November 2008
Feature Story

William Check, PhD

“The tuberculin skin test is dead,” Charles L. Daley, MD, head of the Division of Mycobacterial and Respiratory Infections and professor of medicine at National Jewish Health, Denver, likes to say. He’s quick to add, however, that the test is not there yet but will be soon. So what will be crowned as the new gold standard for detecting Mycobacterium tuberculosis (Mtb) infection? In Dr. Daley’s view, the successor to tuberculin skin testing will be a recently introduced in vitro technology, interferon-gamma-releasing assays, four versions of which have been approved for use in the United States.

“Sales of interferon-releasing assays worldwide are going up,” Dr. Daley says. “Entire institutions have switched over. We don’t have all the answers yet about the performance of these assays, but they are being utilized and we will learn a lot about Mtb infection from them.”

The adoption of interferon-gamma-releasing assays, or IGRAs, will bring a major change, Dr. Daley predicts: “They will increase the importance of laboratories in tuberculosis control.” Tuberculin skin testing is done by nurses, he notes, but IGRAs are done by medical technologists. “Skin testing is not quality controlled,” he adds. “IGRAs will be done in quality-controlled labs.”

Gerald Mazurek, MD, medical officer and epidemiologist in the Division of TB Elimination at the Centers for Disease Control and Prevention, is also optimistic about IGRAs. “Interferon-gamma-releasing assays are evolving into a test that will be better, if they are not already better, than tuberculin skin testing,” he says. They may have higher specificity in some settings. They’re considered particularly suitable in immigrants who have received BCG (bacille Calmette-Guérin) vaccination. “There is concern, and considerable evidence from other countries, that BCG can cause a positive TB skin test,” he says, and adds, “This has not been shown to be a problem in the U.S.” There are many confounders; for example, most people who get BCG received it while residing in a country where TB is prevalent, making it difficult to determine if the positive tuberculin skin test, or TST, is due to infection with TB or to BCG vaccination. “However,” Dr. Mazurek says, “it is pretty clear that [IGRAs] are not affected by BCG vaccination.”

Antonino Catanzaro, MD, professor of medicine in the Division of Pulmonary and Critical Care Medicine at the University of California, San Diego, School of Medicine, quantitates the impact of this difference in reactivity to BCG vaccination. “More than half of the cases of active TB in the U.S. occur in foreign-born individuals, most of whom have received BCG,” Dr. Catanzaro says. “These people have children and other contacts at home who have had BCG. Also, we employ many foreign-born persons as health care workers.” Using the more-specific IGRAs has the potential to reduce greatly the number of false-positive results in case finding, contact tracing, and mandatory health care worker screening programs.

One of the interferon-gamma-releasing assays to come to market is just now starting to have high uptake, according to Dr. Catanzaro. “Maybe 50 to 100 hospitals in the U.S. use it to screen health workers,” he says. “Perhaps twice as many health departments use it. However,” he notes, “it is by no means the standard at this time.”

Probably the biggest deficiency now with regard to IGRAs, in Dr. Mazurek’s estimation, is a lack of data regarding the test’s ability to identify individuals at greatest risk for developing active tuberculosis, and whether TST or IGRA is better for estimating risk. As Dr. Daley puts it, “The most important question is whether a positive response in an interferon releasing assay predicts progression to active disease.”

Identifying those most likely to progress is not in question for tuberculin skin testing, which “does predict progression to TB,” says Ajit Lalvani, MA, DM, FRCP, chair of infectious diseases in the Tuberculosis Immunology Group at the National Heart and Lung Institute, Imperial College, London. He calls tuberculin skin testing “the first biomarker.

“It was validated in large studies in the 1950s and 1960s among household contacts,” he says. “During five- to 10-year followup, nearly all the people who developed TB were in the TST-positive group.”

However, many TST-positive people must be treated to prevent one case of active disease. Historical data suggest that at most one in 10 people who are TST-positive will progress to TB. A major hope for IGRAs is that their higher specificity will improve this situation. “Fewer positive individuals are identified with interferon-gamma assays,” Dr. Mazurek says. “We hope that, instead of one in 10 progressing, maybe one in five will progress. So by improving specificity, which we think these tests have done, perhaps fewer people will need to be treated to prevent one person from developing TB.”

But long-term studies to show that is the case have not been done, he cautions. Much work is going into proving that IGRAs can fulfill this goal, and this work is starting to bear fruit. “Recent studies suggest that [IGRAs] do predict progression,” Dr. Daley says, “and in one study, better than the skin test.”

