2.1 Basic information of the subjects
This study included 479 participants, including 297 ATB patients and 182 controls, and the controls included 103 OPD patients and 79 healthy individuals, and the basic information is shown in Table 1. There was no statistical difference in gender and age distribution between the ATB and OPD groups (p> 0.05). 14.8% ATB and 7.8% OPD had a previous history of TB (p> 0.05). At least one TB-related symptom was reported in 78.1% of ATB patients, compared with 64.1% of OPD patients (p =0.04). A younger age and higher proportion of women in the HC group compared with ATB and OPD (p <0.05). Of 297 ATB patients, sputum smear positive was 34.0%, culture positive 44.5% (118 / 256) and Xpert MTB / RIF positive 54.1% (120 / 222). ATB patients were divided into 6 groups, with 40% within 3 days of medication.
Table 1: Clinical data of the three groups of study subjects
|
ATB patients
n(%)
|
OPD n(%)
|
Healthy controls n(%)
|
Total
|
297
|
103
|
79
|
Age
|
|
|
|
<65y
|
173(58.2)
|
47(45.6)
|
76(96.2)
|
≥65y
|
124(41.8)
|
56(54.4)
|
3(3.8)
|
Sex
|
|
|
|
Female
|
98 (33.0)
|
45(43.7)
|
59(74.7)
|
Male
|
199(67.0)
|
58(56.3)
|
20(25.3)
|
*Any TB symptoms
|
|
|
|
No
|
65(21.9)
|
37(35.9)
|
NA
|
Yes
|
232(78.1)
|
65(64.1)
|
NA
|
History of TB
|
44(14.8)
|
8(7.8)
|
NA
|
Drug-resistant TB
|
43(14.5)
|
NA
|
NA
|
Smear positive
|
101(34.0)
|
6/41(6.83)
|
NA
|
Culture positive(TB)
|
118/256(44.5)
|
0/39(0)
|
NA
|
GeneXpert MTB/RIF
|
|
|
|
Positive or weak positive
|
120/222(54.1)
|
0/37(0)
|
NA
|
Negative
|
119/244(45.9)
|
37/37(100)
|
NA
|
Anti-TB treatment time
|
|
|
|
<3d
|
119(40.0)
|
NA
|
NA
|
3d-2w
|
68(22.9)
|
NA
|
NA
|
2w-2m
|
46(15.5)
|
NA
|
NA
|
2m—4m
|
25(8.4)
|
NA
|
NA
|
4m—6m
|
17(5.7)
|
NA
|
NA
|
>6m
|
22(7.4)
|
NA
|
NA
|
2.2 Distribution of genes and TB scores in the three population groups
We compared the expression of genes GBP5 and DUSP3 and TB scores in the three populations. In ATB patients, the Ct value of GBP5 was significantly lower than in the other two groups (p <0.0001), although the Ct value in the OPD group was also significantly lower than HC group (p <0.05), the difference was relatively small (Figure 1A). The Ct value of the DUSP3 gene was significantly lower in the ATB group than in the HC group (p <0.001), but not significant from the OPD group (Figure 1B). The TB score values varied significantly between the three groups (p <0.0001), and the overall TB score was significantly lower in the ATB group compared to the other two groups (Figure 1C).
2.3 Diagnostic performance of the TB score for the ATB patients
After analysis of the diagnostic performance of TB score, the AUC of TB score was 0.879 (95% CI, 0.844 ~ 0.914), and the optimum sensitivity was 76.09% and 93.27% specific by the Youden index (Figure 2A). When the specificity was fixed at 70%, the sensitivity of the TB score was 96.2%, satisfying the minimum TPP criteria for the triage test. In contrast, the TB score had a lower AUC for identifying ATB and OPD: 0.689 (95% CI, 0.628 to 0.750) (Fig. 2A).
