Medical history, cigarette smoking and risk of acoustic neuroma: An international case-control study

Authors


Abstract

Acoustic neuroma (vestibular schwannoma) is a benign tumor of the vestibulocochlear nerve. Its recorded incidence is increasing but risk factors for this tumor have scarcely been investigated. We conducted a population-based case-control study of risk factors for acoustic neuroma in the UK and Nordic countries, including 563 cases and 2,703 controls. Tumor risk was analyzed in relation to medical history and cigarette smoking. Risk of acoustic neuroma was significantly raised in parous compared with nulliparous women (OR = 1.7, 95% CI: 1.1–2.6), but was not related to age at first birth or number of children. Risk was not associated with a history of allergic disease, past head injury, past diagnosis of a neoplasm or birth characteristics, but was significantly raised for past diagnosis of epilepsy (OR = 2.5, 95% CI: 1.3–4.9). Tumor risk was significantly reduced in subjects who had ever regularly smoked cigarettes (OR = 0.7, 95% CI: 0.6–0.9), but the reduction applied only to current smokers (OR = 0.5, 95% CI: 0.4–0.6), not ex-smokers (OR = 1.0, 95% CI: 0.8–1.3). The reduced risk of acoustic neuroma in smokers and raised risk in parous women might relate to sex hormone levels, or smoking might suppress tumor growth, but effects of parity and smoking on timing of diagnosis of the tumor are also a potential explanation. The raised risk in relation to past diagnosis of epilepsy might be a surveillance artefact or imply that epilepsy and/or antiepileptic medication use predispose to acoustic neuroma. These findings need replication by other studies and possible mechanisms need to be clarified. © 2006 Wiley-Liss, Inc.

Acoustic neuroma (vestibular schwannoma) is a nerve sheath tumor of the vestibulocochlear nerve. It is a benign and slow-growing tumor and accounts for the majority of cranial nerve sheath tumors.1 Presenting symptoms include tinnitus, unilateral hearing loss and gait imbalance.2 The recorded incidence is 1–20 per million per year and is increasing,1, 3, 4, 5, 6 which could reflect a real increase or improved neuroimaging technologies. Very little is known about its aetiology. The dominantly inherited disorder neurofibromatosis II is associated with bilateral acoustic neuroma, but is thought to account for a small minority of cases.5, 7 High to moderate doses of ionizing radiation have been shown to increase risk, evidence of which comes mainly from studies of children who received radiotherapy for tinea capitis8 and of the Japanese atomic bomb survivors.9

There are only a few reports of medical risk factors for acoustic neuroma and they have been based on small numbers of cases (<100). One study reported a raised risk with a history of allergies,10 and another a raised risk among men following serious head injury and following exposure to dental X-rays.11 The majority of studies, however, have investigated the relation to mobile phone use,12, 13, 14, 15, 16, 17, 18, 19, 20, 21 of which most did not find an association, but some did.

We conducted a large case-control study of risk factors for acoustic neuroma, with 563 cases and 2,703 controls, in 5 North European countries. We have previously reported on risk of acoustic neuroma in relation to use of mobile phones,21 and report here on risk factors related to medical history and cigarette smoking.

Methods

A population-based case-control study of acoustic neuroma aetiology was conducted from 5 centers in the Nordic countries and the United Kingdom. The study was conducted in Denmark nationwide, in Finland excluding Northern Lapland and Åland, in the Southern and middle regions of Norway, in the Stockholm, Göteborg and Lund regions of Sweden and in the Thames regions of Southeast England. These centers also contribute data to the Interphone study, a 13-country case-control study of brain tumors in relation mobile phone use,22 but have collaboratively and uniformly extended the study to include a wider age range of patients, a greater range of potential risk factors, and collection of blood samples.

Cases were identified through neurosurgery, neuropathology, oncology, neurology and otorhinolaryngology units in the study areas. Lists of cases were also obtained from the appropriate population-based cancer registries to ensure completeness of ascertainment. Eligible cases were individuals diagnosed with acoustic neuroma between September 1999 and February 2005 (the exact dates within this period varied by centre) at ages 20–69 in the Nordic countries and 18–59 in SE England and resident in the study region at the time of diagnosis. The date of diagnosis was defined as the date of the first imaging examination which unequivocally showed the presence of the tumor, or if that was not available, the date of pathological diagnosis.

In the Nordic centers, controls were randomly selected from the population register for the study region, frequency matched to cases on age, sex and region. In the UK, where there is no such accessible population register, frequency-matched controls were randomly selected from general practitioners' (GPs) patient lists. This is a representative source of population-based controls as it has been estimated that 98% of the UK population is registered with a GP.23 Controls were subject to the same age and residence criteria as cases and had never been diagnosed with a brain tumor.

Subjects were invited by letter to participate in the study. If no reply was received, a repeat letter was sent or the subject was contacted by telephone. Each study was approved by the appropriate local ethics committee. Written informed consent was obtained from all subjects at interview.