In addition to having suboptimal specificity, tuberculin skin testing also has a sensitivity problem, Dr. Lalvani says. “Screening for latent infection should be targeted to those who, if they test positive, have a higher risk of progression, such as HIV-infected persons. Those are exactly the populations most prone to having a false-negative skin test due to impaired cellular reactivity. The good news,” he says, “is that interferon-releasing assays are more sensitive than skin testing [in these groups].”

More-accurate detection of latent Mtb infection and active TB would be a great boon, given that one in three people worldwide is estimated to be infected with Mtb. Eight million to 10 million new cases of active disease arise each year; one-third of those people die.

The current system in the U.S. is good at diagnosing and treating TB as quickly as possible, Dr. Catanzaro says. Still, on average seven people have been infected by the time an active case is diagnosed—it takes about 12 weeks from symptoms to diagnosis. To prevent transmission, programs were developed to identify and treat latent Mtb infection based on skin testing those at high risk, such as children in households of persons with TB. “In my opinion, that is a major reason why we have been better at controlling TB than many other countries,” Dr. Catanzaro says.

The number of new cases in the U.S. peaked in 1992 and then began to decline. More recently, that decline has slowed. Dr. Daley says: “We’ve reached as good as we can do right now. Perhaps improved diagnostics will decrease the incidence further.” He points out that the U.S. epidemic has two components—foreign-born persons and U.S.-born persons. All of the decline in the overall epidemic can be attributed to fewer cases among U.S.-born persons. Cases among immigrants make up about half of all cases now, and that proportion is increasing. “We can’t skin test them, since most are vaccinated with BCG,” Dr. Daley says. “And that’s the population we want to screen.”

An important subset of this group is foreign-born health care workers. Mandatory screening of workers is based on work by Johannes Heimbeck in Norway in the late 1920s, Dr. Catanzaro notes. Heimbeck showed that virtually all tuberculin-negative student nurses became infected during their three-year training. This led to what Dr. Catanzaro calls the “rather draconian” requirement that all health care workers who have face-to-face contact with patients be screened for TB annually. “Literally hundreds of thousands of health care workers must be tested each year,” he says. Screening has made TB rare among such workers. It not only identifies those who contract TB from patients, Dr. Catanzaro says, but also prevents them from giving TB to hospitalized patients. More-specific screening tests would improve both aspects of the health care worker screening program.

Enter the interferon-gamma-releasing assays. Like tuberculin skin tests, IGRAs are immunological assays but with two major differences. First, the two antigens in IGRAs—ESAT-6 and CFP-10—are highly specific for bacteria in the Mycobacterium tuberculosis complex, including the subspecies M. tuberculosis, M. bovis, M. africanum, and M. microti. They are also present in a few other mycobacteria, most notably M. kansasii, M. marinum, and M. szulgai. Purified protein derivative, on the other hand, contains more than 500 antigens. Second, while tuberculin skin testing is performed in vivo, IGRAs are done in vitro using fresh blood. “By stimulating the patient’s T cells with specific antigens in vitro, interferon-gamma [IFN-λ] is released in previously sensitized individuals, allowing for the detection of latent Mtb infection,” Dr. Daley explains. With tuberculin skin testing, the person being tested must return within a narrow time window for the skin reaction to be read. “Even the best programs have only a 50 percent return rate,” Dr. Daley says. As standard in vitro laboratory assays, IGRAs don’t have this drawback.

Three commercial IGRAs are available in the U.S.: Quantiferon-TB Gold (QFT-G, Cellestis Ltd., Carnegie, Australia), Quantiferon-TB Gold In-Tube (QFT-GIT, Cellestis), and T-Spot.TB (Oxford Immunotec, Oxford, UK). QFT-G and QFT-GIT measure the amount of IFN-λ; released into the plasma surrounding stimulated T cells, while T-Spot uses the ELISpot technology that Dr. Lalvani devised in the late 1990s to measure the number of stimulated T cells that produce IFN-λ. After the May 2005 approval of QFT-G, the CDC issued a recommendation concluding that “QFT-G may be used in all circumstances in which the TST is currently used, including contact investigations, evaluation of recent immigrants, and sequential-testing” with noted limitations and warnings (Mazurek GH, et al. MMWR Recomm Rep. 2005;54:49–55). In 2007, the FDA approved the QFT-GIT assay, which includes a third antigen and facilitates testing by including the antigens in the tube used to collect blood. Earlier this year, the FDA approved the premarket application for T-Spot. Dr. Mazurek says guidelines for use of the three available assays are in the works. Another test, developed by Dr. Lalvani and currently called ELISpot(Plus), also contains a third antigen.

