The first explanation brought forward for failure of torcetrapib was its off-target effects. Indeed, patients in the experimental arm of ILLUMINATE study had slightly elevated blood pressure (5). This phenomenon was reproduced in rats, a species that does not express CETP, thus it was clearly an off-target effect (9). However, in addition to a 25% increase in mortality that was observed, this putative off-target effect should also account for lack of 25% decrease in mortality that should have occurred due to lowering of LDL-C level, which did not eventuate, and a 75% decrease in mortality due to elevation of HDL-C, which also did not eventuate. Thus, an off-target effect(s) should have been responsible for about a 125% increase of mortality to offset the expected beneficial effects of treatment if the underlying hypothesis was correct; clearly, a highly unlikely possibility. Finally, another CETP inhibitor, dalcetrapib, did not cause blood pressure elevation, but still did not reduce cardiovascular mortality (6). Overall, given excellent safety record of torcetrapib (10), it is unlikely that failure of CETP inhibitors to improve cardiovascular outcomes was due to off-target effects.

Another explanation was that CETP inhibitors elevate plasma level of HDL-C, but impair functionality of HDL. Inhibition of CETP precludes exchange of cholesteryl esters (CE) and triglycerides (TG) between HDL and apoB-containing lipoproteins (mainly Very Low Density Lipoprotein, VLDL) resulting in reduced catabolism of HDL and accumulation of larger TG-rich HDL particles (11). HDL was implicated in many anti-atherogenic activities, and changes in HDL composition or size may affect some aspects of HDL functionality, at least in vitro. However, when tested in animals (12) and humans (13–15), this suggestion was not confirmed: HDL functionality, at least toward its two most important functions, reverse cholesterol transport (RCT) and anti-inflammatory properties, was unaffected by inhibition of CETP. Anti-oxidation function of HDL was also unaffected in subjects with CETP deficiency (16). Overall, no evidence was produced to support a hypothesis that failure of CETP inhibitors to improve cardiovascular outcomes was due to impaired HDL functionality.

Initial steps of RCT, cholesterol efflux, formation of nascent HDL particles, and early steps of HDL remodeling are very similar in humans and animals. However, late steps of RCT are different: delivery of CE to the liver may occur via direct RCT pathway, selective uptake of CE from HDL by hepatocyte through HDL receptor SR-B1, or indirect RCT pathway, after transfer of CE to VLDL/LDL and uptake of LDL by hepatocyte LDL receptor. The ratio between these two pathways is mainly driven by the activity of CETP and differs between humans and animals.

Preclinical studies supporting beneficial effects of CETP inhibitors were done in rabbits, a species that, like humans, expresses CETP and is susceptible to diet-induced atherosclerosis (17). HDL metabolism in rabbits, however, is different from that in humans (18). In this regard, it is interesting to compare the effects of CETP deficiency/inhibition in species with different CETP activities (12). Mice do not express CETP, but CETP may be introduced in mice systemically following adenovirus-mediated transfection. Moderate overexpression of CETP in mice did not affect HDL-C or LDL-C levels, but increased the rate of RCT; the effect was blocked by torcetrapib (12). The explanation is likely that supplementation of natural direct pathway with additional indirect pathway resulted in overall higher rates of RCT, and CETP inhibition in this context had an adverse effect on the overall rate of RCT (Figure 1A). Hamsters express CETP, but activity is less than in humans and HDL-C levels are fivefold higher than in humans. Inhibition of CETP in this context resulted in further increase in plasma HDL-C levels and increased the overall rate of RCT (12). The explanation is likely that inhibition of relatively minor indirect pathway and stimulation (through higher HDL-C levels) of the predominant direct pathway resulted in overall faster RCT (Figure 1B). Apparently, in hamsters, liver had sufficient levels of SR-B1 to handle modest amount of additional cholesterol delivered through direct pathway. In humans, CETP activity is high, HDL-C levels are low, and indirect pathway accounts for almost 90% of RCT traffic (19). CETP inhibition would block indirect RCT pathway and increase direct RCT pathway, however, the benefits would depend on a balance between the two and on how much of additional cholesterol flow through direct pathway can be handled by liver (Figure 1C). Does human liver have, or can attain sufficient levels of CLA-1 (human analog of SR-B1) to process sharply increased flow through direct RCT pathway or it gets saturated? Several studies investigated the effect of torcetrapib on fecal cholesterol excretion, the endpoint of RCT, and found no change. In rabbits, CETP activity is triple that of humans and HDL-C level is 15% of that in humans (18). The rate of RCT in rabbits and the effects of CETP inhibitors on this rate were not studied, but the effects on development atherosclerosis were beneficial (17) implying that RCT rate was also elevated (Figure 1D). Interestingly, although torcetrapib caused elevation of HDL-C in rabbits, it was still about half of that in normolipidemic humans and less than a third of what was observed in patients treated with torcetrapib (5), in addition there was no reduction in LDL-C. Perhaps, HDL levels in rabbits, in contrast to humans, are still below saturation of the direct pathway, and rabbits, but not humans, can handle a switch from indirect to direct RCT. One conclusion is, however, clear: the overall effect of CETP inhibition is a balance between inhibition of the indirect pathway and stimulation of the direct pathway of RCT and this balance is greatly influenced by metabolic context.

It is important to recognize that in all clinical trials of CETP inhibitors, patients in both arms of the study were receiving statins. Statins inhibit cholesterol biosynthesis in liver causing sharp elevation of the expression of hepatocyte LDL receptors leading to increased catabolic rate of LDL and lowering plasma LDL-C level. Uptake of LDL, however, is also a key step of indirect RCT pathway delivering to liver HDL-derived cholesterol after its transfer to VLDL/LDL. By increasing LDL uptake, statins would stimulate indirect RCT pathway, the very same pathway that is inhibited by CETP inhibitors (Figure 1C). Thus, statins and CETP inhibition have opposing effects on indirect RCT pathway. Furthermore, VLDL/LDL plays an important role in the initial stages of RCT. Most of cellular free cholesterol taken up by HDL ends up in VLDL/LDL prior to esterification, as VLDL/LDL provides a pool with much bigger capacity to hold free cholesterol before relatively slow process of esterification converts cholesterol into esters (20). This finding gave rise to the “shuttle” and “sink” hypothesis (21) also supported by findings that cholesterol efflux to plasma devoid of apoB-containing lipoproteins is much slower and that the availability of apoB-containing lipoproteins may become rate limiting in RCT when HDL-C level are significantly increased (22). Interestingly, humans are the only species where CETP inhibition caused reduction of VLDL/LDL levels (Figure 1), which could also contribute to reductions of overall rate of RCT. In this context, combining CETP inhibitors and statins may have detrimental effect on the actions of both drugs.