Spina Bifida Research Resource: Study design and participant characteristics†
The data in this article have not been presented in full or in part at any meetings.
Abstract
BACKGROUND:
Myelomeningocele is a common serious malformation. In the majority of affected individuals, it is believed to be nonsyndromic and determined by the effects of multiple genetic and nongenetic factors.
METHODS:
The Spina Bifida Research Resource (SBRR) is an ongoing, family-based study, designed to identify maternal and embryonic genes related to myelomeningocele. Families that include at least one individual with myelomeningocele are eligible to participate in the SBRR. Recruitment into the SBRR has occurred in two phases. Descriptive analyses were undertaken to characterize the case individuals enrolled in Phase 1. In addition, the characteristics of subgroups of case individuals, defined by lesion level, were compared.
RESULTS:
During Phase 1, 671 families including 683 case individuals were enrolled. Families in which the case individual(s) were known or suspected to have a recognized pattern of malformations and families that did not complete the study interview were excluded from the present analyses. The case individuals in the remaining families (n = 534) were predominantly female (53%) and non-Hispanic Caucasian (87%), and in the majority (74%) the highest level of the lesion was in the lumbar region. Differences in the characteristics of the case individuals with lumbar and thoracic level lesions were detected.
CONCLUSIONS:
This article provides details regarding study design, recruitment, and data collection, and the characteristics of the SBRR Phase 1 study population. The data available from the SBRR, in combination with a rapidly evolving understanding of the variation within the human genome, provide an unprecedented opportunity to explore the genetic contribution to myelomeningocele. Birth Defects Research (Part A) 2008. © 2008 Wiley-Liss, Inc.
INTRODUCTION
The term spina bifida has been used as a label for a range of conditions (e.g., myelomeningocele, meningocele, occult spinal dysraphism) that involve the caudal neural tube and/or its surrounding vertebrae. Despite the common terminology applied to these conditions, they are likely to result from the disruption of different embryological processes. For example, myelomeningocele, a condition in which a central neural placode (i.e., a splayed-open neural structure) is herniated through a defect in the spine into a saclike membrane (i.e., open), is believed to result from failure of fusion of the neural tube during primary neurulation, whereas meningocele, a skin covered (i.e., closed) condition in which the meninges are herniated through a defect in the spine, is thought to result from an abnormality of secondary neurulation or canalization (Lemire, 1988; Moore, 2006). The specific causes of these various conditions are largely unknown, as is the extent to which they may share common risk factors (Mitchell, 2005).
Myelomeningocele is one of the most common and severe conditions that may be referred to as spina bifida. Given its severity, and its rank as one of the most common birth defects worldwide (Christianson et al., 2006), myelomeningocele (which is referred to simply as spina bifida from here forward) has received considerable attention from both the epidemiologic and genetic communities. Past studies have demonstrated that spina bifida is etiologically heterogeneous, and may occur in association with syndromes and other recognized patterns of malformations (e.g., sequences) of chromosomal (e.g., trisomy 13), Mendelian (e.g., Waardenburg), teratogenic (e.g., diabetic embryopathy), or unknown (e.g., amniotic band) origin (Lynch, 2005; Seaver and Stevenson, 2006). However, in the majority of individuals with spina bifida, the condition appears to be nonsyndromic (i.e., not associated with a syndrome or other recognized pattern of malformations), and a specific causative agent cannot be identified.
The epidemiological characteristics of nonsyndromic spina bifida have been extensively described and tend to be similar across study populations. In general, females are affected more frequently than males. In addition, prevalence varies with race and ethnicity, and exhibits both geographic and temporal variation (reviewed in Mitchell et al., 2004). Several studies have suggested that these characteristics may differ across subgroups of affected individuals, defined by the location of the defect along the caudal neural tube (high vs. low) and/or by the presence of additional, nonsecondary malformations (Khoury et al., 1982a, b; Toriello and Higgins, 1985; Frecker et al., 1988). Such differences have not, however, been convincingly established (Mitchell, 2005). Moreover, although epidemiological studies have investigated a wide range of potential risk factors, the list of factors for which there is convincing evidence of an association with spina bifida is rather short: maternal obesity (Waller et al., 2007), maternal hyperthermia (Moretti et al., 2005), and lack of maternal, periconceptional folic acid supplementation (Pitkin, 2007).
