Clin Microbiol Infect 2011; 17: 306–311
Skull-base osteomyelitis (SBO) occurs secondary to invasive bacterial and fungal infection. Distinguishing between fungal and bacterial aetiologies of SBO has significant therapeutic implications. An 18-year (1990–2007) retrospective review of patients with SBO presenting to Westmead Hospital was performed. Epidemiological, clinical, laboratory and radiology data were collated. Twenty-one patients (median age 58 years) with SBO were identified: ten (48%) had bacterial and 11 (52%) had fungal SBO. Diabetes mellitus (57%) and chronic otitis externa (33%) were the most frequent co-morbidities; immunosuppression was present in five cases (24%). Cranial nerve deficits occurred in ten (48%) patients. The commonest pathogens were Pseudomonas aeruginosa (50% bacterial SBO) and a zygomycete (55% fungal SBO). Compared to bacterial SBO, fungal SBO was more frequently associated with underlying chronic sinusitis, sinonasal pain, facial/periorbital swelling and nasal stuffiness or discharge and the absence of purulent ear discharge (all p <0.05). Bacterial SBO was more frequently associated with deafness, ear pain or ear discharge (all p <0.05). Median time to presentation was longer in patients with bacterial SBO (26.3 weeks vs. 8.1 weeks, p 0.08). Overall 6-month survival was 88% (14/18 patients). All four deaths occurred in patients with fungal SBO. Immunosuppression was a risk factor for death (p <0.05). Early diagnostic sampling is recommended in patients at increased risk of fungal SBO to enable optimal antimicrobial and surgical management.
First described in 1959 by Meltzer and Kelemen , skull-base osteomyelitis (SBO) is an uncommon condition associated with significant morbidity and mortality [2–4]. Subsequently, our understanding of SBO is derived from case reports or small series in the surgical literature [1–15]. The entity has been described most often as a complication of malignant otitis externa secondary to Pseudomonas aeruginosa infection [5,16]. Increasing age, diabetes mellitus and microvascular disease are common risk factors [2,9].
SBO, however, may also occur in the absence of malignant otitis externa and with pathogens other than P. aeruginosa, including fungi [3,7,8,12–14,17]. Fungal SBO has been reported to be mostly caused by Aspergillus, and less commonly, Scedosporium spp. [3,7,12]. Although underlying diabetes and primary or acquired immunodeficiencies have often been evident , fungal SBO has also occurred in the absence of these traditional risk factors [10,12,15].
Early diagnosis, identification of the causative pathogen(s), prompt initiation of appropriate antimicrobial or surgical therapies, and continuation of therapy for an adequate period are essential when managing SBO. Identification of the pathogen often requires surgical biopsy. Because this may be delayed for medical or technical reasons, clinical features or risk factors that discriminate between SBO caused by bacteria and that due to fungi could guide selection of empiric antimicrobial therapy pending definitive diagnosis. Using data from cases at a single institution over 18 years, we compared the epidemiology and clinical characteristics of bacterial and fungal SBO, aiming to identify unique risk factors and clinical associations.
Materials and Methods
A retrospective study of SBO was undertaken at a university hospital over 18 years (1990–2007). Cases were identified following interrogation of medical records using the International Statistical Classification of Diseases and related Health Problems definitions, 10th revision, Australian modification . Key words included: ‘base of skull’, ‘osteomyelitis’, ‘malignant otitis externa’, ‘otitis media’, ‘sinusitis’, ‘diseases of the petrous temporal bone’ and ‘invasive mycoses’. The Microbiology, Otorhinolaryngology and Histopathology databases were also queried for cases of SBO. Approval for the study was obtained from Sydney West Area Health Service Human Research Ethics Committee.
For each patient, clinical information was recorded on a standardized form and included: patient demographics, co-morbidities and predisposing factors within the preceding 90 days, likely source of infection (ear, sinus, other), clinical features, results of microbiological and histopathological investigations, treatment, hospital length of stay and clinical outcome at 6 months. Results of computed tomography (CT), magnetic resonance imaging (MRI) or combined technetium Technitium-99m/gallium citrate (Ga)]-67 scans were assessed by one author (LG), who was blinded to the other results.
