Epidemiology and Genomic Investigation of Clostridioides Difficile in Hospitalized Patients With Nosocomial Diarrhoea
Abstract
Introduction: Clostridioides difficile is associated with hospital-acquired diarrhoea with underreported disease burden from African countries. This study aimed to ascertain the prevalence and the predictors of healthcare facility-onset C. difficile infection (HO-CDI) in symptomatic hospitalized patients admitted at Kenyatta National Hospital (KNH), as well as to characterize the toxin variants, antibiotic resistance determinants, sequence types, and evolutionary strains of the isolates associated with HO-CDI. Methods: A cross-sectional study was conducted in 333 hospitalized patients with hospital-onset diarrhoea at KNH. Patients' demographic, admission, and clinical information were extracted from their medical records. Stool samples from study participants who gave their consent were tested for C. difficile using anaerobic culture-based methods in selective media. HO-CDI cases were confirmed by a positive real-time PCR assay for tpi gene along with one or more toxin genes (tcdA, tcdB, or cdtA/cdtB). E-test strips were used to detect the susceptibility of confirmed isolates to a panel of antibiotics, including vancomycin, metronidazole, rifampicin, ciprofloxacin, tetracycline, clindamycin, erythromycin, and ceftriaxone. Logistic regression was used to examine potential risk predictors in cases of confirmed HO-CDI. The genetic relatedness of selected isolates was determined using multi-locus sequence typing (MLST). The Oxford nanopore MinION technology was used to sequence the entire genome of nine C. difficile strains. PubMLST and MLST (v2.0) were used to perform multilocus sequencing typing on the generated genomes. Various databases, including card, vfdb core, plasmidfinder, resfinder, and virulencefinder, were utilised to detect virulence factors, antimicrobial resistance genes, toxin coding genes, and plasmids replicons in the generated genomes. Phylogeny and metadata overlay were carried out using Phandago to determine the degree of genetic relatedness between the isolates. The sequences were aligned using Roary, and a maximum likelihood tree was constructed using RAXML (v0.9.0). Results: C. difficile was found in 71 (21%) of the patients. One or more toxin genes were present in 69 (97.1%) of the 71 confirmed isolates. An incomplete tcdA gene was present in more than half of the toxigenic isolates. All isolates were sensitive to vancomycin, but three (2.1%) were resistant to metronidazole (MIC >32 mg/L). Resistance to rifampicin (65/71, 91.5%), erythromycin (63/71, 88.7%), ciprofloxacin (59/71, 83.1%), clindamycin (57/71, 80.3%), and ceftriaxone (36/71, 50.7.8%) was observed. Among the resistant isolates, 61 (85.9%) were multidrug-resistant. Significant predictors in the multivariate logistic regression model included chronic obstructive pulmonary disease (odds ratio [OR], 9.51; 95% confidence interval [CI], 1.80-
9
50.10), diabetes (odds ratio [OR], 3.56; 95% CI, 1.11-11.384), chronic kidney disease (odds ratio [OR], 3.88; 95% CI, 1.57-9.62), iron deficiency anemia (OR, 3.67; 95% CI, 1.61–8.34) and hypertension (OR, 2.47; 95% CI, 1.00–6.07). CCI score of 2 (OR, 6.67; 95% CI, 2.07 – 21.48) and ECI scores of 1 (OR, 4.07; 95% CI, 1.72 – 9.65), 2 (OR, 2.86; 95% CI, 1.03 – 7.89), and 3 (OR, 4.87; 95% CI, 1.40 – 16.92) were significantly associated with increased odds of developing HO-CDI. In addition, age, antibiotic exposure, use of more than one antibiotic, surgical interventions and nasogastric feeding were significantly associated with increased odds of developing HO-CDI.
The analysis of the nine assembled genomes revealed that, with the exception of three genomes lacking resistance genes, the majority of isolates conferred antimicrobial resistance. Some of the antimicrobial resistance genes found in the six genomes included those for lincosamides (erm(G) and erm(B)), tetracycline (tet(M)), macrolides (msr(C), msr(D) and msr(A)), rifamycin (RpoB), fluoroquinolone (GyrA), and aminoglycosides (ant(6') and aac(6')). RepUS43 plasmid was found in six isolates in the PlasmidFinder database. Four previously described sequence types were identified (ST37, ST743, ST40, and ST58), while two were novel. The phylogenetic inference analysis of C. difficile isolates' genomes revealed that they belonged to two distinct clades (clades 1 and 4). The Kenya sequences clustered with sequences from Indonesia, the United States of America, and Ghana.
Conclusion: C. difficile diarrhoea was identified in the hospitalized population, and the risk was higher for patients with prior exposure to antibiotics, invasive procedures, and co-morbidities. The presence of diverse sequence types and virulence genes among the few sequenced isolates provides novel insights into C. difficile isolates from this region, forming a basis for future studies using a larger population to investigate the genetic relationship of these isolates.
Publisher
University of Nairobi
Rights
Attribution-NonCommercial-NoDerivs 3.0 United StatesUsage Rights
http://creativecommons.org/licenses/by-nc-nd/3.0/us/Collections
The following license files are associated with this item: