Our study was conducted in three tertiary hospitals, including the First Affiliated Hospital of Zhejiang University School of Medicine (ZYH, equipped with 2,500 beds), the Fourth Affiliated Hospital of Zhejiang University School of Medicine (ZSH, equipped with 1,100 beds) and the Second Affiliated Hospital of Wenzhou Medical University (WYH, equipped with 2,900 beds) from January 2012 to December 2020. The inclusion criteria consisted of (i) CRPA was detected at least once in blood or intravenous catheter culture; (ii) the first episode of bloodstream culture; (iii) patients could be diagnosed as P. aeruginosa BSI (Horan et al., 2008). The primary outcomes were the 30 day mortality. A total of 177 patients with CRPA BSI were identified during the 9 year study period, then 40 cases were excluded due to polymicrobial bacteremia (n = 35), incomplete medical record (n = 2), or donor-derived infection (n = 3). The remaining 137 CRPA BSI patients were eligible for the cohort. 111 patients were from ZYH, 10 from ZSH, and 26 from WYH. Carbapenem-sensitive P. aeruginosa (CSPA) BSI cases detected in the current year were randomly matched with CRPA BSI cases, with 1:1 ratio. If the number of CSPA cases was insufficient in the current year, cases that did not enter the cohort study in the adjacent year were randomly selected. Each hospital followed this matching principle. 137 CSPA bacteremia patients were paired. The study flow chart was shown in Supplementary Figure S1.

We conducted a retrospective, multicenter cohort study to identify clinical characteristics and risk factors for CRPA BSI. Therapy responses and risk factors associated with CRPA BSI mortality were also evaluated. Epidemiology trends of P. aeruginosa isolates in ZYH were analyzed. The data of 274 patients were collected from the electronic medical records. The following data were recorded: antimicrobial susceptibility, demographic, ward, underlying disease, type of resistance, underlying conditions, prior invasive procedure and/or devices, source of bacteremia, exposure to antibiotics within 90 days, conditions after BSI, laboratory examinations, therapeutic medication and mortality.

BSI was defined as patients with at least once positive peripheral and/or central line blood culture of P. aeruginosa, accompanied by signs and symptoms of infection. CRPA was defined as resistance to imipenem or meropenem (the minimum inhibitory concentration (MIC) ≥ 8 μg/mL), and CSPA was defined as susceptibility to imipenem or meropenem (the MIC≤2 μg/mL; CLSI, 2020). Multidrug-resistant P. aeruginosa (MDRPA) was defined as non-susceptibility to at least one agent in three or more antipseudomonal antimicrobials. DTRPA was defined as non-susceptibility to all of the following antimicrobials: ceftazidime, cefepime, ciprofloxacin, levofloxacin, meropenem, imipenem-cilastatin, aztreonam and piperacillin-tazobactam (Tamma et al., 2022). Neutropenia was defined as an absolute neutrophil count <0.5*10^9 cells/L. Appropriate initial antibacterial therapy was defined as receiving one or more antimicrobial agents with in vitro activity within 48 h. Appropriate therapy was defined as receiving at least one antimicrobial with in vitro activity (Gutiérrez-Gutiérrez et al., 2017). Previous antibiotic therapy was defined as antibiotic used for at least 24 h within the last 90 days before onset of BSI. Glucocorticoid therapy was defined as methylprednisolone consumption at least 40 mg per day (or its equivalent) for more than 48 h. Clinical culture samples of P. aeruginosa include: intravenous catheters, bloodstream, feces, urine, bile, cerebrospinal fluid, secretions, drainage fluid, throat swabs, sputum, lavage fluid, brushes, transplant organ preservation fluid, peritoneal fluid and vegetations culture samples.

In this study, microorganisms were identified using the VITEK 2 system (bioMérieux, France). MIC of 12 antipseudomonal agents were performed by the agar dilution method, including ceftazidime, cefepime, ciprofloxacin, levofloxacin, amikacin, gentamicin, tobramycin, imipenem, meropenem, piperacillin-tazobactam, cefoperazone-sulbactam and aztreonam. MIC of polymyxin were performed by the broth microdilution method. The results were interpreted by Clinical and Laboratory Standards Institute standards (CLSI, 2020). Isolates that showed intermediate susceptibility were considered resistant (exclude polymyxin).

