Francisella tularensis antibiotic susceptibility has been mainly evaluated using the broth microdilution (BMD) and the E-test strips (E-test) methods, and less frequently the antibiotic agar dilution method (Caspar and Maurin, 2017). The BMD method from the Clinical and Laboratory Standards Institute (CLSI) is currently the best standardized method (Clinical And Laboratory Standards Institute, 2005, 2016). Experimental conditions include the use of cation-adjusted Mueller Hinton broth (caMHB) supplemented with 2% defined growth supplement (e.g., Polyvitex^® or IsoVitalex^®), a 5 × 10⁵ cfu/ml bacterial inoculum, and a 48 h incubation of cultures at 37°C in aerobic atmosphere (or 5% CO₂ atmosphere for capnophilic strains). In Tables 1–4 we have summarized data obtained with the BMD method using CaMHB medium or the E-test strips method. Some publications reporting high MICs suggesting acquired resistance to antibiotics in F. tularensis but without genetic confirmation were not considered (Baker et al., 1985; Scheel et al., 1993; Martínez-Martínez et al., 2021).

The beta-lactams are poorly active against F. tularensis with MIC90 (i.e., MIC inhibiting 90% of the tested strains) highest than the maximum concentration tested (i.e., > 32–256 mg/L) for penicillins, cephalosporins, monobactams, and carbapenems (Table 1). A few F. tularensis strains displayed low MICs, e.g., 4 mg/L for penicillin A, 0.03 mg/L for ceftriaxone, 0.047 mg/L for imipenem, and 0.094 mg/L for meropenem (Tomaso et al., 2005; Hotta et al., 2013; Heine et al., 2017). In the above studies, the percentage of strains with MICs lower than 2 mg/L among those tested varied from 4 % (Tomaso et al., 2005) up to 66.6% (Hotta et al., 2013) for imipenem, and was 50% for ceftriaxone (Hotta et al., 2013). These two studies used the E-test method. The last study did not specify the distribution of strains according to the MIC (Heine et al., 2017). These data are difficult to interpret until we further understand the mechanisms of beta-lactam resistance in F. tularensis (see below).

The beta-lactam/beta-lactamase inhibitor combinations are no more effective (e.g., amoxicillin/clavulanate and piperacillin/tazobactam) (Ikäheimo et al., 2000; Tomaso et al., 2005; Velinov et al., 2011; Yesilyurt et al., 2011; Georgi et al., 2012; Heine et al., 2017). Recently, ceftobiprole medocaril has been reported to display an MIC of < 0.03 mg/L for the Schu S4 strain (Hahn et al., 2023).

The aminoglycosides gentamicin and streptomycin are highly active against F. tularensis in vitro (Table 2). All tested strains had MICs lower than the CLSI breakpoints (≤8 mg/L for streptomycin and ≤ 4 mg/L for gentamicin). Tobramycin displayed MIC and MIC90 ranges of ≤ 0.03–0.5 mg/L and 0.25–2 mg/L for the BMD method and 0.125–3 mg/L and 0.25–1.5 mg/L for the E-test method (Ikäheimo et al., 2000; Tomaso et al., 2005; Yesilyurt et al., 2011; Heine et al., 2017; Caspar et al., 2018). Amikacin was less effective with MIC and MIC90 ranges of ≤ 0.03–8 mg/L and 2 mg/L for the MBD method and 0.75–16 mg/L and 1–4 mg/L for the E-tests method (Tomaso et al., 2005; Yesilyurt et al., 2011; Heine et al., 2017).

Tetracycline and doxycycline displayed MIC and MIC90 ranges lower than the CLSI susceptibility breakpoint (≤4 mg/L) for all F. tularensis strains tested (Table 3). Tigecycline displayed similar activity, with MIC and MIC90 ranges of ≤0.03–1 mg/L and 0.25–1 mg/L for the BMD method (Heine et al., 2017; Caspar et al., 2018) and 0.094–0.38 mg/L and 0.19–0.25 mg/L for the E-test method (Yesilyurt et al., 2011; Kreizinger et al., 2013).

The fluoroquinolones are the most effective antibiotics against F. tularensis in vitro (Table 3). All F. tularensis strains tested were susceptible to ciprofloxacin and levofloxacin according to CLSI susceptibility breakpoint (≤0.5 mg/L). Moxifloxacin displayed similar activity with MIC and MIC90 ranges of ≤ 0.004–0.25 mg/L and 0.06–0.125 mg/L for the BMD method (Heine et al., 2017; Caspar et al., 2018) and 0.012–0.125 mg/L and 0.032–0.125 mg/L for the E-test method (Johansson et al., 2002; Tomaso et al., 2005; Yesilyurt et al., 2011).

