Idiopathic Sudden Sensorineural Hearing Loss Corticosteroid Treatment, the Diagnostic Protocol and Outcome

(1)

Linköping University Medical dissertation No. 1229

Idiopathic Sudden Sensorineural Hearing Loss

Corticosteroid Treatment, the Diagnostic Protocol and Outcome

Ramesh Nosrati-Zarenoe

__________________________________________________ Department of Clinical and Experimental Medicine

Division of Oto-Rhino-Laryngology Faculty of Health Science,

Linköping University SE-58185 Linköping, Sweden

(2)

Cover picture: Ramin Nosrati

Published articles have been reprinted with the permission of the copyright holders.

Printed in Sweden by Liu-Tryck, Linköping, Sweden, 2011

ISBN 978-91-7393-220-2 ISSN 0345-0082

(3)

To my beloved family Reza, Alma and Arvid

(4)

(5)

ABSTRACT

Idiopathic Sudden Sensorineural Hearing Loss (ISSNHL) is a rapid loss of hearing caused by damage to the cochlea (inner ear) or auditory nerve. Spontaneous recovery has been seen in 32%–81%. The incidence of the ISSNHL has been estimated to be between 5 and 20 per 100,000 per year. Different theories (vascular catastrophes, immunologic damage, infections or intracochlear membrane break) about the etiology have resulted in different treatment policies. The effect of therapy is difficult to evaluate for a single physician who sees just a few patients annually.

The aim of the present thesis was: 1) to investigate the current management and treatment of ISSNHL patients in Sweden with regard to outcome, 2) to evaluate whether, in comparison to placebo, the most common drug given in the treatment of ISSNHL in any way influences the outcome, and 3) to analyze which variables such as background data, concomitant disease, audiogram shape and laboratory tests, best can predict the outcome of ISSNHL.

A national database was developed with half of all ENT clinics in Sweden participating by submitting a questionnaire for each patient with SSNHL. The questionnaire covered the patient’s background, current disorder, past and family history of different diseases, examinations, and treatment. Audiograms at the onset of SSNHL and after three months were requested.

A randomized placebo controlled multicenter trial (RCT) was performed using a modified version of the questionnaire used in the national database. Prednisolone in high tapering dosage, or placebo was given with a total treatment period of eight days. If recovery was complete, treatment stopped, otherwise medication was continued at 10 mg daily to a total of 30 days from beginning. After an initial pure tone audiogram, new audiograms were taken at three follow-up visits: day eight of treatment, after one month, and after three months.

Results from the national database showed that out of 400 patients included in the study with ISSNHL, almost 60% were medically treated, of which nearly 90% were given corticosteroids. Hearing improvement was not statistically associated with receipt of medication. 40% of all patients had an MRI or CT, where 3–4% had acoustic neuroma. 24% of the patients with ISSNHL who had hematological tests taken, had one or more pathological findings. Blood screening varied from simple routine tests to a complete analysis with such tests as HSP70, Anti-Neutrophilic Cytoplasmic Antibodies (ANCA) and Borrelia tests

In the RCT, 47 patients were randomized to Prednisolone and 46 to placebo. No significant difference of hearing recovery was observed between the Prednisolone group and placebo group at either first or final follow-up regarding the effect of treatment. Presence of vertigo had significant negative effect on hearing improvement in both groups. Inflammatory signs in laboratory work-up had a positive prognostic effect, irrespective of treatment.

Conclusion: There is no standard program for management or treatment of ISSNHL in Sweden. The diagnostic protocol varies. MRI is an underused resource to get specific

(8)

8 Abstract

diagnoses for the condition especially acoustic neuromas. Regardless of pathological findings, treatment is mainly limited to corticosteroids or no medication.

Prednisolone in customary dosage does not influence recovery of Idiopathic Sudden Sensorineural Hearing Loss. Whenever a drug or a substance is proven not to work for a specific condition, it should be withdrawn from use and the focus for research within the field turned in a new direction.

(9)

List of original papers 9

LIST OF ORIGINAL PAPERS

This thesis is based on the following papers, which will be referred to in the text by their roman numbers I - IV:

I. Nosrati-Zarenoe R, Arlinger S, Hultcrantz E. Idiopathic sudden sensorineural

hearing loss: results drawn from the Swedish national database. Acta Otolaryngol. 2007 Nov;127(11):1168-75.

II. Nosrati-Zarenoe R, Hansson M, Hultcrantz E. Assessment of diagnostic

approaches to idiopathic sudden sensorineural hearing loss and their influence on treatment and outcome. Acta Otolaryngol. 2010; 130:384-91.

III. Nosrati-Zarenoe R, Hultcrantz E. Corticosteroid treatment of idiopathic

sudden sensorineural hearing loss. Part 1: a randomized triple-blinded placebo controlled trial. Submitted for publication.

IV. Hultcrantz E, Nosrati-Zarenoe R. Corticosteroid treatment of idiopathic sudden sensorineural hearing loss. Part 2: a meta-analysis of a RCT and the Swedish national database. Submitted for publication.

(10)

10____________________________________________________________________ Abbreviations

ABBREVIATIONS

ABR Auditory brainstem responses

AICA Anterior inferior cerebellar arteries

ANA Anti-nuclear antibodies

ANCA Anti-neutrophilic cytoplasmic antibodies

CI Confidence intervals

CRP C-reactive protein

CSF Cerebrospinal fluid

CT Computed tomography

dB HL dB hearing level

ENT Ear, nose and throat

Hb Hemoglobin

HDL High-density lipoprotein

HSP-70 Heat-shock protein 70

IgG antibodies Immunoglobulin G antibodies IgM antibodies Immunoglobulin M antibodies

ISSNHL Idiopathic Sudden Sensorineural Hearing Loss

kHz kilohertz

LDL Low density lipoprotein

MRI Magnetic resonance imaging

OAE Otoacoustic emission

OR Odds ratio

PTA Pure tone average

RCT Randomized controlled trial

SD Standard deviation

SLE Systemic lupus erythematosus

(11)

Introduction______________________________________________________________________11

INTRODUCTION

Idiopathic Sudden Sensorineural Hearing Loss (ISSNHL) involves a rapid loss of hearing that is caused by damage to the cochlea (inner ear) or auditory nerve. This hearing loss can be accompanied by tinnitus and/or vertigo.

A standard definition of ISSNHL does not exist, nor is there a standard method for reporting recovery. However, agreement has been reached with regard to the requirement for a 30 dB HL or more hearing loss in at least three contiguous frequencies1-5_{. The definition of “sudden” can vary from 24 hour to 72 hours in} different studies 2,6,7_.

No standard method exists for audiological assessment with respect to the configuration of hearing loss and hearing recovery. Hearing recovery has been reported in different categories, such as “complete recovery”, “good recovery” and “fair recovery”, but there is no agreement regarding to the actual degree of

improvement indicated by each of these categories. The pattern of hearing loss has often been mixed with the degree of hearing loss8,9_.

Independent of what treatment is given, spontaneous recovery can occur within a few hours to few days after onset. Complete and partial recovery is often combined when reporting spontaneous recovery, which is seen in 32%–81% of the cases2,6,10,11_{. A} higher recovery rate has been reported for patients with hearing loss in the low frequencies than for those with loss in the high frequencies10_.

