{"paper_id":"4e0e9988-860b-457d-b1ea-d91bb0ab0ca2","body_text":"Subpopulations in clinical samples of M. tuberculosis can give\nrise to rifampicin resistance and shed light on how resistance is\nacquired\nViktoria M Brunner1 and Philip W Fowler*1,2,3\n1Nuffield Department of Medicine, University of Oxford, Oxford, UK\n2National Institute of Health Research Oxford Biomedical Research Centre, John Radcliffe\nHospital, Oxford, UK\n3Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial\nResistance, University of Oxford, UK\nAbstract\nThe increasing threat of antimicrobial resistance necessitates accurate and rapid diagnostics.\nWhole genome sequencing (WGS) has become a key tool for diagnosing Mycobacterium tubercu-\nlosis infections, but discrepancies between genotypic and phenotypic drug susceptibility testing can\nhinder effective treatment and surveillance. This study investigates the impact of including resistant\nsubpopulations and compensatory mutations in WGS-based rifampicin resistance prediction, based\non a dataset of 35,538 M. tuberculosis samples. The sensitivity and specificity of resistance classi-\nfication were evaluated with and without considering subpopulations and compensatory mutations.\nBy lowering the fraction of reads required to identify a resistance-associated variant in a sample from\n0.90 to 0.05, the sensitivity significantly increased from 94.3% to 96.4% with no significant impact on\nspecificity. This indicates that a substantial fraction of false negative calls in WGS-based rifampicin\nresistance prediction can be explained by masked resistant subpopulations. Allowing compensatory\nmutations to predict resistance further lowered the false negative rate. Finally, we found that samples\nwith resistant subpopulations were less likely to be compensated than homogeneous resistant sam-\nples, consistent with the recent evolution of resistance in the samples with subpopulations. Further\nanalysis of these samples revealed distinct clusters with differing amounts of within-sample diversity,\npointing towards different mechanisms of resistance acquisition, such as within-host evolution and\nsecondary infections.\nRunning title: Subpopulations give rise to rifampicin resistance in M. tuberculosis.\n*To whom correspondence should be addressed: philip.fowler@ndm.ox.ac.uk, @philipwfowler\n1\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nIntroduction\nM. tuberculosis infections are still responsible for about 1.25 million deaths per year 1 and resistance\nto first-line antibiotics continues to spread. 2 The first step towards successfully treating M. tuberculosis\ninfections is fast and reliable diagnostics, including drug susceptibility testing (DST). Phenotypic DST\ninvolves culturing the bacteria in the presence of different antibiotics 3 and is reliable for most drugs.\nHowever, since M. tuberculosis grows slowly, it is time-consuming despite efforts to reduce culture\ntime. 4 Genotypic DST instead infers from the genetic variants whether a sample is resistant and can be\nfaster than phenotypic DST.5 For this to work, the underlying mechanism must be genetic and the exact\ngenetic variants conferring resistance must also be known. If true, one can then look for the presence\nof specific genetic variation using rapid molecular tests such as Gene Xpert MTB/RIF 6, or use whole\ngenome sequencing (WGS) to scan the entire bacterial genome for resistance-associated variants (RA Vs)\nfor different antibiotics.7 The latter involves WGS of the patient-derived sample, followed by application\nof a high-confidence catalogue of RA Vs, enabling the sample to be classified as susceptible or resistant\nto a panel of antibiotics.8\nThe second edition of the WHO catalogue of mutations contains the most comprehensive list of RA Vs\nto date for predicting rifampicin resistance in M. tuberculosis and achieves 93.3% sensitivity and 96.9%\nspecificity on its training dataset compared to standard phenotypic DST results. Despite being one of\nthe best-performing drugs, the sensitivity for rifampicin remains below the 95% threshold proposed for\nantimicrobial susceptibility test devices by ISO.9 It is, of course, unrealistic to expect perfect agreement\nbetween the results of phenotypic and genotypic DST but any discrepancy creates problems not only for\nscientific efforts investigating disease transmission networks of M. tuberculosis,10 but also for diagnos-\ntics.11 If WGS-based DST is to complement or even replace phenotypic DST in some settings, it needs\nto achieve comparable performance. Hence there is a strong need to close the gap between phenotypic\nand genotypic DST results.