Chi-square test or Fisher’s exact test was used for categorical variables, and t-test or Mann–Whitney U-test for continuous variables to compare groups in our study. Selected variables with p values ≤0.20 in univariate analysis were included in multivariate analyses. We used binary logistic regression to evaluate predictors associated with the development of CRPA BSI and used cox regression to identify risk factors associated with the outcomes of CRPA BSI. The Kaplan–Meier method was used to compare time to mortality. All statistical analyses were performed in IBM SPSS Statistics v.23.0. p values <0.05 were considered statistically significant.

Since ZYH detected the largest number of CRPA BSI cases among the three hospitals, it was chosen for analyzing P. aeruginosa epidemiological characteristics. The non-duplicate P. aeruginosa and CRPA clinical isolates were collected between January 2012 and December 2020 (Figure 1A). A total of 7,384 P. aeruginosa clinical samples were cultured in ZYH during 9 years. From 2014, we recognized a remarkable increase in both P. aeruginosa clinical culture and bloodstream culture samples, and the first downward trend was shown by 2020. The proportion of carbapenem resistant isolates among all P. aeruginosa clinical culture samples increased notably from 27% in 2012 to 48% in 2020, and annual incidence increased year on year. Since 2012, we found CRPA detection rate in bloodstream culture samples increased dramatically from 17% in 2012, to 55% in 2019, even 60% in 2020 (Figure 1B). Moreover, DTRPA detection rate in bloodstream culture samples increase dramatically from 2.1% in 2012 to 25% in 2020.

Antimicrobial susceptibility of the 274 P. aeruginosa isolates is listed in Table 1. Apart from carbapenem resistant phenotypes, the 137 CRPA bloodstream culture isolates in our study also exhibited resistance to ceftazidime, cefepime, ciprofloxacin, levofloxacin, piperacillin-tazobactam, cefoperazone-sulbactam and aztreonam, and the antimicrobial resistance were 60, 56, 51, 54, 45, 62 and 57%, respectively. However, the antimicrobial resistance to amikacin, tobramycin and polymyxin were only 4.1, 8.3 and 5.8%, respectively. Except for amikacin and polymyxin, the remain 11 antibiotics in CRPA group showed significantly higher resistance than that in CSPA isolates. Moreover, the proportion of MDRPA and DTRPA isolates were notably higher in CRPA group than that in CSPA group (68% vs. 7.3%, p < 0.001; 34% vs. 0, p < 0.001). Forty-six DTRPA isolates were detected among all P. aeruginosa isolates (Table 2).

The baseline characteristics of 274 P. aeruginosa BSI patients were presented in Table 2. Most patients were male (69%, n = 188) and elderly older than 55 years (66%, n = 182). 51% (n = 140) of the patients had history of admission at least once prior to their first positive culture. The majority of patients came from ICU (35%, n = 97) and the hematology ward (18%, n = 48). Cardiovascular disease (mainly hypertension) presented 41% (n = 111) of the underlying disease, followed by solid malignant tumor (25%, n = 68), hematological diseases (25%, n = 67), diabetes (19%, n = 53). Among patients with hematological diseases, 53(79%) were acute leukemia or lymphoma. Underlying disease were not statistically different between CRPA and CSPA group, implying that the baseline general conditions were balanced. Except for unknown origin of infection, lung (22%), biliary tract source (16%), catheter related infection (9.9%) and urinary tract source (8.0%) were the most frequent infection sites, and this trend was consistent across the two groups.

In the univariate analysis of patients with CRPA and CSPA BSI, the results revealed that patients infected by CRPA tended to had prior ICU hospitalization, longer hospital stay before infection, glucocorticoid therapy, immunosuppressive therapy, receipt of mechanical ventilation and central venous catheter (CVC), lower hemoglobin, higher Pitt bacteremia score (PBS), and poor prognosis after BSI (including multiple organ failure (MOF), p = 0.003; sepsis/septic shock, p = 0.002; underwent mechanical ventilation, p = 0.015; longer total length of hospital stay, p < 0.001; higher medical expenses, p < 0.001).

