Maternal Age, Exposure to Siblings, and Risk of Amyotrophic Lateral Sclerosis

Abstract

Between 1987 and 2005, the authors conducted a nested case-control study based on the Swedish Multi-Generation Register to investigate whether early life exposures, namely, maternal age at delivery and exposure to siblings, are associated with an increased risk of amyotrophic lateral sclerosis (ALS). The study comprised 768 ALS cases and five controls per case matched by birth year and gender. Odds ratios and their corresponding 95% confidence intervals for ALS were estimated by conditional logistic regression modeling. Low maternal age (≤20 years) and high maternal age (≥41 years) were both associated with higher risk of ALS (odds ratio (OR) = 1.5, 95% confidence interval (CI): 1.1, 2.0 and OR = 1.7, 95% CI: 1.1, 2.4, respectively). The relative risk of ALS increased slightly with increasing number of younger siblings (OR = 1.1, 95% CI: 1.0, 1.1; p = 0.02). Children whose first younger sibling was born after the age of 6 years had the greatest relative risk (OR = 1.8, 95% CI: 1.2, 2.7). Exposure to older siblings was not associated with the risk of ALS. Although the strength of the observed associations was modest, these results provided further support for the theory that early life exposures might contribute to the disease pathogenesis.

Amyotrophic lateral sclerosis (ALS) is the most common neurodegenerative disorder affecting motor neuron functions (1). Given its largely unknown etiology, various environmental risk factors have been proposed to be associated with ALS, but it is not clear whether or not these associations are causal (2). Events operating at a critical or sensitive period (e.g., prenatal and postnatal) could result in a long-term change in the structure or function of the organism (3). Within-household infections in early childhood could also play a role in adult-onset disorders (4). These hypotheses have seldom been investigated for ALS other than by two studies from the 1980s that failed to show any association between parental age and sibship size (as two proxies of early life exposures) and the risk of motor neuron diseases (5, 6).

The low disease incidence and the long incubation between the exposure and outcome make the assessment of these associations a challenge. The availability of several national registers in Sweden provided us a unique opportunity to examine these associations with a large number of cases and, thus, good power to disclose real associations. We compared the maternal age at delivery and number of siblings between ALS patients and controls to evaluate whether these factors are independent risk indicators in the development of ALS.

MATERIALS AND METHODS

Health-care system

The public medical service in Sweden provides equal access to public health care to all citizens, and there is no private in-patient treatment in Sweden. Therefore, hospital-provided medical services are population based and referable to theresidence county of the patients. Since 1964-1965, the Swedish National Board of Health and Welfare has compiled data on individual hospital discharges coded according to the International Classification of Diseases (ICD): Seventh Revision (ICD-7) before 1969, Eighth Revision (ICD-8) from 1969 through 1986, Ninth Revision (ICD-9) from 1987 through 1996, and Tenth Revision (ICD-10) since 1997 (7). Coverage of the In-patient Register has been nationwide since 1987. Follow-up of our study cohort started on January 1, 1987, accordingly. The completeness and accuracy of the Inpatient Register are generally high. The overall underreporting rate of hip fracture was less than 2 percent (8). The false negative rates of trauma, ischemic heart disease, and malignant tumors were estimated to be 5 percent, 7 percent, and 3 percent, respectively, and the false positive rates were estimated to be 1 percent, 2 percent, and 2 percent, respectively (9).

Study design

We conducted a nested case-control study within the Swedish Multi-Generation Register (10). The register includes index persons who were born in 1932 or later together with their biologic or adoptive parents. In total, it contains more than 13 million unique individuals. Complete familial information is available for nearly all individuals who were either alive in 1991 or deceased before 1968, but for only 60 percent of those who died between 1968 and 1990 (10). Individuals deceased before 1991 were deleted from the register by parish civil registration offices, which were responsible for local population registration in Sweden at that time. Among these individuals, 100 percent of those who died before 1968 and about 60 percent of those who died between 1968 and 1990 could be reidentified from other registers and were included in the Multi-Generation Register again. Between 1968 and 1990, there were in total 2,049,487 deaths. Thus, about 800,000 subjects might have a missing familial link, which comprises about 6 percent of the entire population covered by the register. Since 1991, tax offices have been responsible for the local population registration in Sweden and have supplied complete data to the Multi-Generation Register (10).

In order to avoid potential problems due to missing data at the commencement of the register and the complexity of multiple births, as well as to accrue individuals old enough to be at risk of ALS, we included only individuals born as singletons between 1936 and 1975. In total, we accrued 3,989,559 individuals who satisfied all the above criteria, that is, born in Sweden, alive and free of ALS in 1987, and also having retrievable maternal information in the Multi-Generation Register. National registration numbers (a unique identifier assigned to all Swedish residents) were used to follow the cohort of 3,989,559 persons through cross-linkages to the In-patient Register, the Causes of Death Register, and the Emigration Register. Follow-up started on January 1, 1987, and was censored at the date of first diagnosis of ALS, death, emigration out of Sweden, or December 31, 2005, whichever occurred first.