Proven (definite) SBO was defined as skull-base infection in patients with localizing symptoms/signs at presentation who had: (i) radiological or scintigraphic features indicative of bone erosion and/or infection and (ii) isolation and/or visualization of the pathogen from the affected bone(s) and surrounding tissue. Probable SBO was defined as infection in patients with localizing symptoms/signs with evidence of SBO on imaging studies, but from whom a pathogen was recovered from clinical specimens other than bone or tissue (e.g. ear swabs) or in whom a definite response to antimicrobial therapy was evident. The primary source of the pathogen was assigned according to patient clinical features in the context of accompanying microbiological and radiological results.
Clinical data were analyzed using spss, version 15.0.0 (SPSS Inc., Chicago, IL USA). Variables associated with SBO caused by bacteria were compared with those associated with fungal SBO. Analyses were also performed to examine risk factors, presenting symptoms, causative pathogen, and treatments administered associated with survival at 6 months. Univariate analyses were performed using a Student’s t-test (for continuous variables) or the chi-square or Fisher’s exact tests (for categorical variables); p <0.05 was considered statistically significant.
Demography and predisposing factors
From over 500 patients identified through the search, 21 patients met the case definition of SBO; 15 had proven SBO and six had probable infection. The mean patient age was 58 years (range 26–80 years); 66.7% were male.
Diabetes mellitus was the most frequent predisposing factor (12 patients; 57%) followed by chronic otitis externa (33%) and underlying chronic sinusitis (29%; Table 1). Of five immunosuppressed patients, three had rheumatoid arthritis and were receiving corticosteroids alone, methotrexate alone, or corticosteroids, methotrexate and adalimumab (each n = 1); one renal transplant patient was receiving cyclosporine and azathioprine, and a haematopoietic stem cell transplant recipient was receiving corticosteroids.
|Bacterial SBO (n = 10)||Fungal SBOa (n = 11)||Significance|
|Mean age, years (range)||56 (26–80)||56 (32–72)||NS|
|Diabetes||4 (40.0%)||8 (72.7%)||NS|
|Chronic otitis media||4 (40.0%)||2 (18.2%)||NS|
|Chronic otitis externa||5 (50.0%)||2 (18.2%)||NS|
|Chronic sinusitis||0||6 (54.5%)||p <0.02|
|Immunosuppression||1 (10.0%)||b4 (36.4%)||NS|
|Trauma or surgery||3 (30.0%)||1 (9.1%)||NS|
|Mean time to presentation weeks (95% CI)||26.3 (4.2–28.5)||8.1 (0.6–15.5)||NS (p 0.08)|
|Fever||3 (30.0%)||6 (54.5%)||NS|
|Headache||4 (40.0%)||8 (72.7%)||NS|
|Deafness||7 (70.0%)||2 (18.2%)||p 0.03|
|Ear pain||8 (80.0%)||2 (18.2%)||p <0.009|
|Ear discharge||8 (80.0%)||1 (9.1%)||p <0.002|
|Sinonasal pain||1 (10.0%)||8 (72.7%)||p <0.008|
|Facial or periorbital swelling||1 (10.0%)||7 (63.6%)||p 0.024|
|Nasal stuffiness or discharge||3 (30.0%)||9 (81.8%)||p 0.03|
|Cranial nerve involvement||5 (50.0%)||5 (45.5%)||NS|
|Eye/orbit infection||0||2 (18%)||–|
|Brain parenchymal infection||0||1 (9%)||–|
|Site of original infection|
|Ear disease||8 (80.0%)||2 (18.2%)||p <0.01|
|Sinus disease||0||9 (81.8%)||p <0.001|
|Traumatic or uncertain||2 (20.0%)||0||NS|
|Surgery||5 (50.0%)||11 (100%)||p <0.02|
|Antibacterial Therapy||10 (100%)||6 (54.5%)||p <0.04|
|Antifungal therapy||0||11 (100%)||p <0.001|
|Hyperbaric oxygen Therapy||0||2 (18.2%)||NS|
|Survival (6 months)||7/7 (100%)||7/11 (63.6%)||NS (p 0.12)|
|Disease-free survival (6 months)||5/7 (71.4%)||7/11 (63.6%)||NS|
Clinical features and site of bone infection
The median time from symptom onset to SBO diagnosis was 7 weeks (range 1 week–2 years).The most frequent clinical features are summarized in Table 1. Cranial nerve palsies were present in ten (48%) patients affecting the facial (n = 7 patients), ophthalmic (n = 7), optic (n = 4), ophthalmic, trochlear or abducens (each n = 4), and glossopharyngeal (n = 4), nerves. Infection extending into the orbit/eye and brain was uncommon (Table 1).
