Real-World Data on the Effectiveness and Use of Intravenous Fosfomycin for the Treatment of Difficult-to-Treat Infections Caused by Carbapenem-Resistant Gram-Negative Bacteria—A Subgroup Analysis from the FORTRESS Study

Abstract

Introduction

This subgroup analysis of the FORTRESS study aimed to evaluate clinical and microbiological outcomes, treatment patterns, and safety of intravenous fosfomycin (FOS)-containing regimens for the treatment of infections due to carbapenem-resistant (CR) Gram-negative bacteria in a real-world setting.

Methods

Interim data from patients treated with FOS for infections due to CR pathogens were analyzed from the ongoing prospective, multicenter, multinational, observational FORTRESS study. Key outcomes included patient demographics, clinical and infection characteristics at baseline, treatment patterns, and indications of FOS use, as well as clinical, microbiological, and safety outcomes. Exploratory Firth univariate and multivariate logistic regression were performed to identify factors associated with successful clinical response.

Results

The subgroup included 161 patients (median age 60 years, 30.4% female, median APACHE II score 15), of whom 59.6% required intensive care, and 41.6% had sepsis at baseline. The most common indications for FOS therapy were hospital-acquired/ventilator-associated pneumonia (28.0%), bacteremia/sepsis (26.1%), and complicated urinary tract infections (24.8%). Infections were predominantly caused by Klebsiella pneumoniae (62.7%), Pseudomonas aeruginosa (28.6%), and Acinetobacter baumannii (17.4%). Carbapenem resistance was mainly mediated by Klebsiella pneumoniae carbapenemase (KPC) and New Delhi metallo-β-lactamase (NDM). In the majority of patients, FOS was part of a combination regimen (91.3%), with ceftazidime–avibactam being the most frequently used partner antibiotic. Overall, clinical success was achieved in 79.4% of patients, while the successful clinical response and microbiological cure rates were 87.5% and 81.3% at the end of FOS treatment, respectively. All-cause in-hospital mortality was 9.9%. Electrolyte imbalances were the most commonly reported adverse drug reactions, but they were mostly not treatment-limiting.

Conclusions

The real-world data of this subgroup analysis suggest that FOS-containing regimens were associated with generally favorable clinical and microbiological outcomes and acceptable tolerability in patients with severe infections due to CR Gram-negative bacteria.

Supplementary Information

The online version contains Supplementary Material available at https://doi.org/10.1007/s40121-026-01413-5.

Key Summary Points

Why carry out this study?

Infections due to carbapenem-resistant (CR) Gram-negative bacteria are difficult to treat owing to increasing antimicrobial resistance and limited treatment options.

Intravenous fosfomycin (FOS) shows in vitro activity and synergistic potential with other antibiotics against many clinically relevant Gram-negative pathogens and various carbapenem resistance-conferring mechanisms, but prospective real-world data on the use of FOS in patients infected with CR pathogens remain limited.

This subgroup analysis of the ongoing prospective FORTRESS study aimed to describe treatment patterns, clinical and microbiological outcomes, and safety of FOS in patients with infections due to CR bacteria from clinical practice, including a detailed analysis of underlying carbapenem resistance mechanisms of causative agents.

What was learned from the study?

FOS was predominantly used as part of combination regimens (> 90%) for the treatment of various infection types caused by Gram-negative bacteria, including Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii. K. pneumoniae carbapenemase (KPC) and metallo-β-lactamases (MBL) were the most frequently identified resistance mechanisms.

Overall, a clinical success rate of 79.4% at the end of FOS treatment (EOT) was achieved in severely ill patients with infections due to CR bacteria. Successful clinical response and microbiological cure at EOT were 87.5% and 81.3%, respectively. Outcomes were particularly favorable in (KPC) K. pneumoniae infections.

The all-cause in-hospital mortality was 9.9%.

Introduction

Infections caused by carbapenem-resistant (CR) Gram-negative bacteria pose a major challenge to public health worldwide owing to limited therapeutic options and high mortality rates [1, 2]. Among them, carbapenem-resistant Enterobacterales (CRE), Pseudomonas aeruginosa (CRPA), and Acinetobacter baumannii (CRAB) represent one of the clinically most concerning agents and were categorized as critical or high-priority pathogens by the World Health Organization (WHO) [3]. According to the results of a recently published systematic analysis, deaths attributable to CR Gram-negative bacteria have almost doubled between 1990 and 2021 [4].

One of the main mechanisms conferring resistance to carbapenems is the production of carbapenemases, including enzymes of Ambler classes A, B, and D [5]. While Klebsiella pneumoniae carbapenemase (KPC) represents the most prevalent carbapenemase among CRE [6], metallo-β-lactamases (MBLs), e.g., New Delhi metallo-β-lactamase (NDM), have become more frequent in recent years [7], partly explainable by the introduction and selection pressure of newer β-lactam/β-lactamase inhibitors such as ceftazidime–avibactam [8, 9].

The challenging therapeutic landscape, particularly for infections caused by MBL-producing Gram-negative bacteria with only few options, such as aztreonam–avibactam, and the emerging resistance even against newer antibiotics such as cefiderocol or ceftazidime–avibactam [10, 11], emphasizes the need for a broad spectrum of effective agents against CR bacteria. In this context, intravenous fosfomycin (FOS), which belongs to the class of phosphonic acid derivatives, has gained renewed interest for the treatment of severe infections caused by difficult-to-treat Gram-negative bacteria in recent years.

FOS exhibits a unique mode of action by inhibiting the first committed step of bacterial cell wall synthesis [12, 13]. Due to its broad-spectrum activity covering clinically relevant Gram-positive and Gram-negative pathogens, including various carbapenem resistance-conferring mechanisms such as KPC and MBL [13–17] enzyme production, and synergistic interactions with most antibiotic classes [18], FOS has been mainly used as an adjunct to other antibiotics in the targeted or empirical treatment of various infection types, particularly in those caused by CRE or difficult-to-treat/carbapenem-resistant P. aeruginosa [15]. Recent real-life data have also indicated a possible role of FOS-containing combinations in the treatment of A. baumannii infections [19–24].

