Mortality and Cancer Incidence After Exposure to Blue Asbestos in Childhood: A Further 10 Years of Follow‐Up

TL;DR

Exposure to blue asbestos in childhood is linked to higher cancer rates and mortality in adulthood, according to a long-term study.

Contribution

The study confirms and extends prior findings by showing a clear exposure-response relationship for asbestos-related diseases.

Findings

Males had increased risks of all cancers, mesothelioma, melanoma, and lip/mouth/liver/brain cancers.

Females showed elevated risks for all cancers, mesothelioma, and ovarian and brain cancers.

Longer and higher asbestos exposure correlated with higher mesothelioma rates, supporting an exposure-response relationship.

Abstract

The impact of early‐life exposure to asbestos on disease risk remains uncertain. Childhood exposure to blue asbestos at Wittenoom has previously been linked to the development of malignant mesothelioma and various cancers in adulthood, as well as to a greater risk of all‐cause mortality compared with the general population. This study aims to provide an update on mortality and cancer incidence rates after this exposure.

The cohort of all those who lived in the asbestos mining town of Wittenoom as children (less than 15 years of age; 1279 males and 1185 females) was linked to state and national cancer and death registries. We calculated standardized incidence ratios (SIRs) for a range of cancers, and standardized mortality ratios (SMRs) for all‐cause and cause‐specific mortality for the cohort compared with the general Western Australian population.

Compared with the Western Australian population, males from the cohort had an increased risk of all cancers and mesothelioma, as well as melanoma and cancers of the lip and mouth, liver, and brain. Females had a significantly elevated risk of all cancers, mesothelioma, and cancers of the ovary and brain. Higher rates of mesothelioma were observed among those with a longer duration of exposure and higher cumulative exposure, consistent with a known exposure–response relationship. Former Wittenoom children also had a greater risk of all‐cause mortality and mortality from cancer, mesothelioma, and ill‐defined symptoms.

This update confirms earlier studies and shows that exposure to asbestos in childhood is associated with several cancer and mortality outcomes in adulthood.

Tables

Table 1. Demographics and asbestos exposure of former Wittenoom children, by sex.

Characteristic	Male (n = 1279)	Female (n = 1185)	Total (n = 2464)
	n (%)	n (%)	n (%)
Age at first exposure (years)
< 5	826 (64.6)	741 (62.5)	1567 (63.6)
5–9	332 (25.9)	313 (26.4)	645 (26.2)
10 < 15	121 (9.5)	131 (11.1)	252 (10.2)
Year of first exposure
1943–1958	457 (35.7)	421 (35.5)	878 (35.6)
1959–1966	555 (43.4)	488 (41.2)	1043 (42.3)
1967–1985	267 (20.9)	276 (23.3)	543 (22.1)
Length of exposure (years)
< 2	775 (60.6)	701 (59.2)	1476 (59.9)
2 < 5	318 (24.9)	314 (26.5)	632 (25.6)
5 < 10	119 (9.3)	109 (9.2)	228 (9.3)
10–37	55 (4.3)	49 (4.1)	104 (4.2)
Unknown	12 (0.9)	12 (1.0)	24 (1.0)
Cumulative exposure (f/mL‐year)
< 5	793 (62.0)	746 (63.0)	1539 (62.5)
5 < 10	262 (20.5)	215 (18.1)	477 (19.4)
10–69	214 (16.7)	216 (18.2)	430 (17.4)
Unknown	10 (0.8)	8 (0.7)	18 (0.7)
Time since first exposure (years)a
33–39	5 (0.4)	9 (0.8)	14 (0.6)
40–49	149 (11.6)	148 (12.5)	297 (12.0)
50–59	607 (47.5)	558 (47.1)	1165 (47.3)
60–69	480 (37.5)	427 (36.0)	907 (36.8)
70–75	38 (3.0)	43 (3.6)	81 (3.3)
Smoking status
Ever	568 (44.4)	513 (43.3)	1081 (43.9)
Never	299 (23.4)	350 (29.5)	649 (26.3)
Unknown	412 (32.2)	322 (27.2)	734 (29.8)

Table 2. Standardized incidence ratios (SIRs) and standardized mortality ratios (SMRs) (95% CI) among male former Wittenoom children, compared with the Western Australian male population.

Cancer incidencea	N	SIR1 (95% CI)b , c	SIR2 (95% CI)b , d
All	190	1.22 (1.05–1.40)	2.06 (1.78–2.38)
Mesothelioma	48	33.81 (24.93–44.83)	58.97 (43.48–78.18)
Lung cancer	8	0.74 (0.32–1.45)	1.32 (0.57–2.60)
All excluding MM and lung	134	0.93 (0.78–1.10)	1.57 (1.32–1.86)
Bladder	4	1.34 (0.37–3.43)	2.46 (0.67–6.29)
Brain	7	2.26 (0.91–4.65)	3.43 (1.38–7.06)
Colorectal	11	0.69 (0.34–1.23)	1.15 (0.58–2.06)
Lip and mouth cancer	11	2.35 (1.17–4.20)	3.39 (1.69–6.07)
	11	2.09 (1.04–3.73) e	3.01 (1.50–5.39) e
Liver	7	2.58 (1.04–5.31)	4.90 (1.97–10.10)
	7	2.46 (0.99–5.07)e	4.67 (1.88–9.63) e
Melanoma	24	1.02 (0.65–1.52)	1.57 (1.00–2.33)
Prostate	23	0.56 (0.35–0.84)	1.06 (0.67–1.59)

Table 3. Standardized incidence ratios (SIRs) and standardized mortality ratios (SMRs) (95% CI) among female former Wittenoom children, compared with the Western Australian female population.

