The Detection of Airborne Mycobacterium tuberculosis Using Micropore Membrane Air Sampling and Polymerase Chain Reaction^* FREE TO VIEW

During the past decade, the number of reported cases of Mycobacterium tuberculosis (MTb) in the United States and the rest of the world has grown steadily.¹ Decreased public funding for tuberculosis control measures, increases in the homeless population, and the emergence of HIV during this time period are all cited as potential causes.^1,2 However, our understanding of the transmissibility and infectivity of MTb has not kept pace. Studies involving guinea pigs performed by Riley et al³ in the late 1950s were among the first to offer scientific proof of the airborne spread of the disease. But in the nearly 40 years since these classic findings, there have been only modest advances in the understanding of the airborne infectivity of tuberculosis.

MTb bacilli are carried on airborne droplet nuclei produced by aerosolization that can occur from coughing, talking, or even singing.^4,5,6 The most clinically significant respiratory droplets measure between 0.5 and 2.0 μm. Larger droplet nuclei settle rapidly. The smaller droplet nuclei are rapidly dispersed and can remain suspended for an extended length of time, though once settling on surfaces, they do not reaerosolize and are no longer considered infectious.⁷ Transmissibility of MTb infection appears to be dependent on several factors. Among these is the quantity of MTb in the sputum, the radiologic extent of disease especially cavitary lesions, cough frequency and other personal habits, the closeness and duration of exposure, host susceptibility factors, and the presence or absence of specific lesions such as laryngeal tuberculosis.⁸ Effective chemotherapy markedly reduces the release of infectious aerosolized particles.⁹

Droplet nuclei are small enough to bypass mucociliary defenses and settle out in terminal alveoli. Here, they multiply and infect adjacent lymph nodes and are eventually disseminated to the bloodstream. In most immunocompetent patients, the development of cell-mediated immunity after several weeks aborts the infection and prevents development of active disease. Individuals with compromised immune systems (eg, HIV-positive patients) may develop acute progression to active disease within weeks. The overall risk of reactivation of latent infection peaks at 2 years after exposure and is 5 to 10% over a lifetime,¹⁰ though in HIV-infected patients, this figure is substantially higher.¹¹

Airborne droplet nuclei may contain as little as a single tubercle bacillus, yet because of their prolonged period of suspension, it may be reasonably assumed that they can be filtered from the air onto a porous medium, where, using an extremely sensitive technique such as polymerase chain reaction (PCR), even a single organism may be readily detected. Membrane filters (Micropore; 3M Health Care; Minneapolis, MN) are a relatively new technology involving the manufacture of durable polycarbonate membranes with extremely small (0.01 to 20μ m) pores. These membranes have a variety of industrial uses ranging from water filtration to air quality analysis. Filtration and collection of microorganisms from the air are possible due to the extremely small size of the pores. Air filtration sampling has been commonly employed to investigate industrial exposures, but its potential medical uses have only recently been explored. In this study, we wish to introduce a novel, noninvasive application of this technique to detect the environmental presence of aerosolized MTb. This rapid technique combines air filtration and PCR. The technique has been used by us previously to successfully document the presence of aerosolized MTb in our first patient,¹² and to detect airborne Cytomegalovirus.¹³ Others have used similar setups to document aerosolized Varicella-zoster virus¹⁴ and Pneumocystis carinii.¹⁵ To our knowledge, this is the first large study reporting on the rapid noninvasive detection of airborne MTb in multiple patients through the combined use of air filtration and PCR.

The patients selected for this study were identified with preliminary clinical diagnoses of MTb pneumonia based on history, symptoms, physical examination, and chest radiograph. These patients were inpatients of one or more of the affiliated teaching hospitals of the University of South Florida College of Medicine, Tampa, FL. All patients were placed in negative pressure respiratory isolation rooms and had three, consecutive, daily sputum samples submitted for acid-fast bacilli (AFB) staining and culture.

The air samples were taken from the negative pressure isolation rooms of patients selected for the study. The air was filtered through a 37-mm-diameter, 0.2-μm pore polycarbonate track etch membrane (Poretics Corp, Livermore, CA). The membrane filter holder was a 37-mm disposable styrene acrylonitrile two-piece assembly that holds the membrane filter. The filters have been loaded into the membrane holders with reinforcement backing in a class 100 clean room at the manufacturer to ensure sterility. The filter holder was directly connected to a battery-operated pump (Escort Sampling; Hazco Services Inc; Dayton, OH). The pump and filter apparatus were placed within 1 m of the patient’s bed on an adjacent nightstand. Pump and filter were calibrated with a soap bubble apparatus (Gilibrator) to filter air at the rate of 2.0 L/min. Room air was filtered through the membrane for a 6-h period. After sampling, the membrane filter was recovered by removal from the membrane holder using sterile technique under an aseptic hood and stored at −70°C.