The ongoing FORTRESS study evaluates the effectiveness and safety of FOS in a real-world setting across various countries. A recently published comprehensive interim analysis of the FORTRESS study reported clinical success in 75.3% of patients in the overall cohort (539 of 716 patients), while the success rate was 81.8% in infections due to CR pathogens (63 of 77 patients) [25].

The present report constitutes a dedicated subgroup analysis of the FORTRESS study, extending the previously published interim data in both sample size and scope by providing a focused examination of the clinical and microbiological characteristics of bacterial infections caused by CR bacteria, including a detailed assessment of underlying resistance mechanisms, as well as a comprehensive evaluation of clinical and microbiological outcomes, treatment patterns, and safety of FOS in this subgroup.

Methods

The design, inclusion and exclusion criteria, and data collection of the FORTRESS study have been described in detail by Bodmann et al. [25].

The study was conducted in accordance with the Helsinki Declaration of 1964 (and its amendments) and was approved by all ethics committees or other authorities according to national/local requirements. Written informed consent was obtained from each patient or the patient’s legally acceptable representative before any study-specific activity was performed. Full details of the ethical approval have been previously reported [25].

Sample Size

The FORTRESS study aims to enroll a total of 1500 participants across multiple study sites in various countries, including sites in Germany, Italy, Greece, Austria, and the UK at the time of the interim analysis. Recruitment commenced in January 2017. For the current study, the database lock was June 2025, with a total of 1019 finalized patients overall comprising the full analysis (FA)/safety-evaluable (SE) population. Of these, patients with documented infections due to carbapenem-resistant bacteria were included in this interim subgroup analysis. Parts of the current dataset have been published in Bodmann et al. [25].

Definitions and Outcomes

Definitions and outcomes have been described previously [25]. Briefly, the primary endpoint was clinical success, a composite endpoint defined as either clinical cure or clinical improvement requiring concurrent microbiological cure, analyzed at end of FOS treatment (EOT). Microbiological cure was defined as either the elimination of the relevant pathogen(s) at the relevant site(s) of infection or assumed pathogen elimination, when the attending physicians explicitly documented that no follow-up sample was indicated or available due to a confirmed sufficient clinical response.

Secondary endpoints for the current report included successful clinical response (additional endpoint as described in the statistical analysis plan, defined as either resolution or partial resolution of signs and symptoms), clinical failure (assessed at the discretion of the attending physician), microbiological cure at EOT, all-cause in-hospital mortality, and safety (including adverse events [AEs], serious AEs [SAEs], adverse drug reactions [ADRs; defined as possibly related to FOS treatment], serious ADRs [SADRs], and deaths). Deaths documented as part of the safety evaluation may have occurred after the end of hospitalization.

Pretreatment at baseline was defined as intravenous antibiotic therapy during the current hospital stay for the current infection, prior to the start of FOS therapy. Antimicrobial susceptibility testing (AST) and the characterization of resistance mechanisms were performed as part of routine practice according to local guidelines and were therefore not systematically obtained in all patients.

Statistical Analysis

Descriptive statistics (mean, standard deviation, median, interquartile ranges [IQR], and minimum and maximum) were performed for continuous variables. Frequencies and percentages are given for categorical parameters. All statistical analyses were carried out by means of SAS software (version 9.4). Figures were prepared employing GraphPad Prism (version 10.6.1).

Univariate and multivariate Firth’s bias-reduced logistic regression analysis employing R software was performed to investigate the relationship between relevant baseline variables and successful clinical response [26, 27]. The baseline variables for the univariate analysis included age, sex, infection type, intensive care, selected comorbidities, requirement of mechanical ventilation, presence of bacteremia and/or sepsis/septic shock, the use of prior antibiotics for the current infection, FOS treatment characteristics (including time to FOS, dosing, and empirical use), causative agent, mechanisms of resistance, and combination partners. Results of the univariate logistic regression analysis were reported as odds ratio (OR) with 95% confidence interval (CI), and p  < 0.05 was considered statistically significant. For the multivariate logistic regression, only variables with p ≤ 0.1 in the univariate model were included for further analysis.

Results

Patient Disposition

Of the 1019 patients enrolled overall in the FORTRESS study at the time of this interim analysis, 161 received FOS for the treatment of infections caused by CR Gram-negative bacteria. The majority of patients were enrolled in Italy (n = 106) and Greece (n = 36), while 15 and 4 patients were enrolled in Germany and the UK, respectively.

Baseline Demographics and Clinical Characteristics

Baseline demographic and clinical characteristics of the subpopulation are summarized in Table 1. The median age was 60.0 years (IQR: 48.0–70.0), and 30.4% of patients were female (n = 49). Approximately 60% of patients (n = 96) were treated in intensive care units (ICU) at baseline, while 36.6% were mechanically ventilated (n = 59; 36.6%). The median Acute Physiology and Chronic Health Evaluation II (APACHE II) and Sequential Organ Failure Assessment (SOFA) scores in evaluable patients were 15 and 8.5, respectively, and more than 40% of patients had sepsis or septic shock, indicative of a generally severely ill patient population. Bacteremia was present in 29.8% cases, most of which originated from the urinary tract or a pneumogenic focus. Almost every patient had a comorbidity, with cardiovascular, respiratory, and diabetes mellitus being the most common. Notably, 25.5% presented with an electrolyte disorder before starting FOS treatment, and 26.1% were considered immunocompromised.

Table 1.

