Genetic Studies On Osteoporosis With Linkage Exclusion Analyses And Artificial Neural Networks

It is very important and valuable to study the complex disease like osteoporosis, alcohol dependence, and diabetes because of their high prevalence, heavy injury, and huge expenses. Complex disease's etiology is very complicated which is related to many genetic and environmental factors and the interactions among genes and between genes and environment. Based on the complex system, the study on identifying and mapping the disease genes developed more slowly than that on monogenic disease. In current studies, some new approaches will be employed for the genetic research of osteoporosis. It is the first time to study the exclusion regions related to the phenotypes of bone mineral density and bone size with linkage exclusion analyses. It is also the first time to conduct the artificial neural networks to study the important candidate genes which may cause osteoporosis.Low bone mineral density (BMD) is a major determinant of osteoporosis and it is under strong genetic control. A large number of linkage and association studies for BMD variation have been conducted with the results being largely inconsistent. Linkage exclusion analysis is a useful tool for gene mapping but has never been used on BMD. In the present study, we conducted a linkage exclusion mapping for BMD variation on chromosomes 1, 4, 6 and 17 in 1816 individuals from 79 Caucasian pedigrees. BMD at spine (L1-L4) and at hip were measured with dual energy x-ray absorptiometry. Genotyping was analyzed with 37, 30, 19 and 23 micorsatellite markers on chromosomes 1, 4,6 and 17, respectively. The software SOLAR based on the variance components was employeed for multipoint linkage exclusion analyses. For hip BMD variation, several genomic regions were excluded for effect sizes of 10% or greater, including regions of 61-77cM at Ip34, 167-196cM at Iq21-q23, and 261-291cM at lq42-q44, 85-112cM at 4q21-q25 and 146-150cM at 4q31, and 77—85cM at 6pl2~ql3. For spine BMD, we were able to exclude the regions of 168-189cM at Iq21-q23, 92-94cM at 4q21 and 106-107cMat 4q24, and 56-103cM at 17ql2-q25, as having effect sizes of 10% or greater. These results suggest that a number of candidate genes located in the excluded regions, such as interleukin 6 receptor (IL6R) gene, type I collagen a 1 (C0L1A1) gene, and bone morphogenetic protein-3 (BMP3) gene are unlikely to have a substantial effect on BMD variation in this Caucasian population.Bone size is also an important risk factor of osteoporotic fractures and it has strong genetic determination. However, genetic studies on bone size variation are relatively rare. We conducted a linkage exclusion mapping for bone size variation on chromosomes 1, 4, 6 and 17 in 79 Caucasian pedigrees. For hip bone size variation, several genomic regions were excluded at effect sizes of 10% or greater, including regions of 61~77cM at Ip35-p34, 43-59cM at 4pl5-pl3, l-59cM at 6p25-p21, 82-88cM at 17q23-q24 and 113-114cM at 17q25. For spine bone size, at effect sizes of 10% or greater, we excluded regions of 115-122cM at Ip31-p22, 136-141cM at Ip21, 207-260cM at Iq31-q42, 20~89cM at 4pl6-q21, ll-21cM at 6p24-p23, and l~6cM at 17pl3. Theseresults suggested that a number of candidate genes located in the excluded regions, such as growth hormone (GH) gene, tumor necrosis factor alpha (TNFA) gene and bone morphogenetic protein-3 (BMP3) gene, are unlikely to have a substantial effect on bone size variation in this Caucasian population.Using BP artificial neural networks (ANN), the effects of 12 factors on BMD were analyzed. The results showed that sex was a significant factor to affect the BMD. Weight, eight and age had greater effect than genes. Among the genes, estrogen receptor a (ERa ) gene may play an important role in BMD variation, and bone gamma- corboxy-glutamic acid protein (BGP) gene may has limited effect on BMD. Artificial neural networks got similar results to traditional statistics, but ANNs had some advantages. For example, ANNs can analyze many genes' effect and their interactions simultaneously.Individuals with histories of alcohol dependence have a higher risk of osteoporosis. We found a large pedigree with alcohol dependence in Hunan. The consanguineous and drinking information were collected with questionnaire. The pedigree was large and stable, which is useful for the complex disease research. Though the drinking rate was very high, the heritability of alcohol dependence was just only about 10%. Female individuals had stronger effect on the analysis for heritability of alcohol dependence, which may be due to the lack of alcohol dependence individuals in female. The present research is a pilot work for molecular genetical studies and mechanism of osteoporosis induced by alcohol dependence infuture.In conclusion, some genomic regions and candidate genes were excluded to the variation of BMD and bone size. A modified neural network design can be well used for gene identification of complex disease in random population. We found a large pedigree with high drinking rate, high alcohol dependence rate and low heritability, and it will be helpful for the mechanism research of alcohol dependence and relationship between alcohol dependence and osteoporosis.