Recent investigations have focused upon understanding HDL as a heterogeneous particle with pleiotropic functions and suggest that the effects of HDL on CVD risk are not captured by HDL cholesterol content alone [7
]. HDL particles appear to exert anti-inflammatory effects [5
] and play a critical role in cholesterol efflux from macrophages that reside within the artery wall [20
]. In the present study, we demonstrate that acute sex steroid withdrawal increases the serum cholesterol efflux capacity in young, healthy men. Further, we found that the enhanced efflux capacity is specifically attributable to androgen rather than estradiol deprivation. Sex steroid deprivation also confers changes in the protein composition of HDL, changes that could directly impact cholesterol efflux or other HDL effector functions. These findings underscore the complexity of the relationship between HDL and sex steroid exposure.
The modulatory effects of androgens on CVD risk remain poorly understood. Historically, testosterone has been presumed to augment risk on the basis of its HDL-C lowering effect and the earlier onset of CVD in men relative to women. Yet recent prospective epidemiologic data now suggest the converse, as low endogenous circulating androgens predict CVD and mortality risk in men [2
]. Similarly, men who have undergone ADT for the treatment of prostate cancer exhibit a disproportionately increased incidence of CVD compared to age-matched controls [21
]. These inconsistent findings in part may derive from comparison of interventional and observational data, as the latter particularly may be confounded by the presence of co-morbid conditions, such as obesity and chronic illness, that independently can alter testosterone and HDL-C concentrations and overall CVD risk [1
]. Alternatively, the apparent discrepancy may result from the heterogeneous biological processes that underlie atherosclerosis, many of which may be modulated by sex steroids. Further, despite the consistent, negative correlation between endogenous HDL-C concentrations and CVD risk, it remains highly uncertain whether changes in HDL-C due to clinical interventions translate into modified risk [4
Many lines of evidence support the proposal that a primary atheroprotective role of HDL is the mediation of reverse cholesterol transport, the process by which HDL unloads cholesterol from peripheral tissues and transports it to the liver [24
]. Animal studies provide compelling evidence that one key mechanism involves the removal of cholesterol from lipid-laden macrophages, which are of central importance in the pathogenesis of atherosclerosis [26
]. Rader, Rothblat, and colleagues have recently developed an assay to determine the capacity of serum to accept cholesterol from macrophages [(7
)]. As this assay measures the efflux capacity of apoB-depleted serum, it has been used as a surrogate for HDL-mediated function. These investigators demonstrated that low cholesterol efflux capacity as measured by this assay predicted the presence of coronary artery disease in a clinical cohort [7
]. Moreover, the relationship between cholesterol efflux and coronary artery disease persisted after correction for HDL-C, suggesting that this metric of HDL function might prove a key indicator of CVD risk. Indeed, efflux capacity was a better predictor of CVD status than was HDL-C [7
]. These observations raise the question of whether HDL-C alone is an adequate assessment of HDL-associated CVD risk. These findings also underscore the potential disconnect between the cholesterol content and functional significance of HDL.
