Surveillance of multiple congenital anomalies; searching for new associations

The study addresses how to improve detection of human teratogens by identifying congenital anomalies that co-occur more often than expected by chance. Because many teratogens lead to a spectrum of anomalies rather than isolated defects, analyzing patterns of multiple anomalies may reveal unknown teratogens or novel associations. Approximately 2% of births have a congenital anomaly, but specific combinations of multiple anomalies are rare, necessitating large datasets. The aim was to use EUROCAT registry data (2008–2016) to systematically identify pairs or triplets of anomalies occurring together more frequently than expected, determine whether these represent known associations/sequences, and flag potential new associations for further investigation including available genetic testing.

The Coding and Classification Committee evaluated significant associations against published literature to classify them as known associations/sequences or potential new ones. Ten pairs corresponded to established complexes or associations (limb body wall complex, OEIS complex, VACTERL association). Three pairs reflected the known association between neural tube defects and omphalocele. Other known links included PDA with diaphragmatic hernia (via pulmonary hypertension), and sequence-based associations such as oligohydramnios sequence (posterior urethral valves with clubfoot). Several pairs had limited published evidence (e.g., encephalocele with cleft lip; neural tube defects with gastroschisis). The committee used targeted literature searches and case-by-case registry reviews (including karyotype availability) to adjudicate borderline findings and exclude those likely due to undiagnosed genetic syndromes or coding errors.

Design and data sources: Population-based surveillance across 32 EUROCAT congenital anomaly registries, covering 6,599,765 births (2008–2016). From 154,154 cases with one or more major anomalies, cases with chromosomal anomalies, genetic syndromes, skeletal dysplasias, or hereditary skin disorders were excluded, leaving 123,566 cases for analysis.

Case classification: The EUROCAT multiple congenital anomaly algorithm (developed since 2004) classifies cases into: (a) chromosomal syndromes, (b) genetic/environmental syndromes, (c) isolated anomalies or known sequences, and (d) multiple congenital anomalies (two or more major anomalies in different organ systems, pattern not recognized as syndrome/sequence). About 90% are auto-classified into (a)–(c). Approximately 10% flagged as potential multiple anomalies undergo expert review by three EUROCAT geneticists via a web-based system; consensus or moderator decision finalizes classification.

Anomaly subgroups: Sixty EUROCAT subgroups (57 specific, plus neural tube defects (NTDs), congenital heart defects (CHD), and severe CHD) were analyzed. Pairs within the same organ/system, clubfoot with spina bifida or renal dysplasia (sequence-related), and situs inversus with any cardiac anomaly (heterotaxy spectrum) were excluded from pairwise testing.

Statistical analysis—pairs: Among confirmed multiple anomaly cases, for each eligible pair (A,B) the odds ratio quantified the odds of B given A vs B given not A; two-sided Fisher’s exact test provided p-values. Multiple testing control used the Benjamini–Hochberg procedure to estimate adjusted p-values; pairs with adjusted p<0.05 were taken forward. Sex-stratified analyses were also performed (excluding indeterminate/missing sex).

Statistical analysis—triplets: Logistic regression models assessed sets of three anomalies, including interaction terms to estimate the odds ratio for co-occurrence of all three given any two. Known related sets were excluded; FDR control via Benjamini–Hochberg was applied.

Sensitivity analysis: The pairwise analysis was repeated in the full set of anomaly cases (n=123,566), not only multiple cases. As this inflates relative odds (due to inclusion of isolated cases), a stricter significance threshold (adjusted p<0.01) was used for identifying notable pairs.

Post-analytic review: The EUROCAT Coding and Classification Committee reviewed statistically significant pairs to determine whether they reflect known associations/sequences, potential coding issues, undiagnosed genetic syndromes, or represent potential new associations. Registries provided additional case details and updated genetic testing (e.g., karyotyping) where available.