\nHere, we show the performance of WGS-based DST for rifampicin can be trivially improved by per-\nmitting resistant subpopulations to contribute to the final classification. This was already shown to sig-\nnificantly improve the sensitivity of resistance prediction for fluoroquinolones inM. tuberculosis.12 The\nhypothesis is that resistant subpopulations are usually not called by bioinformatics pipelines. Hence\nthe genotypic approach does not lead to a prediction of resistance, yet the samples are phenotypically\nresistant. For many pathogens, resistant subpopulations are not expected since the standard laboratory\nprocess is to pick single colonies. Yet this is not true for M. tuberculosis where e.g. an aliquot taken for\nDNA extraction from a MGIT tube usually contains multiple ‘crumbs’ 13 and so could readily harbour\nresistant subpopulations, reflecting more accurately the patient’s infection.\nWe define the fraction of read support (FRS) as the proportion of reads at a genetic locus that supports a\n2\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nspecific genetic variant. Bioinformatics pipelines conventionally specify a minimum FRS for a genetic\nvariant to be called. The first edition of the WHOM. tuberculosismutation catalogue14 and the CRyPTIC\nconsortium 15 both used a conservative FRS threshold of 0.90 when calling genetic variants. This high\nthreshold prevents sequencing errors leading to spurious variant calls, but as sequencing technologies\nhave improved and error rates decreased, this now seems unduly conservative and ensures genetic sub-\npopulations are not detected. Such subpopulations are expected if an infection has only recently evolved\nresistance, perhaps in response to treatment, or if there has been a secondary resistant infection.\nA second way to improve the performance of catalogue-based predictions is to include more resistance-\nassociated variants. For now, catalogues only contain alleles that (are assumed to) directly cause resis-\ntance. We show that including genetic variants indirectly linked to resistance, such as compensatory\nmutations (CMs), can improve sensitivity without lowering specificity. CMs arise in response to a re-\nduction in fitness caused by rifampicin resistance mutations and hence whilst not causal are only found\nin rifampicin-resistant samples. 16–18 We have previously identified a list of high-confidence CMs using\ntheir strong association with the presence of resistance mutations. 19 We aim to test if including this list\nin the official WHO resistance catalogue20 improves the performance metrics of resistance prediction.\nIn addition to their potential to evade detection, samples with resistant subpopulations are an interesting\ntransition state in M. tuberculosis infections, which may arise through multiple different scenarios (Fig.\n1). The first possibility is within-host evolution of resistance, where the resistant subpopulation gradually\ntakes over a susceptible population during drug treatment (Fig. 1A). Secondly, the evolved drug-resistant\nsubpopulation may revert to a susceptible phenotype or gets out-competed after drug treatment is stopped\n(Fig. 1B); or thirdly the resistant subpopulation is acquired through a secondary infection (Fig. 1C). Dis-\ntinguishing these possibilities requires that we know whether the sample was taken before, during or\nafter drug treatment. Unfortunately, precise data on this is scarce and hence we can seldom distinguish\nbetween scenarios 1 and 2 (Fig. 1A, B). The second unknown variable is the source of resistance: within-\nhost evolution or a secondary infection. We aim to discern the source of resistance by quantifying the\namount of genetic diversity in the samples with resistant subpopulations in our dataset.\n3\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nA within-host evolution of \nresistance during treatment \nB reversion of resistance following cessation of treatment \nC secondary infection with a resistant strain\nresistant to antibiotic\nsusceptible \nto antibiotic\npatient with resistant \nsubpopulation\nsample taken\nFigure 1: Three ways a patient can acquire an infection with a resistant subpopulation. The sus-\nceptible subpopulation is shown in purple, the resistant subpopulation in red. Samples with resistant\nsubpopulations can originate from multiple different scenarios, among them (A) within-host-evolution,\n(B) reversion to susceptible phenotype following cessation of treatment, and (C) secondary infection.