Cancer incidence	N a	SIR1 (95% CI)b , c	SIR2 (95% CI)b , d
All	150	1.02 (0.86–1.20)	1.54 (1.30–1.80)
Mesothelioma	24	62.51 (40.05–93.01)	104.61 (67.02–155.65)
Lung cancer	10	1.13 (0.54–2.08)	1.92 (0.92–3.54)
All excluding MM and lung	116	0.83 (0.69–1.00)	1.25 (1.03–1.50)
Brain	5	2.49 (0.81–5.80)	3.60 (1.17–8.39)
Breast	46	0.84 (0.62–1.12)	1.27 (0.93–1.69)
Colorectal	4	0.32 (0.09–0.82)	0.50 (0.14–1.27)
	4	0.31 (0.09–0.81)	0.49 (0.14–1.25)
Melanoma	14	0.74 (0.41–1.25)	1.05 (0.57–1.76)
Ovarian	9	2.38 (1.09–4.52)	3.51 (1.61–6.67)
	9	2.36 (1.08–4.48) e	3.48 (1.60–6.62) e

Table 4. Mesothelioma incidence rates per 100,000 person‐years at risk, by categories of asbestos exposure, among former Wittenoom children.

	Total		Male		Female
	N	Ratea (95% CI)	N	Ratea (95% CI)	N	Ratea (95% CI)
Total	72	59.4 (52.4–66.4)	48	77.4 (66.2–88.6)	24	40.5 (32.2–48.8)
Age at first exposure
< 5 years	43	57.7 (48.9–66.5)	31	80.2 (65.8–94.5)	12	33.5 (23.8–43.1)
5–9 years	23	70.3 (55.6–85.0)	14	84.6 (62.0–107.2)	9	55.7 (37.1–74.2)
10 < 15 years	6	42.9 (25.4–60.4)	3	44.2 (18.7–69.8)	3	41.6 (17.6–65.6)
Year of first exposure
1943–1958	36	73.8 (61.5–86.2)	25	100.2 (80.1–120.2)	11	46.2 (32.3–60.2)
1959–1966	30	59.5 (48.7–70.4)	19	72.6 (55.9–89.2)	11	45.4 (31.7–59.1)
1967–1985	6	27.1 (16.1–38.2)	4	36.8 (18.4–55.2)	2	17.8 (5.2–30.4)
Length of exposureb
< 2 years	23	32.4 (25.7–39.2)	15	40.5 (30.0–50.9)	8	23.6 (15.2–31.9)
2 < 5 years	29	90.0 (73.3–106.7)	18	114.3 (87.4–141.3)	11	66.8 (46.6–86.9)
5 < 10 years	13	109.3 (79.0–139.6)	8	130.7 (84.5–177.0)	5	86.6 (47.9–125.3)
10–37 years	6	115.3 (68.3–162.4)	6	223.7 (132.4–315.0)	0	—
Cumulative exposurec
< 5 f/mL‐year	21	29.2 (22.8–35.6)	13	35.4 (25.6–45.2)	8	22.7 (14.7–30.8)
5 < 10 f/mL‐year	21	85.0 (66.4–103.5)	17	127.7 (96.7–158.6)	4	35.1 (17.5–52.6)
10–69 f/mL‐year	29	121.2 (98.7–143.8)	17	146.0 (110.6–181.5)	12	97.7 (69.5–126.0)
Time since first exposure
40–49 years	3	25.6 (10.8–40.4)	2	34.3 (10.0–58.5)	1	17.0 (0.0–34.0)
50–59 years	31	56.7 (46.6–66.9)	20	71.9 (55.9–88.0)	11	41.0 (28.6–53.4)
60–69 years	29	58.6 (47.7–69.5)	19	73.0 (56.2–89.7)	10	42.6 (29.1–56.1)
70–75 years	9	183.6 (122.4–244.7)	7	322.8 (200.8–444.8)	2	73.1 (21.4–124.9)

Full text

Introduction

1

Exposure to asbestos has been associated with several malignant and nonmalignant diseases including mesothelioma and various cancers (including cancers of the lung, larynx, and ovary) [1] as well as asbestosis, pleural effusion, diffuse pleural fibrosis [2], cardiovascular disease [3], and systemic autoimmune diseases (SAIDs) [4, 5]. Most asbestos‐related diseases (ARDs) occur in adults as a result of predominantly occupational exposure during adult life [6], and data on long‐term outcomes from childhood exposure remain sparse. The evidence to date suggests that childhood exposure to asbestos is linked most clearly to the development of malignant mesothelioma [7, 8, 9, 10, 11, 12], with some studies also finding associations with subsequent respiratory symptoms [13] and pulmonary abnormalities including pleural thickening [14].