Demographic and baseline characteristics

Demography/clinical characteristic	Patients (n = 161)
Age (years), median (IQR)	60.0 (48.0–70.0)
Sex (female), n (%)	49 (30.4)
Patients in ICU, n (%)	96 (59.6)
APACHE II score, median (IQR) (n = 39)	15 (13–20)
SOFA score, median (IQR) (n = 10)	8.5 (6.0–10.0)
Bacteremia/BSI, n (%)a	48 (29.8)
Sepsis, n (%)	67 (41.6)
Septic shock, n (%)	13 (8.1)
Mechanical ventilation, n (%)	59 (36.6)
Immunocompromisedb	42 (26.1)
Creatinine clearance (mL/min), median (IQR) (n = 112)	82.5 (42.8–120.0)
Creatinine clearance < 60 mL/min, n (%)	39 (24.2)
Concomitant treatment (any), n (%)	132 (82.0)
Comorbidities (any), n (%)c	149 (92.5)
Cardiovascular	85 (52.8)
Hypertension	64 (39.8)
Coronary heart disease	17 (10.6)
Congestive heart failure	11 (6.8)
Oncologic	24 (14.9)
Cirrhosis	9 (5.6)
Respiratory	49 (30.4)
COPD	24 (14.9)
Bronchiectasis	3 (1.9)
Diabetes mellitus	49 (30.4)
Renal insufficiency	29 (18.0)
Renal replacement therapy	7 (4.3)
Electrolyte disorders	41 (25.5)
Hypokalemia	21 (13.0)
Hypernatremia	12 (7.5)
Hyponatremia	11 (6.8)

^aUrinary tract infection (n = 14), pneumonia (n = 10), catheter-associated infection (n = 6), intra-abdominal infection (n = 6), skin/soft tissue infection (n = 3), primary bacteremia (n = 3), unknown/not further specified (n = 6)

^bPatients receiving immunosuppressive/anticancer chemotherapy and/or patients with an immunosuppressive comorbidity

^cMultiple entries per patient possible

APACHE II Acute Physiology and Chronic Health Evaluation II, BSI bloodstream infection, COPD chronic obstructive pulmonary disease, ICU intensive care unit, IQR interquartile range

Indications of FOS Use and Microbiological Findings

The main infection sites for FOS use were hospital-acquired/ventilator-associated pneumonia (HAP/VAP) (28.0%), bacteremia/sepsis (26.1%), and complicated urinary tract infections (cUTI) (24.8%). Less frequently treated infection types included complicated skin and soft tissue infections (cSSTI) (8.7%) and other infection sites, such as bone and joint infections (BJI) (12.4%) (Table 2). However, country-specific differences could be observed (Supplementary Table S1). Overall, infections were predominantly caused by Enterobacterales, which were involved in 72.7% of cases, with K. pneumoniae being the most frequently isolated pathogen (n = 101). Non-fermenting bacteria were isolated in 42.2% of all patients, primarily P. aeruginosa (n = 46) and A. baumannii (n = 28) (Table 2). While K. pneumoniae was the predominant causative agent in bacteremia/sepsis and cUTI (> 50%), non-fermenters were involved in most HAP/VAP cases (42.2% A. baumannii and 33.3% P. aeruginosa, respectively) (Supplementary Table S2). Causative agents stratified by country are summarized in Supplementary Table S3 (Supplementary Material). Overall, the majority of infections were caused by one pathogen (68.9%), while 31.1% of patients had polymicrobial infections. In this context, Gram-positive pathogens were involved in mixed infections, mainly in cSSTI cases, while 1.9% of patients had a concomitant fungal infection.

Table 2.

Indications, causative agents, and identified resistance mechanisms

Characteristics	Patients (n = 161)
Primary infection site, n (%)
HAP/VAP	45 (28.0)
Bacteremia/sepsis	42 (26.1)
cUTI	40 (24.8)
cSSTI	14 (8.7)
Othera	20 (12.4)
Type of infection, n (%)
Monomicrobial	111 (68.9)
Polymicrobialb	50 (31.1)
MDR	161 (100)
Causative bacteria, n (%)c
Enterobacterales	117 (72.7)
Klebsiella pneumoniaed	101 (62.7)
Escherichia coli	8 (5.0)
Proteus spp.	8 (5.0)
Enterobacter spp.	6 (3.7)
Serratia spp.	2 (1.2)
Citrobacter spp.	2 (1.2)
Morganella morganii	1 (0.6)
Non-fermenters	68 (42.2)
Pseudomonas aeruginosa	46 (28.6)
Acinetobacter baumanniie	28 (17.4)
Stenotrophomonas maltophilia	3 (1.9)
Achromobacter xylosoxidans	1 (0.6)
Gram-positive pathogensf	22 (13.7)
Otherg	3 (1.9)
Mechanisms of resistance, n (%)c
Ambler class A	71 (44.1)
KPC	63 (39.1)
CTX-M	12 (7.5)
Ambler class B	35 (21.7)
NDM	23 (14.3)
VIM	10 (6.2)
IMP	1 (0.6)
GIM	1 (0.6)
Unknown MBL	1 (0.6)
Ambler class C	1 (0.6)
AmpC	1 (0.6)
Ambler class D	5 (3.1)
OXA-48	4 (2.5)
OXA-23	1 (0.6)
Double carbapenemase	6 (3.7)
Carbapenemase + ESBL	19 (11.8)
Not reported/unknown	69 (42.9)

^aPatient-level: endocarditis (n = 1), central nervous system infection (n = 1), bone and joint infection (n = 8; including long bone osteomyelitis, prosthetic joint infection, and diabetic foot osteomyelitis), complicated intra-abdominal infection (n = 2), not further specified (n = 5), febrile neutropenia (n = 1), otomastoiditis (n = 1), pleural empyema (n = 1)

^bAt least two different causative agents

^cMultiple entries per patient possible

^dIncludes cases reported as Klebsiella spp.

^eIncludes cases reported as Acinetobacter spp.

^fGram-positive pathogens were identified solely as co-pathogens in mixed infections; patient-level: Enterococcus faecalis (n = 7), E. faecium (n = 4), coagulase-negative Staphylococcus spp. (n = 1), S. capitis (n = 1), S. cohnii (n = 1), S. hominis (n = 2), methicillin-resistant S. aureus (n = 2), methicillin-susceptible S. aureus (n = 4), S. epidermidis (n = 3), Streptococcus agalactiae (n = 2), Bacillus cereus (n = 1), Helcococcus kunzii (n = 1)

^gPatient-level: Candida spp. (n = 3)

AmpC ampicillinase C, cSSTI complicated skin/soft tissue infection, CTX-M cefotaximase Munich, cUTI complicated urinary tract infection, ESBL extended-spectrum β-lactamase, GIM German imipenemase, HAP/VAP hospital-acquired/ventilator-associated pneumonia, IMP imipenemase, KPC Klebsiella pneumoniae carbapenemase, MBL metallo-β-lactamase, MDR multidrug-resistant, NDM New Delhi metallo-β-lactamase, OXA oxacillinase, VIM Verona-integron metallo-β-lactamase

Ambler class A (44.1% of patients) and B carbapenemases (21.7% of patients) were the most frequently identified mechanisms conferring carbapenem resistance, with KPC and NDM being the predominant enzymes, respectively. Double carbapenemase production (3.7% of patients) and Ambler class D carbapenemases (3.1% of patients) were uncommon. In 42.9% of patients, at least one causative agent was identified for which a specific resistance mechanism remained unknown or was not tested for.