Data collection

Trained interviewers conducted a personal interview, which usually took place at the subject's home or hospital, or less often at another place convenient for the subject. For almost half the interviews in Norway, however, and a small minority elsewhere, where face-to-face interviews were not possible, interviews were conducted over the telephone. The interview was computer-assisted, with the answers being entered directly into the questionnaire program on a laptop computer, except in Finland, where answers were recorded on a paper copy of the questionnaire and later entered into the program.

Statistical analysis

The study was conducted as part of larger case-control studies of several types of intracranial tumor, using the same questionnaire, with controls recruited for the entire set of cases. To increase statistical power, we used all participants interviewed as controls who fitted the frequency matching strata of the acoustic neuroma cases.

We analyzed risk of acoustic neuroma in relation to medical history and regular cigarette smoking. Factors related to medical history included a past diagnosis of allergy and use of antiallergenic medication, past diagnosis of epilepsy, past diagnosis of a neoplasm, a history of head injury, birth characteristics and factors related to female sex hormones. Regular cigarette smoking was defined as at least 1 cigarette a day for 6 months or more. Conditions or exposures with an onset less than 1 year prior to diagnosis were ignored for the analysis, with the exception that use of hormonal contraceptives and hormonal treatments was included up to the date of interview because information on start and stop ages was not collected in a way that enabled us to calculate exposures up to the diagnosis date. As controls were not individually matched to cases, we constructed a reference date equivalent to the cases' diagnosis date to truncate exposure for controls, based on interview year and how far back the subject was asked to recall exposures, as described previously.21

Postmenopausal status was defined as having had a bilateral oophorectomy or having stopped having periods at least 1 year before the reference date. True menopausal status could not be assessed for women who had had a hysterectomy without bilateral oophorectomy at an age at which they reported that they were premenopausal. Several risk factors were analyzed for pre- and postmenopausal women separately. For the purpose of such stratified analyses, women with unknown menopausal status were classified as premenopausal if they were under age 46 and as postmenopausal if they were aged 55 or over and were excluded from the analysis if they were aged between 46 and 54 years.

Analyses were performed for each country separately and for the overall pooled dataset. Odds ratios were calculated as estimates of relative risks and were obtained using conditional logistic regression, with strata of country, region within country, 5-year age group at the reference date and sex, and adjusted for highest attained educational level and interview year. Heterogeneity in results between countries was assessed with a log likelihood ratio test.24 In instances where there was evidence of heterogeneity (defining heterogeneity, conservatively, as p < 0.10), results were obtained using a two-stage random-effects model.25 Tests for trend were done on the original values rather than on categorized variables and, where appropriate, were carried out with and without the baseline group (i.e. subjects not affected by the condition or exposure), except for age at onset or time since last exposure, because no value could be assigned to the baseline group. The statistical package STATA was used for these analyses.26 All presented p-values are two-sided.

As exposure to ionizing radiation is a known risk factor for acoustic neuroma, we repeated the analyses after adjustment in the model for past diagnostic medical radiation exposures to the head or neck and after excluding subjects who reported having had radiotherapy to these areas 10 or more years prior to the reference date. To investigate possible residual confounding by socio-economic status, analyses were compared with and without adjusting for highest education level.

Results

A total of 732 cases and 6,464 controls were ascertained, and 567 cases and 3,246 controls were interviewed. The overall participation rate was 77% for cases and 50% for controls. As contacts were initially made by post, some of the nonrespondents may have been written to at the wrong address, i.e. never in fact received the letter; participation rates were 81% for cases and 58% for controls based on subjects for whom we could obtain a response man. Reasons for nonparticipation included refusal (cases 11%, controls 33%), difficulty in contacting the subject (cases 5%, controls 14%), no permission from the treating consultant (cases 3%, controls 0%) and illness or death (cases 1%, controls 1%). Four cases were excluded from the analysis because they had neurofibromatosis, leaving 563 cases in the analysis. In total, 2,703 interviewed controls corresponded to matching strata of the cases and were therefore included in the analysis (Table I).

Table I. Characteristics of Acoustic Neuroma Cases and Controls
CharacteristicCases (n = 563) no. (%)Controls (n = 2,703) no. (%)
Sex
 Male264 (46.9)1,247 (46.1)
 Female299 (53.1)1,456 (53.9)
Age at reference date (yrs)
 18–2931 (5.5)145 (5.4)
 30–3998 (17.4)391 (14.5)
 40–49137 (24.3)632 (23.4)
 50–59223 (39.6)1,051 (38.9)
 60–6974 (13.1)484 (17.9)
Country
 Denmark102 (18.1)782 (28.9)
 Finland91 (16.2)572 (21.2)
 Norway45 (8.0)254 (9.4)
 Sweden144 (25.6)479 (17.7)
 SE England181 (32.1)616 (22.8)
Highest attained educational level
 Primary school92 (16.3)506 (18.7)
 Secondary/high school177 (31.4)809 (29.9)
 Medium level technical/professional144 (25.6)760 (28.1)
 University/higher level technical148 (26.3)623 (23.1)
 Not known2 (0.4)5 (0.2)
Marital status
 Single51 (9.1)310 (11.5)
 Married/cohabiting442 (78.5)2,049 (75.8)
 Separated/divorced50 (8.9)248 (9.2)
 Widowed18 (3.2)92 (3.4)
 Not known2 (0.4)4 (0.1)

