Malignant otitis externa: What is the role of surgery? Authors: Lisa Zhang, Joseph Bonanno, Woo Yul Byun, Yin Ren Published: 2024 ABSTRACT AbstractObjectiveMalignant otitis externa (MOE) is typically managed with long‐term broad‐spectrum antibiotics. The impact of surgical intervention on clinical outcomes is not well described. This study aims to compare clinical outcomes of MOE patients managed with or without surgery.Study Design/SettingRetrospective cohort, academic tertiary referral center.MethodsPatients diagnosed with MOE between January 2010 to September 2022 were included. Univariate analyses compared symptoms at initial presentation and long‐term (≥1 year) outcomes between surgical and non‐surgical patients.ResultsA total of 23 patients were included (78% male, mean age 69 ± 13 years, median follow‐up 305 days). Most (N = 22, 96%) patients were diabetic. Seventeen (74%) underwent surgical intervention (76% tympanomastoidectomy, 24% external auditory canal debridement and biopsy). Poor facial nerve (FN) function at initial presentation (defined as House‐Brackmann [HB] grade ≥3) significantly predicted undergoing surgical intervention (p = 0.02). Comparing surgically managed versus non‐surgical patients at the time of presentation, there were no differences in the degree of hearing loss, severity of diabetes, rate of insulin dependence, incidence of immunosuppression, or the Charlson Comorbidity Index (all p > 0.05). FN outcomes at long‐term follow‐up also did not significantly differ (p > 0.05). No significant differences in the length of stay (9 vs. 6 days, p = 0.2), rate of readmission (31% vs. 17%, p = 0.5) or 5‐year overall survival (53% vs. 66%, p = 0.6) were observed between surgical and non‐surgical patients.ConclusionsLong‐term outcomes for patients with MOE remain poor. Patients with poor FN function at presentation were more likely to undergo surgical intervention. Patient comorbidities, including the severity of diabetes, were not predictive of undergoing surgery. However, surgical intervention for MOE did not appear to lower the length of stay, the rate of hospital readmission, or overall mortality in our cohort.Level of EvidenceIIIKeywords: malignant otitis externa, skull base osteomyelitis, surgery INTRODUCTION Malignant otitis externa (MOE), also known as skull base osteomyelitis or necrotizing otitis externa, is a severe and life‐threatening infection that typically begins in the external auditory canal soft tissue and rapidly progresses to involve the lateral skull base and critical neurovascular structures. Traditionally associated with patients who are elderly, immunocompromised or have poorly controlled diabetes, MOE has historical rates of morbidity as high as 50% with additional long‐term neurological sequelae in an additional 30% of patients. 1 , 2 , 3 , 4 , 5 Previous studies have noted the predominance of two microbial organisms to be causative, Staphylococcus aureus and Pseudomonas aeruginosa, with the later traditionally associated with upwards of 90% of infections and thought to be related to the immunocompromised state. 6 , 7 Standard of care for MOE, which have only been done in the inpatient setting, typically focuses on reversing underlying immunosuppression, stringent glycemic control for diabetic patients, and concurrent treatment with broad‐spectrum, culture‐directed systemic antibiotics for at least 6 weeks. 8 , 9 Although outcomes for MOE have improved over the years, mortality rate of MOE continues to range between 10–20% with additional persistent chronic pain and permanent cranial neuropathy in survivors despite modern aggressive interventions. 1 , 10 , 11 However, the relative rarity of this condition, the non‐specificity of clinical presentation which typically include severe otalgia and otorrhea, and the tremendous variability of clinical management protocols across institutions have resulted in the literature to be dominated by single‐institutional studies with small heterogeneous patient cohorts, which presents additional challenges to standardizing treatment regimens for MOE and improving patient outcomes. In addition to medical management with anti‐pseudomonal antibiotics, a subset of patients may undergo surgical intervention. Although there is no standardized guideline to determine which patient requires surgical intervention, it is typically thought to be due to progressive disease, MOE refractory to medical management with antibiotics alone, or in patients who develop new cranial nerve palsies such as facial nerve (FN) paralysis. 12 , 13 , 14 However, there is no uniform treatment guidelines to determine when surgeries are needed, or the exact type of surgical intervention required. This has resulted in large variations in patient outcomes, and increased patient frustration due to the lack of clear practice guidelines. 3 , 15 Recent literature has also suggested potential improvement in postoperative outcomes in MOE following FN decompression and debridement for patients with MOE associated facial paralysis, although limitations in application have stemmed from small cohort sizes. 