Prognosticating patients with necrotising otitis externa based on response to treatment

Abstract

Introduction

Necrotising otitis externa (NOE) is a severe infection of the temporal bone. The traditional severity based staging system does not fully prognosticate all patients with NOE. We hypothesise that a patient response staging system would more accurately capture the disease process and guide prognosis.

Methods

We carried out a retrospective notes review of patients diagnosed with NOE from January 2017 to December 2018 in a regional tertiary referral centre. Patient outcomes from our proposed patient response staging system were compared to a modified previously published severity based Gleeson staging system with patients requiring prolonged treatment classified as having a poor outcome.

Results

A total of 34 patients were treated for NOE. The majority were male (n=24) and had diabetes (n=25). Patients with the most severe Gleeson staging did not have the worst outcome. Daily delay in resolution of otorrhoea was associated with an increased need for more than six weeks of treatment. Rapid responders are patients who had resolution of otalgia, otorrhoea and C-reactive protein normalisation within 14 days, and all were cured following standard 6 weeks of treatment.

Conclusions

The Gleeson staging system was valuable in assessing the extent of disease and all early Gleeson staged patients had good outcomes. However, patients with higher severity staging on the Gleeson system did not necessarily require prolonged treatment. There is a role for a joint approach in staging patients based on both modified Gleeson and treatment response, which would subsequently guide prognosis, duration of treatment and early diagnosis of potential fungal NOE.

Introduction

Necrotising otitis externa (NOE) or malignant otitis externa is an uncommon, severe infection of the temporal bone and skull base that can be fatal. It is more common in men, the elderly and immunosuppressed patients.¹ The rising incidence of NOE is believed to be secondary to an ageing population and increasing comorbidities.^1,2 Patients typically present with symptoms of otalgia, otorrhoea and, in some cases, cranial nerve involvement. The diagnosis of NOE is based on a strong clinical suspicion coupled with results from microbiology cultures and imaging in the form of computed tomography (CT) and/or magnetic resonance imaging (MRI). Treatment of NOE involves a prolonged course of systemic antibiotics and, in some cases, surgical debridement.³ The use of hyperbaric oxygen for NOE has been described in the literature, especially for refractory or severe cases, but more research is needed regarding its efficacy.^4,5

The most common organism causing NOE is Pseudomonas aeruginosa.³ Multidrug antimicrobial treatment is advocated due to the organism’s propensity to form biofilms and develop multidrug resistant strains.⁶ Cases of fungal NOE have been reported and can be associated with significant morbidity and mortality.⁷

Treatment of NOE involves prolonged combination antimicrobial treatment, usually over a period of six weeks, based on the rationale that bone revascularisation takes four weeks.⁸ Antimicrobial therapy usually involves Ciprofloxacin as there is effective antipseudomonal activity, high oral bioavailability and good bone penetration.⁹ This is usually paired with a broad-spectrum intravenous antibiotic.⁶ Challenges in the management of NOE involve difficulties with determining resolution of infection and hence treatment duration.⁸

Several staging systems have been published in the literature for NOE with increasing severity grading based on a mixture of clinical findings, extent of infection seen on imaging and cranial nerve involvement.^10–14 The aim of an ideal staging system is to guide disease severity and subsequently provide an indication of prognosis and treatment duration. Cranial nerve weakness would, in most staging systems, represent a more severe picture, and these patients were postulated to be less likely to recover and more likely to require aggressive treatment such as surgery.^10,12–14 However, Mani et al reported that cranial nerve deficits did not affect patient outcomes in their series.¹⁵

We hypothesise that a patient response staging system that incorporates clinical signs and biomarkers such as otalgia, otorrhoea and C-reactive protein (CRP) would more accurately capture the disease process and guide prognosis. To our knowledge, this is a novel approach to staging NOE and has not previously been proposed in the literature.

Methods

All patients treated for NOE in our tertiary Ear, Nose and Throat (ENT) referral unit were retrospectively reviewed over a 2-year period from January 2017 to December 2018. Data on age, sex, co-morbidities, microbiology results, clinical and imaging findings, C-reactive protein (CRP), diabetes control, while inpatient (if diabetic) and clinical response in terms of otalgia and otorrhoea were collected from patients’ electronic medical records.

