Hydroxymethylglutaryl-coenzyme A inhibitors (HMG-CoA inhibitors, or statins) were initially identified as fungal extracts in 1976 (1). They were subsequently developed as cholesterol-lowering drugs, and have been shown in numerous clinical trials to reduce both cardiovascular morbidity and mortality (2–6). However, studies also revealed that statins yield a larger mortality benefit than can be readily explained by their cholesterol-lowering effects alone, since their benefits occur too quickly to be explained by effects on atherosclerotic plaque (7), and are greater than lipid-lowering alone would predict (8). Additionally, statins appear to have beneficial effects on human diseases, such as multiple sclerosis (MS) (9) and osteoporosis (10), that have no direct association with cholesterol levels.
These findings, together with growing awareness that atherosclerosis is itself an inflammatory disease, led to the suggestion that, in addition to lowering cholesterol, statins modify atherosclerosis via antiinflammatory mechanisms. A broader hypothesis followed naturally: that statins might have general antiinflammatory and/or immunomodulatory effects. Research over the last 10 years has elucidated a number of mechanisms by which statins may exert antiinflammatory effects (7, 11). In this article, we review the mechanisms of the action of statins, and the specific signaling pathways they modulate. We summarize the known potential antiinflammatory and immunomodulatory effects of statins at the cellular level. Finally, we review the data on the possible efficacy of statins in autoimmune/inflammatory diseases in both animal models and human trials.
Biochemical effects of statins
Statin-sensitive biosynthetic pathways.
Statins inhibit the rate-limiting step of cholesterol synthesis by preventing HMG-CoA from being reduced to mevalonate via HMG-CoA reductase (Figure 1). Mevalonate is the necessary substrate not only for cholesterol biosynthesis, but also for the synthesis of several other biologically important lipid intermediates by means of alternative synthetic pathways; statins also have the potential to inhibit synthesis of these products. Two such intermediates are the 15-carbon isoprenoid farnesyl pyrophosphate (FPP) and the 20-carbon isoprenoid geranylgeranyl pyrophosphate (GGPP); these serve as lipid attachments (via the activity of farnesyltransferase and geranylgeranyl transferase [GGT], respectively) required for proper localization and activation of a variety of proteins, including monomeric GTPases (12, 13).
GTPases, so-named for their intrinsic autolytic GTPase activity, are intracellular switches that play essential roles in numerous cellular processes, including gene expression, actin cytoskeleton regulation, membrane trafficking, proliferation, apoptosis, and migration (14, 15). The GTPase Rho, and Rho-like proteins, regulate adhesion complex formation, as well as a number of inflammatory pathways, such as the JNK and p38 MAP kinase cascades (16, 17). Experimental evidence suggests that statins may inhibit inflammation predominantly by inhibiting Rho family protein activity (18). Other GTPases that may be inhibited by statins include Ras family proteins, which primarily transduce signals from growth factor receptors and regulate the MAP kinase, ERK. Whereas Rho proteins (along with Rab proteins, which regulate vesicular trafficking) are typically geranylgeranylated, Ras proteins (along with heterotrimeric G proteins and nuclear lamins) are farnesylated (14).
Experimental manipulation of statin-sensitive pathways.
It is possible to elucidate how statins exert their antiinflammatory effects by manipulating specific steps in the cholesterol and isoprenoid synthesis pathways. In vivo, it is important to differentiate effects caused by decreased isoprenoid levels from those caused by lowering cholesterol. To this end, a few options are available. Most directly, cholesterol can be measured in statin-treated and untreated conditions. If no difference in cholesterol exists between the conditions, then an antiinflammatory effect cannot be attributed to cholesterol lowering. Because statins do not lower cholesterol levels in mice, all of the antiinflammatory effects of statins in mouse models of disease are, of necessity, cholesterol independent (19). Differentiating a cholesterol-lowering effect from an isoprenoid-lowering effect can also be accomplished (in vivo or in vitro) by using a specific cholesterol synthesis inhibitor such as squalestatin, which acts downstream of isoprenoid intermediate synthesis, and therefore affects cholesterol synthesis exclusively (20).
