Real-World Data on the Use of Intravenous Fosfomycin for the Treatment of Central Nervous System Infections: a Subgroup Analysis from the FORTRESS Study Authors: Dominik Jarczak, Stefan Kluge, Martin Kieninger, Stefan Hagel, Mathias W Pletz, Michael Zoller, Claudia Spies, Sebastian Kintrup, Lukas Antonitsch, Jörg Zundel, Valerio Del Bono, Valentina Galfo, Marco Falcone, Christina Iasonidou, Loredana Sarmati, Laura Campogiani, Abhijit M Bal, George Dimopoulos, Matthias G Vossen, Claudio M Mastroianni, Klaus-Friedrich Bodmann, Carina Herbst, Christian Mayer Published: 2026 ABSTRACT AbstractIntroductionCentral nervous system (CNS) infections are associated with high morbidity and mortality, and their management is challenged by antimicrobial resistance and limited cerebrospinal fluid (CSF) penetration of many antibiotics. Intravenous fosfomycin (FOS) offers broad-spectrum activity, favourable pharmacokinetics and reliable CSF penetration, making it a potential partner in combination therapy for severe CNS infections. This interim subgroup analysis of the prospective FORTRESS study evaluated real-world treatment patterns, effectiveness and safety of FOS in patients with CNS infections.MethodsFORTRESS is an ongoing multicentre, non-interventional, international study enrolling patients receiving FOS for severe infections. We analysed patients with documented CNS infections among 1019 enrolled up to June 2025. Effectiveness endpoints at end of treatment (EOT) included clinical success (composite endpoint defined as successful clinical response plus concomitant microbiological cure), successful clinical response (defined as complete or partial resolution of signs and symptoms), microbiological cure, clinical failure and in-hospital mortality. Safety outcomes included adverse drug reactions (ADRs; including serious ADRs) and laboratory parameters.ResultsA total of 64 patients with CNS infections were included, of which most were critically ill at baseline, with high rates of ICU admission (90.6%), sepsis (42.2%) and mechanical ventilation (35.9%). Included patients were treated for ventriculitis (53.1%), brain abscess/empyema (26.6%), bacterial meningitis (26.6%) and shunt infections (18.8%), mainly caused by Staphylococcus spp. (48.8%). FOS was administered exclusively in combination regimens, generally as second- or third-line therapy, at high daily doses (median 22.1 g/day). At EOT, clinical success was achieved in 82.8% of patients, a successful clinical response in 92.2% and microbiological cure in 85.9%. All-cause in-hospital mortality was 12.5%. Outcomes were especially favourable in bacterial meningitis (94.1% clinical success) and Gram-positive infections (96.6% clinical success). FOS was generally well tolerated. Electrolyte imbalances were the most common ADRs but were mainly mild and did not require treatment modification.ConclusionsThese interim real-world data suggest that FOS, used in combination with other antimicrobial agents, is an effective and well-tolerated treatment option for severely ill patients with CNS infections.Trial RegistrationClinicalTrials.gov identifier, NCT02979951.Keywords: Bacteraemia, CNS infection, CSF, Enterobacterales, Fosfomycin, MDR, Meningitis, Observational study, Staphylococcus aureus, Staphylococcus epidermidis Key Summary Points Why carry out this study? CNS infections are difficult to manage owing to rising antimicrobial resistance and challenges achieving adequate cerebrospinal fluid antibiotic levels.Intravenous fosfomycin (FOS) shows good CNS penetration and broad antibacterial activity, but prospective real-world data on the use of FOS in CNS infections remain limited.This study aimed to evaluate real-world treatment patterns, effectiveness and safety of FOS in CNS infections within the large prospective FORTRESS study. What was learned from the study? Fosfomycin, used exclusively in combination therapy, achieved high clinical success (82.8%), clinical response (92.2%) and microbiological cure (85.9%) rates in severely ill patients with CNS infections, with an overall all-cause in-hospital mortality rate of 12.5%.Effectiveness was particularly strong in bacterial meningitis and infections caused by Gram-positive pathogens such as Staphylococcus aureus and S. epidermidis.Fosfomycin was generally well tolerated; electrolyte imbalances were the most common adverse drug reactions and were mainly mild, without requiring treatment discontinuation. Introduction Bacterial central nervous system (CNS) infections, including bacterial meningitis, encephalitis, ventriculitis, brain abscesses and intracranial empyema, constitute a major cause of morbidity and mortality worldwide [1]. CNS infections are either community-acquired or nosocomial, the latter typically arising following neurosurgical procedures or traumatic head injury. Distinct pathogen profiles are associated with each infection setting. In adults, community-acquired bacterial meningitis is most frequently caused by Streptococcus pneumoniae and Neisseria meningitidis, while nosocomial bacterial meningitis is predominantly attributed to Streptococcus spp., Staphylococcus spp. and Gram-negative bacilli. Moreover, the epidemiology of bacterial meningitis varies considerably across geographic regions and healthcare environments [1]. CNS infections are associated with high rates of neurologic sequelae, systemic complications and mortality; they represent medical emergencies that require rapid initiation of effective targeted antimicrobial therapy. The management of CNS infections remains, however, particularly challenging owing to the rising prevalence of antimicrobial resistance and notably of multidrug resistance (MDR) [1, 2]. Another challenge is the proper penetration of antibiotics into the cerebrospinal fluid (CSF) and extracellular space of the brain. Accordingly, selection of antimicrobial agents is constrained, and higher doses of antibiotics or prolonged infusions are often considered [1–4]. The choice of antibiotics varies according to the underlying pathogen, site of infection, epidemiological determinants and host factors (such as age, immune status, renal and hepatic function, comorbidities), and combination therapy is frequently required to achieve optimal therapeutic