Maximum likelihood phylogeny (GTR substitution model, 1000 resamples) of 277 MDR-MTBC isolates from Karakalpakstan, Uzbekistan sampled from 2001 to 2006. Columns show drug resistance associated …
Genealogical tree of CAO strains in Karakalpakstan, Uzbekistan and effective population size over time based on a (piecewise-constant) Bayesian skyline approach using the GTR substitution model and …
Comparisons between isolates carrying compensatory mutations (in orange) and isolates with no-compensatory mutations (in blue), from the Karakalpakstan dataset. (A) Boxplot showing number of …
Shown are the coefficients and 95% confidence bands of multiple linear regression of the transmission success score, defined as the size of clusters diverging by at most N SNPs and divided by N or, …
Any resistance associated mutations (see methods) for individual antibiotics are depicted with red bars for each isolate. The presence of any putative compensatory mutation in the RNA polymerase …
MDR-MTBC isolates stratified to three Beijing clades, other Beijing isolates and non-Beijing isolates. Proportions of isolates with identified molecular drug resistance mutations (see Supplementary …
Box represents inter quartile range, whiskers represent 95% of the data, outliers shown as black dots; solid black line represents the median.
P-values for pairwise comparisons within groups were calculated with Fisher exact test (two-sided). Beijing CAO 2001/2002 vs 2005/2006 p=0.0018, Beijing CAO 2003/2004 vs 2005/2006 p=0.0002.
Box represents inter quartile range, whiskers represent 95% of the data, outliers shown as black dots; solid black line represents the median. Beijing CAO isolates exhibit more phenotypic drug …
Box-Plot with mean (diamond) and median (horizontal line) number of genotypic drug resistances (see methods) to additional anti-TB drugs (beyond MDR defining rifampicin and isoniazid resistance). …
(A) Boxplot showing number of resistance mutations for the two categories (without or with compensatory mutations). The two categories were significantly different (two-sample t-test p<2.2×10−16). (B…
Shown are the coefficients and 95% confidence bands of multiple linear regression of the transmission success score, defined as the size of clusters diverging by at most N SNPs and divided by N or, …
Each node/isolate is labelled with its transmission index, i.e. number of other isolates with a maximum distance of 10 SNPs. Branches are indicated with number of unique SNPs.
Cohort 1 | Cohort 2 | P value | |
---|---|---|---|
Year of isolate collection (patient diagnosis with MDR-TB) | 2001–2002 | 2003–2006 | |
No. MDR cases diagnosed within time period | 57 | 300 | |
No. Included in this analysis Reasons for non-inclusion: Multiple strain infection No DNA available Patient already in cohort 1 Low DNA quantity | 49 (86%) 6 2 NA 0 | 228 (76%) 1 40 11 20 | 0.094 |
Patient residence (within Karakalpakstan) Nukus Chimbay Other Unknown | 34 (69%) 6 9 0 | 146 (64%) 64 1 17 | 0.49 |
Male | 27 (55%) | 119 (52%) | 0.72 |
Age (median, IQR) Missing age | 32, 27–38 0 | 31, 24–41 49 (21%) | 0.40 |
