1. Introduction
Subjective tinnitus is the phantom perception of sound in the absence of a corresponding external acoustic stimulus. As there are no objective measures, the prevalence of tinnitus has always been self-reported and varies between 5.1 to 42.7% depending on its definition [
1]. However, a rarer form is severe tinnitus, which affects 1–3% of the population to the point of impacting quality of life [
2,
3]. Recent studies converge towards the notion that subjective tinnitus results from a maladaptive plasticity in response to sensory deprivation, such as hearing loss [
4]. Damage in the hearing system (the cochlea) leads to central neural compensatory responses (central gain) along the auditory pathway, something which is observed in both animals and humans [
5]. In addition, limbic structures (e.g., amygdala) have been found to interact with the auditory pathway in animals and humans with tinnitus and is thought to exacerbate tinnitus-related hyperactivity [
6]. Recently, it has been suggested that the lack of effective treatment may be due to the heterogeneity of tinnitus [
7], whose subtypes remain to be established [
8].
Tinnitus is mainly considered a symptom and has been thought to derive only from environmental influences. A handful of studies have made attempts to reveal genetic influences, with mixed results [
9]. For instance, a large multicenter study with 198 families showed that the risk of having tinnitus when having another member of the family with tinnitus was 1.7 [
10]. Another population-based family study (
n = 28,066 participants) determined a heritability of H
2 = 11% [
11]. We previously speculated that possible reasons for these failures may have resided in poor definitions of tinnitus [
12]. In a twin study, when considering laterality as a potential classification of tinnitus, a heritability of 68% was found for bilateral tinnitus in men, while in contrast, unilateral tinnitus showed a heritability of 27%, suggesting a greater involvement of genetic factors in the former [
13]. As twin studies may still show bias due to uncontrolled shared-environment effects [
14], we undertook an adoption study based on national medical registry data in which tinnitus was clinically diagnosed by a medical doctor. The odds ratio (OR) for tinnitus was 2.01 (95% CI, 1.10–3.69) for adoptees of affected biological parents, whereas the OR for adoptees with an affected adoptive parent was 1.04 (95% CI, 0.53–2.04) adjusted for age, sex, county, educational attainment, depression, anxiety, and hearing loss, confirming that the familial transmission of clinically significant tinnitus is genetically influenced (H
2 = 31%) and not due to shared-environment [
15]. These two studies support a genetic contribution to tinnitus in families. Indeed, Clifford et al. recently published the first large genome-wide association study using the UK Biobank and replicated 3 loci and 8 genes in the Million Veterans Program with a single-nucleotide variants (SNP) heritability of 6.3% [
16], providing a first basis of tinnitus as a neurological disorder [
17].
Here, we used familial clustering as an analysis to estimate the relative sibling risk for tinnitus. Familial aggregation studies are used in genetic epidemiology to estimate recurrence risk of a trait within a family, which is the initial step to identify hereditary conditions [
18]. The general approach is to determine whether having a relative with tinnitus increases one’s risk of developing tinnitus. This is assessed by a recurrence risk ratio or sibling recurrence risk (lambda sib [λs]). λs is defined as the risk to siblings of probands (the sample individual) for a specific condition, relative to the population prevalence (see
Supplementary Materials 1) [
19]. λs has been used to estimate the genetic risk for a wide range of diseases or traits. For instance, in schizophrenia, the sibling recurrence rate is 9% whereas the prevalence in the population is 1% (λs = 9).
3. Results
We aimed to quantify the familial aggregation of tinnitus by estimating λs for different forms of tinnitus, namely, bilateral or unilateral, constant, and severe, using data from 2446 participants from the Swedish Tinnitus Outreach Project (STOP). The challenge was to identify estimates for these different forms of tinnitus in the general Swedish population. We identified four large national studies, the combination of which allowed us to determine the prevalence for bilateral, unilateral, constant, and severe tinnitus (
Table 1). Sex prevalence estimates showed a clear bias towards greater prevalence in males for all tinnitus subtypes.
