A modified decision tree approach to improve the prediction and mutation discovery for drug resistance in Mycobacterium tuberculosis.

BACKGROUND: Drug resistant Mycobacterium tuberculosis is complicating the effective treatment and control of tuberculosis disease (TB). With the adoption of whole genome sequencing as a diagnostic tool, machine learning approaches are being employed to predict M. tuberculosis resistance and identify underlying genetic mutations. However, machine learning approaches can overfit and fail to identify causal mutations if they are applied out of the box and not adapted to the disease-specific context. We introduce a machine learning approach that is customized to the TB setting, which extracts a library of genomic variants re-occurring across individual studies to improve genotypic profiling. RESULTS: We developed a customized decision tree approach, called Treesist-TB, that performs TB drug resistance prediction by extracting and evaluating genomic variants across multiple studies. The application of Treesist-TB to rifampicin (RIF), isoniazid (INH) and ethambutol (EMB) drugs, for which resistance mutations are known, demonstrated a level of predictive accuracy similar to the widely used TB-Profiler tool (Treesist-TB vs. TB-Profiler tool: RIF 97.5% vs. 97.6%; INH 96.8% vs. 96.5%; EMB 96.8% vs. 95.8%). Application of Treesist-TB to less understood second-line drugs of interest, ethionamide (ETH), cycloserine (CYS) and para-aminosalisylic acid (PAS), led to the identification of new variants (52, 6 and 11, respectively), with a high number absent from the TB-Profiler library (45, 4, and 6, respectively). Thereby, Treesist-TB had improved predictive sensitivity (Treesist-TB vs. TB-Profiler tool: PAS 64.3% vs. 38.8%; CYS 45.3% vs. 30.7%; ETH 72.1% vs. 71.1%). CONCLUSION: Our work reinforces the utility of machine learning for drug resistance prediction, while highlighting the need to customize approaches to the disease-specific context. Through applying a modified decision learning approach (Treesist-TB) across a range of anti-TB drugs, we identified plausible resistance-encoding genomic variants with high predictive ability, whilst potentially overcoming the overfitting challenges that can affect standard machine learning applications.

A Bayesian space-time model for clustering areal units based on their disease trends.

Population-level disease risk across a set of non-overlapping areal units varies in space and time, and a large research literature has developed methodology for identifying clusters of areal units exhibiting elevated risks. However, almost no research has extended the clustering paradigm to identify groups of areal units exhibiting similar temporal disease trends. We present a novel Bayesian hierarchical mixture model for achieving this goal, with inference based on a Metropolis-coupled Markov chain Monte Carlo ((MC)$^3$) algorithm. The effectiveness of the (MC)$^3$ algorithm compared to a standard Markov chain Monte Carlo implementation is demonstrated in a simulation study, and the methodology is motivated by two important case studies in the United Kingdom. The first concerns the impact on measles susceptibility of the discredited paper linking the measles, mumps, and rubella vaccination to an increased risk of Autism and investigates whether all areas in the Scotland were equally affected. The second concerns respiratory hospitalizations and investigates over a 10 year period which parts of Glasgow have shown increased, decreased, and no change in risk.

Attractor-like Dynamics in Belief Updating in Schizophrenia.