All patients had radiographic or scintigraphic evidence of SBO (CT, n = 17; MRI, n = 11; Tm-99/Ga-67 scintigraphy, n = 16). Three patients had evidence of bilateral disease (14%). The petrous temporal bone was most frequently involved (13/21, 62% cases; Fig. 1) followed by the maxillary sinus wall (7/21; 33%), sphenoid or clivus (5/21; 24%), nasal bones (5/21; 24%), ethmoid or sphenoid sinus wall (4/21; 19%) and squamous temporal bone (2/21; 10%).
The likely source of infection was determined in 20 patients (95%); ten were considered to have had a primary ear infection (including mastoiditis), and nine had primary sinus infection. One patient developed infection post trauma (Table 2).
|Site of disease||Bacterial SBO (n = 10)||Fungal SBO (n = 11)|
|Proven bacterial (n = 4)||aProbable bacterial (n = 6)||Fungal (n = 10)||Mixed bacterial and fungal (n = 1)|
|Ear||Pseudomonas aeruginosa and Staphylococcus aureus (n = 1)|
P. aeruginosa (n = 1)
bAcinetobacter spp. (n = 1)
|c5||Scedosporium apiospermum (n = 2)|
|Sinus||–||–||Aspergillus fumigatus (n = 1)|
Aspergillus flavus (n = 1)
Rhizopus spp. (n = 2)
Mucor spp. (n = 1)
dZygomycete (n = 2)
Dematiaceous fungus (n = 1)
S. aureus and
dZygomycete (n = 1)
|Traumatic inoculation||P. aeruginosa (n = 1)||–||–||–|
Microbiological and histopathological findings
Clinical specimens (e.g. sinus material, pus from ear) were obtained for culture from all patients, including 15 patients who underwent surgical biopsy. Biopsy specimens (bone, 15/15; sinus tissue, 7/15; periorbital tissue, 1/15; ear tissue, 1/15) all demonstrated tissue necrosis with or without visible microorganisms.
Overall, a bacterial pathogen was cultured from seven patients, and a fungal pathogen was cultured, or visualized, in 11 cases (Table 2). Of six patients with probable bacterial SBO, two had P. aeruginosa cultured from purulent ear material, whereas Gram-negative rods were seen in the ear fluid of another. A single patient had mixed bacterial and fungal SBO (Table 2) and was included in the ‘fungal SBO’ group for analyses (see below). Thus, there were ten patients with bacterial SOB and 11 with fungal infection (Table 1).
P. aeruginosa was the most frequent bacterial pathogen (5/10 cases of bacterial SBO, one case of mixed infection; Table 2). Zygomycetes were the most common fungal pathogen (n = 6); in three patients, broad pauci-septate hyphae were seen in bone but no fungus was cultured. There were two cases each due to Scedosporium apiospermum and Aspergillus spp. Of six patients who had blood cultures, one yielded Acinetobacter spp.; the organism was also cultured in this case from pus/tissue from the external ear.
Comparison of bacterial and fungal SBO
Predisposing factors for bacterial and fungal SBO were similar, although a greater proportion of patients with fungal SBO had underlying chronic sinusitis (55% vs. 0%; p <0.02) (Table 1). Thirty-six percent of fungal SBO patients were immunosuppressed compared to 10% of those with bacterial SBO (p: not significant).
There was a trend towards more rapid presentation with fungal SBO (mean time to presentation 8 weeks vs. 26 weeks for bacterial SBO; p 0.08; Table 1). Other differences in presenting symptoms/signs were also observed between the two patient groups: deafness and ear pain or purulent discharge were associated with bacterial SBO (p ≤0.03), whereas sinonasal pain, facial or periorbital swelling and nasal stuffiness/discharge were more frequent in fungal SBO (p ≤0.03). No significant differences in the frequency or magnitude of fever, or in the proportion of cranial nerve deficits, were identified (Table 1).
