Clinical Features and Outcomes of Bloodstream Infections Caused by New Delhi Metallo-β-Lactamase–Producing Enterobacterales During a Regional Outbreak

Abstract

Limited data about New Delhi metallo-β-lactamase (NDM) bacteremia are available. Blood isolates from 40 patients with NDM bacteremia were studied for antibiotic susceptibility and whole-genomic sequencing. NDM bacteremia has high 30-day mortality. In most cases, aztreonam-avibactam is active in vitro. Ceftazidime-avibactam plus aztreonam may represent a feasible therapeutic option.

An outbreak of New Delhi metallo-β-lactamase (NDM)–producing Enterobacterales was recently documented in the northwestern area of Tuscany [1]. From November 2018 to May 2019, colonization or infection by NDM-producing Enterobacterales was documented in 350 patients from 9 different hospitals [1]. We retrospectively reviewed the clinical and microbiological characteristics of 40 patients with documented bloodstream infection (BSI).

METHODS

Cases of NDM-producing BSI were identified by reviewing records from the microbiology laboratories of all 9 institutions. Demographics, clinical and laboratory findings, comorbid conditions, source of infection, source control data, treatment regimens, and 30-day mortality rates were collected from clinical charts. Source of infection and source control were defined as previously described [2]. Septic shock was defined according to the Sepsis-3 definition [3]. The study was approved by the local ethical committee.

Blood isolate identification and susceptibility testing were performed by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-ToF MS; Vitek MS, bioMérieux; or MALDI Biotyper, Bruker Daltonics). Carbapenemase determinants were evaluated by either the Allplex Entero-DR Assay (Seegene) or RESIST-3 O.K.N. ICT immunocromatographic assay (Coris BioConcept, Gembloux, Belgium) and confirmed by real-time polymerase chain reaction as previously described [4]. Antimicrobial susceptibility testing was carried out with reference broth microdilution, except agar dilution for fosfomycin, according to the ISO 20776-1:2006 guidelines [5], and interpreted according to the EUCAST clinical breakpoints (v.9.0 2019; http://www.eucast.org/clinical_breakpoints/).

Isolates were subjected to whole-genome sequencing (WGS) with an Illumina MiSeq platform (Illumina Inc., San Diego, CA, USA) and a paired-end approach (2 × 300 bp). Raw sequences were assembled using SPAdes software [6]. In silico analyses using draft-assembled genomes were performed by dedicated tools available at http://www.genomicepidemiology.org/ (eg, MLST v.2.0) and by the BLAST suite (https://blast.ncbi.nlm.nih.gov/Blast.cgi).

Continuous variables were reported as medians and interquartile ranges (IQRs). The Mann‑Whitney U test was used to analyze non–normally distributed data. Categorical data were expressed as frequency distributions, and the chi-square test or Fisher exact test was used to determine if differences existed between groups. Factors influencing 30-day survival were examined by univariate analysis. All significant variables at univariate analyses (P < .05) were considered for the multivariate model. Hazard ratios (HRs) and 95% confidence intervals (95% CI) were calculated. Statistical significance was established at P ≤ .05.

RESULTS

Overall, during the study period, 47 patients with BSI caused by NDM-producing strains were initially identified in 9 hospitals in the northwestern area of Tuscany (Italy). Of these, 44 patients had a confirmed BSI caused by an NDM-producing strain, whereas 3 patients had an infection caused by strains with a different mechanism of carbapenem resistance (these cases were excluded from further analysis); the data of 4 patients were unavailable. Thus, the final cohort included 40 patients, while 35 strains were available for full microbiological and molecular characterization.

The isolates from bacteremia were identified as Klebsiella pneumoniae (31 patients) and Escherichia coli (4 patients). Preliminary characterization by WGS revealed that most K. pneumoniae isolates belonged to the same clonal lineage, namely sequence type (ST) 147 (n = 30, 96.8%), with a singleton of ST307. The E. coli isolates belonged to 2 different clonal lineages, ST8 (n = 2) and ST2 (n = 2). Among the K. pneumoniae isolates, all those of ST147 carried the bla_NDM-1 gene, whereas those of ST307 carried bla_NDM-5 (as did the 4 E. coli isolates). All K. pneumoniae isolates and the 2 ST8 E. coli additionally carried the bla_CTX-M-15 extended-spectrum β-lactamase (ESBL) gene. A clonal analysis, based on single nucleotide polymorphisms (SNPs) evaluated on the core genome, revealed that all the ST147 isolates were closely related to each other (SNP range, 0–25), strongly suggesting that the outbreak was due to clonal expansion of a single NDM-1-producing K. pneumoniae strain.

