Sociological characteristics
A total of 60 countries with M.tb isolates >10 in the TBprofiler database was selected for analysis (Supplementary Figure 1). According to their geographical distribution, the countries were assigned to 6 continents. The sociological characteristics were described in Table 1. Northern America had the highest median population number (184.37 million, IQR: 111.06-257.69 million), GHSI (79.40, IQR: 77.35-81.45), GDP per capita (49.12 US$1000, IQR: 46.57-51.67 US$1000), and relatively longer median life expectancy (80.65 years, IQR: 79.75-81.54 years). Africa had the lowest level of median age (19.60 years, IQR: 18.10-21.50 years), life expectancy (64.07 years, IQR: 60.19-66.60 years), GHSI (35.50, IQR: 29.00-40.60), and GDP per capita (2.99 US$1000, IQR: 1.73-5.34 US$1000). Asia had the largest median population density with an estimated 230.00 per square kilometer (IQR: 120.38-376.51 per square kilometer).
Table 1
Characteristics of tuberculosis in sixty countries stratified by the six Continents in the World.
Characteristics | Continents | P-value |
Africa | Asia | Europe | Northern America | Oceania | South America |
Isolates (n=29126) | 7648 (26.3) | 7047 (24.2) | 10661 (36.6) | 1941 (6.7) | 224 (0.8) | 1605 (5.4) | |
Lineages (%)* | | | | | | | <0.001 |
Lineage 1 | 384 (5.0) | 1427 (20.2) | 722 (6.8) | 288 (14.8) | 5 (2.2) | 9 (0.6) | |
Lineage 2 | 1439 (18.8) | 3837 (54.4) | 1875 (17.6) | 381 (19.6) | 158 (70.5) | 94 (5.9) | |
Lineage 3 | 396 (5.2) | 583 (8.3) | 1950 (18.3) | 135 (7.0) | 5 (2.2) | 3 (0.2) | |
Lineage 4 | 5068 (66.3) | 1198 (17.0) | 5879 (55.1) | 1136 (58.5) | 55 (24.6) | 1337 (83.3) | |
Lineage 5 | 207 (2.7) | 0 (0.0) | 31 (0.3) | 1 (0.1) | 0 (0.0) | 0 (0.0) | |
Lineage 6 | 95 (1.2) | 0 (0.0) | 23 (0.2) | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
Lineage 7 | 47 (0.6) | 0 (0.0) | 1 (0.0) | 0 (0.0) | 1 (0.4) | 0 (0.0) | |
Lineage 9 | 1 (0.0) | 0 (0.0) | 1 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
M.bovis | 8 (0.1) | 2 (0.0) | 149 (1.4) | 0 (0.0) | 0 (0.0) | 158 (9.8) | |
M.caprae | 3 (0.0) | 0 (0.0) | 3 (0.0) | 0 (0.0) | 0 (0.0) | 4 (0.2) | |
M.orygis | 0 (0.0) | 0 (0.0) | 27 (0.3) | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
Drug Resistance types (%)* | | | | | | | <0.001 |
Sensitive | 5469 (71.5) | 3408 (48.4) | 6964 (65.3) | 1682 (86.7) | 90 (40.2) | 191 (11.9) | |
MDR | 992 (13.0) | 1204 (17.1) | 1109 (10.4) | 25 (1.3) | 93 (41.5) | 542 (33.8) | |
Pre-MDR | 549 (7.2) | 963 (13.7) | 704 (6.6) | 144 (7.4) | 28 (12.5) | 244 (15.2) | |
XDR | 249 (3.3) | 212 (3.0) | 429 (4.0) | 2 (0.1) | 5 (2.2) | 80 (5.0) | |
Pre-XDR | 237 (3.1) | 740 (10.5) | 761 (7.1) | 6 (0.3) | 8 (3.6) | 371 (23.1) | |
Others | 152 (2.0) | 520 (7.4) | 694 (6.5) | 82 (4.2) | 0 (0.0) | 177 (11.0) | |
Population(million)** | 26.38 (11.82-53.77) | 76.90 (34.48-137.41) | 10.10 (5.79-46.75) | 184.37 (111.06-257.69) | 17.22 (13.09-21.36) | 48.04 (36.03-109.42) | 0.009 |
Population density(kilometer−2)** | 79.49 (42.73-104.96) | 230.00 (120.38-376.51) | 93.11 (46.86-205.86) | 19.82 (11.93-27.72) | 10.71 (6.96-14.47) | 55.33 (29.90-66.44) | 0.009 |
Median age(year)** | 19.60 (18.10-21.50) | 31.50 (27.45-34.12) | 42.30 (40.80-43.20) | 39.85 (39.07-40.62) | 30.25 (26.43-34.07) | 29.30 (29.15-31.25) | <0.001 |