Subjects with a diagnosis of schizophrenia (Scz) overweight unexpected evidence in probabilistic inference: such evidence becomes "aberrantly salient." A neurobiological explanation for this effect is that diminished synaptic gain (e.g., hypofunction of cortical NMDARs) in Scz destabilizes quasi-stable neuronal network states (or "attractors"). This attractor instability account predicts that (1) Scz would overweight unexpected evidence but underweight consistent evidence, (2) belief updating would be more vulnerable to stochastic fluctuations in neural activity, and (3) these effects would correlate. Hierarchical Bayesian belief updating models were tested in two independent datasets (n = 80 male and n = 167 female) comprising human subjects with Scz, and both clinical and nonclinical controls (some tested when unwell and on recovery) performing the "probability estimates" version of the beads task (a probabilistic inference task). Models with a standard learning rate, or including a parameter increasing updating to "disconfirmatory evidence," or a parameter encoding belief instability were formally compared. The "belief instability" model (based on the principles of attractor dynamics) had most evidence in all groups in both datasets. Two of four parameters differed between Scz and nonclinical controls in each dataset: belief instability and response stochasticity. These parameters correlated in both datasets. Furthermore, the clinical controls showed similar parameter distributions to Scz when unwell, but were no different from controls once recovered. These findings are consistent with the hypothesis that attractor network instability contributes to belief updating abnormalities in Scz, and suggest that similar changes may exist during acute illness in other psychiatric conditions.SIGNIFICANCE STATEMENT Subjects with a diagnosis of schizophrenia (Scz) make large adjustments to their beliefs following unexpected evidence, but also smaller adjustments than controls following consistent evidence. This has previously been construed as a bias toward "disconfirmatory" information, but a more mechanistic explanation may be that in Scz, neural firing patterns ("attractor states") are less stable and hence easily altered in response to both new evidence and stochastic neural firing. We model belief updating in Scz and controls in two independent datasets using a hierarchical Bayesian model, and show that all subjects are best fit by a model containing a belief instability parameter. Both this and a response stochasticity parameter are consistently altered in Scz, as the unstable attractor hypothesis predicts.

Large-scale genomic analysis of Mycobacterium tuberculosis reveals extent of target and compensatory mutations linked to multi-drug resistant tuberculosis.

Resistance to isoniazid (INH) and rifampicin (RIF) first-line drugs in Mycobacterium tuberculosis (Mtb), together called multi-drug resistance, threatens tuberculosis control. Resistance mutations in katG (for INH) and rpoB (RIF) genes often come with fitness costs. To overcome these costs, Mtb compensatory mutations have arisen in rpoC/rpoA (RIF) and ahpC (INH) loci. By leveraging the presence of known compensatory mutations, we aimed to detect novel resistance mutations occurring in INH and RIF target genes. Across ~ 32 k Mtb isolates with whole genome sequencing (WGS) data, there were 6262 (35.7%) with INH and 5435 (30.7%) with RIF phenotypic resistance. Known mutations in katG and rpoB explained ~ 99% of resistance. However, 188 (0.6%) isolates had ahpC compensatory mutations with no known resistance mutations in katG, leading to the identification of 31 putative resistance mutations in katG, each observed in at least 3 isolates. These putative katG mutations can co-occur with other INH variants (e.g., katG-Ser315Thr, fabG1 mutations). For RIF, there were no isolates with rpoC/rpoA compensatory mutations and unknown resistance mutations. Overall, using WGS data we identified putative resistance markers for INH that could be used for genotypic drug-resistance profiling. Establishing the complete repertoire of Mtb resistance mutations will assist the clinical management of tuberculosis.

Comparison of in silico predicted Mycobacterium tuberculosis spoligotypes and lineages from whole genome sequencing data.

Bacterial strain-types in the Mycobacterium tuberculosis complex underlie tuberculosis disease, and have been associated with drug resistance, transmissibility, virulence, and host-pathogen interactions. Spoligotyping was developed as a molecular genotyping technique used to determine strain-types, though recent advances in whole genome sequencing (WGS) technology have led to their characterization using SNP-based sub-lineage nomenclature. Notwithstanding, spoligotyping remains an important tool and there is a need to study the congruence between spoligotyping-based and SNP-based sub-lineage assignation. To achieve this, an in silico spoligotype prediction method ("Spolpred2") was developed and integrated into TB-Profiler. Lineage and spoligotype predictions were generated for > 28 k isolates and the overlap between strain-types was characterized. Major spoligotype families detected were Beijing (25.6%), T (18.6%), LAM (13.1%), CAS (9.4%), and EAI (8.3%), and these broadly followed known geographic distributions. Most spoligotypes were perfectly correlated with the main MTBC lineages (L1-L7, plus animal). Conversely, at lower levels of the sub-lineage system, the relationship breaks down, with only 65% of spoligotypes being perfectly associated with a sub-lineage at the second or subsequent levels of the hierarchy. Our work supports the use of spoligotyping (membrane or WGS-based) for low-resolution surveillance, and WGS or SNP-based systems for higher-resolution studies.