outcomes (i.e. maximise bactericidal activity, achieve synergistic effects and prevent the emergence of resistance) [1]. Intravenous fosfomycin (FOS) is an antibiotic that re-emerged in recent years as a promising antibacterial agent, notably as combination partner to other antibiotics, for the treatment of various types of difficult-to-treat infections [3, 5–10]. FOS is a low-molecular-weight compound [3], with negligible protein-binding capacity [11]. It is able to cross the blood–brain barrier and reach both the CSF and brain parenchyma, with CSF concentrations further increasing in the presence of meningeal inflammation, such as in bacterial meningitis [3, 4, 12–14]. It therefore meets several key criteria for its use in the treatment of CNS infections [1]. FOS confers several clinical advantages. It is a broad-spectrum bactericidal agent effective against both Gram-positive and Gram-negative pathogens, including MDR, carbapenem-resistant (CR) and extended-spectrum β-lactamase (ESBL)-producing Enterobacterales, CR Pseudomonas aeruginosa, methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE) and penicillin-resistant Streptococcus pneumoniae (PRSP) [2, 3, 15–18]. It exerts antibiofilm activity [3, 19] and shows favourable pharmacokinetic characteristics, high cellular uptake and tissue penetration (including in cerebral tissue, as indicated above) and a favourable safety profile [2, 3]. Most notably, extensive evidence demonstrates that FOS exhibits broad synergy with most antibiotic classes [5, 8, 19–28]. Recent evaluation of published evidence, based on an analysis of 224 patients across 32 publications, supports the consideration of FOS as a valuable therapeutic option, particularly for resistant pathogens, in the treatment of CNS infections, including bacterial meningitis, abscesses and CSF shunt infections [2]. FOS demonstrated high clinical success rates across various CNS infection types, with overall 94% of patients cured. Microbiological outcomes were similarly favourable, with 97% of patients achieving CSF sterilisation [2]. However, available data on the use of FOS for treating CNS infections is currently largely limited to case reports, case series and smaller observational studies [1, 2], while prospective real-world evidence is scarce. The ongoing FORTRESS study, a large prospective, observational, multicentre study, was initiated to collect thorough real-world data on FOS from daily clinical practice across various countries. While a comprehensive interim analysis has recently been published [6], the present work aimed to further evaluate the real-world use, effectiveness and safety of FOS in a subgroup of patients with CNS infections, including a detailed analysis of patient profiles, treatment patterns and microbiological characteristics. Methods Study Population Study participants are a subgroup of patients with documented CNS infection (including bacterial meningitis) enrolled as part of the ‘Fosfomycin IV for Treatment of Severely Infected Patients’ (FORTRESS) study. Enrolled patients were those scheduled for treatment with FOS according to the respective (national) Summary of Product Characteristics (SmPC), as part of clinical routine (real-world setting). Inclusion and exclusion criteria have been detailed in the recent interim analysis of the FORTRESS study [6]. Study Design FORTRESS is an ongoing prospective, international, multicentre, non-comparative, non-interventional clinical study conducted in five European countries at the time of this interim analysis (i.e. Germany, Italy, the UK, Greece and Austria), to document and evaluate the effectiveness and safety of the treatment of severely infected patients with FOS under real-life conditions (ClinicalTrials.gov identifier: NCT02979951). The study was conducted in accordance with the Declaration of Helsinki of 1964 and its amendments and was approved by the ethics committees or other authorities of the respective participating centres, according to national and local requirements, as described previously [6]. Written informed consent was obtained from each participant, as previously reported [6]. In Austria, an informed consent waiver was approved for patients who were incapacitated. Data collection and sample size were formerly described [6]. FORTRESS is an ongoing study initiated in January 2017 and aiming to enrol a total of 1500 participants. For the present interim analysis, the database was locked in June 2025, with documentation completed for 1019 patients (i.e. full analysis [FA]/safety-evaluable [SE] population) from 45 centres. Of these, patients with documented CNS infections/bacterial meningitis (hereinafter referred to as CNS infections) were included in this interim subgroup analysis. Parts of the current dataset have been published in Bodmann et al. [6]. Definitions and Outcomes Definitions and outcomes of the FORTRESS study have been previously described [6]. For this subgroup analysis of CNS infections, the following endpoints were evaluated: clinical success at end of FOS treatment (EOT), which was the primary endpoint of the study, successful clinical response at EOT, clinical failure at EOT, microbiological cure at EOT and all-cause in-hospital mortality. Successful clinical response was defined as either resolution of signs and symptoms or partial resolution of signs and symptoms (as assessed by the treating physician). Microbiological cure was defined as eradication of the relevant pathogen(s) at the relevant site(s) of infection or when the attending investigator explicitly documented that no follow-up sample was available or indicated due to confirmed sufficient clinical response, pathogen elimination was assumed. Clinical success is a composite endpoint defined as successful clinical response plus concomitant microbiological cure. The assessment of clinical failure was at the discretion of the responsible attending physician. In addition, pre-treatments at baseline were defined as intravenous antibiotic therapy during the current hospital stay for the current infection, prior to FOS start. Antimicrobial susceptibility testing (AST) was performed according to local practices and guidelines. For this interim subgroup analysis, safety evaluation included documented sodium