The activity of the macrolides depends on the F. tularensis clade (Table 4). Type B biovar (bv) II strains (currently clade B12) are naturally highly resistant to the macrolides, with erythromycin and azithromycin MICs > 256 mg/L. The other A and B clades display lower MICs to the macrolides although variables between studies. Overall, using the BMD method, erythromycin MIC and MIC90 ranges were 0.5–8 mg/L and 1–4 mg/L, respectively. Azithromycin was more effective, with MIC ranges of 0.125–2 mg/L. No macrolide CLSI susceptibility breakpoint is available for F. tularensis.

Chloramphenicol and rifampicin are considered less effective against F. tularensis, although no CLSI breakpoint is available for this species (Table 4). Moreover, rapid resistance to rifampicin may occur through mutations in the RNA polymerase encoding gene (Bhatnagar et al., 1994; Zaw et al., 2018).

Trimethoprim-sulfamethoxazole (ratio 1:19) was poorly active against F. tularensis, with MICs ≥ 4/76 mg/L for 69 type B strains from Europe (Georgi et al., 2012) and MIC ranges of 0.25/4.75–8/152 mg/L (MIC90, 4/76 mg/L) for 19 type A and 10 type B strains of variable origin (Heine et al., 2017).

Although we did not find MICs for glycopeptides and polymyxins against F. tularensis, it is likely that this bacterium is resistant to both antibiotic classes since its isolation from environmental samples was obtained using a selective medium containing 4 mg/L of vancomycin and 8 x 10⁴ U/L of polymyxin B (Petersen et al., 2009).

A few studies have evaluated the bactericidal activity of antibiotics against F. tularensis by determining the MBCs (minimum antibiotic concentration allowing 3-log or greater reduction of a bacterial inoculum). For fluoroquinolones (ofloxacin and ciprofloxacin) MBCs were close to MICs demonstrating the high bactericidal activity of these antibiotics against F. tularensis (Maurin et al., 2000; Aloni-Grinstein et al., 2015). The MBC/MIC ratios ranged from 2 to 8 for gentamicin and streptomycin compatible with a bactericidal activity of these aminoglycosides, at least for the most susceptible strains (Maurin et al., 2000; Caspar et al., 2014; Aloni-Grinstein et al., 2015). Less favorable MBC/MIC ratios were found for doxycycline (i.e., from 4 up to 512) suggesting lack of bactericidal activity (Maurin et al., 2000; Aloni-Grinstein et al., 2015). One study found no bactericidal activity of doxycycline at concentrations up to 8 mg/L (Caspar et al., 2014). Therefore, tetracyclines should be considered mainly bacteriostatic against F. tularensis. Thiamphenicol, erythromycin, and trimethoprim-sulfamethoxazole displayed MBC/MIC ratios of 2, 4 and 16, respectively (Maurin et al., 2000; Aloni-Grinstein et al., 2015). However, due to their high MICs, these antibiotics can be considered weakly or non-bactericidal on F. tularensis. Rifampicin was more effective (MBC, 1 mg/L; MBC/MIC ratio of 2) (Maurin et al., 2000), but as previously mentioned this antibiotic must not be used as monotherapy.

The activity of antibiotics against intracellular F. tularensis was determined in variable eukaryotic cell systems, including murine macrophage-like cells (P388D1 and J774.1), Vero cells, A549 cells, and MRC5 cells (Maurin et al., 2000; Ahmad et al., 2010; Madrid et al., 2013; Sutera et al., 2014a; Aloni-Grinstein et al., 2015). The cell systems may also vary according to the bacterial inoculum used to infect cell monolayers and the multiplicity of infection (MOI, i.e., the ratio of bacteria to cells), the time antibiotics are added after cell infection, and the length antibiotic exposure of infected cells. The bacterial inoculum used to infect cells varied from 1.2 × 10⁵ CFU (Maurin et al., 2000) to 10⁷ CFU (Sutera et al., 2014a). MOIs varied from 1 to 3,000 (Ahmad et al., 2010; Madrid et al., 2013; Aloni-Grinstein et al., 2015). Antibiotics were usually added to culture supernatant after cell infection and removal of non-phagocytized bacteria (i.e., 1–3 h post-infection). Antibiotic activity was then evaluated after a short (15–24 h) (Ahmad et al., 2010; Madrid et al., 2013; Aloni-Grinstein et al., 2015) or long (3–5d) (Maurin et al., 2000; Sutera et al., 2014a) incubation period of antibiotic exposure of infected cells. The antimicrobial activity was deduced from: (1) changes in the intracellular bacterial load, using the CFU counting (Maurin et al., 2000) or a qPCR method (Aloni-Grinstein et al., 2015); or (2) changes in cell lysis rate caused by bacterial multiplication, after staining cell monolayers with a vital dye (Sutera et al., 2014a) or by measuring the release of lactate dehydrogenase (LDH) (Madrid et al., 2013). The minimum inhibitory extracellular concentration (MIEC) is the antibiotic concentration in the cell culture supernatant allowing complete inhibition of bacterial growth or cell lysis.