Clinical response was recorded based on documentation in nursing and medical notes including patients’ drug charts, where analgesia prescriptions were reviewed. Poorly controlled otalgia was defined as patients requiring escalating analgesia prescription compared with admission and a pain score of more than 5 (out of 10) when questioned during ward round. Good response to treatment in terms of otorrhoea was defined as the moment the ear canal appeared normal on examination and the patient did not require microsuction. A CRP level of less than 10mg/l was considered in normal range. Patients with poor diabetes control were defined as patients with skinprick testing glucose levels consistently above 10mmol/l and requiring inpatient diabetes team review.

A literature review was performed looking at published staging criteria for NOE (Table 1). The Gleeson staging criteria described in Scott-Brown’s Otorhinolaryngology, Head and Neck Surgery 7^th Edition was identified as comparison as parameters used for staging were similar to most published staging systems in the literature.¹⁰ The modified version of the Gleeson staging system was utilised as published by Lambor et al as bone scan is not performed routinely in the management of patients with NOE in our unit.¹⁶ Patients were categorised into four stages as summarised in Table 2.

Table 1 .

Summary of published staging systems for NOE in the literature including various parameters incorporated into the staging system

Studies	Comorbidities	Clinical findings		Imaging studies	Patient response
		Ear canal inflammation	Cranial nerve palsy
Gleeson10	No	Yes	Yes	CT or MRI, Gallium scan	No
Peleg et al11	Yes	No	No	CT	No
Stevens et al12	No	Yes	Yes	All imaging modalities	No
Eveleigh et al13	No	Yes	Yes	No	No
Levenson et al14	No	No	Yes	Technitium scan	No

CT = computed tomography; MRI = magnetic resonance imaging; NOE = necrotising otitis externa

Table 2 .

Modified Gleeson staging system as published by Lambor et al¹⁶

Stage	Characteristics
1	Clinical evidence of soft tissue infection in the external auditory canal and beyond, without bony erosion on high resolution CT of the temporal bone
2	The above features with bone erosion
3a	The above features with facial nerve involvement
3b	The above features with multiple cranial nerve involvement
4	The above (stage 1, 2, 3) features with the presence of intracranial complications such as meningitis, sigmoid sinus thrombosis and empyema

CT = computed tomography

All patients were subsequently staged according to severity of disease based on the modified Gleeson staging criteria and findings were compared with patient response. Patients were followed up for at least a year following completion of treatment to assess for any recurrence.

The antibiotic regime for patients diagnosed with NOE in our unit is dual therapy of intravenous Ceftazidime (third generation cephalosporin) and oral Ciprofloxacin (second generation quinolone) for 6 weeks. If suitable, patients would be managed by the Outpatient Parenteral Antibiotic Therapy (OPAT) team to complete their antibiotic treatment in the community. They would have regular follow up every 1–2 weeks during the treatment period to review the ear and for any side effects from treatment. As the standard treatment for NOE in our unit is 6 weeks, any patients who required treatment for more than 6 weeks were classified as requiring prolonged treatment.

Statistical analysis

Univariate logistic regression analyses were performed to investigate the independent association between each variable and the need for more than 6 weeks of treatment. Due to small sample size, the proportion of patients who responded to treatment with resolution of otalgia, otorrohoea and CRP was used to define response to treatment. A systematic review by Mahdyoun et al showed the cure rate for treatment of NOE was 87–100%, with a recurrence rate of 9.6%.¹⁷ Based on this, a threshold of 75% was chosen; the point at which 75% or more of patients had normalisation of otalgia, discharge and CRP would be defined as a rapid responder. Data analyses were performed using Stata MP/15.1.

Results

Thirty-four patients were treated for NOE during the period studied, with a median age of 83 years (interquartile range, IQR 78–87; range 50–99). The majority of patients treated were men (70.6%, n=24). The median length of stay in hospital was 9.5days (IQR 8–13). A large proportion of patients (73.5%, n=25) had diabetes, with 13 patients tablet-controlled and 12 patients on insulin. All patients had CT imaging while the majority of patients (85.3%, n=29) also had MRI imaging. Only one patient had Gallium scan, which was used to assess post treatment response.