16 Our study intends to build upon this foundation to study additional indications for surgical management of MOE, by systematically examining the specific symptomatology at presentation as well as quantitative measures of medical comorbidities. Clinical outcomes in patients who did versus did not undergo surgical management for MOE are also compared. METHODS Following approval by the Institutional Review Board (Ohio State University Wexner Medical Center IRB protocol #2022H0178), a retrospective chart review of patients diagnosed with malignant otitis externa using ICD‐9 (380.14) and ICD 10 (H60.20–23) codes between January 2010 to September 2022 was conducted. Adult patients who presented either to the outpatient otolaryngology clinic or to the emergency department at a single tertiary care academic referral center were included. Univariate regression analyses were used to compared symptoms at initial presentation, short‐term (≤6 months) follow‐up, and long‐term (≥1 year) follow‐up for patients who underwent surgery in conjunction with intravenous antibiotics versus patients who were exclusively medically managed without surgery. Variables of interest included demographics, presence of medical comorbidities (diabetes mellitus, immunocompromised status, Charlson Comorbidity Index scores), deficits in cranial nerves VII/VIII/IX/X/XI/XII (including facial nerve function and symptoms such as hoarseness, dysphagia, and hearing loss), and other non‐specific symptoms such as headache, otalgia, and vertigo. The Charlson Comorbidity Index has been widely used as a weighted index to predict risk of death within two years of hospitalization in patients with specific comorbidities (including age, cardiac risk factors, stroke, COPD, connective tissue disease, peptic ulcer disease, dementia, liver disease diabetes, hemiplegia, chronic kidney disease, solid tumor, leukemia, lymphoma, and acquired immunodeficiency syndrome [AIDS]). 17 If available, microbiology laboratory data from presentation and intraoperatively (if available) were curated and treatment courses were compared. Ultimate disposition plans for patients, including the length of stay, discharge disposition, and need for long‐term antibiotics at discharge were included. House‐Brackmann (HB) grades were used to quantitatively assess facial nerve function. Long‐term facial nerve outcomes were coded as binary variables, with poor facial nerve outcomes as those with HB ≥3. A significance threshold of 0.05 was used. All statistical calculations were conducted with Stata 17.0 (College Station, TX, USA). RESULTS From 2010 to 2022, 23 patients diagnosed with MOE were included in the study. The mean age was 69 years (standard deviation [SD], 13 years) and 78% (N = 18) patients were male. In the entire cohort, 52% (N = 12) of patients had MOE in the left ear, and one patient had bilateral MOE. Eighteen patients (78%) underwent inpatient admission due to the severity of illness to receive medical management including initiation of intravenous broad‐spectrum antibiotics and treatment of underlying medical comorbidities, while five patients were managed solely on an outpatient basis without hospital admission. Indications for surgical intervention included severity of symptoms at presentation, such as cranial neuropathies or facial nerve weakness as well as worsening symptoms despite conservative medical management. A total of 17 (74%) patients underwent surgical management and 6 patients (26%) did not undergo surgery (Table 1). Median follow‐up time after initial presentation was 305 days (range 171–1258 days). TABLE 1Baseline characteristics comparing patients who underwent surgical management for malignant otitis externa (MOE) versus medical management alone. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05.Patients undergoing medical management (N = 6, 26%)Patients undergoing surgical intervention (N = 17, 74%) p‐valueAge (years)69680.8DemographicsSex1.0Male5 (83%)13 (76%)Female1 (17%)4 (24%)Insurance status0.67Public Insurance5 (83%)12 (71%)Private Insurance1 (16%)3 (18%)None0 (0%)2 (11%)Race0.8White/Caucasian5 (83%)15 (89%)Black/African American1 (16%)2 (11%)Severity of DiabetesDiabetes Mellitus5 (83%)17 (100%)0.09Average HA1C, %6.88.20.2Insulin‐dependence3 (50%)9 (53%)0.8POC glucose, mg/dl2312430.9Diabetic nephropathy3 (50%)4 (24%)0.3Immunocompromised Status1 (17%)1 (6%)0.4Charlson Comorbidity Index55.50.7 Baseline characteristics comparing patients who underwent surgical management for malignant otitis externa (MOE) versus medical management alone. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05. Almost all (n = 22, 96%) of the patients had a diagnosis of diabetes mellitus. The mean point‐of‐care (POC) glucose at initial presentation was significantly higher than normal at 229 mg/dL (SD, 102 mg/dL). Average hemoglobin A1C was 7.9%. Twelve patients (54%) were insulin dependent. Seven patients (32%) had chronic kidney disease as a complication associated with their diabetes. Two (9%) patients were diagnosed with immunocompromised conditions (one was on chronic high dose oral steroids for rheumatoid arthritis, the other had active malignancy requiring chemotherapy). The median Charlson Comorbidity index score was 6 (interquartile range, 4–7). The most common symptoms at presentation included otalgia (N = 21, 91%), otorrhea (N = 15, 65%), and hearing loss (N = 13, 57%). Nine patients (39%) presented with significant facial nerve weakness, defined as having an HB grade of 3 or worse. Microbiology laboratory data obtained at presentation indicated over a third of the external ear cultures grew Pseudomonas aeruginosa (N = 8, 35%) followed by coagulase‐negative Staphylococcus (N = 6, 26%). Five patients (22%) ultimately had negative microbiology cultures. Of these five patients with negative microbiology cultures, four of them underwent surgical management and deep tissue cultures were obtained in three of the cases. Two of these patients grew Pseudomonas aeruginosa and one patient again had negative microbiology cultures. External ear fungal cultures were obtained in seven patients at presentation, all of whom underwent surgical intervention. Deep tissue fungal cultures were obtained in five patients, all of which were negative for fungal growth. A total of 17 (74%) underwent surgical management (76% tympanomastoidectomy with external ear canal debridement, 24% external ear canal biopsy and debridement without mastoidectomy). Of the patients who underwent tympanomastoidectomy, three (17%) underwent facial nerve decompression. All three patients had facial nerve weakness (HB ≥3), with one patient presenting with complete facial nerve paralysis (HB 6). All patients with facial nerve weakness (HB≥3) underwent surgical intervention. Facial nerve weakness at presentation was the only statistically significant predictor of undergoing surgery (HB≥3, p = 0.02). Neither patient age, insurance status, POC glucose at presentation, immunocompromised status, Charlson Comorbidity Index scores, nor any other patient‐reported symptoms were predictive of patients who progressed to requiring surgical management (all p > 0.05, Table 1). At discharge, there was a trend towards higher rates of needing prolonged intravenous antibiotics in the surgical versus non‐surgical cohort, as measured by the need for a peripherally inserted central catheter (PICC) placement (75% vs. 40%, p = 0.15, Table 2). Surgical management was not shown to significantly correlate with length of stay (9 days vs. 6 days) nor the rate of hospital readmission (31% vs. 17%, all p > 0.05). The rate of disposition to home was much lower in the surgical group compared to the non‐surgical group, which trended towards statistical significance (46% vs. 100%, p = 0.06). Patients who were discharged to skilled nursing facilities had significantly worse Charlson Comorbidity scores than those discharged home (7.2 vs. 4.8, p = 0.03). Overall, there were no differences in five‐year overall survival rates in patients who did and did not undergo surgical management (53% vs. 66%, p = 0.58). Patients with worse Charlson Comorbidity scores trended towards worse overall survival (OR 0.7, 95% confidence interval [0.45–1.1]). Mean overall five‐year survival rates in the entire cohort of MOE patients was 57%. TABLE 2Clinical outcomes comparing patients who did versus did not undergo surgical management of MOE. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05.Patients undergoing medical management (N = 6, 26%)Patients undergoing surgical intervention (N = 17, 74%) p‐valueLength of Stay (days)690.2Type of disposition0.06Skilled Nursing Facility (SNF)4 (100%) a 6 (46%) b Home0 (0%)7 (53%)Readmission1 (17%)5 (31%) c 0.5PICC line placement d 2 (40%) e 12 (75%) c 0.15Overall survival4 (66%)9 (53%)0.58 a Total cohort: 4 patients (two patients without complete data). b Total cohort: 13 patients (four patients without complete data). c Total cohort: 16 patients (one patient without complete data). d PICC: peripherally inserted central catheter. e Total cohort: 5 patients (one patient without complete data). Clinical outcomes comparing patients who did versus did not undergo surgical management of MOE. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05. Total cohort: 4 patients (two patients without complete data). Total cohort: 13 patients (four patients without complete data). Total cohort: 16 patients (one patient without complete data). PICC: peripherally inserted central catheter. Total cohort: 5 patients (one patient without complete data). A total of 20 patients (87%) presented for short‐term follow‐up and nine patients (40%) were included for long‐term clinic follow‐up with an otolaryngology subspecialist. At the short‐term follow‐up period (≤6 months), patients who underwent surgical management had worse FN function that trended towards significance compared to those who underwent medical management alone (HB≥3; 47% vs. 0%, p = 0.06, Table 3). However, at the long‐term follow‐up period (≥1 year), there were no differences in facial nerve function between the two cohorts (HB≥3; surgical 43% vs. 0%, p = 0.3). Of the three patients who underwent facial nerve decompression, one patient (33%) returned to normal facial nerve function in the immediate postoperative and two patients were noted to have residual facial nerve weakness at the immediate and long‐term postoperative follow‐up period. Of the five patients who had facial nerve weakness who did not undergo facial nerve decompression, four patients (80%) continued to have poor FN function at the immediate postoperative period. Long‐term facial nerve function was only available for one patient, who continued to have poor facial nerve function. TABLE 3Comparison of FN function at initial presentation, at short‐term follow‐up, and at long‐term follow‐up in medical and surgical MOE patients. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05 and marked (**).Patients who only underwent medical management (N = 6, 26%)Patients who underwent surgical intervention (N = 17, 74%) p‐valueFacial nerve function at presentation0.02**Good (HB 1–2)6 (100%) a 8 (47%)Poor (HB ≥3)0 (0%)9 (53%)Facial nerve function at short‐term (<= 6 months) follow‐upGood (HB 1–2)5 (100%) b 8 (53%) c 0.06Poor (HB ≥3)0 (0%)7 (47%)Facial nerve function at long‐term (>1 year) follow‐upGood (HB 1–2)2 (100%) d 4 (57%) e 0.3Poor (HB ≥3)0 (0%)3 (43%) a Total cohort: 6 patients (one patient without complete data). b Total cohort: 5 patients (two patients without complete data). c Total cohort: 15 patients (two patients without complete data). d Total cohort: 2 patients (four patients without complete data). e Total cohort: 7 patients (ten patients without complete data). Comparison of FN function at initial presentation, at short‐term follow‐up, and at long‐term follow‐up in medical and surgical MOE patients. All percentages are calculated within subgroup analyses (within surgical versus nonsurgical cohorts respectively). Statistically significant predictors were calculated with p < 0.05 and marked (**). Total cohort: 6 patients (one patient without complete data). Total cohort: 5 patients (two patients without complete data). Total cohort: 15 patients (two patients without complete data). Total cohort: 2 patients (four patients without complete data). Total cohort: 7 patients (ten patients without complete data). DISCUSSION Originally described in 1959 by Meltzer, malignant otitis externa (MOE) is a rare but devastating infection that begins in the external ear canal but spreads rapidly to the skull base, likely via the Fissures of Santorini. Due to the rare incidence and the lack of disease‐specific symptoms, it is often mis‐ or under‐diagnosed in the primary care or emergency room setting. Most existing literature is primarily based on single‐center, retrospective cohort studies and case series with limited sample size of under 20 patients. The constellation of vague symptoms at presentation, with persistent otalgia and otorrhea being the most common, highlights how difficult the early diagnosis of this devastating illness can be. 18 In addition, the historical consensus in the literature had suggested that with the advent of fluoroquinolone therapy for Pseudomonas Aeruginosa treatment, there would be little role for surgical management of MOE in the management of these patients. 19 Even with modern aggressive interventions, mortality rate varies significantly across studies. Moreover, difficulty with treatment standardization at the national level have placed otolaryngologists and other specialists in difficult positions of relying on anecdotal data to make complex treatment decisions. 15 However, with the rise of MOE cases refractory to medical or antibiotic management, the topic of surgical debridement in these cases and in patients with severe disease began to again garner interest and debate in the literature. 12 , 13 , 14 The goal of our study was to analyze our institutional cohort of MOE patients managed over a 12‐year period, and identify clinical characteristics of the subset of MOE patients requiring surgical intervention, and compare short‐ and long‐term outcomes between surgically and non‐surgically managed MOE patient populations. Currently, the literature has recognized that there exist more nuances to the traditional risk factors of the elderly, diabetic patient with a Pseudomonas aeruginosa infection. Our microbiology culture data, with 35% of infections caused by Pseudomonas aeruginosa and 25% coagulase‐negative Staphylococcus, brings up similar concerns and raises the concerns for re‐examining empiric treatment regimens to ensure adequate medical therapy prior to progressing to surgery. 