Amoxicillin or ceftriaxone at 10 mg/L, or erythromycin, clarithromycin, or thiamphenicol at 4 mg/L, were ineffective against a F. tularensis in P388D1 cells (Maurin et al., 2000). Ciprofloxacin inhibited F. tularensis growth in P388D1 cells (at 1 mg/L) (Maurin et al., 2000) and in Vero cells (MIEC ranges of 0.06–0.25 mg/L) (Aloni-Grinstein et al., 2015). In MRC5 cells, ciprofloxacin, levofloxacin, and moxifloxacin were effective (MIEC of 0.064 mg/L, 0.125 mg/L, and 0.5 mg/L, respectively) (Sutera et al., 2014a). Gentamicin was found less effective (MIEC, 2 mg/L) in MRC5 cells and ineffective in Vero cells (MIEC > 20 mg/L) (Sutera et al., 2014a; Aloni-Grinstein et al., 2015). However, in a study allowing longer exposure of infected cells to gentamicin to facilitate its intracellular penetration, a bactericidal effect was observed at a concentration of 10 mg/L of this antibiotic (Maurin et al., 2000). Doxycycline displayed MIECs of 0.125–0.5 in Vero cells (Aloni-Grinstein et al., 2015) and 0.5 mg/L in MRC5 cells (Sutera et al., 2014a), and was also effective in P388D1 cells (Maurin et al., 2000). MIECs were 0.5 mg/L for rifampicin in MRC5 cells (Sutera et al., 2014a), and 1–2 mg/L for chloramphenicol in Vero cells (Aloni-Grinstein et al., 2015). Erythromycin displayed MIECs of 1–2 mg/L for two type B bvI isolates grown in MRC5 cells (Sutera et al., 2014a) and > 256 mg/L for the type B bvII LVS strain in Vero cells (Aloni-Grinstein et al., 2015). Azithromycin was bactericidal against the LVS strain at 5 mg/L in J774.1 cells, but only at 25 mg/L in human lung epithelial A549 cells, suggesting better penetration of this antibiotic in the former cells (Ahmad et al., 2010). Finally, the protection level of J774.1 cells infected with the SchuS4 strain was 100% for tetracycline at 50 μM (22 mg/L) and erythromycin at 50 μM (37 mg/L) and 92% for minocycline at 50 μM (23 mg/L) (Madrid et al., 2013).

Overall, data obtained in cell models are difficult to interpret because of methodological heterogeneity between studies. However, the fluoroquinolones had the strongest bactericidal activity against intracellular F. tularensis, gentamicin had a delayed bactericidal activity, and doxycycline was only bacteriostatic. Among other antibiotic classes, rifampicin displayed the lowest MIECs.

Table 6 summarizes data from tularemia case series with a predominance of the UG or GL forms. Many other sporadic UG or GL tularemia cases have been reported in several countries but with similar findings than those reported in case series (Pérez-Castrillón et al., 2001; Jackson et al., 2012; Sobolewska-Pilarczyk et al., 2014; Boone et al., 2015; Formińska et al., 2015; Calin et al., 2017; Pekova et al., 2017; Whitten et al., 2017; Haulrig et al., 2020; Kubiliute et al., 2021). The UG and GL forms are predominant worldwide, except in some countries such as Turkey (see below).

Beta-lactams (penicillin, amoxicillin, amoxicillin-clavulanate, cefuroxime, cefotaxime, ceftriaxone), macrolides (clarithromycin, clindamycin), and antituberculosis drugs often given as empirical treatments in UG and GL tularemia patients were ineffective (Maurin et al., 2011; Jackson et al., 2012; Sobolewska-Pilarczyk et al., 2014; Boone et al., 2015; Formińska et al., 2015; Pekova et al., 2017; Whitten et al., 2017; Haulrig et al., 2020; Kubiliute et al., 2021). Overall, gentamicin, ciprofloxacin (as well as levofloxacin and moxifloxacin) and doxycycline were the most successful treatments. However, several courses of antibiotic therapy were often prescribed because of poor clinical response and chronic evolution especially in the approximately 30% of UG or GL tularemia patients suffering from lymph node suppuration. The antibiotic therapy was adapted empirically according to the clinical and radiological findings. Treatment failure and relapse rates were usually high but variable between studies [e.g., 21.3% in Evans et al. (1985) and 38.6% in Maurin et al. (2011)]. Relapses were more frequent in patients with a significant treatment delay. For example, in the case series from Missouri (USA) (Weber et al., 2012), the mean treatment delay was 13 days (range, 0–82 days), but 21 days for patients evolving to lymph node suppuration. In a French series (Darmon-Curti et al., 2020), treatment delay was significantly longer (p = 0.006) for patients requiring surgery (mean, 52.1 days; range, 7–175 days) compared to those cured without surgery (mean, 32.1 days; range 1–87 days). In some studies, failure rates were higher for doxycycline versus ciprofloxacin (i.e., 4.5 versus 42.8%) (Pérez-Castrillón et al., 2001). Drainage or removal of suppurated lymph nodes were needed for cure in 12.4% up to 57.9% of patients (Maurin et al., 2011; Kızıl et al., 2012; Weber et al., 2012; Boone et al., 2015; Formińska et al., 2015). Many patients (29.7–58%) required hospitalization because of disease severity or more frequently the need for surgical intervention (Maurin et al., 2011; Weber et al., 2012; Darmon-Curti et al., 2020). However, severe cases requiring hospitalization are more likely to be reported than milder cases and thus might be overestimated in the literature. In some patients, complications occurred while under appropriate antibiotic therapy, including continued enlargement of the infected lymph nodes, involvement of new lymph nodes, regional and systemic complications (Jackson et al., 2012; Kızıl et al., 2012; Weber et al., 2012; Formińska et al., 2015; Calin et al., 2017; Haulrig et al., 2020; Kubiliute et al., 2021). A few patients with UG or GL forms died (2/418, 0.48% for the above sporadic cases and case series). Death usually occurred in patients with underlying health condition, e.g., a cardiovascular disease (Weber et al., 2012). Altogether, the above data emphasizes the need for specific medical and surgical treatment recommendations for patients suffering from UG or GL forms of tularemia, especially when experiencing lymph node suppuration.