The most common organism grown was P. aeruginosa (61.8%, n=21). Nine cases had no growth and one patient did not have a microbiology sample sent. There were four cases where fungus was cultured but in all four patients, it was determined to be a commensal rather than a driver of infection and no antifungal treatment was needed. None of the organisms cultured showed multiresistant strains.

Twelve patients (35.3%) had stage 3a or worse based on the Gleeson staging criteria (Table 3). Of these, four patients (33.3%) required prolonged (more than 6 weeks) antimicrobial treatment due to persistent infection as evidenced by continued symptoms of otalgia, otorrohoea and/or elevated CRP towards the end of the initial 6-week treatment. Two of the patients were subsequently thought to have invasive fungal infection due to poor response to treatment and had raised beta-d-glucan blood levels. In both patients, ear swabs did not culture fungal organisms. Patient 8 (Table 3) was initially managed in a district general hospital and completed a 6-week antibiotic treatment. He presented again 3 weeks later with worsening pain (ear, face and neck), and imaging showed progression of disease. Due to worsening infection, he was referred to our centre for further management. Patient 8’s response to treatment was calculated from the day he re-presented. None of Gleeson Stage 1 and 2 patients (n=22) needed prolonged treatment. One patient who had Stage 3a disease died of congestive cardiac failure 4 months after diagnosis of NOE.

Table 3 .

Summary of patients who had stage 3 and 4 NOE based on Gleeson staging system

	Patient	Staging	CRP to normalise (days)	Otorrhoea (days)	Otalgia (days)	Diabetes control	Patient response	Treatment duration (weeks)
1	81, M	3a	Normal	9	9	Good	Rapid	6
2	90, F	3a	7	3	1	Poor	Rapid	6
3	83, M	3a	7	11	4	N/A	Rapid	6
4	81, F	3a	16	8	10	Poor	Slow	6
5	81, M	3a	21	6	1	Good	Slow	6
6	78, F	3a	57	73	30	Poor	Nonresponder	13
7	80, M*	3a	60	90	2	Poor	Nonresponder	16
8	77, M*	3a	21	n/a	12	Good	Slow	12
9	85, M	3b	14	53	2	Poor	Slow	13
10	83, M	4	Normal	4	2	Poor	Rapid	6
11	74, F	4	11	3	2	Good	Rapid	6
12	93, F	4	Normal	7	4	n/a	Rapid	6

*Patients with fungal NOE

CRP = C reactive protein; F = female; M = male; N/A = not applicable; NOE = necrotising otitis externa

Ten (28.6%) patients had cranial nerve involvement as a complication of NOE, with CNVII most commonly affected (n=8) followed by CNXII (n=2) and CNVI (n=1). Following treatment, the majority of patients showed improvement in their cranial nerve function except for two patients. Both of these patients also required prolonged antimicrobial treatment due to persistent disease. Intracranial complications (11.4%, n=4) included venous sinus thrombosis (n=2), dural enhancement (n=1) and cerebellar stroke on the same side of the disease (n=1).

Univariate logistic regression analyses were performed to investigate the independent association between each variable and the need for more than 6 weeks of treatment. Delay in daily resolution of otorrhoea (odds ratio (OR) 1.08, 95% confidence interval (CI) 1.02–1.14, p=0.010) was associated with an increased need for more than 6 weeks of treatment. There was an increased trend of requiring more than 6 weeks treatment for delay in daily CRP improvement (OR 1.13 (95% CI 0.99–1.30, p=0.079)) but this was not statistically significant. No significant association was found for resolution of otalgia, age and gender (Table 4).

Table 4 .

Association between selected factors and outcome of requiring more than 6 weeks of treatment

Factors examined	OR (95% CI)	p value
Days for CRP Normalisationa	1.13 (0.99–1.30)	0.079
Days for resolution of otorrhoeaa	1.08 (1.02–1.14)	0.010*
Days for resolution of otalgiaa	1.09 (0.96–1.24)	0.190
Ageb	0.99 (0.91–1.09)	0.911
Genderc	1.75 (0.25–12.50)	0.577

*p value of <0.05 was deemed statistically significant

^aFor per day delay in CRP normalisation, resolution of otorrhoea and resolution of otalgia

^bFor per year increase in age

^cMales compared with females

CI = confidence interval; CRP = C reactive protein; OR = odds ratio

The proportion of patients who had resolution of otalgia was 79.4% at 8 days post treatment while 76.5% had resolution of otorrhoea at 11 days. It took 13 days for 75.8% of patients to have normalisation of CRP levels. One patient was excluded from CRP analysis as his CRP levels did not normalise during his treatment for NOE, complicated by a concurrent gout flare up. Rapid responders were therefore defined as patients who responded to treatment and had resolution of all three parameters (otalgia, otorrhoea and normalisation of CRP) within 14 days. Otalgia was included as part of the patient response staging, although no association was found on univariate analysis as it is a vital part of decision making process when stopping treatment.

