Table 3.
Diagnostic performance of CFP-10pep assay for TB
| Confirmed TBα | Unconfirmed TBβ | Unlikely TBγ | |||||
|---|---|---|---|---|---|---|---|
| MTB infectiona | No MTB infectionb | Total | MTB infection | No MTB infectionc | Total | ||
| All participants (N = 471) | |||||||
| CFP-10pep +, no. (%) | 6 (100.0) | 34 (79.1) | 17 (85.0) | 51 (81.0) | 6 (11.3) | 16 (4.6) | 22 (5.5) |
| CFP-10pep −, no. (%) | – | 9 (20.9) | 3 (15.0) | 12 (19.0) | 47 (88.7) | 333 (95.4) | 380 (94.5) |
| Sensitivity—% (95% CI) | 100 (54.1–100.0) | 79.1 (64.0–90.0) | 85.0 (62.1–96.8) | 81.0 (69.1–89.8) | – | – | – |
| Specificity—% (95% CI) | – | – | – | – | 88.7 (77.0–95.7) | 95.4 (92.7–97.4) | 94.5 (91.8–96.5) |
| HIV-infected participants (N = 266) | |||||||
| CFP-10pep +, no. (%) | 5 (100.0) | 24 (80.0) | 12 (92.3) | 36 (83.7) | 3 (9.4) | 12 (6.4) | 15 (7.7) |
| CFP-10pep −, no. (%) | – | 6 (20.0) | 1 (7.7) | 7 (16.3) | 29 (90.6) | 174 (93.6) | 203 (92.3) |
| Sensitivity—% (95% CI) | 100 (47.8–100.0) | 80.0 (61.4–92.3) | 92.3 (64.0–99.8) | 83.7 (69.3–93.2) | – | – | – |
| Specificity—% (95% CI) | – | – | – | – | 90.6 (75.0–98.0) | 93.6 (89.0–96.6) | 93.1 (88.9–96.1) |
| HIV-uninfected participants (N = 205) | |||||||
| CFP-10pep +, no. (%) | 1 (100.0) | 10 (76.9) | 5 (71.4) | 15 (75.0) | 3 (14.3) | 4 (2.4) | 7 (4.3) |
| CFP-10pep −, no. (%) | – | 3 (23.1) | 2 (28.6) | 5 (25.0) | 18 (85.7) | 159 (97.6) | 177 (96.2) |
| Sensitivity—% (95% CI) | 100 (2.5–100.0) | 76.9 (46.2–95.0) | 71.4 (29.0–96.3) | 75.0 (50.9–91.3) | – | – | – |
| Specificity—% (95% CI) | – | – | – | – | 85.7 (63.7–97.0) | 97.6 (93.8–99.3) | 96.2 (92.3–98.5) |
CI confidence interval
αConfirmed TB: Mtb culture-positive
βUnconfirmed TB: ≥ 2 types of non-bacteriological TB evidence within 12 weeks
γUnlikely TB: no or insufficient evidence of TB or a confirmed alternative diagnosis (did not have active TB)
aTST positive (induration of ≥ 5 mm for HIV-infected children and ≥ 10 mm for HIV-uninfected children)
bTST negative (induration of < 5 mm for HIV-infected children and < 10 mm for HIV-uninfected children)
cTST-negative or no TST result available for children not suspected of TB who did not reach a protocol-designated visit for TST performance (parent study weeks 96, 144, 196)
CFP-10pep diagnostic sensitivity was calculated using the sample drawn closest (± 24 weeks; ± 1 study visit for the HIV-uninfected cohort following INH/placebo treatment) to the time of their retrospective diagnosis by 2015 NIH criteria