NAME: David Dowdy
POSITION: Assistant professor of epidemiology, Johns Hopkins Bloomberg School of Public Health
Since its introduction in 2009, Cepheid's Xpert MTB/RIF — a PCR-based assay that simultaneously detects Mycobacterium tuberculosis and rifampicin resistance in about two hours — and the benchtop GeneXpert system on which the assay runs have seen significant uptake in resource-poor areas of the world.
The company has not broken out assay sales in its financial reports, but reported in 2011 that it placed 418 GeneXpert systems through its so-called "high-burden developing country," or HBDC program, followed by 506 GeneXpert systems under the HBDC program in 2012.
This uptake has been aided by multiple peer-reviewed studies demonstrating that Xpert MTB/RIF has equivalent or superior sensitivity and specificity to traditional culture-based methods for detecting M. tuberculosis and rifampicin resistance, as well as a 2010 endorsement from the World Health Organization to make Xpert MTB/RIF the initial test in those cases suspected of multidrug-resistant or HIV-associated TB.
Despite the test's sensitivity, specificity, speed, and ease of use, questions remain about how much better Xpert MTB/RIF is than current diagnostic methods, and whether it is appropriate for use in all resource-poor areas of the world, particularly those lacking certain infrastructure required for proper operation and maintenance of the GeneXpert platform.
Researchers have begun investigating these questions. For instance, in April 2012 the National Institutes of Health awarded a team from the University of North Carolina at Chapel Hill nearly $600,000 for the first year of a three-year project to investigate treatment outcomes of South African drug-resistant tuberculosis patients diagnosed with the assay.
And, this month, a team led by researchers at Johns Hopkins University was awarded approximately $231,000 for the first year of a two-year comparative effectiveness study of Xpert MTB/RIF-based diagnosis in Uganda. In their study, the researchers will attempt to provide a framework of how certain characteristics of patients, providers, clinics, labs, and health systems affect the ability of Xpert MTB/RIF to diagnose and control TB.
PCR Insider spoke this week with David Dowdy, the Johns Hopkins epidemiologist serving as principal investigator on the project. Following is an edited transcript of that interview.
What was your impetus for conducting this study? From the grant abstract it sounds like you will be essentially doing a cost-benefit analysis of using the Xpert MTB/RIF test in resource-poor areas of the world.
The impetus behind this grant is that right now people are asking that exact question: 'Is it worth it to us to implement this test on a large scale?' For an individual hospital … that's one question, but when we're talking about district- or country-wide scale-up, the cost of doing that is quite substantial.
On one hand, this test is clearly much more sensitive than the current standard of care, which is sputum smear. It's also easier to use. It takes very little training to get someone to use this machine. But it is quite expensive. Up until this time, most assessments of the cost-effectiveness of this test … have largely taken the test in some sort of idealized scenario. For instance, they take someone walking into a clinic and ask, 'If we give this person a sputum smear versus a GeneXpert test, what would their likely outcomes be?' The challenge is that in real life it's not always that ideal setting. Lots of things can impede the scale-up of this test, whether it's difficulty in getting the test out to peripheral sites; or, the equipment breaks down and you don't have your Cepheid representative on the line right away; or, you can't afford the cartridges for a little while; and there is a certain period of time where you don't have any infrastructure for doing testing, et cetera.
I think there are also questions about how much patients and doctors want this test, and how they will use the test. For example, another real question is: Are the doctors already treating all these cases of TB and just calling them smear-negative, or are they really missing a large number of people?
The idea behind this grant is to try and capture some of these real-world operational characteristics that would be associated with scaling this test up to be able to make a better informed decision than just assessing what might happen in the ideal setting.
It sounds like the pricing of the test is the main issue here. Is there a cost threshold where this test might not even be worthwhile in, for instance, a remote village in a developing country?