and potassium levels, adverse drug reactions (ADRs, defined as possibly caused by FOS administration as assessed by the study physician), serious ADRs (SADRs) and death. Deaths recorded as part of the safety evaluation could also be documented after hospital discharge. The severity of electrolyte imbalances such as hypernatraemia and hypokalaemia was assessed at the discretion of the treating physician in accordance with local protocols. Statistical Analysis Descriptive statistics (mean, standard deviation [SD], median, interquartile range [IQR]) were applied to continuous variables. Counts (n) and frequencies (%) were applied to categorical (dichotomous) variables. Associations between baseline variables and clinical success at EOT were examined using univariate Firth’s bias-reduced logistic regression. Given the small sample size, Firth’s logistic regression was applied to reduce the bias associated with maximum likelihood estimation. Results of the univariate logistic regression analysis were reported as odds ratio (OR) with 95% confidence interval (CI). p-Values < 0.05 indicated a statistically significant association. Univariate Firth’s bias-reduced logistic regression analysis was performed using the R Statistical Software (version 4.4.3) [29] and the logistf (version 1.26.1) package [30]. All other statistical analyses were carried out using the SAS software (version 9.4). Figures were prepared employing GraphPad Prism (version 10.6.1). Results Patients’ Demographics and Characteristics at Baseline Of the 1019 enrolled patients with complete documentation at database lock, 64 presented with a CNS infection. Most (58/64 [90.1%]) patients were enrolled in Germany, while three were enrolled in Italy, two in Austria and one in Greece. The patients’ demographics and characteristics at baseline are summarised in Table 1. Median (IQR) age of patients with CNS infections was 55.0 (45.0–65.5) years, and half (50.0%) of the patients were female. Most (90.6%) patients were in intensive care unit (ICU) at the start of FOS treatment. Documented median (IQR) APACHE II score in evaluable patients was 20.0 (13.0–23.0), indicating a markedly elevated severity of illness. Table 1Demographics and clinical characteristics at baseline (n = 64)CharacteristicsAge in years, median (IQR)55.0 (45.0–65.5)Sex (female), n (%)32 (50.0)Patients in ICU, n (%)58 (90.6)APACHE II score, median (IQR) (n = 11)20.0 (13.0–23.0)GCS score, median (IQR) (n = 57)13.0 (8.0–15.0)Sepsis, n (%)27 (42.2)Septic shock, n (%)1 (1.6)Mechanical ventilation, n (%)23 (35.9)Bacteraemia, n (%)11 (17.2)Concomitant SARS-CoV-2 infection, n (%)2 (3.1)Immunosuppression, n (%)a9 (14.1)Creatinine clearance in mL/min, median (IQR) (n = 58)106.5 (94.0–118.0)Creatinine clearance < 60 mL/min, n (%)5 (8.6)Concomitant treatment, n (%) Any55 (85.9) Glucocorticoids20 (31.3)Comorbidities, n (%)b Any62 (96.9) Electrolyte disorders31 (48.4) Hypokalaemia22 (34.4) Hypernatraemia11 (17.2) Hyponatraemia10 (15.6) Cardiovascular27 (42.2) Oncologic14 (21.9) Renal6 (9.4)CNS infection details, n (%)b Ventriculitis34 (53.1) Brain abscess/empyema17 (26.6) Postoperative meningitis15 (23.4) Ventricular shunt infection12 (18.8) Posttraumatic meningitis2 (3.1)Sample origin, n (%)b CSF30 (46.9) Blood11 (17.2) Abscess5 (7.8) Tissue sample3 (4.7) Sinus tract drainage1 (1.6) Respiratory secretion1 (1.6) Otherc20 (31.3)Imaging results, n (%)b Cerebral oedema22 (36.7) Hydrocephalus20 (31.3) Midline shift7 (10.9) Otherd18 (28.1)Protein concentration in CSF in mg/dL, median (IQR) (n = 45)175.9 (97.6–280.9)APACHE II, Acute Physiology and Chronic Health Evaluation II; CNS, central nervous system; CSF, cerebrospinal fluid; GCS, Glasgow Coma Scale; ICU, intensive care unit; IQR, interquartile rangeaDefined as patients receiving immunosuppressive/anticancer chemotherapy and/or patients with immunosuppressive comorbiditybMultiple entries possiblecPatient-level: biopsy (n = 2), deep nasal swab (n = 1), enteric (n = 1), intraoperative swab (n = 8), CSF (n = 1), nasal smear test (n = 1), native material, tissue (n = 1), shunt material (n = 1), ventricular peritoneal drainage (n = 1), wound swab (n = 4)dPatient-level (excluding abscesses): hygroma (n = 2), vasospasm (n = 2), diffusion abnormality (n = 1), cerebral perfusion disorder (n = 1), CSF circulation disorder (n = 1), signs of cerebral ischemia (n = 1), cerebral infarction (n = 1), cerebral haematoma (n = 1), cerebritis (n = 1), suspected ventriculitis/shunt insufficiency (n = 1), shunt dysfunction (n = 1), cerebellar haemorrhage (n = 1), intracerebral haemorrhage (n = 1), intraventricular haemorrhage (n = 1), CSF accumulation (n = 1), spondylodesis/osteochondrosis/spinal canal stenosis (n = 1), signs of recent neurosurgery (n = 1), suspected ventriculitis (n = 1) Demographics and clinical characteristics at baseline (n = 64) APACHE II, Acute Physiology and Chronic Health Evaluation II; CNS, central nervous system; CSF, cerebrospinal fluid; GCS, Glasgow Coma Scale; ICU, intensive care unit; IQR, interquartile range aDefined as patients receiving immunosuppressive/anticancer chemotherapy and/or patients with immunosuppressive comorbidity bMultiple entries possible cPatient-level: biopsy (n = 2), deep nasal swab (n = 1), enteric (n = 1), intraoperative swab (n = 8), CSF (n = 1), nasal smear test (n = 1), native material, tissue (n = 1), shunt material (n = 1), ventricular peritoneal drainage (n = 1), wound swab (n = 4) dPatient-level (excluding abscesses): hygroma (n = 2), vasospasm (n = 2), diffusion abnormality (n = 1), cerebral perfusion disorder (n = 1), CSF circulation disorder (n = 1), signs of cerebral ischemia (n = 1), cerebral infarction (n = 1), cerebral haematoma (n = 1), cerebritis (n = 1), suspected ventriculitis/shunt insufficiency (n = 1), shunt dysfunction (n = 1), cerebellar haemorrhage (n = 1), intracerebral haemorrhage (n = 1), intraventricular haemorrhage (n = 1), CSF accumulation (n = 1), spondylodesis/osteochondrosis/spinal canal stenosis (n = 1), signs of recent neurosurgery (n = 1), suspected ventriculitis (n = 1) A total of 27 (42.2%) patients presented with sepsis, including one who additionally developed septic shock, and 23 (35.9%) required mechanical ventilation. Bacteraemia was detected in 