Childhood is considered a “critical period of vulnerability” for exposure to many toxic substances [15]. It has been suggested that children may receive a higher dose relative to adults in the same environment due to their higher respiratory rate [16], and thus may be more susceptible than adults to the effects of environmental exposures [15]. The risk of ARDs as a result of exposure to asbestos in childhood, therefore, may be increased due to exposure early in life, the long latency period of many ARDs [17], and a potentially increased susceptibility of children to the effects of asbestos [18]. There are currently conflicting data regarding whether childhood exposure to asbestos confers a higher or lower risk for the development of mesothelioma [9], and very limited data on other long‐term health outcomes, including mortality, in those exposed to asbestos as children.

The township of Wittenoom, Western Australia, provides a unique opportunity to investigate the longer term effects of childhood exposure to asbestos. A crocidolite (blue asbestos) mine and mill operated about 12 km from Wittenoom from 1943 to 1966 [7]. Although the exact numbers are unknown, an estimated 20,000 people lived in the town during the mining operations, of which almost half were either born there or came as children less than 15 years of age [19]. Tailings from the mine containing residual asbestos fibers were distributed around the town and used for paving roads, playgrounds, and the racecourse [20], and all homes in the town were built using asbestos cement sheets [7], meaning that residents as well as workers were exposed to crocidolite. We have established a cohort of 6485 males and 410 females who were employed at the Wittenoom asbestos mine (“Wittenoom workers cohort”) as well as 2160 males and 2608 females who lived in the town but did not work for the asbestos company (“Wittenoom residents cohort”) [19]. Around half of this cohort first arrived in Wittenoom as children aged less than 15 years [7].

We have previously estimated all‐cause mortality and cancer incidence among these Wittenoom children to 2007 and 2009, respectively [7]. Among males, we found significantly elevated risks of malignant mesothelioma, leukemia, cancers of the brain, colorectum and prostate, and all cancers combined. Male children from Wittenoom also had a 50%–83% greater risk of all‐cause mortality compared with the Western Australian male population. Among females, we found excess risks of malignant mesothelioma, brain and ovarian cancers, and all cancers combined, as well as a 20%–47% greater risk of all‐cause mortality compared with the Western Australian female population. We now have access to an additional 10 years of mortality and cancer incidence data. These data are important as chronic diseases, such as cancer, become more prevalent with age and time since first exposure.

Methods

2

Cohort Description

2.1

The town of Wittenoom was established in 1946 to support a crocidolite mine owned by the Australian Blue Asbestos (ABA) company. It was originally established 1.6 km from the mine but moved 12 km away in 1947 [7]. Tailings from the mine were used in the town until the mid‐1960s, and the town was officially degazetted in 2007, although some residents remained in the town until their eviction in 2022, after which the town was completely demolished.

The Wittenoom residents' cohort was established between 1991 and 1993. This cohort consists of residents of Wittenoom who lived in the town up until 1993 but did not work for ABA. Details of how this cohort was established have been previously published [21]. In short, public records including state primary school records, hospital and general practitioner records, Western Australian electoral roll, birth and burial records, non‐mining employment records, and questionnaires sent to former ABA workers were used to identify potential residents. A total of 18,853 records were collected identifying a total of 5097 individuals who had lived in Wittenoom but were not employed by ABA. Further work to refine the cohort has resulted in a final total of 4768 individuals, including 2160 males and 2608 females. Since its establishment, this cohort has been periodically followed up through linkage to state and national registries as well as mailed questionnaires [19].

More than half of the residents' cohort (n = 2545; 1330 males and 1215 females) lived in Wittenoom as children (less than 15 years of age). Of these, 471 (18.5%; 245 males and 226 females) were born in Wittenoom, and a further 1096 (43.1%; 581 males and 515 females) were aged less than 5 years when they arrived in Wittenoom. Most children (n = 2352, 93.1%; 1230 males and 1122 females) left Wittenoom before age 16. These numbers are slightly different from those reported previously (n = 2483) [7] as further work has identified additional residents who were children during their time at Wittenoom.

As the legal age permitted for work in the ABA company was 15 years of age [22], we used age 15 as our cutoff for inclusion in the children's cohort so as to rule out any potential occupational exposure. We also excluded any children who reported that they later worked in the Wittenoom mine or mill (n = 81; 51 males and 30 females), resulting in a final cohort of 2464 individuals (1279 males and 1185 females).

