In this cross-sectional study, we noticed inverse associations between the lipid content of selected lipoproteins and adult bone turnover. An increase in the triglyceride or cholesterol content of VLDL and IDL particles was related to lower P1NP and CTX concentrations, especially in younger individuals. Similar effects were observed for the triglyceride content of LDL and HDL particles and for the Apo-A2 content of HDL particles. These effects were independent of the lipoprotein subclass. Associations between the lipid content of the lipoproteins and the ultrasound-based stiffness index were absent.

Our explorative study showed considerable inverse associations between the lipid content of the examined lipoproteins and both BTMs. Similar associations of P1NP and CTX were expectable, as in non-selected adults, bone formation and resorption processes are coupled [20]. Although with aging, bone resorption outweighs bone formation, a general increase in BTM concentrations is observed in women after menopause [20]. Elevated BTMs in turn, are associated with an increased fracture risk in postmenopausal women [21]. Our data revealed that both BTMs are inversely associated with the triglyceride and the cholesterol content of VLDL and IDL particles. These associations were consistent over the VLDL subclasses, between the sexes, present in younger and older individuals, and therefore of especial interest.

VLDL and IDL are triglyceride-rich particles. They transport triglycerides, but also cholesterol, to peripheral tissues such as muscles or adipose tissue, where they are hydrolyzed to release free fatty acids (FFA) for energy production or storage in intracellular lipid droplets [2, 3]. The observed inverse associations between the triglyceride and cholesterol content in VLDL and IDL particles with P1NP and CTX, point to a stable, possibly protective effect of the triglyceride-rich particles on bone turnover. Bone remodeling, which is crucial to maintain bone strength, is an energy-consuming process and an adequate supply with glucose and FFAs is essential for skeletal health [22]. Osteoblasts use fatty acid oxidation for energy production. Blocking this process results in a low-bone-mass phenotype in mice [23]. Further, FFAs indirectly impact on osteoclast development and function by activating nuclear transcription factors, as NFATc1 and NF-κB (reviewed in [24]). Hence, FFAs and triglycerides play an important role in bone homeostasis [25]. In line with this, Dragojevic et al. [26] reported that bone biopsies from fifty osteoporotic patients vs. fourteen healthy controls not only exhibited lower osteoblastogenesis, but also impaired triglyceride metabolism, characterized by an impaired fatty acid uptake and release from bone cells. Of note, the vast majority of the study participants had normal triglyceride concentrations (< 2.30 mmol/l, 88.1%). Thus, the positive association between triglycerides and bone turnover was largely restricted to the normal range.

Also cholesterol plays an important role in skeletal integrity by exerting multiple direct and indirect effects on bone cells. For example, it is a precursor for the synthesis of steroid hormones like estrogen, testosterone and vitamin D, all of which are critical for bone health. Furthermore, cholesterol represents an essential component of bone cell membranes [22]. Moreover, cholesterol modulates the osteoblastic differentiation of mesenchymal stem cells [27], plays a crucial role in signal transduction during osteoclastogenesis and increases osteoclast viability [28]. Potential protective effects of VLDL-1 particles have been reported from a community-based study, performed in 797 Chinese adults [29], which found that the chance of having osteoporosis decreased with increasing lipid content in VLDL-1 particles, including for example VLDL-1 triglycerides and cholesterol. Another study, restricted to female participants (483 pre- and 118 postmenopausal), reported positive associations of VLDLs with high vs. low BMD in postmenopausal women [30]. In our data, inverse associations between the lipid content of LDL- and HDL-particles and the BTMs were consistent for the triglycerides but largely absent for cholesterol. Only in the subgroup of young male participants, associations of LDL-cholesterol with the BTMs were observed. In the cholesterol-rich LDL- and HDL-particles, the correlation between the triglyceride and the cholesterol content was much lower (Pearson correlation coefficients mainly < 0.5) than in the VLDL and IDL particles (all > 0.7, Additional Fig. 1). It is thus conceivable that the associations with the BTMs observed here are dominated by the triglycerides. However, as functional data is absent, this hypothesis must remain speculative.