Spina bifida is of interest to geneticists because it aggregates within families. That is, the risk of spina bifida among the relatives of an affected individual is increased as compared to the risk in the general population (e.g., the risk to a sib of an affected individual is approximately 30 to 40-fold higher than the risk in the general population). The family recurrence pattern for nonsyndromic spina bifida is, however, inconsistent with simple Mendelian inheritance. Based on the observed familial aggregation patterns, C.O. Carter proposed that spina bifida is inherited as a multifactorial threshold trait (Carter, 1969), that is, a trait for which liability is determined by multiple factors each with a relatively small effect, and affection status is determined relative to a threshold liability value. In contemporary terminology, spina bifida is considered to be a complex trait that is determined by the effects of multiple genetic and nongenetic factors that, individually, have only a small effect on risk.
Although spina bifida aggregates within families, it is a less than ideal candidate for linkage-based genetic studies because families with more than one affected member from whom DNA can be obtained are rare. Hence, it was not until the advent of genetic association studies in the late 1980s and early 1990s that studies aimed at identifying genes that influence the risk of spina bifida truly became feasible. As the initiation of genetic association studies for complex traits coincided with the widespread recognition of the protective effect of maternal folic acid supplementation for spina bifida, it is not surprising that the first association studies of spina bifida focused on genes that are involved in the transport and metabolism of folic acid (van der Put et al., 1995; Whitehead et al., 1995). These early studies raised an interesting issue that had not previously been considered in the design and analysis of genetic association studies. Specifically, because folate-related genes may influence the risk of spina bifida via an effect of the maternal and/or the embryonic genotypes, studies aimed at assessing the effects of such genes must account for the correlation between the maternal and embryonic genotypes in order to obtain valid results. As traditional association-based methods do not account for this correlation (Posey et al., 1996), the study of folate-related genes as risk factors for spina bifida required the development of novel study designs and analytical approaches (e.g., Mitchell, 1997; Weinberg et al., 1998; Weinberg, 2003; Mitchell and Weinberg, 2005).
The Spina Bifida Research Resource (SBRR) was initiated in the mid-1990s with the goal of assembling a study population that would provide a powerful platform for genetic association studies aimed at the identification of maternal and embryonic genes that influence the risk of spina bifida. This report provides details regarding study design, recruitment, and data collection, and the characteristics of the study population assembled thus far.
MATERIALS AND METHODS
Study Design
From its initiation, the goal of the SBRR was to identify maternal and embryonic genes that influence the risk of nonsyndromic defects of the caudal neural tube that are likely to be attributable to an abnormality in the process of primary neurulation (i.e., myelomeningocele or open spina bifida). To this end, the SBRR is based on a novel family-based design, in which the family unit of primary interest is a “pent”. A complete pent includes an individual/pregnancy affected with spina bifida (i.e., case), and his/her parents and maternal grandparents. This design was originally suggested as one that would allow for the evaluation of both maternal and embryonic genetic effects and would not be subject to population stratification bias (Mitchell, 1997). Statistical methods for pent data that provide valid assessment of both maternal and embryonic genetic effects have subsequently been developed (Mitchell and Weinberg, 2005).
Because the pent design does not include an unaffected control group, the evaluation of hypotheses regarding associations between spina bifida and nongenetic risk factors is precluded. However, hypotheses regarding gene-environment interactions can be assessed. Consequently, the study questionnaire captured information relevant to the three most convincingly established risk factors for spina bifida; maternal hyperthermia, obesity, and periconceptional folic acid intake. Further, the questionnaire captured information on variables that can be used to define potentially more homogeneous subgroups of the case families (e.g., lesion level).
Subject Recruitment and Data Collection
The recruitment of families (i.e., pents) into the SBRR has occurred in two phases. Recruitment for Phase 1 was initiated in November 1997 and ended in April 2006, and recruitment for Phase 2 was initiated in October 2006 and is ongoing. Phases 1 and 2 are differentiated by the questionnaire used during the interview (both questionnaires are available at: www.sbrr.info) and the biological samples used as the primary source of DNA (i.e., buccal swabs in Phase 1 and saliva in Phase 2). This report provides details regarding Phase 1 recruitment and data collection, and the characteristics of the cases enrolled during this phase.