Between 1987 and 2005, a total of 779 ALS patients were identified from the In-patient Register out of the cohort (ICD-9 code 335C, ICD-10 code G12.2). Cases younger than 30 years (11 of 779) were excluded because of their potential genetic cause, leaving 768 ALS cases for final analysis. Using the method of incidence density sampling (11), we randomly selected 3,840 controls (five controls per case) that were individually matched to ALS patients on gender and year of birth. These were persons who had not yet died, emigrated out of Sweden, or been diagnosed with ALS at the time of index case diagnosis. Information on the mothers and siblings of the index persons was retrieved from the Multi-Generation Register. Siblings were defined as those having the same biologic mother as the index persons. Educational level at the time of index case diagnosis was identified from the Education Register kept by Statistics Sweden. The Education Register provides information on the highest attained ducational level for all Swedish residents who were alive in 1985 or later. The study was approved by the regional ethics committee at Karolinska Institutet.

Statistical analysis

Odds ratios and their 95 percent confidence intervals were estimated for maternal age (≤20 years, i.e., low maternal age; 21-25 years; 26-30 years; 31-35 years; 36-40 years; and ≥41 years, i.e., high maternal age; referent group: age 26-30 years), number of older siblings (none, one, two, and three or more; referent group: none), number of younger siblings (none, one, two, and three or more; referent group: none), and birth interval (no sibling, more than 6 years, 2-6 years, and less than 2 years; referent group: no sibling) by conditional logistic regression models. Multivariable models includedfurther adjustment for educational level in three groups: 9 years or less as elementary school, 10-12 years as secondary school, and 13 years or more as college, with elementary school as a referent group. The potential modifying effect of birth interval (before school age, i.e., ≤6 years and after school age, i.e., >6 years) between index cases and their first younger siblings on the association between number of younger siblings and ALS risk was further assessed. The p values reported are all two sided. Analyses were performed by SAS, version 9.1, software (SAS Institute, Inc., Cary, North Carolina).

To allay the concern of case underascertainment due to using only hospitalized cases, we performed a sensitivity analysis by including “death certificate only” cases, that is, cases recorded with ALS as the underlying cause of death in the Causes of Death Register but not recorded in the Inpatient Register, given that mortality data are another potential source of case ascertainment (12). Five dead controls per “death certificate only” case, with other causes of death but ALS and death in the same year as the ALS cases, were randomly selected from the general deceased population, with matching by year of birth and gender as in the main analysis. Because data on causes of death during year 2005 were not available, this sensitivity analysis was restricted to cases identified between 1987 and 2004.

RESULTS

The distributions of gender, age at diagnosis, educational level, and sibship size at the time of index case diagnosis for cases and controls are presented in table 1. Sixty-two percent of the ALS cases were men, giving a male:female ratio of 1.6. The mean and median age of ALS patients were 55 years and 56 years (range: 32-69 years). There was nosubstantial difference between cases and controls with regard to the educational level.

TABLE 1.

Characteristics of cases and controls in a nested case-control study on amyotrophic lateral sclerosis, Sweden, 1987-2005

	Cases (n = 768)		Controls (n = 3,840)		Total (n = 4,608)
	No.	%	No.	%	No.	%
Gender
Men	478	62.2	2,390	62.2	2,868	62.2
Women	290	37.8	1,450	37.8	1,740	37.8
Age at diagnosis/date of referral (years)
≤50	221	28.8	1,105	28.8	1,326	28.8
51-60	337	43.9	1,685	43.9	2,022	43.9
≥61	210	27.3	1,050	27.3	1,260	27.3
Educational level
Elementary school	231	30.0	1,164	30.3	1,395	30.3
Secondary school	333	43.4	1,686	43.9	2,019	43.8
College	202	26.3	984	25.6	1,186	25.7
No information	2	0.3	6	0.2	8	0.2
Sibship size
1	121	15.8	690	18.0	811	17.6
2	234	30.5	1,211	31.5	1,445	31.4
3	178	23.2	886	23.1	1,064	23.1
≥4	235	30.5	1,053	27.4	1,288	27.9

After conditioning on gender and year of birth, the risk of ALS increased with both low and high maternal age compared with maternal age between 26 and 30 years (table 2). Adjustment for the number of older and younger siblings slightly increased the association between high maternal age and risk of ALS (table 2). There was no obvious association between the number of older siblings and the risk of ALS (odds ratio (OR) = 1.0, 95 percent confidence interval (CI): 0.9, 1.1; p = 0.73). The risk was weakly associated with the total number of younger siblings (OR = 1.1, 95 percent CI: 1.0, 1.1; p = 0.02). Because there was no association observed between the risk of ALS and the number of older siblings, we investigated only the interval between births of the index persons and their first younger siblings. The highest relative risk of ALS was observed for individuals whose first younger sibling was born after the age of 6 years (table 3). Further adjustment for the highest attained educational level did not change the estimates materially (data not shown).