\n4\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nMaterials and Methods\nDataset sources\nIn addition to collecting >20,000 M. tuberculosis samples, each of which underwent whole genome\nsequencing (WGS) and drug susceptibility testing (DST) using one of two bespoke 96-well broth mi-\ncrodilution plates,15;21 the Comprehensive Resistance Prediction for Tuberculosis: an International Con-\nsortium (CRyPTIC) project also aggregated M. tuberculosis samples with WGS and/or DST data that\nhad been previously published. An early version of this dataset with heterogeneous DST methods was\nused to build the first edition of the WHO catalogue of tuberculosis resistance-associated variants.22\nWGS data processing\nA subset of 41,575 publicly available samples from the European Nucleotide Archive with short-read\npaired-end FASTQ files were uploaded to EIT Pathogena (https://www.eit-pathogena.com) and pro-\ncessed using version d5f9cd0 of the Mycobacterial pipeline. 23 The variant caller, Clockwork, calculates\nthe ‘fraction of read support’ (FRS) and, by default, only calls genetic variants where FRS≥ 0.90 with a\nfilter applied to the remainder.24 Here, we instructed the downstream tool, gnomonicus, to ignore these\nfilters and record all potential genetic variants so that we could detect subpopulations of bacteria due to\nmixed infections. 25 Most samples (37,594) had one or more of 101,020 mutations detected in the RNA\npolymerase (genes rpoA, rpoB, rpoC, rpoZ and sigA). This includes 3,260 so-called null mutations where\nthere are insufficient (< 3) reads to call a variant. To protect against spurious calls due to sequencing\nerrors, variants needed to be supported by at least three reads.\nExamining rpoB in more detail we find 31,347 samples containing 58,789 mutations with a FRS ≥ 0.90\nand 1,372 samples containing 1,940 mutations supported by an FRS< 0.90. A small number of the latter\nsamples have very large numbers of putative mutations, indicative of contamination or poor sequencing\nquality. In both cases the most common rpoB mutations were A1075A and S450L, as expected. Mu-\ntations were flagged as being associated with resistance to rifampicin according to the second edition\nof the WHO catalogue of mutations in M. tuberculosis. 20 We used a published list of high-confidence\ncompensatory mutations 19 to annotate which mutations are compensatory.\nPhenotypic DST data processing\nA total of 52,148 samples had one or more binary rifampicin drug susceptibility test results using a\nrange of methods; the most common were broth microdilution plate (24,172 samples) and mycobacterial\ngrowth indicator tube (23,682). If a sample had been tested using more than one method and all meth-\nods produced the same S/R result then, if present, the CRyPTIC result was retained since it has richer\ndata and has undergone additional quality control. If the methods disagreed on the outcome then the\n5\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nfirst resistant result was retained. This resulted in 48,031 samples with a single rifampicin DST result.\nMerging with the genetic data identified 35,538 samples which have both whole genome sequencing data\nand a single rifampicin DST binary result: this dataset forms the basis of our subsequent analysis. The\nsensitivity and specificity of the genotypic resistance call were calculated with respect to the phenotypic\ndrug susceptibility result. Significance was tested using a proportions z-test.\nReproducibility statement\nAll analysis, figures and tables in this paper can be reproduced using a GitHub repository which contains\nall data tables and code.26\n6\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nResults\nClassifying subpopulations that contain rifampicin RA Vs as resistant improves sensitivity\nIn our dataset of 35,538 samples, we will describe samples with a rifampicin resistance-associated variant\n(RA V) with a fraction of read support (FRS) ≥ 0.90 as homogeneous, and samples containing RA Vs at\nan FRS < 0.90 as heterogeneous. Mutations with an FRS < 0.90 read support will be described as\nminor. Out of the 10,568 samples containing a rifampicin RA V at any level of read support, 10,287 are\nhomogeneous, 261 are heterogeneous and 20 are mixed, i.e. contain at least two rifampicin RA Vs, one\nsupported by FRS ≥ 0.90 and another with FRS < 0.90 (Fig. 2A). Examining the distribution of minor\nrifampicin RA Vs, we observe that RA Vs are seen at all levels of read support, possibly increasing at\nhigher values of FRS (Fig. 2B). This indicates that the full range of values should be considered when\nassessing the impact of the FRS threshold on the performance of resistance prediction.