A steady and sufficient supply with triglycerides and cholesterol is necessary for bone health. While this is beyond dispute, the detrimental effects of dyslipidemia for bone homeostasis are also clear. Cross-sectional studies suggest that triglycerides are inversely associated with whole body-BMD [31]. In addition, hypercholesterolemia, increased LDL-cholesterol and elevated triglycerides are associated with an increased risk of osteoporosis [32, 33]. A Mendelian randomization study [34] further suggested that statins might positively affect BMD and reduce fracture risk. This effect was attributed to their LDL-lowering effect [34]. Several pathophysiological explanations have been presented that describe how dyslipidemia might affect bone homeostasis. All illustrate the detrimental effects of excessive or impaired lipid metabolism. For example, it has been observed in-vitro that cholesterol-treated mouse osteoblasts demonstrate an impaired proliferation and differentiation [35] and that an atherogenic diet with high cholesterol levels increases the expression of RANKL in mice, which enhances osteoclast differentiation and leads to bone loss [36]. A couple of recent reviews provide more details on the action of lipids on osteoblasts [8, 37], osteoclasts [38] or bone health in general [5, 39]. In addition to the direct effects mentioned above, dyslipidemia exerts several indirect effects on bone health, e.g. by its close relation to obesity. An accumulation of body weight and fat mass is, due to high mechanical loading, an effective stimulus for bone formation [40]. On the other side, obesity causes increased oxidative stress, systemic inflammation, insulin resistance and bone marrow adiposity. Pro-Inflammatory cytokines and adipocytes, as leptin or chemerin, are known to be upregulated in obesity and to inhibit bone physiology [41, 42]. The adipokine chemerin may for example cause a shift in the differentiation of mesenchymal stem cells and promote adipogenesis over osteoblastogenesis [43]. Independent of the specific mechanism, it must be noted that for functional bone remodeling a fine balance of enough but not excess supply with triglycerides and cholesterol is necessary. Our results, combined with the existing literature, suggest a U-shaped association between lipoproteins and bone turnover, with low and high levels adversely affecting bone metabolism. We therefore tested for the presence of non-linear associations between total cholesterol and total triglyceride concentrations and P1NP, CTX or stiffness index. For stiffness index, none of the models indicated a non-linear fit, and for cholesterol only one of the eight remaining tested models indicated a non-linear fit. The results were different for triglycerides (Additional Fig. 5), for which we found reverse J-shaped relations in younger men (P1NP and CTX), older men (CTX) and premenopausal women (P1NP and CTX). In detail, a decrease in triglycerides in the range of 0–2 mmol/l is related to a strong increase in BTMs, whereas in the range above 2 mmol/l BTM concentrations are rather stable. However, our data in this range are sparse and an interpretation is hardly possible. We therefore, cannot finally confirm or refute detrimental effects of elevated triglycerides on bone metabolism.

Despite the consistent effects of the lipids on both BTMs in our study, no associations with the stiffness index were observed. This may have different reasons. The single BTM and lipid measurements examined in our study represent an instantaneous picture of an individual’s bone and lipid metabolism. Yet, alterations of bone substance are subject to slow changes. Thus, effects of an impaired lipid metabolism on bone substance may only become apparent after several years of untreated dyslipidemia, but not in our cross-sectional analyses. Next to lipid metabolism, a multitude of factors affect bone turnover. Indeed, intake of various drugs, such as glucocorticoids, but also the individual medical history and lifestyle-related factors, such as low body weight, all have an impact on BMD [44]. We excluded individuals with intake of glucocorticoids, anti-osteoporotic or lipidlowering medication and adjusted our models for waist circumference, physical inactivity, diabetes mellitus and hsCRP concentration to appropriately consider these effects. But again, the long-term effects of individual medical conditions and lifestyle-related factors or their changes on BMD, could not be assessed in the present cross-sectional analyses. This also fits to our observation of fewer associations in older men and postmenopausal women than in youger men and premenopausal women. Indeed, the observation of more associations in younger than in older individuals was unexpected and explanations must remain speculative. In younger age bone turnover is more stable [20] and general health better than in older age. In this period of life, single risk factors may exert a more pronounced effect than in later life, when multiple risk factors simultaneously impact on bone metabolism. It is therefore possible that the effects of alterations in lipid metabolism on bone turnover are simply more visible here. Young and middle-age men and women should thus be aware of their bone health, including sufficient supply with triglycerides, to prevent disturbances in bone metabolism and deterioration of bone substance. With menopause, BTMs and the osteoporotic fracture risk strongly increase in women [44] and metabolic derangements accumulate. The assumed positive effects of a sufficient supply of bone cells with triglyceride or cholesterol may be exceeded by the detrimental effects of obesity. In fact, the average study participant was overweight, with one quarter of them being classified as obese.

Next to the absence of associations between the lipoprotein subclasses and the ultrasound-based bone stiffness index, our data did not show differential effects of the lipoprotein subclasses on bone turnover. Lipoprotein subclasses possess a distinct cardiovascular risk. For example, the atherogenity of small and dense LDL-particles is higher than that of large, less dense particles [9, 12, 45]. However, the lipoprotein subclasses had similar effects on bone turnover. This suggests that differentiating lipoprotein particles based on their density and size does not provide additional insights into its relation with bone turnover. Future studies might therefore examine other aspects of the relationship between lipid and bone metabolism, e.g. the effects of certain fatty acids. Long-chain polyunsaturated fatty acids, especially omega-3 fatty acids, are anti-inflammatory and thought to be beneficial for bone health. Also, saturated fatty acids and, most importantly, the right balance of the different fatty acid types are needed to maintain healthy bones (reviewed in [46]).

The work presented here stands out due to the large number of participants from the general population with 1H-NMR-based lipoprotein quantification. Indeed, our participants cover a large age range, which allowed us to perform analyses stratified by age or menopausal status. Furthermore, the intensive SHIP examinations with highly standardised procedures and validated interviews assured a high data quality and allowed us to carefully select our study population and to adjust our models for interfering covariates. Another strength of our study was the possibility to assess bone turnover and the ultrasound-based stiffness index simultaneously to obtain a detailed picture of the participants’ bone health.

Beside these strengths, our study has at least three limitations. First, our analyses were based on cross-sectional data, which prohibits drawing causative conclusions. Second, BMD measurements based on ionizing radiation represent the gold standard method for osteoporosis diagnostic, but were unavailable in the present population-based study for ethical reasons. However, the present ultrasound-based results provide a comparable fracture risk prediction [47]. Finally, while the SHIP-TREND sample is population-based we excluded selected study participants due to interfering medication intake or medical condition. Our results may therefore not apply to individuals with these criteria. Moreover, our study was performed exclusively in Caucasian participants and may not be directly transferrable to other ethnicities.