Previous TB treatment | 38 (78%) | 228 (100%) | <0.0001 |
First-line resistance profile: HR HRE HRS HRES HRSZ HREZ HRESZ | 1 0 12 (24%) 28 (57%) 1 (2%) 1 7 (14%) | 2 1 41 (18%) 49 (21%) 27 (12%) 1 107 (47%) | <0.0001 |
No. of first-line drugs resistant 2 3 4 5 | 1 12 (24%) 30 (61%) 7 (14%) | 2 42 (18%) 77 (34%) 107 (47%) | <0.0001 |
Availability of second-line drug susceptibility testing (DST) | Ofx, Cap, Proth | Ofx, Cap, Ami, Eth, Cyc, PAS | |
Ofx resistance | 5 (10%) | 6 (3%) | 0.033 |
Cm resistance | 1 (2%) | 53 (23%) | 0.0001 |
Abbreviations: H = isoniziad, R = rifampicin, E = ethambutol, S = streptomycin, Z = pyrazinamide, Ofx = ofloxacin, Cm = capreomycin
Comparisons of Beast runs using a combined dataset of Central Asian outbreak (CAO) isolates originated from Germany (1995 – 2000), Karakalpakstan (2001–2006), and Samara (2008 – 2010) as well as …
Clock model | Demographic model | Marginal L estimate | Mean ESS | Δ marginal L estimate | Subst rate x 10−7 (95%HPD) | MRCA and rpoC N698S mean node age (95%HPD) |
---|---|---|---|---|---|---|
Combined CAO dataset for clock model comparison | ||||||
Strict (no tip dating) | Coalescent constant size | −10131.67 | 4302 | 32.21 | 1.0 (fixed) | 41.5 (30.6–49.1) NA |
Strict (tip dating) | Coalescent constant size | −10099.46 | 4041 | ref | 1.0 (fixed) | 42.9 (34.3–50.3) NA |
Relaxed, lognormal | Coalescent constant size | −10117.21 | 1303 | 17.75 | 0.96 (0.65–1.24) | 57.6 (34.4–84.5) NA |
Combined CAO dataset for molecular clock estimate among CAO strains | ||||||
Relaxed, lognormal | Coalescent constant size | −10117.21 | 1303 | 78.28 | 0.96 (0.65–1.24) | 57.6 (34.4–84.5) NA |
Relaxed, lognormal | Exponential | −10044.41 | 1266 | 5.48 | 0.88 (0.58–1.21) | 40.5 (26.4–53.2) NA |
Relaxed, lognormal | Bayesian skyline | −10038.93 | 924 | ref | 0.94 (0.72–1.15) | 37.1 (25.7–44.0) NA |
CAO-Karakalpakstan dataset for demographic model comparison (under best clock estimate) | ||||||
Strict (tip dating) | Bayesian skyline | −7617.09 | 2874 | 3.79 | 0.94 (fixed) | 32.2 (23.9–37.3) 16.1 (11.6–16.9) |
Strict (tip dating) | Coalescent constant size | −7667.92 | 4231 | 54.62 | 0.94 (fixed) | 37.5 (30.2–45.1) 15.8 (12.3–18.6) |
Strict (tip dating) | Exponential | −7613.30 | 4003 | ref | 0.94 (fixed) | 29.3 (23.5–33.7) 15.4 (12.4–20.1) |
CAO-Karakalpakstan dataset, exponential growth model, upper and lower 95% HPD values (from best clock estimate) | ||||||
Strict (tip dating) | Exponential | −7610.94 | 3926 | ref | 0.72 (fixed) | 36.4 (30.4–43.3) 18.4 (15.8–21.6) |
Strict (tip dating) | Exponential | −7613.30 | 4003 | 2.36 | 0.94 (fixed) | 29.3 (23.5–33.7) 15.4 (12.4–20.1) |
Strict (tip dating) | Exponential | −7621.22 | 4031 | 10.28 | 1.15 (fixed) | 24.4 (20.0–25.8) 12.9 (10.8–14.4) |
CAO-Karakalpakstan dataset, skyline model, upper and lower 95% HPD values (from best clock estimate) | ||||||
Strict (tip dating) | Bayesian skyline | −7611.12 | 2694 | ref | 0.72 (fixed) | 39.5 (30.7–47.8) 18.8 (14.6–21.5) |
Strict (tip dating) | Bayesian skyline | −7617.09 | 2874 | 5.97 | 0.94 (fixed) | 32.2 (23.9–37.3) 16.1 (11.6–16.9) |
