Bacterial genome-wide association study substantiates papGII of Escherichia coli as a major risk factor for urosepsis

BACKGROUND

Urinary tract infections (UTIs) are among the most common bacterial infections worldwide, often caused by uropathogenic Escherichia coli. Multiple bacterial virulence factors or patient characteristics have been linked separately to progressive, more invasive infections. In this study, we aim to identify pathogen- and patient-specific factors that drive the progression to urosepsis by jointly analysing bacterial and host characteristics.

METHODS

We analysed 1076 E. coli strains isolated from 825 clinical cases with UTI and/or bacteraemia by whole-genome sequencing (Illumina). Sequence types (STs) were determined via srst2 and capsule loci via fastKaptive. We compared the isolates from urine and blood to confirm clonality. Furthermore, we performed a bacterial genome-wide association study (bGWAS) (pyseer) using bacteraemia as the primary clinical outcome. Clinical data were collected by an electronic patient chart review. We concurrently analysed the association of the most significant bGWAS hit and important patient characteristics with the clinical endpoint bacteraemia using a generalised linear model (GLM). Finally, we designed qPCR primers and probes to detect papGII-positive E. coli strains and prospectively screened E. coli from urine samples (n = 1657) at two healthcare centres.

RESULTS

Our patient cohort had a median age of 75.3 years (range: 18.00-103.1) and was predominantly female (574/825, 69.6%). The bacterial phylogroups B2 (60.6%; 500/825) and D (16.6%; 137/825), which are associated with extraintestinal infections, represent the majority of the strains in our collection, many of which encode a polysaccharide capsule (63.4%; 525/825). The most frequently observed STs were ST131 (12.7%; 105/825), ST69 (11.0%; 91/825), and ST73 (10.2%; 84/825). Of interest, in 12.3% (13/106) of cases, the E. coli pairs in urine and blood were only distantly related. In line with previous bGWAS studies, we identified the gene papGII (p-value < 0.001), which encodes the adhesin subunit of the E. coli P-pilus, to be associated with 'bacteraemia' in our bGWAS. In our GLM, correcting for patient characteristics, papGII remained highly significant (odds ratio = 5.27, 95% confidence interval = [3.48, 7.97], p-value < 0.001). An independent cohort of cases which we screened for papGII-carrying E. coli at two healthcare centres further confirmed the increased relative frequency of papGII-positive strains causing invasive infection, compared to papGII-negative strains (p-value = 0.033, chi-squared test).

CONCLUSIONS

This study builds on previous work linking papGII with invasive infection by showing that it is a major risk factor for progression from UTI to bacteraemia that has diagnostic potential.

Identification of microorganisms by a rapid PCR panel from positive blood cultures leads to faster optimal antimicrobial therapy - a before-after study

BACKGROUND

The BioFire® FilmArray® Blood Culture Identification Panel 1 (BF-FA-BCIP) detects microorganisms with high accuracy in positive blood cultures (BC) - a key step in the management of patients with suspected bacteraemia. We aimed to compare the time to optimal antimicrobial therapy (OAT) for the BF-FA-BCIP vs. standard culture-based identification.

METHODS

In this retrospective single-centre study with a before-after design, 386 positive BC cases with identification by BF-FA-BCIP were compared to 414 controls with culture-based identification. The primary endpoint was the time from BC sampling to OAT. Secondary endpoints were time to effective therapy, length of stay, (re-)admission to ICU, in-hospital and 30-day mortality. Outcomes were assessed using Cox proportional hazard models and logistic regressions.

RESULTS

Baseline characteristics of included adult inpatients were comparable. Main sources of bacteraemia were urinary tract and intra-abdominal infection (19.2% vs. 22.0% and 16.8% vs. 15.7%, for cases and controls, respectively). Median (95%CI) time to OAT was 25.5 (21.0-31.2) hours with BF-FA-BCIP compared to 45.7 (37.7-51.4) hours with culture-based identification. We observed no significant difference for secondary outcomes.

CONCLUSIONS

Rapid microorganism identification by BF-FA-BCIP was associated with a median 20-h earlier initiation of OAT in patients with positive BC. No impact on length of stay and mortality was noted.

TRIAL REGISTRATION

Clinicaltrials.gov, NCT04156633, registered on November 5, 2019.

653. Direct Identification of Microorganisms in Positive Blood Cultures by the BioFire® FilmArray® Blood Culture Identification Panel Leads to Faster Optimal Antibiotic Therapy: A Before–After Study

Background

Rapid pathogen identification from positive blood cultures may help optimize empiric antibiotic therapy quickly by reducing unnecessary broad spectrum antibiotic use and may improve patient outcomes. The BioFire® FilmArray® Blood Culture Identification Panel 1 (BF-FA-BCIP) identifies 24 pathogens directly from positive blood cultures without subculture. 3 resistance genes are included. We aimed to compare the time to optimal antibiotic therapy between BF-FA-BCIP and conventional identification.

