Several subgroups of juvenile IIM patients demonstrated seasonal birth patterns that differed significantly from other juvenile IIM patients or from controls, while only one subgroup of adult IIM patients demonstrated some evidence of a pattern of birth that differed from that of other adult IIM patients. The lack of a distinct overall seasonal pattern in births of IIM patients may be due to etiologic heterogeneity of the IIM. These findings suggest that some forms of the IIM, particularly in subgroups of patients with childhood–onset disease, are influenced by seasonal environmental exposures during early development, while other forms are not. In the IIM, seasonal influences during gestation or shortly after birth might be more likely to have an impact earlier in life, leading to juvenile-onset disease. However, it is important to note that the sample size was too small to examine the birth distribution for Hispanic adult IIM patients and for adult IIM patients with the anti-p155 autoantibody. Therefore, it is not possible to tell whether the seasonal patterns of birth displayed for these subgroups were unique to juvenile IIM patients, because the birth distributions of adult IIM patients in these subgroups could not be determined.
Studies of the influences of exposures early in life on later disorders such as cardiovascular disease have focused primarily on birth weight and maternal nutrition (25
). The seasonal birth distributions observed in certain subgroups of IIM patients might be explained by several potential genetic or environmental factors, including factors that are not related to maternal nutrition. Phillips has recently highlighted the evidence that prenatal toxic exposures, nutritional influences, antigen exposure, and infection may influence the development of autoimmune disease (2
Parental occupational exposures such as silica (26
) or seasonally applied agricultural pesticides (27
) might play a role. Cord serum levels of pesticides have been associated with childhood asthma (28
). For children fathered by pesticide workers in the Red River Valley of Minnesota, Garry et al. (29
) found that the percent of birth defects was highest for those conceived in the spring, the season when herbicide application occurs in this region. A recent study of a primarily Hispanic population in Salinas Valley, an agricultural region of California, found that environmental exposures during the first year of life, including living with an agricultural worker, having a gas stove in the home, owning pets, and being breast-fed, were associated with the Th1/Th2 immune response at age two (30
Other potential exposures include infectious agents (8
). Outbreaks of illness occur in distinct seasonal patterns for many pathogens (31
). Early-life infection with seasonal pathogens, such as viruses, has been proposed as a mechanism leading to other autoimmune diseases, particularly Type 1 diabetes (8
). It is also possible that the same factor can influence both susceptibility to infection (31
) and susceptibility to autoimmune disease. Thus, meteorological factors that vary by season, such as ultraviolet light, might influence the immune system and the development of autoimmune disease (31
Although there is some evidence for immune system modulation during early development, birth patterns may actually reflect the influence of seasonal factors acting at various time points: pre-conception, during gestation, shortly after birth, or during specific developmental windows later in childhood. Birth season is related not only to the timing of environmental exposures, but also to the timing of social exposures during childhood. For example, birth date is a determining factor in whether a child is amongst the oldest, youngest, or in the middle age-range for his or her grade level. Relative age within a school grade has been found to be associated with academic performance (34
), with children of younger relative age being at a disadvantage. It is possible that differing levels of stress based on relative age within a school grade might influence immune function in children, thus potentially influencing childhood-onset autoimmune disease. While this is speculative, it demonstrates that birth patterns may be a surrogate for exposures at a variety of time points in life.
The seasonal pattern of birth observed for Hispanic juvenile IIM patients, with a peak in October, strongly differed from that of other juvenile IIM patients and from controls. Similarly, Laron et al. (12
) found that birth patterns in patients with Type 1 diabetes can vary by ethnic group so that examining a heterogeneous population as a whole could mask differences between ethnic groups. It is also possible that Hispanics have genetic risk factors that predispose them to susceptibility to certain seasonal environmental risk factors (35
The finding of a difference in birth pattern between juvenile-onset DM patients with and without the p155 autoantibody supports the hypothesis that a seasonal environmental exposure during gestation or shortly after birth might trigger an immune response that produces this autoantibody (17
). It is interesting to note that adult IIM patients with anti-synthetase autoantibodies did not display a seasonal pattern of birth, but a previous study found that adults with anti-synthetase autoantibodies have seasonal patterns in the onset
of their IIM (36
). Thus, agents that influence the p155 autoantibody response might have a greater effect during early development, while factors that influence the production of anti-synthetase autoantibodies may have a greater influence later in life.
