We conducted a genome-wide association study for nonsyndromic cleft lip with or without cleft palate (NSCL/P) in 401 affected individuals and 1,323 controls, with replication in an independent sample of 793 NSCL/P triads. We report two new loci associated with NSCL/P at 17q22 (rs227731, combined P = 1.07 × 10 −8 , relative risk in homozygotes = 1.84, 95% CI 1.34-2.53) and 10q25.3 (rs7078160, combined P = 1.92 × 10 −8 , relative risk in homozygotes = 2.17, 95% CI 1.32-3.56).NSCL/P is one of the most common human birth defects. In European populations, NSCL/P has a prevalence ranging from 1 in 700 to 1 in 1,000. We recently reported a susceptibility locus for NSCL/P at chromo some 8q24.21 from a genome wide association study in 224 individuals with NSCL/P (cases) and 383 population based controls 1 . This locus is the second susceptibility locus to have been unequivocally identified for NSCL/P to date, the first being the IRF6 locus 2 .To identify additional cleft susceptibility loci, we enlarged our sample by genotyping an additional set of 177 NSCL/P cases and adding the genotypes of 940 population based controls of central European origin. Genotyping was performed using Illumina BeadChips (Human610 Quad and HumanHap 550k).Following quality control (Supplementary Methods and Supplementary Fig. 1), association analysis of 521,288 SNPs having a minor allele frequency (MAF) of ≥1% in controls was performed in 399 cases and 1,318 controls.After excluding markers from the previously described 8q24.21 locus, 20 SNPs with P < 10 −5 remained. Five chromosomal loci (8q12.3, 10q25.3, 13q31.1, 15q13.3 and 17q22) were located within these 20 top SNPs, and the associations at these loci were further supported by at least three more SNPs with P < 10 −4 ( Supplementary Fig. 2 and Supplementary Table 1). Two additional regions were considered to be promising NSCL/P susceptibility loci (6p22.1, 11q14.2), as they contained at least four markers with P < 10 −4 .To replicate the genome wide association study (GWAS) findings, we selected the 20 top SNPs (P < 10 −5 ) as well as additional backup markers for each of the seven previously mentioned loci, resulting in two replication assays. We included additional SNPs with P < 10 −4 in the two replication assays, giving highest priority to SNPs with the lowest P values. Thus, a total of 56 markers were genotyped in a replication sample of 793 NSCL/P triads of European origin. Genotyping using matrix assisted laser desorption/ionization time of flight (MALDI TOF) mass spectrometry (Sequenom Inc.) was successful for 45 markers (representing 32 different loci), which were then analyzed by the transmission disequilibrium test in 665 triads (128 triads were excluded after quality control, Supplementary Methods).Of the 45 SNPs successfully genotyped, 11 (representing six differ ent loci) showed P < 0.05 in the replication sample (Supplementary Table 2). Two of these SNPs remained significant after correction for multiple testing by a conservative Bonferroni procedure (17q22: rs227731, P corr ...

Our study did not provide support for NOG being the causal gene at 17q22. However, the observation of a significant excess of rare variants in GREM1 supports the hypothesis that this is the causal gene at chr. 15q13. Because no single causal variant was identified, future sequencing analyses of GREM1 should involve larger samples and the investigation of regulatory elements.