Allergies

Risk of acoustic neuroma was not significantly related to past diagnosis of asthma, hay fever or eczema, and there was no trend of risk with duration of having each condition (Table II). Tumor risk was not related to a history of seasonal or nonseasonal allergic nasal catarrh and conjunctivitis, food allergy, contact allergy or other types of allergy specified by the participant. The odds ratio for having any of the above conditions was 0.9 (95% CI: 0.8–1.1), and there was no trend of risk with the number of conditions reported (p trend = 0.6). Risk was not significantly associated with use of antihistamines, nasal spray, eye drops or desensitizers nor with the frequency of use of antiasthmatic medication (not in table).

Table II. Risk of Acoustic Neuroma in Relation to Apart Diagnosis of Asthma, Hay Fever and Eczema
Condition1StatusEver diagnosed with conditionDuration of the condition (years)p trend
NoYes<1010–19≥20Not known
  • 1

    Asthma status not known for 3 cases and 3 controls, hay fever status not known for 2 cases and 2 controls and eczema status not known for 3 cases and 1 control.

  • 2

    Odds ratios from conditional logistic regression model stratified by country, region, 5-year age category at the reference date and sex, and adjusted for highest education level and interview year.

AsthmaCase499613112162 
Control2,451249112526619 
OR (95% CI)21.01.1 (0.8–1.5)1.3 (0.8–2.0)1.1 (0.6–2.1)1.1 (0.6–2.0) 0.5
Hay feverCase4591022722494 
Control2,2394621139223126 
OR (95% CI)21.00.9 (0.7–1.1)1.0 (0.6–1.6)1.0 (0.6–1.6)0.9 (0.6–1.2) 0.3
EczemaCase461993526362 
Control2,19750517211020419 
OR (95% CI)21.01.0 (0.8–1.3)1.0 (0.7–1.5)1.2 (0.8–1.9)0.9 (0.6–1.3) 0.7

Epilepsy

Fourteen cases and 33 controls reported a diagnosis of epilepsy more than 1 year prior to the reference date (OR = 2.5, 95% CI: 1.3–4.9) (Table III). Risk of acoustic neuroma was significantly raised in patients who were first diagnosed with epilepsy 10 or more years ago (OR = 4.1, 95% CI: 1.9–8.8), but not for those with more recently diagnosed epilepsy (OR = 0.5, 95% CI: 0.1–3.5). Risk was higher in patients with epilepsy who reported ever use of antiepileptic medication than in those who did not, and was significantly raised for epilepsy diagnosed under age 20 years (OR = 5.8, 95% CI: 2.2–15.6), but not at older ages. For 9 out of 14 cases with past diagnosis of epilepsy we were able to ascertain the reason for the imaging examination that lead to the tumor diagnosis; all 9 patients had been examined because of symptoms indicative of a tumor.

Table III. Risk of Acoustic Neuroma in Relation to Past Diagnosis of Epilepsy and Ever Use of Antiepileptic Medication
FactorCases1 no. (%)Controls1 no. (%)OR (95% CI)2
  • 1

    Information not collected for 29 cases and 28 controls.

  • 2

    Odds ratios from conditional logistic regression model stratified by country, region, 5-year age category at the reference date and sex, and adjusted for highest education level and interview year.

  • *

    p < 0.05.

  • **

    p < 0.01.

  • ***

    p < 0.001.

Diagnosis of epilepsy
Ever
 No519 (97.2)2,637 (98.6)1.0
 Yes14 (2.6)33 (1.2)2.5 (1.3–4.9)**
 Not known1 (0.2)5 (0.2) 
Time since diagnosis (years)
 <101 (0.2)14 (0.5)0.5 (0.1–3.5)
 ≥1013 (2.4)19 (0.7)4.1 (1.9–8.8)***
Age at diagnosis (years)
 <209 (1.7)10 (0.4)5.8 (2.2–15.6)***
 20–393 (0.6)13 (0.5)1.1 (0.3–4.1)
 ≥402 (0.4)10 (0.4)1.4 (0.3–6.8)
Ever used antiepileptic medication
 No6 (1.1)17 (0.6)1.9 (0.7–5.2)
 Yes8 (1.5)16 (0.6)3.2 (1.3–8.0)*

Head injury

Risk of acoustic neuroma was nonsignificantly reduced in participants who reported a history of head injury resulting in loss of consciousness, hospitalization or both (OR = 0.8, 95% CI: 0.6–1.1). There was heterogeneity in odds ratios between countries for this variable (p heterogeneity = 0.05), which was driven by a significantly reduced risk in Denmark (OR = 0.4, 95% CI: 0.2–0.7). No significant associations with tumor risk were found for other countries (range of odds ratios: 0.8–1.3, pooled OR excluding Denmark = 0.9, 95% CI: 0.7–1.1, p heterogeneity = 0.7). Risk was not significantly related to the total number of such head injuries experienced, age at first injury, or time since first injury in the pooled dataset, either with or without the Danish data, or by sex (results not shown).

