Corticosteroid-Resistant and Corticosteroid-Dependent AIED Patients Treated With Anakinra: A Decade of Experience

Abstract

Objective

Patients with autoimmune inner ear disease (AIED) are particularly vulnerable if they stop responding to corticosteroids or if they are dependent on high-dose corticosteroids to maintain hearing, as they have no therapeutic alternatives. We initially demonstrated the efficacy of anakinra in a small open-label clinical trial in corticosteroid-resistant AIED and are conducting a larger phase 2 placebo-controlled clinical trial. A large number of screened patients were ineligible or refused participation; however, for those we could obtain insurance coverage, we have used anakinra to improve and/or stabilize hearing.

Study Design

Retrospective case review.

Setting

Tertiary referral center.

Patients/intervention

A total of 49 patients representing three cohorts were treated with anakinra from 2013 to 2023, and hearing results followed over time: cohort A, corticosteroid-resistant AIED patients (n = 28); cohort B, corticosteroid-dependent AIED patients (n = 12); and cohort C, AIED patients who have participated in our anakinra trials with hearing improvement but were unable to maintain hearing off drug (n = 9).

Outcome measures/results

The anakinra response rate was 46% for cohort A, 83% for cohort B, and 78% for cohort C. In cohort A, a positive durable response was measured by hearing improvement by PTA: in this cohort of 13 patients, 14 ears improved by an average of 14 dB PTA. If we adopt a more conservative measure of improvement in both the PTA and WRS domains, eight ears in eight patients improved by 22% in WRS in this PTA-responsive cohort. After removing three responders who exhibited a WRS ceiling effect and three patients deemed nonresponders as they improved by WRS only, the overall rate of improvement drops to 32% or 7 of 22 evaluable patients. In cohort B, a positive response was considered to achieve hearing stability, with a concomitant reduction in corticosteroid dose by at least 50%. In cohort C, a positive response was to achieve hearing stability.

Conclusion

Anakinra seems to be efficacious in some patients with corticosteroid-resistant and corticosteroid-dependent AIED.

BACKGROUND

Patients with autoimmune inner ear disease (AIED) have a highly variable course. Historically, 70% initially respond to corticosteroids (1). However, continued responsiveness in the face of ongoing fluctuations is uncommon, as many patients become refractory to continued corticosteroid use. Although few longitudinal studies exist, one study showed that only 14% of patients remain corticosteroid responsive after 3 years (1). Furthermore, a subgroup of patients with AIED are corticosteroid dependent, and weaning of corticosteroids often meets with progressive hearing loss. We initially observed increased expression of an interleukin-1 (IL-1) decoy receptor in their peripheral blood immune cells (PBMCs) in response to autologous perilymph in a small cohort of AIED patients undergoing cochlear implantation (2). We subsequently observed decoy receptor expression was predictive of corticosteroid response in peripheral blood immune cells of patients with AIED. Corticosteroid-resistant patients had high basal levels of decoy receptor, which led us to interrogate plasma IL-1 levels. Indeed, corticosteroid-resistant AIED patients exhibited high plasma levels of interleukin-1 (IL-1) (3). Moreover, PBMC from these corticosteroid-resistant patients could not inhibit cellular release of IL-1 with corticosteroids; however, anakinra was able to block IL-1 release in vitro and became the basis of our early efficacy trial of anakinra. This trial (NCT01267994) initially intended to enroll 29 patients; however, after the 14th recruited patient, the trial was halted because the efficacy target was achieved: 70% of patients demonstrated improved pure-tone average (PTA), and 80% achieved improvement in word recognition score (WRS) (4). Improvement strongly correlated with a reduction in plasma IL-1beta. We are currently conducting a phase 2 placebo-controlled trial to determine the efficacy of anakinra in these corticosteroid-resistant AIED patients (NCT03587701).

Anakinra is a biologic agent that targets the interleukin-1 receptor antagonist that works by blocking the activity of IL-1, which is responsible for the development of inflammation. This protein-based therapy is administered by daily subcutaneous injection that is not presently FDA approved for the treatment of AIED. At the time of our original studies to evaluate use of anakinra for immune-mediated hearing loss, anakinra was an appealing option primarily because given its short half-life, if an untoward effect was observed, no lasting effect would be enacted, unlike the other anti–IL-1 therapeutics available that have a substantially longer half-life.

There is precedent to use anakinra for immune-mediated hearing loss. Anakinra has been used to treat and is approved for use in Muckle-Wells syndrome (MWS), a rare autosomal-dominant autoinflammatory disease that predominantly affects children and causes a high-frequency, progressive sensorineural hearing loss, transient skin rashes, and systemic amyloidosis (5). Typically in MWS, the hearing loss does not fluctuate, and the hearing loss generally stabilizes or improves in the presence of anakinra (6).

Approximately half of the patients with AIED experience vertigo and vestibular symptoms during periods of hearing decline (7,8). Moreover, the hearing loss sustained by AIED patients is often a low-frequency sensorineural hearing loss with evidence of endolymphatic hydrops (9) and therefore exhibits similar audiometric characteristics to patients with Menière's disease. Corticosteroid response rates in Menière's disease patients are between 40 and 50% (10,11), demonstrating that they are more corticosteroid resistant than those with either sudden hearing loss or AIED.