Life expectancy(year)** | 64.07 (60.19-66.60) | 75.26 (71.12-76.97) | 81.63 (76.05-82.40) | 80.65 (79.75-81.54) | 73.97 (69.24-78.70) | 75.86 (75.05-76.72) | <0.001 |
GHSI (points)** | 35.50 (29.00-40.60) | 47.05 (37.15-49.15) | 64.60 (52.30-68.70) | 79.40 (77.35-81.45) | 51.65 (39.73-63.58) | 53.40 (45.45-58.35) | <0.001 |
GDP per capita (US$1000)** | 2.99 (1.73-5.34) | 10.48 (6.23-16.98) | 39.75 (24.77-46.68) | 49.12 (46.57-51.67) | 24.24 (14.03-34.44) | 15.30 (12.49-17.34) | <0.001 |
Geographical distribution of M.tb isolates data
We collected 29126 individual isolates from included 60 countries. Among them, 18 countries were ranked as the top 30 high TB-burden countries[15]. Countries with >1000 strains were South Africa (4209), United Kingdom (4052), Netherlands (2430), Vietnam (1981), China (1911), Canada (1814), Malawi (1747), Thailand (1483), and Russia (1074), respectively. South Africa in Africa, Vietnam in Asia, the United Kingdom in Europe, Canada in Northern America, Austria in Oceania, Peru in South America had the largest number of isolates strains in separated corresponding continents (Figure 1).
Global distribution of isolates lineages
Most M.tb isolates belonged to the Euro-American lineage 4. A striking feature of lineage 4 was identified as the widespread global distribution lineage, and it was the majority lineage in 4 of the 6 continents: Africa (66.3%), Europe (55.1%), Northern America (58.5%), and South America (83.3%). Globally, lineage 4 accounted for more than half of all lineages (50.38%). Our map also showed the widespread distribution of East Asian lineage 2, which was identified in 2 of the 6 continents and was the majority lineage in Asia (54.4%) and Oceania (70.5%). Lineage 2 accounted for more than one-quarter of all lineages (26.73%) (Table 1 and Figure 2). South America had the highest proportion of Mycobacterium Bovis (M.bovis) with an estimated 9.8%, followed by Europe (1.4%).
Lineage 1 to lineage 6 accounted for 99.03% of all isolates. Among them, the largest proportion of each lineage were the sublineages of lineage 1.2.1.2.1 (635, 22.40%), lineage 2.2.1 (6458, 82.97%), lineage 3 (3072, 69.21%), lineage 4.8 (1975, 13.46%), lineage 5.1.1 (23.9, 38.08%), and lineage 6.3.1 (62, 52.54%), respectively (Figure 2). Our geographical map of major lineages also demonstrated that lineage 1 was more prevalent in South Asia (India and the Philippines). Lineage 2 was identified as the major lineage in Eastern Asia (China, South Korea, and the eastern part of Russia) and Oceania (Austria). Lineage 4 was identified as the major lineage in Europe, North American, South American, and Africa. We also found that lineage 3 was the major lineage in Norway (Europe), Saudi Arabia (Western Asia), and Iran (Southern Asia). Lineage 5 and lineage 7 were the major lineage only in Ghana (Africa) and Ethiopia (Africa), respectively (Figure 3).
Global drug-resistance and resistant lineages
The proportion of strains resistant to at least one anti-TB drug was 28.5% in Africa, 34.7% in Europe, 13.3% in Northern America, 88.1% in South America, 51.6% in Asia, and 59.8% in Oceania. The MDR and XDR strains accounted for 13.0% in Africa, 17.1% in Asia, 10.4% in Europe, 1.3% in Northern America, 41.5% in Oceania, and 33.8% in South America (Table 1).