11 (17.2%) patients, and few (3.1%) had a concomitant SARS-CoV-2 infection. Most (96.9%) patients presented with comorbidities, primarily electrolyte disorders (48.4%), such as hypokalaemia (34.4%), hypernatraemia (17.2%) and hyponatraemia (15.6%), followed by cardiovascular (42.2%) and oncologic (21.9%) disorders (Table 1). More than half (53.1%) of the CNS infections included in this study were attributed to ventriculitis, followed by brain abscess/empyema (26.6%), bacterial meningitis (26.6%; 15/17 postoperative [88.2%]) and ventricular shunt infection (18.8%). Most (64.1%) diagnoses were established using CSF (46.9%) and blood samples (17.2%). CNS infections were often associated with complications, notably cerebral oedema (36.7%) and hydrocephalus (31.3%), as assessed by neuroimaging. Protein concentration in the CSF (evaluable in 45 patients), a key diagnostic marker of CNS infections [31], was elevated at baseline (i.e. prior to FOS initiation), with a median (IQR) of 175.9 (97.6–280.9) mg/dL (normal range: < 50 mg/dL) (Table 1). Microbiological Description at Baseline The pathogens identified in this subset of patients prior to receiving FOS treatment are listed in Table 2. Staphylococcus spp. was identified in almost half (48.8%) of patients, with S. epidermidis being the most common pathogen (23.4%), followed by S. aureus (15.6%). Enterobacterales (28.1%) and the Streptococcus anginosus group (7.8%) were the next most frequent isolates. Non-fermenters and enterococci were uncommon (Table 2). Table 2Microbiology and resistance profiles at baseline (n = 64)Microbiological characteristicsn (%)aIsolated pathogensb Staphylococcus spp.31 (48.8) Staphylococcus epidermidis15 (23.4) Staphylococcus aureus10 (15.6) Other coagulase-negative staphylococcic7 (10.9) Enterobacterales18 (28.1) Enterobacter spp.5 (7.8) Escherichia coli5 (7.8) Klebsiella pneumoniae4 (6.3) Serratia spp.2 (3.1) Citrobacter spp.1 (1.6) Proteus spp.1 (1.6) Streptococcus anginosus groupd5 (7.8) Non-fermenters2 (3.1) Pseudomonas aeruginosa1 (1.6) Acinetobacter baumannii1 (1.6) Enterococcus faecalis1 (1.6) Othere4 (6.3)Type of infection Monomicrobial37 (57.8) Polymicrobial13 (20.3) Negative culture/no sample14 (21.9)Resistance patternsb Multidrug resistance (MDR)7 (10.9) ESBL-positive2 (3.1) Carbapenem resistance (CR)1 (1.6)CNS, central nervous system; ESBL, extended-spectrum beta-lactamaseaNumber and percentage of patientsbMultiple entries possible per patientcPatient-level: S. capitis (n = 3), S. hominis (n = 1), S. lugdunensis (n = 2), S. caprae (n = 1), S. haemolyticus (n = 1)dPatient-level: S. intermedius (n = 2), S. constellatus (n = 1), S. anginosus (n = 2)ePatient-level: Prevotella oralis (n = 1), Fusobacterium nucleatum (n = 1), Cutibacterium acnes (n = 1), Corynebacterium mucifaciens (n = 1), Candida spp. (n = 1) Microbiology and resistance profiles at baseline (n = 64) CNS, central nervous system; ESBL, extended-spectrum beta-lactamase aNumber and percentage of patients bMultiple entries possible per patient cPatient-level: S. capitis (n = 3), S. hominis (n = 1), S. lugdunensis (n = 2), S. caprae (n = 1), S. haemolyticus (n = 1) dPatient-level: S. intermedius (n = 2), S. constellatus (n = 1), S. anginosus (n = 2) ePatient-level: Prevotella oralis (n = 1), Fusobacterium nucleatum (n = 1), Cutibacterium acnes (n = 1), Corynebacterium mucifaciens (n = 1), Candida spp. (n = 1) Overall, 10.9% of isolated bacteria were classified as MDR, while 3.1% were ESBL-positive (exclusively identified in Klebsiella pneumoniae). Notably, all Staphylococcus spp. isolates, including both S. epidermidis and S. aureus, were methicillin susceptible. Carbapenem resistance (CR) was identified in one Acinetobacter baumannii isolate (Table 2). The majority (37/64 [57.8%]) of CNS infections were monomicrobial, while 13/64 (20.3%) were polymicrobial (Table 2). It is noteworthy that no pathogen was identified in 14/64 (21.9%) patients. Evaluation of susceptibility to FOS was documented in 31 (48.4%) patients, of whom 21 (67.7%) had baseline isolates susceptible to FOS. FOS Treatment Characteristics The patterns of FOS usage in patients with CNS infections are displayed in Table 3. FOS was administered as combination therapy in all (64/64 [100%]) enrolled patients, mostly with two partner antibiotics (51.6%), but also with one (29.7%) and ≥ 3 (18.8%) companion antibiotics. Meropenem was the antibiotic most frequently combined with FOS (65.6%), followed by vancomycin (60.9%) and flucloxacillin (15.6%). FOS was mainly (53.1%) administered as second-line treatment but also as first-line option in 26.6% of patients. Notably, more than half of the patients received FOS empirically, and the main rationales for its use included the presence of a difficult-to-treat infection (50.0%), a nosocomial infection (45.3%), lack of effectiveness or failure of pretreatment (32.8%) and its favourable tissue penetration properties (17.2%). Table 3Antibiotic infection management (n = 64)Treatment characteristicsFOS therapy Combination therapy, n (%)a64 (100) 1 partner antibiotic19 (29.7) 2 partner antibiotics33 (51.6) ≥ 3 partner antibiotics12 (18.8) Most frequently used companion antibiotics, n (%)a, b Meropenem42 (65.6) Vancomycin39 (60.9) Flucloxacillin10 (15.6) Ceftriaxone6 (9.4) Metronidazole5 (7.8) Rationale for FOS use, n (%)b Difficult-to-treat infection32 (50.0) Nosocomial infection29 (45.3) Insufficient effectiveness/treatment failure of pre-treatment21 (32.8) Favourable tissue penetration11 (17.2) Insufficient effectiveness of potential alternatives5 (7.8) Side effects of pre-treatment1 (1.6) Line of treatment, n (%) First-line17 (26.6) Second-line34 (53.1) Third-line13 (20.3) Empirical use35 (54.7) Duration of FOS treatment in days, median (IQR)13.0 (8.0–17.0) Time to start of FOS treatment in days, median (IQR)c2.0 (0.5–6.5) Targeted daily dose in g, median (IQR)d22.1 (15.0–24.0)Prior antibiotic therapye Prior antibiotics, n (%)47 (73.4) Duration of prior antibiotic treatment in days, median (IQR) (n = 47)5 (2.0–9.0) 1–3 days, n (%)19 (40.4) 4–7 days, n (%)11 (23.4) > 7 days, n (%)17 (36.2)CNS, central nervous system; FOS, intravenous fosfomycin; IQR, interquartile rangeaMaximum 1 day after FOS startbMultiple entries possiblecEstimated