We compared patient outcomes based on the Gleeson staging criteria and the proposed patient response staging system (Table 3). Only Gleeson Stage 3 and 4 were included in the table as no Gleeson Stage 1 and 2 patients required prolonged antimicrobial treatment. Four patients who were Gleeson stage 2 were classified as slow responders due to a prolonged period of otorrhoea and/or elevated CRP. Three patients had the most severe Gleeson stage but did not require prolonged treatment. Among patients who had Gleeson Stage 3 and 4, 50.0% (n=6) of patients were rapid responders and all completed a standard 6-week treatment without complications or recurrence. Of the six patients characterised as slow/nonresponders, four required prolonged treatment. None of the rapid responding patients required prolonged treatment.

Proposed NOE management guideline

A patient management guideline was developed incorporating patient response staging (Figure 1). Once diagnosis is confirmed, patients are referred for suitability for OPAT and have a daily review of signs and symptoms of otalgia and otorrhoea. CRP is assessed every 3–4 days. Patients are subsequently categorised into three main categories of response to treatment: rapid, slow and nonresponder. Rapid responders are patients with resolution of otalgia, otorrhoea and normalisation of CRP (less than 10mg/l) within 14 days. Slow responders are patients who took longer than 14 days to show improvement in otalgia and/or otorrhoea and/or CRP. Slow responders require close monitoring of symptoms with a low threshold to continue treatment for longer than 6 weeks. Patients who do not show any improvement are categorised as nonresponders. These patients are expected to have a worse prognosis and further investigations will be required including assessment for fungal infections and repeat biopsies.

Figure 1 .

Proposed guideline for management of patients with NOE incorporating patient response staging system. CRP = C reactive protein; CT = computed tomography; ENT = ear, nose and throat; MRI = magnetic resonance imaging; NOE = necrotising otitis externa; OPAT = outpatient parenteral antibiotic therapy. EAC = external auditory canal; MC + S = microscopy, culture & screen; FBC = full blood count; U&E = urea and electrolytes; LFT = liver function test; ANCA = antineutrophil cytoplasmic antibodies; PO = oral; PICC = peripherally inserted central catheter; OPD = out patient department.

Discussion

NOE is a serious, potentially life threatening infection of the temporal bone that is increasing in incidence in the UK.⁴ There is consensus that treatment of NOE requires prolonged intraveneous antibiotics, preferably dual therapy due to the potential for Pseudomonas sp. to develop resistance.¹⁷ However, there is still a lack of consensus on the ideal staging and prognostication of patients with NOE, namely predicting patient suitability for stopping treatment.¹⁷

Different factors have been postulated to affect outcome, such as patients with diabetes, comorbidities (heart failure, liver disease), cranial nerve involvement, extent of disease (based on imaging findings) and age.^{3,11–13,18,19} There remains heterogeneity of evidence in the literature, with conflicting results suggesting diabetes control and cranial nerve involvement does not indicate poorer outcomes.^1,15 Extent of disease found on imaging did not affect outcome of patients in our series, which corresponded with findings from Loh and Loh and Sudhoff et al.^20,21 Interestingly, we did not find any association between patient age and likelihood of requiring more than 6 weeks of treatment.

Prognostication of patients with NOE remains a challenge with heterogeneity in data supporting different systems in the literature. We postulate a role for prognosticating patients based on response to treatment, where patients are divided into rapid, slow or nonresponders. In our series of 34 patients with NOE, the Gleeson staging system was valuable in assessing extent of disease and all patients with Gleeson stage 1 and 2 disease had good outcomes. As for patients at a more severe stage (Gleeson Stage 3 and 4), the modified Gleeson staging system could not identify patients who would have a poor outcome and require prolonged antimicrobial therapy.