Recruitment during Phase 1 of the SBRR focused on families that included an individual or pregnancy affected with nonsyndromic spina bifida. Potential cases included individuals/pregnancies from all races and ethnicities, of both sexes, and all ages, with or without additional birth defects. Recruitment occurred onsite at two children's hospitals and through mailings conducted in association with various clinics and organizations. Participation included an interview, which was usually conducted with the mother of the case. In addition, each participating family member was asked to self-collect a sample of buccal cells using cytobrushes (two brushes/subject) or to undergo phlebotomy to obtain a blood sample. No other biological samples or sources of information (e.g., medical records) were collected.
Onsite recruitment for Phase 1 of the SBRR occurred in conjunction with the Spina Bifida Program (11/97–9/04) and the Center for Fetal Diagnosis and Treatment (3/99–6/03) at The Children's Hospital of Philadelphia in Pennsylvania, and the Spinal Dysfunction Program of the Nemours Alfred I. duPont Hospital for Children in Wilmington, Delaware (10/99–10/02). Recruitment at these sites was conducted by a member of the SBRR staff. In general, potential participants were provided with written information about the SBRR prior to a scheduled clinic appointment, and met with an SBRR staff member during their appointment to determine their eligibility for, and interest in, study participation. Once appropriate assent and/or consent was/were obtained, the interview was conducted onsite, in-person or scheduled as a telephone interview for a later date. In addition, biological samples were collected onsite and/or sample collection kits with prepaid return mailers were provided to each study participant.
During Phase 1, recruitment also occurred through mailings conducted in association with other spina bifida clinics and local spina bifida associations (n ∼ 30). In general, recruitment packets that included information about the SBRR and a response form with a postage paid return envelope were provided by the SBRR to a participating organization (e.g., spina bifida clinic or association). The packets were addressed by members of the participating organization and mailed to the individuals served by that organization. Individuals who received packets and who were interested in participating in the SBRR either returned the response form or called the SBRR via a toll-free telephone number. Potential subjects spoke with a member of the SBRR staff to determine study eligibility, and consent/assent documents were mailed to eligible individuals. Once appropriate consent/assent was obtained, a telephone interview was scheduled, and a sample collection kit with prepaid return mailer was provided to each participant.
The study protocol was approved by the Institutional Review Boards of Texas A&M University, University of Pennsylvania School of Medicine, and each of the participating hospitals. Informed consent, and when appropriate, assent was/were obtained for each study participant.
The Phase 1 study interview was based on a paper and pencil questionnaire administered by a member of the SBRR staff. Information collected during the interview included: maternal health and pregnancy histories; parental education, occupation, and reproductive histories; periconceptional exposures (i.e., exposures that occurred early in pregnancy, before the mother realized that she was pregnant); details of the case individual's lesion and the presence of additional malformations; and a three-generation family history focused on the presence of congenital malformations in the relatives of the case. Additional malformations reported for the case were classified as secondary defects (e.g., hydrocephalus, club foot, spine, rib, or sterum defects), minor defects (e.g., cleft gum, high arched palate, rocker bottom feet, pilonidal sinus, hernias other than diaphragmatic), and major defects (e.g., heart defects, diaphragmatic hernia, polydactyly, anal atresia).
The questionnaire and family history data are stored in a Progeny database (Progeny Software LLC, South Bend, IN) and can be exported to a variety of data formats (e.g., Excel) for analysis. In addition, DNA was extracted from all biological samples provided by Phase 1 participants, as previously described (Jensen et al., 2006). Each DNA sample was split into two aliquots, with one maintained as a working sample and the other stored as a backup and for future use.
Statistical Methods
Descriptive analyses were undertaken to characterize the case individuals enrolled during Phase 1 of the SBRR. These analyses focused on demographics, factors that have previously been reported to be associated with the risk of spina bifida (e.g., maternal obesity), and common exposures (e.g., maternal use of alcohol and cigarettes). All variables were summarized using counts and proportions.