The pricing, if anything, is the most clear of all these parameters in that there is a negotiated price per cartridge. If you're in an approved resource-limited setting, it's $9.98 a cartridge, and the machines sell for about $17,000. Again, this is only in the resource-limited, high-burden setting. You can't get those prices in the US, but you can get them in any other country in this program. It's transparent, and everyone knows what they're paying.
To be fair, I think that there has been enough donor support of this. For example, part of the reason the test costs $9.98 is that the Gates Foundation [along with United States President's Emergency Plan for AIDS Relief, the United States Agency for International Development, and Unitaid] agreed with Cepheid to buy it down to that price to make it available (PCR Insider, 8/9/2012).
We know what the price is, but what is likely to vary widely from one setting to another is how easy it is to implement. You think about trying to put one of these machines in 100 different clinics in a district — you've got to find a way to transport them out there; to do the quality assurance and maintenance once it's out there; you have to find a way to train your staff; you have to consider whether you are still going to keep the same infrastructure for sputum smears. Do you put this in a central lab and try to find a way to transport all the sputum specimens to this lab, and thus not have to buy as many of the machines and can keep them running at higher volume? Or, do we try to distribute them way out into the periphery where, for much of the day, they will sit there unused because there are not as many samples coming in; but, when that test comes in, you can get that test right away and start them on treatment the same day.
It's those kinds of things we are trying most to characterize.
This test received a well-publicized endorsement from the World Health Organization. I don't know how these endorsements are made, but it seems like they would have done this kind of analysis before giving the test a stamp of approval. Do you have any insight into that?
WHO did commission a more detailed type of analysis that was in the works, and they probably had some kind of preliminary results from that analysis by the time they made that recommendation. But these sorts of factors are very different from one setting to the next. We are doing this in Uganda, but we are certainly not claiming that the factors would be exactly the same in India, or Vietnam, or the US, or even in South Africa.
It's hard to know just how much of this analysis should or should not be done before giving approval. We face the same problems here in the US with a new drug. The [US Food and Drug Administration] may approve a new drug for a new indication based on the fact that the drug is more effective than the current standard of care, but that doesn't mean that it's necessarily a good idea for our health system to adopt it at scale.
Have you personally interacted with some of these clinics already?
Most of my experience in the field in Africa is in Malawi, which, if anything, is an even poorer country than Uganda. This will be my first time working with these clinics in Uganda. But most of our collaborative team has worked there in the past.
But I can certainly speak to the Malawian context knowing that in some ways Uganda is not terribly different.
Did you have any anecdotal experiences that spurred you to pursue this research? Situations where you saw the test not really helping, or not being easy to implement?
Yes, I think a lot of this comes from a trial we have in Malawi, where we're looking at this test in terms of its impact on mortality. There, we've tried to put these [GeneXperts] into local clinics. And it's hard. I'll give one story. Many of these clinics don't have electricity. The [GeneXpert] needs electricity. In this particular clinic we had to buy a solar panel in addition to the machines. While the GeneXpert is appropriately marketed as being about the size of a cappuccino machine, and that I could carry it myself without any difficulty, a solar panel is quite massive, and required a team of four people to move it. And, if you just leave the solar panel out at night, then people will come and steal parts off of it, so you have to lock it up, and in the morning roll it out into the sun. Another problem is there is a rainy season in Malawi, and if you take this out in the rain, you're basically bringing this out into mud. And it was virtually impossible — there was just one step they had to get down and get back up, but the solar panel would get so stuck in the mud that they would have a hard time moving it. Just this became a limiting factor in how often they could run the GeneXpert. They had to go back to doing smears, or we would have to get a universal power supply or something. It was not feasible in our existing setup to even run the machine every day. So if I was a patient coming in during the rainy season, that test might not be available to me. We made sure it was available, because we had money from a trial to put in extra power supplies, and such, but that wouldn't happen in the real world.
So it sounds like the infrastructure in many of these areas is not yet there to enable this test to be as valuable as it could be.
