Celiac disease (CD) is an autoimmune mediated enteropathy with the presence of autoantibodies in the sera (1) and intestinal mucosa, and also inflammation and villous atrophy of the mucosa (2). The cause of CD is still uncertain, but it is aggravated by the ingestion of the gliadin fraction of gluten in genetically susceptible individuals (3). The incidence of CD, which is frequently associated with type 1 childhood diabetes mellitus (T1DM) (4–8) and other autoimmune diseases, has increased in the past several decades (9). These changes have been attributed to rapid changes in lifestyle in the modern world and to the greater spread of and exposure to perinatal infections (10–13). There is recent evidence of an association between seasonal viral infections, such as rotavirus (14–16) and adenovirus (17), and the onset of CD. In the case of T1DM, evidence of perinatal enterovirus infection has been reported (18–21).
Studies carried out by our group have found that children and even young adults with different autoimmune diseases, such as T1DM (22–24), Graves disease, Hashimoto thyroiditis (25), multiple sclerosis (26), and atopic dermatitis (manuscript in preparation), have a month of birth (MOB) whose seasonality is different from that of the general population in the respective area or country. These findings are suggestive of a seasonal environmental factor in the etiology of autoimmune diseases, endemic virus infections acting as the trigger in the perinatal period (27). On the basis of our findings, we investigated the seasonality of MOB in a cohort of children with CD diagnosed in Israel compared with that in the general population.
PATIENTS AND METHODS
Four hundred thirty-one Jewish children with CD (239 girls and 192 boys) and their first-degree relatives were identified from medical records of the Departments of Gastroenterology of the Schneider Children's Medical Center, Petah Tikva, and Safra Children's Hospital, Ramat Gan, both affiliated with Tel Aviv University. These cohorts represent all of the patients with a biopsy-proven diagnosis of CD between 1991 and 2007. Eighty-one boys and 138 girls were younger than 24 months, and 111 boys and 101 girls were older than 24 months at the time of diagnosis. Of these children, 13% had associated autoimmune diseases, predominantly T1DM. Forty-six girls and 29 boys (17.4%) had family members with CD. The diagnosis of CD was ascertained by serological methods and histology.
The seasonality and rhythm of MOB of the children with CD (n = 431) were analyzed and compared with those in the general Jewish population of Israel (n = 1,040,558) born in the same time period, obtained from data published by the Population Bureau of Statistics.
Rhythmicity was analyzed by use of the cosinor method (28). Cosine approximation Yi = M + A × COS (ωti + φ) yielded the following parameters: M = the time series mean (midline estimating statistic of rhythm), A = AMPLITUDE (half of the peak-to-trough variation), and Acrophase (φ) − the peak of the calculated rhythm (ωt) − the period of the rhythm. (2D Table curve, Jandel Scientific, San Rafael, CA). The statistical significance of the rhythms is expressed by R and P values. The data were compared with the pattern of total live births. Differences between groups were analyzed by use of the χ2 test. Statistical significance is expressed by P ≤ 0.05. The study was approved by the ethical committees of the hospitals.
Our investigation revealed that the children with CD had a different seasonality of MOB from that found in the general population of Israel (Figs 1–3). Figure 1 shows that both the girls with CD (A) (n = 239) and the boys with CD (B) (n = 192) had a different yearly MOB seasonality from that of the general population (C and D). Whereas in the general population both sexes had the same yearly MOB rhythm, the patients with CD differed between the sexes (P < 0.02), the girls having more than 1 peak. In girls with CD, the MOB pattern was rhythmic, with periods of 4 + 8 months (mesor = 20.2, amp4 = 3.8, amp8 = 2.3, R = 0.74, P < 0.01). In the general population, the rhythm of MOB of girls (n = 505,908) was 12 months (mesor = 42,160, amp = 3050, R = 0.85, P < 0.01), with a peak in September. No significant differences were found in MOB patterns between boys with CD (n = 192) and those in the general population (n = 534,650; 8 months rhythm, mesor = 17.46, amp = 6.9, R = 0.953, P < 0.01 for patients with CD, and 12 months rhythm, mesor = 44555, amp = 3277.4, R = 0.86, P < 0.01 for the general population).