Moreover, hospital stay before infection in CRPA group was one-fold than CSPA group [16 (6, 29) days vs. 8 (2, 17) days; p < 0.001]. The difference of antibiotics exposed within 90 days prior to recovery between carbapenem resistance group and carbapenem susceptibility group lay mainly in carbapenems (68% vs. 29%, p < 0.001), meropenem (39% vs. 6.6%, p < 0.001), BLBLIs(53% vs. 36%, p = 0.004)and tigecycline (21% vs. 8.8%, p = 0.004). Neither daily dose [3.00 (1.50, 3.00) grams vs. 3.00 (2.13, 3.00) grams; p = 0.484) nor total consumption [24.00 (6.00, 34.50) grams vs. 12.75 (5.25, 21.00 grams; p = 0.180)] of meropenem expose was different between the two groups.

When entering the multivariate logistic analysis, prior ICU hospitalization (OR 4.160; 95% CI 1.985–8.719; p < 0.001), immunosuppressive therapy (OR 2.112; 95% CI 1.061–4.203; p = 0.033), and exposure to carbapenems within 90 days (including meropenem, imipenem, biapenem) were independent risk factors associated with development of CRPA bacteremia (Table 3).

During the 30 days follow-up, 94 (34%) of 274 P. aeruginosa infected patients died and 23 (50%) of 46 DTRPA infected patients died. The 7 day, 14 day and 30 day crude mortality in CRPA group and CSPA group were 27% vs. 17% (p = 0.041), 31% vs. 23% (p = 0.103), 39% vs. 29% (p = 0.075), respectively. Distinction between the two groups became inapparent as the disease progressed for more than 2 weeks. Of 274 eligible patients in our study, 201 (73%) received appropriate therapy after BSI onset, whereas that rate of CRPA BSI patients died within 7 day and 30 day were only 35 and 41%, respectively. The median PBS of CRPA BSI patients died within 7 days was 6 and that patients deceased within 30 days was 5; the median PBS was 2 of surviving patients. In 274 P. aeruginosa BSI patients, the 30-day survival probability of patients with CRPA BSI was significantly worse than that of those with CSPA BSI [Hazard Ratio (HR) 1.732, 95% CI 1.117–2.686, p = 0.014, log-rank test; Figure 2A]. The survival of patients with MDRPA BSI was significantly worse than that of patients with non-MDRPA BSI (HR 2.301, 95% CI 1.452–3.646, p < 0.001; Figure 2B). The survival of patients with DTRPA BSI was significantly worse than that of patients with non-DTRPA BSI (HR 2.138, 95% CI 1.143–4.000, p = 0.002; Figure 2C).

The univariate analysis for 7-day crude mortality of CRPA BSI was shown in Supplementary Table S1. Multivariate cox regression analysis (Table 3) identified the following as independent risk factors for 7-d mortality: hematological disease (HR 2.785; 95% CI 1.349–5.750, p = 0.006), higher PBS (HR 1.193; 95% CI 1.092–1.304, p < 0.001), MOF (HR 3.825; 95% CI 1.493–9.798, p = 0.005). Receipt of appropriate therapy was identified as independent factor associated with better outcome (HR 0.327; 95% CI 0.154–0.693, p = 0.004).

The univariate analysis revealed that the patients deceased within 30 days tend to had hematological disease, glucocorticoid therapy, MOF and sepsis or septic shock, lower platelet, higher C-reactive protein and procalcitonin, agranulocytosis and higher PBS (Table 4). No association between different sources of CRPA bacteremia and 30-day mortality. In multivariate cox regression (Table 3), the 30-day crude mortality of CRPA BSI was independently associated with MOF (HR 7.098; 95% CI 3.435–14.667; p < 0.001) and higher PBS (HR 1.092; 95% CI 1.009–1.182; p = 0.030), whereas receipt appropriate therapy improved prognosis (HR 0.312; 95% CI 0.164–0.593; p < 0.001). This conclusion was consistent with the results of the survival analysis (Figures 2D,E).