Birth characteristics

Risk of acoustic neuroma was not related to being first born (OR = 1.1, 95% CI: 0.9–1.4), having been delivered by caesarean section (OR = 0.5, 95% CI: 0.2–1.3), being a twin (OR = 1.1, 95% CI: 0.5–2.2), or birth weight (recorded as ‘light’/‘normal’/‘heavy’) (p trend = 0.9), length of gestation (born ‘early’/‘on time’/‘late’) (p trend = 0.5), maternal age (p trend = 0.7) or paternal age (p trend = 0.7) (results not shown).

Previous neoplasms

On the basis of the pooled data except Denmark, where participants with a previous neoplasm were excluded from the study, 30 cases (6.5%) and 161 controls (8.4%) reported having been diagnosed with a tumor, leukemia or lymphoma prior to the reference date (OR = 0.9, 95% CI: 0.6–1.4). Eight cases and 65 controls specified that their neoplasm was malignant (OR = 0.7, 95% CI: 0.3–1.4). Tumors of the breast were reported by 9 cases and 34 controls (OR = 1.6, 95% CI: 0.7–3.5), including breast cancer by 4 cases and 20 controls (OR = 1.3, 95% CI: 0.4–4.1). Likewise, there were no significant differences between cases and controls in frequencies of other site-specific tumors (results not shown).

Female reproductive factors and sex hormones

The odds ratio for acoustic neuroma was 1.3 (95% CI: 0.5–3.3) in postmenopausal women compared with premenopausal women (Table IV), with a range of 0.2–5.7 between countries (p heterogeneity = 0.04). Risk was highest for menopause under age 40 (OR = 1.9, 95% CI: 0.7–5.0), but there was no significant trend of risk with age at menopause. Ten cases and 40 controls had a surgically induced menopause (OR = 2.1, 95% CI: 0.9–4.8) (not in table).

Table IV. Risk of Acoustic Neuroma in Relation to Menarche, Menopause and Childbearing
FactorCases1 (n = 291) no. (%)Controls1 (n = 1,449) no. (%)OR (95% CI)2
  • 1

    Eight cases and 7 controls excluded because information on female sex hormones was not collected for them.

  • 2

    Odds ratios from conditional logistic regression model stratified by country, region, and 5-year age category at the reference date, and adjusted for highest education level and interview year.

  • 3

    Menopausal status could not be assessed because woman had a premenopausal hysterectomy without bilateral oophorectomy.

  • 4

    The age at menopause was calculated as the age the woman's periods stopped or of bilateral oophorectomy, whichever occurred first.

  • 5

    Results obtained from two-stage random effects model. Test for between-country heterogeneity χ2(df = 8) = 16.3, p = 0.04

  • *

    p < 0.05.

Age at menarche (years)
 <1249 (16.8)192 (13.3)1.0
 12–14177 (60.8)990 (68.3)0.8 (0.6–1.2)
 ≥1557 (19.6)260 (17.9)1.1 (0.7–1.8)
 Not known8 (2.8)7 (0.5)Trend (all subjects) p = 0.15
Menopausal status
 Premenopausal139 (47.8)667 (46.0)1.0
 Postmenopausal115 (39.5)654 (45.1)1.3 (0.5–3.3)5
 Not known due to hysterectomy326 (8.9)95 (6.6)1.4 (0.6–3.3)5
 Not known-other reasons11 (3.8)33 (2.3) 
 Age at menopause (years)4
  <407 (2.4)28 (1.9)1.9 (0.7–5.0)
  40–4945 (15.5)250 (17.3)1.4 (0.8–2.3)
  ≥5059 (20.3)348 (24.0)1.5 (0.8–2.8)
  Not known4 (1.4)28 (1.9)Trend (postmenopausal only) p = 0.8
Ever had live birth
 No43 (14.8)248 (17.1)1.0
 Yes248 (85.2)1,201 (82.9)1.7 (1.1–2.6)*
 Age at first live birth (years)
  <2024 (8.2)123 (8.5)1.5 (0.8–2.8)
  20–2489 (30.6)460 (31.7)1.6 (1.0–2.6)*
  25–2982 (28.2)400 (27.6)1.7 (1.1–2.7)*
  30–3437 (12.7)163 (11.3)2.0 (1.2–3.6)*
  ≥357 (2.4)44 (3.0)1.2 (0.5–3.1)
  Not known9 (3.1)11 (0.8)Trend (parous only) p = 0.6
 Number of children
  149 (16.8)235 (16.2)1.7 (1.0–2.9)*
  2124 (42.6)574 (39.6)1.8 (1.1–2.9)*
  356 (19.2)281 (19.4)1.6 (1.0–2.8)
  ≥419 (6.5)111 (7.7)1.3 (0.7–2.6)
   Trend (all subjects) p = 0.3; trend (parous only) p = 0.6