6 Nonetheless, many patients who present to subspecialists with possible MOE have already been treated with at least one, but sometimes multiple courses of oral and ototopical antibiotics, thereby significantly lowering the likelihood of obtaining culture data of diagnostic significance. For patients with severe symptoms, usually referring to patients presenting with FN paralysis or other cranial neuropathies, the consensus has generally recommended more aggressive approaches, including surgical management such as tympanomastoidectomy with or without FN decompression. 14 , 16 Prior studies have reported that up to 32% patients with skull base osteomyelitis will have concurrent FN palsy, and close to 20% of patients could also develop other lower cranial nerve palsies involving the IX, X, XI or XII cranial nerves. 20 , 21 , 22 The presence of cranial neuropathy suggests more severe and widespread disease and portends increased overall mortality. 23 This is consistent with our data, in which poor FN function at presentation was the only predictor of patients who ultimately underwent surgical management. Our results also demonstrate in the immediate postoperative period, the patients who underwent surgical management continued to trend towards worse FN outcomes. However, in the long‐term follow‐up period, FN outcomes in the surgical group are similar to those in those who were medically managed, suggesting an overall stabilization to improvement in FN function in the surgical cohort. 16 Interestingly, our data does not demonstrate any other patient‐reported symptoms to be significant predictors of which patients required surgical management. This also included severity of traditional risk factors of disease, such as uncontrolled diabetes mellitus (POC glucose levels, rate of insulin independence) or immunocompromised status. We also did not see any of these risk factors to be predictive of FN function, raising the potential possibility that carrying the diagnosis of these risk factors, rather than an exacerbation of disease, to be correlated with the development of MOE and progression to surgical management. This is consistent with recent literature that showed similarly no significant differences in major symptoms of patients with the diagnosis of MOE, but instead that patients who were elderly or diabetic often had worse outcomes. 18 This is further supported by our data looking at postoperative planning for patients managed surgically. While not statistically significant, possibly due to small cohort sizes, patients who underwent surgical management trended towards longer inpatient stays, need for long‐term IV antibiotics, and higher rates of disposition to skilled nursing facilities for post‐hospitalization needs. Previous studies have also identified diabetic patients with MOE to have significantly longer hospitalization rates than their nondiabetic counterparts, likely suggesting a role that these medical risk factors may play in disease‐specific rates of morbidity and mortality following MOE treatment. 18 These all provide interesting avenues of further research, either with multi‐institutional collaborations or with existing national level large‐scale databases. Our study has similar limitations to other published series such as small sample size with the utilization of single tertiary academic center data, but the trend towards improvement in FN function following surgical management is a promising indication for determining patient populations that early surgical intervention should be considered. We also recognize that there is the potential that with appropriate conservative treatment, patients would similarly recover FN function. However, surgical decompression for other causes of facial nerve paralysis, including idiopathic, traumatic or iatrogenic injury, has demonstrated more promising results with early surgical intervention and similar findings with relation to MOE may exist. 24 We hope that our study presents evidence that patients with more severe disease presentation (including facial nerve paralysis or disease unresponsive to initial antibiotic trial) may benefit from earlier surgical intervention for debridement of necrotic tissue and deep tissue cultures collection. CONCLUSIONS Long‐term survival following MOE diagnosis remains poor. Facial nerve weakness at presentation is the primary predictor of patients who ultimately underwent surgical management of MOE, with improvement in short‐ and long‐term facial nerve outcomes. Medical comorbidities, including diabetes and immunocompromise status, did not significantly predict surgical management. Patients who underwent surgery did not have longer length of hospitalizations nor higher rates of readmission or mortality. FUNDING INFORMATION This research is supported by National Institute on Deafness and Other Communication Disorders (NIDCD) grant 1K08DC020761–01 (YR). DISCLOSURES We have no disclosures. This paper will be presented as a poster at the American Neurotology Society as part of the Combined Otolaryngology Spring Meetings 2023. Zhang L, Bonanno J, Byun WY, Ren Y. Malignant otitis externa: What is the role of surgery? Laryngoscope Investigative Otolaryngology. 2024;9(6):e70029. doi: 10.1002/lio2.70029 REFERENCES