Abstract
AbstractBackgroundWhile Neisseria gonorrhoeae poses an urgent public health threat because of increasing antimicrobial resistance, much of the circulating population remains susceptible to historical treatment regimens. Point-of-care diagnostics that report susceptibility could allow for reintroduction of these regimens, but development of such diagnostics has been limited to ciprofloxacin, for which susceptibility can be predicted from a single locus.MethodsWe assembled a dataset of 12,045 N. gonorrhoeae genomes with phenotypic resistance data for tetracycline (n = 3,611) and penicillin (n = 6,935). Using conditional genome wide association studies (GWAS), we sought to define genetic variants associated with susceptibility to penicillin and tetracycline. We evaluated the sensitivity and specificity of these variants for predicting susceptibility and non-resistance in our collection of gonococcal genomes.FindingsIn our conditional penicillin GWAS, the presence of a genetic variant defined by a non-mosaic penA allele without an insertion at codon 345 was significantly associated with penicillin susceptibility and had the highest negative effect size of significant variants (p = 5.0 × 10−14, β = -2.5). In combination with the absence of blaTEM, this variant predicted penicillin susceptibility with high specificity (99.8%) and modest sensitivity (36.7%). For tetracycline, the wild type allele at rpsJ codon 57, encoding valine, was significantly associated with tetracycline susceptibility (p = 5.6 × 10−16, β = -1.6) after conditioning on the presence of tetM. The combination of rpsJ codon 57 allele and tetM absence predicted tetracycline susceptibility with high specificity (97.2%) and sensitivity (88.7%).InterpretationAs few as two genetic loci can predict susceptibility to penicillin and tetracycline in N. gonorrhoeae with high specificity. Molecular point-of-care diagnostics targeting these loci have the potential to increase available treatments for gonorrhea.FundingNational Institute of Allergy and Infectious Diseases, the National Science Foundation, and the Smith Family FoundationResearch in ContextEvidence before this studyWe searched PubMed with the terms “Neisseria gonorrhoeae” and “diagnostic” or “assay” plus “penicillin” or “tetracycline” for reports in any language published up to July 1, 2021. We additionally searched for “Neisseria gonorrhoeae” and “genome wide association study”. We found that previously proposed molecular diagnostics for penicillin and tetracycline susceptibility either exclusively focused on plasmid-mediated resistance (i.e., targeting blaTEM or tetM) or did not include variants in genes encoding antibiotic targets (e.g., did not include penA or rpsJ). Targets for molecular surveillance have focused on resistance-associated alleles rather than susceptibility-associated alleles. We did not find any previous penicillin or tetracycline conditional genome wide association studies (GWAS) in N. gonorrhoeae.Added value of this studyTo identify targets for molecular diagnostics that predict penicillin and tetracycline susceptibility, we conducted GWAS conditioning on the presence of plasmid-mediated resistance determinants to detect chromosomal loci with the highest association with susceptibility. We discovered a sequence (penA_01) that differentiates susceptible isolates from those with a resistance-associated insertion at codon 345 and from those with mosaic penA alleles, which is associated with penicillin susceptibility. We also found that rpsJ codon 57 was the chromosomal locus contributing the most to tetracycline susceptibility. The combination of these chromosomal loci and the absence of plasmid encoded determinants predicts penicillin and tetracycline susceptibility with high specificity in both a large global collection of N. gonorrhoeae and a validation dataset consisting of recently published genomes from CDC’s Gonococcal Isolate Surveillance Program (GISP) surveillance collected in 2018.Implications of all the available evidenceThe chromosomal loci penA_01 and rpsJ codon 57 in combination with plasmid loci blaTEM and tetM are candidates for the development of point-of-care molecular diagnostics for penicillin and tetracycline susceptibility. The loci may be combined with the currently available ciprofloxacin susceptibility diagnostics to predict susceptibility to multiple antibiotics. Additionally, our study suggests that conditional GWAS focused on variants associated with susceptibility may be a promising approach to identify minimal sets of loci for molecular diagnostics and surveillance.
Publisher
Cold Spring Harbor Laboratory