Methods

We performed a single-center retrospective case-control before-after study of 386 cases (November 2018 to October 2019) with BF-FA-BCIP compared to 414 controls (August 2017 to July 2018) with conventional identification. The primary study endpoint was the time from blood sampling to implementation of optimal antimicrobial therapy. Secondary endpoints were time to effective therapy, length of hospital stay, and in-hospital and 30-day mortality. Outcomes were assessed using cause-specific Cox Proportional Hazard models and logistic regressions.

Results

We included 800 patients with comparable baseline characteristics. Main sources of blood stream infection (BSI) were urinary tract infection and intra-abdominal infection (19.2% vs. 22.0% and 16.8% vs. 15.7% for case and control groups, respectively). Overall, 212 positive blood cultures were considered as contaminations. Identification results were available after a median of 21.9 hours by the BF-FA-BCIP and 44.3 hours by the conventional method. Patients with BF-FA-BCIP received the optimal therapy after a median of 25.5 hours (95%CI 21.0 - 31.2) as compared to 45.7 hours (95%CI 37.7 - 51.2) in the control group (Figure 1). We found no effect of the identification method on secondary outcomes.

Kaplan-Meier curve representing the probability of implementing the optimal therapy at any given time according to the identification method (Standard vs. BF-FA-BCIP).

Shaded ribbons represent the 95 % confidence interval (CI). The vertical dashes represent censored data. The vertical dotted lines represent the median time, i.e. the time at which 50 % of the patients obtained the optimal therapy, for the two methods. Median (95 % CI) time to optimal therapy is 45.7 (37.7 - 51.4) hours with the Standard method and 25.5 (21.0- 31.2) hours with Biofire. The tables below the curves present the numbers expecting optimal therapy according to the bacteria identification method, as well as the number of censored data in parenthesis. Panel A shows data from 0 to 900 hours. Panel B shows the data from 0 to 90 hours to better visualize how the probability to implement optimal therapy varies in the first 72 hours.

Conclusion

In conclusion, rapid pathogen identification by BF-FA-BCIP was associated with an almost 24h earlier initiation of the optimal antibiotic therapy in BSI. However, the overall benefit for individual patients seems to be limited. Future studies should assess the cost-effectiveness and impact on the prevention of antibiotic resistance using this diagnostic approach.

Disclosures

All Authors: No reported disclosures

Clinical impact of the type VI secretion system on virulence of Campylobacter species during infection

BACKGROUND

The clinical course of Campylobacter infection varies in symptoms and severity depending on host factors, virulence of the pathogen and initiated therapy. The type VI secretion system (T6SS) has been identified as a novel virulence factor, which mediates contact-dependent injection of enzymes and toxins into competing bacteria or host cells and facilitates the colonisation of a host organism. We aimed to compare the clinical course of Campylobacter infection caused by strains with and without the T6SS and identify possible associations between this putative virulence factor and the clinical manifestations of disease.

METHODS

From April 2015 to January 2017, patients with detection of Campylobacter spp. were identified at the University Hospital of Basel and the University Children's Hospital of Basel and included in this case-control study. Presence of the T6SS gene cluster was assayed by PCR targeting the hcp gene, confirmed with whole genome sequencing. Pertinent clinical data was collected by medical record review. Differences in disease- and host-characteristics between T6SS-positive (case) and -negative (control) were compared in a uni- and multi-variable analysis. Hospital admission, antibiotic therapy, admission to intensive care unit, development of bacteraemia and in-hospital mortality were considered as clinical endpoints.

RESULTS

We identified 138 cases of Campylobacter jejuni infections and 18 cases of Campylobacter coli infections from a paediatric and adult population. Analyses were focused on adult patients with C. jejuni (n = 119) of which 16.8% were T6SS-positive. Comparisons between T6SS-positive and -negative C. jejuni isolates did not reveal significant differences regarding clinical manifestations or course of disease. All clinical endpoints showed a similar distribution in both groups. A higher score in the Charlson Comorbidity Index was associated with T6SS-positive C. jejuni isolates (p < 0.001) and patients were more likely to have a solid organ transplant and to be under immunosuppressive therapy.

CONCLUSIONS

Our study does not provide evidence that T6SS is associated with a more severe clinical course. Interestingly, T6SS-positive isolates are more commonly found in immunocompromised patients: an observation which merits further investigation.