Diagnostic process
The flow diagram of patients' diagnostic process is presented in Fig. 1. Week 1, 70 patients were diagnosed with smear/NAAT-positive outcome either based on sputum or BALF and 67 out of them immediately started anti-TB therapy. Sixty-one among the rest patients, except for 16 inactive PTB patients and 7 patients with positive culture of common bacteria, were undergone about two weeks of empiric antibacterial therapy (azithromycin + cefixime). Week 2–3, 27 out of 36 nonresponders to empiric antibacterial therapy subsequently started anti-TB therapy. Week ཞ8, final diagnosis of all patients was made. Three nonresponders were confirmed to be infected by non-tuberculosis mycobacterium and 2 of them were excluded from the cohort of anti-TB. Five additional patients started anti-TB therapy due to the positive result of Mtb culture. The overall anti-TB rate was 94.2% (97/103). The anti-TB delay rates caused by empiric antibacterial therapy (Week 2–3) and Mtb culture (Week ཞ8) were 33.0% (34/103) and 4.9% (5/103), respectively. Seven cases were grouped into gray zone (gz-PTB) because they were IGRA-positive, but well responded to antibiotic treatment or had a positive result of common bacterial culture. Among active PTB patients, 82 (79.6%) cases, 33 (32.0%) based on sputum and additional 49 (47.6%) depended on BALF, were microbiologically confirmed. Among those patients undergone anti-TB therapy without microbiologic diagnosis, 15 out of 34 cases were confirmed by Mtb culture. All patients were divided into different groups based on the final diagnosis (Table 1).
Table 1
Demographics and clinical characteristics of subjects
|
Active PTB
|
Inactive PTB
(n =16)
|
gz-PTB†
(n =7)
|
Non-PTB
(n =28)
|
Characteristics
|
Sputum-based
(n =33)
|
BALF-dependent
(n =49)
|
Clinical
(n =21)
|
Gender (M/F)
|
24/9
|
31/18
|
16/5
|
13/3
|
4/3
|
13/15
|
Age (Years old)
|
43.4±14.8
|
47.2±15.3
|
40.8±16.5
|
50.2±12.6
|
58.5±12.2
|
53.0±15.7
|
TB contact
|
2
|
3
|
0
|
0
|
0
|
0
|
Smoking
|
7
|
14
|
4
|
6
|
1
|
4
|
T2DM
|
5
|
12
|
1
|
2
|
1
|
1
|
GHbA1C (%)
|
6.78±0.43
|
6.62±0.38
|
5.84±0.33
|
5.95±0.33
|
5.82±0.24
|
5.61±0.17
|
WBC (109/L)
|
7.08±1.96
|
6.43±2.68
|
5.98±1.63
|
6.34±2.00
|
6.77±3.34
|
6.82±3.22
|
Neutrophil (%)
|
68.3±10.5
|
62.3±11.6
|
63.2±9.32
|
63.0±11.6
|
60.1±17.2
|
64.1±11.1
|
ESR (mm/h)
|
43.3±5.67
|
40.3±5.22
|
17.4±4.64
|
27.6±7.48
|
32.0±9.46
|
33.2±5.27
|
CD4+T (106/L)
|
507±32.0*
|
554±44.0
|
541±56.0
|
613±61.0
|
720±175
|
736±103
|
Cavitary lesion
|
24
|
13
|
9
|
0
|
1
|
3
|
IGRA (+)
|
32/32‡
|
40/48
|
18
|
15/15
|
7
|
0
|
Sputum (+)
|
|
|
|
|
|
|
Smearing
|
25
|
0/48
|
0/19
|
0
|
0
|
0/26
|
TB-DNA
|
21/24
|
0/33
|
0/15
|
0/11
|
0/5
|
0/19
|
Xpert test
|
12/12
|
0/6
|
0/3
|
0/2
|
0/3
|
0/6
|
BALF (+)
|
|
|
|
|
|
|
Smearing
|
21
|
8
|
0
|
0
|
0
|
0/27
|
TB-DNA
|
27
|
9/46
|
0
|
0
|
0
|
0/27
|
Xpert test
|
30/32
|
34/47
|
0/20
|
0/14
|
0
|
0/21
|
Mtb culture (+)
|
25
|
34/47
|
0
|
0
|
0
|
0/27
|
PTB, pulmonary tuberculosis; BALF, bronchoalveolar lavage fluid; T2DM, type 2 diabetes; GHbA1C, glycated hemoglobin A1c; WBC, white blood cell; ESR, erythrocyte sedimentation rate; IGRA, interferon-γ release assay; Xpert, GeneXpert MTB/RIF; Mtb, Mycobacterium tuberculosis; †gz-PTB, gray zone-PTB, IGRA-positive, but well responded to antibiotic treatment or had a positive result of common bacterial culture; ‡Actual cases involved; *P < 0.05 when compared with inactive PTB patients. |
BALF levels of sCD14, IL-17 and IL-22 is similar between PTB and non-PTB
BALF levels of sCD14, IL-17 and IL-22 have been reported to be increased in PTB and non-PTB patients [18, 23, 25–27], but there is a lack of direct comparison. Here, all of them in BALF were found to be comparable between PTB and non-PTB patients (Fig. 2A). Though BALF level of IL-17 was significantly higher in gz-PTB, the significance is limited due to its small sample size.