Genetic mutations underlying isoniazid-resistant Mycobacterium tuberculosis in Khyber Pakhtunkhwa, Pakistan.

Tuberculosis, caused by Mycobacterium tuberculosis, is a major public health issue in Pakistan. Isoniazid is a first-line pro-drug that requires activation through an enzyme called catalase peroxidase, but is subject to widespread resistance, driven by mutations in katG and inhA genes and other loci with compensatory effects (e.g., ahpC). Here, we used whole genome sequencing data from 51 M. tuberculosis isolates collected from Khyber Pakhtunkhwa province (years 2016-2019; all isoniazid phenotypically resistant) to investigate the genetic diversity of mutations in isoniazid candidate genes. The most common mutations underlying resistance were katG S315T (37/51), fabG1 -15C>T (13/51; inhA promoter), and inhA -154G>A (7/51). Other less common mutations (n < 5) were also identified in katG (R128Q, V1A, W505*, A109T, D311G) and candidate compensatory genes ahpC (-54C>T, -51G>A) and oxyS (M249T). Using DynaMut2 software, the mutants exhibited various degrees of stability and flexibility on protein structures, with some katG mutations leading to a decrease in KatG protein flexibility. Overall, the characterisation of circulating isoniazid resistant-linked mutations will assist in drug resistant TB management and control activities in a highly endemic area of Pakistan.

Spatio-temporal disease risk estimation using clustering-based adjacency modelling.

Conditional autoregressive models are typically used to capture the spatial autocorrelation present in areal unit disease count data when estimating the spatial pattern in disease risk. This correlation is represented by a binary neighbourhood matrix based on a border sharing specification, which enforces spatial correlation between geographically neighbouring areas. However, enforcing such correlation will mask any discontinuities in the disease risk surface, thus impeding the detection of clusters of areas that exhibit higher or lower risks compared to their neighbours. Here we propose novel methodology to account for these clusters and discontinuities in disease risk via a two-stage modelling approach, which either forces the clusters/discontinuities to be the same for all time periods or allows them to evolve dynamically over time. Stage one constructs a set of candidate neighbourhood matrices to represent a range of possible cluster/discontinuity structures in the data, and stage two estimates an appropriate structure(s) by treating the neighbourhood matrix as an additional parameter to estimate within a Bayesian spatio-temporal disease mapping model. The effectiveness of our novel methodology is evidenced by simulation, before being applied to a new study of respiratory disease risk in Greater Glasgow, Scotland from 2011 to 2017.

Characterisation of drug-resistant Mycobacterium tuberculosis mutations and transmission in Pakistan.

Tuberculosis, caused by Mycobacterium tuberculosis, is a high-burden disease in Pakistan, with multi-drug (MDR) and extensive-drug (XDR) resistance, complicating infection control. Whole genome sequencing (WGS) of M. tuberculosis is being used to infer lineages (strain-types), drug resistance mutations, and transmission patterns-all informing infection control and clinical decision making. Here we analyse WGS data on 535 M. tuberculosis isolates sourced across Pakistan between years 2003 and 2020, to understand the circulating strain-types and mutations related to 12 anti-TB drugs, as well as identify transmission clusters. Most isolates belonged to lineage 3 (n = 397; 74.2%) strain-types, and were MDR (n = 328; 61.3%) and (pre-)XDR (n = 113; 21.1%). By inferring close genomic relatedness between isolates (< 10-SNPs difference), there was evidence of M. tuberculosis transmission, with 55 clusters formed consisting of a total of 169 isolates. Three clusters consist of M. tuberculosis that are similar to isolates found outside of Pakistan. A genome-wide association analysis comparing 'transmitted' and 'non-transmitted' isolate groups, revealed the nusG gene as most significantly associated with a potential transmissible phenotype (P = 5.8 × 10-10). Overall, our study provides important insights into M. tuberculosis genetic diversity and transmission in Pakistan, including providing information on circulating drug resistance mutations for monitoring activities and clinical decision making.