Dividing the patients by diagnosis younger or older than age 24 months (Fig. 2) revealed a different seasonality of MOB between the sexes in both age groups and, furthermore, a different pattern in the girls according to age at diagnosis (P < 0.02). For boys, no differences were found in MOB patterns between those diagnosed before 24 months of age and after that age (n = 81.8 months, rhythm, mesor = 7.27, amp8 = 2.55, R = 0.86, P < 0.01, for those diagnosed before 24 months [Fig. 3A]; n = 111.6 + 8 months rhythm, mesor = 10.02, amp8 = 3.7, amp6 = 1.3, R = 0.97, P < 0.001 for those diagnosed after that age [Fig. 3B]), and both peak in July. In girls, patients who received their diagnoses after the age of 24 months (n = 138) differed (P < 0.005) from those who received their diagnoses before the age of 24 months (n = 101), 4 + 8 months, mesor = 8.61, amp8 = 2.5, amp4 = 1.62, R = 0.74, P < 0.01.
Figure 3 shows that children with CD in the family had a different seasonality of MOB from children without such a family history (P < 0.001). Because of the small number, it was not possible to analyze the cases of children with CD and associated T1DM. In the boys with CD in a first-degree relative, there were peaks in January and August (n = 29; 4 + 8 months rhythm, mesor = 2.76, amp8 = 2.03, amp4 = 1.1, R = 0.822, P < 0.01), whereas in those with no CD in the family, the MOB peaked in June (n = 163; 4 + 8 months rhythm, mesor = 14.7, amp8 = 5.2, amp4 = 1.3, R = 0.9, P < 0.01).
In girls, the pattern of MOB was different between those 2 populations. In girls with CD in the family (n = 46; 6 + 8 months, mesor = 4.37, amp8 = 3.1, amp6 = 2.01, R = 0.793, P < 0.01), the peak was in the spring, whereas in girls with no CD in the family (n = 193; 4 + 8 months, mesor = 16.06, amp8 = 2.8, amp4 = 3.3, R = 0.71, P < 0.01) the peak was in the late summer.
This study shows that the seasonality of MOB in children with CD differed significantly from the pattern in the general population. The pattern also differed between girls and boys, between those whose cases were diagnosed before or after age 24 months, and between children with and without CD in first-degree family members.
Our findings extend the observation by Ivarsson et al (29) that Swedish children with CD show a peak of births during the summer months by revealing a variability in the pattern of MOB between the sexes and in children with a family history of CD.
The difference from the MOB pattern in the general population and excess births in the summer months of children with CD may be explained by endemic viral infections, such as enteroviruses, rotaviruses, and influenza, during the late autumn and winter. The same explanation has been postulated for childhood T1DM (22–24), assuming that the infected mother transmits the virus to the fetus in whom, if the fetus is genetically susceptible, the specific autoimmune disease process will be triggered. Additional pathogens such as gliadin in the case of CD will accelerate the damage and lead to the clinical disease (3). The difference in MOB seasonality in children with CD below and above age 24 months at diagnosis may denote differences in etiology, the viral cause predominating in the younger group.
The difference in MOB pattern between girls and boys and between children with and without CD in the family can be explained by a combination of genetic susceptibility and multiple environmental factors, not only viral infections. Whereas the viral hypothesis in the perinatal cause of T1DM is based on a series of direct (17–19) and indirect (21–24) findings, the link between virus infections of the mother (17) and CD is sparse.
In the present study we examined whether there is a possible linkage between the MOB, age at diagnosis of CD, and patients with or without a family history of CD. Our findings in children younger than 24 months support the reports that children born in the summer have a higher incidence of CD (29). However, we found that in boys the peak is indeed in the summer, but in girls there is an additional peak in the spring, which points to a second subpopulation of CD. Furthermore, only girls whose cases were diagnosed before the age of 2 years showed rhythm in MOB, whereas those whose cases were diagnosed after the age of 24 months were not rhythmic. The sex ratios for children with CD diagnosed before 24 months (0.59) compared with those with CD diagnosed after 24 months (1.1) supports the suggestion of different causes.
These findings show the complex interaction of genetic susceptibility and environmental factors acting in the perinatal period and mediating autoimmune diseases. Identifying the etiology in the subpopulations of CD patients may lead to new approaches in the primary prevention of CD, such as disrupting the autoimmune process by vaccines or other medications (30).