Due to high rate of spontaneous recovery, not every potential patient ends up seeking help so that the incidence of ISSNHL is difficult to estimate. In 1944, De Kleyn reported an increase of ISSNHL between the years 1936 and 1942 in Amsterdam12_. Since 1958, the overall incidence has been reported to be 5–20 per 100,000 per year13. Recently, there have been two epidemiological studies that have shown that the prevalence of patients seeking help has increased. One study showed an increase in Japan from 3.9 to 27.5 per 100,000 persons annually14_{, and another study in Germany} showed a prevalence of 160 per 100,000 persons per year in the city of Dresden15_. Since the etiology of ISSNHL still remains unknown, a standardized treatment does not exist. Different theories with regard to its etiology have resulted in different treatment policies over the years in Sweden and in many other countries16-19_{. These} treatments have included anti-stress treatment with bed rest and blockage of the ganglion stellatum in the 1950’s20, treatment with dextran 40 and other hyperosmolar hemodiluting drugs in the 1970’s21_{and corticosteroid treatment since the 1980’s}2_. The randomized placebo controlled trial by Wilson et al. 1980 has been the foundation for a therapy tradition, which over many years, has become internationally the most common treatment for ISSNHL. Corticosteroid dosage has been successively increased without any evidence of better effect22_{. Several later investigations with} corticosteroids have not been able to demonstrate any specific effect even if

administered through the round window23. Moreover, the Wilson study has later been criticized by Cochrane reports 2006 and 2009 as not fulfilling modern standards for RCT24,25_.

(12)

12 Introduction

Due to its low incidence, its rate of spontaneous recovery and its unknown etiology, the effect of therapy is difficult to evaluate for a single physician who sees just a few patients annually. By compiling the data from a large number of patients, it was possible to identify prognostic factors and the impact of the various treatments used in Sweden. The most common treatment for ISSNHL used in Sweden was then compared to placebo with regard to treatments influence on outcome.

(13)

Background 13

BACKGROUND

The auditory system

The auditory system is described as the peripheral and central auditory system. The peripheral auditory system is comprised of three components, which are the outer, middle and inner ear. The central auditory system includes the auditory pathways and auditory cortex26.

Fig. I. The peripheral auditory system.

The outer ear

The outer ear consists of the pinna/auricle and the ear canal/external auditory meatus. The pinna/auricle collects and directs sound waves that are traveling through the air into the ear canal/external auditory meatus, which, in turn, transmits the sound waves to the tympanic membrane (eardrum). The ear canal forms a resonance channel that amplifies the sound pressure up to 15–25 dB in 2–5 kHz.

The middle ear

The middle ear is an air-filled cavity that includes the tympanic membrane (eardrum) and the ossicular chain. The ossicular chain consists of three interconnected bones, which are the malleus27_{, incus (anvil) and stapes (stirrup). The malleus is attached to} the tympanic membrane, and the footplate of the stapes inserts into the oval window of the inner ear. The incus is located between the malleus and the stapes. Thus, the movements of the tympanic membrane (eardrum) will set the malleus, incus and stapes into motion. Attached to the ossicular chain are the stapedius and tensor tympani muscles. The sound is not amplified evenly across the ossicular chain, and contractions of these muscles protect the inner ear by reducing the intensity of sound transmission to the inner ear.

(14)

14 Background

The inner ear

The inner ear consists of two systems that function independently. These include the sensory organ of hearing, which is called the cochlea, and the organ of balance, which is the vestibular system.

The cochlea is spiral-shaped and contains fluid-filled chambers. The two outer chambers, which are the scala vestibuli and scala tympani, are a part of the bony labyrinth and are filled with perilymph. The middle chamber, which is the scala media and is also called the cochlear duct, is filled with endolymph. Both perilymph and endolymph are clear solutions that contain electrolytes and proteins, and they are chemically quite different from each other. The perilymph is rich in sodium salts, whereas the endolymph is rich in potassium salts. The cochlear duct separates the two chambers from each other by the Reissner's membrane and the basilar membrane, on which the organ of Corti lies. The third partition consists of the stria vascularis, which is a rich bed of capillaries and secretory cells that are responsible for the production of endolymph. The scala vestibuli ends at the oval window, where the footplate of the stapes sits, and the scala tympani ends at the round window. A vibration coming from the stapes into the inner ear via the oval window moves the perilymph in the scala vestibule, which, in turn, vibrates the endolymph in the scala media, the perilymph in the scala tympani, the basilar membrane and the hair bundles of the hair cells in the organ of Corti.

The organ of Corti consists of approximately 3500 inner hair cells and 15,000 outer hair cells. The hair cells lie on the basilar membrane and are covered by the tectorial membrane. The basal parts of the organ of Corti are responsible for detecting the highest frequencies that we can perceive, and the frequencies that can be detected gradually decrease as they move towards the apical parts. The inner hair cells transform the sound vibrations in the fluids of the

cochlea into electrical signals. The outer hair cells amplify the low-level sounds by mechanically enhancing the motion of the tectorial membrane in order to increase the stimulation of the inner hair cells. They also add to the frequency resolution.

Fig. 2. Cross section of Cochlea.

Both the outer and inner hair cells are associated with afferent and efferent neurons. The afferent neurons, which carry information to the brain, contact mainly the inner hair cells, and the efferent neurons, which carry information back to the hair cells, connect mostly to the outer hair cells.

(15)

Background 15

The labyrinth artery is an end artery and is the sole blood vessel supplying the inner ear. The artery divides into the cochlear artery and the anterior vestibular artery. The cochlear artery further divides into the main cochlear artery and the vestibulocochlear artery. The main cochlear artery lies below the organ of Corti and has more capillaries in the apical portion of the cochlea then in the basal part. The distance between the vas spirale and the spiral border near the inner hair cells is shorter in the apical part than in the basal part28.

Glucocorticoids are one of five groups of steroid hormones that are released by the adrenal glands in response to stress. Receptors that steroid hormones bind to are ligand-activated proteins and are found in the cytosol and the nucleus. The effect of glucocorticoid is grouped into two categories namely metabolic and immunological. Cortisol is the term for glucocorticoids in humans and its metabolic effect is to increase and maintain normal concentrations of glucose in blood. By interaction with the glucocorticoid receptor, glucocorticoids can either up-regulate the expression of anti-inflammatory protein or down-regulate the expression of pro-inflammatory protein. The density of Glucocorticoid receptors varies in different tissues29_{. In} cochlea, the spiral ganglion neurons and spiral ligament have the highest expression. The hair cells and stria vascularis have lower expression30-34.

The central auditory system

A vibration that comes from the middle ear is transmuted into a neural signal in the cochlea. The neural signal transfers from the hair cells to the spiral ganglion, which is found in the spiral bony structure located centrally in the cochlea (modiolus). Axons from the ganglion cells are bundled together to form the auditory portion of the eighth cranial nerve. The auditory nerve carries the signal into the brainstem and synapses in the cochlear nucleus. From the cochlear nucleus, auditory information is split into at least two streams, which include the ventral cochlear nucleus and the dorsal cochlear nucleus. The ventral cochlear nuclear cells project to a collection of nuclei called the superior olive, which helps to determine the direction of the sound. The dorsal

cochlear nucleus analyzes the quality of sound. Both streams of information proceed to the sensory thalamus and from there to the auditory cortex that is located in the temporal lobes.

Hearing impairment

Hearing loss can be classified into four categories35.