All patients who received anakinra were initially treated with dexamethasone and were unsuccessful in achieving meaningful clinical improvement. Three types of patients were considered for anakinra therapy: patients who were treated in our clinical trials that initially achieved benefit from anakinra but failed to maintain hearing during the observational period and had exceeded the protocol-defined number of re-treatments allowed, steroid-dependent AIED patients, or steroid-resistant AIED patients who did not meet the study criteria for enrollment in our phase 2 clinical trial of anakinra (NCT03587701). The purpose of this study was to determine whether these steroid-resistant or steroid-dependent cohorts of patients treated with anakinra derived a hearing benefit from use of this treatment in terms of hearing improvement or hearing stability. Specifically, we queried whether the steroid-resistant group (cohort A) experienced hearing improvement with use of anakinra. By comparison, we asked whether steroid-dependent AIED patients (cohort B) or AIED patients who benefitted from anakinra but could not maintain hearing in its absence (cohort C) experienced hearing stability through the ongoing use of anakinra. A flow diagram outlines which patients were included in each of the cohorts (Fig. 1).

FIG. 1.

Inclusion criteria for cohorts A, B, and C treated with anakinra. Flow diagram of patients treated with anakinra. To be eligible, all patients were initially treated with high-dose corticosteroids. Those patients who did not meet eligibility for our ongoing anakinra trial but were able to secure insurance coverage for the medication were included in cohort A, those patients who were steroid dependent and could not be weaned from high-dose oral corticosteroids were included in cohort B, and finally, those patients who previously derived benefit from anakinra during their clinical trial participation but could not maintain hearing in its absence were included in cohort C.

METHODS

This study was approved as a retrospective chart review by the Northwell Institutional Review Board. Additionally, all but eight patients had enrolled in a prospective study of inflammatory mediators at the time of their office visit, and they demonstrated a decline in hearing. All patients included in this study had AIED and were treated with and demonstrated no hearing improvement to corticosteroids before use of anakinra. Clinical characteristics of patients who were diagnosed with AIED included bilateral sensorineural hearing loss with active hearing fluctuation in one or both ears evolving in less than 90 days, with no evidence of retrocochlear pathology by MRI (4,12,13).

We treated a cohort of 49 AIED patients with anakinra from 2013 to 2023 who fell into one of three categories: they were either corticosteroid resistant who did not meet the criteria for our clinical trial/refused participation (A), were corticosteroid dependent (B), or patients who demonstrated response to anakinra during clinical trial participation but could not maintain hearing in the absence of continued anakinra (C). An algorithm for use of anakinra for these AIED patients is shown in Figure 1. Inclusion criteria for consideration of anakinra were failure to respond to corticosteroids, as defined as <5-dB improvement in the five-frequency pure-tone average (PTA) of 250, 500, 1000, 2000, and 4000 Hz and <12% improvement in word recognition score (WRS) as we have used previously (4), or, in the case of steroid-dependency, inability to reduce the corticosteroid dose below 1 mg/kg without incurring further hearing loss. Table 1 shows the demographics of these patients included in this study and the attribution for use. All patients underwent quantiferon gold testing to screen for tuberculosis before use of anakinra. Additionally, C-reactive protein (CRP) levels and a variety of autoimmune markers were screened based on the clinical history of the patient. Genetic testing for Muckle-Wells syndrome was offered to all patients. In many cases, the test was not performed (NP) as the patient opted not to be screened or their insurance did not cover the test. Of note, clinical trial participation required Muckle-Wells testing; therefore, all group C patients were required to be tested and did not carry the mutation. For all patients, the PTA values of 250, 500, 1000, 2000, and 4000 Hz are shown. The audiometric pattern is also described in Table 1. All patients tested at our center for word recognition score WRS were tested with recorded speech using a different track number for each assessment, although tested at the same level of difficulty for each assessment. Many patients remain on anakinra. For patients on ongoing therapy, the ending PTA was derived from the audiogram performed closest to January 1, 2024. We defined a clinical response as we have done previously: a greater than 5-dB improvement in the five-frequency PTA and a greater than or equal to 12% improvement in the WRS (4), unless the pretreatment WRS was better than 88%, in which case, a ceiling effect precluded the WRS response.

TABLE 1.

Demographics of AIED patients on anakinra (all cohorts)