Exposure Assessment

2.2

Comprehensive information on how estimates of asbestos exposure were derived has been published previously [7, 23, 24]. In brief, periodic dust surveys in and around the township took place between 1966 and 1992. Interpolation between these dust surveys was used to allocate exposures for each year (ranging from 0.5 fiber/milliliter of air [f/mL] in 1966 to 0.01 f/mL in 1992). For exposures from 1958 to 1966, exposure assessments were based on the results of the monitoring conducted in 1966 (0.5 f/mL‐year). For exposures from 1943 to 1957, exposure levels were assumed to be twice as high as those from 1958 onwards, as a new mill was commissioned in 1957 and the original mill was in use before this time. Therefore a level of 1.0 f/mL was assigned for these years. Cumulative exposure for each resident was estimated by summing over all their years of residence the product of fiber concentration for each year and the length of time they spent in Wittenoom during that year. These exposure levels were then adjusted by a factor of 4.2 ([24 × 7]/[8 × 5]) to allow for a continuous 24‐h exposure (rather than the 40‐h week method used to determine occupational exposure levels) [24, 25].

These exposure estimates have been shown to have adequate internal validity through agreement with lung fiber burdens [26]. In addition, Wittenoom exposures have been found to be comparable to exposure levels reported at other crocidolite mines internationally [27].

Ascertainment of Cases

2.3

The Wittenoom residents' cohort is regularly linked to state and national cancer and death registries, including the Western Australian Cancer Registry and associated Mesothelioma Registry, National Cancer Statistics Clearing House, Western Australian Registrar General's Mortality Database, and National Death Index. Cancer incidence is available from 1982 to the end of 2018 (i.e., follow‐up ended 31 December 2018). The International Classification of Diseases (ICD) for Oncology, Third Edition (ICD‐O‐3) was used to define cancers.

The date of death was available from 1950 to the end of 2019, with coded cause of death available to the end of 2017. Coded cause of death was not available for 43 (10.3%) deaths. The ICD Revisions 7–10 (as relevant to the time period) were used to define causes of death as listed on the death certificate. Follow‐up for the all‐cause mortality analysis ended December 31, 2019; for the analysis relating to specific causes of death, follow‐up ended December 31, 2017.

This study has approval from the Western Australian Department of Health Human Research Ethics Committee.

Statistical Analysis

2.4

Standardized incidence ratios (SIRs) were calculated as the ratio of observed to expected cancers, separately for males and females. Age‐ (in 5‐year groupings), calendar period‐ (in 5‐year increments), sex‐, and cancer‐specific incidence rates for the Western Australian population for the period 1982–2018 were obtained from the Western Australian Cancer Registry. These rates were then applied to the person‐years accrued by the cohort to estimate the expected number of cancers.

Standardized mortality ratios (SMRs) were calculated as the ratio of observed to expected deaths. Expected numbers of deaths were estimated by applying age‐ (in 5‐year groupings), calendar period‐ (in 5‐year periods), sex‐, and cause‐specific mortality rates for the Western Australian population for the period 1970–2017 to the person‐years accrued by the cohort. For all‐cause mortality, expected numbers of deaths were estimated for 2019. Death rates for the Western Australian population before 1970 were not available, and so rates for the period 1970–1974 were applied to population statistics to estimate expected deaths for the period 1950–1969.

For both SIRs and SMRs, 95% confidence intervals (CIs) were calculated by assuming a Poisson distribution for the observed numbers.

For each outcome, participants were censored on either the date of death (SMR) or of the cancer diagnosis (SIR). Approximately 20% of the cohort have been lost to follow‐up (date last known to be alive of 2004 or earlier). Therefore, two censoring dates were used to estimate expected deaths and cancers, representing a minimum and maximum estimate of effect. The first censoring date (SIR1/SMR1) assumed that all participants not known to be dead or to have developed cancer were alive and/or cancer‐free at the end of follow‐up. This is likely to overestimate the person‐years at risk and therefore underestimate the SIR/SMR. The second censoring date (SIR2/SMR2) censored those lost to follow‐up at their date last known to be alive. This underestimates the person‐years at risk, as many of those lost to follow‐up will still be alive and contributing to person‐years at risk, and thus overestimates the SIR/SMR.

Dates last known to be alive were collated from various records including questionnaires sent to cohort members, public records including the Western Australian Electoral Roll, hospital records, and participation in an annual asbestos review program. For a small number of participants (n = 136, 5.5%), records have not been updated since the date when participants left Wittenoom.

As smoking is known to be a risk factor for some of the cancer outcomes studied, adjustments were made where appropriate using the method outlined by Axelson and Steenland [28]. Smoking data for the cohort were available for 867 males (67.8%) and 863 females (72.8%). This information had been compiled from several questionnaires sent to former Wittenoom residents from 1979 to 2012 and information provided by those participating in a cancer prevention program from 1990 to 1995 [29]. Smoking data for the Australian population for the year 2010 were obtained from the Australian Institute of Health and Welfare (AIHW) [30]. The proportions of never, former, and current smokers by sex in the Australian population were adjusted to the age distribution of the cohort. Risk estimates for the association between smoking and the outcome of interest were obtained from the literature (detailed in relevant sections below). The estimated rate ratios of the cohort compared with the Australian population, due to smoking differences alone, were calculated for each relevant outcome following the method outlined by Axelson and Steenland [28]. The expected numbers of cancers or deaths, where appropriate, were multiplied by this smoking adjustment factor. Adjustments for smoking were only made where there was a significantly elevated risk in the Wittenoom cohort and there was substantial evidence for the relationship between smoking and the outcome of interest.

