilized vial with PANTA antibiotic mixture was reconstituted with 15 ml of MGIT growth supplement. In this study additionally , 0.015µg of Vancomycin was added to 15ml of reconstituted vial to eliminate contamination by Gram positive organisms.  Aseptically 0.8 mL of this mixture and 0.5 ml of the processed sample was added to the MGIT tubes and recapped. The inoculated tubes were scanned and placed inside the appropriate place shown by BACTEC MGIT 960 instrument and incubated at 37°C. BACTEC MGIT 960, a fully automated system monitors the incubated tubes for increasing fluorescence every hour for 42 days or until it is flagged positive.
Processing of Positive MGIT Tubes:
MGIT tubes flagged positive by the BACTEC MGIT 960 automated system was removed for further processing and observed for granular turbidity. Broth was vortexed and two to three ml was transferred to sterile centrifuge tube and centrifuged at 3500 rpm for 20 minutes. Supernatant was discarded and ZN smears were made from MGIT tubes flagged positive by the machine or from the colonies grown on LJ media. Identification of MTBC was confirmed by a combination of standard conventional tests and rapid immunochromatographic MPT64 antigen detection kits.
Extrapulmonary specimens particularly body fluids are paucibacillary to abacillary in nature. Ziehl-Neelsen technique which is widely used for sputum smear microscopy is less sensitive for EPTB specimens. Conventional culture employing LJ media is less sensitive for EPTB specimens and also consumes more time which delays the diagnosis and treatment. Automated liquid culture systems like MGIT are more sensitive compared to LJ and make the identification faster. Varying percentages of culture positivity have been reported by different researchers using different isolation techniques. (Table:.2)
The egg-based solid media LJ has been widely used for the isolation of M. tuberculosis in spite of its lesser sensitivity compared to liquid media.  A low positivity rate of 1%, 4% and 5% on LJ has been reported. [7-9] The present research records culture positivity of 5.8% by LJ. Though some researchers [10-13] had reported a high positivity rate like 45%, 46.8%, 48.9% and 50.7% on LJ, others [14-20] have reported moderate positivity ranging from 9.14% to 32.6%. Chakravorty et al  reported 0% by conventional culture but 7.9% from the same samples using USP culture.
The highest isolation rate of 83.3%, 50.7%, 45% and 32.6% positivity from lymphadenitis samples using LJ was reported by few authors.[22″,13″,23″,20] In the present study, 18% of the lymphadenitis / cold abscess specimen and 20% of the lymph node biopsies grew M. tuberculosis complex on LJ. Kamal et al and Noussair et al [24″,25] reported 61.3% and 17% positivity from lymph node biopsies respectively. Patwardhan et al  reported a sensitivity of 74.5%, specificity of 100%, PPV of 1% and NPV of 0.34% for LJ positivity of lymph node aspirates.
From pleural fluids some authors could not isolate any MTBC. In this present study 1.3% LJ positivity is recorded which is almost similar to the findings of Sharma et al . Other authors [15, 22, 28] reported positivity from pleural fluid on LJ ranging from 13.8% to 22.9%. From ascitic fluid, mycobacteriologists [9, 15, 22, 26, 28] have reported positivity ranging from 0% to 33.3%, compared to 1.6% in the current study. Different authors [15, 22, 26, 29] reported varying isolation rates on LJ from urine samples of GUTB patients ranging from 11.1% to 50%. Present research work records 11.5% positivity of urine samples of GUTB patients, a finding similar to Iqbal et al . This is in contrast to the experience of Rishi et al and Siddiqui et al [9, 28], who could not isolate MTBC from any urine specimen on LJ.
In 2011 a study conducted reports 62.5%  MTBC from bone marrow aspirates whereas Subramani et al  reported 0%, similar to the present study, which included a single bone marrow specimen which was both LJ and MGIT negative. In the present research MTBC did not grow on any LJ slants inoculated with pericardial fluid and synovial fluid. Siddiqui et al  had similar experience. This is in contrast to reports by others [26, 31, 17, 22] who recorded a positivity of 83.4%, 39.13%, 33.3% and 16.7% respectively from synovial fluid. Subramani et al  could not isolate any MTBC from CSF, which is similar to the finding in the present study. Some researchers [15, 29, 32, 33] recorded CSF positivity varying from 10.9% to 25%.