Other seasonal variants discussed in the literature, such as vitamin D (31), ultraviolet irradiation (32), and nutritional changes, are not applicable to the Israeli population living in a Mediterranean climate. Low birthweight, as suggested by Sandberg-Bennich et al (13), also did not apply to our study population.
Most studies of biological rhythm rely on 2 types of time series analytic approaches. One approach involves the fit of time series data by a mathematical model (eg, cosine function) with a predetermined period. The other approach involves subjecting the data to spectral analysis to ascertain information on the different ranges of cycles in the data. Epidemiological studies on yearly MOB distribution using the Poisson regression test (33) or the Walter and Elwood test (34) have been found to be of limited use for small populations (35–39). The advantage of the cosinor analysis is that in addition to statistical significance it provides parameters regarding the rhythms (28). Therefore, we analyzed the data using this method.
Celiac disease is diagnosed at an early age. Its link to T1DM has been demonstrated (4–7), and its natural history is only partially known (3), making a perinatal seasonal enteral virus infection a plausible candidate (13,27). Therefore, further studies are indicated to investigate the theory of perinatal viral infection as a trigger, which may open a route to possible prevention of this disease.
The authors thank Prof Gabriel Dinari and Prof Yoram Boyanover for permission to use medical records from their departments, Mrs Gila Waichman for technical assistance in manuscript preparation, and Dr Gabriela Halpern for English editing.
1. Maki M. The humoral immune system in celiac disease. Baillieres Clin Gastroenterol 1995; 9:231–249.
2. Korponay-Szabo IR, Haltunen T, Szalai Z, et al
. In vivo targeting of intestinal and extraintestinal transglutaminase 2 coeliac autoantibodies. Gut 2004; 53:641–648.
3. Branski D, Fasano A, Troncone R. Latest developments in the pathogenesis and treatment of celiac disease. J Pediatr 2006; 149:295–300.
4. Saukkonen T, Salvilahti E, Reijonen H, et al
. Coeliac disease: frequent occurrence after clinical onset of insulin dependent diabetes mellitus. Childhood Diabetes in Finland Study Group. Diabet Med 1996; 13:464–470.
5. Cronin CC, Shanahan F. Insulin dependent diabetes mellitus and celiac disease. Lancet 1997; 349:1096–1097.
6. Hansen D, Bennedbaek FN, Hansen LK, et al
. High prevalence of celiac disease in Danish children with type 1 diabetes mellitus. Acta Paediatr 2001; 90:1238–1243.
7. Simmons JH, Klingensmith GJ, McFann K, et al
. Impact of celiac autoimmunity on children with type 1 diabetes. J Pediatr 2007; 150:461–466.
8. Goh C, Banerjee K. Prevalence of coeliac disease in children and adolescents with type 1 diabetes mellitus in a clinic based population. Postgrad Med J 2007; 83:132–136.
9. Cavell B, Stenhammer L, Ascher H, et al
. Increasing incidence of childhood celiac disease in Sweden: results of a national study. Acta Paediatr 1992; 81:589–592.
10. Bach JF. The effect of incidence on susceptibility to autoimmune and allergic disorders. N Engl J Med 2002; 347:911–920.
11. Zanoni G, Navone R, Lunardi C, et al
. In celiac disease, a subset of autoantibodies against transglutaminase binds toll-like receptor 4 and induces activation of monocytes. PLoS Med 2006; 3:1637–1653.
12. Honeymarn MC, Coulson BS, Stone NL, et al
. Association between rotavirus infection and pancreatic islet autoimmunity in children at risk of developing type 1 diabetes. Diabetes 2000; 49:1319–1324.
13. Sandberg-Bennich S, Dahlquist G, Kallen B, et al
. Coeliac disease is associated with intrauterine growth and neonatal infections. Acta Paediatr 2002; 19:30–33.
14. Stene LC, Honeyman MC, Hoffenberg EJ, et al
. Rotavirus infection frequency and risk of celiac disease autoimmunity in early childhood: a longitudinal study. Am J Gastroenterol 2006; 101:2333–2340.
15. Troncone R, Auricchio S. Rotavirus and celiac disease: clues to the pathogenesis and perspectives on prevention. J Pediatr Gastroenterol Nutr 2007; 44:527–528.
