Is There Evidence for Aetiologically Distinct Subgroups of Idiopathic Congenital Talipes Equinovarus? A Case-Only Study and Pedigree Analysis

Background Idiopathic congenital talipes equinovarus (CTEV) is a common developmental foot disorder, the aetiology of which remains largely unknown. Some aspects of the epidemiology suggest the possibility of aetiologically distinct subgroups. Previous studies consider CTEV as a homogenous entity which may conceal risk factors in particular subgroups. We investigate evidence for aetiologically distinct subgroups of CTEV. Methods Parents of 785 probands completed a postal questionnaire. Family pedigrees were compiled by telephone. Case-only analysis was used to investigate interactions between risk factors and sex of the proband, CTEV laterality and CTEV family history. Results The male∶female ratio was 2.3∶1, 58% of probands were affected bilaterally and 11% had a first-second degree family history. There were modest interactions between family history and twin births (multivariate case - only odds ratio [ORca] = 3.87, 95%CI 1.19–12.62) and family history and maternal use of folic acid supplements in early pregnancy (ORca = 0.62, 95%CI 0.38–1.01); and between sex of the proband and maternal alcohol consumption during pregnancy (female, positive history and alcohol consumed: ORca = 0.33, 95%CI 0.12–0.89). Previous reports of an interaction between maternal smoking and family history were not confirmed. Relatives of female probands were affected more often than relatives of male probands. Conclusions These results provide tentative evidence for aetiologically distinct CTEV subgroups. They support the ‘Carter effect’, suggesting CTEV develops though a multifactorial threshold model with females requiring a higher risk factor ‘load’, and suggest areas where future aetiological investigation might focus. Large multi-centre studies are needed to further advance understanding of this common condition.


Introduction
Congenital talipes equinovarus (CTEV) is a common developmental disorder with birth prevalence of 1-4.5 per 1000. [1] Affected feet are inclined inwards, axially rotated outwards, and point downwards, with concomitant soft tissue abnormalities. [2] Severity ranges from cases that resolve with manipulation to those requiring multiple operations with disability and discomfort persisting into later life. Although some cases occur with other neuromuscular and neurological disorders, most affected children have idiopathic CTEV. [3] Mechanical, neurological, muscular, bony, connective tissue and vascular mechanisms for idiopathic CTEV have all been proposed. [3] Although genetic and lifestyle/environmental factors are thought to be aetiologically relevant, the genetic model is unclear and little is known about non-genetic risk factors. [3] However, some aspects of the epidemiology suggest areas worthy of further study: twice as many males as females are affected [4][5][6][7] and there is evidence of the 'Carter effect' (higher risk in relatives of affected females); [8,9] 7-21% [10,11] of families report CTEV in first-degree relatives, and one study suggests that family history modifies the association between CTEV and maternal smoking; [10] around half of affected children have bilateral CTEV [1,4,12,13] and mouse studies suggest the number of affected feet is a marker for genetic load. [14] These observations raise the possibility of aetiologically distinct CTEV subgroups. Previous studies have considered idiopathic CTEV as a homogenous entity that may have concealed risk factors relevant, or more important, in particular subgroups.
The ECCE (Exploring Causes of Clubfoot in Europe) study comprises the largest reported series of idiopathic CTEV with primary data collection. Here, we investigate interactions between epidemiological risk factors and family history, the proband's sex, and laterality of the condition. We also report family pedigree analyses.

Ethics Statement
The Grampian Research Ethics Committee approved the study and written consent was obtained from each participating family (most often the mother signed on behalf of her partner and participating children).

Subjects
Subjects were recruited May 2001-May 2003 through two support groups, steps [15] in the United Kingdom and VOK [16] in the Netherlands. The support groups approached families by mail on behalf of the investigators. A parent of the affected child (generally the mother) completed a questionnaire that included: nature of the condition (laterality, treatment, other medical conditions), maternal reproductive history, parental lifestyle (tobacco, alcohol, folic acid supplement and oral contraceptive [OC] use in the periconceptional period of the index pregnancy), and CTEV family history. On questionnaire receipt, a clinical geneticist (ZM) reviewed details of the foot defect and any additional conditions to exclude syndromic cases and non-CTEV conditions. Pedigrees were elicited by telephone from families who reported CTEV in family members other than the proband.