time from first (suspected) diagnosis to start of FOS treatmentdAverage targeted daily doseeFor current infection Antibiotic infection management (n = 64) CNS, central nervous system; FOS, intravenous fosfomycin; IQR, interquartile range aMaximum 1 day after FOS start bMultiple entries possible cEstimated time from first (suspected) diagnosis to start of FOS treatment dAverage targeted daily dose eFor current infection The median (IQR) time from diagnosis to start of FOS treatment was 2.0 (0.5–6.5) days. The median (IQR) targeted daily dose of FOS was 22.1 (15.0–24.0) g, and patients were treated for a median (IQR) duration of 13.0 (8.0–17.0) days (Table 3). Clinical and Microbiological Outcomes Clinical and microbiological outcomes (including clinical success, successful clinical response and microbiological cure) at EOT are shown for the total population (n = 64; Fig. 1 and Table 4), by infection type (Fig. 1) and by FOS usage, causative agents and baseline characteristics (Table 4). Clinical and microbiological outcomes of FOS-treated patients with CNS infections. The percentage (%) of patients with clinical success, successful clinical response, and microbiological cure was assessed in the whole study population (n = 64) and for the respective infection types: ventriculitis (n = 34), meningitis (n = 17, including 15 postoperative and 2 posttraumatic meningitis), brain abscess/empyema (n = 17) and ventricular shunt infection (n = 12). Multiple diagnoses per patient were possible. Patients may have had concomitant bacteraemia Table 4Clinical and microbiological outcomes by antibiotic use, causative agents, baseline characteristics and combination partners at end of FOS treatment (EOT)Subgroup/patient populationaCases, nClinical success, n (%)Successful clinical response, n (%)Clinical failure, n (%)Microbiological cure, n (%)All patients6453 (82.8)59 (92.2)2 (3.1)55 (85.9)FOS as first-line No4739 (83.0)45 (95.7)0 (0)40 (85.1) Yes1714 (82.4)14 (82.4)2 (11.8)15 (88.2)Daily dosageb ≤ 16 g/day2620 (76.9)24 (92.3)1 (3.8)21 (80.8) > 16 g/day3833 (86.8)35 (92.1)1 (2.6)34 (89.5)Bacteraemia at baseline No5343 (81.1)49 (92.5)2 (3.8)45 (84.9) Yes1110 (90.9)10 (90.9)0 (0)10 (90.9)Sepsis at baseline No3731 (83.8)35 (94.6)0 (0)32 (86.5) Yes2722 (81.5)24 (88.9)2 (7.4)23 (85.2)Prior antibioticsc No1714 (82.4)14 (82.4)2 (11.8)15 (88.2) Yes4739 (83.0)45 (95.7)0 (0)40 (85.1)Targeted therapy No3530 (85.7)32 (91.4)2 (5.7)31 (88.6) Yes2923 (79.3)27 (93.1)0 (0)24 (82.8)Polymicrobial at baseline No5141 (80.4)47 (92.2)2 (3.9)43 (84.3) Yes1312 (92.3)12 (92.3)0 (0)12 (92.3)Causative agent at baselined S. epidermidis1515 (100)15 (100)0 (0)15 (100) S. aureus1010 (100)10 (100)0 (0)10 (100) Enterobacterales1811 (61.1)15 (83.3)0 (0)13 (72.2) GP pathogense2928 (96.6)29 (100)0 (0)28 (96.6) GN pathogensf126 (50.0)10 (83.3)0 (0)8 (66.7) GP + GN pathogens87 (87.5)7 (87.5)0 (0)7 (87.5)Combination partner at baselined,g Meropenem4236 (85.7)39 (92.9)1 (2.4)37 (88.1) Vancomycin3932 (82.1)36 (92.3)2 (5.1)33 (84.6) Flucloxacillin109 (90.0)10 (100)0 (0)9 (90.0)EOT, end of treatment; FOS, intravenous fosfomycin; GN, Gram-negative; GP, Gram-positiveaWith or without bacteraemiabAverage targeted daily dosecFor current infectiondMultiple entries possibleePatients presenting with infections caused solely by GP pathogensfPatients presenting with infections caused solely by GN pathogensgMaximum 1 day after FOS start Clinical and microbiological outcomes by antibiotic use, causative agents, baseline characteristics and combination partners at end of FOS treatment (EOT) EOT, end of treatment; FOS, intravenous fosfomycin; GN, Gram-negative; GP, Gram-positive aWith or without bacteraemia bAverage targeted daily dose cFor current infection dMultiple entries possible ePatients presenting with infections caused solely by GP pathogens fPatients presenting with infections caused solely by GN pathogens gMaximum 1 day after FOS start Overall, clinical success was achieved in 53/64 (82.8%) patients, alongside 59/64 (92.2%) successful clinical response and 55/64 (85.9%) microbiological cure (Fig. 1 and Table 4). Clinical failure (2/64 [3.1%]) (Table 4) and all-cause in-hospital mortality (8/64 [12.5%]) were relatively low. The best clinical and microbiological outcomes were observed in patients with meningitis (n = 17) with 94.1% clinical success and microbiological cure and 100% successful clinical response (Fig. 1). In comparison, outcomes were slightly lower in ventriculitis patients (n = 34), with 79.4% clinical success, 88.2% successful clinical response and 85.3% microbiological cure (Fig. 1). In dichotomous analyses, clinical success was more frequent in patients receiving a daily dose > 16 g versus ≤ 16 g (86.8% versus 76.9%), with baseline bacteraemia (90.9% versus 81.1% without) and polymicrobial infection (92.3% versus 80.4% for non-polymicrobial) (Table 4). Clinical success was also higher in CNS infections caused by Gram-positive pathogens compared with Gram-negative infections (96.6% versus 50.0%), notably against S. epidermidis (100%) and S. aureus (100%). Univariate Firth’s bias-reduced logistic regression was used to evaluate a possible association between baseline variables and the probability of achieving clinical success at EOT (Table 5). Of the variables investigated, four showed a significant association (p < 0.05), all indicating lower odds of clinical success (OR < 1), specifically in patients aged 60–69 years compared with < 60 years (OR = 0.184; p = 0.019), those under mechanical ventilation at baseline (OR = 0.264; p = 0.041), those diagnosed with Enterobacterales infection (OR = 0.162; p = 0.006) and high protein level in CSF at baseline (OR = 0.997; p = 0.048). In that regard, protein level in CSF strongly decreased from a median (IQR) of 175.9 (97.6–280.9) mg/dL at baseline (n = 45) (Table 1) to 64.4 (36.0–108.1) mg/dL at EOT in evaluable patients (n = 34) (of whom 31 patients had values available at both baseline and EOT). Table 5Univariate Firth’s bias-reduced logistic regression analysis for clinical success at end of FOS treatment (EOT) VariableUnivariate analysis (n = 64)OR95% CIp-ValueAge (continuous)0.9770.927–1.0230.3420Age (categorical) < 60 yearsRef 60–69 years0.1840.039–0.7530.0185 70–79 