\nThe sensitivity increases from 94.3% to 96.4% as the minimum FRS required to call a RA V is decreased\nfrom 0.90 to 0.05; over the same range the specificity falls slightly from 98.1% to 97.9% (Fig. 2C, Table\n1). The increase in sensitivity when the FRS is dropped from 0.90 to 0.05 is statistically significant,\nwhilst the decrease in specificity is not (Fig. 2D). Over the same range the negative predictive value\n(NPV) also significantly increases whilst there is no significant change in the positive predictive value\n(PPV , Fig. S1, Table 1). Taken together, this indicates that applying a conservative FRS threshold masks\nresistant subpopulations, resulting in falsely predicting some samples as susceptible to rifampicin.\nCompensatory mutations can identify rifampicin resistance by association with high specificity\nEven when a low FRS threshold is set (e.g. 0.05) we still observe false negatives in our genotypic\nresistance prediction and this is unlikely to improve much upon further lowering of the FRS threshold\n(Fig. 2C). Another cause for false negative calls is low coverage or sequencing errors in genomic regions\nassociated with resistance, hence it would be useful if there was redundancy when predicting resistance\nusing genetic features. We have previously identified 51 compensatory mutations (CMs) 19, which are,\nby definition, only present in resistant samples. Hence by allowing samples to be predicted as rifampicin-\nresistant if they contain either a RA V and/or a CM one might expect to reduce the false negative rate. To\ntest this, we appended these 51 CMs to the WHO catalogue of RA Vs for rifampicin.\nIncluding CMs reduced the number of false negative calls from 591 to 568 when applying an FRS\nthreshold of 0.90 to call variants (Table 1), corresponding to a 3.9% decrease. The resulting improvement\nin sensitivity is not significant, probably due to the small number of affected samples compared to the\nsize of the dataset. If we repeat the experiment setting the FRS threshold to 0.05, the number of samples\nfalsely predicted to be resistant increases from 529 to 663. Most (99%) of these erroneous calls are due to\nsamples which, despite containing the putative CMrpoC F452L, are not actually resistant. If we exclude\n7\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nsensitivity\nspecificity\n%\nFRS ≥ 0.90\nFRS ≥ 0.05\n10,287 (97.3%) samples have ≥1 RIF RAV at FRS ≥ 0.9\n261 (2.5%) samples have  ≥1 RIF RAV at FRS < 0.9 \n20 (0.2%) samples with ≥1 RIF RAV at FRS < 0.9 and ≥1 RIF RAV at FRS ≥ 0.9  \n0.2 0.4 0.6 0.8 1.0\nminimum fraction of reads (FRS) required to call a RAV\nC\nD\nA B\nFigure 2: Lowering the fraction of read support (FRS) threshold for calling rifampicin (RIF)\nresistance-associated variants (RA Vs) increases sensitivity with no significant effect on specificity.\n(A) The majority of samples containing a RA V are homogeneous, meaning they show a RIF RA V at≥\n0.90 FRS. The heterogeneous samples make up for 2.5% of samples, but we rarely see samples with both\na RA V at≥ 0.90 FRS and a RA V with lower read support. (B) The distribution of FRS for the RA Vs\nin the 258 heterogeneous samples with 0 .05 ≤ FRS < 0.9. (C) Decreasing the FRS threshold required\nto support a variant call that is a known RA V increases sensitivity of the prediction with little effect on\nspecificity. The slopes of a linear regression for sensitivity and specificity are -0.02 and 0.002, respec-\ntively. (D) Sensitivity is significantly improved if the FRS threshold is lowered from 0.90 to 0.05 (z-test,\np-value = 8e-13). There is no significant change in specificity.\n8\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nRA Vs CMs rpoC F452L FRS min TP FP TN FN sen spec PPV NPV\nD D 0.90 9819 488 24640 591 94.3% 98.1% 95.3% 97.7%\nD D 0.05 10036 529 24599 374 96.4% 97.9% 95.0% 98.5%\nD D D 0.90 9842 490 24638 568 94.5% 98.0% 95.3% 97.7%\nD D D 0.05 10047 663 24465 363 96.5% 97.4% 93.8% 98.5%\nD D 0.05 10047 531 24597 363 96.5% 97.9% 95.0% 98.5%\nD D 0.05 4452 225 24903 5958 42.8% 99.1% 95.2% 80.7%\nTable 1: Fraction of read support (FRS) and corresponding contingency table values and performance\nmetrics for different scenarios of the catalogue-based predictions for rifampicin resistance. Scenarios are\nshown for different FRS thresholds, and are either using RA Vs and/or CMs for prediction or not. In one\nscenario, the CM rpoC F452L was removed from the analysis. ”sen” shows sensitivity and ”spec” the\nspecificity of the resistance prediction.\nthis CM post hoc, the specificity does not decrease when allowing CMs to predict resistance at low FRS\n(Table 1).