At onset, OP tularemia looks like streptococcal pharyngitis leading to beta-lactam treatment. Due to oral route of infection, complications include lymph node suppuration, parapharyngeal abscess, digestive involvement, and systemic diseases. Large OP tularemia case series have been reported in Turkey where these infections predominate and occur as outbreaks mainly related to the consumption of contaminated water (Meric et al., 2008; Ulu-Kilic et al., 2013; Erdem et al., 2014; Gozel et al., 2014; Table 7). Many other small series or sporadic OP tularemia cases have been reported involving more than 200 patients (Arikan et al., 2003; Chitadze et al., 2009; Sencan et al., 2009; Komitova et al., 2010; Kızıl et al., 2012; Dentan et al., 2013; Eren Gok et al., 2014; Karakas et al., 2014; Uzun et al., 2015; Esmaeili et al., 2021; Binay et al., 2023).

Patients with the OP form of tularemia did not respond to empirical treatment with beta-lactams, macrolides or antituberculosis drugs (Arikan et al., 2003; Sencan et al., 2009; Willke et al., 2009; Cağlı et al., 2011; Erdem et al., 2014; Binay et al., 2023). Adapted antibiotic treatments most often included an aminoglycoside, ciprofloxacin, or doxycycline in descending order of prescription frequency. These antibiotics were often combined or given sequentially. For example, among antibiotic treatments given to 1,034 tularemia patients (including 85.3% OP) in Turkey, an aminoglycoside was prescribed in 36% of patients (28% for streptomycin), a fluoroquinolone in 20% (18% for ciprofloxacin), a tetracycline in 12.5% (12% for doxycycline), and a combined treatment in 28.8% (Erdem et al., 2014). Global failure rates were high but varied between studies, e.g., 30.9% for Ulu-Kilic et al. (2013), 37.9% for Meric et al. (2008), and 50.4% for Erdem et al. (2014). Relapse rates also varied between studies, e.g., 9.09% in Cağlı et al. (2011) and 78.9% in Kızıl et al. (2012). These variations might reflect differences in the recruited patients. Failures and relapses were less frequent when antibiotics were administered within 2–3 weeks of symptoms onset (Meric et al., 2008; Komitova et al., 2010). Doxycycline was associated with higher failure and relapse rates then aminoglycosides and ciprofloxacin (Meric et al., 2008; Erdem et al., 2014). For example, Meric et al. (2008) reported a global failure rate of 37.9% among 145 OP cases, including 9.6% each for aminoglycosides and fluoroquinolones, but 18.6% for tetracyclines. Lymph node suppuration was frequent [e.g., 41.2% in Erdem et al. (2014)]. Surgical drainage or excision of suppurated lymph nodes were required in about 31% (Karakas et al., 2014; Binay et al., 2023) up to 69.7% (Cağlı et al., 2011) of patients with OP tularemia. Among 254 patients (including 217 OP forms) reported as sporadic cases and in small case series (see references above), lymph node resection or drainage was performed in 51.2% (130/254) of patients. Lymph node excision was found more effective than drainage or needle aspiration of lymph node suppuration to prevent relapses (Kızıl et al., 2012). Although OP tularemia can evolve over several months and occasionally lead to severe complications, no fatal cases were reported in the above case series and sporadic cases.