Susceptibility patterns determined in vitro for the 35 characterized isolates are detailed in Table 1. All K. pneumoniae strains (n = 31) were resistant to expanded-spectrum cephalosporins, carbapenems, and β-lactamase inhibitor combinations, while they were susceptible to aztreonam (ATM)-avibactam (AVI; MIC ≤ 1 mg/L). The E. coli isolates (n = 4) showed a similar broad spectrum of β-lactam resistance, but they were more frequently resistant to ATM-AVI, with some differences between the ST8 (n = 2) and ST2 (n = 2) isolates: Whereas the former showed high-level resistance to ATM (MIC > 32 mg/L) and were resistant to ATM-AVI (MIC, 8 mg/L), the latter showed a lower resistance level to aztreonam (MICs, 2 and 8 mg/L) and were not frankly resistant to the AZT-AVI combination (MICs, 2 and 4 mg/L, respectively). Beyond AZT-AVI, the most active antibiotics were colistin and fosfomycin (susceptibility rate of 91.4% and 80.6%, respectively). Almost all strains (97.2%) were resistant to aminoglycosides.

Table 1.

In Vitro Susceptibilities of 35 NDM-Producing Isolates Collected From Patients With BSI Admitted to 9 Hospitals Across Tuscany in 2018–2019

	MIC, mg/L	Susceptibility Rates, %
Bacterial Species (Isolate No.) and Antimicrobial Agent Tested	Range	S	I	R
Klebsiella pneumoniae (n = 31)
Ceftriaxone	>4	-	-	100
Ceftazidime	>64	-	-	100
Cefepime	>16	-	-	100
PIP-TAZ	>128/4	-	-	100
Ciprofloxacin	>1	-	-	100
Levofloxacin	>8	-	-	100
Amikacin	>32	-	-	100
Gentamycin	≤0.5 to >8	3.2	-	96.8
Meropenem	4 to 64	-	3.2	96.8
Ertapenem	1 to >2	-	-	100
TMP-SMX	≤1/19 to >8/152	3.2	-	96.8
Tigecycline	≤0.25 to >4	80.6	-	19.4
Colistin	≤0.5 to >8	90.3	-	9.7
Aztreonam	>32	-	-	100
Fosfomycina	4 to 64	80.6	-	19.4
CLZ-TAZ	>64/4	-	-	100
CAZ-AVI	>32	-	-	100
MER-VAB	4 to >64	3.2	-	96.8
AZT-AVI	≤0.25 to 1	100	-	-
Escherichia coli (N = 4)
Ceftriaxone	>4	-	-	100
Ceftazidime	>64	-	-	100
Cefepime	>16	-	-	100
PIP-TAZ	>128/4	-	-	100
Ciprofloxacin	>1	-	-	100
Levofloxacin	>8	-	-	100
Amikacin	>32	-	-	100
Gentamycin	>8	-	-	100
Meropenem	64 to >64	-	-	100
Ertapenem	>2	-	-	100
TMP-SMX	>8/152	-	-	100
Tigecycline	≤0.25 to 1	75	-	25
Colistin	≤0.5 to 1	100	-	-
Aztreonam	2 to 32	-	25	75
Fosfomycina	≤8 to 64	75	-	25
CLZ-TAZ	>64/4	-	-	100
CAZ-AVI	>32	-	-	100
MER-VAB	64 to >64	-	-	100
AZT-AVI	2 to 8	-	50	50

Abbreviations: AZT-AVI, aztreonam-avibactam; CAZ-AVI, ceftazidime-avibactam; CLZ-TAZ, ceftolozane-tazobactam; MER-VAB, meropenem-vaborbactam; PIP-TAZ, piperacillin-tazobactam; TMP-SMX, trimethoprim-sulfamethoxazole.

^aMIC for fosfomycin determined by agar dilution.

The clinical characteristics of the patient population and comparison between survivors and nonsurvivors are illustrated in Table 2. Half of the bacteremic patients were cared for in medical wards, and 47.5% had malignancy. The majority of patients (67.5%) had previous documented rectal NDM colonization. The overall 30-day mortality rate was 42.5%. Septic shock occurred in 32.5% of patients. The median age and Charlson comorbidity index score were significantly higher in nonsurviving patients than in those who survived. At the same time, primary bacteremia (unknown focus) was more common among patients who died, whereas central venous catheter (CVC)–related bacteremia was more common among survivors.

Table 2.

