In this study, a genome-wide association analysis was conducted to identify the pathway(s) involved in BMD determination at a clinically important skeletal site—wrist UD. The ROA pathway was shown to be significantly associated with wrist UD BMD (p = .005) and was the only pathway, among a total of 963 pathways tested, that achieved statistically significant FDR and FWER values. To further confirm our findings in the discovery cohort, 2187 subjects from the FHS with arm BMD data available were studied as a replication study. The ROA pathway showed significant association with arm BMD in the FHS sample even after the multiple-testing adjustment, which further confirmed our findings in the discovery cohort. This is the first time that the autophagy-related biologic process has been implicated as an underlying factor for wrist BMD variation and hence a risk factor for wrist osteoporosis.
Autophagy is a conserved housekeeping function of eukaryotic cells that permits sequestration of cytoplasmic components in membrane-bound vesicles and delivers them to lysosomes for degradation. Autophagy is subject to suppression or further induction in response to different stresses, starvation, specific hormonal regulation, and other stimuli.(24
) The core process of autophagy and the functional interactions among the genes in the ROA pathway are depicted in . The autophagy pathway has numerous proposed physiologic functions. Growing evidence suggests that malfunctions of autophagy contribute to many human diseases, for example, myopathies, liver disease, neurodegeneration, heart disease, and cancer.(23
Fig. 3 Autophagy and the functional interactions among the genes in the ROA pathway. Basically, the autophagic process can be divided into three stages: vesicle nucleation, vesicle elongation and completion, and vesicle breakdown and degradation. ATG1, ULK, (more ...)
Although this article is the first to support the ROA pathway's importance to osteoporosis, the pathway's relevance to bone metabolism has been suggested in previous studies. Autophagy has been recognized as an intermediate stage during the terminal differentiation of chondrocytes. Specifically, during endochondral ossification, induction of autophagy occurs prior to apoptotic hypertrophic maturation in terminally differentiated chondrocytes.(27
) These chondrocytes morphologically exhibit autophagic characteristics and express autophagic proteins.(27
) The induction of autophagy enhances the survival of chondrocytes in hypoxic microenvironments (eg, low protein, glucose, and O2
concentrations) and facilitates synthesis of the calcified extracellular matrix.(30
Some genes in the ROA pathway also have been considered as important modulating factors for bone development or remodeling. For example, IFN-α can suppress the proliferation of osteoprogenitor cells(33
) and modify the expression of a number of important cytokines that are regulators for human osteoprogenitor cell growth and differentiation.(35
) IFN- α also can inhibit the differentiation of peripheral blood mononuclear cells (PBMCs) to osteoclasts.(38
) Interferon-γ (IFN-γ) is considered to be an antagonist of RANKL and is a well-known potent inhibitor of osteoclast function and formation.(39
) IFN-γ also exhibits antiproliferative actions on primary osteoblast cells.(43
) Finally, it has been reported that in cultured osteoblasts, AMPK influences the expression of cyclooxygenase 2 (COX-2), which potentially could impact fracture healing because COX-2 is a crucial mediator in mechanically induced bone formation.(46
In addition to highlighting the potential contribution of the ROA pathway to human wrist BMD deviation, this study also demonstrates the important advantage of pathway-based GWA analysis. None of the genes in the ROA pathway reached a significant level when considered separately; thus the potential impact of the ROA pathway on wrist BMD would not have been detected by individual gene/SNP GWA analyses. By considering critical information about the interaction of a set of functionally related genes and their joint effects, pathway analysis may be a useful paradigm for revealing the polygenic nature of complex diseases. Recently, some candidate pathway association studies have been done, suggesting that the joint actions of common gene variants within pathways may play a major role in predisposing to complex diseases. Examples include the axon guidance pathways for Parkinson disease (PD)(47
) and amyotrophic lateral sclerosis (ALS).(48
) Differing from the candidate pathway strategies,(47
) in the current genome-wide analysis we systematically studied a large number of pathways annotated by public pathway databases with the aim of identifying pathways associated with UD BMD. This hypothesis-free strategy could make full use of GWA data and would be more useful for discovering new disease-related processes and generating novel hypotheses about pathogenic mechanisms of disease.
It would be interesting to clarify the detailed mechanism of the ROA pathway on wrist BMD. Considering the existing evidence linking some ROA pathway genes with osteoblastogenesis or osteoclastogenesis, it can be imagined that the involvement of autophagy in bone may not be limited to chrondrocytes alone. Since autophagy is another mechanism for programmed cell death and shows close interaction with apoptosis, whether antophagy is involved in the survival control of osteoblasts/osteoclasts/osteocytes is an important hypothesis to test. Also, it will be interesting to study whether some lifestyle factors will interact with the ROA pathway or if the ROA pathway would be a candidate target for osteoporosis therapy.
In this study we found that the ROA pathway is associated with human wrist BMD variation. However, we cannot exclude other pathways for their significance in wrist BMD variation, even though these pathways did not pass the significance criteria adopted in this study. Regarding the statistical aspect of this study, it should be noted that some genes inevitably are shared by different pathways. Although the overlap of genes among different pathways will not affect the pathways' relative ranking in terms of NES values, interdependence between pathways will lead to a decreased power by affecting FDR and FWER when the causal genes are shared by multiple pathways (owing to the permutation procedure for multiple-testing adjustment). A larger sample size or more specifically annotated pathways will be helpful to improve the power of this pathway-based GWAS.
It also should be noted that under the significance criteria adopted in this study, the ROA pathway was not detected to be associated with hip or spine BMD in the discovery cohort (data not shown). The possible reasons for the inconsistent association signals detected at these three skeletal sites may include statistical errors (type I error at the wrist or type II error at the hip/spine) and genetic heterogeneity between different skeletal sites. In particular, type II error at the hip/spine may be the major reason. The type II error may be caused by the relatively low statistical power of this newly developed pathway-based method (as discussed earlier). This low statistical power, when complicated by the possible genetic heterogeneity (ie, smaller effects of the ROA pathway on hip and spine than on wrist), may lead to negative association signals at the hip and spine.
In summary, we applied a novel pathway-based GWA analysis method to systematically screen functional pathways/gene sets associated with wrist UD BMD and thus osteoporosis. The significant enrichment of ROA pathway genes among the top-ranking genes associated with wrist UD BMD, together with the pathway's functional relevance to bone metabolism, strongly supports an important role of autophagy in human wrist BMD variation. Further detailed and specific functional studies of the ROA pathway will provide new insights into the pathway's relevance to the physiology of bone and the etiology of wrist osteoporosis.