• Dataset: 6,599,765 births across 32 registries; 123,566 non-genetic anomaly cases included (2008–2016).
• Multiple anomalies: 8804 cases (7.1%; 95% CI 7.0–7.3) had two or more major anomalies in different organ systems (non-significant male-female difference: males 7.1% vs females 6.8%).
• Scope: 1386 distinct anomaly-pair combinations analyzed.
• Significant associations: 31 statistically significant positive pairs (FDR-adjusted p<0.05 in multiples-only analysis). Analyses using all anomaly cases with adjusted p<0.01 yielded a similar set. No triplet combinations were statistically significant.
• Known vs new: Of the 31 pairs, 20 were classified as known associations/sequences (e.g., limb body wall complex; OEIS complex; VACTERL; NTD with omphalocele; PDA with diaphragmatic hernia; oligohydramnios sequence). One pair was attributable to coding errors. Eleven pairs were flagged for deeper review as potential new associations.
• After literature and case review, six pairs remained as new associations warranting surveillance: (1) encephalocele with anophthalmos/micropthalmos; (2) cleft lip with anophthalmos/microphthalmos (e.g., OR in multiples-only 3.62; adj p<0.0001; all cases OR 4.43; adj p<0.0001); (3) hydrocephaly with hypoplastic right heart (e.g., OR 3.34; adj p=0.0178); (4) atrioventricular septal defect (AVSD) with duodenal atresia/stenosis (e.g., OR 2.65; adj p=0.0382; all cases OR 4.81; adj p<0.0001); (5) tetralogy of Fallot with duodenal atresia/stenosis (e.g., OR 2.53; adj p=0.0476; all cases OR 2.89; adj p=0.006); (6) severe CHD with duodenal atresia/stenosis (e.g., OR 2.02; adj p=0.0043).
• Some pairs initially flagged were deemed not new due to likely undiagnosed genetic syndromes (e.g., microcephaly with congenital cataract; encephalocele with Ebstein’s anomaly) or had only weak literature support (e.g., anencephalus or broader NTDs with gastroschisis; encephalocele with cleft lip).

Systematic surveillance within EUROCAT identified statistically enriched co-occurring anomaly pairs and distinguished known from potentially novel associations. Most significant pairs reflected established complexes (limb body wall, OEIS) and associations (VACTERL) or known pathophysiology (e.g., PDA with diaphragmatic hernia). Importantly, six pairs emerged as plausible new associations independent of known chromosomal/genetic syndromes, underscoring the value of large-scale, standardized, multi-registry surveillance to detect rare clustering that may signal shared etiologies, including potential teratogenic exposures.

Comparisons with prior methods (e.g., observed-to-expected co-occurrence adjusting for clustering) show broadly consistent findings; by restricting the primary analysis to multiple-anomaly cases and excluding known chromosomal/genetic disorders, this study targeted discovery of novel non-genetic clusters. Similar results from the sensitivity analysis including all anomaly cases support robustness. No triplet exceeded significance thresholds, likely reflecting rarity and limited power.

The six new pairs, particularly those linking duodenal atresia/stenosis with several forms of CHD and the associations involving ocular and cranial defects, merit etiologic exploration and thorough genetic evaluation. Ongoing surveillance with attention to genetic testing results (e.g., ensuring karyotype/molecular testing) is crucial to confirm whether these represent true non-genetic associations, potential shared developmental pathways, or unrecognized syndromic conditions.

Using a standardized algorithm and multi-registry EUROCAT data, the study screened 1386 anomaly-pair combinations and identified 31 significant co-occurrences, the majority being known associations. Six pairs were classified as new associations after expert and literature review: encephalocele with anophthalmos/micropthalmos; cleft lip with anophthalmos/microphthalmos; hydrocephaly with hypoplastic right heart; and duodenal atresia/stenosis with AVSD, tetralogy of Fallot, or severe CHD. These findings support the utility of systematic co-occurrence surveillance to generate hypotheses about shared etiologies and potential teratogens. Future work should prioritize detailed genetic testing and longitudinal surveillance to validate these associations, assess causal mechanisms, and monitor temporal or geographic clustering suggestive of environmental exposures.

• Genetic data completeness: Detailed genetic information (e.g., karyotype/molecular testing) was not available for all cases; analyses relied on registry-provided data, and individual cases could not be recontacted.
• Potential undiagnosed genetic syndromes: Some flagged associations may reflect undiagnosed chromosomal or single-gene disorders, especially where karyotype information was missing (noted for duodenal atresia–CHD pairs).
• Coding/misclassification: Occasional coding errors were identified and can influence apparent associations; efforts were made to exclude known sequences and related anomalies.
• Rarity and power: Specific combinations are rare, limiting power to detect associations, particularly for triplets, where none reached statistical significance.
• Generalizability: Findings pertain to European registry data (EUROCAT) and to non-genetic anomaly cases by design; applicability to broader or different populations may vary.