Genetic diversity of candidate loci linked to Mycobacterium tuberculosis resistance to bedaquiline, delamanid and pretomanid.

Tuberculosis (TB), caused by Mycobacterium tuberculosis, is one of the deadliest infectious diseases worldwide. Multidrug and extensively drug-resistant strains are making disease control difficult, and exhausting treatment options. New anti-TB drugs bedaquiline (BDQ), delamanid (DLM) and pretomanid (PTM) have been approved for the treatment of multi-drug resistant TB, but there is increasing resistance to them. Nine genetic loci strongly linked to resistance have been identified (mmpR5, atpE, and pepQ for BDQ; ddn, fgd1, fbiA, fbiB, fbiC, and fbiD for DLM/PTM). Here we investigated the genetic diversity of these loci across >33,000 M. tuberculosis isolates. In addition, epistatic mutations in mmpL5-mmpS5 as well as variants in ndh, implicated for DLM/PTM resistance in M. smegmatis, were explored. Our analysis revealed 1,227 variants across the nine genes, with the majority (78%) present in isolates collected prior to the roll-out of BDQ and DLM/PTM. We identified phylogenetically-related mutations, which are unlikely to be resistance associated, but also high-impact variants such as frameshifts (e.g. in mmpR5, ddn) with likely functional effects, as well as non-synonymous mutations predominantly in MDR-/XDR-TB strains with predicted protein destabilising effects. Overall, our work provides a comprehensive mutational catalogue for BDQ and DLM/PTM associated genes, which will assist with establishing associations with phenotypic resistance; thereby, improving the understanding of the causative mechanisms of resistance for these drugs, leading to better treatment outcomes.

Characterization of rifampicin-resistant Mycobacterium tuberculosis in Khyber Pakhtunkhwa, Pakistan.

Tuberculosis (TB), caused by Mycobacterium tuberculosis, is endemic in Pakistan. Resistance to both firstline rifampicin and isoniazid drugs (multidrug-resistant TB; MDR-TB) is hampering disease control. Rifampicin resistance is attributed to rpoB gene mutations, but rpoA and rpoC loci may also be involved. To characterise underlying rifampicin resistance mutations in the TB endemic province of Khyber Pakhtunkhwa, we sequenced 51 M. tuberculosis isolates collected between 2016 and 2019; predominantly, MDR-TB (n = 44; 86.3%) and lineage 3 (n = 30, 58.8%) strains. We found that known mutations in rpoB (e.g. S405L), katG (e.g. S315T), or inhA promoter loci explain the MDR-TB. There were 24 unique mutations in rpoA, rpoB, and rpoC genes, including four previously unreported. Five instances of within-host resistance diversity were observed, where two were a mixture of MDR-TB strains containing mutations in rpoB, katG, and the inhA promoter region, as well as compensatory mutations in rpoC. Heteroresistance was observed in two isolates with a single lineage. Such complexity may reflect the high transmission nature of the Khyber Pakhtunkhwa setting. Our study reinforces the need to apply sequencing approaches to capture the full-extent of MDR-TB genetic diversity, to understand transmission, and to inform TB control activities in the highly endemic setting of Pakistan.

Robust barcoding and identification of Mycobacterium tuberculosis lineages for epidemiological and clinical studies.