Filter membranes were minced under sterile conditions and the fragments placed in sterile tubes containing 1 mL of Buffer AE (Qiagen Inc; Chatsworth, CA) and approximately 60 mg of 0.3-μm diameter glass beads (Fisher Scientific; Pittsburgh, PA). Tubes were placed in a microprocessor-controlled ultrasonic water bath (Lab-Line model 9331; Lab-Line Instruments Inc; Melrose Park, IL) and sonicated at 35 Khz for 30 min.

A one-tube nested PCR amplification scheme was carried out on 10-μl aliquots of the sonicated product according to the method of Wilson et al,^16,17 with some modifications following procedure optimization in our laboratory.¹⁸ The target for the PCR was the insertion sequence IS 6110 that is present in members of the MTb complex.^19,20 All reactions were run with positive and negative controls; positive controls consisted of dilutions of culture-proven MTb clinical isolates and negative controls consisted of all PCR reagents without MTb DNA. The concentration of dsDNA within these MTb culture lysates was measured with an RNA/DNA spectrophotometric calculator (Gene Quant II; Pharmacia Biotech; Piscataway, NJ). The concentration obtained was then divided by 4.0 femtograms, the average amount of genomic material within a single organism. This calculated value represents the number of organisms of MTb (genomic equivalents) within the culture lysate. Outer oligonucleotide primers Tb294 (5-ggacaacgccgaattgcg-aagggc-3′)^16,17 and Tb850 (5′-taggcgtcggtgacaaaggccacg-3′)^16,17 were used at a concentration of 200 nM (DNA Synthesis Laboratory; University of Florida; Gainesville, FL), while inner oligonucleotide primer Tb670 (5′-agtttggtcatcagcc-3′)^16,17 was used at a concentration of 10 μM (DNA Synthesis Laboratory; University of Florida; Gainesville, FL). The second inner oligonucleotide primer Tb505 (5′-acgaccacatcaacc-3<29)>^16,17 was used at a concentration of 20 μM (Oligos Etc Inc; Newton, CT), following multiple experiments during the optimization stage that showed it to be a more labile oligonucleotide than the others. PCR amplification was performed in a thermal cycler (Biometra TRIO-Thermoblock; Biometra Inc; Gottingen, Germany) programmed to amplify DNA in two stages. The first stage comprised 30 cycles of denaturation (93°C × 1 min), primer annealing (65°C × 2.5 min), and extension (72°C × 8 min). A 580-bp product was generated by the outer primers. The second stage consisted of 20 cycles of denaturation (93°C × 1 min), primer annealing (50°C × 2.5 min), and extension (72°C × 8 min), where product from the first 30 cycles became target for the inner primers. At the annealing temperature of 50°C, inner primers preferentially bind to their complementary target sequence and create a 181-bp product. To prevent contamination, preparation of reaction mixtures and detection of products were each performed in separate rooms. In addition, well-recognized approaches to decrease the risk of amplicon carryover such as aerosol-resistent pipette tips, frequent glove changes, master mixes, and single-tube reagent aliquots were used throughout the procedure.

A probe (5′-cgcaaagtgtggctaaccctgaaccgtga-3′) complementary to a 30-base sequence located within the inner, 181-bp, amplified product was designed in our laboratory. The probe was end-labeled with [(-³²P] adenosine triphosphate (DuPont New England Nuclear; Boston, MA) as previously described.^12,13 Hybridization mixtures were prepared by combining the labeled probe with formamide (Clontech; Palo-Alto, CA), sodium chloride, sodium phosphate, EDTA (SSPE) buffer (Fisher Scientific; Pittsburgh, PA), and Gel Loading Dye II (Ambion Inc; Austin, TX). Equal volumes of PCR amplicon and hybridization mixture were mixed and hybridized by heating to 95°C for 10 min followed by incubation at 37°C for 1 h. Hybridized samples were electrophoresed at 200 V for 45 min on 10% polyacrylamide mini gels (Bio-Rad Laboratories; Hercules, CA). Gels were sealed in plastic wrap, placed in autoradiography cassettes (Fisher Scientific; Pittsburgh, PA), overlaid with x-ray film (Kodak XAR; Eastman Kodak Company; Rochester, NY), and cassettes were stored at −70°C for 24 h to generate autoradiograms. The film was then processed in an automatic developer. A 100-bp reference ladder (GibcoBRL; Gaithersburg, MD) was run with patient samples on polyacrylamide gels.²¹ This reference ladder displays band lengths in 100-bp increments ranging from 100 to 1,500 bp.