Ever-parous women had a significantly raised risk compared with nulliparous women (OR = 1.7, 95% CI: 1.1–2.6). Risk in parous women was raised in all countries (range 1.4–3.4, p heterogeneity = 1.0). There was, however, no significant trend in risk with age at first live birth or number of children among parous women (Table IV), or with time since having the first or last live birth (not in table). Risks were similar after adjusting for marital status. There was suggestive evidence that the parity-association in females was different between premenopausal (OR = 2.0, 95% CI: 1.2–3.5) and postmenopausal (OR = 1.1 95% CI: 0.5–2.2) women (p heterogeneity = 0.10). For males, risk was not raised in those who had at least one child compared with no children (OR = 0.9, 95% CI: 0.5–1.7), but with heterogeneity in odds ratios between countries (range 0.4–4.0, p heterogeneity = 0.02). On the basis of the data on tumor size from 51 cases in Denmark,18 because such data were not available for other countries, parous women had on average significantly smaller tumors at diagnosis than nulliparous women (mean 1.4 cm vs. 2.7 cm, respectively, Kruskal-Wallis test p < 0.0001). In the pooled data set, risk of acoustic neuroma in parous women was not related to ever-breastfeeding (OR = 1.1, 95% CI: 0.7–1.8) or to total number of months of breastfeeding (p trend = 0.4) (not in table).

There was no association of risk of acoustic neuroma with ever having used oral contraceptives (OR = 0.9, 95% CI: 0.7–1.3), or hormone replacement therapy (OR = 1.1, 95% CI: 0.8–1.6), or with use of long-acting hormonal contraceptives, hormonal treatment for gynecological disorders or hormonal fertility treatment (not in table).

Cigarette smoking

Cases were significantly less likely to have ever been regular cigarette smokers up to 1 year prior to the reference date than controls (OR = 0.7, 95% CI: 0.6–0.9) (Table V). Risk of acoustic neuroma was, however, only reduced in people who were still smokers 1 year before diagnosis, i.e. ‘current’ smokers (OR = 0.5, 95% CI: 0.4–0.6), and was not reduced in ex-smokers compared with never-smokers (OR = 1.0, 95 % CI: 0.8–1.3), except among those who had stopped within 5 years prior to diagnosis (OR = 0.5, 95% CI: 0.3–1.0). Significantly reduced risks in ‘current’ smokers were observed in each country in the study, with a range in odds ratios between 0.38 and 0.54 (between-country heterogeneity p = 0.9). Risk was also reduced when smoking was evaluated up to the date of diagnosis (OR = 0.5, 95% CI: 0.4–0.6) or up to the date of interview (OR = 0.4, 95% CI: 0.3–0.6). Results were very similar between the sexes, e.g. in relation to ever-smoking, 0.7 (95% CI: 0.5–1.0) in males and 0.8 (95% CI: 0.6–1.0) in females (not in table).

Table V. Risk of Acoustic Neuroma in Relation to Regular Cigarette Smoking
FactorCases no. (%)Controls no. (%)OR (95% CI)1
  • 1

    Odds ratios from conditional logistic regression model stratified by country, region, 5-year age category at the reference date and sex and adjusted for highest education level and interview year.

  • 2

    Current smoker as at 1 year prior to the reference date.

  • 3

    Prior to diagnosis or equivalent reference date for controls.

  • 4

    Number of years smoked multiplied by average number of cigarette packs (packs of 20) smoked per day.

  • *

    p < 0.05,

  • **

    p < 0.01,

  • ***

    p < 0.001.

Ever regular smoker
 No308 (54.7)1,297 (48.0)1.0
 Yes255 (45.3)1,406 (52.0)0.7 (0.6–0.9)**
 Smoking status
  Ex-smoker170 (30.2)705 (26.1)1.0 (0.8–1.3)
  Current smoker284 (14.9)701 (25.9)0.5 (0.4–0.6)***
  Ever-smoked, unclear if current1 (0.2)  
 Age started smoking (yrs)
  <1543 (7.6)262 (9.7)0.6 (0.4–0.9)**
  15–19150 (26.6)807 (29.9)0.8 (0.6–1.0)*
 ≥2061 (10.8)336 (12.4)0.8 (0.6–1.1)
  Not known1 (0.2)1 (0.04)Trend (ever smokers only) p = 0.5
 Years since started smoking
  <106 (1.1)44 (1.6)0.5 (0.2–1.3)
  10–1927 (4.8)134 (5.0)0.6 (0.4–1.0)
  20–2947 (8.3)287 (10.6)0.7 (0.5–1.0)
  ≥30174 (30.9)940 (34.8)0.8 (0.6–1.0)
  Not known1 (0.2)1 (0.04)Trend (all subjects)p = 0.01; trend (ever-smokers only) p = 1.0
 Total years smoked
  <1064 (11.4)260 (9.6)1.0 (0.7–1.3)
  10–1989 (15.8)364 (13.5)1.0 (0.7–1.3)
  20–2945 (8.0)320 (11.8)0.6 (0.4–0.8)**
  ≥3056 (10.0)457 (16.9)0.5 (0.3–0.7)***
  Not known1 (0.2)5 (0.2)Trend (all subjects)p < 0.001; trend(ever-smokers only) p < 0.001
 Years since stopped smoking3
  ≥3025 (4.4)109 (4.0)1.1 (0.7–1.8)
  20–2954 (9.6)178 (6.6)1.3 (0.9–1.8)
  10–1952 (9.2)204 (7.6)1.0 (0.7–1.4)
  5–928 (5.0)114 (4.2)1.0 (0.7–1.6)
  >1–411 (2.0)95 (3.5)0.5 (0.3–1.0)*
  Current smoker84 (14.9)701 (25.9)0.5 (0.4–0.6)***
  Not known1 (0.2)5 (0.2)Trend (ever-smokers only) p < 0.001
 Smoking pack-years4
  <10123 (21.8)543 (20.1)0.9 (0.7–1.1)
  10–1969 (12.3)359 (13.3)0.8 (0.6–1.1)
  20–2926 (4.6)218 (8.1)0.5 (0.3–0.7)**
  ≥3036 (6.4)270 (10.0)0.6 (0.4–0.8)**
  Not known1 (0.2)16 (0.6)Trend (all subjects) p < 0.001; trend (ever-smokers only)p = 0.02