Year of birth was categorized as being prior to 1997, 1997–1998, and after 1998. The first category (i.e., prior to 1997) corresponds to case individuals conceived prior to the implementation of mandatory folic acid fortification of the US food supply in January 1998. The second category (i.e., 1997–1998) corresponds to case individuals conceived during a transition period that included voluntary folic acid fortification of foods, and/or whose mothers may not have yet fully benefited from mandatory fortification. The third category (i.e., after 1998) corresponds to case individuals who would have been conceived at least 3–4 months following mandatory fortification. Maternal body mass index (BMI) was calculated as weight(kg)/(height[m])2 using maternal prepregnancy height and weight information obtained during the study interview.
The characteristics of subgroups of the cases, defined by the reported highest level of the spina bifida lesion, were compared using either the chi-square or Fisher's exact test. All statistical analyses were conducted using SAS version 8.02 (SAS Institute Inc., Cary, NC).
RESULTS
During Phase 1 (11/97–4/06), 671 families including 683 case individuals were enrolled in the SBRR. These families included 660 (98.4%) with a single case, nine with two affected sibling cases each, one with three affected sibling cases, and one with an affected father-child case pair. Among the case individuals, 31 were reported to have, or were suspected of having a syndrome or other recognized pattern of malformations (Table 1). Among the remaining case individuals, the majority had an open defect involving the caudal neural tube (n = 627 cases in 617 families), 13 had a lipomeningocele, two had anencephaly, two had sacral agenesis, and eight had a closed spina bifida.
| Cases | Families | |
|---|---|---|
| Known or suspected syndrome | 10 | 10 |
| Deletion 22q11 | 1 | 1 |
| Abnormality of 10p | 1 | 1 |
| Chromosome abnormality, unspecified | 1 | 1 |
| VATER | 1 | 1 |
| Hay-Wells (ankyloblepharon-ectodermal defects-cleft lip/palate) | 1 | 1 |
| Waardenburg | 2 | 2 |
| Heterotaxy | 1 | 1 |
| Amniotic band | 1 | 1 |
| Unspecified | 1 | 1 |
| Teratogenic exposure | 14 | 12 |
| Anticonvulsant medication and/or maternal epilepsy1 | 13 | 11 |
| Isotretinoin | 1 | 1 |
| Maternal pregestational diabetes | 7 | 7 |
- 1 Reported anticonvulstants included: carbamazepine, phenytoin, phenobarbital, valproic acid, and acetazolamide.
All subsequent analyses were based on the case individuals with open spina bifida who were not identified as having, or suspected of having, a syndrome or other recognized pattern of malformations, and included a single case individual from each family (n = 617 cases). In general, data for the oldest case individual within a sibship were retained for analysis. In the single family that included a father-child case pair, data for the child case were retained. A completed study interview was available for 534 (86.5%) of these case individuals.
Details regarding recruitment and sample collection for the 534 case individuals with completed interviews are provided in Table 2. Briefly, 40.4% of these cases were recruited onsite at one of three clinics, 43.4% were recruited by mail, and 16.1% were recruited through other sources (e.g., website, advertisements, referral). Between 5.6% and 20.2% of the cases were recruited in any given year. The study interview was completed by the mother of the case, with or without input from the case individual or other relatives, for approximately 96% of the cases for which this information was recorded. On average, the study interview was completed approximately 13 years after the birth of the case or the termination of an affected (case) pregnancy. Biological samples were obtained for approximately 95% (n = 505) of these case individuals.
| Variable | n (%) |
|---|---|
| Recruitment method | |
| Onsite* | |
| CHOP, Spina Bifida Program | 125 (23.4) |
| CHOP, Center for Fetal Diagnosis and Treatment | 45 (8.4) |
| duPont, Spinal Dysfunction Program | 46 (8.6) |
| Spina bifida clinic | 27 (5.0) |
| Spina bifida association | 205 (38.4) |
| Other (self-referral, website, advertisements) | 86 (16.1) |
| Year of Recruitment | |
| 11/97–1998 | 74 (13.8) |
| 1999 | 30 (5.6) |
| 2000 | 44 (8.2) |
| 2001 | 42 (7.9) |
| 2002 | 72 (13.5) |
| 2003 | 108 (20.2) |
| 2004 | 86 (16.1) |
| 2005–4/06 | 78 (14.6) |
| Interview participants (n = 441)† | |
| Mother only | 366 (83.0) |
| Mother and case or other relative(s) | 57 (12.9) |
| Case only | 13 (2.9) |
| Father only | 4 (0.9) |
| Other relative only | 1 (0.2) |
| Biological sample obtained from case | |
| Yes | 505 (94.6) |
| No | 29 (5.4) |
- * CHOP, Children's Hospital of Philadelphia; duPont, Nemours Alfred I. duPont Hospital for Children.