Only sporadic OG tularemia cases have been reported (Pärssinen and Rummukainen, 1997; Komitova et al., 2010; Mahy et al., 2011; Altuntas et al., 2012; Eren Gok et al., 2014; Darmon-Curti et al., 2020). Patients usually suffered from painful unilateral conjunctivitis with lymphadenopathy (mainly in the periauricular and submandibular areas), which defines the Parinaud’s oculoglandular syndrome (Weber et al., 2012; Eren Gok et al., 2014). Diagnosis was often delayed because F. tularensis is only a rare cause of this syndrome and therefore evoked in the second line. Beta-lactams (e.g., amoxicillin-clavulanate or ampicillin-sulbactam) and macrolides (e.g., pristinamycin) were ineffective (Mahy et al., 2011; Altuntas et al., 2012; Darmon-Curti et al., 2020). Specific antibiotic treatment was usually an aminoglycoside or a fluoroquinolone for 10–14 days, or doxycycline for 2–3 weeks, or an antibiotic combination. Eren Gok et al. (2014) advocated adding a topical antibiotic treatment, including ciprofloxacin drops or tobramycin drops and ointment for 7 days. In most patients, ocular symptoms regressed under appropriate treatment within 7–10 days. Four (22.2%) of the 18 above cases required drainage or excision of suppurated lymph nodes. All cases recovered without sequelae.

The same treatment options were used for treatment of keratitis and dacryocystitis (Steinemann et al., 1999; Celik et al., 2013; Köse and Hoşal, 2019). Three-weeks treatment with doxycycline was effective to treat posterior uveitis in a 52-year-old man, but with loss of vision of the involved eye (Terrada et al., 2016). Although treatment of F. tularensis intraocular infection is not standardized, it should be emphasized that ciprofloxacin intravenously has more favorable intraocular pharmacokinetics properties than doxycycline and gentamicin (Kulshrestha et al., 1981; Mounier et al., 1988; Fiscella et al., 1998; Morlet et al., 2000).

Pneumonic tularemia is most often of mild to moderate severity (see above for severe type A pneumonia) (Table 8). Type B pneumonia typically have a prolonged course, with intermittent fever, mild respiratory symptoms, progressive weight loss, and development of mediastinal or hilar lymphadenopathy. Diagnosis is difficult and often delayed because of poor specificity of clinical and radiological findings. It should be suspected when pulmonary hypermetabolic lymph nodes are found on FDG-PET/CT scan in a clinical and epidemiological context compatible with tularemia (Martinet et al., 2021).

Pneumonic tularemia has been reported in case series (Centers for Disease Control and Prevention (CDC), 2009; Väyrynen et al., 2017; Darmon-Curti et al., 2020) or as sporadic cases (Matyas et al., 2007; Fritzsch and Splettstoesser, 2010; Mahy et al., 2011; Rimawi et al., 2014; Fachinger et al., 2015). For the above three case series, representing 116 PN cases, no death was reported. Most patients had respiratory symptoms and 50% up to 90% had abnormal lung radiological findings, including frequent hilar or mediastinal lymphadenopathy. Väyrynen et al. (2017) reported a delay in hospital admission of 1–45 days (mean 9.9 days) and a hospital stay of 0–24 days (mean, 8.6 days). Among 84 patients from two series, five patients (5.9%) required admission to an intensive care unit due to respiratory distress [ Centers for Disease Control and Prevention (CDC), 2009; Väyrynen et al., 2017]. In these series from the USA and Finland, respectively, a fluoroquinolone was administered in 41 and 77.6% of patients, a tetracycline in 49 and 15.5%, an aminoglycoside in 47 and 6.9%, and these antibiotics were combined in most patients. Complications included sepsis, pleural effusion and empyema, endocarditis, pericarditis, and splenic hematoma [ Centers for Disease Control and Prevention (CDC), 2009; Väyrynen et al., 2017; Darmon-Curti et al., 2020].

When considering case series and sporadic cases, PN tularemia most frequently occurred in male patients (Matyas et al., 2007; Fritzsch and Splettstoesser, 2010; Rimawi et al., 2014; Fachinger et al., 2015; Ughetto et al., 2015; Väyrynen et al., 2017, 2017). Empirical treatment with beta-lactams (sultamicillin, ceftriaxone) or macrolides (azithromycin, clindamycin) was ineffective. The antibiotic treatment varied depending on the clinical presentation (mode of onset, severity), underlying medical condition of the patient, and type of infection (A versus B). For type A infections, gentamicin was the first-line treatment, occasionally combined with doxycycline or a fluoroquinolone (Matyas et al., 2007). However, ciprofloxacin alone was successful in some patients with severe pneumonia (Rimawi et al., 2014). Doxycycline alone, often administered upon suspicion of Rocky Mountain Spotted Fever, ehrlichiosis, or Lyme disease was less effective (Matyas et al., 2007). Patients with a prolonged course of type B pneumonia recovered after receiving gentamicin, ciprofloxacin, or a combination of both (Fritzsch and Splettstoesser, 2010; Mahy et al., 2011; Fachinger et al., 2015). Doxycycline administered for 3 weeks was reported as either effective (Darmon-Curti et al., 2020) or poorly effective (Fachinger et al., 2015). In a few cases, patients recovered while receiving no appropriate antibiotic therapy (Ughetto et al., 2015; Darmon-Curti et al., 2020). Surgery was occasionally required for excision of mediastinal or hilar lymph nodes, or empyema or pleural effusion drainage [ Centers for Disease Control and Prevention (CDC), 2009; Rimawi et al., 2014; Fachinger et al., 2015; Darmon-Curti et al., 2020].