Among ever-smokers, there was a significant trend of decreasing risk with increasing number of years of smoking (p < 0.001) and with pack-years of smoking (p = 0.02) (Table V). Risk was significantly reduced for having smoked for 20 or more years and for smoking 20 or more pack-years, but not for shorter durations or cumulative amounts.

When considering current and ex-smokers separately, there were significant trends of risk among current smokers with total years smoked, time since started smoking and pack-years of smoking if never-smokers were included in the analysis, but nonsignificant trends if they were not included (Table VI). For ex-smokers, no trends of risk were observed, except for a borderline significant trend of risk with duration of smoking when excluding never-smokers (p trend = 0.06). There was no trend of risk with the daily number of cigarettes smoked among smokers overall, or for current and ex-smokers separately (not in table).

Table VI. Risk of Acoustic Neuroma in Relation to Cigarette Smoking, by Smoking Status at 1 Year Prior to the Reference Date
FactorSmoking status 1 year prior to the reference date
Current smokerEx-smoker
CasesControlsOR (95% CI)1CasesControlsOR (95% CI)1
  • Trend (all subjects) p = 0.5; trend (ex-smokers only) p = 0.2

  • 1

    Odds ratios from conditional logistic regression model stratified by country, region, 5-year age category at the reference date and sex and adjusted for highest education level and interview year.

  • 2

    Number of years smoked multiplied by average number of cigarette packs (packs of 20) smoked per day.

  • *

    p < 0.05.

  • **

    p < 0.01.

  • ***

    p < 0.001.

Total years smoked
 Never smoked3081,2971.03081,2971.0
 <109470.7 (0.3–1.6)552131.0 (0.7–1.4)
 10–19181220.5 (0.3–0.9)*712421.2 (0.9–1.7)
 20–29141770.3 (0.2–0.6)***311430.9 (0.6–1.4)
 ≥30433550.5 (0.4–0.8)**131020.5 (0.3–1.0)*
 Not known  Trend (all subjects) p < 0.001; trend (current smokers only) p = 0.5 5Trend (all subjects) p = 0.4; trend (ex-smokers only) p = 0.06
 Years since started smoking
  <106320.8 (0.3–2.0)0120 (0–)
  10–1913870.4 (0.2–0.8)*14471.0 (0.5–2.1)
  20–29131620.3 (0.2–0.6)***341251.3 (0.8–2.0)
  ≥30524200.6 (0.4–0.8)**1225201.0 (0.8–1.3)
  Not known  Trend (all subjects)p < 0.001; trend(current smokers only)p = 0.8 1Trend (all subjects)p = 0.9; trend (ex-smokers only)p = 0.8
 Smoking pack-years2
  <10261950.5 (0.3–0.8)**973481.1 (0.9–1.5)
  10–19251750.6 (0.4–0.9)*441841.0 (0.7–1.5)
  20–29131400.4 (0.2–0.7)**13780.7 (0.3–1.3)
  ≥30201880.5 (0.3–0.8)**16820.8 (0.4–1.5)
  Not known 3Trend (all subjects) p < 0.001; trend (current smokers only) p = 0.9 13Trend (all subjects)p = 0.5; trend (ex-smokers only)p = 0.2

On the basis of Danish data only, never-smokers had a mean tumor size of 1.6 cm (SD = 0.83), ex-smokers 1.5 cm (SD = 0.70) and current smokers 1.4 cm (SD = 0.85); there was no significant difference in tumor size between these groups (analysis of variance p = 0.6).

Further analyses

Additional adjustment of odds ratios in the model for past diagnostic medical radiation exposure, or exclusion of the 5 cases and 13 controls who had radiotherapy to the head more than 10 years prior to the reference date, did not affect the findings. Odds ratios for individual conditions or exposures were very similar with and without adjustment for educational level.

