active TB: IGRA not sufficiently accurate

a study in England found that commercially available IGRAs (interferon-g release assays, specific for M. tuberculosis) do not have sufficient accuracy for the diagnostic evaluation of suspected tuberculosis (see tb igra not accurate lancetinfdz2019 in dropbox, or doi.org/10.1016/S1473-3099(18)30613-3).

Details:

-- 845 adults with suspected tuberculosis were prospectively followed for 6 to 12 months for the subsequent diagnosis of tuberculosis, comparing the accuracy of commercially available IGRAs

-- patients greater than 16 years old were enrolled from 10 National Health Service hospitals in five English cities

-- patients are classified as culture-confirmed tuberculosis, highly probable tuberculosis (clinical and radiologic features highly suggestive of TB and unlikely caused by other diseases), clinically indeterminate diagnoses, and those where active tuberculosis was excluded (mostly with a history of exposure to TB and a negative PPD screening test, defined as <15 mm, and no clinical evidence of TB)

--64% were outpatient, median age 38, 59% male, 49% from Indian subcontinent/22% black/20% white/5% Asian, 50% working/31% unemployed/11 current students, BMI 23, 77% had BCG, 22% had known contacts with TB, 50% had no comorbidities, 16% were HIV-positive/10% diabetes/8% asthma

    -- more patients who had culture-confirmed tuberculosis or highly probable TB were treated as inpatients, male, from the Indian subcontinent, were in the UK for less time (4.8 years), less likely to be HIV-infected, more likely to be students, and fewer were white

-- they compared the commercially available QuantiFERON-TB Gold In-Tube (QFT) and the T-SPOT assays, as well as second-generation IGRAs that also incorporate novel M. tuberculosis antigens ESAT-6 and CFP-10

Results:

-- 363 of the 845 people were diagnosed with active TB (either culture-confirmed or highly probable TB)

    --36% had pulmonary TB; 52% had extra-pulmonary (mostly in lymph nodes, small percentages in multiple other locations) and 12% had both

    --6% had drug-resistant TB

-- in those diagnosed with tuberculosis, they found the following test characteristics:

-- T-SPOT

    -- sensitivity: culture-confirmed TB: 84.9% (79.5-89.0%); highly probable TB: 73.1% (63.3- 81.1%). No significant difference between cases of pulmonary versus extra-pulmonary TB

    -- specificity: 86.2% (82.3-89.4%); for those with active TB excluded: 93.5% (86.6-97.0%)

    -- positive likelihood ratio 5.90 (4.55-7.66); negative likelihood ratio 0.22 (0.17-0.27)

--QFT

    -- sensitivity: culture-confirmed TB: 70.6% (64.4-76.1%); highly probable TB: 59.4% (49.4-68.7%). No significant difference between cases of pulmonary versus extra-pulmonary TB

    -- specificity: 80.4% (76.1-84.1%); for those with active TB excluded: 93.4% (86.4-96.9%)

    -- positive likelihood ratio 3.44 (2.76-4.27); negative likelihood ratio 0.41 (0.33-0.48)

   

-- second-generation IGRAs (they tested two of them, with very similar results):

    -- sensitivity: culture-confirmed TB: 94.0% (90.0-96.4%); highly probable TB: 77.8% (68.2-85.1%). No difference between pulmonary and extra-pulmonary TB.

    -- Specificity: 80.0% (75.6-83.8%); for those with active TB excluded: 91.3% (83.8-95.5%)

    -- positive likelihood ratio 4.46 (3.62-5.49); negative likelihood ratio 0.13 (0.10-0.19)

--the second-generation IGRA tests were more sensitive and specific than T-SPOT (by a little: relative sensitivity 1.08 (1.04-1.11), p<0.0001; relative specificity 0.92 (0.91-0.96), p<0.0001)

Commentary:

-- the sensitivities of these IGRAs is still too low to be relied upon for the diagnosis of active TB; but it was notable that the sensitivity for QFT was substantially worse than for the T-SPOT, though both decreased as the likelihood of active tuberculosis decreased; QFT down to the mid-60% and T-SPOT to the low 80% range

-- IGRAs are  supposedly very specific tests targeting specific Mycobacterium tuberculosis protein antigens (these tests measure T-cell release of interferon-g after stimulation by these very specifc TB specific antigens). so, it is a bit curious that their specificity is not essentially 100% (low 90's just does not seem good enough....). Possibilities:

    --several patients who did not have a TB-related infection (their 4th category of TB clinically excluded) may have had LTBI but with waning immunity so in many the PPD was negative (they did not comment if the PPD remained negative after a repeat PPD, the booster effect, see below). And the vast majority of patients did come from areas with high TB prevlence

    --maybe the "specific" mycoplasma TB proteins are not so specific afterall.  maybe some of these proteins are similar to those from other microbes? Maybe some untested nontuberculous mycobacteria? or elsewhere?