- † Information regarding interview participants was not recorded for 93 families.
Among the 534 case individuals with a completed study interview, the highest level of the spina bifida lesion, as reported during the study interview, was lumbar in 394 (73.8%), thoracic in 83 (15.6%), sacral in 45 (8.4%), and was unknown or unspecified in 12 (2.2%). No case individual was reported to have a lesion that extended into the cervical region. As there is evidence that the epidemiological characteristics, and thus perhaps the etiology, of spina bifida differs by lesion level, the characteristics of the study cases and their mothers were determined for the full case group, as well as subgroups of cases defined by lesion level (i.e., sacral, lumbar, thoracic). Further, to assess heterogeneity between subgroups, the distribution of variables among cases with sacral or thoracic level lesions was compared to that observed in the cases with lumbar lesions.
The full case group included 284 (53.2%) females and 250 (46.8%) males, and was predominantly non-Hispanic Caucasian (87.2%). A female predominance was also observed in each of the subgroups (Table 3). Neither the sex ratio nor the distribution of race and ethnicity was significantly different in the subgroups of case individuals with sacral or thoracic, as compared to those with lumbar level lesions. Additional major malformations were reported in 7.7% of all case individuals. The most commonly reported major malformations involved the heart and kidney (Table 4). The proportion of case individuals with an additional major malformation was similar in case individuals with sacral and lumbar level lesions, but was significantly higher in those with thoracic as compared to lumbar level lesions (14.5 vs. 6.1%, respectively). In general, the types of malformations reported in the cases from the various subgroups were similar (Table 4).
| Variable | All cases (n = 534)* | Sacral (n = 45) | Lumbar (n = 394) | Thoracic (n = 83) |
|---|---|---|---|---|
| Case characteristics | ||||
| Sex | ||||
| Female | 284 (53.2) | 26 (57.8) | 205 (52.0) | 48 (57.8) |
| Male | 250 (46.8) | 19 (42.2) | 189 (48.0) | 35 (42.1) |
| p-value† | — | .46 | — | .40 |
| Race/Ethnicity | ||||
| Non-Hispanic Caucasian | 462 (87.2) | 40 (90.9) | 344 (87.8) | 71 (85.5) |
| Hispanic | 40 (7.6) | 3 (6.8) | 28 (7.1) | 6 (7.2) |
| Other | 28 (5.3) | 1 (2.3) | 20 (5.1) | 6 (7.2) |
| p-value | — | .68 | — | .65 |
| Additional malformations | ||||
| No | 493 (92.3) | 42 (93.3) | 370 (93.9) | 71 (85.5) |
| Yes | 41 (7.7) | 3 (6.7) | 24 (6.1) | 12 (14.5) |
| p-value | — | .75 | — | .01 |
| Year of birth | ||||
| Prior to 1997 | 361 (67.6) | 33 (73.3) | 244 (61.9) | 75 (90.4) |
| 1997–1998 | 30 (5.6) | 3 (6.7) | 25 (6.4) | 1 (1.2) |
| After 1998 | 143 (26.8) | 9 (20.0) | 125 (31.7) | 7 (8.4) |
| p-value | — | .26 | — | <0.0001 |
| Maternal characteristics | ||||
| Education | ||||
| Did not complete high school | 22 (4.6) | 2 (5.1) | 17 (4.6) | 2 (3.2) |
| Completed high school | 107 (22.3) | 9 (23.1) | 71 (19.4) | 22 (34.9) |
| Some college | 127 (26.4) | 13 (33.3) | 100 (27.3) | 12 (19.0) |