Conductive hearing loss is caused by difficulties in the transmission of sound into the inner ear. A diagnosis can be made via observation of an air-bone gap on audiometry, which indicates that hearing is better when sound is transmitted in such a way that it bypasses the middle ear ossicular chain. The air-bone gap should be more than 10 dB HL.

Sensorineural hearing loss occurs without an air-bone gap, since air conduction is equal to bone conduction. A diagnosis is made through audiometry. Patients with cochlear damage have no Otoacoustic Emissions-testing (OAE), and those with auditory nerve damage fail the Brainstem Auditory Evoked Responses-testing (ABR).

(16)

16 Background

Mixed hearing loss is a combination of sensorineural hearing loss and conductive hearing loss.

Central hearing loss is caused by damage to the central pathways. The diagnosis cannot be made by pure tone audiometry, since that is often normal in affected individuals. A patient with central hearing loss usually has poor scores on their speech reception threshold or word recognition scores.

Etiologic hypotheses for ISSNHL

Idiopathic diseases are ‘idiopathic’ as long as the causative agent is not known. A cause for SSNHL can be found in 10% of all cases36,37_{. An acute hearing loss can be a} symptom caused by a multitude of known diseases within the vascular system or by different traumas and tumors within the hearing tracts38,39_{. Hallberg (1956) stated that} ”Sudden unilateral or bilateral impairment of hearing is a symptom, not a disease.”40_. Thus far, diagnosis has been based on the patient's audiogram, medical history, and physical examination, while identification of a specific cause for the sudden hearing loss is rather unusual.

Vascular theory

In 1949, Rasmussen suggested vascular occlusion or ischemia as a mechanism for ISSNHL41_{. Vascular or hematological diseases that are associated with SSNHL, such} as Buerger’s disease42_{, leukemia}43_{and sickle cell anemia}44_{have also been reported.} The labyrinth artery is an end artery that solely carries red blood cells and oxygen into the inner ear28_{. Tissue injury that results from oxygen deprivation and ischemia can} occur in the cochlea within 60 seconds6_{. A total but temporary blood circulation} blockage, can cause damage to the hair cells, the ganglion cells and the spiral ligament and can also cause neuronal loss and an alteration of the tectorial membrane after 30 minutes. This damage is irreversible even after the blood flow is restored45. The labyrinth artery is extremely vulnerable to blood pressure oscillation and abnormalities in blood flow46_{. Blood flow has an inverse relationship with blood viscosity}47_{. Low} blood flow causes anoxia due to hyperviscosity, which results in cochlear

hypofunction and the inability to maintain cochlear metabolism48. A correlation between ISSNHL and blood viscosity has been shown49-51_.

Different treatments regimes for ISSNHL, such as fibrinogen apheresis52_,

Rheopheresis53, dextran infusion54, hyperbaric oxygen55 and pentoxifylline56, have been developed in response to this vascular theory. There is evidence both to support and refute such treatment54-58_.

Immunologic theory

McCabe (1979) was the first to suggest autoimmunity as a cause of SSNHL, based upon clinical data, pathological findings in autoimmune tests and positive response to steroid therapy59_{. Hearing loss may be a consequence of local autoimmune processes} within the inner ear60,61_{or of systemic autoimmune diseases, such as Cogan’s}

(17)

Background 17

Although the inner ear was initially considered as an immunologically privileged organ, it is capable of producing a strong immune response. The presence of

antibodies against antigens in the inner ear and the formation of immune complexes in the stria vascularis, endolymphatic sac and ducts support the immunologic theory65-67_. Autoimmunity can cause damage to the cellular components of the organ of Corti and affect the stria vascularis and spiral ligament. Autoimmunity can also cause

dysfunction of the endothelial cells and fibrocytes II, which leads to the impaired diffusion of K+ through marginal cells to the endolymph fluid. This in turn can affect the supporting cells of the organ of Corti, which precedes a late effect upon the hair cells68. The existence of glucocorticoid receptors in the stria vascularis and supporting cells also suggests their role as immune targets in the inner ear69.

Treatment of ISSNHL with corticosteroids was the result of this immune theory. Since this therapy was first used in the 1980’s, there have been many studies on the impact of corticosteroids on SSNHL2,10,70-73. However, the only two randomized, double-blinded, placebo-controlled studies (19802_{, 2001}10_{) with low power showed} contradictory results.

Infectious theory

Herpes zoster that causes sensorineural hearing loss was first reported by Hunt in 190774. SSNHL has also been reported to be caused by known infectious diseases, of both bacterial and viral origins, such as by Borrelia and Syphilis75,76_{, the mumps}77,78 and rubella79,80_.

Both an acute and latent viral attack may damage the inner ear and cause hearing loss81_{. Upper respiratory infection}82_{, as an example of an acute viral infection, and} members of the herpes virus family83_{, as an example of latent infections, have been} proposed to cause SSNHL. Virus infection can damage the organ of Corti, the ganglion cells, the nervous fibers, the tectorial membrane and the stria vascularis84. Two antiviral drugs, acyclovir and valacyclovir, are used to treat SSNHL as a result of this theory4,85_{. However, the four published randomized, placebo-controlled clinical} studies that used acyclovir and valacyclovir did not show any benefit for the treatment of ISSNHL3-5,86.

Membrane breaks

Rupture of the oval or round window can cause loss of the perilymph and result in pressure alteration between the chambers that contain perilymph and endolymph87,88_. There have been studies on the temporal bones that support this theory89,90. The membrane break does not happen spontaneously, but rather occurs after a sudden pressure alteration in the middle ear that can be caused by head injuries, barotraumas or intense physical exercise40,91_{. The fact that many of these patients report hearing} loss upon awaking6_{and that not all individuals with high intracranial and} intra-abdominal pressure, such as women in childbirth or weight lifters, experience SSNHL argues against this theory. In addition, not all of the temporal bones studies found evidence of active or healed ruptures of the oval or round window, basilar membrane or Reissner’s membrane81,84,92_.

(18)

18 Background

Surgery with the intention to repair oval or round window perilymph fistulae combined with strict bed rest has been used in cases of ISSNHL with a history of recent trauma or barotrauma93,94_.

Genetic Theory

Hearing loss can be inherited. Wilde (1853) was the first to report the genetic cause of congenital deafness95. Epidemiological data about postlingual hereditary hearing loss are still unknown. Genetic hearing loss divides into two categories, Syndromic hearing loss and Nonsyndromic hearing loss. Two-thirds of all genetic hearing losses are nonsyndromic hearing loss meaning “without other clinical findings”. Seventy percent of all genetic hearing losses are recessive hearing loss, fifteen percent are dominant and the remaining fifteen percent are other form of inheritance96,97.

Many genes whose mutation causes nonsyndromic hearing loss have been identified within the last few years. WFS1exon 8, connexin 26, 30 and 31 are genes which have been identified in stria vascularis, basal membrane, spiral limbus and spiral

ligament98,99.

Very few studies are done about the heredity of ISSNHL. Two papers were published 2010 on the subject100,101. Gäckler et al., found more individuals with family history of SSNHL among those who themselves had experienced SSNHL100. In the second paper, the researchers were looking for a difference in gene mutation between SSNHL patients and healthy volunteers, but found none101_.

(19)

Aims 19

AIMS

The general objectives of this thesis were to find predictive factors for hearing

recovery, to investigate the treatment policy of ISSNHL in Sweden and to evaluate the efficacy of these treatments on hearing recovery in comparison to placebo.