We next estimated the λs for all potential subtypes of tinnitus by comparing prevalence of tinnitus within participants from the Swedish Tinnitus Outreach Project who were recruited for the purpose of tinnitus subtyping studies (
Table 2). We could not identify articles in which λs included confidence intervals and
p-values taking uncertainty in both the population and the familial prevalence estimates into account. Hence, we implemented a percentile bootstrap approach to provide accurate estimates and significance values for the calculated λs (see
Supplementary Materials 1). The recurrence ratios for participants with bilateral was λs
Bil = 1.79 (95% CI (1.55–2.04)) and did not differ from that of unilateral tinnitus λs
Unil = 1.99 (95% CI (1.45–2.56)). The λs of constant tinnitus increased to 2.29 (95% CI (2.01–2.58)). Interestingly, with increasing severity, tinnitus showed the highest lambda scores, reaching 7.27 (95% CI (5.56–9.07)) for severe tinnitus.
With a sex-specific bias in the prevalence of these different forms of tinnitus, we performed sex stratified analyses. In contrast to our expectations, we found consistently higher λs for women than for men, although being significant only for constant tinnitus (Constant T♂: 1.58 (1.31–1.86); Constant T♀: 3.32 (2.75–3.92)) and severe tinnitus (Severe T♂: 5.03 (3.22–7.01); Severe T♀: 10.25 (7.14–13.61)). The sexual dimorphism was also obvious for bilateral tinnitus, although not being significant between the two sexes (Bilateral T♂: 1.50 (1.23–1.78); Bilateral T♀: 2.16 (1.74–2.60)). Overall, the estimated recurrence risk ratios appear to be greater in women for some types of tinnitus tested here, suggesting a greater genetic susceptibility in women in particular to constant and severe and potentially bilateral tinnitus.
4. Discussion
The present study reveals that the greatest recurrence risk for siblings occurs in subjects with severe tinnitus. These findings are consistent with our recent adoption study using national registry-based data revealing a genetic contribution to the familial transmission of “clinically significant” tinnitus and a lack of shared-environment effects [
15]. Since a large proportion of subjects have tinnitus severe enough to seek medical support, it is thus very likely that the generation of severe tinnitus is genetically influenced, more so than any other form of tinnitus. Although the adoption study could not reveal sex differences in tinnitus liability due to limitations in the sample size, the present work suggests that a greater risk occurs in women.
We previously revealed in a twin study a greater heritability for bilateral tinnitus in men (H
2: 68% (95% CI: (63–73))) when compared to women (H
2: 41% (95% CI: (23–58))) [
13]. Noteworthy, this gender pattern was inverted (62% in females, 22% in males) when focusing on younger groups (<40 years of age), albeit not reaching significance likely due to sample size limitations [
13]. Here, 27.4% of the participants were below 34 years of age, suggesting that young age may be an important contributor to phenotypic expression of genetically transmitted tinnitus in women. In this regard, while the prevalence of tinnitus is greater in men, there are increasing reports showing that tinnitus is more bothersome and psychologically impactful in women [
1,
30,
31]. It is thus possible that the development of constant and/or severe tinnitus is more genetically influenced in women than in men. The sibling recurrence risk observed here in women with severe tinnitus (9.73) is in the range of what has been found for schizophrenia (9) and bipolar disorders (7.9) [
32]. However, the sex bias we observe for tinnitus suggests a strong influence of sex in the pathophysiology of constant and severe tinnitus. The results of this study have substantial implications for further genetic studies mapping gene variants related to tinnitus, indicating that sex and severity are key elements of a subtype that is genetically determined. A recent large genome-wide association study only revealed a handful of loci associated with a broad definition of tinnitus [
17], but the selection of severe cases and stratification by sex may lead to the identification of novel pathways with a sex-specific involvement in the pathophysiology of tinnitus. This approach has been proven effective in elucidating the genetic landscape of autism [
33], bipolar disorder [
34], major depressive disorder [
35], and obsessive-compulsive disorder [
36].