Positive and negative controls were run with each sample. Positive controls consisted of purified dilutions of MTb culture lysates. Serial dilutions subjected to PCR analysis (data not shown) revealed an assay sensitivity of less than one organism of MTb (Fig 1). As many as four products of different lengths are obtained following polyacrylamide gel electrophoresis, representing the four possible primer pairings in the one-tube nested PCR reaction.^16,17 Although all four bands hybridize with a DNA probe homologous to a sequence within the inner product, the strongest product is the inner, short amplicon of 181 bp in length. Examples of polyacrylamide gels obtained from positive and negative patient samples are shown in Figures 2 and 3.

A total of 10 patients were sampled over a 12-month period (Table 1). The median age was 37.5 years. Seven of 10 patients had positive AFB cultures for MTb. PCR results correlated with sputum cultures in six of the seven patients who had positive cultures. MTb was detected in patient C’s room air 22 days after therapy initiation, suggesting he was still expectorating tuberculosis bacilli into his room environment> 3 weeks after starting therapy.

Only one patient in the group had positive cultures and a negative membrane air sampling with PCR analysis (MAS/PCR). Patient G was a 37-year-old HIV-positive Ethiopian man who presented with an interstitial infiltrate and a right pleural effusion. Equipment problems forced us to abbreviate his sampling period to 120 min.

Three of the study patients had negative sputum cultures. All also had negative MAS/PCR runs. Patient H was a 30-year-old HIV-positive individual with a CD4 count of <50 admitted to the hospital with fever, chills, and weakness. He was thought to have disseminated Mycobacterium avium complex, although this was not cultured in blood or sputum during his hospital admission. Patient I was a 27-year-old HIV-positive woman with a CD4 count <50 who presented with a left lung cavitary lesion, fever, and a productive cough. No culture-proven source of her infection was identified during her hospitalization and she responded to antibiotic therapy. Patient J was a 36-year-old HIV-positive man admitted to the hospital for mental status changes. No pulmonary source of infection was identified in this patient.

The results of smears for AFB were also noted in Table 1 and are of interest. AFB smear results were positive for patients A, C, D, and E, and negative for patients H, I, and J. Patient B had a history of positive smears and cultures for tuberculosis, and given his clinical presentation at the time of present hospital admission, he was placed on a regimen of treatment presumptively. However, of interest, patient F had been smear positive on August 23, 1996, but was smear negative on October 6, 1996, which was the date of air sampling. Patient G, in whom equipment problems forced us to abbreviate the sampling time to 120 min, only had rare AFBs seen on smear on the day before and the day after the air sampling.

We have performed air filtration of patient rooms in cases suspicious for pulmonary tuberculosis, and have shown that PCR amplification of the hospital air samples can detect the offending pathogens following as little as a 6-h filtration time. Sampling periods were set at an arbitrary standard of 6 h because it was believed that this was the maximum practical interval with which both sampling and specimen processing on the same day could be accomplished. Certain environmental variables existed for the study and included the following: (1) the rate of room air exchanges per hour in the different rooms in which patients were sampled; (2) the relationship between the last positive AFB smear and culture and the sampling session; and (3) the exact distance between the study subject and the membrane/pump assembly (although this was kept between 1 to 1.5 m from the patient’s bed on an adjacent nightstand).

Prior to the start of the study, it was not known whether Micropore air filtration at the modest rate of 2 L/min could recover MTb organisms successfully. It was recognized that in a negative pressure isolation room with an average of 6 to 10 air exchanges per hour, HEPA filters within the air handling system were expected to remove 99.9% of all airborne contaminants within 69 min.⁷ It would be expected that nearly all droplet nuclei would be removed by ventilation and filtration systems. It is possible that the close proximity of the filtration pump to the patient’s bed facilitated in the filtration of infectious droplet nuclei prior to its recirculation. In a previous MTb outbreak²² on a commercial aircraft cabin with HEPA filtration of recirculated air, purified protein derivative skin test conversions occurred in passengers who were in physical proximity to the index case, whereas passengers in the forward part of the aircraft were not exposed.

Smear-positive patients possess AFB concentration within their sputum of about 10⁶ to 10⁷ AFB per milliliter.⁹ A rapid decline in mycobacterial counts is expected with the institution of therapy. Most sputum-positive patients were sampled within 3 days of onset of a four-drug regimen. Patients who were sampled >3 days after therapy initiation who had positive membranes tended to be nonprimary presenters, ie, they were previously diagnosed and had either failed to respond to therapy or had been noncompliant.