While statin effects in vitro are, of necessity, independent of effects on serum cholesterol levels, statins may still alter cell responses by lowering the intrinsic cholesterol content of cell membranes. In particular, cholesterol in cell membranes tends to be concentrated in lipid rafts, focal patches of membrane that are enriched in protein signaling assemblies. A number of these assemblies have been shown to regulate inflammatory responses. For example, T cell activation of the IκB kinase complex, which leads to activation of NF-κB, depends on lipid rafts (21). Lowering cholesterol may disrupt these rafts and alter the inflammation signaling properties of the cells (22).
Similarly, the membranes of phagocytic vacuoles are enriched with cholesterol in leukocytes such as neutrophils (23). Whether statins alter vacuolar membranes, with effects on phagocytosis, inflammation, and/or antigen processing, remains to be determined. Changes in cholesterol levels may also directly alter intracellular signaling molecules via membrane-independent effects. For instance, the activity of oxysterol-binding protein, a scaffolding protein that regulates ERK-1/2 activity, is directly regulated by interaction with cholesterol (24). Failure of the selective cholesterol synthesis inhibitor squalestatin to reproduce the effects of statins on any particular inflammatory phenotype suggests that the statin effect under study is not mediated by alterations in the cholesterol content of the cell.
Specific farnesyltransferase inhibitors (FTIs) prevent the attachment of FPP to Ras and other farnesylated proteins, resulting in nascent unfarnesylated Ras isoforms (H-Ras, N-Ras, K-Ras) (25). Similarly, inhibition of GGT using specific GGT inhibitors (GGTIs) prevents Rho protein (RhoA, RhoB, Cdc42, Rac-1, and others) geranylgeranylation (12). Unprenylated Ras and Rho proteins mislocalize to the cytosol and are generally inactive. Thus, FTIs and GGTIs replicate specific, but not global, effects of statins. By comparing the in vitro and/or in vivo effect of FTIs or GGTIs with the effect of statins, one can assess which subset of prenylated proteins (farnesylated or geranylgeranylated) may be responsible for a given statin effect.
Alternatively, one can test whether a specific statin effect can be overcome by repletion with exogenous FPP or GGPP. If a statin-mediated effect is duplicated by inhibiting transfer of a particular isoprenoid intermediate, and rescued by exogenous supplementation of that intermediate, it is likely that the statin effect is caused by decreased levels of that isoprenoid (and, in turn, lower levels of activated prenylated proteins). A cruder method of substrate rescue often used is rescue with mevalonate, the cholesterol intermediate just downstream of HMG-CoA (further elaborated below). Since mevalonate synthesis is localized upstream of the branch point between cholesterol and isoprenoid synthesis, when mevalonate rescue successfully reverses a statin effect, nothing is learned about which mevalonate product may be responsible for that effect.
A more specific determination of which prenylated protein may be the target of a given statin effect is possible by direct measurement of the activated protein (infrequently reported in the statin literature) or by pharmacologic or genetic inhibition or stimulation of a specific GTPase (7).
Statins as antiinflammatory drugs: effects on cells and tissues
Numerous studies have confirmed that statins have a wide range of effects on cells and tissues involved in inflammation and/or autoimmunity.
Inhibition of leukocyte–endothelial adhesion.
Statins have been reported to inhibit interactions between leukocytes and endothelial cells (ECs) that necessarily precede leukocyte egress from the vasculature (26–28). Different groups of adhesion molecules sequentially mediate leukocyte rolling, adhesion, and diapedesis, or transmigration through the vascular wall (29). The first step, leukocyte rolling, occurs by the interaction of selectins on one cell with sialyl-Lewis residues on a cognate cell (29). The next step, leukocyte-tight adhesion, occurs through the interaction of leukocyte integrins (e.g., lymphocyte function–associated antigen 1 [LFA-1], Mac-1 [CD11b/CD18]) with counterligands on ECs (e.g., intercellular adhesion molecule 1 [ICAM-1]) (29). Finally, transmigration of leukocytes through the endothelium and the vascular wall is mediated by chemokines, such as monocyte chemotactic protein 1 (MCP-1) (30).
