Scientists can predict ciprofloxacin resistance in Klebsiella pneumoniae from genome data

A global research consortium has shown that ciprofloxacin resistance in Klebsiella pneumoniae can be predicted from genome sequence data with enough accuracy to support surveillance, although the accuracy falls short of what clinical decision-making would require. The work, published in Genome Medicine, was carried out by the KlebNET-GSP AMR Genotype-Phenotype Group, a large collaboration led by Kara K. Tsang and Kathryn E. Holt, with more than one hundred contributing researchers.

The team assembled a matched genotype-phenotype dataset of 12,167 isolates from the Klebsiella pneumoniae species complex, drawn from 27 countries and collected between 2001 and 2021. Each isolate carried both whole genome sequence data and laboratory-measured ciprofloxacin susceptibility, allowing the researchers to link specific genetic features to observed resistance phenotypes.

From this discovery dataset, the group built a rule-based classifier that predicts whether an isolate is resistant or susceptible to ciprofloxacin. The rules categorize the number of mutations in the quinolone resistance determining regions of two target genes, gyrA and parC, together with the number of plasmid-mediated quinolone resistance genes. The classifier also accounts for the presence or absence of aac(6′)-Ib-cr, an enzyme that can acetylate ciprofloxacin and thereby reduce its activity.

When tested against the discovery dataset, the classifier achieved categorical agreement, sensitivity and specificity all above 96 percent, with major and very major error rates below 4 percent. Performance remained similar across diverse isolate sources, including human, animal and other origins, and across different species and lineages within the K. pneumoniae species complex.

External validation provided a more conservative picture. Using independently contributed datasets comprising 7,030 isolates from the species complex, the classifier reached 90.18 percent categorical agreement, with a 95 percent confidence interval of 89.49 to 90.88 percent. Major errors occurred at 8.74 percent, very major errors at 4.19 percent and minor errors at 10.14 percent, each with reported confidence intervals.

The classifier has been implemented in Kleborate, a command-line tool that is also integrated into the Pathogenwatch web platform, making it publicly accessible to researchers and public health laboratories. The authors note that no publicly accessible tools for predicting ciprofloxacin resistance from whole genome sequence data previously existed, despite the resistance determinants themselves being known.

With the tool in place, the team mapped the global distribution of ciprofloxacin resistance determinants across 31,319 genomes of the species complex available in Pathogenwatch, representing 109 countries and sampling years from 2000 to 2023. This analysis revealed a significant positive association between national quinolone consumption rates and predicted ciprofloxacin resistance, with an R-squared of 0.20 and a p-value of 0.004.

The authors emphasize that their findings support surveillance rather than clinical use. Ciprofloxacin resistant K. pneumoniae is common or emerging in many geographies, and knowledge of local resistance rates is important for informing empirical therapy, but the study concludes that unexplained resistance remains and that the observed accuracy is insufficient for clinical applications.

The work also demonstrates the value of aggregating genotype-phenotype data at scale to explore resistance mechanisms and develop predictors. At the same time, the researchers highlight the complexities involved in combining phenotype data generated by different laboratory assays and testing standards, a challenge inherent to any effort to harmonize susceptibility data contributed from many laboratories worldwide.

Subject of Research: Prediction of ciprofloxacin resistance in Klebsiella pneumoniae from whole genome sequence data

Article Title: Ciprofloxacin resistance in Klebsiella pneumoniae: phenotype prediction from genotype and global distribution of resistance determinants

Article References: Tsang, K. K., Amke, I., Aanensen, D. M., Abdulahi, J., Aiken, A., Bachman, M. A., Baker, S., Barry, K., Biffignandi, G. B., de Andrade Belitardo, E. M. M., Bloise, I., Boutiba-Ben Boubaker, I., Fortuin, C., Brisse, S., Campino, S., Cantón, R., Chiaverini, A., Cirillo, D. M., Clark, T. G., … on behalf of the KlebNET-GSP AMR Genotype-Phenotype Group (2026). Ciprofloxacin resistance in Klebsiella pneumoniae: phenotype prediction from genotype and global distribution of resistance determinants. Genome Medicine. https://doi.org/10.1186/s13073-026-01785-w

Image Credits: AI Generated

DOI: 10.1186/s13073-026-01785-w

Keywords: Klebsiella pneumoniae, ciprofloxacin resistance, whole genome sequencing, antimicrobial resistance, Kleborate, Pathogenwatch, genotype-phenotype prediction, gyrA, parC, surveillance, quinolone consumption, plasmid-mediated quinolone resistance