1. Tuberculin Skin Test Variability and Impact of Tuberculin Skin Test on Subsequent QuantiFERON®-TB Gold In-Tube Test Results Whitworth WC1, Ilieva-Hughes EI2, Goodwin DJ3, Mazurek GH1 1Centers for Disease Control and Prevention (CDC), Division of TB Elimination, Atlanta, GA, USA; 2Emory University, Rollins School of Public Health, Atlanta, GA, USA; 3United States Air Force School of Aerospace Medicine, Wright-Patterson Air Force Base, Dayton, OH, USA
Introduction
Methods
Results
The tuberculin skin test (TST) and interferon-gamma release assays (IGRAs), such as the QuantiFERON®-TB Gold In-Tube Assay (QFT-GIT), are used to detect M. tuberculosis (MTB) infection. Although IGRAs are increasingly replacing the traditional TST, the TST is still widely used, especially in resource-limited settings. The TST is also the preferred method for testing for children under 5 years old.
Some people infected with MTB may have a negative TST result due to a waning response to TB antigens with time, age, or concomitant illness. They may have a positive reaction to a subsequent TST or IGRA because injection of TB antigens for the initial TST stimulates recovery of the immune response. This is commonly referred to as the “booster phenomenon” (Figure 1) and may incorrectly be interpreted as TST conversion due to a new MTB infection. To facilitate recognition of conversion due to boosting, TST should be repeated within 1 to 3 weeks for adults who are to be retested periodically, such as health care workers. This two-step approach can reduce the likelihood that a boosted reaction to a subsequent TST will be misinterpreted as a recent infection.
Figure 1. TST Boosting.
Information on the relative within-subject variability of TST and QFT-GIT, and the impact of TST boosting on subsequent TST and QFT-GIT results is limited. The four goals of these analyses were to assess:
1. TST within-subject variability by comparing TST results administered the same day in different arms.
2. TST – QFT-GIT agreement by comparing TST and QFT-GIT results from both tests administered the same day.
3. TST boosting of a subsequent TST by comparing results of TSTs administered 1 week apart.
4. TST boosting of a subsequent QFT-GIT by comparing results of QFT-GITs administered 1 week apart with a TST administered on the same day but after the first QFT-GIT.
Figure 2. Experimental Design
1. TST Within-Subject Variability (continued)
Quantitative
3. TST Boosting of a Subsequent TST
Qualitative
2nd TST, 1 week later (either arm)
2nd TST, 1 week later (either arm)
Figure 3a. Scatter plot.
Figure 3b. Bland-Altman (difference) plot.
pos
neg 45 7 52 27 44 71 72 51
123
pos
neg 41 5 46 31 77 72 51
123
1st TST
Left Arm
1st TST
Right Arm
Discordance = 27.6% (34/123); Kappa = 0.46
42.3% (52/123) positive with 1st TST (left arm) vs. 58.5% (72/123) positive with 2nd TST, one week later (either arm), p <
Boosting: Of 71 negative with 1st (left arm) TST, 38.0% (27/71/) were positive with TST one week later.
Discordance = 29.3% (36/123); Kappa = 0.44
37.4% (46/123) positive with 1st TST (right arm) vs. 58.5% (72/123) positive with 2nd TST, one week later (either arm), p <
Boosting: Of 77 negative with 1st (right arm) TST, 40.3% (31/77/) were positive with TST one week later.
Results
Subject Characteristics
Table 1. Subject characteristics, N = 158 (97 CDC, 61 USAF).
Quantitative
Category
n (%)
Age, years
16 (10.1%)
Therapy for LTBI
Yes
114 (72.2%)
35 (22.2%) No
43 (27.2%)
45 (28.5%)
Unknown
1 (0.6%)
42 (26.6%)
≥ 60
20 (12.7%)
Exposure to Active TB
60 (38.0%)
74 (47.1%)
Sex M
70 (44.3%)
24 (15.2%) F
88 (55.7%)
BCG
32 (20.3%)
Race/Ethnicity
White, non-Hispanic
79 (50.0%)
112 (70.9%)
Black, non-Hispanic
37 (23.4%)
14 (8.9%)
Asian/Pacific
18 (11.4%)
Hispanic
15 (9.5%)
Native American
Region of Birth
United States/Canada
113 (71.5%)
Other
8 (5.1%)
Asia
Central America/Caribbean
12 (7.6%)
Africa
7 (4.4%)
Last Positive TST
Europe/Russia
4 (2.5%)
Pacific
3 (1.9%)
Southeast Asia
2 (1.3%)
Middle East
65 (41.1%)
South America
54 (34.2%)
Years Lived Outside USA
None
66 (41.8%)
1 - 10
63 (39.9%)
Therapy for TB
154 (97.5%)
Multivariate
Model: Probability of negative to positive test conversion. Covariates same as in Assessment #1.
Significant Associations: No factors were significantly associated with a negative to positive test conversion.
Multivariate
Model: Probability of 1) discordant result or 2) positivity. Covariates: age (categories), sex, race/ethnicity, birth region, prior TB exposure, prior TB treatment, prior LTBI therapy, BCG history, lived outside US for > 1 year, year of previous positive TST (categories), study site (CDC or AF), ever lived/worked in jail, homeless shelter, or healthcare facility, TST placer, TST reader.
Significant Associations:
Discordance: No factors were significantly associated with discordance.
Left Arm TST Positivity (Odds Ratio, 95% CI): Right Arm TST Positivity:
Ever lived outside U. S. (ref = no): 2.50 ( ) No factors were significantly associated with right arm TST positivity.
TST reader JAB (ref = others): 0.16 ( )
Methods
4. TST Boosting of a Subsequent QFT-GIT
Qualitative
2nd QFT-GIT
1 week later
Discordance = 27.4% (40/146); Kappa = 0.43
21.2 % (31/146) positive with 1st QFT-GIT vs % (69/146) positive with 2nd QFT-GIT, one week later, p <
Boosting: Of 115 negative with 1st QFT-GIT, 33.5% (39/115) were positive with QFT-GIT one week later.
Subject Selection
Non-diseased subjects recruited in 2010 at the Centers for Disease Control and Prevention (CDC, Atlanta, GA) and Lackland Air Force Base (San Antonio, TX) as part of a larger experimental study investigating QFT-GIT reproducibility.
Inclusion: self-reported prior-positive TST to enrich the number of QFT-GIT positives (prior unpublished assessments in similar cohorts have found that 40% to 50% of persons with self-reported prior-positive TST results were positive by QFT-GIT vs. < 3% for the general U.S. population).
Exclusion: TST in the past 3 years or < 18 years old.
TST
By Mantoux method, injecting 0.1 ml (5 TU) of Tubersol (PPD, Connaught Laboratories, Inc., Toronto, Ontario, Canada) following American Thoracic Society (ATS)/CDC guidelines.
Transverse induration measured at the PPD injection site 48 hours after injection by trained healthcare workers according to ATS/CDC guidelines.
Induration diameter ≥ 10 mm is interpreted as a positive test result.
QFT-GIT
Performed according to QFT-GIT manufacturer’s package insert (Cellestis, 2010).
ELISAs were performed on a Triturus automated ELISA workstation (Grifols USA, Miami, FL), using eight
Interferon gamma (IFN-γ) calibrators (8, 4, 2, 1, 0.5, 0.25, 0.125, and 0 IU/mL) in duplicate to create standard curves. [IFN-γ] (IU/mL) calculated from standard curve using software developed at the CDC, and test results interpreted as indicated in the Cellestis package insert and CDC guidelines.
Statistical Analysis
Paired sample statistical methods used to compare pairs of TST/TST or TST/QFT-GIT results from subjects.
Qualitative Measures (positive/negative test interpretations): compared in 2 x 2 tables using McNemar’s test for significance of differences in marginal percentages, and reporting agreement (Kappa, % discordance), and conversion and reversion rates (percentages).
Quantitative Measures (TST mm induration diameter or QFT-GIT IU/mL IFN-γ): mean, median, difference in distribution location (Wilcoxon Signed Rank test), intraclass correlation coefficient (ICC), standard error (SE, within-subject standard deviation), Bland-Altman 95% limits of agreement (LOA, range that encompasses 95% of paired differences).
Logistic regression: Odds of: discordance, positivity (assessments 1 & 2), - to + conversion (assessments 3 & 4). All models refined used stepwise backward elimination and α = 0.05 for covariate entry and retention.
pos
neg 30 1 31 39 76
115 69 77
146
1st QFT-GIT
2. TST – QFT-GIT Agreement
Qualitative
Quantitative
TST Left Arm
TST Right Arm
pos
neg 20 11 31 55 62
117 75 73
148
pos
neg 22 9 31 48 68
116 70 77
147
QFT-GIT
QFT-GIT
Multivariate
Model: Probability of negative to positive test conversion. Covariates same as in Assessment #1.
Significant Associations: No factors were significantly associated with negative to positive test conversion.
1. TST Within-Subject Variability
Qualitative
Summary/Conclusions
TST Right Arm
Discordance = 13.6% (21/154); Kappa = 0.73
51.9 % (80/154) positive with left arm vs. 48.7% (75/154) positive with right arm, p = 0.28
Number with readings of zero: 49 (R), 53 (L), 37 (R and L)
Two reader-pairs (SC/GM at CDC and DG/JB at USAF) responsible for 63% of subject readings.
Discordance = 44.6% (66/148)
Kappa = 0.11
20.9% (31/148) positive with QFT- GIT vs. 50.7% (75/148) positive with TST (left arm), p >
Discordance = 39.0% (57/147)
Kappa = 0.20
21.1% (31/147) positive with QFT- GIT vs. 47.6% (70/147) positive with TST (right arm), p >
pos
neg 67 13 80 8 66 74 75 79
154
TST Left
Arm
1. TST Within-Subject Variability: Good agreement (discordance = 13.6%, Kappa = 0.73, SE = ± 4.35 mm). No factors associated with discordance. Ever living outside the U. S. and TST reader significantly associated with TST positivity.
2. TST – QFT-GIT Agreement: Poor agreement (discordance = 39.0% to 44.6%, Kappa = 0.11, 0.20). 47.6% to 50.7% positive with TST vs. 21% positive with QFT-GIT. No factors were significantly associated with discordance or QFT-GIT positivity.
3. TST Boosting of a Subsequent TST: Boosting: 38.0% to 40.3% of those with initial negative TST became positive. Conversion rates are greater than within-subject variability (discordance) of 13.6%, P < No factors were significantly associated with negative to positive TST conversion.
4. TST Boosting of a Subsequent QFT-GIT: Boosting: 33.5% of those with initial negative QFT-GIT became positive. Conversion rate is greater than QFT-GIT within-subject variability (discordance, previously reported) of 5.0%, P < No factors were significantly associated with negative to positive QFT-GIT conversion.
Multivariate
Model: Probability of 1) discordant result or 2) positivity. Covariates same as in Assessment #1.
Significant Associations:
Discordance: No factors were significantly associated with discordance.
QFT-GIT Positivity: No factors were significantly associated with QFT-GIT positivity.
Contact:
Emilia Ilieva-Hughes
William C. Whitworth
NCHHSTP
Division of Tuberculosis Elimination
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention, the Department of Defense, or the United States Air Force.