Ron Harbeck, PhD, medical director of the Advanced Diagnostics Laboratory at Denver’s National Jewish Health, has experience with both types of IGRAs. “One important fact about these tests is that blood needs to get to the lab within a relatively short time after drawing,” he says. That’s because both IGRAs require viable T cells. Delivery time is 12 hours for QFT-G and 16 hours for QFT-GIT. T-Spot requires isolation of cells before antigenic stimulation, so blood for this assay needs to be in the lab within eight hours. “Our lab does a lot of [cellular] isolation, but it could be an issue for community hospitals,” Dr. Harbeck says. In QFT-GIT the tube contains antigens in dry form, so the tubes need to be shaken to dissolve the antigens after the blood is collected. The advantage is that the reaction starts immediately when whole blood is added. The disadvantage is that it introduces handler processing that can lead to variability. “People are not used to shaking tubes vigorously,” Dr. Harbeck says. “This could be counterintuitive to phlebotomists.”

Regarding assay performance, Dr. Daley says that both [IGRAs] are at least as sensitive as and more specific than the tuberculin skin test. “Quantiferon is more specific than T-Spot, but the latter is the most sensitive of all assays,” he says. In one review of published studies, among patients with active infection, sensitivities were 0.78 for QFT-G, 0.70 for QFT-GIT, and 0.90 for T-Spot (Pai M, et al. Ann Intern Med. 2008;149:177–184). For comparison, pooled sensitivity of tuberculin skin testing was 0.77. Sensitivity of the next-generation ELISpot(Plus) for active TB is even higher, according to Dr. Lalvani (Dosanjh DP, et al. Ann Intern Med. 2008;148:325–336). Very high sensitivity makes it possible to use a negative result reliably to rule out TB, “which will revolutionize assessment of patients with suspected TB in routine practice,” he says.

Dr. Lalvani points in particular to the higher sensitivity of IGRAs for latent Mtb infection among those at higher risk for progression to active disease. “Among immunocompetent adults infected more than five years ago,” he says, “the lifetime probability of progression to TB is about five percent. In children under two years or immunocompromised adults—those who are HIV-positive or being treated with immunosuppressive drugs—the risk of progression to full-blown TB is much higher.” Dr. Lalvani says that both IGRAs are more sensitive among these groups than tuberculin skin testing, but that this finding has been more clearly demonstrated for T-Spot. In one study in South Africa, for example, more than 70 percent of 40 HIV-positive people were positive by tuberculin skin testing and T-Spot. However, the authors noted, in those with CD4 counts of less than 100 cells/µL, response was significantly impaired for tuberculin skin testing but not for T-Spot (Lawn SD, et al. BMC Infect Dis. 2007;7:99).

There’s still “a nagging concern” about the sensitivity of IGRAs for detecting latent TB infection, Dr. Mazurek says. “This is because there is no gold standard with which to confirm the presence of latent TB infection. Immigrants do have fewer positive results with IGRAs than with TST. That could be due to better specificity. But it may also be due to lower sensitivity.” He raises the possibility that IGRAs are less sensitive for remote infection.

An indirect approach to this question is to ask whether positive test results increase relative to TB incidence. Comparing countries where TB incidence is less than 20/100,000 versus greater than 100/100,000, the rate of positivity increases for TST and IGRAs, but so does the degree of discordance. “That makes it hard to say that all of the discordance is due to BCG,” Dr. Mazurek argues. “In the U.S., the majority of disease is due to reactivation of remote infection.”

In the 2008 Pai review in the Annals of Internal Medicine, specificity for both Quantiferon assays among people at very low risk for latent Mtb infection was 0.99 among nonvaccinated persons and 0.96 among BCG-vaccinated persons. Specificity for T-Spot was 0.93. Dr. Lalvani has written that “both [T-Spot and QFT-G] have very high specificity, approaching 100 percent” (Lalvani A. Chest. 2007;131:1898–1906). His estimate was based on an indirect method: studying BCG-vaccinated individuals at ultralow risk of latent Mtb infection owing to absence of epidemiologic risk factors for Mtb exposure.

For tuberculin skin testing, pooled specificity in the Pai review was 0.97 among non-BCG-vaccinated people but 0.59 among those who are BCG-vaccinated. “Lower specificity leads to overtreatment,” Dr. Daley reiterated.

Dr. Lalvani points out another important implication of lower specificity. As control programs drive down the prevalence of active and latent TB, the proportion of positive tuberculin tests that are false increases. In this context, he says, “The high specificity of interferon-releasing assays is an enabling characteristic for TB elimination programs. Without that you would never get there.”