Number	Patient Profile	Cohort	Anakinra Response?	Hearing Loss Pattern at Anakinra Commencement	Duration of Use?	CRP Level at Start of Therapy (Elevated Values in Bold)	MWS Testing?	Other Meds at Start of Anakinra Use
1a	67-yr-old WF	A	Yes	R: upsloping L: upsloping	5 yr, restarted, ongoing	<0.20 mg/dl	NP	T-HCTZ, Lipitor, NAC
2	23-yr-old WM	A	Initially yes then no, then steroid response again	R: upsloping L: upsloping	1.5 yr	<0.20 mg/dl	NP	T-HCTZ, Monteleukast, NAC
3a	19-yr-old AM	A	Yes	R: flat L: flat	0.5 yr, ongoing	<3 mg/L	Neg	None
4	81-yr-old WF	A	Yes	R: upsloping L: upsloping	1 mo	3 mg/L	NP	Metformin, rovustatin, Ezetimibe, Synthroid, amlodipine
5	52-yr-old WM	A	Yes	R: downsloping L: downsloping	3 mo ongoing	3 mg/L	Neg	T-HCTZ Betahistine Methotrexate
6a	67-yr-old WM	A	Yes	R: downsloping L: profound/CI	0.5 yr, ongoing	<3 mg/L	NP	Albuterol
7a	62-yr-old HF	A	No	R: flat L: profound/CI	4 mo	<0.20 mg/dl	NP	NAC, T-HCTZ
8a	41-yr-old WF	A	No	R: downsloping L: downsloping	3 mo	<3 mg/L	NP	Ministrin, ozempic
9a	59-yr-old WM	A	No	R: flat L: flat	7 mo	5 mg/L	NP	T-HCTZ, simvastatin, Cialis, Eszopiclone
10	47-yr-old BM	A	Yes	R: flat L: downsloping	2 yr	<3 mg/L	NP	T-HCTZ, NAC, ProAir
11a	18-yr-old WF	A	Yes	R: flat L: flat	1 yr, ongoing	<3 mg/L	HET	NAC, monteleukast
12	21-yr-old WM	A	Yes	R: downsloping L: profnd	2.5 yr	4.19 mg/dl	HET	NAC
13a	44-yr-old WF	A	Yes	R: upsloping L: upsloping	11 mo, ongoing	<3 mg/L	NP	Betahistine, amitryptaline
14	54-yr-old WF	A	No	R: profound/CI L: downsloping	3 mo	<3 mg/L	NP	NAC, T-HCTZ, betahistine
15a	20-yr-old BM	A	No	R: profnd/CNT L: borderline nml	1 mo	<3 mg/L	Neg	Methotrexate
16a	46-yr-old WM	A	No	R: upsloping L: flat	3 mo	<0.10 mg/dl	NP	T-HCTZ, olmasartan, xyzal
17	35-yr-old WM	A	Unclear, mult d/c	R: downsloping L: flat	4 yr	0.40 mg/dl	NP	T-HCTZ, NAC, FLUoxetine
18a	58-yr-old WM	A	No	R: downsloping L: profnd/CNT	8 mo	0.32 mg/L	NP	T-HCTZ, NAC, Crestor
19a	56-yr-old WM	A	Initially small response then no, discontin	R: upsloping L: upsloping	7 mo	<0.20 mg/dl,	NP	Bupropion HCl, gabapentin, Cymbalta, atenolol, Armodafinil, T-HCTZ
20a	38-yr-old WF	A	No, but became steroid responsive again	R: upsloping L: normal	11 mo	4.25 mg/dl	Neg	T-HCTZ, NAC, gabapentin, metoprolol, methotrexate
21	49-yr-old HM	A	No	R: flat/profnd L: flat/profnd	2 mo	0.20 mg/dl	Neg	Losartan, Aspirin
22a	44-yr-old WM	A	No	R: downsloping L: down/profnd	2 mo	0.69 mg/dl	NP	NAC
23a	68-yr-old BM	A	Initially small response then respond to steroid again	R: upsloping L: upsloping	4 mo	0.47 mg/dl	NP	T-HCTZ, NAC, zetonna, monteleukast
24	47-yr-old WM	A	No	R flat/profnd L: normal	1 mo	NP	NP	T-HCTZ, NAC
25	73-yr-old WM	A	No	R: downsloping L: flat	1 mo	<0.10 mg/dl	Neg	T-HCTZ, tamulosin, tizanidine,
26	57-yr-old BM	A	No	R: flat L: downsloping	4 mo	7 mg/L	NP	Amlodipine, rovustatin, coumadin, T-HCTZ
27	69-yr-old WF	A	No	R: downsloping L: profound	3 mo	0.36 mg/dl	NP	Uric acid
28	43-yr-old HF	A	Yes	R: upsloping L: downsloping	4 mo	0.82 mg/L	NP	Betahisitne, T-HCTZ, synthroid
29	37-yr-old WM	B	No	R: downsloping L: downsloping	1 mo	<0.20 mg/dl	NP	Prednisone 60 mg, acteazolamide, NAC
30a	61-yr-old WF	B	Yes	R: upsloping L: profnd/CNT	2 mo	<3 mg/L	NP	Prednisone 60 mg, betahistine, sertraline, acetazolamide
31a	59-yr-old AF	B	Yes	R: downsloping L: cochlear implant	7 yr	<0.20 mg/dl	NP	Prednisone 10 mg
32a	67-yr-old WF	B	Yes	R: flat L: downsloping	7 yr	<0.20 mg/dl	Neg	Prednisone 60 mg, Baraclude, Synthroid, Verapamil, Metoprolol, Pradaxa
33	19-yr-old WM	B	Yes	R: downsloping L: downsloping	3 yr, transition to ilaris	NP	Neg	Methotrexate
34a	36-yr-old WF	B	Yes	R: flat L: profnd/CNT	15 mo, ongoing	5 mg/L	NP	Prednisone 60 mg, T-HCTZ, Methotrexate,
35a	51-yr-old WF	B	Yes	R: normal L: upsloping	15 mo	0.28 mg/dl	NP	Prednisone 50 mg, NAC, metformin, toprol, belsomra, T-HCTZ, rovustatin,
36	32-yr-old HM	B	Yes	R: normal L: upsloping	5 yr, ongoing	0.5 mg/L	NP	NAC, T-HCTZ, monteleukast
37	47-yr-old WF	B	Yes	R: upsloping L: flat	20 mo, ongoing	0.1 mg/dl	Neg	Prednisone 50 mg, Ajovy, stelara, Eletriptan
38	46-yr-old WF	B	No	R: upsloping L: flat	7 mo	0.4 mg/dl	Neg	Prednisone 60 mg, T-HCTZ, zyrtec
39a	14-yr-old WM	B	No	R: downsloping L: downsloping	2 mo	2.0 mg/L	Neg	Dexamethasone
40	27-yr-old WM	B	Yes	R: downsloping L: downsloping	5 yr then canakinumab for 5 yr	0.4 mg/dl	Neg	Prednisone 60 mg Methotrexate
41	60-yr-old WM	C	Yes	R: downsloping L: flat	10 yr	NP	NP	None
42	20-yr-old WM	C	Yes	R: upsloping L: upsloping	3 mo	0.99 mg/L, before open label 0.22 mg/L	Neg	Betahistine, T-HCTZ, levothyroxine
43	23-yr-old WF	C	Yes	R: profound/CNT L: upsloping	2.2 yr	0.35 mg/L, before restart 0.94 mg/L	Neg	T-HCTZ
44	57-yr-old HM	C	Yes	R: upsloping L: flat	1.4 yr, ongoing	6.18 mg/L, before restart 7.20 mg/L	Neg	Insulin, metformin, NAC, Glipizide, Crestor, Lisinopril
45	22-yr-old WM	C	Yes	R: upsloping L: flat	2.5 yr, ongoing	2.07 mg/L, before restart 1.21 mg/L	NP	T-HCTZ
46	41-yr-old WF	C	Yes	R: upsloping L: upsloping	2.5 yr, ongoing	0.51 mg/L, before restart 0.61 mg/L	Neg	Acetazolamide, Betahistine
47	52-yr-old WM	C	Yes	R: upsloping L: flat	3 yr, restarted, ongoing	0.40 mg/dl, before restart: 0.53 mg/dl	Neg	Acetazolamide
48	49-yr-old WM	C	Yes	R: upsloping L: downsloping	0.5 yr, ongoing	1.00 mg/L, before restart 0.16 mg/L	Neg	T-HCTZ
49	59-yr-old WF	C	Yes	R: flat L: flat	6 mo	2.94 mg/L, before restart 0.27 mg/dl	Neg	Betahistine, T-HCTZ, Levothyroxine, Celexa, Losartan