I think the infrastructure is there in some places, but not as widely as we would like to think.
For example, South Africa is rolling this out on a country-wide level and, not universally, but in many places in South Africa they do have the infrastructure for doing this. Even before GeneXpert it was standard of care to do a culture on every TB suspect. If you can find the infrastructure to do a culture, you can find infrastructure to do [GeneXpert].
We would like to think that kind of infrastructure exists everywhere, but it clearly does not. Part of the reason we wanted to do this grant was to be very explicit about what the infrastructure requirements would be, and then to look in a setting where many of those might not be in place, and see what the true drivers are of the ability to use this test. I gave you a story about one clinic in Malawi, but who knows if that's an isolated incident or a real problem on a large scale.
Ideally, what would you hope to be the outcome of this study in terms of bettering TB diagnostics? Do you anticipate that you'll find some places are just better off doing smear microscopy? Or is it a matter of finding that certain things are lacking, and then Cepheid or other organizations might be able to address these to make Xpert MTB/RIF better?
I think it's a combination. The first goal is to help delineate where this test is best scaled up, and where it might be a better decision to wait. I'm not saying that's a universal yes-or-no decision, but to help people in any given local circumstance make the decision that's right for them.
And the second goal is to identify the major gaps, the pieces of infrastructure that really could make this work. And then — whether it's working with Cepheid or the ministries of health on a larger scale —trying to find ways to fill those gaps so we can take what is a good technology and make sure that it has the greatest impact possible on health.
It sounds like quite a massive undertaking, with lots of different factors to consider. Will you be looking at more regions outside of Uganda?
The goal here is to focus the actual data collection on Uganda … but then use [data] and literature from other settings to help us generalize what we find in Uganda. It's not going to be perfect, and we're trying to not make this a massive undertaking. But if we can at least collect some data on the ground level to help inform these decisions … then, if you're a TB-control officer in Bangladesh, you can look at this study and say either 'I can see how that either does or doesn't apply to my setting;' or, 'That's so specific to that setting that I need to conduct this kind of study in my own setting.' But at least we will have given them the tools and an example by which they can conduct that study.
The grant also mentions that you would use that same generalization method to look at other diagnostic tests besides Xpert MTB/RIF. There are several other molecular tests on the horizon or just coming through the pipeline now. Will you look at other new molecular tests, or existing tests, or both?
We will be measuring these operational characteristics and seeing how they might overlay on to these newer molecular tests. We won't actually be deploying those tests, but we'll look both at tests that are currently under development or just coming to market; or, there is the possibility of other tests. The major concern here is that if you make a decision to scale up Xpert MTB/RIF across your country, you're probably also making the decision to not scale up whatever else might come up over the next few years, because you've committed yourself to this course of action. It's important to know when the right decision is to either wait until something better comes along, and to know how much better that might be for you, or to go with what you already have.
Who are the other collaborators in this besides Johns Hopkins?
We have investigators at the [University of California, San Francisco], the London School of Hygiene and Tropical Medicine, and the Academic Medical Center of Amsterdam. And in Uganda we are also working with folks at Makerere University in Kampala.
Is there any incentive to get Cepheid on board with this? Have you been in touch with them?
I'm sure they know about it because we've had informal discussions. We certainly haven't been trying to hide it. We're trying to keep this somewhat independent of Cepheid because we don't want this to be perceived as something that is industry-sponsored. I'm certainly not trying to keep it a secret; nor am I trying to put out an analysis that is going to be either particularly supportive or antagonistic of them. My primary goal here is to make sure people know how to use this test in an appropriate way, and I think that's a goal that they share, as well.
Once some of the early data comes out, do you anticipate getting other funding or organizations on board to further the study?
Absolutely, if we find things that are worth following up on, we are interested in doing what we need to do to answer the questions at hand. The science is really the driver, and if we find scientific questions that require more funding to answer, then we'll go find it.