Differences in birth patterns were found between juvenile IIM patients with and without the HLA allele DRB1*0301 (16
), with similar results for the linked risk factor allele, HLA DQA1*0501. The birth distributions of juvenile and adult IIM patients positive and negative for DQA1*0501, respectively, had similar mean birth dates, indicating that this allele might interact with environmental factors to influence both juvenile-onset and adult-onset IIM. The DQA1*0501 allele binds class II associated invariant chain peptide (CLIP) differently than other HLA alleles, suggesting that this allele has access to peptides earlier in the antigen processing pathway, thus potentially encountering novel peptides that induce autoimmunity (37
). Similar to the present findings, HLA alleles have also been associated with birth season of schizophrenia patients (6
) and with season of onset of Type 1 diabetes (38
). Thus, genetic risk factors potentially interact with environmental influences during early development to result in disease onset later in life.
Despite attempts to decrease possible confounders in this study, several limitations remain. One potential limitation is that the IIM patients or controls might not be representative of the general population or that the controls might come from a population with different characteristics than the population of IIM patients. A control group from NIH was used due to the availability of birth dates for this sample, in contrast to broader population-based data.
There is some evidence that seasonal patterns of birth in the general population might be associated with socio-demographic factors such as maternal age, marital status, maternal education, and birth order (40
). Information on these variables was unavailable for both the patients and controls so the possibility of confounding by any of these factors cannot be excluded.
A limitation of the study is that the geographical distribution of patients and controls within the U.S. was not the same. The majority of controls were from Maryland, Virginia, or Washington, D.C., while juvenile IIM patients were from throughout the contiguous U.S., and information was not available on the specific geographic location of adult IIM patients. While there is some potential for these geographic differences to bias the comparisons of patients to controls, significant differences were not found in our data when birth distributions of those from Maryland, Virginia, and Washington, D.C. were compared to those from the rest of the contiguous U.S.
It is possible that any seasonal variation in birth patterns might have been diluted by pooling data from across the contiguous U.S. There is some evidence that seasonal patterns of birth in the general population differ between northern and southern states and that this difference might be due to decreased conception during times of extreme heat (41
). Although meteorological factors can vary across different regions of the U.S., our analyses found no significant differences in birth distributions between northern and southern states. Therefore, while the pooling of our data across the U.S. is a potential limitation, the effect on the results is likely to be minimal.
It is also possible that any seasonal variation in birth patterns might have been diluted by pooling data over a somewhat wide time period because the seasonal distribution of births can vary over time (42
). However, many seasonal occurrences, such as infectious disease outbreaks, occur at relatively constant time points from year to year (31
), and consequently the effect of pooling data across years should be minimal. No significant seasonal birth patterns were found when patients' birth dates were analyzed by decade. Some decade-matched analyses were performed in an attempt to correct for differences in birth years between patients and controls, and these results were reassuringly similar to those of the entire study group (data not shown).
Another limitation of this study is that several subgroup analyses were conducted, and some subgroups had small sample sizes. Independent studies of birth season in Hispanic, p155, and HLA subgroups of IIM are needed to verify the seasonal birth pattern findings in these patient groups. This is an exploratory analysis with small sample sizes and therefore all results with uncorrected p-values of less than 0.05 are considered to be of interest as they highlight potential associations that can be further assessed in future studies.
This work provides evidence that seasonal birth distributions of Hispanic juvenile IIM patients, juvenile IIM patients with the p155 autoantibody, and juvenile IIM patients with certain HLA alleles differ from the birth distributions of patients in other subgroups or from the birth distribution of a population of individuals not known to have an autoimmune disease. These subgroups of myositis patients may have been affected by specific seasonally varying environmental agents around the time of conception, during a certain period of gestation, or during early childhood. Our data suggest that early environmental influences have a greater influence on childhood-onset myositis than on adult-onset myositis. The finding of seasonal patterns of birth in certain subgroups of IIM patients, but not in the overall patient population, is consistent with the notion that different IIM subgroups are distinct entities that have different underlying causes. Further investigations, including studies of interactions among factors, are needed to elucidate the role of the environment in the development of the IIM.