NDM-Producing Enterobacterales BSI: Comparison Between Survivors and Nonsurvivors (Tuscany, Italy, 2018–2019)

	All Patients (n = 40)	Survivors (n = 23)	Nonsurvivors (n = 17)	P
Age, median (IQR), y	70.5 (55.25–77.75)	63 (48–76)	74 (67–82.5)	.018
Male sex	28 (70)	17 (73.9)	11 (64.7)	.530
Ward of hospitalization
Medical wards	20 (50)	12 (52.2)	8 (47.1)	.687
ICU wards	13 (32.5)	8 (34.8)	5 (29.4)
Surgery	7 (17.5)	3 (13)	4 (23.5)
Comorbidities
Cardiovascular disease	20 (50)	9 (39.1)	11 (64.7)	.110
Malignancy	19 (47.5)	9 (39.1)	10 (58.8)	.218
COPD	12 (30)	5 (21.7)	7 (41.2)	.185
Diabetes	12 (30)	8 (34.8)	4 (23.5)	.443
Chronic renal diseases	7 (17.5)	4 (17.4)	3 (17.6)	.983
Charlson comorbidity index, median (IQR)	4 (2–7)	3 (0–5)	6 (3–8.5)	.010
Immunosuppressive therapy,a previous 30 d	15 (37.5)	7 (30.4)	8 (47.1)	.283
Source of infection
Unknown	10 (25)	3 (13)	7 (41.2)	.067
Urinary tract	10 (25)	7 (30.4)	3 (17.6)
Intravascular device	9 (22.5)	8 (34.8)	1 (5.9)
ABSSSI	6 (15)	3 (13)	3 (17.6)
Respiratory tract	3 (7.5)	2 (8.7)	1 (5.9)
Intra-abdominal	2 (5)	0	2 (11.8)
NDM-producing strain rectal colonization	27 (67.5)	17 (89.5)	10 (62.5)	.058
Source control	20 (50)	12 (52.2)	8 (47.1)	.749
SOFA score, median (IQR)	4 (2–6)	4 (2–6)	5 (3–6.5)	.229
Length of hospital stay, median (IQR), d	23 (13–38)	26.5 (17.25–41.75)	19 (10–33)	.187
Septic shock	13 (32.5)	7 (30.4)	6 (35.3)	.746
Antibiotic regimens
No in vitro active antibiotic therapy	14 (35)	5 (21.7)	9 (52.9)	.054
CAZ-AVI + ATM	12 (30)	10 (43.5)	2 (11.8)
Colistin-based regimenb	9 (22.5)	4 (17.4)	5 (29.4)
Othersc	5 (12.5)	4 (17.4)	1 (5.9)

Data are presented as No. (%), unless otherwise indicated. P values < .05 (indicating statistical significance) were reported in bold.

Abbreviations: ABSSSI, acute bacterial skin and skin structures infection; ATM, aztreonam; CAZ-AVI, ceftazidime-avibactam; COPD, chronic obstructive pulmonary disease; CVC, central venous catheter; ICU, intensive care unit; IQR, interquartile range.

^aIncluding steroidal and nonsteroidal immunosuppressive therapy.

^bColistin was used in combination with meropenem (4 cases), fosfomycin (3 cases), tigecycline (1 case), AZT + piperacillin-tazobactam (1 case).

^cOther therapies include: 1 patient treated with fosfomycin + tigecycline + amikacin (death); 1 patient treated with meropenem + tigecycline + fosfomycin; 1 patient treated with fosfomycin alone; 1 patient treated with tigecycline + meropenem; 1 patient treated with tigecycline alone.

The most common regimens were ceftazidime-avibactam (CAZ-AVI) plus ATM (30%), followed by a colistin-based regimen (22.5%). In patients treated with CAZ-AVI plus ATM, CAZ-AVI was administered at the dosage of 2.5 g every 8 hours (in 3 patients in continuous infusion) and ATM at the dosage of 2 g every 8 hours. Dosages were eventually adjusted according to renal function. Among 26 patients (65%) receiving an antibiotic regimen showing in vitro activity against the isolated pathogen, the mortality rate was 30.8% (8/26), whereas patients who did not receive any active antibiotic therapy (n = 14, 35%) had a mortality rate of 64.3% (9/14, P = .041). Two patients treated with the combination of CAZ-AVI plus ATM died: 1 patient had multiple comorbidities, a severe cardiac dysfunction, and developed septic shock at time of diagnosis, while the other was an intensive care unit (ICU) patient with acute cerebral hemorrhage who was unresponsive to surgical therapy. In no cases was development of resistance to ATM-AVI detected in vivo. Time to death (IQR) was 8 (1–18.5) days in those who received in vitro active therapy, compared with 6 (1–8) days in patients who did not (P = .481). Cox regression analysis identified Charlson comorbidity index score (hazard ratio [HR], 1.221; 95% confidence interval [CI], 1.022–1.459; P = .028) and age (HR, 1.040; 95% CI, 1.012–1.079; P = .038) as factors independently associated with 30-day mortality.