As there was a high incidence of mesothelioma observed in the Wittenoom cohort, incidence rates per 100,000 person‐years at risk were calculated for various asbestos exposure categories and separately by sex. These were calculated by dividing the number of cases of mesothelioma in each category by the number of person‐years at risk in the same category, multiplied by 100,000. Non‐cases were censored at their date last known to be alive.

All analyses were conducted in Stata 18.0 (College Station, Texas).

Results

3

A total of 2464 children (1279 males and 1185 females) were nonoccupationally exposed to asbestos at Wittenoom (Table 1). The median age at first exposure was 3, with an interquartile range (IQR) of 0.2–6.9 years. Length of exposure ranged from 0.1 years to 37.2 years (median 1.6 years; IQR 0.6–3.3), and time since first exposure ranged from 34 to 76 years (median 58 years; IQR 53–63). Estimated cumulative asbestos exposure ranged from 0.01 to 68.7 f/mL‐year, with a median of 3.4 f/mL‐year (IQR 1.4–7.6).

There were no significant differences between males and females in terms of age at first exposure (p = 0.647; median ages 2.9 and 3.1, respectively), length of exposure (p = 0.402; median 1.6 years for both), time since first exposure (p = 0.147; median 58 years for both), or cumulative asbestos exposure (p = 0.428; medians 3.5 and 3.3 f/mL‐year, respectively). Males were significantly more likely than females to have ever smoked (p = 0.009).

To the end of 2018, there were 384 cases of cancer in 340 individuals (55.9% male). From 1950 to the end of 2019, a total of 379 former Wittenoom children (66.0% male) had died of any cause.

Cancer Incidence

3.1

Males

3.1.1

A total of 223 cancers in 190 males were diagnosed from 1982 to the end of 2018. The most frequently diagnosed cancers were mesothelioma (n = 50), melanoma (n = 27), cancer of the prostate (n = 26), and colorectal cancer (n = 15). Regardless of censoring method, former Wittenoom male children had an increased risk of all cancers and mesothelioma compared with the Western Australian male population (Table 2). Higher risks of lip and mouth cancer and liver cancer (both censoring methods) and brain cancer and melanoma (SIR2 only) were also observed. There was a significantly decreased risk of prostate cancer for SIR1 only.

As both lip and mouth and liver cancer have been found to be associated with smoking, these SIRs were also adjusted for smoking (Table 2). Smoking rates for the former male Wittenoom children (using known data) were 28.9% current smokers, 36.9% ex‐smokers, and 34.3% never smokers. Australian data for males in 2010 (adjusted to the age distribution of the former Wittenoom children) were 21.0% current smokers, 33.7% ex‐smokers, and 45.2% never smokers. For lip and mouth cancer, a relative risk of 3.43 was used for current smokers and 1.40 for ex‐smokers, and for liver cancer, a relative risk of 1.56 for current smokers and 1.49 for ex‐smokers [31]. The SIRs for lip and mouth cancer were adjusted by a factor of 1.13 and for liver cancer by a factor of 1.05. Both SIRs for lip and mouth cancer remained significant, and SIR2 for liver cancer also remained significant.

Females

3.1.2

There were 161 cases of cancer in 150 females to the end of 2018. The most frequently diagnosed cancers were breast cancer (n = 49), mesothelioma (n = 26), melanoma (n = 14), and lung cancer (n = 10) (Table 3). Increased risks of mesothelioma and ovarian cancer were found regardless of censoring method. In addition, there were significantly elevated risks for all cancers, all cancers excluding mesothelioma and lung, and brain cancer (SIR2 only), and a significantly decreased risk of colorectal cancer (SIR1 only).

The rates for colorectal cancer and ovarian cancer were adjusted for smoking, as there is evidence of an association between these cancer types and smoking. Smoking rates for the former female Wittenoom children were 22.8% current smokers, 36.8% ex‐smokers, and 40.4% never‐smokers. When adjusted to the age distribution of the Wittenoom cohort, Australian data for females in 2010 were 17.8% current smokers, 28.1% ex‐smokers, and 54.1% never‐smokers. For ovarian cancer, a relative risk of 1.06 was used for both current and ex‐smokers [32]; for colorectal cancer, a relative risk of 1.08 was used for current smokers and 1.16 for ex‐smokers [31]. SIRs for ovarian cancer were adjusted by a factor of 1.01 and for colorectal cancer by a factor of 1.02. Both SIRs for ovarian cancer remained significant, and SIR1 for colorectal cancer also remained significant.