In general, Dr. Daley says, T-Spot has a lower rate of indeterminate results—about one percent—than either Quantiferon assay (5.6 to 8.2 percent). “Like the TST, [IGRAs] are primarily tests to identify infection and not disease,” he says. Neither type of test differentiates latent infection from active disease.

Sheldon Campbell, MD, PhD, associate professor in the Department of Laboratory Medicine at Yale University School of Medicine and clinical pathologist at VA Connecticut in West Haven, says given their advantages, interferon-releasing assays have become useful in some public health settings because of their ability to do accurate single-visit screening of populations with high rates of BCG immunization.

Several aspects of the performance of IGRAs are still under investigation. Treatment of latent Mtb infection appears not to change responses of QFT-G or QFT-GIT. For T-Spot.TB, on the other hand, several studies suggest decreased reactivity after successful treatment (for example, Ewer K, et al. Am J Respir Crit Care Med. 2006;174:831–839).

Reproducibility of IGRAs is still being elucidated. Like tuberculin skin tests, IGRAs have what Dr. Daley calls “wobble.”

“Serial testing shows fluctuation around baseline,” he explains. Fluctuation affects reversion and conversion rates. “We don’t really have a validated definition of either rate for interferon-releasing assays,” Dr. Daley says. “Depending on which criteria you use, you get vastly different conversion rates.” There is a definition of conversion for tuberculin skin tests. In fact, there are two definitions. “In the U.S., we use an increase in duration of 10 millimeters,” Dr. Daley says, “while Canada uses six millimeters.” Neither definition has ever been clinically validated.

Of paramount importance to TB control programs is whether IGRAs predict progression from latent Mtb infection to active disease. While the question is not yet settled, there is strong evidence that they do identify those who will progress. Dr. Catanzaro cites a study in Germany in which 601 recently exposed household contacts of TB cases were investigated. Both tuberculin skin testing and QFT-GIT were done. Two hundred fifty people did not follow the treatment recommendation. (“In Germany patients have to go to private doctors to get treatment,” Dr. Catanzaro says.) After two-year followup, six untreated contacts had developed active TB. Five were among the 219 TST-positive individuals, for a progression rate of 2.3 percent, while all six were among 41 QFT-GIT-positive people, for a progression rate of 14.6 percent (Diel R, et al. Am J Respir Crit Care Med. 2008;177:1164–1170). “Quantiferon In-Tube is much more predictive of getting active TB than tuberculin testing,” Dr. Catanzaro concludes. In this population, based on TST positivity, 219 additional people would have to be treated to prevent five cases of TB; using QFT-GIT positivity, all six cases could be averted by treating 41 additional people, allowing a more focused and efficient prevention effort.

Dr. Lalvani and colleagues recently published data for the T-Spot assay from community-based contact investigations in Turkey using a different study design. Among 908 children and adolescents with recent household exposure, isoniazid preventive therapy was given to 688 (76 percent). During followup, 15 contacts developed active tuberculosis, even though the majority of them had been treated. “This seems surprising at first,” Dr. Lalvani says. “But it is to be expected because isoniazid is not 100 percent effective in preventing TB. Without treatment, about 50 cases would have been expected.” Eleven cases occurred among the 381 people with a positive T-Spot, while 12 occurred among the 550 with a positive TST (Bakir M, et al. Ann Intern Med. 2008. Oct. 20 [Epub ahead of print]). “[T-Spot] testing could allow more focused targeting of preventive therapy to fewer contacts,” the investigators concluded.

“We now know,” Dr. Lalvani says, “that a positive interferon-gamma [T-Spot] result is a useful and valid marker of latent TB, because it predicts subsequent development of active TB.” Although screening with T-Spot would allow more focused preventive treatment, “you still need to treat 20 to 30 contacts for every case of TB you prevent,” he says. “That’s why we are now validating our next generation of tests to more precisely target latently infected people who are most likely to develop TB.”

Dr. Lalvani and colleagues are also developing cytokine profiles that can be measured by novel cytokine detection assays to evaluate response to therapy. In one study in 23 patients with active tuberculosis, they found that levels of IFN-λ and IL-2 changed in characteristic ways after curative treatment (Millington KA, et al. J Immunol. 2007;178:5217–5226).

At the present time, Dr. Campbell says, “Between Quantiferon Gold and T-Spot, there’s too little data to choose. The Quantiferon-TB Gold In-Tube is cur­rently easier to deploy.”