^aPatients included in these biological studies.

AF indicates Asian female; AM, Asian male; BM, Black male; CD, congenitally deaf ear; CI, cochlear implant; CNT, could not test; CRP, C-reactive protein (several laboratories used, bold values are above the laboratory reference range and are abnormal); d/c, discontinuations; HET, heterozygous; HF, Hispanic female; HM, Hispanic male; MWS, Muckle-Wells syndrome; Neg, negative; NP, not performed; NAC, N-acetyl cysteine; Pred, Prednisone; T-HCTZ, triamterene-hydrochlorothizide; WF, White female; WM, White male.

Notably, patients enrolled in our early-stage clinical trial or ongoing phase 2 placebo controlled trial were included in this cohort as long as they exhibited hearing improvement with anakinra during the initial drug period as well as sustained a decline in hearing in the absence of drug that extended beyond the 1-month retreatment period allowed in those studies. This group consisted of nine patients (attribution C). The remainder of the patients either were steroid resistant and failed to meet enrollment criteria or refused clinical trial participation (attribution A, N = 28 patients) or corticosteroid dependent who were unable to be tapered from corticosteroids without experiencing a hearing decline (attribution B, N = 12 patients).

In vitro testing of peripheral blood immune cells (PBMCs) was performed on a subset of patients (n = 21) who consented for a prospective, IRB-approved observational study of patients with AIED at the time of hearing decline. Under this protocol, PBMCs were harvested from heparinized peripheral blood by density centrifugation for 20 minutes at 400g, plated at a density of 1 × 10⁶ cells/ml in RPMI with 10% fetal bovine serum, 4.1 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin as previously described (3). These PBMCs were stimulated with either 4 μg/ml dexamethasone or 100 ng/ml anakinra (Sobi Pharmaceuticals, Uppsala, Sweden) for 16 hours at 37 degrees and compared with unstimulated (3). IL-1β was measured in culture supernatants using a sandwich ELISA per the manufacturer's recommendations in duplicate samples to ensure reproducibility (R&D Systems, Minneapolis, MN).

RESULTS

The 49 patients who received anakinra belong to three cohorts, and therefore, the analysis will be stratified based on cohort. In both the steroid-dependent group (cohort B) and the patients with previous positive experiences with anakinra (cohort C), the goal of continued therapy was to maintain hearing, and therefore, measuring hearing improvement would not represent an appropriate outcome measure. Conversely, for the newly treated, corticosteroid-resistant patients given anakinra (cohort A), hearing improvement is an important measure of treatment efficacy.

Effects of Anakinra in Corticosteroid-Resistant Patients (Cohort A)

The subgroup of corticosteroid-resistant patients who did not participate in the two clinical trials of anakinra represents 28 patients. These patients either were ineligible for clinical trial participation or preferred to attempt insurance coverage for an off-label indication. Thirteen patients responded to anakinra with improvement in hearing as measured by either pure-tone average (PTA) or word recognition score (WRS), representing a 46% response rate, as shown in Tables 2 and 3. We defined a positive response in our previous clinical trial of anakinra as a greater than 5-dB improvement in the PTA or a greater than 12% improvement in the WRS (4). Historically, a 15-dB improvement in one frequency or a 10-dB improvement in two frequencies has been used to define a positive response in other studies of AIED patients (12); however, this methodology is imprecise and does not account for worsening hearing at other frequencies that were previously unaffected. When we used these historical criteria (12) in our original clinical trial (4), a similar response rate was observed. Durable clinical hearing improvement was observed in 14 ears of 13 patients by PTA, with an average improvement of 14 dB (Table 2). One additional patient achieved improvement in both the PTA and WRS domains; however, the response was lost quickly and therefore was not counted as a durable benefit (Table 3, subject 4). Improvement in the WRS was observed in eight ears of eight patients, with an average improvement of 22%. Notably, three patients had a starting WRS score of 92% or greater and therefore would be incapable of achieving a 12% improvement in WRS in response to therapy (Table 3). If we adopt a more conservative measure of improvement in both the PTA and WRS domains, eight ears in eight patients improved by 22% in WRS in this PTA-responsive cohort. After removing three responders who exhibited a WRS ceiling effect and three patients deemed nonresponders as they improved by WRS only, the overall rate of improvement drops to 32% or 7 of 22 evaluable patients. The range of anakinra use was 4 to 292 weeks with an average use of 71 weeks in anakinra responders. Two patients were heterozygous for the Muckle-Wells mutation in NLRP3 and therefore, although corticosteroid resistant, may not respond similarly to the rest of the cohort. Neither of these two patients experienced PTA hearing improvement: one experienced improvement in WRS only, and one patient experienced further hearing decline in the absence of treatment. Notably, 14 subjects did not respond to anakinra, although 3 improved in WRS score only, which was not counted as a response. Average use of anakinra in patients who failed to respond to anakinra was 25 weeks with a range of 4 to 54 weeks. The patients who used anakinra for 54 weeks that failed to improve hearing reported substantial improvement in systemic symptoms including dry eyes, joint pain, and skin rashes and had requested continuation of therapy. Rarely therapy was held for the development of infection; however, no patients were stopped for severe infections or neutropenia or required hospitalization. Additionally, no patients experienced a new diagnosis of malignancy of any type.

