Sniffing out tuberculosis - TB breath tests under investigation

Animal-based techniques

Animals can smell TB too. In fact, an organisation called Apopo (see www.apopo.org), which in the late 1990s started using rodents to sniff out land mines in Africa, has now begun training them to smell TB in human sputum samples.

It works surprisingly well and they do it very quickly.

In one blinded study with 819 culture confirmed samples (with 67 culture positives and 752 negatives), one trained Giant African Pouch Rat was able to detect 73% of the TB positive specimens and two rats detected 86.5% of the positive samples. Specificity (as compared to culture) was 89.1%.

What’s most amazing is that the rats screened all 819 specimens in four hours and forty minutes. In contrast, many small peripheral labs in resource limited settings can have trouble screening more than 20 smear microscopy slides in one day (although WHO sets as a standard 40 per lab technician). According to the Apopo site, one rat can screen 100 slides in 20 minutes.

“There are some obstacles to implementation though: the training and maintenance of rats,” said Dr McNerney. “Training rats takes months and I think it’s quite a special relationship between the rat and the trainer so I don’t know how easy it’s going to be to roll this out to the world.”

She noted that even though these rodents are “friendly little chaps” nothing like sewer rats, “the other issue is acceptability - do TB patients want to be diagnosed/screened by a rat? And apparently this has been an issue in Tanzania.”

Honeybee’s are also quite sensitive to smell and can be trained to detect TB. They are a lot cheaper, too, according to Dr McNerney, who said that the London School of Hygiene has been looking at bees. “We've not done any clinical work with our bees. We're looking for funding at the moment.”

Although it is difficult to imagine diagnosis with animals being implemented on a wide scale, the work has provided a sort of proof of concept that smelling TB can work. But nonetheless, they might be worth reconsidering given that the attempts to duplicate ‘smelling’ with a lab test are all rather high tech.

What if you could just breathe into a device and within minutes, a healthcare worker could tell whether you have tuberculosis (TB) — without even having to consult the lab? Although such a simple point-of-care breath test for TB doesn’t exist yet, several groups are working towards that goal, according Dr Ruth McKerney, of the London School of Hygiene and Tropical Medicine. She gave an overview of several of the approaches to “smelling TB” being investigated at the STOP TB Working Group on New Diagnostics/FIND meeting in Paris.

Although smelling TB might sound far fetched, it is actually one of the most ancient TB diagnostic techniques and was even described by Hippocrates. In addition, “a lot of people tell you that TB hospitals have a funny smell,” said Dr McNerney.

The zNose

One possibility, which isn’t exactly simple but is at least portable, is called the zNose, which utilises ultra-high speed gas chromatography but which has been developed and field-tested by Electronic Sensor Technology (a company in California) for a wide range of applications (see www.znose.com). Nevertheless, given its cost and complexity, even a portable zNose would probably not be permanently housed at the most peripheral laboratories in resource-constrained settings. Instead, remote sampling/concentrator devices could be used at the remote site to collect samples of breath or headspace vapours. The device, which is smaller than a pack of cigarettes and battery operated, concentrates VOCs onto little resin-coated sticks (called Slicksticks). These Slicksticks can then be safely stored or transported at ambient temperature to wherever the closest zNose is.

Using the Slicksticks and zNose, researchers from the London School of Hygiene were able to detect TB and distinguish between TB and other environmental TB within a few minutes, according to one poster presentation (McNerney). Another plus was that the reagent costs are less than US$ 1 per specimen. Clinical diagnostic studies explore the further development of a TB diagnosis application for the zNose are being planned in collaboration with the University Teaching Hospital in Lusaka, Zambia.

Electronic noses

For example, electronic noses (e-noses) are instruments that recognize odours (or volatile compounds in the air). The technology was first pioneered by the food industry and has been further developed for security purposes (to detect weapons, drugs or explosives for example).

There are several different e-nose formats that all have the same basic setup, which essentially involves some sort of sensor array that detects volatile compounds in breath or from a sample of air collected from over a specimen (referred to as headspace vapours) and then some conductive property of the array converts what it gathers into electronic signals for data processing.

