Study setting and design
Leftover portions of sputum specimens submitted for routine tuberculosis diagnostic investigations were collected from the Microbiology Laboratory at Mulago National Referral Hospital for inclusion in this study. Sputum specimens were collected from TB suspects; follow up specimens for the purpose of treatment monitoring were excluded from this study.
Laboratory testing
All sputum specimens with at least 2 ml volume were included in this study. Direct FM was performed at the Mulago Hospital Microbiology Laboratory. Specimens were refrigerated immediately after preparation of routine direct FM smears and transported to the FIND laboratory on the day of collection.
An additional direct sputum smear was stained by Ziehl-Neelsen method, and examined according to World Health Organisation (WHO)/International Union Against Tuberculosis and Lung Disease (IUATLD) guidelines [8]. Grading charts were used for all smear readings. Quantification of smear results was as follows: 1-9 acid-fast bacilli (AFBs) per 100 fields (report exact count); 10-99 AFBs per 100 fields (1+); 1-10 per field (2+) and >10 per field (3+).
Specimens were then homogenized and split for concentration fluorescent microscopy using either magnetic bead or conventional NALC-NaOH decontamination. Following aseptic addition of 6-8 sterile 3 mm diameter glass beads, specimens were vortexed for 30 seconds to homogenize the specimens. Two 1 ml portions of homogenized sputum were transferred to separate sterile tubes using a sterile transfer pipette. Tubes were coded to ensure that the technicians performing and reading the results of each method were not aware of the results of alternative reads. On alternate days the first portion was transferred either to the tube for magnetic bead processing or to the decontamination/centrifugation tube to reduce sampling bias. Culture results were also interpreted independently from the smear results. Un-blinding was undertaken at the end of the study at the time of data analysis. All manipulations with potentially infectious clinical specimens were performed in a Class II safety cabinet in a BSL 3 Laboratory.
Standard decontamination and centrifugation method
One tube was decontaminated by standard N-acetyl cysteine (NALC)-sodium hydroxide (NaOH) procedure according to standard methodology (1.5% final concentration NaOH) [9]. Centrifugation was carried out at 3000 g for 20 minutes, using sealed aerosol-free buckets, which were only opened inside the biosafety cabinet. Following buffering and centrifugation 0.1 ml phosphate buffer pH6.8 was added to re-suspend the pellet.
A smear was prepared from 1 drop (approximately 40 μl) of the decontaminated suspension and allowed to air dry. Slides were stained with Auramine O and read according to WHO/IUTALD guidelines [8]. Fluorescence microscopy was performed using Primo Star iLED microscope (Carl Zeiss MicroImaging GmbH) at 400× magnification. For quantification of fluorescent stained smears, 1-19 AFBs per length (report exact count); 20-199 AFBs per length (1+); 5-50 AFBs on average per field (2+); >50 per field on average (3+).
Following smear preparation, an additional 0.5 ml phosphate buffer was added to the processed sputum suspension and mixed using a vortex mixer. Of this, 0.5 ml was used to inoculate MGIT culture and 0.1 ml to inoculate a Lowenstein-Jensen medium slant. Positive cultures were identified as M. tuberculosis species using the Capilia TB Neo assay (Tauns, Numazu, Japan).
Magnetic bead processing method
One tube was processed according to the prototype magnetic bead processing protocol (Microsens Medtech Ltd., London, UK) (Figure 1).
The following components were provided by the manufacturer: bleach tablets (for thinning solution), adhesive wells, capture buffer, ligand-coated magnetic beads and magnetic workstation.
A master thinning solution (10X concentration) was prepared by dissolving 1 bleach tablet in 100 ml distilled water. On a daily basis, this master solution was diluted 1:10 in distilled water to make a working concentration. Decontamination agent consisted of 4% N-acetyl cysteine and 6% sodium hydroxide (Sigma Chemical Co.) dissolved in distilled water. Decontamination agent was prepared daily.
Adhesive wells were firmly attached to microscope slides according to the manufacturer's instructions. Slides were labelled with the appropriate specimen codes and placed on the magnetic workstation (Figure 2).
One millilitre of prepared thinning solution was added to the 1 ml sputum specimen. The top of the sputum container was closed tightly before thoroughly agitating the solution and sputum and incubating for 15 minutes at room temperature.
One ml of Decontamination Agent, 1.0 ml of Capture Buffer and 2 drops (80 μl) of magnetic beads were added to each capture tube (labelled with appropriate specimen code), and mixed gently before the tubes were transferred to the magnetic workstation.
The thinned sputum was mixed with a vortex mixer and 1 ml transferred to the appropriate capture tube. Tubes were incubated for 10 minutes at room temperature to allow mycobacteria to be captured onto the beads. The entire volume of the resulting suspension was transferred carefully to the appropriate slide well using a sterile Pasteur pipette. After 1-3 minutes for capture of the beads on to the surface of the slide (visible as clearing of the liquid around the magnet), the liquid was removed using a transfer pipette.
The wells were gently filled with distilled water. The water was then removed using a transfer pipette. As much water as possible was removed to shorten the drying time required for the slides. Adhesive wells were removed from the slides using forceps, and the slides allowed to air dry at room temperature in the biological safety cabinet.
Magnetic bead slides were stained with auramine and examined as described for concentrated FM smears.
Slides were stored in a closed slide box for re-checking of discrepant results. Specimens with a discrepancy between the magnetic bead FM result and culture were blindly re-checked by a second reader.
Examination time
A selection of corresponding direct ZN, magnetic bead and concentrated FM slides were examined for varying lengths of time (0.5, 1, 3 and 5 minutes) and the results obtained at the different time points were recorded. Slides selected were those in which any of the methods were positive, plus a random selection of slides that were negative by all methods. The incremental gain in positive results was calculated at each time point, based on the total number of positives recorded by each method.
Assessment of hands on time and complexity
The total time required for processing samples according to the magnetic bead protocol, as well as the hands on time, was recorded for 3 batches of between 8-10 specimens. A qualitative assessment of the complexity of the method compared with direct ZN and concentration by centrifugation was made.
Sample size
A sample size of 288 specimens was powered to detect a 20% difference in sensitivity between magnetic bead-FM and direct ZN, assuming sensitivity of direct ZN to be 50% and magnetic bead to be 70% compared with culture. The estimated prevalence of culture positives in the study population was 20%. A drop-out rate of 15% due to loss of culture results due to no growth or contamination was included in the sample size calculation. It was estimated that the sensitivity of concentrated FM and magnetic bead were similar and the study was not powered to detect a small difference in sensitivity between these methods.
Data analysis
Statistical tests were performed using Intercooled STATA 8.0 software (Statacorp LP, College Station, TX, USA) and Microsoft Excel 7.0 (Microsoft Corporation). Results were considered significant at p < 0.05.
Ethical considerations
The study was approved by Makerere University, Department of Medicine, Mulago Hospital and Infectious Disease Institute Ethical Committee. Patient identifiers were removed from sputum specimens prior to transfer to the FIND laboratory. Study data were kept in a secure fashion, with access restricted to study personnel only.