Typhoidal tularemia is often a severe life-threatening disease (see above) but can also correspond to less severe clinical manifestations, in particular for type B infections (Han et al., 2004; Maurin et al., 2011; Nakamura et al., 2018). This form has been often reported in patients with underlying medical conditions, especially immunosuppression (Han et al., 2004; Maurin et al., 2011; Nakamura et al., 2018). Although tularemia is rarely fatal, a large proportion of deaths occur in patients with a TY form (Han et al., 2004; Maurin et al., 2011; Nakamura et al., 2018). Treatment data are scarce because TY forms are often mixed with other forms in case series (Hofinger et al., 2009; Maurin et al., 2011; Weber et al., 2012; Darmon-Curti et al., 2020). Whatever the severity of the disease, most patients have F. tularensis bacteremia and should receive an aminoglycoside (gentamicin or streptomycin, for 10–14 days) as a first-line antibiotic (Dennis et al., 2001; Bossi et al., 2004). We previously suggested combining an aminoglycoside with a fluoroquinolone in patients with severe tularemia for rapid extracellular and intracellular efficacy because the intracellular penetration of aminoglycosides is slow (Caspar and Maurin, 2017), but the superiority of combined antibiotic therapy compared to monotherapy is not established. Fortunately, empirical treatment in patients with TY tularemia often combined a beta-lactam (amoxicillin clavulanate, cefoperazone, or imipenem) with an antibiotic active against F. tularensis such as an aminoglycoside or a fluoroquinolone (Han et al., 2004; Maurin et al., 2011; Nakamura et al., 2018). When an aminoglycoside was used, an oral relay with a fluoroquinolone or a tetracycline for 2 weeks was often considered. It is notable that relapses usually do not occur in patients with TY tularemia receiving appropriate antibiotic therapy.

Skin involvement in tularemia patients include primary lesions such as the inoculation eschars and secondary lesions (tularemids) that occur in approximately 10–20% of cases (Syrjälä et al., 1984; Evans et al., 1985; Cerný, 1994; Meric et al., 2008; Ulu-Kilic et al., 2013; Erdem et al., 2014; Gozel et al., 2014; Şenel et al., 2015; Polat et al., 2018). Secondary lesions might not be considered as complications, but frequently occur in patients with systemic infections (Syrjälä et al., 1984; Cerný, 1994; Polat et al., 2018). They include skin eruptions in 3–4% of patients (Erdem et al., 2014; Şenel et al., 2015; Polat et al., 2018), mainly of the macular or maculopapular type, or vesicular, vesiculopapular, or pustular types. Erythema multiforme has been reported in 2.8–11.3% of patients (Erdem et al., 2014; Şenel et al., 2015), and erythema nodosum in 2.6–13% of cases and mainly in women (Syrjälä et al., 1984; Akdiş et al., 1993; Erdem et al., 2014; Polat et al., 2018). Other skin manifestations include the Sweet syndrome [6.5% of 168 cases in Polat et al. (2018), urticaria (1.2–3.3% of cases (Şenel et al., 2015; Polat et al., 2018)], acne-like rash, Henoch-Schönlein purpura, and dermatitis (Syrjälä et al., 1984; Erdem et al., 2014; Şenel et al., 2015; Polat et al., 2018). Skin lesions can be combined (Polat et al., 2018). The skin lesions usually require no specific treatment and disappear quickly (a few days to 2–3 weeks) after the administration of streptomycin, gentamicin, doxycycline or a fluoroquinolone (Akdiş et al., 1993; Erdem et al., 2014; Marquart and Clifford, 2015; Şenel et al., 2015; Holland and Michelow, 2016).

Soft tissue infections in tularemia patients are rare and mainly include cellulitis and subcutaneous abscesses (Junkins and Snyder, 2011; Şenel et al., 2015; Arslan et al., 2016; Whitsell and Becker, 2020; Kreutzmann et al., 2021). They may occur through direct inoculation or hematogenous spread of F. tularensis to the skin or extension of a lymph node suppuration to the adjacent tissue.

A 49-year-old woman with facial cellulitis after a scalp tick bite was treated with amoxicillin-clavulanate plus gentamicin, then doxycycline plus ceftriaxone due to suspicion of Lyme disease (Arslan et al., 2016). After treatment failure, streptomycin was given for about 1 week for suspected tularemia with full recovery. A hand abscess with axillary and epitrochlear lymphadenopathy developed after a cat bite in a 66-year-old woman despite pasteurellosis prophylaxis with amoxicillin-clavulanate (Whitsell and Becker, 2020). She was cured after surgical drainage of the abscess and doxycycline treatment for 1 month. A 15-year-old boy developed OG tularemia with submandibular lymph node infection extending to the parotid gland and masseter muscle (Kosker et al., 2013). He was cured after suppurated lymph node excision and treatment with streptomycin (10 days) and tetracycline (8 weeks). A 60-year-old man with OG tularemia and orbital cellulitis did not improve under sulbactam-ampicillin treatment (Kreutzmann et al., 2021). He was cured after excision of suppurated cervical lymph nodes and treatment with ciprofloxacin relayed by doxycycline because of side effects.