Where available, the FOS minimum inhibitory concentrations (MICs) for K. pneumoniae and P. aeruginosa isolates ranged from 2–512 mg/L (MIC₅₀: 32 mg/L; MIC₉₀: 256 mg/L; n = 62) and 2–1024 mg/L (MIC₅₀: 16 mg/L; MIC₉₀: 256 mg/L; n = 14), respectively.

Treatment Characteristics

The treatment patterns of FOS use in this subset of patients are detailed in Table 3. In the majority of patients (91.3%), FOS was administered as part of a combination regimen, while 8.7% received monotherapy (mainly for cUTI). The most frequently used partner antibiotics were ceftazidime–avibactam (36.0%) and colistin (26.1%), followed by meropenem (11.2%), cefiderocol (8.1%), and amikacin (6.8%). In this context, ceftazidime–avibactam was the main backbone used for various infection types caused by K. pneumoniae (Supplementary Table S4), whereas colistin was the primary combination partner in A. baumannii-associated cases (mainly HAP/VAP). On a resistance mechanism level, ceftazidime–avibactam was the preferred option in infections due to KPC-producing bacteria but was also used when MBL producers were involved.

Table 3.

Treatment patterns

Treatment characteristic	Patients (n = 161)
FOS therapy
Combination therapy	147 (91.3)
1 partner antibiotic	106 (65.8)
2 partner antibiotics	26 (16.1)
≥3 partner antibiotics	15 (9.3)
Monotherapy	14 (8.7)
Frequently used partner antibioticsa,b,c
Ceftazidime–avibactam	58 (36.0)
Colistin	42 (26.1)
Meropenem	18 (11.2)
Cefiderocol	13 (8.1)
Amikacin	11 (6.8)
Rationale for FOS usea
Difficult-to-treat infection	93 (57.8)
Nosocomial infection	78 (48.4)
Insufficient effectiveness/treatment failure of pretreatment	31 (19.3)
Favorable tissue penetration	37 (23.0)
Insufficient effectiveness of potential alternatives	20 (12.4)
Line of treatment
1st line	68 (42.2)
2nd line	75 (46.6)
3rd or 4th line	18 (11.2)
Empirical treatment	33 (20.5)
Duration of FOS treatment (days), median (IQR)	13.0 (8.0–17.3)
Estimated time from first (suspected) diagnosis to start of FOS treatment (days), median (IQR)	2.0 (0.3–5.0)
Targeted daily dose (g), median (IQR)	16.0 (15.1–24.0)
Prior antibiotic treatmentd
Prior antibiotics, n (%)	89 (55.3)
Duration of prior antibiotic treatment (days), median (IQR) (n = 89)	5 (3.0–10.0)
1–3 days, n (%)	30 (18.6)
4–7 days, n (%)	25 (15.5)
≥7 days, n (%)	34 (21.1)

^aMultiple entries per patient possible

^bN > 10

^cStart of combination partner maximum 1 day after FOS start

^dFor current infection

FOS intravenous fosfomycin, IQR interquartile range

FOS was mainly used due to the presence of difficult-to-treat infections (57.8%), nosocomial infections (48.4%), its favorable tissue penetration (23.0%), and treatment failure/insufficient effectiveness of prior antibiotics (19.3%). Overall, patients were treated with FOS for a median of 13.0 days (IQR 8.0–17.3), with a median daily dose of 16.0 g (IQR 15.1–24.0). In 42.2% of patients, FOS was used as a first-line option, and around 20% of patients received FOS empirically. Notably, more than half of patients had received pretreatment with other antibiotics for the current infection, with a median duration of prior antibiotic therapy of 5 days (IQR 3.0–10.0) (Table 3).

Clinical and Microbiological Outcome

Overall, the primary composite endpoint (i.e., clinical success) was achieved in 79.4% of patients (127/160), with a successful clinical response in 87.5% (140/160) and a microbiological cure rate of 81.3% (130/160) at the end of FOS treatment (Fig. 1 and Supplementary Table S5). The overall all-cause in-hospital mortality rate was 9.9% (16/161) and was highest in A. baumannii infections (21.4%, 6/28) and those caused by other Enterobacterales (20.0%, 5/25) (Supplementary Table S5). Figures 1 and 2 show the clinical and microbiological outcomes by infection type and causative pathogen, respectively. Clinical success and microbiological cure rates varied by indication and were highest in bacteremia/sepsis (87.8% each) and cUTI (82.5% and 85.0%) cases and lowest in cSSTI (64.3% and 71.4%) (Fig. 1).

Fig. 1.

Clinical and microbiological outcomes by indication. The percentage (%) of patients with clinical success, successful clinical response, and microbiological cure was assessed in the overall population and for different types of infections. Patients may have had concomitant bacteremia

Fig. 2.

Clinical and microbiological outcomes by causative agents. The percentage (%) of patients with clinical success, successful clinical response, and microbiological cure was assessed in the overall population and for the most frequently identified causative pathogens. Patients may have had concomitant bacteremia

Regarding outcomes by causative agent, clinical success rates were 83.0% for patients with infections due to K. pneumoniae, 79.2% for other Enterobacterales, 75.0% for A. baumannii, and 73.9% for P. aeruginosa. Similarly, microbiological cure rates were highest in K. pneumoniae-associated cases, while the rates for other Enterobacterales, A. baumannii, and P. aeruginosa were slightly lower (Fig. 2). Interestingly, in evaluable patients with reported minimum inhibitory concentrations (MIC), clinical success and microbiological cure in K. pneumoniae infections was lower when the MIC of the causative isolate was above the species-specific European Committee on Antimicrobial Susceptibility Testing (EUCAST) epidemiological cut-off (ECOFF) of 128 mg/L (Supplementary Table S5).