Strict (tip dating) | Bayesian skyline | −7619.71 | 2763 | 8.59 | 1.15 (fixed) | 25.2 (20.2–30.7) 11.8 (10.1–13.9) |
Abbreviations: HPD = Highest posterior density interval
Data from 277 MDR-MTBC isolates from Karakalpakstan, Uzbekistan, stratified to the particularly successful variant termed Central Asian outbreak (CAO) and other Beijing isolates. Pairwise …
Beijing CAO (n = 173) | Other Beijing (n = 64) | P-value | All (n = 277) | |
---|---|---|---|---|
rpoB mutations outside RRDR, excluding codon 170,400,491 variants wild type | 25 (14.5%) 147 (85.0%) | 12 (18.8%) 52 (81.3%) | 0.43 | 43 (15.5%) 234 (84.5%) |
rpoC variants wild type | 95 (54.9%) 78 (45.1%) | 18 (28.1%) 46 (71.2%) | 0.0002 | 126 (45.5%) 151 (54.5%) |
rpoA variants wild type | 5 (2.9%) 168 (97.1%) | 2 (3.1%) 62 (96.9%) | 1.00 | 7 (2.5%) 270 (97.5%) |
Abbreviations: CAO = Central Asian outbreak, RRDR = rifampicin resistance determining region
Group | S | E | Z | Km | Am | Cm | Fq | Thio | PAS | Pre-XDR XDR |
---|---|---|---|---|---|---|---|---|---|---|
Beijing CAO (n = 201) | 201/201 100.0% | 195/201 97.0% | 152/201 75.6% | 97/201 48.3% | 37/201 18.4% | 37/201 18.4% | 6/201 3.0% | 121/201 60.2% | 99/201 49.3% | 100/201 49.8% |
Beijing clade B (W148) (n = 103) | 103/103 100.0% | 83/103 80.6% | 44/103 42.7% | 61/103 59.2% | 18/103 17.5% | 18/103 17.5% | 23/103 22.3% | 75/103 72.8% | 12/103 11.7% | 64/103 62.1% |
Beijing clade A (n = 187) | 184/187 98.4% | 183/187 97.9% | 163/187 87.2% | 177/187 94.7% | 0/187 0.0% | 0/187 0.0% | 33/187 17.6% | 180/187 96.3% | 7/187 3.7% | 179/187 95.7% |
Other Beijing (n = 100) | 91/100 91.0% | 73/100 73.0% | 52/100 52.0% | 39/100 39.0% | 20/100 20.0% | 23/100 23.0% | 14/100 14.0% | 32/100 32.0% | 15/100 15.0% | 45/187 24.1% |
Non-Beijing (n = 114) | 69/114 60.5% | 63/114 55.3% | 30/114 26.3% | 39/114 34.2% | 14/114 12.3% | 14/114 12.3% | 3/114 2.6% | 34/114 29.8% | 34/114 29.8% | 40/114 35.1% |
Abbreviations: S = streptomycin, E = ethambutol, Z = pyrazinamide, Km = kanamycin, Am = amikacin, Cm = Capreomycin, Fq = fluoroquinolone, Thio = thioamide, PAS = para aminosalicylic acid
Best model is assumed to have the lowest criteration value. Shown are the top 10 AIC models. Substitution model used for Bayesian inference marked in bold.
Subst. model | -lnL | AIC | Δ AIC | BIC | Δ BIC |
---|---|---|---|---|---|
GTR | 8837.6437 | 18567.2875 | 0.0 | 21041.0025 | 7.0748 (2) |
GTR + I | 8837.6747 | 18569.3494 | 2.0619 (2) | 21048.6109 | 14.6832 (5) |
GTR + G | 8838.9842 | 18571.9684 | 4.6809 (3) | 21051.2299 | 17.3022 (6) |
GTR + I + G | 8839.0077 | 18574.0153 | 6.7278 (4) | 21058.8233 | 24.8955 (8) |
TPM1uf | 8845.426 | 18576.852 | 9.5645 (5) | 21033.9277 | 0.0 |
TPM1uf + I | 8845.4446 | 18578.8891 | 11.6016 (6) | 21041.5113 | 7.5836 (3) |
TPM1uf + G | 8846.7354 | 18581.4709 | 14.1834 (7) | 21044.093 | 10.1653 (4) |
TPM1uf + I + G | 8846.7697 | 18583.5395 | 16.252 (8) | 21051.7081 | 17.7804 (7) |
SYM | 8860.6478 | 18607.2955 | 40.008 (9) | 21064.3712 | 30.4435 (9) |
SYM + I | 8860.6826 | 18609.3652 | 42.0777 (10) | 21071.9874 | 38.0596 (12) |
34 Central Asian outbreak (CAO) specific SNPs with annotations.