BALF levels of sCD14, IL-17 and IL-22 are associated with active PTB
Compared with patients with inactive PTB, patients with active PTB had significantly higher levels of sCD14 and IL-22, but comparable level of IL-17 (Fig. 2B). Since active PTB patients (Table S1), especially those patients diagnosed based on sputum (Table 1), were slightly younger than inactive PTB patients, the correlation analyses between age and BALF levels of sCD14, IL-17and IL-22 were conducted (Fig. S1). The un-correlations suggest that the age was not the contributor of the differences in BALF levels of sCD14 and IL-22 between patients with active and inactive PTB.
BALF levels of sCD14, IL-17 and IL-22 are associated with bacteriological confirmation manners in active PTB patients
The final diagnosis of active PTB was made in sputum-based, BALF-dependent and clinical manners (Table 1). Among these three manners, sputum-based and BALF-dependent manners had significantly higher BALF levels of sCD14, IL-17 and IL-22 than clinical manner (Fig. 3A). The IL-17 level of sputum-based manner was also significantly higher than that of BALF-dependent manner. Though there were slight differences in age among these groups (Table 1), those differences were not the contributor of the associations between bacteriological confirmation manners and BALF levels of sCD14, IL-17 and IL-22 since the age was not correlated with the BALF levels of those factors (Fig. S1). In BALF-dependent group, only one patient neither Xpert test nor Mtb culture was positive. The rest patients were divided into Xpert test-, Mtb culture- and Xpert test/Mtb culture-positive groups. BALF levels of sCD14, IL-17 and IL-22 in Mtb culture-positive group were relatively lower, but not significant than those in Xpert test- or Xpert test/Mtb culture-positive groups (Fig. 3B).
BALF levels of sCD14, IL-17 and IL-22 predict bacteriological confirmation failure
The associations with PTB bacteriological confirmation manners suggest that BALF levels of sCD14, IL-17 and IL-22 can predict bacteriological confirmation failure. Indeed, BALF levels of sCD14, IL-17 and IL-22 were significantly higher in bacteriologically-confirmed patients, and all of them were significant predictors of bacteriological confirmation (Fig. 4A). When the sensitivities of bacteriological confirmation failure prediction were set at test thresholds of the WHO minimal requirement for triage tests (≥ 90%) [16], the specificities of BALF levels of sCD14, IL-17 and IL-22 were 54.9%, 67.1% and 62.2%, respectively, very close to the WHO requirement (≥ 70%) [16], especially for those of IL-17 and IL-22 (Table 2).