TABLE 2.

Summary of results of all cohorts

Cohort		Average PTA Pre L/R	Average PTA Post L/R	Average WRS Pre L/R	Average WRS Post L/R	Weeks on Anakinra	Average Hearing Stability Index
A	Anakinra responder	47/39	40/33	72/79	83/89	71	N/A
A	Anakinra responder (active ear only)	Active ear 53	Active ear 39	Active ear 67 (56a)	Active ear 83 (78a)	71	N/A
A	No response anakinra	62/58	61/58	58/71	59/65	25	N/A
B	Anakinra responder	56/33	53/28	71/94	66/94	340	0.009/−0.03
B	No response anakinra	46/46	66/66	78/40	90/14	6	4.5/1.8
C	Anakinra responder	51/47	41/41	80/72	81/79	133	−0.20/−0.19
C	No response anakinra	N/A	N/A	N/A	N/A	N/A	N/A

^a WRS calculated from those patients with a starting WRS score of 88% or less, to account for those patients with a ceiling effect where a 12% improvement in WRS cannot be achieved if the pretreatment WRS exceeds 88%.

N/A indicates not applicable for cohort A, as clinical improvement is measured by hearing improvement not hearing stability.

TABLE 3.

Steroid-resistant patients treated with anakinra not enrolled in clinical trial (cohort A)

Subject No.	Pre Anakinra PTA L/R	Pre Anakinra WRS L/R	Duration Anakinra (wk)	Post PTA L/R First Improvement Time and Amount	Post WRS L/R	Clinical improve?	Reason for D/C
1	38/39	100/92	292	11 wk, 27/36	96/96	Yesa	Stablea restarted several mo post d/c 2nd to hearing decline
2	48/29	76/92	83	4 wk, 39/29	76/96	Yesa	Hearing decline as
3	85/29	8/96	50b	11 wk, 70/24	64/96	Yesc	Ongoing
4	48/73	88/80	8 wk	4 wk, 47/56	92/92	Yesc, initial response then decline	At decline, became steroid responsive again
5	17/48	100/96	11 wk, ongoing	11 wk, 15/37	100/96	Yesa	Ongoing
6	CI/66	CI/16	33b	16 wk; CI/59	CI/40	Yesc	Ongoing, poor CI perform
7	CNT/41	CNT/88	12 wk; tried again 3 yr later 12 wk	4 wk; CNT; 33	CNT/92	Yesb	Underwent CI; unable to effectively use CI
8	35/35	92/96	12	8 wk, 46/36	92/88	NO	No benefit
9	69/66	56/60	31	8 wk, 71/52	48/76	Yesc	Stopped responding
10	25/55	88/44	120b	4 wk, 29/35	88/80	Yesc	Ongoing
11	57/57	80/84	104b	Drop in hearing with change to canikinumab (PTA 70/74 and WRS 48/56)		No, intended for maintenance in MWS, hearing worsened when changed to canakinumab	Ongoing, MWS+
12	94/44	100/8	94	7 wk 93/41	100/48	Stable hearing, no sig improve, except WRS	MWS heterzygote
13	34/21	100/100	58b	8 wk 6/15	100/100	Yesa	Ongoing
14	50/CI	40/CI	8	8 wk 47/CI	60/CI	No except WRS	No benefit
15	25/105	100/0	4	4 wk 12/110	100/0	No	No benefit
16	62/47	64/92	7	7 wk 80/49	0/92	No	Placed on oral steroids, WRS 48 as after 2 wk
17	71/26	36/92	144?	12 wk 66/26	76/100	Yesc, erratic use, unknown	Poor compliance
18	120/60	24/92	43	6 wk, 100/61	44/84	Not significant except WRS	No benefit
19	66/26	44/100	28	10 wk 62/23	72/100	Yesb	Stable, end WRS 64/100
20	12/32	100/100	54	6 wk, 14/38	100/92	No, but continued as systemic SX improved	Stopped 2nd to incr systemic infections
21	78/70	4/16	12	12 wk, 79/63	8/24	No	No benefit
22	96/63	4/80	8	8 wk, 85/53	4/80	No	No benefit
23	47/50	88/68	27	13 wk, 41/49	96/84	Yesb
24	13/73	100/16	4	4 wk, 12/72	92/16	No	No benefit
25	80/56	76/80	7	7 wk, 84/55	64/88	No	No benefit
26	39/52	88/76		4 wk, 37/55	80/60	No, vertigo improvement, no hearing improvement	Hearing stable vertigo significantly improved, patient choice
27	115/47	0/88	12	12 wk, 113/51	0/84	NO	No benefit
28	12/44	96/88	40	12 wk, 12/11	96/100	Yesc	Hearing stable and improved patient choice

Three of 14 nonresponders experienced improvement in WRS only, which we did not consider as a meaningful clinical improvement.

^aPTA with WRS ceiling. If the pretreatment WRS was 92% or greater, WRS improvement could not be achieved because a 12% or greater improvement could not be achieved due to the ceiling effect.

^bPTA without WRS. Benefit measured by PTA improvement alone.

^cPTA and WRS. Benefit measured by both PTA and WRS.