BACKGROUND: Tuberculosis, caused by bacteria in the Mycobacterium tuberculosis complex (MTBC), is a major global public health burden. Strain-specific genomic diversity in the known lineages of MTBC is an important factor in pathogenesis that may affect virulence, transmissibility, host response and emergence of drug resistance. Fast and accurate tracking of MTBC strains is therefore crucial for infection control, and our previous work developed a 62-single nucleotide polymorphism (SNP) barcode to inform on the phylogenetic identity of 7 human lineages and 64 sub-lineages. METHODS: To update this barcode, we analysed whole genome sequencing data from 35,298 MTBC isolates (~ 1 million SNPs) covering 9 main lineages and 3 similar animal-related species (M. tuberculosis var. bovis, M. tuberculosis var. caprae and M. tuberculosis var. orygis). The data was partitioned into training (N = 17,903, 50.7%) and test (N = 17,395, 49.3%) sets and were analysed using an integrated phylogenetic tree and population differentiation (FST) statistical approach. RESULTS: By constructing a phylogenetic tree on the training MTBC isolates, we characterised 90 lineages or sub-lineages or species, of which 30 are new, and identified 421 robust barcoding mutations, of which a minimal set of 90 was selected that included 20 markers from the 62-SNP barcode. The barcoding SNPs (90 and 421) discriminated perfectly the 86 MTBC isolate (sub-)lineages in the test set and could accurately reconstruct the clades across the combined 35k samples. CONCLUSIONS: The validated 90 SNPs can be used for the rapid diagnosis and tracking of MTBC strains to assist public health surveillance and control. To facilitate this, the SNP markers have now been incorporated into the TB-Profiler informatics platform ( https://github.com/jodyphelan/TBProfiler ).

The therapeutic milieu reconceptualized for the 21st century.

The notion of the therapeutic milieu has come under fire for lack of relevance to current inpatient psychiatric care environments. Yet, in different fields of health care, scholars are suggesting a need to build healing environments. A view of the therapeutic milieu as an optimal healing environment based on continuous healing relationships, patient-centered care, safety as a systems priority, and cooperation among clinicians provides a framework to organize care in a holistic manner that supports positive health outcomes. This approach provides a platform for nurses and other clinicians to expand the view of a milieu traditionally limited to the unit environment to one that includes a broad systems context.

Estimating the health impact of air pollution in Scotland, and the resulting benefits of reducing concentrations in city centres.

Air pollution continues to be a key health issue in Scotland, despite recent improvements in concentrations. The Scottish Government published the Cleaner Air For Scotland strategy in 2015, and will introduce Low Emission Zones (LEZs) in the four major cities (Aberdeen, Dundee, Edinburgh and Glasgow) by 2020. However, there is no epidemiological evidence quantifying the current health impact of air pollution in Scotland, which this paper addresses. Additionally, we estimate the health benefits of reducing concentrations in city centres where most LEZs are located. We focus on cardio-respiratory disease and total non-accidental mortality outcomes, linking them to concentrations of both particulate (PM10 and PM2.5) and gaseous (NO2 and NOx) pollutants. Our two main findings are that: (i) all pollutants exhibit significant associations with respiratory disease but not cardiovascular disease; and (ii) reducing concentrations in city centres with low resident populations only provides a small health benefit.

A model to estimate the impact of changes in MMR vaccine uptake on inequalities in measles susceptibility in Scotland.

An article published in 1998 by Andrew Wakefield in The Lancet (volume 351, pages 637-641) led to concerns surrounding the safety of the measles, mumps and rubella vaccine, by associating it with an increased risk of autism. The paper was later retracted after multiple epidemiological studies failed to find any association, but a substantial decrease in UK vaccination rates was observed in the years following publication. This paper proposes a novel spatio-temporal Bayesian hierarchical model with accompanying software (the R package CARBayesST) to simultaneously address three key epidemiological questions about vaccination rates: (i) what impact did the controversy have on the overall temporal trend in vaccination rates in Scotland; (ii) did the magnitude of the spatial inequality in measles susceptibility in Scotland increase due to the measles, mumps and rubella vaccination scare; and (iii) are there any covariate effects, such as deprivation, that impacted on measles susceptibility in Scotland. The efficacy of the model is tested by simulation, before being applied to measles susceptibility data in Scotland among a series of cohorts of children who were aged 2.5-4.5, in September of the years 1998 to 2014.

Relocating the Menninger Hospital: a nursing perspective.

The move of the Menninger Hospital represented a special challenge to the direct care nursing staff. As the efforts to implement an effective translocation of the clinical services moved forward, the goal included an effective transfer of the long-established therapeutic community and practices of Menninger into the new setting. This article reviews the challenges, processes, and problems in achieving that goal.