Paper I

 To analyze which variables, such as background data, concomitant disease, audiogram shape and laboratory tests, can best predict the outcome of ISSNHL.  To investigate the treatment policy of ISSNHL in Sweden.

 To evaluate the effects of treatments on outcomes.

Paper II

 To explore the different diagnostic test batteries for ISSNHL, that are currently used in Sweden.

 To evaluate if and how, positive diagnostic findings result in treatment modifications.

 To investigate whether such treatment modifications influence the outcome of ISSNHL.

Paper III

 To evaluate whether, in comparison to placebo, corticosteroids (Prednisolone) used in high tapering dosage in any way influence the outcome of ISSNHL.

Paper IV

 To analyze a larger patient group by meta-analysis of data from the RCT together with corresponding material drawn from the Swedish national database for ISSNHL.

(20)

(21)

Methods 21

METHODS

Paper I and II

The first two publications (I, II) in this thesis are based on data from a national database in Sweden for sudden sensorineural hearing loss.

Swedish national database for SSNHL

The database began gathering data from patients with SSNHL in winter 2002–2003. Approximately half of all ENT clinics in Sweden have been contributing data to the database.

To build the database, a questionnaire was developed that covered the patient’s past medical history, potential precipitating events that preceded the SSNHL, traumas, family history of different diseases, especially hearing loss, the current disease, the diagnostic protocol including laboratory, radiological and further audiological examinations and all other treatments. The time course of the hearing loss’s onset and associated symptoms, such as tinnitus and vertigo, were also requested. The

questionnaire included the results of an ENT examination. Information on radiological investigations (MRI or CT), laboratory work-ups and the use of BRA or a vestibular work-up were requested. This information request was phrased in general terms so as not to influence the doctor's own diagnostic practices on their decision-making process. The questions with regard to treatment included the use of corticosteroids, antiviral therapy, rheological treatment or “other” drugs, and prescription of rest or surgery of a suspected fistula. In the case that pathological test results were performed, the complete lab-sheet for that patient was requested. See Appendix I. For each patient, after informed consent, a questionnaire was completed by the otorhinolaryngologist, and two audiograms were requested—one to have been taken during the first visit to the ENT clinic due to the symptoms of SSNHL and the other taken after three months. When available, a pure tone audiogram was requested for patients known to have a prior diagnosed hearing loss before onset of SSNHL.

Paper III

A randomized triple-blind1 placebo-controlled multicenter trial on the effect of corticosteroids on ISSNHL was performed between January 2006 and September 2010. Fourteen public otorhinolaryngological centers in Sweden were successively enrolled and each center contributed for 1 to 4.8 years.

Patients asked to participate were those aged 18–80 years, presenting with sudden onset of hearing loss developing within 24 hours, and without any known etiology (no earlier or present ear disease). The pure tone average in the affected ear of the three contiguous frequencies most affected should be ≥30 dB HL. Enrollment and treatment was to be started within 7 days from onset. Exclusion criteria were the

1_{Triple-blind trial is a trial in which neither the subject, the person administering the treatment nor the person} evaluating the response to treatment knows which treatment a particular subject is receiving.

(22)

22 Methods

common medical reason for not using corticosteroids: pregnancy, diabetes, chronic infections, peptic ulcer, uncompensated heart disease, recent surgery, or psychiatric disease. The patients' ordinary medication for concomitant disease was permitted except vascular, antiviral or corticosteroid treatment.

A case report form (CRF) was collected for each patient. The CRF consisted of a questionnaire, audiograms, information on radiological investigations (MRI or CT), laboratory work-ups, brainstem response audiometry (BRA) and vestibular work-up, according to the praxis at the different clinics, and information on adverse events and/or serious adverse events. The questionnaire was a modified version of one used for the Swedish database for ISSNHL (see Sppendix II). After an initial pure tone audiogram, new audiograms were taken at three follow-up visits: day eight of treatment, after one month, and three months. In cases where patients were known to have prior diagnosed hearing loss, a copy of any available previous audiogram was included in the CRF.

All patients received written information about Sudden Sensorineural Hearing Loss, the aim of the study and instructions for taking the drugs. Prednisolone as 10 mg capsules, or placebo was given as a single dose of 60 mg daily for three days; the dose was then reduced by 10 mg per day, with a total treatment period of eight days. If recovery was complete (complete recovery=difference between the initial audiogram and audiogram at the follow-up <10 dB), treatment stopped, otherwise medication was continued at 10 mg daily to a total of 30 days from beginning. Follow-up visits were scheduled for day eight ± one day, 1 month and 3 months after randomization. If recovery was complete at day eight, the next follow-up was at 3 months.

Paper IV

Meta-analysis was used to strengthen and confirm the results from the randomized triple-blind placebo controlled clinical trial on the effect of corticosteroids on ISSNHL (Paper III). The material from the RCT (Paper III) was augmented with corresponding data drawn from the Swedish national database for SSNHL.

The patients included in the analysis had been treated with Prednisolone, treated with placebo or received no treatment. The patients from the database were selected if they had received the same or equivalent treatment as in the RCT, which was Prednisolone as 10 mg capsules as a single dose of 60 mg daily for three days, thereafter reduced by 10 mg per day, with a total treatment time of at least eight days. The control group from the RCT was the placebo treated patients, with audiograms at the start and after 3 months. The control group from the database was those who had not received Prednisolone or other specific treatment, but been examined with audiograms at the initial visit and after 3 months.

Assessment of hearing loss and hearing recovery

Sudden Sensorineural Hearing Loss was defined as a hearing loss of at least three contiguous frequencies between 0.125 kHz to 6 kHz, with a mean of 30 dB HL or more and that occurred within 24 hours.

(23)

Methods 23

The hearing loss was characterized by a comparison of the audiogram taken at the first visit after onset of SSNHL to an audiogram that was taken not more than two years before the acute hearing loss. If no previous audiogram was available, hearing was compared to the non-affected ear in its present state.

Four frequency regions were created to describe the hearing loss: Low frequency region:

 Pure-tone average (PTA) of low frequencies (125, 250, 500 Hz) > PTA of mid frequencies (1000, 1500, 2000 Hz) and high frequencies (3000, 4000, 6000 Hz) by at least 10 dB.

 Hearing loss in the low and mid frequencies – PTA of low frequencies > PTA of mid frequencies with a difference less than 10 dB.

Mid frequency region:

 PTA of mid frequencies > PTA of low- and high frequencies by at least 10 dB.  Hearing loss in the low and mid frequencies – PTA of mid frequencies > PTA

of low frequencies with a difference less than 10 dB. High frequency region:

 PTA of high frequencies > PTA of low- and mid-frequencies by at least 10 dB.  Hearing loss in the mid and high frequencies – PTA of high frequencies > PTA

of mid frequencies with a difference less than 10 dB. Flat loss:

 The differences between the PTA for all three frequency regions were less than 10 dB.

The audiogram taken at the first visit and the audiogram obtained three months after the onset of SSNHL were compared with respect to the PTA, characterizing the loss to determine the degree of hearing recovery and remaining hearing loss (Table I).

(24)

24 Methods

Table I. Hearing improvement and remaining hearing loss after recovery.

Improvement

Large improvement >30 dB Moderate improvement 10 – 30 dB No improvement ± 10 dB

Worsening >-10 dB

Hearing loss after recovery

No remaining hearing loss Difference between initial audiogram and audiogram at the follow-up <10 dB Partial recovery The difference ≥10 dB and

the improvement ≥10 dB No regress The difference ≥10 dB and

the improvement <10 dB

Categorization of laboratory tests

The pathological results of laboratory tests were categorized by one or more pathological values prior to analysis.