\nOne could, of course, use exclusively CMs to predict rifampicin resistance. However, as one might\nexpect, only a moderate sensitivity is achieved (42.8%, Table 1), since merely a fraction of resistant sam-\nples exhibit compensation. Interestingly, the specificity of predicting resistance is higher (99.1%) than\nwhen simply applying the WHOv2 catalogue of mutations (Table 1). This suggests that one could use\nCMs to predict rifampicin resistance in the absence of a RA V , which may be particularly useful when\ndealing with samples with low coverage. The combined effect of lowering the FRS threshold to 0.05 and\npermitting CMs to predict resistance is a reduction of false negative calls by 38.6%. Hence more than\na third of the resistant samples incorrectly classified as susceptible are hereby explained and corrected\n(Table 1), a handy improvement.\nHeterogeneous resistant samples are less likely to show compensation\nWe hypothesised that resistant heterogeneous samples are less likely to have had the time to evolve a\ncompensatory mutation. The proportion of heterogeneous samples with a compensatory mutation is\nindeed significantly less than the fraction of homogeneous samples which are compensated (43.5% v\n34.2%, Fishers exact p-value = 3.63e-4, Fig. 3A).\nIn the compensated heterogeneous samples, the CM FRS and RA V FRS values are correlated (Fig. 3B,\np = 1.69e-22). This is consistent with these samples being composed of a susceptible strain and a ri-\nfampicin resistant strain that has acquired a compensatory mutation. The reduced rate of compensation\nin heterogeneous samples suggests that the resistant subpopulations have, on average, evolved resistance\nmore recently than homogeneous resistant samples.\n9\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nA B\nFigure 3: Heterogeneous resistant samples are less likely to also have a compensatory mutation\n(CM) than homogeneous resistant samples. (A) Percentage of samples showing CMs at any FRS in\nhomogeneous vs heterogeneous resistant samples (Fishers exact p-value = 3.63e-4). (B) For the hetero-\ngeneous samples that are compensated, this plot shows the significant correlation between the FRS of the\nrespective resistance and compensatory mutation (p = 1.69e-22). The grey line indicates the hypothetical\nline of perfect correlation.\nOver a third of heterogeneous resistant samples are the result of secondary infections\nWe can gain some insight into the source of resistance by examining the genetic diversity in the hetero-\ngeneous samples. If the resistant subpopulation arose by within-host evolution (Fig. 1A), one would\nexpect to see relatively few, if any, other minor mutations in the sample, given the low mutation rate\nof M. tuberculosis. A similar logic applies when a sample is taken midway through reversion of resis-\ntance following cessation of treatment (Fig. 1B); again relatively few other minor mutations would be\nexpected. If the patient has been infected more than once (Fig. 1C), the sample will probably contain\ndifferent strains and therefore we would expect a much greater number of minor mutations. This will not\nnecessarily always be true: if the co-infecting strain is part of the same outbreak, the number of minor\nmutations will again be small. Only when the number of minor mutations is high, e.g. if the sample\ncontains multiple (sub)lineages, can we draw a definite conclusion since in this case secondary infection\nis the only viable scenario.\nTo investigate this, we measured the number of minor alleles in all 248 heterogeneous resistant samples\nwith a singular minor RA V . The latter condition is required to ensure that only a single resistant subpop-\nulation is present. The resulting distribution is very broad; some samples have no or relatively few minor\nmutations, whilst others have several thousand (Fig. 4A).\n10\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nA B\nNumber of mutations with FRS < 0.9 Number of mutations with FRS < 0.9\nFigure 4: Quantifying genetic diversity within heterogeneous resistant samples using the amount\nof minor mutations detected. (A) Genetic diversity within the heterogeneous samples with resistant\nsubpopulations, as measured by the amount of minor mutations per sample. Minor mutations are all\nvariants in these samples which display an FRS below 0.90. (B) Same as in A, but the data is divided\ninto three different batches based on the sample mixture type assigned by mykrobe (v0.12.1): A single\nlineage, multiple sublineages or multiple lineages per sample.\nMapping the mixture type identified by mykrobe (single, multiple sublineages, or multiple lineages per\nsample) onto the distribution neatly segregates the data into three partially overlapping subgroups (Fig.