We found patient response to treatment in resolution of otalgia, otorrhoea and CRP to be a useful indicator of patient prognosis, especially for patients with cranial nerve or intracranial involvement. Patients who had resolution of otalgia, otorrhoea and normalisation of CRP within 14 days of starting treatment did not require prolonged (more than 6 weeks) treatment. This included patients who had significant intracranial complications, which ranged from patients with extensive thrombus extending from the transverse sinus to internal jugular vein to a patient with cerebellar stroke who were all rapid responders and required only the standard 6-week antibiotic treatment.

We had two cases of likely fungal NOE with increased beta-d-glucan blood levels, and both patients required more than 6 weeks of antimicrobial treatment including antifungals. Historically, fungal NOE has been associated with more extensive inflammation and significant complications and are difficult to diagnose.^13,22,23 They are also associated with a worse prognosis, but evidence in the literature is inconclusive if patients with fungal NOE truly have a poor prognosis due to more aggressive infection or if it is secondary to delays in diagnosis and host factors such as immunosuppression.^13,22 Outcome of patients with fungal NOE is variable. In the literature, patients with fungal NOE are often diagnosed late, presenting with multiple cranial nerve involvement, which could lead to death.^7,13,22

In contrast, there is also evidence that, despite significant complications from fungal NOE, patients could be treated effectively with antifungals and be cleared of disease.^22,23 In our cohort, both patients with fungal NOE had cranial nerve involvement with CNVII and CNXII palsy, respectively. They responded to systemic antifungal therapy and were symptom free at 1 year follow up with resolution of the cranial nerve palsies. Surgical debridement for fungal NOE remains controversial as a systematic review by Mion et al could not draw conclusion about its effectiveness.²⁴ A high index of suspicion for fungal NOE is needed, especially for patients who do not respond to treatment, and potentially a biopsy as ear swabs are often nondiagnostic.^20,22 With the patient response staging system, fungal infections could be diagnosed earlier as these patients would usually fall under the slow/nonresponder category necessitating further investigations.

Post-treatment assessment of patients involved a combination of clinical assessment and measuring inflammatory markers. We did not perform post-treatment imaging routinely unless the patient showed signs of deterioration or persistent infection. Al-Noury and Lotfy found signal and bony changes seen in MRI and CT can persist for up to 12 months following treatment for NOE and thus would not be a useful investigation to guide end of treatment.²⁵ In our cohort, one patient had post-treatment Gallium imaging as the patient had a recurrence of NOE having been treated for NOE 6 months prior to the recurrence, which preceded the study period. All patients with NOE were followed up for at least 1 year as recurrence rate has been reported to be up to 20% in the literature.^12,17

The aim of staging patients into rapid or slow/nonresponders would be to guide management of patients especially with regards to severity of infection in each specific patient and subsequently duration of treatment. A systematic review by Mahdyoun et al found patients could be treated effectively with 4–6 weeks’ treatment.¹⁷ By adopting this staging system, patient treatment could be personalised as patients who are rapid responders could potentially be treated with 4 weeks of antimicrobials instead of all patients having a blanket treatment of 6 weeks thus reducing the risk of adverse drug reactions.²⁶ Further work is being performed to validate this hypothesis.

Limitations

This study was performed retrospectively with analysis of electronic patient records. Unfortunately we could not obtain complete patient notes prior to digitisation of patient records in 2017 to increase our sample size. Due to small sample size, it was difficult to show conclusive association between rapid patient response and prognosis. With larger sample size in the future, we aim to review the number of days for resolution of symptoms for a patient to be classified as a rapid responder.

Conclusion

Patients with the most severe Gleeson staging did not necessarily have the worst prognosis in terms of treatment duration and outcome. Staging patients based on response to treatment may be more accurate in predicting patient outcome for patients with cranial nerve and intracranial involvement. Previously considered poor prognostic indicators such as cranial nerve involvement, extensive involvement of disease based on imaging and patient age did not affect patient outcome in our cohort if they are rapid responders to treatment. However, due to rarity of disease and small sample size, more research is needed to show conclusive association between rapid patient response and prognosis. There may be a role for a joint approach in staging patients based on both modified Gleeson and treatment response that would subsequently guide prognosis, duration of treatment and early diagnosis of potential fungal NOE.