 “Arteriosclerosis associated variables”: LDL-cholesterol/HDL-cholesterol ratio>3, Total cholesterol >5 mmol/L and C-reactive protein (CRP) >3 mg/L in patients with or without earlier known

cardiovascular disease.

 “Inflammation/infection”:

CRP >10 mg /L, Erythrocyte sedimentation rate >20 mm, Leukocyte count >10 x 109_{ml/L, Hemoglobin count (Hb) <120 g/L,} Thrombocyte count >150 x 109_{ml/L and Borrelia tests “positive” (IgG} antibodies and IgM antibodies) in patients with or without ongoing clinical infection.

 “Autoimmune variables”:

HSP-70, Cardiolipin, Antiphospholipid, Anti-Neutrophilic Cytoplasmic Antibodies (ANCA) and Antinuclear antibodies (ANA) “positive”.

(25)

Statistical methods 25

STATISTICAL METHODS

Statistical analysis was performed using Minitab software, version 13.32 for Windows, for Paper I, version 15 for Windows, for Paper II and version 16 for Windows, for Paper III and IV.

Descriptive statistics were used in all four papers to show the characteristics of the subjects. The data was expressed as the number of cases and percentage. Parametric data was expressed as mean ± standard deviation65.

Paper I and II

Ordinal/Ordered logistic regression was used for all analyses with regard to hearing recovery. This method performs a logistic regression on an ordinal response variable (categorical variables that have three or more possible levels with a natural ordering) with the help of both continuous and categorical predictors.

The estimated probability for hearing improvement and no remaining hearing loss in relation to the prognostic factor, the frequency regions (low, mid, high and “flat loss”) and the frequency regions (low, mid and high) in relation to the number of days from the onset of ISSNHL was expressed as an odds ratio (OR) and 95% confidence intervals (CI).

Seasonal variance and gender differences with regard to age distribution, the presence of tinnitus and/or vertigo, the comparison between treatment options and laboratory tests, the interval between the onset of hearing loss and the first visit to the ENT clinic and the pure-tone average between different frequency regions were performed using χ2-test. The level of significance was set at p<0.05.

Paper III and IV

The analyses of primary and secondary endpoints were performed according to modified Intention-To Treat (modified ITT) and Per Protocol (PP) with comparisons of pure tone audiograms.

Multiple regression was used with selected variables (age, heredity for hearing loss, vertigo, tinnitus, time from onset of SSNHL to first ENT visit, prescribed rest or sick leave, affected frequency regions) which were forced into the model together with Prednisolone and placebo. The first step in the multiple regression analysis was to study interactions between treatment and some selected variables The interactions studied were between treatment & age, treatment & time from onset of SSNHL to first ENT-visit, treatment & affected frequency regions, treatment & baseline pure tone average for the affected frequencies. The goal was to see if any of the selected interactions had a significant covariance with recovery. Interactions, that were not significant, were thereby removed from further analyses.

The second step in the multiple regression analyses was to include variables with stepwise forward method if p<0.05 to see if any single variable had a significant covariance with recovery. Three dummy variables were created out of the four categories of frequency regions and used to indicate the absence or presence of some

(26)

26 Statistical methods

categorical effect that might be expected to shift the outcome. These three dummy variables were not tested individually in the stepwise forward process; the choice was between including none or all (partial F-test).

A single-sided null-hypothesis was used in testing the effect of Prednisolone and placebo, since a deterioration in hearing due to corticosteroids was not expected:

H0 = the efficacy of Prednisolone on recovery ≤10 dB H1 = the efficacy of Prednisolone on recovery >10 dB

For the other variables, double-sided null-hypothesis was tested that the variable had no effect on recovery.

(27)

Ethical considerations 27

ETHICAL CONSIDERATIONS

Paper I and II

The Medical Research Ethic Committee of Linkoping’s University, Sweden (registrations number 02-337) approved the database. Participants were given oral information about the database by their local ENT doctors. A written informed consent was not requested, as there was no intervention that was performed and as the

participant’s identity (security number) was not reported in the database.

The data were handled confidentially. No more than the participant’s birth date and gender were provided on the questionnaires, audiograms and requested lab-sheets.

Paper III

The study, EudraCT 2005-001487-32 was approved by the regional ethics review board and Swedish Medical Products Agency in accordance with the Declaration of Helsinki and good clinical practice guidelines.

All patients received written information about Sudden Sensorineural Hearing Loss, the aim of the study and instructions for taking the drugs.

A case report form (CRF) was collected for each patient. The data were handled confidentially. The CRFs were coded with the number of study drug bottle, the participant’s birth date, gender and the name of the clinic.

(28)

(29)

Material 29

MATERIAL

Paper I

Three hundred patients with acute hearing loss who had initially received the diagnosis of “Sudden deafness/Sudden Sensorineural Hearing Loss” by their ENT-doctor, were included in the Swedish national database for SSNHL.

Of the three hundred patients in the database, ninety-two either had conductive hearing loss, Mb Ménière, other known disorders of the inner or middle ear, a hearing loss less than a 30 dB HL, less than three contiguous frequencies involved or a hearing loss that occurred over more than a 24-hour period. These ninety-two individuals were

excluded from analysis. See Table II.

Paper II

Four hundred patients (three hundred from Paper I and one hundred additional patients) who were initially diagnosed with SSNHL and, thereby, reported to the database for SSNHL were evaluated.

Table II. Final diagnosis after examination for patients initially reported as SSNHL (n=92).

Diagnoses Number

Acoustic neuromas 5

Trauma acoustic trauma 2

head trauma 2

trauma after water irrigation of

the ear 2 barotraumas 3 Subdural hematoma 1 Myeloma 1 Mb Ménière 5 Hydrops 1

Progressive hearing loss 1

Bleeding in Pons

(infarct at the root entry zone) 1

Transient ischemic attacks 1

Coronary disease 3

Did not fulfill the criteria Hearing loss less than 30 dB HL 44

for SSNHL Hearing loss did not occur

(30)

30 Material

Paper III

One hundred and three patients with ISSNHL were included randomized evenly to treatment with either Prednisolone or placebo. Ten patients were excluded from the analyses, four of them in the treatment group and six in the placebo group. See Fig. III. Ninety-three patients were analyzed by modified Intention-To Treat (modified ITT) forty-seven received Prednisolone and forty-six placebo. For eight of ninety-three patients, the hearing loss was evaluated by comparison of an audiogram taken within two years before the ISSNHL. Six of these had been treated with placebo.

Eighty-seven out of ninety-three patients who took the study drug according to the protocol under the first eight days of the treatment period were analyzed by PP; forty-five belonged to the treatment group and forty-two placebo. Fourteen of those eighty-seven patients either should have continued taking the study drug up to 30 days or did continue, but not in accordance with the protocol.

Seventy-three out of ninety-three patients who did take the study drug according to the protocol were analyzed as Total Per Protocol (total PP); thirty-eight belonged to the treatment group and thirty-five placebo.

Paper IV

A total of one hundred ninety-two patients were analyzed, eighty-seven from the RCT and one hundred and five from the Swedish national database.

Ninety-nine of the total one hundred two had received Prednisolone and ninety-three placebo or no treatment.

The hearing of each patient was evaluated at the first visit at the ENT clinics and after three months.