| Four or more years of college | 224 (46.7) | 15 (38.5) | 178 (48.6) | 27 (42.9) |
| p-value | — | .62 | .06 | |
| Pre-pregnancy body mass index | ||||
| <18.5, underweight | 28 (5.6) | 4 (9.3) | 19 (5.1) | 5 (6.4) |
| 18.5–24.9, normal | 331 (65.8) | 30 (69.8) | 241 (65.1) | 50 (64.1) |
| 25.0–29.9, overweight | 89 (17.7) | 5 (9.3) | 71 (19.2) | 14 (18.0) |
| ≥30.0, obese | 55 (10.9) | 5 (11.6) | 39 (10.5) | 9 (11.5) |
| p-value | — | .28 | — | .93 |
| Gestational diabetes | ||||
| No | 490 (95.7) | 40 (95.2) | 364 (96.0) | 74 (93.7) |
| Yes | 22 (4.3) | 2 (4.8) | 15 (4.0) | 5 (6.3) |
| p-value | — | .68 | — | .36 |
| Drugs to stimulate ovulation | ||||
| No | 505 (94.6) | 43 (95.6) | 372 (94.4) | 79 (95.2) |
| Yes‡ | 29 (5.4) | 2 (4.4) | 22 (5.6) | 4 (4.8) |
| p-value | — | .27 | — | 1.00 |
| First trimester fever | ||||
| No | 507 (94.9) | 45 (100.0) | 371 (94.2) | 80 (96.4) |
| Yes | 27 (5.1) | 0 (0.0) | 23 (5.8) | 3 (3.6) |
| p-value | — | .15 | — | .60 |
| Periconception exposures/maternal behaviors (prior to pregnancy recognition) | ||||
| Cigarette smoke | ||||
| No | 393 (74.3) | 31 (68.9) | 293 (75.3) | 60 (72.3) |
| Yes, mother smoked | 112 (21.2) | 10 (22.2) | 79 (20.3) | 20 (24.1) |
| Yes, second hand smoke only | 2 (4.5) | 4 (8.9) | 17 (4.4) | 3 (3.6) |
| p-value | — | .33 | — | .71 |
| Maternal use of alcohol | ||||
| No | 326 (61.5) | 28 (62.2) | 248 (63.3) | 40 (49.4) |
| Yes | 204 (38.5) | 17 (37.8) | 144 (36.7) | 41 (50.6) |
| p-value | — | .87 | — | .02 |
| Maternal use of hot tub, sauna or tanning bed | ||||
| No | 513 (96.1) | 45 (100.0) | 375 (95.2) | 81 (97.6) |
| Yes | 21 (3.9) | 0 (0.0) | 19 (4.3) | 2 (2.4) |
| p-value | — | .24 | — | .55 |
| Maternal use of multivitamins | ||||
| No | 302 (64.4) | 27 (65.8) | 210 (61.0) | 59 (79.7) |
| Yes | 167 (35.6) | 14 (34.2) | 134 (39.0) | 15 (20.3) |
| p-value | — | .61 | — | .002 |
- * Includes case individuals for whom the level of the spina bifida lesion was unknown or unspecified (n = 12).
- † P-value from chi-square or Fisher exact test of the comparison with cases that have lumbar level lesions.
- ‡ Maternal use of clomiphene (clomid) was reported in 18 of the 29 cases in this group, including 15 cases with lumbar, two cases with thoracic and one case with an unknown/unspecified lesion.
| Malformation | All cases (n = 534)* | Sacral (n = 45) | Lumbar (n = 394) | Thoracic (n = 83) |
|---|---|---|---|---|
| Congenital heart defect | 15 | 1 | 10 | 4 |
| Kidney anomaly† | 15 | 1 | 6 | 7 |
| Craniosynostosis | 4 | 0 | 4 | 0 |
| Polydactyly/syndactyly‡ | 4 | 0 | 3 | 1 |
| Cleft palate | 2 | 0 | 1 | 1 |
| Anal atresia | 2 | 0 | 1 | 1 |
| Omphalocele | 2 | 0 | 1 | 0 |
| Diaphragmatic defect | 1 | 1 | 0 | 0 |
| Total | 45 | 3 | 26 | 14 |
| # of cases with additional malformations | 41 | 3 | 24§ | 12‖ |
- * Includes case individuals for whom the level of the spina bifida lesion was unknown or unspecified (n = 12).
- † Single, horseshoe or fused kidney, or double or absent ureter.
- ‡ Pre-axial polydactyly (n = 2); postaxial polydactyly (n = 1); syndactyly (n = 1).
