Connect to share and comment

A diverse look at global health issues.

Using giant rats to detect TB

A Q&A with Dr. Poling of APOPO about training giant pouched rats to detect tuberculosis in sub-Saharan Africa (VIDEO)
An African giant pouched rat; used to detect tuberculosis in Tanzania. (Brian Johnson/APOPO/Courtesy)

You might know them as pesky rodents, or lab animals— but for Bart Weetjens, rats were the cute and cuddly pets he grew up with in Belgium.  Years later, Weetjens, a product engineer, was working on an analysis of landmines in sub-Saharan Africa. He quickly realized that landmines were a dangerous and costly problem for the people living there.  He thought through the problem, brainstorming ways to solve the problem cheaply, and came across an article about gerbils. Weetjens remembered the rats of his childhood, and an idea formed.  

APOPO, the organization Weetjens founded, has been using African giant pouched rats- a noctural species native to sub Saharan Africa- to detect landmines since 2000.  In 2003, they began using the rats to detect tuberculosis, which has a distinct smell the rats can pick up on. APOPO has been working with DOTS [Directly Observed Treatment Short course] centers in Tanzania. DOT Centers are the internationally recommended standard for treating TB- the purpose of the center is both to diagnose and treat TB. A recent study, published in the Pan African Medical Journal, reports APOPO's most recent amassing of data, which shows that using the rats works.  

GlobalPost had a conversation with Dr. Alan Poling, a professor at Western Michigan University, who is part of APOPO's team.  He speak about training the rats, where he believes the science is leading, and next steps for APOPO.

How did you come across the concept of using a rat to detect TB? Is it because you'd used them for the detection of landmines?

The notion of using the pouched rats to detect tuberculosis didn't originate with me, it actually came from Bart Weetjens who's the founder of APOPO, the organization that does the TB detection work. Essentially Bart and his colleagues had been using the rats to detect landmines, so they knew that the animals had an excellent sense of smell and they also knew that years ago physicians reported that when they were treating people with tuberculosis, sometimes they could detect a characteristic odor on the breath of those patients or in the sputum [mucus coughed up from the lower airways] of those people. So they reasoned that if humans, who have a relatively poor sense of smell, could detect tuberculosis at least occasionally, then perhaps the pouched rats could do it easily.

Why pouched rats as opposed to other animals?

The pouched rats were chosen because APOPO had already been working with them for landmine detection. The choice to use them there was based on several characteristics of the animals. One was that they have an excellent sense of smell. The other is that they live a relatively long time; they live for three-eight years in captivity, so they have potentially a long working life. They are native to sub-Saharan Africa and Baart and his colleagues intended to do their work their work in that area. So they thought that the rats should certainly be hearty, they could hold up to the rigors of the climate, the parasites and so forth. Also, the rats are big enough that they're easy to manage, but they're small enough not to activate any landmines when they walk across them.

How are the rats trained?

The training process is fairly elaborate and it starts with the socialization procedure. When the rats are about three weeks old, we start to handle them; they're handled extensively for about a month. People expose them to a huge variety of sights and smells, and sounds. They go for rides in vehicles, they go into people’s offices, they hear music, they see other animals.

This makes the animals gentle, and easy to handle. It also establishes humans as things of value —people feed the rats preferred foods; they love peanuts, they love bananas. 

They get used to being around people, and also they come to be vary resistant to startling; any new stimuli don't bother them and of course this is hugely valuable when you're using them in landmine detection.

After the socialization process, we start teaching them to detect the scent of mycobacterium tuberculosis. With all of this training, we use operant conditioning. Basically, the animals are rewarded for correct identification responses. We place the rats in a metal chamber, that has a little hole in the center of the floor. Beneath hat the hole, we place a little pot that contains a sputum sample that we know to be positive for mycobacterium tuberculosis (which is the bacillus that causes TB). The rats naturally poke their noses in the holes. As soon as the rats pokes her or his nose into the hole, we reinforce or reward that behavior with a bite of banana.

Gradually, we require longer and longer periods of the pausing with the nose in the hole, until the rats are pausing for four or five seconds. When the rat does that regularly, we start adding sputum samples that don't contain mycobacterium tuberculosis, and we never reward the animal for pausing above those samples. It's trial and error for the animal and soon enough they learn that pausing above a sample that doesn't smell like TB doesn't pay off, but that pausing above a samples that smell like TB does pay off. 

Watch a video of the rats in action: 

How are the rats being used?

We present samples [that have already been checked through microscopy] from the DOTS [Directly Observed Treatment Short course] centers to the rats. If the rat pauses for a specified period above a sample that was deemed TB positive by a microscopist, then we reward that response with a bite of food. If the rat pauses in the other sample, we don't do anything, but we record that response. These are the interesting responses— the DOTS centers miss a lot of TB positive smears, so if our rats detect a center negative sample as TB positive, then we do a recheck of that. We either do a second microscopy, or use a device called the Gene Xpert, and it actually looks for genetic material characteristic of mycobacterium tuberculosis. When we confirm that a rat’s positive, but DOTS’s negative sample actually confirms TB, then that's a new case indication.