Mortality

3.2

Males

3.2.1

At the end of 2019, 250 males who arrived at Wittenoom as children had died of any cause. Boys from Wittenoom had a 48%–111% greater risk of all‐cause mortality than the Western Australian male population, depending on the censoring method used (Table 2). Former Wittenoom male children also had a greater risk of mortality from cancer (both censoring methods), and in particular mesothelioma, with SMRs ranging between 40.49 (95% CI 29.53–54.17) and 63.06 (95% CI 46.00–84.38). The risk of mortality from lung cancer was not significantly increased.

In addition, mortality from diseases of the circulatory system, as well as from accidents, injuries and poisonings and ill‐defined symptoms (ICD‐10 code R00‐R99; deaths where the underlying cause was not clearly defined), was significantly increased in the former Wittenoom male children when compared with the general Western Australian male population, for SMR2 only (Table 2). As diseases of the circulatory system have been found to be associated with smoking, this SMR was adjusted for smoking. Relative risks were obtained from an Australian cohort study investigating the association between smoking and total cardiovascular disease mortality [33]. A relative risk of 2.75 was used for current smokers and 1.20 for former smokers, and the SMR was adjusted by a factor of 1.10. Following adjustment, SMR2 remained significant.

Females

3.2.2

At the end of 2019, there were 129 deaths among females who had arrived at Wittenoom as children. Girls from Wittenoom had a 29%–86% greater risk of all‐cause mortality than the Western Australian female population, depending on which censoring method was used (Table 3). Similar to the males, former Wittenoom female children had a greater risk of mortality from cancer and mesothelioma in particular (both censoring methods), whereas the risk of mortality from lung cancer was not significantly increased. The risk of mortality from all cancers excluding mesothelioma and lung cancer was significantly increased for SMR2 only. Mortality from ill‐defined signs and symptoms was also significantly increased in the former Wittenoom female children when compared with the general Western Australian female population, regardless of censoring method.

Mesothelioma Rates

3.3

Mesothelioma incidence rates for various asbestos exposure categories are presented in Table 4. Those who had a longer duration of exposure and higher cumulative exposure had a higher rate of mesothelioma, providing evidence of an exposure–response relationship. Incidence rates also increased with time since first exposure. Age at first exposure did not display a linear trend, with those first exposed between the ages of 5 and 9 years having higher mesothelioma rates than other age groups. For all variables, the incidence rate was higher for former male than female children, although the CIs crossed in some instances.

Discussion

4

The current study has found that those who were exposed to asbestos as children in Wittenoom have an increased risk of mesothelioma and ovarian cancer (in women only), as well as some non‐asbestos‐related cancers, compared with the general Western Australian population. Among males, in addition to mesothelioma, elevated risks of all cancers combined as well as melanoma and cancers of the lip and mouth, liver, and brain were observed, whereas females had elevated risks of cancers of the ovary and brain. In addition, increased risks of all‐cause mortality and mortality from cancer and mesothelioma were found for both males and females. For males, mortality from diseases of the circulatory system, as well as from accidents, injuries, and poisonings, and from ill‐defined symptoms, was significantly elevated when compared with the general Western Australian population, whereas for females, mortality from ill‐defined symptoms was also significantly increased. These findings suggest that the risk of mortality among those exposed to asbestos in childhood is increased for both ARDs and non‐ARDs.

The increase in the risk of mesothelioma is expected. The current study found evidence of an exposure–response relationship, with those who had been exposed to asbestos for a longer time and at a greater cumulative dose having higher rates of mesothelioma. In addition, incidence rates increased with time since first exposure. This is consistent with previous studies of the Wittenoom children's cohort [7], as well as European studies of environmental or nonoccupational asbestos exposure and mesothelioma [34, 35]. However, age at first exposure was not linearly associated with mesothelioma rates, with the highest rates observed among those aged between 5 and 9 years at first exposure. Thus, it may be that there are critical windows of exposure during childhood, and that the risk conferred by asbestos exposure varies at different ages and developmental stages [15]. For example, lung development is known to continue through early childhood, and thus exposures during this time may contribute to greater disease risks in the future [36]. Accordingly, this study found that the rate of mesothelioma among those first exposed between the ages of 10 and 15 years was approximately half to two‐thirds that of those exposed at earlier ages.

Previous research with this cohort has consistently found a higher risk of mesothelioma in those first exposed as adults (aged ≥ 15 years) than those first exposed as children [20, 22, 25]; however, the wider literature has produced conflicting evidence. A study of occupationally‐exposed Chinese workers found higher mesothelioma mortality rates among those first exposed as adults [37], whereas studies in the United Kingdom [12] and Italy [10] have found higher rates in those first exposed at earlier rather than later ages. These inconsistencies may be related to latency, with those first exposed at older ages having effectively had less time to develop mesothelioma by follow‐up, and difficulties in determining the precise age at first exposure to asbestos [9]. In our study, the age of arrival at Wittenoom is well‐documented, and so the age at first exposure is clear. Regarding latency, the difference in rates by age at first exposure has decreased over time, while remaining significantly higher among those first exposed as adults, [22], and thus it is possible that with increasing follow‐up, this difference may diminish further.