\n4B). The first is the 163 samples which, according to mykrobe, have one and only one lineage (la-\nbelled ‘single’ in Fig. 4B). The remaining 85 samples (34 %) are assessed as either containing multiple\nsublineages belonging to the same lineage, or different lineages. These are labelled ‘sublineages‘ and\n‘lineages‘, respectively, on Fig. 4B. The only plausible explanation for the latter two cases is if the pa-\ntient picked up a secondary infection and we therefore infer that at least 34 % of heterogeneous resistant\nsamples in our dataset arose through secondary infection. Interestingly, there is also a significant associ-\nation between a heterogeneous resistant sample having a compensatory mutation and having more than\none (sub)lineage (Fishers exact p-value = 2.38e-04).\nWithin the heterogeneous samples which contain multiple M. tuberculosis lineages, we can clearly see\ntwo different clusters (Fig. 4B). When evaluating the mix of lineages contained in these samples, we\nfound that samples with a greater number of minor mutations are likely to contain Lineage 1, one of the\nancient strains of M. tuberculosis. The modern lineages (Lineages 2-4) are more similar to the reference\nstrain (H37Rv, Lineage 4) from which mutations are defined. Accordingly, the cluster with only modern\nlineages shows on average 358 minor mutations per sample fewer than the samples containing Lineage\n1 (Fig. S2).\n11\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nDiscussion\nWe have demonstrated that it is important to consider within-host diversity when predicting drug re-\nsistance in M. tuberculosis using whole genome sequencing. Lowering the minimum fraction of read\nsupport (FRS) required to call a resistance-associated variant (RA V) from 0.90 to 0.05 significantly\nimproves the sensitivity of genotypic resistance classification by +2.1% to 96.4%, with no detectable\neffect on specificity. A higher sensitivity improves any treatment decisions and leads to more effective\nsurveillance. Importantly, it also takes rifampicin prediction on this dataset above the required minimum\nthreshold of 95% in both sensitivity and specificity to pass the ISO standard for antimicrobial suscepti-\nbility test devices. 9 It would therefore seem sensible for genetic workflows that process M. tuberculosis\nsamples to identify and call rifampicin RA Vs present in a minority, as well as a majority, of reads in the\nrpoB gene.\nAllowing the presence of high-confidence compensatory mutations (CMs) in the RNA polymerase to\nalso identify rifampicin resistance by association would also seem a trivial way to boost sensitivity. The\nimprovement in sensitivity on this dataset was not significant, probably due to the moderate read depth\nensuring there were relatively few resistant samples with a compensatory mutation where the resistance\nmutations could not be resolved but the CMs could. In scenarios with much lower read depth (for\nexample in multiplexed samples) allowing compensatory mutations to predict rifampicin resistance could\nprovide a necessary boost to the specificity. In future work we will investigate the read depths at which\nit becomes beneficial to consider CMs for calling rifampicin resistance by randomly down-sampling\nsequencing reads.\nAs noted above, it is vital that any list of compensatory mutations is accurate if they are to be used to\ninfer resistance. 19 The mutation rpoC F452L appears to not be resistance-associated when present at\nlower FRS, which could either be an error or point to an inverse causal relationship between this CM and\nthe presence of RA Vs. This raises the question of whether F452L should be considered a CM or rather\nprovides an advantageous genetic background for the acquisition of rifampicin resistance, by raising\nfitness prior to resistance emergence. Whilst speculation, this is supported by the observation that rpoC\nF452L was shown to stabilise the open promoter complex and increase the elongation rate of the M.\ntuberculosis RNA polymerase not only in the resistant rpoB S450L mutant, but also in the wild type.27\nOverall, reducing the minimum FRS threshold to 0.05 and permitting the presence of CMs to predict\nrifampicin resistance raises sensitivity to 96.5%. The remaining discordance can likely be explained by\nseveral factors: there may be additional, unknown rare resistance mutations in the RNA polymerase that\nhave not been classified as such by the second edition of the WHO resistance catalogue. Tackling this\nproblem is non-trivial but could involve the use of machine learning models trained to predict the effect\nof individual rpoB mutations.28 There are also almost certainly errors in the laboratory processes, such\n12\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nas mislabelling or measurement error, that we can minimise but not eradicate.