- § One case with both a heart defect and craniosynostosis, and one with both a horseshoe kidney and polydactyly.
- ‖ One case with both a fused kidney and anal atresia, and one with a heart defect and a single kidney.
There were no significant differences between case individuals with sacral and lumbar lesions with regard to year of birth, maternal characteristics, or exposures that occurred early in pregnancy (Table 3). In contrast, case individuals with thoracic and lumbar level lesions differed with respect to year of birth (p < .0001), maternal educational level (p = .06), maternal use of alcohol (p = .02), and maternal use of multivitamins (p = .002). In general, cases with thoracic level lesions were more likely to be born prior to folic acid fortification, and their mothers were less educated, more likely to drink alcohol, and less likely to use a multivitamin as compared to case individuals with lumbar level lesions (Table 3).
As case individuals were born over a wide time span (1932–2006), the observed differences between cases with thoracic and lumbar lesions may reflect temporal trends that are unrelated to the variables of interest; in particular folic acid fortification and maternal use of multivitamins. Consequently, additional analyses of year of birth and maternal use of multivitamins were undertaken using data only from case individuals with thoracic or lumbar lesions, born in 1990 or later (Table 5). However, even within this subgroup of cases, individuals with thoracic level lesions were more likely to have been born prior to folic acid fortification (p = .002), and their mothers were less likely to have used multivitamins (p = .01), as compared to case individuals with lumbar level lesions.
| Lumbar | Thoracic | |
|---|---|---|
| Year of birth | ||
| 1990–1996 | 74 (35.6) | 17 (73.9) |
| 1997–1998 | 23 (11.0) | 0 (0.0) |
| 1999–2006 | 111 (53.4) | 6 (26.1) |
| p-value* | .002 | |
| Multivitamin use | ||
| No | 103 (49.5) | 18 (66.7) |
| Yes | 105 (50.5) | 5 (21.7) |
| p-value | .01 | |
- * p-value from chi-square or Fisher exact test.
DISCUSSION
It is important to note that the SBRR is not a population-based study, and that the information provided in this article is not intended as a description of the general population of individuals who have spina bifida. As the SBRR is not population-based, the characteristics of the study participants reflect factors that are associated with study participation as well as factors related to the development of spina bifida. Hence, this article serves only to provide a descriptive overview of this relatively large and unique study population.
Direct comparison between the characteristics of the SBRR case individuals and other study populations is problematic due to differences in case definitions, ascertainment sources, and phenotypic assessments across study populations. However, the syndromes and additional malformations observed in the SBRR case individuals are, in general, consistent with other reports in the literature (see for example: Kallen et al., 1998; Seaver and Stevenson, 2006; Stoll et al., 2007). The one exception to this generalization is the case individual with Hay-Wells (ankyloblepharon-ectodermal defects-cleft lip/palate) syndrome, as neural tube defects have not been reported as part of the phenotypic spectrum for this syndrome.
Data from subsets of the families enrolled in Phase 1 of the SBRR have been used to evaluate the association between spina bifida and several putative candidate genes: MTR and MTRR (Doolin et al., 2002), NOS3 (Brown et al., 2004), T (Jensen et al., 2004a), ABCC2 (Jensen et al., 2004b), and NAT1 (Jensen et al., 2005, 2006), and additional candidate gene studies based on the full Phase 1 group are forthcoming. The information contained in the present article will provide a useful supplement to the publications describing these studies, because it includes a more comprehensive description of the study design and participants than a typical Material and Methods section. Moreover, the information contained in this article may be helpful to others who are interested in initiating similar studies.
The information presented in this article will also be used to guide subsequent statistical analyses using the SBRR Phase 1 data, and may be useful to other groups that are investigating genetic risk factors for spina bifida using family-based study designs (e.g., Boyles et al., 2006; King et al., 2007). In particular, the differences observed between cases with lumbar and thoracic level lesions suggest that future analyses should include consideration of subgroups of case families defined by lesion level. Although the observed differences between the thoracic and lumbar subgroups may be reflective of differences in the factors that influence participation of cases with lumbar and thoracic level lesions (e.g., maternal educational level), or differences in the accuracy with which exposure information is reported, it is also possible that they result from differences in the risk factor profiles for lesions that occur at different levels of the neural tube. Additional evidence supporting the inclusion of subgroup analyses based on lesion level in studies of potential spina bifida risk factors includes previous studies that suggest that the familial aggregation patterns of spina bifida are related to lesion level (Toriello and Higgins, 1985; Hall et al., 1988) and studies suggesting that variation in the MTHFR and BRCA1 genes is associated with lesion level in individuals with spina bifida (Volcik et al., 2003; King et al., 2007).