The present research records culture positivity of 5.8% by LJ and 11% by MGIT from 565 EPTB specimens. Panicker et al  reported that 33% have grown in both LJ and BACTEC 460, but 53% of LJ negative cases were detected by only by BACTEC. Similarly in our study 48.4% have grown in both LJ and MGIT, whereas another 48.4% have grown exclusively in MGIT which would have been missed if only LJ is used for culture.
Mycobacteriologists have documented 5.7% to 25% culture positivity using automated systems from pleural fluid [27, 34, 28]. The present study reports additional 5% positivity using MGIT automated system from pleural fluid compared to LJ. Studies conducted in lymph node aspirate reports 18% culture positivity by LJ and an additional 8.2% by MGIT.  In another study BacT/ALERT documented 32.6% from LJ and 43.1%  in MGIT. Using BACTEC 460 culture for lymph node aspirates/cold abscesses 33.5%  positivity was reported. In the present study MGIT grew an additional 8% MTBC from lymph node aspirates/cold abscess samples compared to LJ. The isolates included 16 M. tuberculosis; one isolate of M. bovis and two isolates M. scrofulaceum.
Using BACTEC automated systems a moderate rate of positivity was reported – 17.4% from ascitic fluid.  This present study also records 4.8% positivity of ascitic fluid from MGIT, which is similar to other studies . Urine specimens showed positivity of 11.5% on LJ and an additional 4% positivity by MGIT, but Siddiqui et al  could not isolate any MTBC from urine specimens. Maurya et al  reported 3.1% positivity using BACTEC culture from synovial fluid. In the present study all synovial fluids were negative by MGIT, which is similar to another study. 
Maurya et al  reported 7.2% positivity from gastric aspirates, 6.7% from biopsy materials. In the present study, while all the gastric aspirates were negative by LJ and MGIT culture, biopsy materials grew M.tuberculosis 20% each on LJ and in MGIT. From pericardial fluid 2.4% positivity and from bone marrow aspirate 4.2% was reported by Maurya and his team.  Siddiqui et al  could not isolate MTBC from pericardial fluid in MGIT, which is similar to the present research work. Various authors [36, 37, 34, 28] have reported CSF positivity in automated systems ranging from 5.4% to 27.4%. The present study has not recorded any positivity from CSF either from LJ or MGIT.
Among the different clinical specimens processed, MTBC isolated from Lymph node aspirate/Cold abscess and TB spine was found to be statistically significant. ( p <0.01) Proportions test was applied to analyze the isolation of MTBC from EPTB cases. Isolation using MGIT is superior to LJ (p<.0001). But this study records a low rate of isolation from body fluids. Use of large volume body fluids might increase the positivity.
Several researchers [4, 7, 12″,15, 16, 20, 23, 25, 28, 32, 33, 37, 38] have reported the mean detection time or the turnaround time for LJ ranges from 24 days to 42 days. Studies [20, 32, 33, 37, 38, 42] have reported the TAT for automated mycobacterial culture systems ranging from 8.24 days to as high as 21.8 days for EPTB specimens. In the present study the normality of the data was tested using Shapiro-Wilk, and found highly significant (p= .000). Hence median was reported for MGIT as 16.5 days ranging from 6 -56 days, with a standard deviation of 12.4. For LJ the normality of the data was tested using Shapiro-Wilk, and found no significance (p=0.07). So, the data follows the normal distribution and hence the mean was reported for LJ as 28.7 days ranging from 10 -56 days with a standard deviation of 8.76. This study thus shows that automated liquid culture system MGIT is superior to LJ in terms of sensitivity and turnaround time. However, including liquid media like MGIT and egg based solid media like LJ together will increase the percentage positivity.