CI indicates cochlear implant; CNT, could not test; L/R, left/right; PTA, pure-tone average; WRS, word recognition score.

Effects of Anakinra in Corticosteroid-Dependent Patients (Cohort B)

Twelve patients had steroid-dependent AIED, and upon attempts to taper oral corticosteroids, worsening hearing was observed. As a result, anakinra was instituted, and oral corticosteroids subsequently tapered. Starting dose of prednisone ranged from 10 to 60 mg, with the majority of patients starting at 50 to 60 mg as shown in Tables 4 and 5. One 14-year-old child was on 5 mg of dexamethasone, which correlated to 33 mg of prednisone and corresponded to approximately 0.5 mg/kg at the time of initial presentation. To easily compare interventions, the ending pure-tone average (PTA; here to consist of 250, 500, 1000, 2000, and 4000 Hz) is subtracted from the starting PTA and is then divided by the number of weeks an intervention is used, and is defined as the hearing stability index. For the 12 steroid-dependent patients, the hearing stability index is shown for both corticosteroids and anakinra with the starting and ending dose of prednisone shown in Tables 4 and 5. Negative values suggest hearing improvement, where the magnitude of improvement correlates to the greater negative value. Positive decimal values suggest stability with small but tolerable declines in hearing, and positive values of 1.0 or greater suggest failure to maintain hearing: see Table 2 for average values for anakinra responders and nonresponders. For those 12 corticosteroid-dependent patients transitioned to anakinra, 9 out of 12 had negative hearing stability index values to suggest improvement in one or both ears as well as a significant reduction of the corticosteroid dose (percent reduction is shown in Tables 4 and 5). Two patients had positive values greater than 1.0 with clear worsening hearing, and anakinra therapy was discontinued. One of these two patients underwent a cochlear implant as he had significant complications from ongoing corticosteroids including development of a cushingoid body habitus. By comparison, for the corticosteroid therapy group, while on maximal tolerable steroids, although 6 out of 12 had a negative hearing stability index value in one ear, in 2 of these patients, the other ear had a value of greater than 1 suggesting worsening hearing. Comparing the hearing stability index for anakinra and prednisone (and removing two patients who had an index >1.0 for both therapies or incomplete data) showed that in 14 ears, the hearing stability index value was more negative and therefore more beneficial in the anakinra-treated patients. By comparison, only three ears showed better scores with prednisone (Table 4). Word recognition scores are shown as well (Table 5).

TABLE 4.

Corticosteroid-dependent patients treated with anakinra: PTA (Cohort B)

Subject No.	First Encounter PTA L/R	No. Weeks on Pred	Pred Hearing Stability Index	Starting Pred Dose on Ana	Ending Pred Dose on Ana	% Reduction of Steroid Dose	Ana Starting PTA L/R	Ana Ending PTA L/R	No. Weeks on ANA	Ana Hearing Stability Index
29	31/42	17	1.0/0.29	60 mg	5 mg—return to 60 mg	92%, then 0%	48/47	82/49, treated with steroids 54/48	4	8.5/0.5 1.5/0.25
30	100/22	20	*/0.07	60 mg	11 mg	82%	DNT/29	DNT/11	22	*/−0.50
31	106/39	76	*/0.09	10 mg	2 mg	80%	CI/46	CI/35	369	/*−0.03
32 On pred before first visit	93.3/27.5	25	−0.3/−0.13	60 mg	8 mg	87%	85.8/24.2	76.0/32.0	360	−0.002/0.02
33	20.8/44.2	128	0.19/0.02	80 mg + MTX	0 mg	100%	45.0/46.7	43.3/50.0	1,600	−0.001/0.002
34	116/46	29	0.07/−0.52	60 mg + MTX	0 mg	100%	118/31	CI/11	115	*/−0.17
35	34/9	16	−0.38/−0.13	50 mg	22 mg	54%	40/7	39/7, bone line 23 L	43	−0.02/0
36	57/4	80	−0.05/−0.10	50 mg	10 mg	80%	53/12	46/16	213	−0.03/0.02
37	69/29	8	−0.50/2.25	60 mg	0 mg	100%	65/47	64/16	132	−0.008/−0.23
38 On pred before first visit	58/23	14	−0.14/1.14	40 mg + EMBREL	30 mg	25%	56/39	59/52	23	0.13/0.57
39 On pred before first visit, cushingoid	52/79	468a	Significant missing datab	Dex 5 mg In past with MTX or immuran	Dex 1.25 mg	75%	43/45	49/83	12	0.50/3.17
40	21/44	129	0.19/0.02	60 mg	0 mg	100%	45/47	43/50	520	−0.004/0.006

Bold values suggest a hearing stability index consistent with progressive worsening of hearing in response to either prednisone or anakinra.

^aApproximate value.

^b Significant missing data, audiograms from other country, no clear documentation for changes in steroid dosing, or correlation to audiogram. Follow-up audiograms with same provider show worsening hearing, over 1-year gap from last known audiogram to our first one.

*Hearing is not measurable, either profound loss or cochlear implant in place.

ANA indicates anakinra; CI, cochlear implant; CNT, could not test; Dex, dexamethasone; L/R, left/right; MTX, methotrexate; Pred, prednisone; PTA, pure-tone average; WRS, word recognition score.

TABLE 5.