Statistical analysis
The analysis included unrelated index children with idiopathic CTEV. Case-only methods [17,18] were used to investigate whether CTEV risk factors differed by presence/absence of CTEV family history; sex of the proband; or laterality of the condition. Analysis contrasted sub-groups of cases with particular combinations of these ''stratification variables'' and risk factor exposures (e.g. male/female proband and maternal folic acid use/non-use), with the ''association'' between the stratification variable and risk factor (strictly the interaction, or departure from a multiplicative relationship) expressed as a case-only odds ratio (ORca). The stratification variables reference categories were: no family history; male; and unilateral CTEV. The primary analysis concerned first or second-degree family history. Using logistic regression, a ''minimally adjusted'' ORca was computed for each risk factor adjusted for country. Factors where the likelihood ratio test (LRT) p value was #0.1 in minimally adjusted analysis were considered for inclusion in multivariate models. Final multivariate models included country and variables where p#0.1 for the LRT comparing the multivariate model containing the variable with the model that did not. The family history analysis was repeated stratifying by sex, since sex differences have been reported. [10] Using the pedigrees, the total numbers of affected and unaffected first and second-degree relatives were determined. The ratio of affected to total relatives was calculated overall and by sex of the relative, proband, and relative and proband. Associations were assessed using the chi-square test.

Results
Of 1504 invited families, 827 completed questionnaires (participation rate = 55%). 42 families were excluded because the foot condition was not idiopathic CTEV. This analysis includes 785 probands.

Participant characteristics
The male:female ratio was 2.3:1 (Tables 1, 2). More than half had bilateral CTEV (58%). In unilateral cases the right foot was affected most often (56% right, 44% left). CTEV in first-second degree family members was reported by 11% of families, in firstthird degree relatives by 16% and in 'any' family member by 26%.

Family history associations
Factors that interacted with first-second degree family history in relation to CTEV risk were: maternal OC use, maternal use of folic acid-containing supplements, maternal ethnicity, twin birth and birthweight (Tables 3, 4, 5). Compared to those with no family history, probands with a family history were more likely to have a twin, have mothers who were non-Caucasian, and have mothers who took OCs in early pregnancy; they were less likely to have mothers who took folic acid supplements periconceptionally.
Maternal smoking in the periconceptional period was less common in those with a family history, reflected in an inverse, but non-statistically significant, ORca (multivariate ORca = 0.64, 95%CI 0.34-1.22, p = 0.16). The risk estimates were similar for smoking in the three months pre-conception and in the first trimester (data not shown). There was no association with paternal smoking.
After stratifying by sex, males with a family history were more likely than those without to have mothers who took OCs in early pregnancy (multivariate ORca = 4.35, 95%CI 1.01-18.78, p = 0.07) and to have a twin (ORca = 5.28, 95%CI 1.31-21.32, p = 0.03), and less likely to have mothers who took folic acidcontaining supplements first trimester (ORca = 0.59, 95%CI 0.31-1.10, p = 0.10) or who had previously had a miscarriage

Associations by proband sex
The factors which interacted with sex to affect CTEV risk were: maternal gravidity and miscarriage history, chorionic villus sampling in the index pregnancy, forceps delivery, birthweight, and proband birth year. Compared to males, females were more

Pedigree analysis
CTEV in first-degree relatives was reported in 5.7% (45/785) of families; 5.7% (45/785) had affected second-degree relatives, 1.0% (8/785) had affected first and second-degree relatives and 10.5% (82/785) had affected first or second-degree relatives. Of those with a first-degree family history, 38 had one affected relative (15 sibs, 14 fathers, nine mothers), six had two affected relatives (three sib/ mother-pairs, one sib/father-pair and two sib-pairs) and one had three affected relatives (mother and two sibs). Regardless of degree of relatedness, 139 families reported one affected relative, 46 reported two, 13 reported three, two reported four and five reported five. CTEV risk to any first-degree relative was 2.2% and to any first or second-degree relative 1.2% (Table 6). Male relatives were affected more often than female relatives (first-second degree 1.4% vs 1.0%, p = 0.05. Table 7) and relatives of female probands were affected more often than relatives of male probands (first-second degree 1.6% vs 1.0%, p = 0.01, Table 8). Male relatives of female probands had the highest absolute risk (first-second degree 2.0%, p = 0.02, Table 9).

Strengths and limitations
Most previous CTEV studies have either been based on routine data, which gives large sample sizes but lack certainty about the diagnosis of CTEV, or on small clinical series from single centres, which may be highly selected. In addition, studies do not always distinguish clearly between syndromic and idiopathic CTEV. The current study is the largest reported series of idiopathic CTEV involving primary data collection, and we carefully reviewed questionnaires to exclude syndromic CTEV and other foot conditions. The case-only design is statistically powerful for the investigation of interactions. [17,18] The key assumption underpinning the design is independence in the population between the stratification variable and risk factor; [19] if violated, risk estimates may be biased. We are not aware of any evidence to suggest the factors considered are not independent.
Recall accuracy and diagnostic reliability are challenges in family history analyses. We confirmed positive reports by telephone interview and additional questionnaires where possible, and restricted most analyses to first-second degree history, which may be more accurately reported.
Study participants were accrued from two national support groups, raising the possibility that they might not be representative of all idiopathic CTEV. For the results to be seriously biased, the ( probability of participation would need to have been associated with family history, laterality or proband sex. The sex ratio and laterality distribution mirrors patterns seen else-where. [1,4,6,11,[20][21][22][23][24][25][26][27] Moreover, the proportion with a family history corresponds with the upper limit of estimates from two US series, [4,28] is consistent with the UK Talipes series, [26] and is slightly lower that in series of 120 Scottish children. [22] This suggests our results are unlikely to be seriously biased.