Most membrane positive patients in the study (ie, five of six) had documented positive sputum smears on or after the date in which they were subjected to MAS/PCR. Two patients had documented positive sputum smears at 1 and 5 days prior to the sampling session, respectively. The significance of sputum smear/culture positivity in relationship to MAS/PCR result is difficult to assess, as patients did not provide sputum specimens daily but only periodically, but it is reasonable to presume that smear/culture positivity at or near the sampling date would increase assay sensitivity. Of interest is the fact that one of the patients, patient F, had been sputum smear positive several months prior to the present hospital admission, but was smear negative on the day of air sampling. This presumes that the lower level of AFB present in his sputum was still at levels detectable by a very sensitive amplification technique. Also of interest is the fact that patient E’s empty isolation room was sampled approximately 3 weeks after the previous sampling session and 7 days following hospital discharge. PCR results on the membrane were negative, suggesting that no “background” level of circulating MTb existed in this frequently used isolation room.

Careful handling and storage of membrane filters was essential for MAS/PCR accuracy. Six additional patients sampled after the initial 10 were removed from the study because their membranes were not handled according to the exact stipulated protocol. Plastic containers with membranes inside were frozen prior to the DNA lysis procedure rather than being processed immediately following the filtration process. It is speculated that freezing of the membranes within their plastic holders, followed by thawing and refreezing of the membranes prior to removal from the membrane holders (a modification of original protocol to facilitate their processing), may render the MTb complex DNA on the membrane surface less available for amplification. Results on these membranes were negative. This second set of patients had also been sampled for various intervals ranging from 4 to 11 h and it was believed that this variability could also obscure our ability to reach conclusions once too many uncontrolled variables were included. In addition, therapy had already been underway at the sampling time of these patients.

Potentially, Micropore MAS/PCR could have numerous epidemiologic applications. The background levels of circulating MTb within hospital wards, microbiology laboratories, tuberculosis clinics, and other similar institutions (eg, correctional facilities) is not well characterized. In such settings, the use of a technique for detecting the presence of circulating MTb complex DNA could be used to document environmental contamination in nonisolated areas.

Hospital respiratory isolation rooms are designed to be effective barrier spaces for the containment of highly infectious respiratory pathogens. Air handling mechanisms and HEPA filtration units are designed to minimize levels of circulating microorganisms. We have shown, however, that significant detectable quantities of AFB are still present within the air of these rooms. Little is known, however, concerning how long AFB organisms may circulate within the isolation room prior to being filtered by ventilation or air handling mechanisms. We will soon be examining these issues and several others through additional studies. In certain noninstitutional settings, there has been great public concern involving the potential infectious risks associated with a group of individuals sharing a confined airspace with poor air circulation. Public transportation has received particular scrutiny. Most notably, Kenyon et al²² presented a well-researched epidemiologic investigation documenting the transmission of MTb aboard a transoceanic commercial airline flight. Transmission was documented by purified protein derivative conversion and appeared to have occurred to passengers in close proximity to the index case, rather than via recirculated air. Routine surveillance of commercial airline flights using the membrane sampling technique could improve our understanding of the relative incidence of exposure incidents within the commercial airline system.

Several limitations to the use of MAS/PCR exist. A potential limitation is the lack of a means for assessing the viability of airborne AFB. Because of the limited size of the membranes, the investigators were unable to culture portions of the membrane filters to determine whether any viable MTb organisms remained. Given the environmental exposure and desiccating effects of 6 h on the air filter, it is unlikely that any living organisms could have remained. Certain instruments, such as the Anderson air sampler (Graseby; Smyrna, GA), are commonly used for collecting and culturing airborne microorganisms and could potentially be used concomitantly for this purpose. A quantitative assay of the concentration of airborne AFB as detected by MAS/PCR using a curve of known controls, not performed at the time, could permit specific quantification of mycobacteria and could have many potential applications, such as helping us gauge the level of potentially infectious mycobacteria in room air. Despite some of the limitations presented in this study, we believe that MAS/PCR could be a clinically useful tool in the noninvasive diagnosis of infectious tuberculosis in patients who cannot physically or willingly provide sputum for analysis. Patients could very easily be sampled either within their isolation rooms or in controlled ventilation booths to allow for optimal collection of exhaled organisms. However, the greatest application of this technology is in the area of hospital epidemiology and infection control, and in the documentation of the transmissibility of MTB within noninstitutional settings where groups of people are involved. More must be known, however, regarding the most ideal sampling interval and filtration rates for clinical and other uses, and future studies will address these issues.

ACKNOWLEDGMENT: The authors would like to acknowledge Dr. Richard McClusky for his preliminary studies on detection of airborne cytomegalovirus by PCR in our laboratory and Dr. Santo V. Nicosia for his support in the creation of the molecular laboratory.