The data processing step is very difficult because of the many molecules than can be in a sample of air/gas. Expensive and sophisticated software (and a fair amount of computing power) is needed to perform pattern recognition and neural networking. In effect, the software must be taught to recognise the pattern of odours caused by an infection.

One e-nose, the Bloodhound sensor from Scensive Technologies UK (www.scensive.com) is currently being investigated by the University of Leads for TB and veterinary use (Mycobacterium bovis in cattle and badgers in England).

In a recent study, after being trained to recognise TB, and the Bloodhound sensor was able to discriminate between three different mycobacteria (with 91% specificity for TB) as well as other lung infections (Fend). Its sensitivity with culture-proven TB was 89%, and it could sense MTB when there were as few as 1000 bacilli in a specimen —significantly more sensitive than smear microscopy. The authors of the recent paper acknowledge that the method used in the paper is “not yet fully optimized for ‘field’ application.”

However, according to Dr Arend H.J. Kolk of the Royal Tropical Institute in Amsterdam, an improved version of this test is going into large-scale studies in Tanzania. “The elegance of the e-nose is you don't need sample treatment. It is very fast, and the electronic nose is ready after a few seconds so you can sample and sample and sample. So, the running cost is very low. In fact, we use just a sputum cup, and punch this and connect this with the e-nose and smell it.”

Another very high-tech approach, more precise than e-noses, is being developed by Dr Michael Phillips of Menssana Research in the US, using gas chromatography and mass spectrometry (GC-MS) to detect the specific volatile organic compounds (VOCs) associated with TB infection. In in vitro studies, he identified over 130 different VOC from cultured mycobacteria (Phillips). After inputting this data into very sophisticated fuzzy logic software to establish patterns associated with TB, he evaluated the breath of 42 patients hospitalised with TB (23 cultures confirmed) compared to 59 healthy controls. It worked almost perfectly, distinguishing between the ill and healthy patients, as well as those who were culture positive from those who were not.

“That’s very exciting, but there are some problems,” said Dr McNerney. “GC-MS is a wonderful instrument, but it's very sophisticated and very expensive and it's not close to patients. Again it's proved there's something out there to be detected; that some organic compounds and volatile compounds are there that can be analyzed. But what we have to have is a diagnostic test where you put it in a little box and give it to someone to go off and use. We need it to be something that can be calibrated and standardised, portable, affordable and easy to use. We want something simple.”

Bio-optical breathalyser device

Another device that is closer to the market is a bio-optical breathalyser device, developed by a Cambridge-based company, Rapid Biosensor Systems, (see www.rapidbiosensor.com). This is a ‘near-patient’ screening test for pulmonary tuberculosis that takes under five minutes.

The test does not really smell TB but rather uses an immunoassay system and fluorimetry to detect TB in expectorant — in a way, it’s more like a portable fluorescence smear microscopy lab where everything is contained in two small parts: a disposable cough sample collection device, which looks a little like an asthma inhaler and admits a nebulising spray; and a very compact battery-powered reader, with an LED and two buttons (stop and start), that the collection device plugs into.

Dr McNerney shared preliminary reports from a field trial in Gujarat, India. In the study, the device detected 62 pulmonary TB patients (25 smear positive/x-ray positive, 4 smear positive/x-ray negative, 21 smear negative/x-ray positive, and 12 smear negative/x-ray negatives).

“That's very encouraging,” said Dr McNerney, “they're picking up people who wouldn't be picked up by smear and they're doing this in five minutes, rather than your smear taking all day.”

A total of 111 non-tuberculosis patients were negative by the test. There were no false positives, but the test still isn’t 100% sensitive as it missed 35 culture positive cases (18 smear positives and 17 smear negatives).

But for a rapid screening test that can be performed without a technician and where the patient is sitting, this is still quite promising. If these data are borne out by other studies, and the platform proves robust in the field, and the technology is affordably priced, this test could be used at more remote sites without highly trained lab technicians. For the bulk of patients with pulmonary disease, it could decrease the number of clinic visits or the necessity to visit a remote peripheral lab. The results could also be used to get people on anti-TB treatment within hours of their presenting to the health clinic.