Overall, bacterial SBO was more likely than fungal SBO to have arisen from an otogenic focus of infection (80% vs. 18.2%; p <0.01). Conversely, fungal SBO was more likely to have originated from the sinuses (82% vs. 0%; p <0.001; Table 1). Absence of ear discharge was the most sensitive predictor of fungal SBO, with a sensitivity of 91% and specificity of 80%, whereas the presence of sinonasal pain yielded a sensitivity of 73% and specificity of 90% for fungal SBO.
Treatment and outcomes
Surgical debridement (n = 16) was performed more frequently with fungal SBO (100% vs. 50%, p <0.02). Antibacterial therapy was administered to all patients with bacterial SBO and six with fungal SBO prior to diagnosis. The median length of antibacterial therapy was 12 weeks [bacterial SBO, 13 weeks (range 4–26 weeks); fungal SBO, 5 weeks (range 1–18 weeks)]. Fifteen patients received an anti-pseudomonal β-lactam antibiotic (ticarcillin/clavulanic acid (n = 10) or ceftazidime (n = 5), 11/16 ciprofloxacin and 5/16, vancomycin. All patients with fungal SBO received antifungal therapy [lipid amphotericin B formulation (7/11), itraconazole (2/11), voriconazole (3/11), posaconazole (4/11) and caspofungin (1/11)]. The median duration of therapy was 28 weeks. The median length of hospital stay for 19 patients was 3 weeks (bacterial SBO, 2.6 weeks; fungal SBO, 5 weeks; p: not significant).
Six-month survival was 77% (14/18 patients); three patients with bacterial SBO were lost to follow-up. Four deaths occurred, all in patients with fungal SBO (median time from diagnosis to death: 8 weeks; range 2–18 weeks). The cause of death in all cases was progressive fungal disease; secondary haemorrhage from mycotic aneurysms [internal carotid artery (n = 1), superior cerebellar artery (n = 1)] was contributory in two cases. Cranial nerve abnormalities persisted in all patients. The median time from symptom onset to initiation of therapy was 5 weeks (range 1–13 weeks) in those who died compared to 11 weeks in survivors (range 2–53 weeks; p: NS). Immunosuppression was the only factor associated with decreased survival by univariate analysis (40% vs. 92% in non-immunosuppressed patients, p <0.05). Fungal SBO per se was not associated with decreased survival (Table 1).
SBO, usually a complication of uncontrolled otogenic, odontogenic or sinus infection, is uncommon [3,11,14]. Large, adequately powered epidemiological studies have not been published. The prevalence of SBO and the relative contribution of bacteria and fungi are difficult to estimate from previous case series as a result of small patient numbers, the frequent inclusion of patients with malignant otitis externa without confirmed SBO and variation in the diagnostic methods used [6,9,16]. The present study, using strict case definitions for SBO, reveals that fungi accounted for a significant proportion (approximately 50%) of SBO in contemporary hospital practice. We have also identified clinical variables associated with fungal, or bacterial, SBO that may guide initiation of early diagnostic sampling and/or selection of empiric antimicrobial therapy.
The significant morbidity of SBO in the present study is consistent with previous reports [3,4,6]. Almost half (48%) of patients had persistent cranial nerve abnormalities, comparable to that reported elsewhere (21–43%) [3,6,19]. Although extension into brain was uncommon [6, present study], cerebral involvement has been associated with high mortality in reported cases despite surgical intervention; the single patient in our series succumbed to fungal infection before surgery could be considered. A large proportion of patients suffered disfiguring surgery with an attendant prolonged hospital stay (median 3 weeks). As noted in other studies [3,6,16], diabetes mellitus and chronic ear disease were predisposing factors. However, confirmation of underlying chronic sinusitis as a risk factor for SBO is required.