Parental smoking
Reports of associations between foot deformities, including CTEV, and maternal smoking during pregnancy are inconsistent. [5,7,10,[29][30][31][32][33] One US case-control study of idiopathic CTEV reported a greater than multiplicative interaction between smoking and family history, such that maternal smoking increased risk only in children with a family history (OR 20.30, 95%CI 7.90-52.17). [10] We, in contrast, found no evidence of any interaction between family history and maternal (or paternal) smoking in the three months before, or first trimester of, the index  pregnancy. If anything our risk estimates suggested a less than multiplicative interaction, although they were not statistically significant.
In our study maternal smoking prevalence in the first trimester was 15% (22% in the three months before the pregnancy or first trimester) compared with 38% in the first trimester among cases in the US study. This difference could be due to differences in data collection methods (interview versus postal questionnaire), study location or subjects' period of birth (1968-1980 vs 1941-2003 [.70% 1991-2000]). The US study defined family history as 'probable' CTEV in first-degree relatives, but when we restricted our analysis to first-degree relatives and first trimester smoking the risk estimate was further from unity (multivariate ORca = 0.59, 95%CI 0.21-1.69, p = 0.30). The CTEV-smoking relationship, in those with or without a family history, thus remains controversial, and a role for smoking in CTEV cannot be entirely ruled out.

Perinatal factors and other maternal exposures during index pregnancy
The observed significant (p = 0.04) interaction between a positive family history and twin births is novel and may have become evident because, unlike previous studies of CTEV and twinning, [4,34] we stratified by family history. It could be interpreted as consistent with the uterine constraint hypothesis for CTEV. [3] As with other congenital anomalies, [35] there is some evidence of a role for folate metabolism in CTEV. [13,36,37] The borderline significant interaction between family history and maternal folic acid supplement use (p = 0.09) provides some further support for this. Although recall accuracy might be a concern, it seems unlikely this would be differential by family history. Since our results suggest supplement use might be associated with reduced CTEV risk in those without a family history further investigation is warranted.
Although observed in a subgroup analysis, the significant interaction (p = 0.02) between maternal alcohol consumption and family history in females is intriguing (mothers of female probands with a family history were less likely to report alcohol consumption). It is unlikely the finding reflects avoidance of 'risky' behaviour during pregnancy in women aware of a family history, as the association was not seen in males. Although alcohol is teratogenic, [38] it has rarely been considered in relation to CTEV and further investigation would be valuable.
The suggestion of an interaction between family history and maternal OC use in early pregnancy is of interest, especially as the effect was strongest in males. Increased risk of congenital limb deficiencies in offspring of mothers who had taken relatively highdose OCs in the periconceptional period has been reported, [39] suggesting our finding could be due to specific OC types (e.g. higher-dose or anti-androgenic OCs). We could not explore further as we did not have information on types of OCs used. However, while some studies report modest increased risks of birth defects, including limb deformities, with OC use, [40] the FDA concluded they were not teratogenic [41] and it is unclear how much of the maternal hormones reach the fetus and whether exogenous hormones are more likely to cross the placental barrier than endogenous (P Fowler, personal communication). Moreover, since our result was only borderline significant it may be due to chance.

Carter effect
Our results add to growing evidence for the Carter effect and a multifactorial threshold model in CTEV. The observed higher CTEV risk in relatives of female probands is consistent both with early work from Wynne-Davis et al, based on 144 UK cases born in 1940-1961, [8] and a recent US study which described increased CTEV transmission from mothers to their offspring compared with fathers. [9] Although other studies found CTEV risk was independent of the proband's sex, these included relatively few pedigrees (n,175). [4,26] The somewhat different risk factor pattern in females and males also points towards the possibility that a higher ''load'' of risk factors (whether genetic and/or environmental) in families of affected girls might predispose to CTEV.

Conclusions
Using the largest series of idiopathic CTEV with primary data collection so far reported, we set out to (1) follow-up previous observations suggesting the possibility of risk factor heterogeneity and (2) generate hypotheses for future study. Our results provide support for the 'Carter effect', suggesting that females require a higher risk factor 'load' before developing CTEV. Beyond this, although we found only tentative evidence for aetiologically distinct subgroups, our results do suggest some areas worth further exploration, including the relationships between family history and twinning and maternal use of folic acid supplements and alcohol during the index pregnancy. Large multi-centre studies, with sufficient power to fully explore risk factors in different case subgroups, are needed to further elucidate the aetiology of this common, but poorly understood, condition.