Initial studies showed that statins down-regulate expression of adhesion molecules on ECs and leukocytes; these studies have been described in previous reviews (8, 31). However, subsequent in vitro studies demonstrated that statins actually increase the expression of these molecules in inflammatory settings, suggesting that statin effects may be context dependent (32–35). Perhaps of greater significance is that a number of ex vivo human studies showed that statins down-regulate the soluble concentration of these molecules. Shed (soluble) adhesion molecules are well-known markers for atherosclerotic disease (36–38), and several studies indicated that statins down-regulate circulating soluble ICAM-1, vascular cell adhesion molecule 1 (39–41), and E- and P-selectin (42, 43). However, not all studies that examined the effects of statins on the soluble levels of these molecules have replicated these findings (44, 45).
While the effects of statins on adhesion molecules in vitro may be inconsistent, statins consistently inhibit endothelial–leukocyte adhesion in complex models. For example, statins decrease leukocyte–endothelial adhesion in vitro in experiments performed under conditions of physiologic flow. Intravital microscopy, which allows for real-time in vivo examination of study animals, confirms that statins inhibit leukocyte–EC interaction in post-mesenteric venules of rats (27, 46, 47). Statin inhibition of leukocyte–endothelial adhesion has also been directly observed using in vivo confocal microscopy (48).
A number of direct and indirect mechanisms may account for the observed inhibitory effect of statins on intercellular adhesion. Importantly, statins inhibit the formation of focal adhesion complexes (FACs) in human ECs. Because FACs represent the tether points and signaling foci for transmembrane adhesion molecules, subsequent disruption of adhesion should be unsurprising (18). This effect is likely due to inhibition of several members of the Rho family that work together to form these complexes (49). The in vitro inhibition of FACs by statins is reversed with the addition of mevalonate (18).
A novel way in which statins may regulate leukocyte–endothelial adhesion is by directly binding to a novel regulatory site, the L-site (named for lovastatin), on the integrin LFA-1 found on leukocytes (50). LFA-1, a β2 integrin, is critical in the development of inflammatory arthritis in the K/BxN serum transfer mouse model of arthritis. Ablating LFA-1, either by using a CD11a-null mouse or by coadministration of K/BxN serum with monoclonal antibodies to LFA-1, prevents induction of (and ameliorates established) disease (51). When statins (excepting pravastatin, which does not interact with the L-site) bind to LFA-1, they effect allosteric changes in the integrin that prevent ICAM-1 binding. This effect is both cholesterol- and isoprenoid-independent, and may interfere not only with endothelial–leukocyte interaction, but also with T cell activation, since LFA-1 is a weak T cell costimulator. Since this novel docking site on LFA-1 was discovered, statin derivatives with much higher affinity for the L-site have been developed, such as LFA-878, which was recently shown to be effective in a rat model of inflammation (52). LFA-878 possesses no activity as an HMG-CoA reductase inhibitor.
Statins inhibit the in vitro and in vivo production of MCP-1, a major chemoattractant for monocytes and T lymphocytes, and, as noted earlier, a signal for leukocyte diapedesis. Simvastatin inhibits the production of MCP-1 by human ECs stimulated by C-reactive protein (CRP), interleukin-1β (IL-1β), or lipopolysaccharide (LPS) in vitro (53), and both lovastatin and simvastatin decrease MCP-1 production in human peripheral monocytes (54, 55). These effects are reversed with the addition of mevalonate (54). Withdrawal of cerivastatin from pretreated vascular smooth muscle cells induces MCP-1 production, an effect that is replicated when these cells are coincubated with cerivastatin plus GGPP, which suggests that it occurs via geranylgeranylated proteins (56). Rosuvastatin diminishes MCP-1 production in the vessel walls of hypercholesterolemic mice (54), and atorvastatin and pravastatin decrease MCP-1 expression in vessel walls of hypercholesterolemic swine (57). Fluvastatin, atorvastatin, cerivastatin, and simvastatin significantly decrease circulating MCP-1 levels in patients with hypercholesterolemia (58–61). Statins also decrease levels of the chemokine IL-8 (59, 62), another important regulator of leukocyte adhesion and chemoattraction. Statin effects on these and other soluble mediators may therefore indirectly alter the conditions for leukocytes to migrate to sites of inflammation.