years0.2110.032–1.5080.1142Type of infection Brain abscess/empyema0.8920.235–4.0200.8713 Ventriculitis0.6230.160–2.2010.4649 Ventricular shunt infection0.9170.215–5.3380.9133 Bacterial meningitis3.0800.637–30.1970.1762ICU1.2600.120–7.3160.8172Mechanical ventilation0.2640.066–0.9480.0411Complicationsa0.6540.115–2.6620.5713Protein level in CSF [mg/dL] (continuous)0.9970.994–1.0000.0475Concomitant glucocorticoid use at baseline0.4760.131–1.7670.2599Presence of bacteraemia at baseline1.6900.333–16.8790.5583Presence of sepsis at baseline0.8440.238–3.0830.7921Prior antibiotics1.1220.249–4.2480.8713Duration of prior antibiotics0.8970.783–1.0110.0731Duration of prior antibiotics (categorical) 1–3 daysRef 4–7 days0.3080.025–2.6500.2781 ≥ 7 days0.1840.017–1.0810.0616FOS treatment FOS start within ≤ 2 daysb1.5340.434–5.6070.5037 High-dosing regimenc1.9310.544–7.100.3055 Duration of FOS treatment1.0280.949–1.1360.5257 Targeted therapy0.6520.178–2.3060.5037 First-line FOS treatment1.1220.249–4.2480.8713Causative agent Enterobacterales0.1620.039–0.6000.0064 Polymicrobial2.1080.425–20.8840.3914Combination partner Vancomycin0.9070.232–3.2270.8817 Meropenem1.7650.479–6.3600.3843CI, confidence interval; CNS, central nervous system; CSF, cerebrospinal fluid; EOT, end of treatment; FOS, intravenous fosfomycin; Ref, reference (for statistical comparison); ICU, intensive care unit; OR, odds ratioaPatient-level: hygroma (n = 2), vasospasm (n = 2), diffusion abnormality (n = 1), cerebral perfusion disorder (n = 1), CSF circulation disorder (n = 1), signs of cerebral ischemia (n = 1), cerebral infarction (n = 1), cerebral haematoma (n = 1), cerebritis (n = 1), suspected ventriculitis/shunt insufficiency (n = 1), shunt dysfunction (n = 1), cerebellar haemorrhage (n = 1), intracerebral haemorrhage (n = 1), intraventricular haemorrhage (n = 1), CSF accumulation (n = 1), signs of recent neurosurgery (n = 1)bEstimated time from (suspected) diagnosis to start of FOS treatmentc> 16 g FOS/dayBold values indicate statistical significance (p< 0.05) Univariate Firth’s bias-reduced logistic regression analysis for clinical success at end of FOS treatment (EOT) CI, confidence interval; CNS, central nervous system; CSF, cerebrospinal fluid; EOT, end of treatment; FOS, intravenous fosfomycin; Ref, reference (for statistical comparison); ICU, intensive care unit; OR, odds ratio aPatient-level: hygroma (n = 2), vasospasm (n = 2), diffusion abnormality (n = 1), cerebral perfusion disorder (n = 1), CSF circulation disorder (n = 1), signs of cerebral ischemia (n = 1), cerebral infarction (n = 1), cerebral haematoma (n = 1), cerebritis (n = 1), suspected ventriculitis/shunt insufficiency (n = 1), shunt dysfunction (n = 1), cerebellar haemorrhage (n = 1), intracerebral haemorrhage (n = 1), intraventricular haemorrhage (n = 1), CSF accumulation (n = 1), signs of recent neurosurgery (n = 1) bEstimated time from (suspected) diagnosis to start of FOS treatment c> 16 g FOS/day Bold values indicate statistical significance (p< 0.05) Safety Evaluation In total, 27 of 64 (42.2%) patients with CNS infections experienced at least one adverse drug reaction (ADR; i.e. possibly causally related to FOS treatment), and 2/64 (3.1%) had a serious ADR (SADR). A total of eight (12.5%) patients died, but none of the deaths were considered related to FOS treatment. Most frequently observed ADRs were hypernatraemia and hypokalaemia in 12/64 (18.8%) and 13/64 (20.3%) patients, respectively. All 12 hypernatraemia cases were assessed as mild by the attending physician, while 10 of 13 hypokalaemia cases were assessed as mild and two as moderate and one as severe. However, not all of these electrolyte disorders necessitated corrective intervention, and those that did were readily manageable. Consequently, FOS was neither discontinued nor was the dose reduced in any of the patients with reported electrolyte imbalances. Other ADRs were elevated liver enzymes in two patients and one case each of nausea, leukopenia, abdominal pain, cholelithiasis and fever, the latter requiring discontinuation of FOS treatment. Despite the documentation of hypernatraemia and hypokalaemia as common ADR, the overall sodium and potassium levels in the study population (when documented) were in a comparable range at baseline and after 13 days of FOS treatment (i.e. at the median duration of FOS therapy; see Table 3). Median (IQR) sodium levels at baseline and at day 13 were 139.0 (136.5–141.5) mmol/L (n = 12) and 142.0 (140.0–147.0) mmol/L (n = 23), respectively, while median (IQR) potassium levels at baseline and at day 13 were 3.85 (3.30–4.35) mmol/L (n = 12) and 3.80 (3.60–3.90) mmol/L (n = 22), respectively. Discussion This report is a subgroup analysis of patients with CNS infections from the ongoing FORTRESS study, a prospective, observational, multicentre, international study aiming to evaluate the real-world use, effectiveness and safety of FOS for the treatment of severely infected patients. Almost all (96.9%) patients included in this analysis had several comorbidities, including notably cardiovascular and electrolyte disorders, and most patients presented with severe illness (90.6% in ICU, 42.2% with concomitant sepsis, 35.9% on mechanical ventilation, median APACHE II score of 20). FOS was used across a broad range of CNS infections, including ventriculitis, brain abscess/empyema, bacterial meningitis and ventricular shunt infection, in line with the results of a comprehensive review by Tsegka and colleagues [2]. CNS infections in our cohort were caused by a wide spectrum of pathogens, including Gram-positive (mostly Staphylococcus spp., but also the Streptococcus anginosus group and Enterococcus faecalis) and Gram-negative (mostly Enterobacterales, but also P. aeruginosa and Acinetobacter spp.) bacteria. The spectrum of causative agents identified in the present study, notably the predominance of Gram-positive clinical isolates (Staphylococcus spp. [48.8%] in particular), reflects the typical microbial aetiology of nosocomial CNS infections [32, 33]. Bacterial infections treated with FOS were in part polymicrobial (20.3%), and causative pathogens showed a relatively low rate of phenotypical antibiotic resistance. Overall, 10.9% of isolated pathogens were classified