TABLE 2.

Odds ratios and their 95% confidence intervals of amyotrophic lateral sclerosis according to familial characteristics, Sweden, 1987-2005

	No. of cases	No. of controls	Odds ratio*	95% confidence interval	Odds ratio†	95% confidence interval
Maternal age (years)
≤20	74	273	1.6	1.2, 2.1	1.5	1.1, 2.0
21-25	212	966	1.3	1.0, 1.6	1.2	1.0, 1.5
26-30‡	195	1,130	Referent		Referent
31-35	164	818	1.2	0.9, 1.5	1.2	1.0, 1.5
36-40	84	502	1.0	0.7, 1.3	1.0	0.8, 1.4
≥41	39	151	1.5	1.0, 2.2	1.7	1.1, 2.4
No. of elder siblings
0‡	397	1,974	Referent		Referent
1	236	1,174	1.0	0.8, 1.2	1.1	0.9, 1.3
2	81	395	1.0	0.8, 1.3	1.1	0.8, 1.4
≥3	54	297	0.9	0.7, 1.2	0.9	0.7, 1.3
No. of younger siblings
0‡	287	1,571	Referent		Referent
1	223	1,157	1.1	0.9,1.3	1.1	0.9, 1.3
2	142	649	1.2	1.0, 1.5	1.2	0.9, 1.5
≥3	116	463	1.4	1.1, 1.8	1.3	1.0, 1.8
Birth interval between index persons and their first younger siblings (years)
No younger sibling‡	287	1,571	Referent		Referent
>6	106	445	1.3	1.0, 1.7	1.3	1.0, 1.7
2-6	268	1,325	1.1	0.9, 1.3	1.1	0.9, 1.3
<2	107	499	1.2	0.9, 1.5	1.2	0.9, 1.5

^*Conditioning on matching variables (i.e., year of birth and gender).

^†Conditioning on matching variables (i.e., year of birth and gender) for all estimations. In addition, birth order and number of younger siblings were adjusted for in estimating the role of maternal age; maternal age and number of younger siblings were adjusted for in estimating the role of birth order; maternal age and birth order were adjusted for in estimating the role of number of younger siblings on the risk of amyotrophic lateral sclerosis. Maternal age was adjusted for in estimating the role of birth interval between index persons and their first younger siblings.

^‡Referent group.

TABLE 3.

Odds ratios and their 95% confidence intervals of amyotrophic lateral sclerosis by combination of number of younger siblings and birth interval between index persons and their first younger siblings, Sweden, 1987-2005

	No. of cases	No. of controls	Odds ratio*	95% confidence interval	Odds ratio†	95% confidence interval
No younger sibling‡	287	1,571	Referent		Referent
One younger sibling
First sibling aged <6 years	152	816	1.0	0.8, 1.3	1.0	0.8, 1.3
First sibling aged ≥6 years	71	341	1.1	0.9, 1.5	1.2	0.9, 1.6
More than one younger sibling
First sibling aged <6 years	223	1,008	1.2	1.0, 1.5	1.2	1.0, 1.5
First sibling aged ≥6 years	35	104	1.8	1.2, 2.8	1.8	1.2, 2.7

^*Conditioning on matching variables (i.e., year of birth and gender).

^†Conditioning on matching variables (i.e., year of birth and gender) and further adjusted for maternal age.

^‡Referent group.

Individuals identified as the first child of his or her mother could probably have older siblings not included in the register if the older siblings were born before 1932. To address this concern, we reanalyzed the data on number of older siblings by restricting the analysis to individuals born after 1942 (10 years after initiation of the Multi-Generation Register). Exposure to older siblings was still not associated with ALS in the restricted analysis (OR = 0.9, 95 percent CI: 0.8, 1.0). In addition to the 668 cases identified from the In-patient Register between years 1987 and 2004, there were additionally 109 “death certificate only” cases identified through the Causes of Death Register (14.0 percent of the total 777 cases identified from both registers) during this period. Sensitivity analysis by including all 777 cases found that the relative risks of ALS associated with low maternal age (OR = 1.3, 95 percent CI: 1.0, 1.8), high maternal age (OR = 1.5, 95 percent CI: 1.0, 2.2), number of younger siblings (OR = 1.1, 95 percent CI: 1.0, 1.2), and number of older siblings (OR = 1.0, 95 percent CI: 0.9, 1.0) were all similar to those found in the main analysis, if diminished to some extent.