When analyzed by underlying resistance mechanisms, successful clinical response rates were similar for infections caused by MBL (91.4%) and KPC (87.1%) producers (Supplementary Table S5). Of note, microbiological cure rates were comparable in patients with infections due to single carbapenemase producers and those producing additional β-lactamases (i.e., another carbapenemase or extended-spectrum β-lactamases [ESBL]).

Supplementary Table S5 (Supplementary Material) summarizes the analysis of outcomes by various clinical and treatment factors. The additional analyses revealed that clinical outcomes were lower in patients with sepsis or septic shock and when FOS was used as monotherapy in the treatment of cUTI, while there was no difference between patients treated empirically or targeted with FOS. Among backbone antibiotics used, clinical success rates were 84.6% with cefiderocol, 84.2% with ceftazidime–avibactam, 83.3% with meropenem, 78.6% with colistin, and 63.6% with amikacin.

Univariate and Multivariate Regression Analysis

Potential factors associated with successful clinical response were explored by univariate and multivariate Firth’s bias-reduced logistic regression analysis (Table 4). While most variables tested in the univariate analysis were not significant covariates, including age, sex, mechanical ventilation, and empirical treatment, patients with an oncologic comorbidity had a significant association with a reduced probability of successful clinical response (OR 0.316, 95% CI 0.113–0.953). In contrast, an immunocompromised status was the only variable associated with significantly better outcome (OR 5.239, 95% CI 1.265–48.322). Other factors showing a trend toward better outcomes were the use of ceftazidime–avibactam as combination partner (OR 2.242, 95% CI 0.805–7.573) and infections caused by K. pneumoniae (OR 2.238, 95% CI 0.886–5.775), while the presence of sepsis (OR 0.428, 95% CI 0.163–1.080) and A. baumannii infections (OR 0.424, 95% CI 0.155–1.253) had a tendentially negative effect on successful clinical response. In the subsequent multivariate analysis including five variables with p ≤ 0.1 from the univariate models, baseline sepsis (OR 0.285, 95% CI 0.093–0.800) and an oncologic comorbidity (OR 0.273, 95% CI 0.088–0.877) were independently associated with a reduced probability of successful clinical response, whereas an immunocompromised status showed a significant positive association (OR 4.575, 95% CI 1.026–44.248).

Table 4.

Exploratory univariate and multivariate Firth’s bias-reduced logistic regression analysis for successful clinical response at EOT

Variable	Univariate analysis (n = 160)			Multivariate analysis (n = 160)
	OR	95% CI	p-Value	OR	95% CI	p-Value
Age (continuous)	0.990	0.959–1.020	0.5141
Age (categorical)
< 60 years	Ref
≥ 60 years	0.874	0.341–2.196	0.7746
Sex (female versus male [ref])	1.300	0.484–3.990	0.6146
Type of infectiona
Monomicrobial (yes versus no [ref])	0.505	0.198–1.308	0.1561
HAP/VAP (yes versus no [ref])	0.868	0.331–2.500	0.7821
Bacteremia/sepsis (yes versus no [ref])	1.878	0.622–7.448	0.2799
cUTI (yes versus no [ref])	1.281	0.453–4.365	0.6561
ICU at baseline (yes versus no [ref])	0.614	0.215–1.587	0.3210
Comorbid condition at baselinea
Oncologic (yes versus no [ref])*	0.316	0.113–0.953	0.0413	0.273	0.088–0.877	0.0301
Immunocompromised (yes versus no [ref])*	5.239	1.265–48.322	0.0189	4.575	1.026–44.248	0.0457
Electrolyte imbalance (yes versus no [ref])	0.751	0.285–2.170	0.5794
Mechanical ventilation (yes versus no [ref])	0.675	0.267–1.740	0.4091
Presence of sepsis at baseline (yes versus no [ref])*	0.428	0.163–1.080	0.0724	0.285	0.093–0.800	0.0167
Duration of prior antibiotics (continuous)b	1.123	0.976–1.478	0.1645
FOS treatment characteristics
Time to FOS treatment (continuous)*	0.973	0.937–1.002	0.0634	0.973	0.934–1.004	0.0829
Time to FOS treatment (categorical)
Administration within 72 h from diagnosis to start of FOS	Ref
Administration after 72 h from diagnosis to start of FOS	0.786	0.312–2.020	0.6101
Dosing (> 16 g versus ≤ 16 g [ref])	0.821	0.324–2.163	0.6807
Empirical (yes versus no [ref])	1.380	0.452–5.512	0.5935
Causative agent at baselinea
K. pneumoniae (yes versus no [ref])*	2.238	0.886–5.775	0.0878	2.606	0.930–7.657	0.0684
P. aeruginosa (yes versus no [ref])	0.552	0.216–1.468	0.2270
A. baumannii (yes versus no [ref])	0.424	0.155–1.253	0.1159
Resistance mechanisma
MBL-positive (yes versus no [ref])	1.498	0.492–5.969	0.4993
KPC-positive (yes versus no [ref])	0.927	0.367–2.438	0.8732
Combination partner at baselinea
Ceftazidime–avibactam (yes versus no [ref])	2.242	0.805–7.573	0.1272
Colistin (yes versus no [ref])	1.377	0.488–4.687	0.5618
Meropenem (yes versus no [ref])	0.618	0.190–2.551	0.4709

Bold values indicate statistical significance (p < 0.05)

^aMultiple entries per patient possible

^bN = 89

^*Variables were included in the multivariate analysis. The selection was based on the results of the univariate analysis (p ≤ 0.1)

CP carbapenemase, cUTI complicated urinary tract infection, ESBL extended-spectrum β-lactamase, FOS intravenous fosfomycin, HAP/VAP hospital-acquired/ventilator-associated pneumonia, ICU intensive care unit, KPC Klebsiella pneumoniae carbapenemase, MBL metallo-β-lactamase, OXA oxacillinase

Safety

Adverse drug reactions (ADRs) were reported in 51 of 161 patients (31.7%), with hypokalemia (23.0%, 37/161) and hypernatremia (12.4%, 20/161) being the most frequently reported ADRs. Serious ADRs (SADRs) were uncommon (3.1%). Notably, except for one case, electrolyte imbalances were mild to moderate and did not require the discontinuation of FOS. Other ADRs included one case each of vomiting, hyperkalaemia, hypomagnesemia, and suspected development of resistance under FOS therapy. The latter led to discontinuation of FOS. Overall, 17 (10.6%) patients died but none was considered related to FOS treatment.