as MDR. Notably, all isolated Staphylococcus spp., including the two most frequently documented pathogens S. aureus and S. epidermidis, were methicillin-susceptible (MSSA and MSSE, respectively). In the context of the predominantly German patient collective in our study, the absence of MRSA among S. aureus CNS infections is consistent with the low prevalence (< 5%) of MRSA reported for Germany in the 2025 annual EPI-Net report, a European antimicrobial resistance surveillance project [34]. Conversely, our findings contrast with the usually higher rates of methicillin-resistant S. epidermidis (MRSE) documented for this region [35, 36]. In line with its recommended use and previous studies [2, 3, 5–10], FOS was exclusively used in combination with other antibiotics, most frequently with ≥ 2 companion antibiotics. Compared with another observational study conducted in Germany and Austria [7], FOS was less frequently used as first-line option in the present study (55.5% versus 26.6% in our study). Consistent with previously published data [2, 4, 12], FOS was administered at high dose, with a median daily dose of 22.1 g. A likely rationale for these high-dose FOS regimens was to reach a sufficient antibiotic concentration in the CSF to improve clinical success, a generally accepted approach in the treatment of critically ill patients with CNS infections [1]. We found that the clinical and microbiological outcomes of patients with CNS infections treated with FOS-containing combination regimens were overall favourable, with clinical success, successful clinical response and microbiological cure achieved at EOT in 82.8%, 92.2% and 85.9% of patients, respectively. These outcomes are in a comparable range as those reported by Putensen et al. in their (smaller) population of patients treated for CNS infections (n = 45), with 89.1% clinical success and 81.8% microbiological cure [7] and those reported in other studies, as summarised in a review article [2]. Except for bacterial meningitis, for which clinical success was achieved in 94.1% of patients, clinical success rates were slightly lower and comparable between patients with ventriculitis, brain abscess/empyema and ventricular shunt infections. While most evidence on FOS use in CNS infections pertain to the treatment of bacterial meningitis [2], data on its clinical and microbiological effectiveness in ventriculitis are currently scarce and limited to case reports. In this context, Gatti and colleagues reported on the successful use of FOS in combination with ceftazidime–avibactam in a patient with post-neurosurgical ventriculitis caused by CR P. aeruginosa. In this patient, FOS was administered as continuous infusion with a total daily dose of 24 g, reaching optimal pharmacodynamic target attainment and CSF-to-plasma ratios ranging from 0.42 to 0.5 [37]. In another case report, a patient with post-neurosurgical ventriculitis and concomitant bacteraemia due to KPC-producing K. pneumoniae was successfully treated with meropenem–vaborbactam and high-dose FOS [38]. In our cohort, ventriculitis was the most frequently treated CNS infection type (n = 34), with successful clinical response and microbiological cure achieved in 88.2% and 85.3% of patients, respectively, expanding the limited body of evidence and supporting the use of FOS in combination therapy in this challenging clinical condition. Similarly, data on the effectiveness of FOS for the treatment of brain abscesses are limited and originate from small observational studies and case reports [2, 39–42]. In a study from the 1980s, FOS was used in 25 patients with brain abscesses and six with spinal abscesses, mainly as monotherapy, with an overall favourable clinical outcome and a high rate of CSF sterilisation [2, 43]. In our cohort, a successful clinical response and microbiological cure were achieved in 88.2% and 82.4% of patients with brain abscess/empyema, respectively. In this context, brain abscesses are particularly challenging and typically require a combination of surgical measures and antimicrobial therapy. Due to its physicochemical properties, FOS rapidly crosses the blood–brain barrier, reaching adequate concentration in the CSF and brain parenchyma [4, 12–14, 44]. Together with its ability to penetrate into abscess fluid [45] and the favourable outcomes in the present study, FOS may be regarded as a useful combination partner for the treatment of brain abscesses. In complement to the clinical and microbiological outcome analyses, the assessment of the protein concentration in the CSF was informative of a possible successful response to FOS combination therapy. Protein concentration in the CSF was elevated at baseline, prior to FOS start (median 175.9 mg/dL) and was strongly reduced at EOT (median 64.4 mg/dL). This agrees with CSF protein levels being typically elevated in CNS infections due to inflammation-induced disruption of the blood–brain barrier, intrathecal release of cytokines, antibodies and cellular proteins from activated immune cells and infection-related neuronal or glial injury, with concentrations in bacterial meningitis often exceeding 100 mg/dL (as opposed to < 50 mg/dL in healthy adults) [31]. Altogether, these clinical and microbiological outcome results suggest a beneficial role of FOS as combination partner, especially when administered in doses > 16 g per day, in the treatment of CNS infections in patients who are critically ill. The safety profile of FOS use in the current cohort was also favourable, with mild-to-moderate hypernatraemia and hypokalaemia being the most frequent ADRs (18.8% and 20.3%, respectively), in line with other studies [2, 7, 46–48]. However, it should be noted that electrolyte disorders were already documented as baseline comorbidities, with 17.2% and 34.4% of patients with hypernatraemia and hypokalaemia, respectively, and that overall sodium and potassium blood concentrations remained relatively stable during FOS treatment. Moreover, none of these documented ADRs led to FOS discontinuation or dose adjustment. Thus, while electrolytes should be regularly monitored during FOS therapy, imbalances are mostly manageable. In this context, the incidence of hypokalaemia may be reduced by supplementary potassium or prolonged