Corticosteroid-dependent patients treated with anakinra: WRS (Cohort B)

Subject No.	First Encounter WRS L/R	No. Weeks on steroids	Starting Pred Dose When Starting ANA	Ending Pred Dose on Ana	Starting Ana WRS L/R	Ending Ana WRS L/R	No. Weeks on Anakinra
29	100/68	17	60 mg	5 mg—return to 60 mg	92/68	100/28 Post steroid 92/76	4
30	0/96	20	60 mg	11 mg	0/96	DNT/100	22
31	0/92	76	10 mg	2 mg	CI/100	CI/100	369
32	16/100	25	60 mg	8 mg	24/100	16/100	360
33	96/92	128	80 mg + MTX	0 mg	92/84	92/88	1,600
34	0/100	29	60 mg + MTX	0 mg	0/100	CI/100	115
35	100/92	16	50 mg	22 mg	100/96	100/100	43
36	68/100	80	50 mg	10 mg	92/100	80/100	213
37	72/100	8	50 mg	0 mg	76/88	36/100	132
38	76/96	14	60 mg	30 mg, worsened went back to 60 mg	24/96	40/52	23
39	Unknown	Longstand	5 mg Dex	1.25 mg Dex	64/12	80/0	12
40	96/92	129	60 mg	0 mg	92/84	100/100	520

ANA indicates anakinra; CI, cochlear implant; CNT, could not test; Dex, dexamethasone; L/R, left/right; MTX, methotrexate; Pred, prednisone; PTA, pure-tone average; WRS, word recognition score.

Effect of Anakinra in Clinical Trial Participants (Cohort C)

Anakinra proved significantly beneficial in hearing recovery in our early efficacy open-label trial. Nine clinical trial participants from the early efficacy trial experienced benefit of anakinra and a subsequent hearing decline in the absence of drug (Tables 2 and 6). As a result, they required ongoing anakinra. The duration of ongoing anakinra varied with a range of 12 to 520 weeks. Their data are shown in Table 6, and their hearing stability index was calculated. Seven out of nine patients had a negative value for their hearing stability index, corresponding to a 78% response rate. One patient has been on anakinra for 520 weeks. Although he has had some mild progression of his hearing loss, we have attempted to stop therapy on several occasions, which has resulted in progression of his hearing loss.

TABLE 6.

Cohort C: Clinical trial participants treated with ongoing anakinra

Subject No.	Starting Ana PTA at Time of Retreat L/R	Ending Ana PTA L/R	Starting Ana WRS L/R	Ending Ana WRS L/R	Hearing Stability Index	No. Continuous Weeks On Anakinra
41	41/22	64/34	92/80	96/64	0.04/0.02	520
42	42/39	28/36	96/100	96/100	−1.16/−0.25	12
43	53/CD	36/CD	92/CD	100/CD	−0.13/CD	130
44	66/74	68/41	40/28	16/84 (previous month 60/84)	0.03/−0.49	68
45	16/67	26/87	100/4	100/4	0.07/0.15	130
46	50/56	22/45	80/80	100/92	−0.21/−0.08	130
47	60/58	CI/44	76/88	CI/96	CI/−0.09	156
48	39/43	42/25	88/92	88/92	0.11/−0.69	26
49	94/15	85/13	52/100	52/100	−0.38/−0.07	26

ANA indicates anakinra; CI, cochlear implant; CD, congenitally deaf; Dex, dexamethasone; L/R, left/right; Pred, prednisone; PTA, pure-tone average; WRS, word recognition score.

Correlation of Pretreatment Cytokine Expression With Clinical Response

We concurrently study inflammatory mediators in the peripheral blood immune cells (PBMCs) of patients with AIED. Enrollment is voluntary in this IRB-approved study. We interrogated cytokine expression in the A and B groups of patients. We specifically did not study the C group as they already had experienced anakinra at the time of inclusion into this cohort. Of the patients in the A and B groups, there were 15 corticosteroid-resistant patients and 6 corticosteroid-dependent patients who participated in the study that had blood drawn immediately before starting anakinra. The patients included in these biological studies are denoted by an italicized letter “a” in Table 1. Interestingly, corticosteroid-dependent patients increase their cellular release of IL-1 in response to dexamethasone, which may explain their ongoing need for high-dose corticosteroids (Fig. 2). Paradoxically, the corticosteroid-resistant patients seem to repress IL-1 release when treated with corticosteroids, inferring they clinically should respond to corticosteroids (Fig. 2); however, they do not. As a result, we further bisected the corticosteroid-resistant patients treated with anakinra from Figure 2 based on their ability to respond to anakinra. Strikingly, anakinra responders dramatically drop their cellular release of IL-1 in response to anakinra, whereas anakinra has no effect in the nonresponders (Fig. 3).

FIG. 2.

In vitro, pretreatment, IL-1β release from PBMC in corticosteroid-resistant (cohort A) and corticosteroid-dependent (cohort B) AIED patients. Fifteen patients in cohort A and six patients from cohort B participated in a study of inflammatory mediators in AIED and had their blood taken before starting anakinra for isolation of their peripheral blood mononuclear cells (PBMCs). These PBMCs were cultured in the presence of either dexamethasone (DEX) or anakinra (ANA) and compared with untreated (UNTX) as previously described (1). Cellular release of IL-1 into the culture media was measured by ELISA.

FIG. 3.

In vitro, pretreatment, IL-1β release from PBMC in corticosteroid-resistant (cohort A) AIED patients based on response to anakinra. The ELISA data from the 15 patients who were corticosteroid resistant were segregated based on their clinical response to anakinra. A dramatic reduction of IL-1 release in response to anakinra was observed in the anakinra responders as compared with those who did not derive benefit from this therapy.

DISCUSSION

Corticosteroid-resistant and corticosteroid-dependent AIED patients with fluctuating hearing loss represent two of the most difficult patient populations to treat medically. Based on the study by Broughton et al. (1), we anticipate that the majority of AIED patients initially responsive to corticosteroids will become refractory over time. Our early efficacy study of anakinra suggested that there may be an opportunity to treat these patients with an IL-1 inhibitor. We are currently engaged in a phase 2, placebo-controlled trial to validate these results (NCT03587701), again using the initial 84-day drug period to measure whether a clinical hearing improvement was observed in this three-arm study of 42 days of anakinra/placebo, placebo/anakinra, or anakinra/anakinra. Clinical hearing improvement during the 84-day drug period was again used to determine if drug was to be administered in the event of a hearing decline during the 264-day observational period.