While IGRAs are being incorporated into many public health programs, hospitals also are adopting them. “I would say they are being embraced with almost more vigor” in hospital settings, Dr. Mazurek says. “I think it is so they can use IGRAs in employee screening programs. Once brought in for that, they make them available for other patients.”

That’s been Dr. Harbeck’s experience. Some hospitals in Denver are running QFT-G, he says. “The test is being used by employee health at institutions where the immigrant populations from endemic areas are high,” Dr. Harbeck notes. National Jewish Health has not yet adopted an IGRA on a routine basis. “Employee health considers it too expensive,” he says. Because he has done validation studies on the IGRAs, he has them in his lab. “We have a large pulmonary clinic, and often a physician will order an in vitro test for a patient undergoing diagnosis,” he says. “We think this test would prevent some followup expense, particularly if patients go to radiology as part of their TB workup.”

Dr. Harbeck is conducting a study now comparing QFT-GIT to T-Spot in a population that has been underestimated and underdiagnosed in the past as having latent or active TB and in which tuberculin skin testing has not been an accurate indicator of disease. “According to preliminary results, the two IGRAs appear to be fairly comparable at this time,” Dr. Harbeck says. “In our hands both are good assays.”

Sooner or later any public health department or hospital that wants to replace tuberculin skin testing with an IGRA will have to face the financial facts: In vitro assays are several times more expensive than TST. But “based on at least three analyses,” Dr. Daley says, “[IGRAs] are more cost-effective than skin testing.” Nurses administer and read skin tests, and every person who is skin tested will not return to have the test read. “Nurse time in the U.S. is expensive,” Dr. Daley says. “You also have to add in the fact that, because of the lower specificity of TST, people will get treated who should not be treated, particularly those who were BCG-vaccinated, and that creates extra cost and hepatotoxicity.”

Dr. Lalvani readily concedes that the cost of an individual IGRA is much higher than the cost of tuberculin skin testing, “especially so for [T-Spot],” he says. However, like Dr. Daley, Dr. Lalvani cites health economic analyses that have shown that using IGRAs instead of TST will save money in the medium term. These savings arise from two sources: the reduction in outlay from not treating false-positive cases, and “because of [IGRAs’] improved sensitivity in vulnerable groups, you would miss fewer people with latent infection and at high risk of progression to TB disease.” Treating active TB is much more expensive. “The vast majority of money in TB control programs is spent on patients with active TB,” he says. In addition, during progression each latently infected person would infect several others.

Dr. Mazurek agrees that money will be saved by using IGRAs with greater specificity, provided there is not a significant reduction in sensitivity. “We currently do not know if IGRAs are as sensitive as TST in predicting who will develop TB. While fewer IGRA tests are positive,” he says, “IGRAs may not be identifying as many people who will develop TB as the TST. Those missed may develop TB,” which ultimately may result in greater expense. Long-term studies are needed to determine if IGRAs are cost-effective, Dr. Mazurek says, and, in the short term, personnel costs for testing will probably rise. “You won’t have as many people doing skin tests,” he agrees. “But more people will be drawing blood for IGRAs and performing IGRAs in the laboratory. Until we can use IGRAs for everyone, we need to maintain the ability to do skin testing.”

Dr. Campbell sums up the economic arguments. “At this time, the IGRAs are quite costly,” he says. “It’s nice to have them available, in that they’re pretty well proven to give fewer false-positives in BCG-immunized persons than the TST and can therefore save expensive and potentially toxic prophylactic treatment when used as a confirmatory test.

“The difficulty lies in adequately accounting for the cost of the TST,” he continues. “PPD itself is very inexpensive, but the cost of followup visits for reading of the TST is highly variable and difficult to quantify. Some cost-effectiveness studies have shown the IGRAs to be superior, but I don’t know how generalizable those results are.”

Work on evaluating IGRAs continues. Dr. Daley is coordinating a four-center CDC-funded study in which more than 2,000 health care workers will be tested with all three tests every six months for two years. “We will get point prevalence figures and find out which assay is most stable and predictive,” he says.

Dr. Lalvani is further developing cytokine detection technologies to create a T-cell profile with IFN-λ and IL-2. “Because you are measuring individual T cells, that allows you to measure more than one cytokine at the single-cell level,” he says. “So we can use the assay to quantify treatment response in individual patients.” Three different profiles represent two different categories of T cells. Activated effector T cells make only IFN-λ and represent a high pathogen burden and high antigen load. Cells that make both cytokines or only IL-2 are central memory T cells and predominate when patients have cleared Mtb or successfully suppressed it. Monitoring treatment response should allow modifications that increase treatment efficacy.