Discussion

The results of this subgroup analysis from the ongoing FORTRESS study provide valuable insights into the treatment patterns, effectiveness, and safety of FOS for infections caused by CR Gram-negative bacteria, including clinical and microbiological characteristics of these infections. Generally, patients in this cohort were severely ill, as indicated by the high percentage of patients requiring intensive care, frequent presence of sepsis/septic shock, and high APACHE II scores before FOS treatment start. Furthermore, almost all patients had concurrent comorbidities, most notably cardiovascular and electrolyte disorders, diabetes mellitus, and an immunocompromised status.

Overall, clinical success at EOT in this cohort was achieved in 79.4% of patients, while the all-cause in-hospital mortality was 9.9%. In line with previous reports, most patients were infected by Enterobacterales species and P. aeruginosa [28, 29]. However, it is worth noting that FOS-containing regimens were also frequently used in infections involving A. baumannii (17.4%). Among patients infected by A. baumannii (mostly HAP/VAP cases), clinical success was slightly lower, and the mortality rate was higher (21.4%) compared with the overall population. In this context, FOS-containing combination regimens have been increasingly used for infections caused by multidrug-resistant (MDR)/CR A. baumannii in recent years [15, 19–21, 23, 30, 31]. In a retrospective observational study conducted by Oliva and colleagues, patients receiving FOS-containing regimens had a clinical cure rate of 72.7% (8 of 11 patients) [22], which is comparable to the results of the current study. Furthermore, in another retrospective study, the combination of cefiderocol plus FOS was associated with 30-day survival in patients with bacteremic ventilator-associated pneumonia caused by CRAB in patients with coronavirus disease 2019 (COVID-19) [20], while the same combination resulted in a significantly higher survival of patients with severe A. baumannii infections at 14 days compared with colistin plus FOS in an Italian multicenter study [21]. In the subset of patients with CRAB involvement in the present report, colistin was the most frequently used partner antibiotic to FOS (n = 17), followed by cefiderocol (n = 7), with comparable clinical success rates (82.4% versus 85.7%, respectively) but higher mortality in patients treated with FOS combinations including colistin (29.4% versus 14.3%, respectively). However, due to the relatively low patient numbers, these results should be interpreted carefully and warrant further investigations.

Among Enterobacterales, K. pneumoniae was the most frequently isolated pathogen in our cohort, with the majority of patients treated with FOS combination regimens including ceftazidime–avibactam. In these patients, a successful clinical response was achieved in 91.0%, while the microbiological cure rate was 85%. In this context, the combination of FOS with ceftazidime–avibactam is widely used in clinical practice for the treatment of patients infected with KPC-producing K. pneumoniae in some countries [22, 32–35]. In a retrospective matched cohort study, patients with bloodstream infections caused by KPC K. pneumoniae treated with ceftazidime–avibactam/FOS had a numerically higher clinical cure rate and significantly less deaths associated with secondary infections compared with ceftazidime–avibactam alone, while the 30-day mortality was comparable between both arms [32]. In another observational study from Oliva et al., FOS-containing regimens (mostly including ceftazidime–avibactam) achieved clinical cure in 96.2% of patients (25 of 26) and microbiological cure in all evaluable patients (100%, 22 of 22) with severe KPC K. pneumoniae infections [22]. Furthermore, regimens including FOS were associated with a lower 30-day mortality compared with non-FOS-containing antibiotic combination therapies in K. pneumoniae infections (7.7% versus 37.5%, p = 0.072), although they were not statistically significant [22].

While KPC is still one of the most common carbapenemases among CR Enterobacterales (CRE), a global increase in the frequency of MBL enzymes has been observed in recent years [36, 37]. MBL-producing CRE are of great concern, as there are only very limited treatment options available, and infections involving these bacteria are associated with high mortality [2, 7]. With the exception of aztreonam–avibactam, the activity of the available options against MBL-producing CRE is limited, with increasingly nonsusceptibility rates even for novel agents such as cefiderocol [38, 39]. FOS exhibits in vitro activity against MBL-producing Enterobacterales, with most data being available for K. pneumoniae [40–43]. In addition, several studies have demonstrated synergism of FOS with other antibiotics, most notably β-lactams [41–45]. In this regard, FOS plus ceftazidime–avibactam appears to be a particularly promising combination, with synergy rates up to 72.7% against MBL-producing K. pneumoniae found in a recent report [45]. In our study, four patients with infections due to MBL K. pneumoniae treated with a FOS-containing regimen including ceftazidime–avibactam (without additional aztreonam) had a successful clinical response and microbiological cure (4 out of 4), while the overall clinical response and microbiological cure rates in infections involving MBL producers (35 patients) were 91.4% and 85.7%, respectively. While these data indicate a possible role of FOS-containing combination regimens in the treatment of MBL-related infections, clinical data in this indication remain currently limited [46–48]; further clinical evidence is warranted to better define the role of FOS-containing regimens in these difficult-to-treat infections.

In line with published real-world data [29, 49, 50], FOS was primarily used as part of combination regimens (> 90%). This may be attributed not only to its broad synergistic activity with most antibiotic classes [18] but also its primary use in difficult-to-reach/difficult-to-treat infections in severely ill patients [15], where monotherapy may be insufficient. In this context, a recent study found that combining FOS with meropenem had a great impact on pharmacokinetic/pharmacodynamic (PK/PD) targets required for bactericidal effects [51], enabling probability of target attainments of ≥ 90% for a 1-log and 2-log target in Monte Carlo simulations, even at high FOS MICs [52]. Interestingly, incorporating fosfomycin and meropenem MICs in their PK/PD model, Farooq et al. derived combined PK/PD breakpoints for this combination for dosing regimens used in clinical practice [52], which may be useful considering the current lack of FOS clinical breakpoints for most clinically relevant pathogens.

