A new, sophisticated diagnostic test for tuberculosis now being rolled out promises to be faster and more accurate than the old methods and much easier to use. But the first trials of the GeneXpert MTB/RIF Assay test in real-life situations have proved that while all this is true, it did not make any significant difference to treatment outcomes.
The problem with the traditional TB tests is that they need trained lab technicians and tend to be either inconclusive or extremely slow. Some cases of TB can be spotted simply by inspecting a sputum sample under a microscope, but not all; chest X-rays can also be a guide as to whether the disease may be present. But to be sure, the samples have to be cultured in a laboratory, and the process takes around eight weeks to produce a result. By then a patient may have wandered off and disappeared, and even if he does then start treatment, he has had two more months to infect other people.
The GeneXpert system looks for TB's DNA markers and can give a result within two hours, with a high degree of accuracy. The system is automated; all the necessary chemicals are contained in a cartridge, which just has to be slotted into the machine, a fresh cartridge for each test. So it is very easy to use. It can also be used to look for the DNA markers which show whether the particular strain of TB is resistant to Rifampicin, one of the most commonly used drugs.
Impressed by its capabilities, the World Health Organization (WHO) has backed its introduction and the Global Fund to Fight AIDS, Tuberculosis and Malaria is helping pay for it. But it is still expensive, even at discounted prices for developing countries - US$17,000 for the machine and $10 for each cartridge. And critics point out that it needs a reliable power supply and the kind of clean, cool environment not available in many rural clinics.
Now researchers from Cape Town University have looked at what actually happened when the system was installed in primary health care TB clinics in South Africa, Zimbabwe, Zambia and Tanzania. It was operated by nurses who had been given just one day's training, and patients presenting at the clinics with TB-like symptoms were randomly assigned to either conventional testing or to testing with the GeneXpert machines. Their results have now been published in the London-based medical journal, the Lancet.
The results in many ways were encouraging. The nurses were capable of operating the equipment and confident about their ability to do so. More of the GeneXpert group were able to start treatment on the first day they came to the clinic, 23 percent as opposed to 15 percent of those diagnosed by smear microscopy. If you look at those treated within three days, the gap had started to close, with 32 percent of the GeneXpert group having started treatment, as opposed to 27 percent of those relying on traditional diagnosis. After eight weeks, when the culture test results were available, there was only a 1 percent difference between the groups - but as the study remarks, "This issue is important for tuberculosis control, because these patients would have continued to transmit tuberculosis if left untreated."
So it looks as if most patients who needed treatment got it in the end, regardless of how they were diagnosed, and when they followed up the patients six months later to see how well they were doing, they could not detect any difference in death rates, or in their state of health. (They measured morbidity using the Karnofsky performance status index, which scores patients from 100 percvent = in perfect health to 0 percent = dead).
One reason, they suggest, is that these clinics did not actually wait eight weeks for the results of the lab tests to start treating their patients if, despite a negative initial smear, they were pretty sure they had TB. They took chest X-rays as well as sputum samples, assessed their symptoms, and were prepared to go ahead on empirical evidence. But they could not know which cases were drug-resistant until they got the full test results. Of course, these were good clinics, with X-ray facilities available and a lot of experience in diagnosing tuberculosis. But less well-equipped clinics would be likely to struggle with more basic aspects of the new technology.
In a comment in the Lancet, Christian Wejse, of Aarhus University in Denmark, writes: "Considering (these) findings. the substantial financial burden of Xpert MTB/RIF rollout needs to be reassessed to see if it provides value for the cost. Placing very expensive equipment in health-care facilities in rural Africa that might have no electricity and poorly trained, underpaid staff is going to be a difficult undertaking. Are the incremental gains in same-day diagnosis and treatment initiation, as well as reduced loss to follow-up, enough to justify this investment?"
A more targeted role?
But Professor Alison Grant of the London School of Hygiene and Tropical Medicine, who herself works on improving TB diagnosis and treatment in southern Africa, thinks this technology does have a role, although not for universal use. She told IRIN: "Policymakers need to know how best to invest money to improve TB control. This may differ between countries. In this study, similar to others, not all patients with a positive test result started TB treatment (8 percent with Xpert versus 15 percent with smear microscopy), and systems need to be strengthened to minimize these losses, regardless of what test is used.
"A major advantage of Xpert MTB/RIF is that it can identify people with drug-resistant TB very rapidly. In this study the proportion of people with drug-resistant TB was low, and there might be a much greater benefit from using Xpert in settings where drug-resistant TB is more common, providing that people identified as having drug-resistant TB start on the correct treatment promptly."