16. Kagnoff MF, Austin RK, Hubert JJ, et al
. Possible role for a human adenovirus in the pathogenesis of celiac disease. J Exp Med 1984; 160:1544–1557.
17. Dahlquist G, Frisk G, Ivarsson SA, et al
. Indications that maternal coxsackie B virus infection during pregnancy is a risk factor for childhood-onset IDDM. Diabetologia 1995; 38:1371–1373.
18. Otonkoski T, Roivainen M, Vaarala O, et al
. Neonatal type 1 diabetes associated with maternal echovirus 6 infection: a case report. Diabetologia 2000; 43:1235–1238.
19. Varela-Calvino R, Peakman M. Enteroviruses and type 1 diabetes. Diabetes Metab Res Rev 2003; 10:431–441.
20. Jun HS, Yoon JW. A new look at viruses in type1 diabetes. Diabetes Metab Res Rev 2005; 19:8–31.
21. Viskari H, Ludvigsson J, Uibo R, et al
. Relationship between the incidence of type I diabetes and maternal enterovirus antibodies: time trends and geographical variation. Diabetologia 2005; 48:1280–1287.
22. Laron Z, Shamis I, Nitzan-Kaluski D, et al
. Month of birth and subsequent development of type 1 diabetes (IDDM). J Pediatr Endocrinol Metab 1999; 12:397–402.
23. Willis JA, Scott RS, Darlow BA, et al
. Seasonality of birth and onset of clinical disease in children and adolescents (0-19 years) with type 1 diabetes mellitus in Canterbury, New Zealand. J Pediatr Endocrinol Metab 2002; 15:645–647.
24. Laron Z, Lewy H, Wilderman I, et al
. Seasonality of month of birth of children and adolescents with type 1 diabetes mellitus in homogenous and heterogenous populations. Isr Med Assoc J 2005; 7:381–384.
25. Krassas GE, Tziomalos K, Pontikides N, et al
. Seasonality of month of birth of patients with Graves' and Hashimoto's diseases differ from that in the general population. Eur J Endocrinol 2007; 156:631–636.
26. Laron Z, Rotstein A, Kahana E, et al
. Multiple sclerosis and celiac disease patients similar to childhood type 1 diabetes have an abnormal seasonality of birth. Pediatr Diab 2005; 6(Suppl 3):12.
27. Laron Z. Does childhood diabetes start in utero? Riv Ital Pediatr 2001; 27:397.
28. Bingham C, Arbogast B, Guillaume GC, et al
. Inferential statistical methods for estimating and comparing cosinor parameters. Chronobiologia 1982; 9:397–439.
29. Ivarsson A, Hernell O, Nystrom L, et al
. Children born in the summer have increased risk of celiac disease. J Epidemiol Community Health 2003; 57:36–39.
30. Barzilai O, Ram M, Shoenfeld Y. Viral infection can induce the production of autoantibodies. Curr Opin Rheumatol 2007; 19:636–643.
31. McGrath J. Does “imprinting” with low prenatal vitamin D contribute to the risk of various adult disorders? Med Hypotheses 2001; 56:367–371.
32. Ponsonby AL, McMichael A, van der Mei I. Ultraviolet radiation and autoimmune disease: insights from epidemiological research. Toxicology 2002; 181–182:71–78.
33. Molina EC. Poisson's Exponential Binomial Limit. Princeton, NJ: Van Nostrand; 1942.
34. Walter SD, Elwood JM. A test for seasonality of events with a variable population at risk. Br J Prev Soc Med 1975; 29:18–21.
35. de Prins J, Hecquet B. Data processing in chronobiology studies. In: Touitou Y, House E, editors. Biological Rhythms in Clinical and Laboratory Medicine. Berlin: Springer-Verlag; 1992. pp. 90–113.
36. St Leger AS. Comparison of two tests for seasonality in epidemiological data. Appl Statist 1976; 25:280–286.
37. Karvonen M. Seasonality in the clinical onset of insulin dependent diabetes mellitus in Finnish children. Am J Epidemiol 1996; 143:167–176.
38. Roger JH. A significance test for cyclic trends in incidence data. Biometrika 1977; 64:152–155.
39. McKinney PA. Seasonality of birth in patients with childhood type I diabetes in 19 European regions. EURODIAB Seasonality of Birth Group. Diabetologia 2001; 44(Suppl 3):B67–74.