Besides its synergistic properties, the broad-spectrum activity covering both Gram-negative and Gram-positive pathogens makes FOS also a viable option for empirical therapy and polymicrobial infections. In the present study, FOS was used empirically in around 20% of cases. While FOS is often perceived as a second-line or salvage option, 42.2% of patients received FOS as first-line therapy in this cohort, highlighting its importance in the treatment of infections due to CR bacteria.

Although FOS is primarily used as adjunct to the antibiotic backbone for the treatment of critically ill patients, recent studies have demonstrated its efficacy as monotherapy in the treatment of cUTI caused by MDR/ESBL-producing Enterobacterales, primarily E. coli [53–56]. In the present study, eight patients received FOS as monotherapy for the treatment of cUTI episodes due to CR bacteria, mainly involving K. pneumoniae (5 out of 8), with a clinical success rate of 75.0%. In this context, in a recent study by Rodríguez-Gómez and colleagues, FOS was given as monotherapy for the treatment of cUTI episodes caused by K. pneumoniae, with an overall clinical cure rate of 70.2% [57]. Our data may add to the existing evidence and suggest a possible role for FOS alone as a therapeutic option in the treatment of cUTI caused by CR Enterobacterales. However, further data are needed to confirm these preliminary results, as the patient numbers in the present cohort are relatively small.

The most frequently reported ADRs were hypokalemia and hypernatremia, similar to previously published data [25, 58, 59]. Electrolyte imbalances associated with FOS use may be due to the relatively high amount of sodium applied. It is important to note, however, that all except for one case in this subgroup were mild to moderate in intensity and manageable, and a considerable number of patients already presented with electrolyte disorders at baseline. In this regard, some data have indicated that a prolonged infusion time and/or the administration of prophylactic potassium may reduce the incidence of hypokalemia [60, 61]. Moreover, the correct preparation of the FOS solution plays a crucial role, as studies have shown that electrolyte imbalances may be more frequent when sodium-containing solvents are used instead of the recommended water for injections or glucose solution [59, 62]. In accordance with other clinical studies, development of resistance under FOS therapy was rare (0.6%) and may be explained by the predominant use of FOS in combination regimens, as data from numerous in vitro experiments have shown that combinations containing FOS can effectively suppress the emergence of resistant subpopulations [45, 63–66]. However, resistance development under FOS therapy appears to be also uncommon when used as monotherapy [53, 54].

Our study has several strengths, including its prospective, multicenter, multinational design, extensive baseline and follow-up documentation, and thorough monitoring, resulting in a robust and high-quality dataset that allowed for detailed description of the patient population and the use, clinical and microbiological outcomes, and safety of FOS in real-world clinical practice across Europe.

However, we acknowledge several limitations of the current study. First, the lack of a comparator arm prevents a direct comparison to other non-FOS-containing regimens, and the predominant use of FOS as part of combination regimens precludes a direct causal attribution of clinical and microbiological outcomes to FOS alone. Second, results may be subject to potential selection bias owing to the nonsystematic inclusion of patients treated with FOS; country-specific differences in the use of FOS and an overall highly heterogeneous cohort should be considered when interpreting the results. Third, due to the observational nature of our study, availability of certain baseline parameters, such as SOFA and APACHE II scores or resistance profiles of identified pathogens, is limited. Fourth, local practices regarding AST of FOS may have varied across study sites. Lastly, the limited sample size of this subgroup resulted in relatively wide confidence intervals for some tested variables, and the reported associations of both the univariate and multivariate logistic regression should therefore be considered exploratory and interpreted with caution, especially the positive association in immunocompromised patients.

Conclusions

The ongoing FORTRESS study provides valuable insights into the treatment patterns, clinical and microbiological outcomes, and safety of FOS-containing regimens from real-world practice across various countries, including clinical and microbiological characteristics of underlying infections. The results of the present subgroup analysis suggest that FOS-containing regimens were associated with generally favorable outcomes and acceptable tolerability in patients with infections due to CR bacteria, including those caused by Enterobacterales, P. aeruginosa, A. baumannii and irrespective of the underlying carbapenem resistance mechanism. These findings may help inform antibiotic selection in critically ill patients with infections due to CR bacteria. Future prospective studies should aim to compare FOS-containing versus other regimens for specific pathogens and infection sites to further optimize the use of FOS.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Klaus-Friedrich Bodmann contributed to the conception and design of the work; was involved in the acquisition, analysis, and interpretation of the data; and wrote and edited the manuscript. Thomas Borrmann and Christian Mayer provided support with the data analysis. Alessandra Mularoni, Giovanna Russelli, Valentina Galfo, Giusy Tiseo, Francesco Alessandri, Giancarlo Ceccarelli, Eleni Mouloudi, Stavrina Avgeropoulou, Loredana Sarmati, Laura Campogiani, Ivan Gentile, Kai Zacharowski, Stefan Kluge, Dominik Jarczak, Christina Iasonidou, Stefan Hagel, Alessandro Capone, Alessandra Bandera, Daniele Roberto Giacobbe, Annalisa Saracino, Pavlos Myrianthefs, Markos Marangos, Michael Zoller, Jan T. Kielstein, Carlo Tascini, Antonio Cascio, Abhijit M. Bal, Francesca Ferretti, Valerio Del Bono, Despina Hatzilia, Stelios F. Assimakopoulos, Evangelos J. Giamarellos-Bourboulis, Marco Falcone, Mathias W. Pletz, Alessandra Oliva, Matteo Bassetti, Matthias G. Vossen, George Dimopoulos, and Claudio M. Mastroianni were involved in the acquisition of data. All authors revised the draft and approved the final version of the manuscript.

