Diagnostic advantages of local versus nonlocal definitions of uropathogenic E. coli (UPEC) in semi-rural Ecuador

Abstract

Urinary tract infections (UTIs) are among the most common outpatient morbidities globally and can progress to invasive infections, including sepsis. Uropathogenic Escherichia coli (UPEC) is the most common cause of UTIs and third-generation cephalosporin-resistant (3GCR) UPEC are particularly important clinically; however due to its genetic heterogeneity identifying putative UPEC outside the context of clinical infection is challenging. As the molecular epidemiology of UPEC can have regional variation, local definitions may provide improved diagnostic power over existing molecular definitions. We evaluated the validity of previously established (i.e., “nonlocal”) definitions when applied to local samples collected in peri urban communities near Quito, Ecuador and compared them to a locally derived definition. For this study, we evaluated 124 virulence factor genes associated with UPEC in extended spectrum ß-lactamase (ESBL)-producing UPEC from clinical urine samples ( n = 137), i.e., urine culture associated with UTI, and E. coli selected for 3GC resistance from fecal samples ( n = 176) collected from healthy children in nearby communities. Using the 65 virulence genes found to be statistically more prevalent ( p < 0.05) in clinical vs. community samples we created receiver operating characteristic (ROC) curves based on the sensitivity and specificity of a sequence of 65 tests (i.e., the presence of all 65 genes, the presence of the 64 genes with the strongest associations, down to the presence of the single gene with the strongest association). The optimal local definition for ESBL-producing UPEC was the presence of ≥ 1 of upaH , cnf1 , or sat (Youden’s index = 0.505, 95% CI: 0.416, 0.596; sensitivity = 56.2%, 95% CI: 47.5, 64.7; specificity = 94.3%, 95% CI: 89.8, 97.2), which showed better diagnostic performance among isolates than nonlocal definitions of UPEC (YI ranged from 0 to 0.361). Additionally, when evaluating all possible cutoffs (i.e., minimum total number of genes present to qualify as putative UPEC) that could be utilized with nonlocal UPEC definitions, the best performing cutoffs among our local isolates according to Youden’s index differed from cutoffs seen in the literature. These findings suggest that nonlocal definitions of UPEC, based on the presence of specific sets of virulence genes, may not perform well in some settings, and local definitions derived from local clinical and commensal samples may offer diagnostic advantages.