Approximately twenty cases of tularemia meningitis have been reported worldwide, mostly in the United States (Hofinger et al., 2009; Venkatesan et al., 2020). It was usually an inaugural manifestation of tularemia and diagnosis was based on F. tularensis isolation from the cerebrospinal fluid (CSF) (Gangat, 2007; Hofinger et al., 2009; Contentin et al., 2011; Venkatesan et al., 2020). This disease is spontaneously fatal (Venkatesan et al., 2020) and empirical treatments for meningitis (e.g., ceftriaxone, amoxicillin, vancomycin, and acyclovir) are ineffective (Weiner et al., 2004; Hofinger et al., 2009; Contentin et al., 2011; Venkatesan et al., 2020). Brain involvement has been rarely reported, including rhombencephalitis (Barbaz et al., 2013), meningitis with cerebral microabscesses (Gangat, 2007), and central nervous system vasculitis (Çoban et al., 2021).

Streptomycin, gentamicin, doxycycline, ciprofloxacin and chloramphenicol have been most frequently used for treating tularemia meningitis patients (Lovell et al., 1986; Hofinger et al., 2009; Contentin et al., 2011; Venkatesan et al., 2020; Ducatez et al., 2022). These antibiotics were often administered in combination or successively for a total duration of 2 to 8 weeks (Lovell et al., 1986; Hofinger et al., 2009; Contentin et al., 2011; Venkatesan et al., 2020; Ducatez et al., 2022). A 68-year-old man with tularemia meningitis was cured after receiving thiamphenicol for 6 days, then gentamicin for 18 days, then ciprofloxacin for 4 weeks (Contentin et al., 2011). In contrast, a 64-year-old man was cured after only 2 weeks of ciprofloxacin. Although tularemia meningitis treatment is not standardized, most cases occurring over the past two decades have recovered without sequelae while receiving variable antibiotic treatments (Gangat, 2007; Hofinger et al., 2009; Contentin et al., 2011; Venkatesan et al., 2020; Ducatez et al., 2022). The use of phenicols for tularemia meningitis seems no longer justified due to the risk of fatal aplastic anemia. These antibiotics might be useful in case of brain involvement.

Francisella tularensis endocarditis is a rare complication of tularemia and mostly occurred in men older than 55 years (Tancik and Dillaha, 2000; Salit et al., 2013; Yeom et al., 2015; Gaci et al., 2017; Olivo et al., 2019; Darmon-Curti et al., 2020; Kaeppler et al., 2020). F. tularensis infection involved native cardiac valves (Tancik and Dillaha, 2000; Salit et al., 2013; Gaci et al., 2017), a prosthetic valve (Kaeppler et al., 2020), or a pacemaker (Gaci et al., 2017). Beta-lactams (amoxicillin-clavulanate, piperacillin-tazobactam, cefotetan, cefazolin, ceftriaxone), erythromycin, and vancomycin were ineffective, including in some patients receiving gentamicin concomitantly (Tancik and Dillaha, 2000; Salit et al., 2013; Gaci et al., 2017; Kaeppler et al., 2020). Thus, the first-line treatments for endocarditis caused by Staphylococcus or Streptococcus species are not suitable for tularemia endocarditis. Most patients with tularemia endocarditis received gentamicin intravenously for at least 2 weeks. Ciprofloxacin or another fluoroquinolone was often combined to gentamicin or administered orally as a relay. The total duration of the specific treatment varied from 6 weeks to up to 12 weeks (Olivo et al., 2019). In some patients complications occurred while on appropriate antibiotic therapy, including a pulmonary embolism despite 14 days of intravenous gentamicin and ciprofloxacin (Olivo et al., 2019). Interestingly, in many patients, cardiac valve vegetations were still detected by echocardiography up to 6 months after the end of treatment (Salit et al., 2013; Olivo et al., 2019; Kaeppler et al., 2020). In one patient with severe complications (e.g., pulmonary embolism) and poor compliance to the antibiotic therapy, mitral valve replacement was performed. In other patients, persistence of cardiac vegetations was not associated with relapse and patients recovered without complications (Salit et al., 2013; Kaeppler et al., 2020). Although no definite recommendation can be raised from these anecdotal tularemia endocarditis cases, a 2-week course of intravenous gentamicin alone or combined with intravenous fluoroquinolone followed by 4 weeks oral administration of a fluoroquinolone seems a good alternative in the absence of severe complications. Cardiac surgery was rarely performed and mainly justified by a high risk of hemodynamic failure. Most patients were infected with F. tularensis type B and no death was reported.