Otogenic P. aeruginosa infection accounted for 50% of bacterial SBO. This is consistent with the adage that SBO complicating malignant otitis externa is almost uniformly caused by this bacterium [16,20]. However, other bacteria, and fungi are also important causes of SBO [3,6]. Fungal SBO is increasingly reported in the literature [3,7,10,12,15,17,21], although no systematic long-term epidemiological studies have been conducted. This apparent rise may reflect the increasing use of immunosuppressive therapies, in parallel with the rise of other forms of invasive fungal disease (IFD) [21,22]. Indeed, four of five immunosuppressed patients in our study had fungal SBO (Table 1). Importantly, however, fungal SBO may also occur in immunocompetent individuals [10,15, present study] and should be considered in all patients presenting with symptoms/signs of SBO, especially given evidence from other forms of IFD that delay in antifungal therapy results in increased mortality [23,24].
Most cases of fungal SBO have been due to Aspergillus or Scedosporium spp., reportedly arising from contiguous spread of ear infection [7,10,12,17], although one report noted fungi to be a rare cause of invasive otitis externa . In comparison, we observed that fungal SBO occurred primarily as a result of antecedent sinus infection and that zygomycetes were the most frequent pathogen. The reasons for the relative prevalence of zygomycetes are not readily apparent but are of interest. Zygomycetes are well-known pathogens of invasive fungal sinusitis [3,17,21] and, as such, may have influenced the aetiology of SBO. Although not statistically significant, the majority (>70%) of fungal SBO patients had diabetes, which is a risk factor for zygomycosis . By using strict case definitions to determine SBO, we also excluded instances of ‘colonization’, typically with fungi (e.g. Aspergillus spp.) other than zygomycetes. It is possible that local institutional epidemiology may have influenced the prevalence of specific fungi, although there are no data to support this.
Fungal SBO is often considered only following failure of antibacterial therapy . Kountakis et al.  noted that only one of ten patients with Aspergillus SBO had received antifungal therapy prior to surgical biopsy. In our study, 55% of patients with fungal SBO were treated with antibacterial drugs for a median of 5 weeks prior to diagnosis. It remains uncertain how long one should wait prior to performing diagnostic sampling.
Previous data attempting to identify risk factors for the development of SBO have not distinguished between disease caused by different microorganisms. Although the number of cases included in the analysis is small and therefore the results need to be interpreted with caution, our analyses suggest that there may be differences in clinical risk factors and associations for bacterial and fungal SBO. Patients with fungal SBO were more likely to have underlying chronic sinus disease, symptoms attributable to invasive sinus infection (sinofacial pain, periorbital swelling, nasal stuffiness/discharge), but with a relative paucity of features attributable to ear infection. Indeed, the absence of purulent ear discharge was a sensitive (91%) predictor of underlying fungal SBO (Table 1). These associations warrant further investigation through larger prospective multi-centred epidemiological studies. Nonetheless, the results may assist in identifying patients most likely to benefit from early antifungal therapy. Given the diversity of fungal pathogens identified, diagnostic sampling should be performed, where possible, prior to commencing therapy. However, because no clinical association was 100% sensitive or specific for presence of fungal SBO, clinical failure with antibacterial therapy should also prompt similar considerations.
Antimicrobial therapy in SBO is targeted to eradication of the causative pathogen. Because P. aeruginosa infection predominates in most case series of bacterial SBO, including the present study, initiation of antibiotics with activity against P. aeruginosa is appropriate pending microbiological diagnosis. In the absence of local epidemiological data to guide empiric antifungal therapy, agents with activity against all major pathogens are necessary. Because zygomycetes were responsible for more than 50% of fungal SBO in our series, the use of regimens including high-dose amphotericin B formulations is advised pending definitive diagnosis. The role of surgical resection is likely also influenced by pathogen. Early surgery is associated with improved survival in patients with zygomycosis . Aggressive surgical debridement is recommended in fungal SBO  but is probably unnecessary in patients with bacterial SBO [6,20].
The limitations of this small retrospective study are recognized. The number of patients accessible was insufficient to fully examine significant differences. Longitudinal data were not collected beyond 180 days. The study, however, used rigorous inclusion criteria and the frequent use of biopsy to confirm SBO and identify causative pathogens was evident.
In conclusion, the present study highlights the importance of pathogens other than P. aeruginosa, including fungi, as causes of SBO. Certain clinical variables may enable recognition and differentiation of fungal from bacterial SBO, which may assist clinicians in identifying patients who are at increased risk of either type of infection and who may benefit from early diagnostic sampling.
The authors have no financial disclosures to make and declare that they have no conflicts of interest.