DISCUSSION

The main findings of our study were that both low and high maternal ages at delivery, as well as exposure to younger siblings but not older siblings, were associated with an increased risk of ALS. Children whose first younger sibling was born after they started school experienced the highest relative risk of ALS.

In contrast to our findings, two case-control studies from the 1980s did not find an association between maternal age and risk of motor neuron disease (5, 6). The lack of association is probably due to the case-control design that is prone to various biases and confounding, small number of cases accrued, and use of only mortality data (5, 6). Given the modest strength of the association, a lack of power or precision of the study would most likely end up with a null finding. On the other hand, a meta-analysis based on 11 individual studies showed that both low and high maternal ages were related to an increased risk of Alzheimer’s disease and that the effect size was almost identical to our findings (13). Fetal growth is profoundly affected by even transient changes in the maternal environment. Fetal origin hypothesis proposed that an abnormal intrauterine environment, such as insufficient nutrition supply, associated with either too early or too late pregnancy might alter organ development and functioning in the fetus and thereby increase disease susceptibility later in life (3). Future studies should explore further the potential associations between other markers of fetal development, such as brood size, birth weight, and handedness, and the risk of ALS. Immaturity of the uterine or cervical blood supply may also predispose mothers to subclinical infections (14), and these infections might be acquired by the fetus or newborn babies. In addition to biologic risks associated with young or old maternal age, sociodemographic factors, such as inadequate prenatalcare, being unmarried, emotional stress during pregnancy, alcohol consumption, smoking, and so on, could also play potential roles in the growth of the child before and after birth (15). The effect of these factors on the development of chronic neurodegenerative disorders such as ALS is not clear, but our results suggest that further investigation is warranted.

In our study, having younger but not older siblings increased the risk of ALS. Our null finding about exposure to older siblings or birth order is in accordance with the results reported by a previous study in the 1980s from Japan (5). However, there is no study about the association between exposure to younger siblings and the risk of ALS to date to our knowledge. An Australian study found that the relative risk of multiple sclerosis decreased with increasing numbers of younger siblings and that the strength of the inverse association decreased with increasing birth interval, with the strongest association observed within a birth interval of less than 2 years (16). In contrast, we observed that the association between exposure to the first younger sibling and the risk of ALS increased with increasing birth interval, with the greatest association appearing after a birth interval longer than 6 years. A potential explanation is that repeated exposure to active infections carried by infant siblings in early life would affect the risk of neurologic disorders later in life. The explanation can be further supported by the “delayed-infection” hypothesis related to the pattern and timing of common infections in early life (17), which proposes that lack of early infections could lead to abnormal immunologic response to common infections in later life and thus modulate the risk of several diseases (17). A long birth interval may also indicate problems related to the fetal or postnatal development of the index cases, which could be related to their predisposition to neurodegeneration in adult life and also cause delay of a second pregnancy of their mothers (18).

In this nested case-control study, we included all incident hospitalized ALS cases diagnosed between 1987 and 2005 identified from a population that was defined independently of our study purpose. Underascertainment of cases by using only hospitalized cases could be a concern. However, the hospitalization rate of ALS is fairly high (86 percent, i.e., 668 hospitalized cases out of 777 total cases identified from both registers, assuming that all the “death certificate only” cases were correctly diagnosed). We believe that in-hospitalization data in Sweden have a better quality than mortality data given that most diagnoses at discharge are performed by neurologists. The large number of cases accrued in the study made it unfeasible for us to review the medical records and verify the correctness of diagnosis. However, the accuracy of the In-patient Register is generally high as reported (9).

Another minor concern is the missing familial information for the 40 percent of individuals deceased between 1968 and 1990 in the Multi-Generation Register (10). To allay this concern, we linked all siblings of cases and controls to the Causes of Death Register. Between 1968 and 1990, 19 percent of the sibling deaths occurred among siblings of cases, and before 1968 or since 1991, 20 percent. Thus, this missing information did not seem to be different for cases and controls. Another limitation of the present study is the lack of adjustment of other potential confounders (e.g., smoking and environmental toxins) but educational level. Although associations between these factors and ALS risk were suggested, none was confirmed. In addition, it is unlikely that they are associated with the exposure of interest in our study. Otherwise, the nested case-control design within the strictly defined cohort of our study preserves the validity of a cohort study, thereby eliminating bias due to selection forces and differential misclassification of the exposure among cases and controls. The independent purpose for data collection, prospective exposure measurement, and long-term follow-up make other selection problems and reverse causality unlikely.

In conclusion, both low and high maternal ages, as well as exposure to younger siblings, are associated with increased risk of ALS. The association between risk of ALS and number of younger siblings was stronger among individuals whose first younger sibling was born when the index persons were at school age. Future studies are needed to explore the underlying mechanism of these associations.

Abbreviations

ALS

amyotrophic lateral sclerosis

CI

confidence interval

ICD

International Classification of Diseases

OR

odds ratio