While our findings suggest that androgen withdrawal increases cholesterol efflux from macrophages, prior work in vitro
has suggested that androgen supplementation with physiologic and supraphysiologic doses of testosterone might augment HDL3
-mediated cholesterol efflux from macrophages [28
]. Although it is possible that both sex steroid withdrawal and replacement produce similar changes in efflux, these apparently discrepant results instead may be attributable to the different experimental methods employed. In our study, we manipulated sex steroids in men for 4 weeks, allowing time for in vivo
modifications of HDL to occur, and then examined functional capacity of total HDL. In contrast, previous analyses applied sex steroids in vitro
and only examined efflux mediated by the small, dense HDL3
subfraction. Notably, these investigators found that testosterone treatment increased hepatocyte and macrophage expression of scavenger receptor B1 (SRB1), a receptor implicated in HDL uptake in the liver [28
]; accordingly, the reduction in HDL-C often observed with testosterone treatment potentially could result from accelerated hepatic uptake of HDL-derived cholesterol and paradoxically predict reduced atherosclerotic risk. Of note, the assay employed in our study quantifies cholesterol efflux predominantly mediated by the ABCA1 transporter; future studies are warranted to determine whether sex steroids exert differential effects on pathways mediated by other cholesterol transport proteins in vivo
Previous models also have demonstrated the in vivo
effects of sex steroids on HDL-C, HDL-associated proteins, and cholesterol efflux capacity. The synthetic steroid tibolone decreased whereas conjugated equine estrogens increased serum efflux capacity in a primate model [29
]; the effect of tibolone may be dose-dependent, as decreases in cholesterol efflux have been observed only at higher doses in association with substantial reductions in HDL-C [29
]. Acute sex steroid suppression in men further has been shown to exert effects on serum concentrations of HDL-C and HDL-associated proteins [31
]. A novel aspect of our study is inclusion of a treatment group rendered selectively estrogen-deficient to discriminate between androgen- and estrogen-mediated effects on HDL-C and cholesterol efflux. In addition, we examined cholesterol efflux using a macrophage-based assay that strongly associates with coronary artery disease [7
], but larger trials with purely HDL-specific assays are needed to fully define the impact of androgens on HDL-mediated cholesterol efflux. Importantly, too, no data have yet demonstrated that modifying HDL efflux capacity alters CVD risk. Thus, intervention trials with long-term follow-up are required to understand the clinical implications of our findings.
In contrast to anticipated findings, selective estradiol deprivation (Group 3) did not confer changes in HDL-C. In a previous study with a similar design, estradiol deprivation was associated with a decrease in HDL-C [34
], and supraphysiologic, exogenous estradiol administration to men undergoing ADT increased HDL-C concentrations [35
]. The fact that we did not observe changes in HDL-C with selective estradiol withdrawal may be a function of our small sample size or may reflect the lesser degree of estradiol suppression achieved in our study compared to previous studies using aromatase inhibitors [34
The functional capacity of HDL appears to be determined in part by its protein composition, and increasing evidence suggests that the protein constituents are modifiable and associate with CVD risk [6
]. We therefore investigated whether sex steroid withdrawal modulates the HDL proteome. After only 4 weeks of sex steroid deprivation, we found significant increases in HDL-associated clusterin and apoA-IV, though only the change in clusterin remained significant after correction for multiple comparisons. Consistent with these findings, clusterin is an androgen-responsive target gene in the prostate also known as testosterone-repressed prostate message-2 (TRPM-2), and elevations in prostatic clusterin have been observed in men after ADT [37
]. Interestingly, low levels of HDL-associated clusterin have been correlated with CVD risk factors in patients with metabolic syndrome [8
]. We also observed a partially sustained increase in HDL-clusterin after restoration of endogenous sex steroids, suggesting that even transient changes in sex steroid exposure might confer more enduring effects on HDL protein composition. Finally, the observed increase was evident in HDL-associated but not serum clusterin and therefore implicates sex steroids in the specific modulation of HDL protein composition.
Although the in vivo
significance of apoA-IV remains uncertain, in vitro
data suggest it may play a role as a regulator of cholesterol ester transport protein (CETP), phospholipid transfer protein (PLTP), and lecithin-cholesterol acyltransferase (LCAT) [38
], enzymes that modify the lipid, phospholipid, and protein composition of HDL. HDL isolated from patients with CVD exhibited significant enrichment in apoA-IV compared to HDL from healthy controls [9
], though overexpression of apoA-IV significantly attenuated atherosclerosis in a transgenic mouse model [40
]. Rodent models have demonstrated sex steroid regulation of apoA-IV, but these findings pertained to mRNA expression rather than HDL-associated protein [41
]. Although the increase in apoA-IV did not remain statistically significant after correction for multiple comparisons, the relationship between sex steroid exposure and HDL-associated apoA-IV merits further study in larger trials.
As the HDL proteome was evaluated only in Group 1 subjects, additional investigation is needed to determine whether androgens or estrogens mediate the observed effects on HDL protein composition and the impact of longer-term changes in sex steroids on the HDL proteome. Larger-scale studies eventually might be able to identify composite protein signatures, analogous to a genetic haplotype, that correspond to changes in HDL function.