infusion times [49, 50]. The strengths of the current analysis include its prospective, multicentre, real-life design and its extensive baseline data documentation that allowed both a thorough description of the patient population and a robust evaluation of effectiveness and safety. Limitations of this analysis include its non-comparative design, the potential confounding effect of the partner antibiotics co-administered with FOS, a potential selection bias among FOS-treated patients (as they were not systematically enrolled in the study), a potential country-specific bias due to most patients being recruited in Germany and the small sample size of this subset of analysed patients. Moreover, owing to the non-interventional nature of the present study, complete data were not available for all variables in all patients, including APACHE II scores, CSF protein levels at EOT and electrolyte measurements. Another limitation is the restriction to univariate regression analysis, as the small number of patients who did not achieve clinical success precluded multivariable adjustment for potential confounders. The overall limited sample size also resulted in relatively wide confidence intervals for some variables, and the reported associations of the univariate regression analysis should therefore be interpreted cautiously and considered exploratory. Conclusions FORTRESS provides valuable real-world evidence on treatment patterns, patient profiles and the effectiveness and safety of FOS in European clinical practice. The present interim subgroup analysis suggests that FOS, used in combination with other antibiotics, is an effective and safe treatment option for CNS infections. These findings may help inform and optimise antibiotic management in severely ill patients with CNS infections. Acknowledgements We are grateful to all participating study sites and FORTRESS investigators for their valuable contributions to this study. We thank the participants of this study. We are thankful to Katja Eifert and Regina Rothmann for their technical expertise. Medical Writing, Editorial and Other Assistance Editorial and medical writing support was provided by Anne Rascle (AR Medical Writing, Regensburg, Germany), in accordance with Good Publication Practice (GPP 2022) guidelines (https://www.ismpp.org/gpp-2022; accessed on 27 January 2026), and this support was sponsored by InfectoPharm. Author Contributions Dominik Jarczak and Stefan Kluge contributed to the conception and design of the work, were involved in the acquisition, analysis and interpretation of the data and drafting of the manuscript. Carina Herbst and Christian Mayer provided support with data analysis. Martin Kieninger, Stefan Hagel, Mathias W. Pletz, Michael Zoller, Claudia Spies, Sebastian Kintrup, Lukas Antonitsch, Jörg Zundel, Valerio Del Bono, Valentina Galfo, Marco Falcone, Christina Iasonidou, Loredana Sarmati, Laura Campogiani, Abhijit M. Bal, George Dimopoulos, Matthias G. Vossen, Claudio M. Mastroianni and Klaus-Friedrich Bodmann were involved in the acquisition of data. All authors revised the draft and approved the final version of the manuscript. Funding This study was funded by InfectoPharm Arzneimittel und Consilium GmbH, Heppenheim, Germany. All participating study centres received contractually agreed study fees by InfectoPharm 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 for open access publication. Data Availability The datasets generated and/or analysed 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 and subject to certain criteria, conditions and exceptions. Declarations Conflict of Interest The following authors declare potential conflicts of interest unrelated to the submitted work: Stefan Kluge received research support from Biotest, CytoSorbents, Daiichi Sankyo and Fresenius Medical Care. He also received lecture fees from ADVITOS, bioMérieux GmbH, CSL Behring, Gilead, MSD, Pfizer and Shionogi. He received consultant fees from ADVITOS, Gilead and Pfizer. Stefan Hagel has received honoraria from Pfizer, MSD, InfectoPharm, Philips, Advanz Pharma, Beckman Coulter, Shionogi, Thermo Fisher and Tillots. He also participated in advisory boards for Advanz Pharma, Shionogi and Pfizer. 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 having participated in advisory boards from bioMérieux, Sanofi, Pfizer, GSK, AstraZeneca, Shionogi, InfectoPharm, Tillotts and Moderna. Claudia Spies participated in advisory boards for Takeda Pharmaceutical, Lynx Health Science GmbH, Mundipharma, PAION Pharma GmbH. 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. Matthias G. Vossen received consulting fees and speaker honoraria from Astro Pharma. Klaus-Friedrich Bodmann has received honoraria for lectures from Abbott, Accelerate, Correvio, InfectoPharm, Pfizer, Weber & Weber and Shionogi. Carina Herbst and Christian Mayer are employees of InfectoPharm, Heppenheim, Germany. Dominik Jarczak, Martin Kieninger, Michael Zoller, Sebastian Kintrup, Lukas Antonitsch, Jörg Zundel, Valerio Del Bono, Valentina Galfo, Marco Falcone, Christina Iasonidou, Abhijit M. Bal, George Dimopoulos and Claudio M. Mastroianni have nothing to declare. Ethical Approval The study was conducted in accordance with the Declaration of Helsinki of 1964 (and its amendments) and was approved by the ethics committees or other authorities of the respective participating centres, according to national and local requirements, as described previously [6]. Written informed consent was obtained from each patient or the patient’s legally acceptable representative before any study-specific activity was performed. In Austria, an informed consent waiver was approved for patients who were incapacitated. Footnotes Prior presentation: Data from patients included in the current manuscript/report have been part of the publication by Bodmann et al. (2025) (https://pubmed.ncbi.nlm.nih.gov/40106180/). Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Dominik Jarczak and Stefan Kluge contributed equally. Carina Herbst and Christian Mayer share senior authorship. References Associated Data Data Availability Statement The datasets generated and/or analysed 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 and subject to certain criteria, conditions and exceptions.