At the final follow-up, audiograms were missing for two patients in Prednisolone group (RCT), one patient in corticosteroid group (the Swedish national database) and two patients in the no treatment group (the Swedish national national database).

(31)

Material 31

FIG. III. TRIAL PROFIL

(32)

(33)

Results 33

RESULTS

Paper I

The aim of Paper I was to analyze which variables best can predict the outcome of ISSNHL, to investigate the treatment policy of ISSNHL in Sweden and to evaluate the effect of treatments on outcome.

None of the patients in the Swedish national database were bilaterally affected. Of three hundred patients reported to the database, two hundred and eight have been evaluated according to the defined criteria for ISSNHL

Descriptive data for the two hundred and eight patients with ISSNHL (age range 15–87 years, mean 57) are presented in Table III. All two hundred and eight patients had a physical examination, 68% had laboratory work-up, and 37% had MRI. The time between onset of ISSNHL and the first visit at the ENT clinics varied, with a mean of 16.4±31.5 days.

T V

able III. Profiles of the ISSNHL patients (n=208).

ariables Number (%)

Gender Female 98 (47)

Male

Age Mean ± SD (years) 56.1 ± 16

Range (years) 15 - 87

Affected ear Left 96 (46)

Right

Prevalence of associated Tinnitus 107 (51.4)

symptoms _Vertigo

With nystagmus 1 (0.48)

Without nystagmus 7 (3.4)

No info. about nystagmus 1 (0.48)

Tinnitus and vertigo

Without nystagmus 38(18.3)

No info about tinnitus or vertigo 1 (0.48)

No associated symptoms 36 (17.3)

Interval between onset of Median (days) 5

hearing loss and first visit at the

ENT clinics Range (days) 0 - 212

110 (53)

112 (54)

All variables in the questionnaire (Appendix I) were analyzed using ordinal logistic regression in order to look for interactions with hearing recovery and remaining hearing loss as dependent variables.

(34)

34 Results

Independent of treatment or no therapy, “Heredity for hearing loss” and older age were significantly associated with outcome. “Heredity for hearing loss” was

associated with a significantly lower odds for improvement, with an odds ratio of 3.02 (95% CI 1.34–6.80, p=0.008), but was not associated with the remaining hearing loss. An older age was related to a reduced chance of both improvement of hearing in dB (OR 1.05, 95% CI 1.03–1.08, p=0.000) and the remaining hearing loss (OR 1.03, 95% CI 1.0–1.05, p=0.008).

Hearing loss in the low frequency region occurred in 42% of all cases, hearing loss in the mid frequency region in 26%, hearing loss in the high frequency region in 30% and 3% had a “flat loss”. The reported number of patients with “flat loss” was too low for statistical analysis (n=6); their hearing improved partly or totally irrespective of treatment (Table IV). Patients with hearing loss in the mid frequency region had a significantly higher chance for improvement when compared to those with hearing loss in low (p=0.000) or in high frequency region (p=0.000), if they visited an ENT clinic the same day as the onset of ISSNHL. The probability for improvement of the hearing loss decreased with number of days to ENT visit regardless of the frequency region.

Total or partial recovery for patients with hearing loss in mid frequency region was significantly more likely when compared to those with hearing loss in low (p=0.002) or high frequency region (p=0.014), if they visited an ENT clinic the same day as the onset of ISSNHL. The probability for no remaining hearing loss decreased with number of days before the first visit regardless of the frequency region.

T

P

able IV. Treatment and outcome for patient with “flat loss” (n=6).

atient No. Treatment Improvement Hearing loss after recovery

1 Prednisolon 60 mg/day Large No remaining hearing loss 1 No medical treatment Moderate Partial recovery 1 Prednisolon 60 mg/day Moderate Partial recovery

1 Prednisolon 60 mg/day Moderate Partial recovery

1 No medical treatment Large No remaining hearing loss 1 Prednisolon 80 mg/day Moderate Partial recovery

Corticosteroid therapy was used for one hundred and five (50%) of the total two hundred and eight patients and ninety (44%) received no medical treatment. Of the remaining eleven patients, three received antiviral therapy and five received antibiotics. In all, 38% of the patients treated with corticosteroids had a large improvement (>30 dB) and 34% had moderate improvement (10–30 dB) compared with 28% with large and 36% with moderate improvement among those who were not medically treated. The differences were not significant.

The main corticosteroid used was Prednisolone (80%). The dosage, duration of treatment and tapering schedule varied from 80 mg to 25 mg per day for five days to four weeks. No significant difference in outcome was seen as regards the dosage

(35)

Results 35

given. Among those who were treated with Prednisolone, 68% received high dose (>50 mg/day) and 32% received low dose (≤50 mg/day). Large improvement (>30 dB) was noticed among 40% of patients treated with high dose (>50 mg/day) Prednisolone and moderate improvement (10–30 dB) among 33% compared to 30% showing large and 37% moderate improvement among those who were treated with low dose (≤50 mg/day). See Table V. The differences between the groups were not significant.

(36)

36 Results

Paper II

The aim of Paper II was to explore the different diagnostic test batteries for ISSNHL that are currently used in Sweden, to evaluate if and to what extent, positive diagnostic findings result in treatment modifications and to investigate whether such treatment modifications influence the outcome of ISSNHL.

Of four hundred patients reported to the database, three hundred were evaluated as suffering from ISSNHL. Descriptive data on the three hundred patients evaluated as suffering from ISSNHL are presented in Table VI.

One hundred and fifty-eight (40%) of the four hundred patients with SSNHL had an MRI or CT, and twenty-two had pathological findings. Out of the twenty-two with pathological findings, ten patients had accidental findings that were not connected to the hearing tracts and thereby were defined as ISSNHL. Out of remaining twelve patients: five had acoustic neuroma, one had subdural hematoma, one had a pons infarction and five had different vascular abnormalities that might have had a possible connection to SSNHL. The five individuals who had different vascular abnormalities had all a low frequency hearing loss (Table VII).

No significant association was found between pathological MRI-findings and either hearing recovery or remaining hearing loss for the patients with ISSNHL.

T V

able VI. Profiles of the ISSNHL patients (n=300).

ariables Number (%)

Gender Female 146 (49)

Male 154

Age Mean ± SD (years) 57.4 ± 16

Range (years) 8 - 87

Affected ear Left 141 (47)

Right 159

Prevalence of associated Tinnitus 149 (49.7)

symptoms _Vertigo

Tinnitus and vertigo

No info about tinnitus or vertigo 1 (0.33)

No associated symptoms 61 (20.3)

Interval between onset of Median (days) 5

hearing loss and first visit at the

ENT clinics Range (days) 0 - 498

(51)

(37)

Results 37

Hematological tests were taken in two hundred and fifty-eight (65%) cases of the total four hundred. The blood screening varied from simple routine tests to a complete analysis (CRP, Hb, HSP70, ANCA tests, and Borrelia tests), covering most hypotheses concerning SSNHL (e.g., infections, vascular catastrophes, autoimmune disease or membrane rupture).

Of the three hundred patients who were classified as having ISSNHL, one hundred ninety six had one or more laboratory tests taken, and forty-seven (24%) of these had one or more pathological findings.