Emerging evidence suggests serotonin could play a role in the mechanism of tinnitus severity. The 5-HTTLPR polymorphism in the promoter of the serotonin transporter gene
SLC6A4 was found associated with greater tinnitus severity [
37]. Although no studies have replicated these findings, a pilot GWAS on tinnitus showed an enrichment in serotonin signaling but without identifying genome wide significant variants due to limited sample size [
38]. In support of the potential contribution of serotonin to tinnitus severity, serotonin has been shown to enhance signaling only from the multisensory input in the dorsal cochlear nucleus (DCN), while decreasing input from auditory fibers [
39]. More specifically, serotonin increased excitability and spontaneous firing of vertical cells, leading to a vertical cell-mediated inhibitory activity in fusiform cells. Such mechanism could underlie the enhanced multisensory processing in the DCN observed during tinnitus in animal models [
40,
41], and whose bimodal somatosensory and auditory stimulation can decrease the severity of tinnitus [
42]. Additional studies are needed to clarify the role of serotonin on tinnitus severity.
The present findings showing a high familial transmission of severe tinnitus may explain the low heritability of tinnitus reported in previous studies [
10,
11]. In one of the studies, tinnitus was defined as “
Nowadays, do you ever get noises in your head or ear (tinnitus) which usually last longer than five minutes?”, which is very broad and may encompass a large number of subtypes and thus overestimate prevalence. The second one used the question “
Are you bothered by ringing in your ears?”, which does not specifically define whether tinnitus is constantly and presently perceived, nor whether it is severe. Thus, the inclusion of separate items for the present percept, severity and duration is important to consider when classifying tinnitus.
To our knowledge, there are no studies providing recurrence risk ratio along with measures of dispersion (e.g., CI,
p-value) taking uncertainty in both population and familial prevalence estimates into account. We provide here a bootstrap-based methodology allowing to generate such values when the ratio uses two distinct data sources for population and familial prevalence estimates—an approach that can be used in other diseases. Indeed, estimates of the recurrence risk ratio are highly dependent on available prevalence estimates in the population. Previous systematic reviews have revealed the prevalence for constant, or severe tinnitus, however the reported values varied a lot between studies likely due to the formulation of the questions addressing tinnitus, but also regional effects [
1]. Thus, the use in the present study of different national cohorts with data on specific subtypes (laterality, severity, and whether it is constant or intermittent) is a major strength, something that cannot be achieved with the current medical registry data codes for tinnitus.
This study has several limitations. The first one is the ecological design of the study: Data are analyzed at the population or group level, rather than at the individual level. Prevalence of different types of tinnitus in the general population is computed from a set of large studies; the same is computed among siblings of tinnitus subjects; finally, the two prevalence estimates are compared. The source for these estimates is different with potentially differences in mean age and differences in the assessment of tinnitus. Furthermore, due to the different sources of population and familial estimates, it was not possible to control for chronic ear diseases that could have contributed to tinnitus. In this regard, information on hearing (e.g., audiometry, or self-reported hearing ability) is also missing, but given the study design, it would require similar individual audiometric or hearing information within each population study and the one in which sibling risk is estimated. In tinnitus studies, it is important to consider auditory measures beyond the clinical standard which is set at 8 kHz [
43]. Many individuals have normal hearing thresholds up to 8 kHz, but 84.8% of these (
n = 589) show elevated thresholds (>20 dBHL) at higher frequencies when assessed up to 16 kHz (C.R.C., unpublished data). The second is a bias in recall whereby 27.2% of participants could not recall whether they have a family member that is affected. In the cases where affected sibling(s) were reported, there is no information whether the tinnitus type matches the one of the proband, nor whether the affected siblings were clinically diagnosed, or their tinnitus was self-reported. Since the prevalence of self-reported tinnitus is very close to that of diagnosed tinnitus (severe tinnitus: 2.55% (2.45–2.65),
n = 92,287; ICD
9–10 H93.1: 2.77% (2.65–2.89),
n = 74,351), diagnosed tinnitus may be used as a proxy of severe tinnitus. Hence, the use of national registry data would thus allow addressing the above-mentioned limitations with adjustments for multiple variables likely to contribute to tinnitus and determine the genuine sibling risk for tinnitus.