Next, we analyzed whether age, sex had an effect on the diagnostic performance of TB score in R language. A new model was included in the ROC model of TB score identifying ATB from OPD, and Delong’s tested no statistical difference in AUC between the two ROC models (p> 0.05) (Figure 2B). With the treatment of ATB patients, the results of pathogenic examination turned negative. We compared the positive detection rate of TB culture, Xpert MTB / RIF test and TB score for the enrolled ATB population. According to the results, the detection rate of ATB by TB score was much higher than culture and Xpert MTB / RIF in both the type of sample of sputum and BALF obtained from invasive sampling (Table 2). When TB score and TB culture were combined detection, the diagnostic efficacy was not significantly improved (p>0.05) (Figure 2C), and when TB score and Xpert MTB / RIF were combined diagnosed, the diagnostic efficacy was significantly improved compared with TB score alone (p <0.00001) (Figure 2D).
Table 2 : Comparison of TB score, tuberculosis culture and Xpert MTB / RIF detection in the detection rate of ATB patients.
Sample type
|
Sputum sample
|
BALF
|
Detect method
|
Xpert MTB / RIF
|
MTB culture
|
And Xpert MTB / RIF
|
MTB culture
|
Positive number
|
69
|
63
|
51
|
51
|
Negative number
|
59
|
100
|
42
|
42
|
Total
|
128
|
163
|
93
|
93
|
ATB, relevance ratio
|
0.5391
|
0.3865
|
0.5484
|
0.5484
|
The detection rate of ATB by TB score
|
0.7969
|
0.7546
|
0.7849
|
0.7634
|
2.3 Effect of treatment duration of ATB patients on TB score
To assess the effect of treatment duration on the detection of TB scores, we divided ATB patients into six subgroups according to the duration of anti-TB drug use and observed the distribution of TB scores in the different subgroups. The results showed that after the start of treatment, the mean TB score of TB patients increased significantly, and decreased to the initial level 2 months later. With the further increase of treatment duration, the mean TB score increased from -1.591 to -0.972, gradually approached the cutoff value and tended to stabilize (Figure 3A). To reduce the interference of treatment factors, we separately analyzed ATB patients within 3 days and found that TB scores within 2 months of medication was higher than that of patients within 3 days of medication, but the difference was not statistically significant (p> 0.05). However, when administered, older than 2 months, the patients' TB scores changed significantly (p <0.01) (Figure 3B). Next, we did the ROC curve to further analyze whether the difference in treatment duration will affect the detection of ATB patients by TB score.The results showed that the AUC of TB score was 0.895 (95%CI: 0.857~0.934) in distinguishing ATB patients and healthy people within 3 days of medication, and the optimal sensitivity was 79.68% and the specificity was 92.41%. When OPD patients were used as a negative control, the AUC of ATB patients within 3 days of medication was 0.715 (95%CI: 0.642 ~0.770), and the optimal sensitivity was 76.47% and the specificity was 61.17%. For ATB patients within 2 months of medication, the AUC distinguishing TB score from OPD patients was 0.720 (95%CI: 0.651 to 0.790), similar to ATB patients within 3 days (Figure 3C).
2.3 Correlation between bacterial load and TB score
We collected concurrent laboratory findings from ATB patients and performed descriptive analysis of TB scores for smear microscopy, Xpert MTB / RIF testing, TB culture, and IGRA test results as shown in Table 3. Patients with smear positive ATB had lower TB scores (mean: -1.92) than those with smear negative (mean: -1.16). Similarly, ATB patients with high Xpert MTB / RIF semi-quantitative results ( mean : -2.07 ) were significantly lower than those with very low semi-quantitative results ( mean : -1.35 ).The TB scoreof ATB patients with positive tuberculosis culture ( mean : -1.83 ) was lower than that of ATB patients with negative culture ( mean : -1.18 ).In addition, the TB score of ATB patients who tested negative for IGRA (mean: -1.82) was lower than the TB score tested positive for IGRA (mean: -1.47).