The current study also found higher mesothelioma incidence rates among male as opposed to female former children. This pattern was also observed in earlier studies of this cohort [7], as well as studies of adult Wittenoom residents [20], but contrasts with a 2017 review, which found a generally higher disease burden in females than males [38]. This may be due to the exposure estimates in the current study being based on period and duration of residence but not on more nuanced details such as activity patterns; it may be that male children were more likely to participate in outdoor activities than female children, and to have experienced subsequent occupational asbestos exposure on leaving Wittenoom, leading to higher levels of asbestos exposure. However, this is speculation only.

Among both sexes, increased risks of all cancers, mesothelioma, and brain cancers were observed, consistent with past analyses in this cohort. Although an association between occupational asbestos exposure and brain cancer has been reported in a small number of studies [39, 40, 41], more recent studies have not found any evidence of an association between asbestos exposure and brain cancer [42]. In the current study, the relationship was based on a small number of cases (< 10), and although the SIRs were above 2 for all associations, a significant finding was only observed using SIR2, which we acknowledge may overestimate the incidence rate. However, given that this relationship was observed in both sexes and that there is still a poor understanding of environmental risk factors for brain cancer [43], some further investigation may be required.

For females, a significantly increased risk of ovarian cancer was also observed. Past analyses in this cohort showed a significant relationship for SIR2 only [7]; however, in the current study, the relationship was observed for both censoring methods and with narrower CIs. There has been considerable debate in the literature over the relationship between asbestos exposure and ovarian cancer [5]. Although the International Agency for Research on Cancer (IARC) concluded that there was sufficient evidence for the relationship in 2009 [44], a more recent review and meta‐analysis found that although there was an overall increased risk of ovarian cancer with asbestos exposure, this relationship was attenuated when restricted to studies of confirmed ovarian cancer incidence only [45]. Past studies have noted the potential for peritoneal mesothelioma to be misdiagnosed as ovarian cancer, particularly in older studies [5]. The majority of cases in the current study had been diagnosed since 2000 (and three since 2010), and thus we can be more confident that these are “true” ovarian cancers. Previous analyses investigating cancer incidence among women exposed to asbestos in Wittenoom in adulthood have found no increased risk of ovarian cancer among workers or residents [46], suggesting that early life exposure may be important.

Among males, some cancer associations were observed that have not previously been found in this cohort. Specifically, childhood exposure to asbestos was found to be associated with cancers of the lip and mouth and liver, as well as melanoma. The associations with lip and mouth cancer and liver cancer were observed for all censoring methods and remained after controlling for smoking. These associations have not been reported in earlier analyses of this cohort [7], nor in analyses of the wider Wittenoom cohorts, although associations between occupational asbestos exposure and cancers of the lip and mouth [47, 48] and liver, particularly intrahepatic cholangiocarcinoma (ICC) [49], have been observed in previous studies. An increased risk of mortality from liver cirrhosis has been observed in former Wittenoom workers [50], with liver cirrhosis in turn being associated with an increased risk of liver cancer [51]. Further, a preliminary study conducted in an area of high asbestos pollution in Italy found asbestos fibers in the livers of patients with cholangiocarcinoma [52]. The findings in the current study were based on a relatively small number of cases (n = 7), and further follow‐up of this cohort is needed to clarify whether this is a true association.

With regard to melanoma, previous analyses in this cohort showed some suggestion of an increased (nonsignificant) risk of melanoma in those exposed to asbestos as children [7]. A nonsignificant increased risk has also been observed in those with potential residential exposure to asbestos [53]. In the current study, this relationship was significant for SIR2 only. It is likely that these relationships are not related to asbestos exposure per se, but to lifestyle factors, such as alcohol consumption and sun exposure, in this cohort.

In contrast, previously observed associations were not found in the current analysis, specifically the associations with colorectal and prostate cancer and leukemia in males. In the current study, childhood exposure was not significantly associated with colorectal cancer incidence in males. Our previous analysis in this cohort based on 10 cases found a significantly increased risk for SIR2 only, which attenuated when controlling for smoking [7]. A recent meta‐analysis found a significant association between occupational asbestos exposure and colorectal cancer [54]; however, the relationship with childhood exposure is not yet fully understood. For prostate cancer, a reversal of previous trends was found, with a significantly decreased risk observed for SIR1, in contrast with the previously observed increased risk (SIR2 only) [7]. Although there was an increase in the number of cases observed, from 12 to 23, there has also been a concomitant increase in prostate cancer cases in Australia over time [55]. Previous studies have produced conflicting evidence about the relationship between asbestos exposure and prostate cancer [56, 57, 58]; the true nature of the relationship, particularly among those exposed to asbestos as children, remains unclear. With regard to leukemia, our previous analysis in this cohort found an increased risk of leukemia among males [7]; however, there was no significant relationship found in the current study, and no additional cases reported (results not reported). This is consistent with a recent Danish cohort study, which found that although long‐term occupational asbestos exposure was associated with an increased risk of leukemia, there was no association with early environmental exposure [59].