There was some evidence of heterogeneity in the parity effect by cigarette smoking status and vice versa. Risk in relation to being ever-parous was 2.4 (95% CI: 1.2–4.9) among ever-smokers and 1.3 (95% CI: 0.7–2.3) among never-smokers and risk in relation to ever-smoking was 0.3 (95% CI: 0.1–0.8) among nulliparous and 0.8 (95% CI: 0.6–1.1) in parous women (p heterogeneity = 0.04). There was no evidence for heterogeneity in odds ratios by smoking or parity status for other variables presented in this article (results not shown).

Discussion

Our population-based case-control study showed a strong inverse association of acoustic neuroma risk with cigarette smoking, and positive associations with parity and epilepsy, but no association with other factors related to medical history. To our knowledge, this study is much the largest and most comprehensive investigation to date into the aetiology of acoustic neuroma.

The reduced risk of acoustic neuroma in ever-smokers was confined to current smokers and those who had recently stopped smoking. Cigarette smoking has not generally been associated with a reduced risk of central nervous system or brain tumors,27, 28, 29 but there have not been any previous analyses of acoustic neuroma risk. There is evidence for an inverse association of smoking with endometrial cancer30 and (more inconsistently) with postmenopausal breast cancer.31 These are thought to be due to an antiestrogenic effect of compounds in cigarette smoke,30, 32 so one possibility is that our finding for acoustic neuroma is also a hormonal effect. Cigarette smoking has effects on hormone levels in both sexes, with raised androgen and sex hormone-binding globulin levels found in smokers.30, 33, 34, 35, 36 An inverse association has also been reported between cigarette smoking and the risk of Parkinson's disease,37 which might be due to nicotine as a neuroprotective agent.38

The relationships of acoustic neuroma risk to duration of smoking and cumulative pack years would accord with an etiological explanation, but the restriction of the relation to current smokers and recent quitters would fit better with an effect of smoking on either tumor growth or presentation because the tumor is likely to exist for several years before diagnosis, being symptomatic or only mildly symptomatic for long periods before eventual diagnosis. In a very large case series, 95 percent of patients reported prediagnostic hearing deficits, for an average of 3.7 years, and 61 percent of patients reported symptoms related to vestibular nerve disturbances, for an average of 2.1 years before eventual diagnosis with acoustic neuroma.2 The timing of the eventual diagnosis is likely to be influenced by the person's awareness and individual response to symptoms, which could include social and psychological factors related to smoking. In addition, cigarette smoking has been associated with hearing loss,39 which might make additional tumor-induced hearing loss less noticeable. The inverse association of risk with smoking might therefore be due to detection bias, in that smokers might be less likely to present to a doctor or be referred for specialist examination, because they fail to recognize hearing loss early, or they or their doctor incorrectly attribute their symptoms to smoking. Arguing against detection bias, one might expect that if such bias caused late presentation, then current smokers would have larger tumors at diagnosis than ex- or never-smokers; we found no evidence for this, based on limited data. Symptom incidence and duration have been reported not to correspond well with tumor size, however.2

Our study showed a significantly increased risk of acoustic neuroma in ever-parous women compared with nulliparous women, but we found no clear associations with age at first birth or number of children. Furthermore, risk was not associated with age at menarche, menopausal status, exogenous hormones such as hormone replacement therapy, or breastfeeding. There was some evidence that the association with parity was stronger in ever-smokers and in premenopausal women, although the effect of smoking itself was protective in females overall. The interpretation of this is unclear, and such data-driven subset analyses need to be treated with caution; it is possible that these findings are due to chance. The observation that in males there was no association with having children makes an explanation based on bias less likely. Parous women have lower prolactin and free oestradiol levels and higher levels of sex hormone-binding globulin than nulliparous women,40, 41, 42 as well as lower levels of androgens.41, 43 The raised risk in parous women could, however, also be an artifact; mothers with children might present with the tumor earlier than women with no children because they become aware of hearing deficits earlier, either because their immediate family notice their hearing loss, or because they visit a physician more frequently (e.g. when accompanying their children). Support for this was found in that parous women were diagnosed with smaller tumours than nulliparous women, based on a subset of the data. There is little evidence from other sources about whether a hormonal mechanism is plausible, except that acoustic neuroma has been reported to express steroid receptors,44, 45, 46 tumors in pregnant women have reported to be larger and more vascular47, 48, 49, 50 and a moderate excess female to male sex ratio in incidence rate has been observed in several studies,3 although not in the largest one.1 A nationwide Danish study showed a marked female excess of acoustic neuroma at postmenopausal but not at younger ages.3

The raised risk of acoustic neuroma in patients with a past diagnosis of epilepsy could be etiological, in that epilepsy or its treatment predispose to acoustic neuroma, or could be a surveillance artefact. The latter could occur if acoustic neuroma is diagnosed incidentally following MRI scanning for epilepsy or if patients with epilepsy are more readily referred for further investigation than subjects without epilepsy when they present with symptoms related to the tumor. Acoustic neuroma has been diagnosed in 2–7 per 10,000 asymptomatic subjects investigated with MRI,51, 52 and an estimated 0.6% of the European population in the age group 20–64 years has active epilepsy.53 We found no evidence from case notes, however, that tumors after epilepsy had been diagnosed incidentally. In addition, a surveillance artefact as a possible explanation is not supported by our finding that risk was only raised for epilepsy of 10 or more years' duration, but not of shorter durations. Elevated risks of glioma and meningioma have been reported in patients with epilepsy,54, 55, 56 although the interpretation of these studies is hampered by epilepsy being an early symptom of these tumors.