-- and, the bottom line issue regarding sensitivity and specificity is that there is no "gold standard" by which to measure these tests, no matter how many decimal points of accuracy are implied by their sensitivity and specificity statistics.  All of the studies from which these numbers derive have their own limitations which undercut the true accuracy of their statistics (eg, see below on the PPD sensitivity numbers)

-- there are very real concerns about the overall reliability of the IGRAs in diagnosing latent TB infection (LTBI):
    -- QFT is quite susceptible to mechanical factors (temperature, sample agitation, time prior to incubation), with changing results from positive to negative in about 20%!!!
    -- another QFT study found that the 2 samples of blood from the same patient at the same time had a discordance rate of 8%
    -- and another found significant discrepancies in QFT when samples from the same patient were sent to different labs

    -- http://gmodestmedblogs.blogspot.com/2016/03/the-uspstf-just-circulated-draft.html reviews a couple of articles finding pretty dramatic inconsistencies in IGRA results, including 8% of patients with positive results reverting to negative

--as in the above article, the USPSTF also found lower sensitivity of QFT vs T-SPOT: see http://gmodestmedblogs.blogspot.com/2016/09/uspstf-ltbi-screening-recommendations.html for the 2016 USPSTF recommendations (full evidence reported in tb ltbi review jama2016 in dropbox, or doi:10.1001/jama.2016.10357), including:

    --“Pooled analyses of the T-SPOT.TB test (a type of IGRA) indicate sensitivity of 90% (16 studies; n=984) and specificity of 95% (5 studies; n=1810). Pooled analyses of the QuantiFERON-TB Gold In-Tube test (another type of IGRA) indicate sensitivity of 80% (24 studies; n=2321) and specificity of 97% (4 studies; n=2053). The USPSTF identified no studies that evaluated the accuracy and reliability of sequential screening strategies.”
-- this active TB study does have some limitations, including that it included only adults and that there were few patients with diabetes or HIV (both of which can decrease IGRA test performance). but it did follow these pretty sick patients for 6-12 months, so it is pretty likely that their TB classification system (culture-confirmed and highly probable TB) are reasonably accurate

-- So, what is the real utility of these IGRA tests (which seem to have taken over as the leading tests in hospitals, and now US immigration is demanding IGRAs instead of PPDs for all people >2yo applying for permanent residence)

    -- seems reasonable to consider IGRAs is in cases where the patient reports a positive PPD in the past, though it is not adequately documented. In their 2013 LATENT TUBERCULOSIS INFECTION: A GUIDE FOR PRIMARY HEALTH CARE PROVIDERS, the CDC recommends “The TST (tuberculin skin test, aka PPD) should not be performed on a person who has written documentation of either a previous positive TST result or treatment for TB disease” given the potential for a severe local reaction.  so, to me, i would get an IGRA prior to starting LTBI therapy.

   -- overall, i thought these tests would be great given the high percentage of my patients who had BCG vaccination. But the striking inconsistencies of these tests (as in the above blog) have largely chilled my enthusiasm, and I have still been using PPDs

   -- but, given the more impressive data on the sensitivity/specificity of T-SPOT tests, it does seem that this test in particular is at least as reliable as PPDs. though, i might add, the comparison studies between T-SPOT and PPD did not include giving a booster dose of PPD and this might have decreased the PPD sensitivity substantially (this "booster test" is done because of waning immunity over time, wherein a second PPD is done after 1-2 weeks and is more likely to be positive in a patient with LTBI because the PPD itself boosts the T-cell response). 

   --and, the T-SPOT does have important advantages in terms of ease of doing (and does not require the patient to have the PPD planted correctly, or return in 48-72 hours and then have it read correctly)

so, this study in the UK did find fairly good sensitivity and specificity of T-SPOT for patients with active TB, much better than QFT, but clearly not good enough to accurately rule-in or rule-out active TB

geoff​

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org

to get access to blogs since 8/15/17:

1. go to http://gmodestmedblogs.blogspot.com/ to see them in reverse chronological order

2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

3. or you can just type in a name in the search box and get all the blogs with that name in them

to access older blogs from the BMJ website, from October 2013 until 8/15/17: go to http://blogs.bmj.com/bmjebmspotlight/category/archive/ 

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list