This is called second line screening. The first line screen is done by the DOTS center, and then the rats do a second screen. They've increased new case detection by over 40 percent. They've looked at over 20,000 patients and significantly increased case detection.

Why are the rats not being used for a first line screening if they're able to identify more positives than microscopists?

Well I think they're probably two reasons that rats aren't being used more widely. One is we're just starting to accumulate a good database to actually show how well the animals perform. In medicine, (and rightly so) people don't accept any technique unless there's good evidence that it works and we're just now publishing the science that shows that the rats perform at least reasonably well.
I think another reason they haven't been widely adopted there is an understandable reluctance to use rats as a diagnostic. Many people view the approach as almost medieval. Medicine is characteristically high tech. Look at the Gene Xpert. You're actually looking at specific kinds of DNA and it's a neat shiny device, and it works quite well. The problem is that it's relatively expensive. And the rats seem primitive, hard to use, not appropriate. We hope that if we amass the data showing they are relatively accurate, they may be more used more widely.

How expensive is it for you to train a single rat?

It's hard to say. We can train a rat in half a year in half a year easily. And I would say that a TB detection rat could be trained for perhaps $2,000. [It actually costs 6,000 Euros and takes approximately 9 months to train a Mine Detection Rat. Time and costs are likely to be considerably less for a TB Detection Rat (because the smell is easier for them to detect than landmines), however a full cost analysis is yet to be conducted.]

How much would the machines cost?

Well, the machines vary in prices. A fairly simple Gene Xpert will cost in the neighborhood of ten thousand dollars. But the expense comes from actually doing the test. For us in Morogoro, Tanzania, it costs roughly $15-20 per test with the Gene Xperts. The cost per sample is very low with a rat. They're doing a lot more quickly. They just walk along and stick their noses in the holes. And so they can evaluate ten samples, in very, very little time.

How did you get involved in working with these rats?

Basically, a colleague sent me an email that contained an ad from APOPO. They were looking for a behavioral scientist who knew discrimination training in animals, a person who could play a lead role in teaching the rats to find landmines and detect tuberculosis and who was willing to work resource poor environment. I've been involved for about three years.

What's the next step for both you and APOPO?

The land mine detection work is moving forward. We're ready to have the rats go to work in Angola and in Thailand. So the logistics just need to be ironed out. The TB detection- we're looking at other possibilities. We may actually start using the rats in Mozambique, I'm not sure that'll come to fruition, but it's something that we're pushing forward. What I'd like to do is to make sure that we adequately evaluate the animal's performance and that we're able to produce an exportable product with defined and reproducible characteristics. So you can tell people that this is what you can expect from the animals. They'll find 8 out of 10 cases of active tuberculosis, or whatever it might be. They'll have a 20 percent false positive rate, or a two percent false positive rate. That's the sort of information that the medical personnel are going to need, because there are always alternatives. People want to know how much does it cost to use these animals. Is it a good investment?

Since we're really just developing the technology we haven't really done the good cost analysis. We can kind of make ball park estimates, but we haven't targeted the speed and training and maximal efficiency and utilization. We're sure that the rats do a good job relatively cheaply but just how cheaply we can't say.
The point of the rats is so that they can have a more speedy diagnoses…

Yeah! Two things you want in a resource poor environment- tenability, it has to be cheap enough that you can use it and you also want speed. You want to be able to detect tuberculosis and treat it as quickly as you possibly can. And we feel as though the rats are relatively accurate, and they're certainly fast. And their speed makes them relatively inexpensive.

Are these rats moving around? Can you go into one village, test all of the samples there, and then move on to the next?

You could do it that way, or what we've been doing is that we've been bringing samples to our lab. We bring the samples from Dar es Salaam, just because there's more people there, and it's more easy to get the samples you need for research. But, you could either have the rats in a centralized location and send people to get the sputum samples, or potentially you could have a mobile lab, that you could move about. We haven't pursued the logistics of the various options, we joke about putting a rat in a cage, and starting out through the hinterlands, testing people, but we haven't done that. And of course another part of that is there's no point in detecting tuberculosis unless you treat it.

Is that actually happening? Are people actually getting the treatment they need after they get detected?

Well, they are. Because we're working with the DOT centers. So, their job is to treat as well as to detect tuberculosis. When we make a new case detection the first thing that we do, is we call the DOT center and we say patient one of seven was found positive by our rats and the DOT centers call that person or track them down in some other way. But we haven't been providing treatment that's been coming throughout the DOT centers.