Although no significant differences were observed between the sacral and lumbar subgroups in Phase 1 of the SBRR, the power to detect such differences was limited by the relatively small number of cases with sacral level lesions. It would, therefore, seem prudent to distinguish between sacral, lumbar, and thoracic level lesions in any subgroup analysis. However, given the relatively small numbers of case individuals with sacral and thoracic level lesions in Phase 1 of the SBRR, such analyses will necessarily focus on the subgroup of families in which the case has a lumbar level lesion.
It is also possible that there are differences in the risk factor profiles of case individuals with and without additional malformations. However, due to the relatively small number of case individuals with additional malformations (n = 41), and the association between additional malformations and lesion level, it was not possible to formally assess such differences using the SBRR Phase 1 data. It may, however, also be prudent to consider the presence/absence of additional malformations in any subgroup analyses.
Further examination of potential differences in the risk factor profiles of subgroups of individuals with spina bifida, defined by lesion level and the presence of additional malformations, would be of interest. Such studies would require large, population-based study samples, such as that provided by the National Birth Defects Prevention Study (Yoon et al., 2001), and would likely benefit from the use of methods such as polytomous logistic regression (Glynn and Rosner, 2004) for assessing heterogeneity of the exposure effect across subgroups defined by lesion level and the presence/absence of additional malformations.
As with any study, the SBRR has some weaknesses. The study design is not population-based; recruitment through clinics and support organizations is likely to have biased the study sample against individuals with severe forms of spina bifida (e.g., thoracic level lesions) and those with other serious, associated malformations (e.g., congenital heart defects) that are associated with increased mortality. The study design also does not allow for the evaluation of the main effects of potential, nongenetic risk factors. Further, all data are based on the response of the interviewee (usually the mother of the case) and are therefore subject to the usual sources of reporting error (e.g., under-reporting, misclassification). Although not discussed in detail in this article, the primary source of DNA for Phase 1 participants was buccal cells, which provide a limited source of DNA that is generally of lower quality than DNA from blood or saliva (Rogers et al., 2007). In addition, the available study sample has limited power for the analysis of subgroups defined by lesion level or other variables and for analyses that involve large numbers of comparisons (e.g., genome-wide association studies).
The SBRR also has several strengths. It is one of the largest family-based studies of spina bifida. In addition, the SBRR is unique in its design and potential for resolving the effects of the maternal and embryonic genotype. To further strengthen the SBRR, all Phase 1 participants who previously provided a buccal sample are being recontacted and asked to provide a saliva sample as a new source of DNA, and the recruitment of additional case families is ongoing (all new study subjects are asked to provide either blood or saliva as a source of DNA).
In summary, the SBRR is a large, ongoing, family-based study designed to identify maternal and embryonic genes that influence the risk of spina bifida. It is one of only a few family-based studies of spina bifida worldwide, and employs a unique design that allows for analyses that can differentiate between maternal and embryonic genetic effects and that are not subject to bias arising from population stratification (see Pent-2 approach in Mitchell and Weinberg, 2005). Analyses of subsets of the SBRR Phase 1 data have confirmed previously reported associations between spina bifida and specific genetic variants (e.g., T locus; Jensen et al., 2004a), and identified new associations (e.g., NAT1; Jensen et al., 2005, 2006). The data available from the SBRR, in combination with the rapidly evolving understanding of the variation within the human genome (e.g., single nucleotide polymorphisms, copy number variants: Newton-Cheh and Hirschhorn, 2005; McCarroll and Altshuler, 2007), provide an unprecedented opportunity to explore the genetic contribution to spina bifida.
Acknowledgements
The author is grateful to the many individuals who have assisted with the recruitment of study subjects and to all of the individuals who have participated in this study. The author thanks Minghua Mei for assistance with data management, and Katy Hoess and Barbara Weyland for subject recruitment and data collection.