Funding

The FORTRESS study was funded by InfectoPharm Arzneimittel und Consilium GmbH, Heppenheim, Germany. All participating study centers received contractually agreed study fees by InfectoPharm Arzneimittel und Consilium GmbH during the conduct of the present study covering the time expenditure to an amount that was approved as applicable by national relevant authorities/ethics committees. However, no payments or honoraria were made to the authors with respect to the preparation of this manuscript. InfectoPharm also funded the journal’s Rapid Service fee.

Data Availability

The datasets generated and/or analyzed during the current study are not publicly available owing to InfectoPharm’s internal policies. InfectoPharm will provide access to related study documents upon reasonable request from qualified researchers, subject to certain criteria, conditions, and exceptions.

Declarations

Ethical Approval

The study was conducted in accordance with the Helsinki Declaration of 1964 (and its amendments) and was approved by all ethics committees or other authorities according to national/local requirements. Written informed consent was obtained from each patient or the patient’s legally acceptable representative before any study-specific activity was performed. Full details of the ethical approval have been previously reported [25].

Conflict of Interest

The following authors declare potential conflicts of interest unrelated to the submitted work: Klaus-Friedrich Bodmann has received honoraria for lectures from Abbott, Accelerate, Correvio, InfectoPharm, Pfizer, Weber & Weber, and Shionogi. Loredana Sarmati received travel grants from Gilead, Merck, Pfizer, and Advanz. She also received honoraria for lectures from Merck, Gilead, Abbvie, Angelini, AstraZeneca, and GSK. Laura Campogiani received a research grant from Gilead and honoraria for lectures from Menarini, MICOM, and Nadirex. She also reports support for attending meetings from Menarini, Shionogi, and Pfizer. Ivan Gentile reports departmental research grants from Advanz Pharma, as well as consulting fees and honoraria for lectures from MSD, Pfizer, GSK, Basilea, InfectoPharm, Angelini, Shionogi, Advanz Pharma, Abbott, and AstraZeneca. He also participated in advisory boards for MSD, Pfizer, GSK, Basilea, InfectoPharm, Angelini, Shionogi, Advanz Pharma, Abbott, and AstraZeneca. Kai Zacharowski received speaker fees from CSL Behring, Masimo, Pharmacosmos, Boston Scientific, Salus, iSEP, Edwards, Hemosonics, Baxter Deutschland GmbH, and GE Healthcare. Stefan Kluge received research support from Biotest, CytoSorbents, Daiichi Sankyo, and Fresenius Medical Care; consultant fees from ADVITOS, Gilead, and Pfizer; and lecture fees from ADVITOS, bioMérieux GmbH, CSL Behring, Gilead, MSD, Pfizer, and Shionogi. Stefan Hagel has received honoraria from Pfizer, MSD, InfectoPharm, Philips, Advanz Pharma, Beckman Coulter, Shionogi, Thermo Fisher, and Tillotts. He also participated in advisory boards for Advanz Pharma, Shionogi, and Pfizer. Daniele Roberto Giacobbe reports investigator-initiated grants from Pfizer, Shionogi, Advanz Pharma, bioMérieux, Tillotts Pharma, and Menarini; travel support from Pfizer, bioMérieux, and Menarini; and speaker/advisor fees from Menarini, Shionogi, Pfizer, Advanz Pharma, bioMérieux, and MSD. Markos Marangos reports consulting fees from MSD, Menarini, and GSK; honoraria for lectures and presentations from MSD, Menarini, Gilead, Norma, GSK, and Pfizer; and support for attending meetings from MSD, GSK, Gilead, and Pfizer. Stelios F. Assimakopoulos has received honoraria from Pfizer, Gilead, GSK, MSD, Menarini, Angelini, InfectoPharm, Norma, Uni-pharma, and Elpen; participated in advisory boards for Pfizer, MSD, GSK, Gilead, and Menarini; and received research support or grants from Pfizer, Gilead, and Elpen. Evangelos J. Giamarellos-Bourboulis reports honoraria from Abbott Products Operations, bioMérieux, Brahms GmbH, GSK, InflaRx GmbH, Sobi, and Xbiotech Inc.; independent educational grants from Abbott Products Operations, bioMérieux Inc., MSD, UCB, and Swedish Orphan Biovitrum AB; and funding from the Horizon 2020 European Grants ImmunoSep and RISCinCOVID, and the Horizon Health grants EPIC-CROWN-2, POINT, and Homi-Lung (granted to the Hellenic Institute for the Study of Sepsis). Mathias W. Pletz received grants from Pfizer, consulting fees from Sanofi and Pfizer, and honoraria for lectures and presentations from Pfizer, MSD, Sanofi, Janssen, GSK, AstraZeneca, Shionogi, InfectoPharm, bioMérieux, and Moderna. He also reports support for attending meetings from Pfizer, MSD, Sanofi, bioMérieux, and Shionogi, and participated in advisory boards for bioMérieux, Sanofi, Pfizer, GSK, AstraZeneca, Shionogi, InfectoPharm, Tillotts, and Moderna. Alessandra Oliva has received speaker’s honoraria for educational meetings from Shionogi; honoraria for participation in a scientific board from Mundipharma, Advanz-Pharma, and Pfizer; and speaker’s honoraria from InfectoPharm. Matthias G. Vossen received consulting fees, speaker honoraria, and payment for expert testimony from Astro Pharma. Alessandra Mularoni, Giovanna Russelli, Valentina Galfo, Giusy Tiseo, Francesco Alessandri, Giancarlo Ceccarelli, Eleni Mouloudi, Stavrina Avgeropoulou, Dominik Jarczak, Christina Iasonidou, Alessandro Capone, Alessandra Bandera, Annalisa Saracino, Pavlos Myrianthefs, Michael Zoller, Jan T. Kielstein, Carlo Tascini, Antonio Cascio, Abhijit M. Bal, Francesca Ferretti, Valerio Del Bono, Despina Hatzilia, Marco Falcone, George Dimopoulos, Matteo Bassetti, and Claudio M. Mastroianni have nothing to declare. Thomas Borrmann and Christian Mayer are employees of InfectoPharm, Heppenheim, Germany.