Three loci tested did not give significant association with BMI although, for two of these, statistical power was limited because only 3 carriers were identified – this number of carriers permits only moderate significance (P = 7.45×110−3) even for the 593 kb deletions of chromosome 16p11.2 that are known to be strongly associated with morbid obesity [12], [13]. However, in contrast to the case-control replication analysis, this quantitative approach provided limited evidence to support involvement in BMI of 2 loci (in addition to the confirmed association with deletions in the SH2B1 region), albeit only at or near nominal significance insufficient to survive correction for multiple-testing. There was suggestive evidence for association with BMI of duplications near to the KIF2B gene on chromosome 17q22 (P = 0.103; mean BMI change = +2.3 kg.m−2 [–0.4 – +5.4]); and deletions at a second locus within the FOXP2 gene on chromosome 7q31.1 were nominally associated with reduced BMI (P = 0.0476; mean BMI change = –2.3 kg.m−2 [–4.4 – –0.03]). However, this latter effect was opposite to the increased risk of obesity originally reported [16].