We found no association of risk of acoustic neuroma with a history of allergies, past head injuries, birth characteristics or past diagnosis of a neoplasm, the latter based on small numbers. Previous data on acoustic neuroma risk in relation to these topics is scarce. Our findings contradict those from a much smaller study reporting raised risks of acoustic neuroma in relation to past allergic disease,10 and suggest that the inverse association of allergy seen for glioma, and, less consistently, with meningioma,57 does not apply to acoustic neuroma. Our findings also do not accord with a previous report of a 2-fold increased acoustic neuroma risk in men 30 or more years after serious head injury,11 but are consistent with a cohort study of subjects hospitalized for head trauma, which did not find raised risks of intracranial schwannoma after excluding the first year of follow-up (Standardized incidence rate = 0.8, 95% CI: 0.4–1.7).58

Risk of acoustic neuroma has been reported to be positively related to household income and educational level.59 Our analyses were adjusted for educational level, although this adjustment barely affected the results. Ionizing radiation is an established cause of this tumor but only few subjects reported past radiotherapy to the head or neck, and exclusion of these subjects or adjustment of odds ratios for past diagnostic radiation exposure did not affect the results. Raised risks of acoustic neuroma have also been reported after loud noise exposure,11, 60 but it is unknown whether this is real or artefactual, and it seems unlikely to explain our findings.

In conclusion, our study shows that acoustic neuroma risk is inversely associated with cigarette smoking and, in females, positively related to being ever parous. These findings could reflect a hormonal aetiology or a suppressive effect of smoking on tumor growth, but an effect of smoking and parity on the timing of presentation with the tumor is a possible alternative explanation. The increased risk found in relation to a past diagnosis of epilepsy could be due to a surveillance artifact or could imply that epilepsy and/or use of antiepileptic medication predispose to acoustic neuroma. These findings need repetition in future large studies, and possible mechanisms need to be clarified.

Acknowledgements

The UICC received funds for this study from the Mobile Manufacturers' Forum and the GSM Association. Provision of funds to the Interphone study investigators via UICC was governed by agreements that guaranteed Interphone's complete scientific independence. The views expressed in the publication are those of the authors and not necessarily those of the funders. The Nordic-UK collaborative group thanks all participants for their valuable contribution to this study. We thank the IARC team, in particular Dr. E. Cardis, Dr. I. Deltour and Dr. L. Richardson, for their input in this study, and James Doughty and Jan Ivar Martinsen for programming work. We thank Professor M. Brada (Royal Marsden NHS Trust, UK) and Dr. P. Cayé-Thomasen (Gentofte University Hospital of Copenhagen) for support and valuable advice. The Finnish centre thanks Dr. J. Jääskeläinen (Helsinki University Hospital), Dr. S. Valtonen (Turku University Hospital), Prof. J. Koivukangas (Oulu University Hospital), Prof. M. Vapalahti (Kuopio University Hospital), Dr. T. Kuurne (Tampere University Hospital) and Prof. R. Sankila (Finnish Cancer Registry). The Swedish centre thank the Swedish Regional Cancer Registries and the hospital staff; especially the following key persons at the hospitals: Dr. J. Boethius, Prof. I. Langmoen, Dr. T. Mathiesen, Dr. I. Ohlsson Lindblom and Dr. H. Stibler (Karolinska University Hospital), Dr. J. Lycke, Dr. A. Michanek and Prof. L. Pellettieri (Sahlgrenska University Hospital), Prof. T. Möller and Prof. L. Salford (Lund University Hospital). The Southeast England centre thank the Thames Cancer Registry, and the following consultants and their teams for their support: Mr. G. Brookes, Mr. A.D. Cheesman, Prof. M.J. Gleeson and Mr. N.D. Kitchen (National Hospital for Neurology and Neurosurgery), Mr. R. Bradford (Royal Free Hospital), Mr. C. Hardwidge, Mr. J.S. Norris and Dr. M. Wilkins (Princess Royal Hospital), Mr. M.M. Sharr, Prof. A.J. Strong and Mr. N. Thomas (King's College Hospital), Prof. A. Bell, Mr. H. Marsh and Mr. F. Johnston (St George's Hospital), Mr. K.S. O'Neill and Mr. N.D. Mendoza (Charing Cross Hospital), Mr. R. MacFarlane (Addenbrooke's Hospital) and Mr. A.R. Aspoas and Mr. S. Bavetta (Oldchurch Hospital).

Ancillary