Many corticosteroid-resistant patients presented here were ineligible for clinical trial participation, most commonly for too long a period of time from the hearing decline and use of corticosteroids to trial screening. As a result, this cohort was less likely to mount a response to anakinra as treatment from the time of original hearing decline was often delayed either rendering them ineligible for trial participation or due to the often lengthy appeal process in obtaining insurance coverage. We would therefore anticipate a lower response rate to anakinra based on a treatment delay. Consistent with this hypothesis, the durable anakinra response rate in the A group was 46%, or 13 of 28 patients (Tables 2 and 3). The clinical hearing improvement was 14 dB, which is superior to the published corticosteroid response rate of 4.1 dB in AIED patients (13), although the overall number of patients participating was smaller by comparison. Similarly, the improvement in WRS in cohort A was 22%. Although this was only in 8 of the 13 patients, 3 patients had a ceiling effect in their WRS score (starting value of 92% or greater) and therefore could not achieve a 12% benefit. Three patients deemed nonresponders in cohort A achieved an average WRS improvement of 27% (patients 12, 14, 18 in Table 3). Arguably, we could consider these patients responders; however, given the absence of PTA improvement, it is unclear as to whether this improvement is durable. If, however, we adopted an “either PTA or WRS score” as a meaningful indicator of clinical response, 18 of 28 would have been deemed a responder, or 64% (Table 3). Again, the improvement with anakinra was superior to the published improvement in WRS of 7% (13). Some caution should be exercised as, although the magnitude of the response is greater, a lower number of patients were included in these studies. However, given that corticosteroid-resistant patients do not have therapeutic alternatives available to them, it seems prudent to consider use of anakinra in this population. Although a few patients experienced initial hearing improvement, they developed further hearing decline. Interestingly, these patients did respond to corticosteroids with hearing improvement after anakinra use. This would suggest a change in the polarization of the immune response in this group. Notably, biologically, we could clearly distinguish anakinra responders in this cohort based on the dramatic in vitro reduction of IL-1 release from PBMC in response to anakinra before treatment (Fig. 3), although not all patients in cohort A participated in this adjunct study.

This study has several weaknesses. First, the response to anakinra cannot readily be compared with a reference therapy as the patients treated all received anakinra. Although to receive anakinra the patients had to have either failed to respond to corticosteroids or failed to maintain hearing as corticosteroids were tapered, regardless, corticosteroids were administered before the use of anakinra. Second, the data collected were from AIED patients outside of a clinical trial. Therefore, in addition to expecting a lower clinical response rate as described above, patients were not always seen at the same treatment intervals, although a concerted effort was made to standardize the evaluation periods for any patient in anakinra. Nonresponders received a maximum of 84 days of treatment, and if no clinical response was observed, it was recommended the treatment be discontinued. Third, the hearing stability index was applied to a limited data set and therefore may not be a durable measure. This measure was devised to be able to compare results of long-term, ongoing therapy with variable duration of use in different patients. Although predominantly used for corticosteroid-dependent patients (cohort B), we also employed this measure for patients dependent on anakinra to maintain hearing (cohort C). For both cohorts, durable hearing stability was achieved with anakinra, with cohort B averaging 340 weeks of treatment and cohort C achieving 133 average weeks of treatment (Table 2). Future studies will determine the clinical utility of this measurement.

Corticosteroid-dependent patients represent a challenge to the treating clinician. Side effects from prolonged steroid use are common and pose a major health risk to these patients. As such, any agent that can reliably reduce the corticosteroid requirements in this group of patients would be beneficial. Of the 11 corticosteroid-dependent patients, 8 could reduce their steroid requirement by 75% or more, and 3 could discontinue corticosteroids entirely. There were three patients who showed minimal or unsustained improvement: one that benefitted but relapsed and required high-dose corticosteroids and 2 patients with a more marginal reduction in corticosteroid requirements.

Elevated CRP is a surrogate marker for interleukin-6, which is influenced by interleukin-1 expression and correlated with corticosteroid resistance in AIED patients (14). CRP levels were measured 47 out of 49 patients before the institution of anakinra, as seen in Table 1. Notably, only 7 of the 49 patients had an elevated CRP immediately before anakinra therapy (15% of all AIED patients), and 6 of the 7 were corticosteroid resistant (patients 12, 17, 20, 22, 23, 26). Surprisingly, despite the relationship between CRP, IL-1, and IL-6, of these six patients, four did not derive clinical benefit from anakinra, and one stopped responding and anakinra was discontinued. For the patients who originally achieved benefit with anakinra that required ongoing therapy (cohort C), CRP was measured in eight of the nine subjects at the time anakinra was reinstituted. Of these, four out of eight had a rising CRP level (patients 43, 44, 46, 47) before restarting; however, an equal number of patients who required retreatment did not (patients 42, 45, 48, 49). Therefore, an elevated CRP is a poor diagnostic and prognostic marker of AIED.

CONCLUSIONS

This is the first larger study to demonstrate the benefit of IL-1 antagonism in AIED patients. Although not all patients experienced benefit, 46% of corticosteroid-resistant patients derived benefit from anakinra, 73% of corticosteroid-dependent patients achieved a substantial reduction in corticosteroids while maintaining or improving hearing, and 78% of patients who previously participated in the anakinra trial achieved lasting benefit from ongoing therapy. Notably, CRP is not a reliable marker to predict response, as CRP was elevated in only 15% of patients treated with anakinra, and over 50% of those with an elevated CRP did not improve with anakinra.