Discussion

This is the first European report on a relatively large group of patients with bacteremia caused by NDM-producing Enterobacterales.

A preliminary genomic characterization of isolates revealed that the majority of episodes of BSI were caused by an NDM-1-producing strain of K. pneumoniae belonging to ST147. Clonal expansion, therefore, appears to be the major mechanism underlying the large outbreak of NDM-producing strains ongoing in this area, although further characterization of other isolates will be necessary to confirm the contribution of this clone and the diversity of circulating strains.

In the Tuscany cluster, BSIs caused by NDM-producing strains involved patients frequently cared for in medical wards, with multiple comorbidities and severe underlying diseases (such as cancer), and were associated with a high 30-day mortality rate. Mortality was greatly influenced by advanced age and by comorbid conditions, calculated by the Charlson comorbidity index. Moreover, mortality was significantly higher among patients not receiving an in vitro active antibiotic therapy. This latter finding underlines the importance of early detection of the molecular mechanism underlying the carbapenem resistance phenotype; it is noteworthy that some patients who died, according to previous local epidemiological data showing a predominance of KPC-producing strains, were empirically treated with CAZ-AVI alone or in combination with aminoglycosides [7].

Data about treatment of BSI caused by NDM-producing Enterobacterales are very limited [8, 9]. Treatment regimens used in previous reports include colistin alone or in combination with aminoglycosides or meropenem, and the association of fosfomycin with meropenem [10–13]. In our series, all strains were in vitro resistant to ATM, while full susceptibility was restored when combination ATM + AVI was used. The efficacy of this combination relies on the activity that the monobactam ATM typically retains against MBLs but not against ESBLs (ie, all strains in our study produced CTX-M-15). In vitro studies demonstrated a synergistic effect of the combination of CAZ-AVI plus AZT against NDM-producing isolates [14, 15], and a clinical study including 5 patients with bacteremia caused by NDM-producing Enterobacterales showed that combination therapy with CAZ-AVI plus AZT is an effective therapeutic option [16]. Of interest, compared with other regimens, in our study surviving patients were more frequently treated with the combination of CAZ-AVI plus AZT. Although this finding was not statistically significant at Cox regression analysis, this is a potentially interesting finding that should be confirmed in larger, multicentric cohorts.

We used CAZ-AVI at a dosage of 2.5 g every 8 hours and ATM at a dosage of 2 g every 8 hours in all cases (with dose adjustment according to renal function). Recently, therapeutic drug monitoring of CAZ-AVI and ATM performed in a child with BSI caused by KPC and NDM-producing Enterobacter spp. showed adequate serum concentrations of the combination CAZ-AVI plus ATM, both used at a dosage of 50 mg/kg every 8 hours [17]. If we translated this dosage in adult patients, the dose would have to be increased to 3–3.5 g every 8 hours for both ATM and CAZ-AVI. Overall, we observed a good clinical response of this combination therapy at standard dosages. Further pharmacokinetic studies are needed to assess the optimal dosage of ATM and CAZ-AVI in patients with BSI due to NDM-producing strains.

After the identification of the outbreak, several rapid actions were implemented to contain the spread of NDM-producing strains across health care facilities in the northwestern area of Tuscany. The measures included (i) mandatory screening by rapid molecular tests on admission and during hospitalization for all at-risk patients in medical wards and for all patients in the ICU, oncology, oncohematology, transplant unit, cardiac surgery, infectious diseases, acute rehabilitation; (ii) adoption of contact precautions in all cases; (iii) collection of data in a regional database and obligatory notification of all cases to a centralized laboratory; (iv) development of practical guidelines for the clinical management of NDM + cases; (v) dedicated medical and nursing staff and, in the tertiary hospital, identification of a dedicated ward for patients with colonization/infection by NDM-producing Enterobacterales; (vi) educational meetings for all the professionals involved [18].

Our study has some limitations: (i) the number of BSIs is limited, but it is the largest cohort described in Europe; (ii) the multivariate analysis might be affected by the low number of cases; (iii) a high proportion of patients did not receive any active antibiotic therapy, which may have influenced the mortality rate. Nevertheless, this finding reflects clinical practice and the difficulties associated with early identification and appropriate treatment of infections caused by carbapenem-resistant Enterobacterales with multiple mechanisms of resistance.

In conclusion, the epidemiology of carbapenemase-producing Enterobacterales strains is changing over time, and new clones carrying new molecular mechanisms of resistance are emerging in some countries such as Italy. Bacteremia mediated by NDM-producing Enterobacterales is a highly lethal condition and requires prompt recognition and treatment. Although the combination of CAZ-AVI plus AZT appears to be a good therapeutic option in patients with NDM-producing bacteremia, the optimal regimen for the treatment of this infection is not defined and further studies are needed.