\nWe can also gain insight into the timing and origin of rifampicin resistance by examining those samples\nwith resistant subpopulations. The significantly lower prevalence of compensation in heterogeneous\nsamples suggests that resistance emerged more recently, on average, than in homogeneous samples.\nFurthermore the number of other putative minor mutations in the heterogeneous samples hints at the\nprobable sources of resistance. Our data are consistent with secondary infection, as opposed to within-\nhost evolution, being the root cause of rifampicin resistance in at least 34% of the heterogeneous samples.\nThe striking level of genetic diversity detected within these samples is only achievable if the patient was\nsubject to a secondary infection, which likely introduced the resistant subpopulation. The significant\nassociation of CMs with samples containing multiple (sub)lineages indicates these samples are also more\nlikely to be compensated. This is consistent with the secondary infection scenario, since the infecting\nstrain could already have been compensated before being transmitted.\nStepping back, is not surprising that our dataset of M. tuberculosis samples contains genetically hetero-\ngeneous samples due to how clinical samples are grown and colonies extracted. For example, aliquots\ntaken from positive MGIT tubes will contain multiple ‘crumbs‘ and therefore contain multiple colonies.\nThis is in sharp contrast to laboratory practices for other pathogens where single colony picks from solid\nmedia are commonplace. One should therefore, perhaps, consider Mycobacterial genetic sequencing as\ninherently metagenomic.\nIn conclusion, this study has shown how to improve the sensitivity of WGS-based rifampicin resis-\ntance prediction by including subpopulations, compensatory mutations, and population diversity. It also\nprovides useful insights into how resistance emerges clinically. Overall, these findings will facilitate\nimproved diagnostic strategies and more effective management of drug-resistant tuberculosis.\n13\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nAcknowledgements\nWe would like to thank Tim Peto, Nicole Stoesser and David Eyre for helpful suggestions.\nFunding\nThis work was supported by the National Institute for Health Research (NIHR) Health Protection Re-\nsearch Unit (HPRU) in Healthcare Associated Infections and Antimicrobial Resistance at Oxford Uni-\nversity in partnership with UK Health Security Agency (NIHR200915), the National Institute for Health\nResearch (NIHR) Oxford Biomedical Research Centre (BRC) and the Ellison Institute of Technology,\nOxford Ltd. VMB is supported by the Biotechnology and Biological Sciences Research Council (grant\nnumber BB/T008784/1). For the purpose of open access, the author has applied a CC BY public copy-\nright licence to any Author Accepted Manuscript version arising from this submission. The findings and\nconclusions in this report are solely the responsibility of the authors and do not necessarily represent the\nofficial views of the NHS, the NIHR, UKHSA, the Department of Health and Social Care or the Ellison\nInstitute of Technology, Oxford Ltd.\nTransparency declaration\nPWF receives consultancy fees from the Ellison Institute of Technology, Oxford Ltd.\n14\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nReferences\n[1] World Health Organisation (2024) Global tuberculosis report. ISBN: 978-92-4-010153-1.\n[2] S. Bagcchi (2023) WHO’s Global Tuberculosis Report 2022,The Lancet Microbe4:e20. Publisher:\nElsevier.\n[3] World Health Organisation (2018) Technical manual for drug susceptibility testing of medicines\nused in the treatment of tuberculosis. ISBN: 978-92-4-151484-2.\n[4] R. Ghodbane, D. Raoult, and M. Drancourt (2014) Dramatic reduction of culture time of Mycobac-\nterium tuberculosis, Scientific Reports 4:4236.\n[5] L. J. Pankhurst, C. del Ojo Elias, A. A. V otintseva, et al. (2016) Rapid, comprehensive, and afford-\nable mycobacterial diagnosis with whole-genome sequencing: a prospective study, Lancet Resp\nMed 4:49–58.\n[6] D. Hillemann, S. R ¨usch-Gerdes, and E. Richter (2007) Evaluation of the GenoType MTBDRplus\nassay for rifampin and isoniazid susceptibility testing of Mycobacterium tuberculosis strains and\nclinical specimens, Journal of Clinical Microbiology 45:2635–2640.\n[7] D. Papaventsis, N. Casali, I. Kontsevaya, et al. (2017) Whole genome sequencing of Mycobac-\nterium tuberculosis for detection of drug resistance: a systematic review, Clinical Microbiology\nand Infection 23:61–68.