The current study's mortality findings were also broadly similar to past findings in this cohort [7]. In particular, increased risks of all‐cause mortality and mortality from all cancers, with and without mesothelioma, were observed for both sexes. In addition, mortality from ill‐defined signs and symptoms was found to be increased for both sexes in the current analysis. This is consistent with past research, including previous analyses of the Wittenoom children's cohort [7], studies of Wittenoom workers and residents [23, 46, 60], and studies of other asbestos‐exposed cohorts [58, 61]. In the current study, the majority of deaths due to ill‐defined symptoms (15 of 18) occurred in the period 1950–1976, with the remaining three occurring between 2000 and 2014. This finding is thus possibly an artifact of coding, with diseases such as mesothelioma being historically difficult to diagnose and poorly recorded, particularly among women [62]. Diagnostic accuracy has improved over time, and given that there is a statutory requirement to report all cancer diagnoses to the Western Australian Cancer Registry, we can be reasonably confident that we have accurately captured most if not all of the cancer diagnoses in the cohort.

This study has used more than 65 years of follow‐up to investigate the long‐term risks of childhood exposure to asbestos, extending upon past analyses in this cohort by up to 10 years. This long follow‐up is an important strength of the current study, alongside the use of time‐based quantitative asbestos exposure measurements. These exposure measurements were based on the period and duration of residence at Wittenoom, rather than personal measurements, and do not account for differences between cohort members including in terms of activity patterns. In addition, there is no information available on potential asbestos exposure after leaving Wittenoom. However, post‐Wittenoom exposure is likely to be of lower magnitude than exposure at Wittenoom for most, and to mixed fibers rather than the crocidolite present at Wittenoom. The Wittenoom exposures also represent the first exposure to asbestos encountered by this cohort, with sufficient latency to investigate health outcomes. As these measurements have been shown to have adequate agreement with lung fiber burdens [26] and are comparable with measurements reported internationally [27], we are confident in using them as measures of quantitative asbestos exposure while recognizing the uncertainties.

It is unclear how representative the former Wittenoom residents are of the general Western Australian population, although the information we have available suggests that they are more likely to live in regional or rural areas of the State [7]. As living in remote areas is known to be associated with poorer health outcomes [63], this may have influenced the findings seen here. In addition, a number of the former Wittenoom children were from the families of new immigrants to Australia. However, intergenerational social and occupational mobility is relatively high in Australia [64], and it is unclear whether migration history influenced the later exposures and health outcomes of these second‐generation Australians. The generalizability of this cohort to other asbestos‐exposed cohorts is also unknown. However, this cohort is unique in that it comprises a population with a relatively intense exposure (median 3.4 f/mL‐year) exclusively to crocidolite over a median period of approximately 18 months [50]. This enables the effects of duration and intensity of crocidolite exposure during childhood to be investigated. For some outcomes, there are still too few cases to adequately determine if there is a risk from early life exposure (e.g., only three cases of laryngeal cancer to date); this points to the need for further follow‐up of this cohort. Finally, there were a number of participants who were lost to follow‐up. We applied two different censoring methods to account for this, one representing a minimum estimate and the other a maximum estimate of effect; it is likely that the true effect lies somewhere between the two presented. It is possible that those lost to follow‐up differ from those retained in the cohort, including in terms of health outcomes or exposures, and so the estimates presented here should be interpreted with some caution.

This study provides an update of the cancer incidence and mortality rates among adults exposed to asbestos at Wittenoom as children, providing an additional 10 years of follow‐up data. An increased risk of cancers, including mesothelioma, as well as all‐cause mortality, was observed. The risk of mesothelioma from childhood exposure remained lower than the risk from nonoccupational exposure in adult life. Although it is possible that some findings may be accounted for by participants' lifestyle and behavioral characteristics, it is also likely that at least some outcomes, particularly the observed increased risk of brain cancer, are related to their childhood asbestos exposure.

These findings have important implications for public health research and policy. The observed associations between exposure in early life and later cancer and mortality outcomes underscore the importance of long‐term surveillance and health monitoring among those substantially exposed to environmental hazards such as asbestos in childhood. These findings also highlight the need for further research into both established and emerging health outcomes related to childhood exposures.

Author Contributions

Peter Franklin, Nicholas de Klerk, Fraser Brims, and Alison Reid were involved in the conception and design of this work. Peter Cinquini assisted in the acquisition of data; analysis and interpretation of data were performed by Renee N. Carey and Alison Reid. Renee N. Carey drafted the work, and all authors revised it critically. All authors have provided final approval of the version to be published and agree to be accountable for all aspects of the work.

Disclosure

The authors have nothing to report.

Ethics Statement

The work in this article was performed at the University of Western Australia and Curtin University. The Western Australian Department of Health Human Research Ethics Committee provided ethical approval. This project used preexisting or routinely collected data, and so informed consent was not required or provided.

Conflicts of Interest

The authors declare no conflicts of interest.