The group categorized as having pathological “arteriosclerosis-associated variables” (n=13) was significantly older than those with normal laboratory tests (n=148), 65 ± 11 years and 56 ± 17 years, respectively, p=0.016. The twenty-seven patients with ”inflammation markers” were of the same age as those with normal laboratory tests. Of these, fifteen had positive Borrelia findings. Seven patients were regarded as having pathological “autoimmune variables”. No age difference was seen between those with pathological “autoimmune variables” and those with normal laboratory tests.

There was no association between any of the laboratory tests and either hearing improvement or remaining hearing loss, even when the tests were evaluated separately or after the tests were compared to those who had normal laboratory findings.

Table VII. Patients with low frequency loss and vascular changes as possible cause of ISSNHL shown by MRI (n=5). Treatment and outcome.

Patient Diagnosis Treatment Hearing recovery

1 ISSNHL Corticosteroids + Antiviral therapy No improvement 2 ISSNHL Corticosteroids No improvement 3 ISSNHL Corticosteroids Moderate improvement 4 ISSNHL Prescribed rest Moderate improvement 5 Myeloma Cytostaticum Moderate improvement

Medical treatment was given to one hundred eighty-two (61%) of three hundred patients with ISSNHL. Of these, one hundred sixty-six patients were given corticosteroids either as single treatment or in combination with antiviral therapy, antibiotics or blood thinning therapy (acetylsalicylic acid). Three patients were only given antiviral therapy, six others only antibiotics, two were only given acetylsalicylic acid, and five received various other drugs. Ninety-six patients received no treatment at all.

Medical treatment had no significant correlation with either hearing improvement or remaining hearing loss

Seventy-seven of three hundred patients with ISSNHL (26%) were prescribed “to rest” or “to stay home on sick leave” for up to four weeks; this was the only treatment for twenty-two of these patients. Patients who had been prescribed rest or been on sick

(38)

38 Results

leave during the first period of the disease had higher odds for hearing improvement (OR 0.46, 95% CI 0.27 – 0.79, p=0.005), regardless of any other treatment.

40% of the patients had their hearing loss in the low frequency region, 28% in the mid frequency region, 23% in the high frequency region and 9% had a “flat loss”. Patients with hearing loss in the mid frequency region had significantly higher odds for hearing improvement compared to those in the groups with a hearing loss in the low frequency region, the high frequency region or a “flat loss”.

The odds for a residual hearing loss was lower for patients with hearing loss in the mid frequency region when compared to those with a hearing loss in the low frequency region (OR 0.45, 95% CI 0.26 – 0.78, p=0.005), the high frequency region (OR 0.31, 95% CI 0.16 – 0.58, p=0.000), or a “flat loss” (OR 0.43, 95% CI 0.18 – 1.01, p=0.05). All variables in the questionnaire were analyzed using ordinal/ordered logistic

regression looking for interactions with hearing recovery and remaining hearing loss as dependent variables. Three factors were significantly related to outcome

independent of treatment or no therapy, namely “Heredity for hearing loss”, age, and presence of vertigo at onset; for details see Table VIII.

T re

able VIII. Outcome measures after ISSNHL with respect to hearing improvement and maining hearing loss using ordinal logistic regression (n=300).

Hearing improvement Remaining hearing loss Variable

OR 95% CI p-value OR 95% CI p-value

Increase of age 1.03 1.01 – 1.05 0.000 0.98 0.96 – 0.99 0.001 Vertigo at onset 1.77 1.06 – 2.95 0.03 0.40 0.24 – 0.67 0.000 Heredity for hearing loss 2.08 1.10 – 3.94 0.02 ns ns ns Prescribed to rest 0.46 0.27 – 0.79 0.005 ns ns ns

(39)

Results 39

Paper III

The aim of Paper III was to compare treatment with corticosteroids (Prednisolone) used in high tapering dosage to treatment with placebo with respect to influence on the outcome of ISSNHL.

Ninety-three patients were analyzed by modified ITT; forty-seven received Prednisolone and forty-six placebo (Fig. III). Audiograms were missing for four of them at final follow-up. Those missing were not included in the analysis—two had gotten Prednisolone. The baseline characteristics data and hearing improvement of ninety-three patients included in the modified ITT analyses are described in Table IX. Of the ninety-three patients in modified ITT analysis, fifty had one or more laboratory tests taken. One or more pathological findings were discovered in seventeen of twenty-six patients with laboratory work-up in the Prednisolone group compared with nine of twenty-four in the placebo group. Almost all of the patients with abnormal findings had “inflammations markers”.

MRI or CT was done on forty patients in modified ITT analysis. Abnormal findings were seen in thirteen of twenty-two patients with MRI or CT in the Prednisolone group compared with ten of eighteen in the placebo group. The majority of the findings were different vascular abnormalities. Acoustic neuroma was found in one patient.

Variables

Table IX. Baseline characteristics and hearing improvement of 93 patients included in a modified Intention-To Treat analysis.

Prednisolone (n=47) Placebo (n=46) p-value Gender Female 23 17 Male 24 29 0.297 Age Mean ± SD (years) 56.8 ± 12.7 53.8 ± 13.5 0.281

Range (years) 26 - 80 26 - 79 Affected ear Left 25 22

Right 22 24 0.680 Initial pure tone Mean ± SD 66.2 ± 21.2 63.5 ± 16.9 0.507 average for the affected

frequencies (dB HL) Range 32 – 110 32 – 100 Improvement at day eight (dB) Mean ± SD Range 25.5 ± 27.1 -40 to +87 26.4 ± 26.2 -32 to +75 Improvement at day ninety (dB) Mean ± SD Range 39 ± 20.1 -5 to +87 35.1 ± 38.3 -17 to +84 Affected frequency Low frequency region 14 13 regions Mid frequency region 22 24 High frequency region 10 6 “Flat” loss 1 3

0.573

Prevalence of Tinnitus 30 38 0.061 associated symptoms Vertigo 11 14 0.49

No associated symptoms 10 6 0.411 Mean ± SD 3 ± 1.9 3,2 ± 2.3 0.660 Median 3 2

Time from onset of SSNHL to the first at

0

(40)

40 Results

Effect of treatment at the first and final follow-up (day eight and ninety)

No significant difference of hearing recovery was observed either at day eight or day 90 between the Prednisolone group and placebo group regarding the effect of treatment.

The estimated treatment efficacy at the first follow-up was –3.1 dB (p=0.563) and –0.83 dB (p=0.887) at day 90 for the ninety-three patients in modified ITT analysis. (See Table X)

For PPanalysis, the estimated treatment efficacy at the first follow-up was –2.48 dB (p=0.662) and for the total PP analysis –2.91 dB (p=0.675), meaning those patients receiving Prednisolone recovered insignificantly less than the placebo group. The estimated treatment efficacy at the final follow-up for PPanalysis was –0.18 dB (p=0.976) and for the total PP analysis –0.06 dB (p=0.994).

Total recovery occurred in twenty patients after eight days of medication with the study drugs, eleven of those had received Prednisolone (ns). At the final follow-up total recovery had been reached for thirty-nine patients, nineteen of those had received Prednisolone (ns). These thirty-nine patients with total recovery at the final follow-up had a mean hearing loss of 60.5 dB ± 17.4. Thirteen of them had a hearing loss <50 dB HL, six treated with Prednisolone and seven placebo. The range of the hearing loss for the remaining twenty-six patients was 52–90 dB HL.

Table X. Prognostic factors at the first and final follow-up (day eight) on 93 patients in modified Intention-To Treat analysis.