\n[8] C. J. Meehan, G. A. Goig, T. A. Kohl, et al. (2019) Whole genome sequencing of Mycobacterium\ntuberculosis: current standards and open issues, Nature Reviews Microbiology 17:533–545.\n[9] International Organization for Standardization (2021). ISO 20776-2: Clinical laboratory testing\nand in vitro diagnostic test systems - Susceptibility testing of infectious agents and evaluation of\nperformance of antimicrobial susceptibility test devices.\n[10] C. J. Worby, M. Lipsitch, and W. P. Hanage (2014) Within-Host Bacterial Diversity Hinders Accu-\nrate Reconstruction of Transmission Networks from Genomic Distance Data,PLOS Computational\nBiology 10:e1003549.\n[11] K. Faksri, O. Kaewprasert, R. T.-H. Ong, et al. (2019) Comparisons of whole-genome sequencing\nand phenotypic drug susceptibility testing for Mycobacterium tuberculosis causing MDR-TB and\n15\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\nXDR-TB in Thailand, International Journal of Antimicrobial Agents 54:109–116.\n[12] A. E. Brankin and P. W. Fowler (2023) Inclusion of minor alleles improves catalogue-based predic-\ntion of fluoroquinolone resistance in Mycobacterium tuberculosis, JAC-Antimicrobial Resistance\n5:dlad039.\n[13] A. A. V otintseva, L. J. Pankhurst, L. W. Anson, et al. (2015) Mycobacterial DNA extraction for\nwhole-genome sequencing from early positive liquid (MGIT) cultures, J Clin Microbiol 53:1137–\n1143.\n[14] World Health Organization (2021) Catalogue of mutations in Mycobacterium tuberculosis complex\nand their association with drug resistance, 1st edition. ISBN: 978-92-4-002817-3.\n[15] The CRyPTIC Consortium (2022) A data compendium associating the genomes of 12,289 My-\ncobacterium tuberculosis isolates with quantitative resistance phenotypes to 13 antibiotics, PLOS\nBiology 20:e3001721.\n[16] G. Brandis, M. Wrande, L. Liljas, et al. (2012) Fitness-compensatory mutations in rifampicin-\nresistant RNA polymerase, Molecular Microbiology 85:142–151.\n[17] A. K. Alame Emane, X. Guo, H. E. Takiff, et al. (2021) Drug resistance, fitness and compensatory\nmutations in Mycobacterium tuberculosis, Tuberculosis 129:102091.\n[18] T. Song, Y . Park, I. C. Shamputa, et al. (2014) Fitness costs of rifampicin resistance in mycobac-\nterium tuberculosis are amplified under conditions of nutrient starvation and compensated by mu-\ntation in the β ′ subunit of RNA polymerase, Molecular Microbiology 91:1106–1119.\n[19] V . M. Brunner and P. W. Fowler (2024) Compensatory mutations are associated with increased\nin vitro growth in resistant clinical samples of Mycobacterium tuberculosis, Microbial Genomics\n10:001187.\n[20] World Health Organization (2023) Catalogue of mutations in Mycobacterium tuberculosis complex\nand their association with drug resistance, 2nd edition. ISBN: 978-92-4-008241-0.\n[21] The CRyPTIC Consortium (2022) Epidemiological cutoff values for a 96-well broth microdilution\nplate for high-throughput research antibiotic susceptibility testing of M. tuberculosis, Eur Respir J\n60:2200239.\n16\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint \n\n[22] T. M. Walker, P. Miotto, C. U. K ¨oser, et al. (2022) The 2021 WHO catalogue of Mycobacterium\ntuberculosis complex mutations associated with drug resistance: a genotypic analysis, The Lancet\nMicrobe 3:e265–e273.\n[23] J. Westhead, C. S. Baker, M. Brouard, et al. (2024). Enhancement and validation of the antibiotic\nresistance prediction performance of a cloud-based genetics processing platform for Mycobacteria.\nbioRxiv preprint. doi:10.1101/2024.11.08.622466.\n[24] M. Hunt, B. Letcher, K. M. Malone, et al. (2022) Minos: variant adjudication and joint genotyping\nof cohorts of bacterial genomes, Genome Biology 23:147.\n[25] J. Westhead and P. W. Fowler (2023). gnomonicus. https://github.com/oxfordmmm/gnomonicus\nHTT you.\n[26] V . Brunner and P. W. Fowler. Accompanying code to reproduce analysis and figures.\nhttps://github.com/fowler-lab/tuberculosis-rnap-subpopulations.\n[27] M. A. Stefan, F. S. Ugur, and G. A. Garcia (2018) Source of the Fitness Defect in Rifamycin-\nResistant Mycobacterium tuberculosis RNA Polymerase and the Mechanism of Compensation by\nMutations in the β ′ Subunit, Antimicrobial Agents and Chemotherapy 62:e00164.\n[28] C. I. Lynch, D. Adlard, and P. W. Fowler (2025) Predicting rifampicin resistance in M. tuberculosis\nusing machine learning informed by protein structural and chemical features, ERJ Open Research\ndoi:10.1183/23120541.00952-2024.\n17\n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted April 9, 2025. ; https://doi.org/10.1101/2025.04.09.647945doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}