To further explore the relationship between MTB bacterial burden and TB score, we divided ATB patients with positive for Xpert MTB / RIF test into four groups: high, moderate, low and very low load based on the semi-quantitative results of bacterial burden. Spearman correlation analysis showed that the bacterial load was significantly negatively correlated with TB score ( p< 0.01 ). In the group comparison, ATB patients with high / medium load had significantly lower TB scores than those with low load (p <0.01) and those with very low load (p<0.001). However, although the TB score was lower in patients with low load than those with very low load, the difference was not statistically significant (p> 0.05). Similarly, we found that only 5 out of the 18 enrolled clinically diagnosed cases were identified by the TB score (Figure 4A). Next, we separately analyzed the performance of the TB score in patients with bacteria-positive ATB (Culture or Xpert MTB / RIF test positive). ROC results showed that the AUC of TB score distinguishing patients with positive ATB and healthy persons was 0.952 (95%CI: 0.651 ~ 0.790), optimal sensitivity was 90.54% and specificity was 93.67%; AUC was 0.778 (95%CI: 0.716-0.839), and optimal sensitivity was 89.86% and specificity was 61.17% (Figure 4B).
Considering the influence of treatment duration and bacterial volume load on TB score simultaneously, we analyzed bacteria-positive ATB patients within 3 days of medication. The ROC curve results showed that the AUC was 0.936 (95%CI: 0.890 ~ 0.981) and 0.788 (95%CI: 0.717~ 0.859), respectively, with the optimal sensitivity of 84.38% and 81.25%, and specificity of 97.47% and 70.87%, respectively (Figure 4C). When considering the effect of treatment duration, bacteria-negative ATB patients(Culture and Xpert MTB / RIF test were negative)within 3 days were analyzed, the AUC of TB score was 0.808 (95%CI: 0.715-0.901), with an optimal sensitivity of 70.00% and a specificity of 86.08%. However, the TB score that the AUC between bacteria-negative patients and OPD patients was only 0.595 (95%CI: 0.492 and 0.699) (Figure 4D).
Table3:Comparison of 3-gene TB score in participants with ATB against smear, Xpert MTB/RIF, culture time to positivity and IGRA results.
|
TB score
|
N
|
Minimum
|
Maximum
|
Mean
|
Medium
|
IQR
|
Varience
|
Standard deviation
|
Smear
|
Positive
|
101
|
-5.10
|
0.15
|
-1.92
|
-1.85
|
-2.65, -1.85
|
1.01
|
1.00
|
|
Negative
|
182
|
-3.80
|
0.55
|
-1.16
|
-1.02
|
-1.67, -0.43
|
0.85
|
0.92
|
MTB/RIF
|
Negative
|
102
|
-3.14
|
0.35
|
-1.25
|
-1.06
|
-1.92, -0.57
|
0.86
|
0.93
|
|
Very low
|
29
|
-3.70
|
0.06
|
-1.35
|
-1.31
|
-1.71, -0.81
|
0.65
|
0.80
|
|
Low
|
45
|
-3.65
|
-0.05
|
-1.56
|
-1.34
|
-2.14, -0.89
|
0.78
|
0.88
|
|
Medium
|
45
|
-4.23
|
-0.44
|
-2.20
|
-2.25
|
-2.91, -1.58
|
0.84
|
0.91
|
|
High
|
11
|
-4.35
|
0.15
|
-2.07
|
-2.05
|
-3.13, -0.95
|
1.82
|
1.35
|
Culture
|
Positive
|
114
|
-5.10
|
0.15
|
-1.83
|
-1.68
|
-2.60,-1.08
|
1.03
|
1.01
|
|
Negative
|
142
|
-3.70
|
0.35
|
-1.18
|
-0.99
|
-1.71,-0.52
|
0.83
|
0.91
|
IGRA
|
Positive
|
210
|
-4.23
|
0.35
|
-1.47
|
-1.33
|
-2.19,-1.33
|
0.91
|
0.95
|
|
Negative
|
27
|
-5.10
|
0.00
|
-1.82
|
-1.78
|
-2.65,-0.81
|
1.64
|
1.28
|