First follow-up Final follow-up Predictor

Coef. SE Coef P-value Coef. SE Coef P-value

Constant 46.80 21.47 39.22 25.76

Treatment -3.114 5.361 0.563* -0.827 5.807 0.887*

Age -0.256 0.189 0.180 -0.346 0.093

Heredity for hearing loss 10.287 6.013 0.091 8.700 6.397 0.178

Vertigo -15.706 5.749 0.008 -18.526 0.004

Tinnitus 4.233 5.788 0.467 -1.764 0.776

Time from onset of SSNHL to the first at the ENT clinics

0.350 1.218 0.775 1.509 1.326 0.259

Prescribed rest or sick leave 0.166 6.565 0.980 -3.367 7.414 0.651

Frequency regions 0.262 ! 0.524 !

Abnormal findings of radiological investigations -22.462 6.148 0.000 ns Abnormal findings of laboratory work-ups 15.178 5.628 0.009 14.424 5.928 0.018

0.203 6.252 6.163

* A single-sided null-hypothesis, that there is less than 10 dB difference of treatment efficacy between the Prednisolon and placebo group was used. ! p-value refers to the partial F-test where null hypothesis is that there is no difference between regions.

(41)

Results 41

Predictive factors at the first and final follow-up (day eight and ninety)

At the first follow-up, three factors were significantly related to outcome regardless of treatment:

The effect of vertigo was –15.7 dB (p=0.008) in the modified ITT analysis, –13.7 dB (p=0.032) in the PPanalysis and –14.5 dB (p=0.041) in the total PP analysis. Abnormal radiological findings had an effect of –22.5 dB (p=0.000) in the modified ITT analysis, –24.3 dB (p=0.001) in the PP analysis and –29.7 dB (p=0.001) in the

total PP analysis.

The effect of abnormal laboratory findings was +15.2 dB (p=0.009) in the modified ITT analysis, +16.4 dB (p=0.008) in the PPanalysis and +17.3 dB (p=0.031) in the total PP analysis.

At the final follow-up, modified ITT analysis and PPanalysis showed slightly different prognostic factors for recovery than at the first follow-up, eight days after the

treatment:

The effect of vertigo was –18.5 dB (p=0.004) in the modified ITT and –20.0 dB (p=0.003) in the PP analysis.

The baseline PTA for the affected frequencies had an effect of +0.4 dB (p=0.014) for in the modified ITT analysis and +0.4 dB (p=0.025) in the PPanalysis. Thus after 90 days, the mean hearing gain was 40% of the initial hearing loss.

The effect abnormal findings of laboratory work-up was +14.4 dB (p=0.018) in the modified ITT analysis and +15.4 dB (p=0.016) in the PPanalysis.

Total PP analysis of seventy-three patients who did take the study drug according to the protocol showed that the only factor affecting the outcome was presence of vertigo. The ability to recover dropped with 20.5 dB (p=0.007) for patients with vertigo.

(42)

42 Results

Paper IV

The aim of Paper IV was to analyze a larger patient group by meta-analysis of data from the RCT together with corresponding material drawn from the Swedish national database for ISSNHL.

A total of one hundred ninety-two patients had been treated according to inclusion, eighty-seven from the RCT (III) and one hundred and five from the Swedish national database (II). At final follow-up audiogram was missing for five patients. Ninety-six of the total one hundred eighty-seven had received Prednisolone, forty-two placebo and

forty-nine no treatment.

Baseline characteristics data for all groups are described in Table XI.

No significant difference in hearing recovery was observed regarding the effect of treatment between the total Prednisolone group and Placebo group plus the control

group without treatment.

The estimated treatment efficacy was 6.10 dB (p=0.06), meaning that patients receiving Prednisolone recovered more than the placebo/no treatment group, but not significantly so.

The recovery with respect to frequency region that was affected by the hearing loss is presented in Table XII.

Analyzing all patients together, three variables were significantly related to outcome regardless of treatment: age at initial contact, presence of vertigo at the onset of hearing loss and abnormal findings of laboratory work-ups.

Age and presence of vertigo at the onset of ISSNHL was associated with less hearing improvement. The effect of age was –0.35 dB (p=0.003) which means that hearing improvement was less with increasing age of the patient (0.35 dB per year). The effect of vertigo was –15.2 dB (p<0.001).

“Abnormal laboratory findings” at the onset of the disease were associated with better prognosis for hearing improvement. “Abnormal laboratory findings” effect on hearing improvement was +11.0 dB (p=0.009). Most of these findings were signs of

inflammation/infection.

Comparison between the placebo group and the control group (no medical treatment)

No significant difference in hearing recovery was observed between the placebo group and control group without treatment. The estimated placebo efficacy was 5.90 dB (p=0.208), meaning that patients receiving placebo recovered not significantly more than those not medically treated.

(43)

(44)

(45)

Discussion 45

DISCUSSION

Sudden sensorineural hearing loss is one of the most mysterious and controversial unsolved entities in otolaryngology. The lack of a standard definition for SSNHL, the low incidence and the fact that some spontaneous recovery occurs in up to 80% of cases, make any evaluation of treatment impossible for those ENT doctors who only see a few patients a year.

The criteria for SSNHL and for the assessment of hearing recovery were developed by the Sudden Deafness Research Committee of the Ministry of Health and Welfare in Japan (1973 and 1981, respectively). However, no specific instruction with regard to how sudden the onset of hearing loss should be, the degree of hearing loss or the affected frequencies were actually provided. Although many authors have selected a hearing loss of 30 dB HL in three contiguous frequencies for their studies1-5_{, the onset} of hearing loss differs from 24 to 72 hours2,6,7_{. In the studies for this thesis (I–IV),} hearing loss over a period of 24 hours was selected in order to stress the concept of “sudden” hearing loss and to avoid including patients with Mb Ménière or

endolymphatic hydrops, where the hearing loss usually develops within a few days. The seemingly low prevalence of ISSNHL in Sweden seen in the present thesis is far below 5–20 per 100.000 per year presented earlier13. However, in one ENT clinic in the countryside to which all patients in the region were referred, the prevalence was estimated to be 8 per 100.000 per year. In city regions, where many patients were treated in private practice, selection made it impossible to estimate

prevalence/incidence. The number of patients to treat for ISSNHL seemed to decline over the study period, which could be observed at all fourteen contributing clinics (III). This may be due to decreasing interest for the study (III). No other reason is evident for a real decrease of incidence. In Japan, an epidemiological study for the last 30 years showed the opposite: an increasing prevalence of ISSNHL14_.

One reason not to be included in the trial was doctor and patient bias. A prerequisite to perform a RCT is to be unbiased. Many of the ENT doctors treating at the contributing clinics had already formed opinions about the effect of corticosteroids and therefore had difficulties in answering patients’ common question: ”How would you treat it if you got this disease, Doc?” This made it difficult for them to recruit patients to the study. Patients today have often studied Internet even before the first visit and already formed a clear opinion about which treatment they wanted; a common reason for patients to refuse to participate was that they wanted corticosteroids.

Radiological and laboratory examination

An acute hearing loss can be a symptom that can be caused by a multitude of known diseases within the vascular system or by different traumas and tumors within the hearing tracts38,42,43,62,63_{. In most cases of SSNHL, it will not be possible to arrive at a} specific diagnosis. However, the assessments of these possible mechanisms should still

Idiopathic Sudden Sensorineural Hearing Loss Corticosteroid Treatment, the Diagnostic Protocol and Outcome