Due to the limited samples of rare isolates of lineage 7, M.bovis, Mycobacterium caprae (M.caprae), and Mycobacterium orygis (M.orygis), we only selected the isolates of lineage 1 - lineage 6 for further drug resistance analysis. The lineage 2 had the highest prevalence of DR-TB (0.564, 95% CI: 0.553-0.576), MDR-TB (0.388, 95% CI: 0.378-0.399), and XDR-TB (0.072, 95% CI: 0.066-0.078). The lineage 4 had the second-highest prevalence of DR-TB (0.326, 95% CI: 0.319-0.334) which was similar to that of lineage 3 in DR-TB (0.326, 95% CI: 0.310-0.343), MDR-TB (0.219, 95% CI: 0.212-0.225), and XDR-TB (0.025, 95% CI: 0.022-0.027). On the contrary, lineage 6 had the lowest risk of developing DR-TB, MDR-TB, and XDR-TB with the estimated proportion of 0.119 (95% CI: 0.066-0.191), 0.042 (95% CI: 0.014-0.096), and 0, respectively. In addition, lineage 3 had the second-lowest risk of being resistant strains with the prevalence of MDR-TB (0.172, 95% CI: 0.158-0.185). Lineage 5 was also identified as isolates with the second-lowest prevalence of DR-TB (0.318, 95% CI: 0.259-0.381).
Interaction between GHSI and GDP
An overview of the interaction between ln(GHSI) and ln(GDP per capita) can be given by the twisted surface in Figure 4A. The surface was twisted, indicating that the direction of the ln(GDP per capita) effect changed with the varying values of ln(GHSI). The highest prevalence of DR-TB can be found at consecutive low values of both ln(GHSI) and ln(GDP per capita), and the ln(GHSI) did have a positive impact on the prevalence of DR-TB for low values of ln(GDP per capita). With an increase of ln(GDP per capita), the ln(GHSI) effect was attenuated. Furthermore, for higher values of ln(GDP per capita), the ln(GHSI) was negatively associated with the percentage of isolates with DR-TB.
The interaction plots in Figure 4B demonstrated that the effect estimator of ln(GHSI) on the prevalence of DR-TB was significantly lower than 0 for exceeding the threshold value of ln(GDP per capita) at 2.35. The non-significance region could also be read from this plot at the range of ln(GDP per capita) <2.35. In brief, the prevalence of DR-TB had a significantly negative association with ln(GHSI) in countries with ln(GDP per capita) ≥2.35.
Association between GHSI and DR-TB
Overall analysis on 60 countries suggested no significant relationship between the prevalence of DR-TB and ln(GHSI) (r =-0.24, 95% CI: -0.46, 0.02) (Figure 5). Subgroup analysis found that 24 countries with ln(GDP per capita) <2.35 (low income: 8, 33.33%; lower middle income: 15, 62.50%; upper middle income: 1, 4.17%) had no significant association (r =-0.13, 95% CI: -0.51, 0.28) between ln(GHSI) and DR-TB. A significantly negative association (r =-0.58, 95% CI: -0.76, -0.31) was observed in 36 countries with ln(GDP per capita) ≥2.35 (lower middle income: 1, 2.78%; upper middle income: 14, 38.89; high income: 21, 58.33%) (Figure 5) .
External verification of association between GHSI and DR-TB
To verify the representativeness, extrapolation, and extensive suitability of above conclusion, we utilized external data extracted from the top 30 high TB-burden countries[15]. No overall association was found between ln(GHSI) and the prevalence of MDR/RR in these countries among new cases (r =-0.03, 95% CI: -0.38, 0.34) or previous treated ones (r =-0.06, 95% CI: -0.41, 0.31). Subgroup analysis found that the coefficient of the prevalence of MDR/RR on ln(GHSI) turned from non-significant (ln[GDP per capita] <2.35) to significant negative (ln[GDP per capita] ≥2.35). For new cases, there was no significant association (r =0.07, 95% CI: -0.37, 0.49) in countries whose ln(GDP per capita) <2.35, but we observed a significantly negative association (r =0.70, 95% CI: -0.93, -0.08) in countries with ln(GDP per capita) ≥2.35. Similar results were also found in the previous treated cases in countries with ln(GDP per capita) <2.35 (r =0.06, 95% CI: -0.38, 0.48) and ≥2.35 (r =-0.71, 95% CI: -0.93, 0.08), respectively (Figure 6).