Two cases of F. tularensis aortitis have been reported (Briere et al., 2016; Darmon-Curti et al., 2020; Kuzmova et al., 2023). A 85-year-old man developed fever and a systolic murmur due to minimal aortic insufficiency (Briere et al., 2016). Enhanced CT and FDG-PET/CT scans revealed an inflammatory infrarenal aortic aneurysm. After 5 days of vancomycin plus gentamicin, doxycycline was administered due to F. tularensis positive blood cultures. Nineteen days later, part of the aorta was replaced by an aorto-aortic allograft. F. tularensis was PCR detected in the removed aortic tissues. The patient further received 3 months of doxycycline plus levofloxacin and was considered cured. An aortic endograft infection was reported in a 79-year-old man (Kuzmova et al., 2023). F. tularensis was isolated from the aortic and periaortic tissues. He was cured after replacing the aortic graft and gentamicin plus ciprofloxacin treatment for 2 weeks then ciprofloxacin for 10 weeks

Clary et al. (2022) reported a finger interphalangeal joint arthritis caused by Francisella tularensis inoculated through a cat bite. The patient was cured after receiving gentamicin for 4 weeks. Most reported osteoarticular infections involved knee or hip prostheses, in persons aged 49–84 years, and more frequently men than women. These infections occurred from 1 month up to 25 years after surgical placement of the prosthesis (Cooper et al., 1999; Rawal et al., 2017; Chrdle et al., 2019; Azua and Voss, 2020; Darmon-Curti et al., 2020; Ponderand et al., 2023). Suspected modes of infection included cleaning a rabbit barn (Chrdle et al., 2019), oral contamination (abdominal symptoms), gardening (Chrdle et al., 2019), respiratory infection (Ponderand et al., 2023), skin abrasion after falling (Darmon-Curti et al., 2020), and tick bite (Cooper et al., 1999). Symptoms mainly included fever, joint pain, and difficulty in moving the infected joint. Diagnosis was often delayed of several weeks to months (Cooper et al., 1999) and usually established by F. tularensis isolation from osteoarticular or blood samples. Local complications were occasionally observed [e.g., hip hematoma (Darmon-Curti et al., 2020)], but no systemic complication which might be related to the fact that only F. tularensis type B strains were involved. Cure was usually obtained by combining an antibiotic treatment with surgical replacement of the infected joint prosthesis. Four patients were reported to be cured after doxycycline treatment (Rawal et al., 2017; Chrdle et al., 2019; Azua and Voss, 2020; Escovar et al., 2023). One patient was cured after receiving doxycycline for 6 weeks plus knee prosthesis replacement (Chrdle et al., 2019). Another patient denied surgery and was considered cured only after receiving doxycycline for 1 year (Rawal et al., 2017). The last patient also received a long term treatment with doxycycline (Azua and Voss, 2020). A patient with infected knee prosthesis was administered a lifelong doxycycline treatment because the prosthesis could not be totally removed (Ponderand et al., 2023). A patient with knee prosthesis infection who denied surgical treatment relapsed after receiving 10 days of gentamicin and 21 days of doxycycline, but then was cured with ciprofloxacin for 20 days (Chrdle et al., 2019). Some patients received an antibiotic therapy (e.g., doxycycline for 3 months, gentamicin, or ofloxacin for 6 weeks), then the infected prosthesis was replaced and antibiotics were further administered after surgery (e.g., ciprofloxacin for 3 months, ciprofloxacin plus doxycycline for 9 weeks, or amikacin for 5 days plus ciprofloxacin for 6 weeks) (Darmon-Curti et al., 2020; Ponderand et al., 2023). A complex case was reported in a 68-year-old man with a knee arthroplasty because of rheumatic arthritis (Cooper et al., 1999). The patient first experienced a postoperative prosthesis infection caused by Enterococcus faecalis. One year later, knee inflammation led to prosthesis removal with insertion of a joint spacer with erythromycin-containing cement. After several antibiotic courses with poor clinical response his symptoms improved under ciprofloxacin plus rifampicin combination. A new total knee arthroplasty was placed 3 months later. Diagnosis of tularemia was available at that time likely after several months of evolution of the disease. The patient further received 6 months of ciprofloxacin plus rifampicin and was considered cured. Overall, the therapeutic management of F. tularensis prosthetic joint infections was highly variable between patients, reflecting diagnostic difficulties and the need for treatment adaptation to the context and clinical evolution of each patient. However, all patients were considered cured from their F. tularensis infection.

Many other complications can occur in tularemia patients, including skin necrosis (Syrjälä et al., 1984), tonsillar phlegmon and otitis (Gürkov et al., 2009; Guerpillon et al., 2016), myocarditis and pericarditis (Landais et al., 2008; Frischknecht et al., 2019; Darmon-Curti et al., 2020), and pleurisy and pulmonary abscess (Gill and Cunha, 1997; Thomas and Schaffner, 2010). The cases reported for each type of complication are too few to deduce a therapeutic approach.