Table 2
Sensitivities and specificities of BALF levels of sCD14, IL-17 and IL-22 in predictions of bacteriological confirmation, antibiotic treatment, Xpert test and Mtb culture in PTB patients
Items
|
Sensitivity†
(%)
|
Specificity
(%)
|
Cut-off value
(µg or ng/mL)
|
IGRA (+) patients
|
Sensitivity
(%)
|
Specificity
(%)
|
Bacteriological
confirmation failure
|
|
|
|
|
|
sCD14
|
90.4
|
54.9
|
≤ 3.05
|
-
|
-‡
|
IL-17
|
90.4
|
67.1
|
≤ 1.23
|
-
|
-
|
IL-22
|
90.4
|
62.2
|
≤ 2.04
|
-
|
-
|
Empiric anti-bacterial nonresponder §
|
|
|
|
|
|
sCD14
|
91.2
|
33.3
|
≤ 3.72
|
100.0
|
66.7
|
IL-17
|
91.2
|
37.0
|
≤ 2.17
|
92.0
|
66.7
|
IL-22
|
91.2
|
14.8
|
≤ 5.80
|
96.0
|
66.7
|
Xpert test (+)
|
|
|
|
|
|
sCD14
|
90.3
|
37.3
|
≥ 1.47
|
-
|
-
|
IL-17
|
90.3
|
31.4
|
≥ 0.72
|
-
|
-
|
IL-22
|
90.3
|
59.2
|
≥ 1.52
|
-
|
-
|
Mtb culture (+)
|
|
|
|
|
|
sCD14
|
91.5
|
27.5
|
≥ 1.46
|
-
|
-
|
IL-17
|
91.5
|
8.6
|
≥ 0.64
|
-
|
-
|
IL-22
|
91.5
|
36.2
|
≥ 1.42
|
-
|
-
|
IGRA, interferon-γ release assay; Xpert, GeneXpert MTB/RIF; Mtb, Mycobacterium tuberculosis; † minimal requirement of WHO (≥ 90%); ‡ IGRA is unsuitable for bacteriological confirmation in PTB; § nonresponders to empiric anti-bacteria (potential PTB patients). |
BALF levels of sCD14, IL-17 and IL-22 predict nonresponders to empiric antibacterial therapy in active PTB patients
At the end of Week 1, 61 smear/NAAT-negative patients needed empiric antibacterial therapy. On Week 2–3, there were a total of 36 nonresponders. Since 2 of them were diagnosed with infection of nontuberculous mycobacterium at last (Fig. 1), the exact anti-TB delay rate on Week 2–3 was 33.0% (34/103). The nonresponders in active PTB patients had significantly lower BALF levels of IL-17 and IL-22 than responders, and the BALF levels of IL-17 and IL-22 were significant predictors of empiric antibacterial therapy (Fig. 4B), suggesting that BALF levels of IL-17 and IL-22 may help avoiding anti-TB delay by predicting nonresponders to empiric antibacterial therapy. When matching the WHO minimal requirement for sensitivity of a triage test (≥ 90%) [16], nonresponders-predicting specificities of the BALF levels of sCD14, IL-17 and IL-22 were 33.3%, 37.0% and 14.8%, which were far below the WHO requirement for specificity (≥ 70%) (Table 2). However, together with IGRA that is usually used to differentiate TB from non-TB, the sensitivities of BALF levels of sCD14, IL-17 and IL-22 all were superior to WHO requirement and the specificities (66.7%) were very close to the WHO requirement (Table 2).
BALF levels of sCD14, IL-17 and IL-22 predict results of Xpert test and Mtb culture
In this study, Xpert test and Mtb culture were the major bacteriological confirmation measures, especially in patients who etiologically diagnosed dependent on BALF (Table 1). Therefore, the associations of bacteriological confirmation manners suggest that BALF levels of sCD14, IL-17 and IL-22 may have predictive values for the results of Xpert test and Mtb culture. Among 119 cases of active and inactive PTB patients, Xpert test was conducted in 113 cases with 62 positive and 51 negative results, respectively. The BALF level of IL-22 rather than sCD14 and IL-17 was significantly higher in Xpert-positive patients, but all of them were significant predictors of Xpert test (Fig. 4C). When matching the WHO minimal requirement for sensitivity of a triage test (≥ 90%) [13], Xpert test-predicting specificities of BALF levels of sCD14, IL-17 and IL-22 were 37.3%, 31.4% and 59.2%, respectively (Table 2). Those specificities were far below the WHO requirement for specificity (≥ 70%) [16]. Mtb culture was conducted in 117 cases with 59 positive and 58 negative results. The BALF levels of sCD14, IL-17 and IL-22 in Mtb culture-positive patients were all insignificantly higher, but sCD14 and IL-22 were significant predictors of Mtb culture (Fig. 4D). Similarly, Mtb culture-predicting specificities were very low if matching the WHO minimal requirement for sensitivity (Table 2).
Significance of serum levels of sCD14, IL-17 and IL-22 is limited in PTB patients
Compared with BALF, serum is more convenient to be obtained in clinical practice. A total of 76 out of 119 PTB patients had serum data of sCD14, IL-17 and IL-22. The average serum levels of sCD14 and IL-22 but IL-17 were significantly lower than average BALF levels in active PTB patients and to some extent among different diagnostic manners (Table S2). The serum levels of sCD14, IL-17 and IL-22 were comparable in patients with active and inactive PTB, and were not associated with bacteriological conformation manners (Fig. S2). Concordantly, serum levels of sCD14, IL-17 and IL-22 were unable to predict the results of Xpert test and Mtb culture (Fig. S3), and the bacteriological confirmation failure prediction performance was far below the minimal requirement of WHO (Table S3).