Effects on endothelial cell nitric oxide synthase (eNOS), inducible NOS (iNOS), and oxidative products.
Endothelial cell NOS is expressed by vascular endothelium and generates NO, which exerts protective effects on ECs and prevents their activation (63, 64). Statins up-regulate eNOS expression in vitro (62, 65, 66) and in vivo (67) by prolonging eNOS messenger RNA (mRNA) survival (68). In vitro, cotreatment of statin-treated ECs with either mevalonate or GGPP reverses this effect, suggesting the involvement of one or more Rho proteins (69, 70). Direct measurement of Rho activity in mevastatin-treated ECs shows that Rho inhibition correlates with statin-induced eNOS up-regulation, and that Rho inhibition was rescued by GGPP but not by FPP (69). In support of a role for Rho in down-regulating eNOS expression, direct inhibition of Rho with Clostridium botulinum C3 transferase or by transfection of a dominant-negative (DN) RhoA mutant increases eNOS expression. Conversely, activation of Rho by Escherichia coli cytotoxic necrotizing factor 1 decreases eNOS expression. Thus, statins appear to increase eNOS levels and eNOS activity by down-regulating a pathway that involves Rho proteins (69).
In contrast to eNOS-derived NO, iNOS-derived NO is frequently used by cells as a proinflammatory signal; statins inhibit the induction of iNOS in several cell types. Lovastatin and mevastatin were shown to inhibit LPS-induced iNOS expression and activity in rat macrophages and microglia (71). That initial study led investigators to study the central nervous system–protective effects of statins, because iNOS is known to mediate neuronal toxicity. Pitavastatin prevents ischemia-induced brain injury in a rodent model, largely by preventing iNOS expression while maintaining eNOS expression (72). Similar results were seen with atorvastatin in a rat spinal injury model (73). Simvastatin decreases iNOS induction in embryonic cardiac myoblasts stimulated with IL-1β or tumor necrosis factor α (TNFα), an effect that is reversed by GGPP and duplicated with a Rho kinase inhibitor (74). Lovastatin, atorvastatin, and fluvastatin decrease both LPS- and interferon-γ (IFNγ)–stimulated iNOS expression in murine macrophages by reducing iNOS transcription, an effect that is reversed by mevalonate, GGPP, or FPP (75). Atorvastatin, cerivastatin, and pravastatin all decrease TNFα/IFNγ-stimulated iNOS expression in mouse aortic endothelium (76).
In addition to effects on NO production, statins inhibit formation of oxygen radicals by ECs (77, 78). Two mechanisms appear responsible for this observed effect. First, statins prevent activation of the NADPH oxidase complex, which generates superoxide, apparently due to their inhibitory effect on the Rho family GTPase Rac-1 (77, 79). Second, statins inhibit angiotensin-induced NADPH oxidase activation by down-regulating the concentration of the angiotensin II type 1 receptor (79).
Inflammatory cytokines and other secreted mediators: regulation of transcription.
Statins reduce the production of a number of inflammatory cytokines. Simvastatin and fluvastatin decrease IL-6 and IL-1β production by stimulated human umbilical vein endothelial cells (HUVECs); this is reversed with the addition of mevalonate, FPP, or GGPP (80). Atorvastatin and simvastatin markedly decrease IL-1β production by peripheral blood mononuclear cells (PBMCs) of patients with coronary artery disease (81). In separate, randomized prospective studies of patients with hypercholesterolemia and hypertension, treatment with simvastatin resulted in decreased circulating IL-1β, and IL-1β production by isolated PBMCs, respectively (82, 83). Both atorvastatin and simvastatin significantly decrease levels of circulating IL-6 and TNFα, in addition to IL-1β, in patients with hypercholesterolemia (39, 41, 59, 84). Carotid plaques resected from patients taking statins contain significantly lower concentrations of IL-6 (P = 0.0005), suggesting that statins do, indeed, alter local inflammation in atherosclerotic lesions (85).
Matrix metalloproteinases (MMPs) are neutral proteases that act extracellularly to digest collagen and other connective tissue molecules. MMP dysregulation plays a critical role in tissue destruction in rheumatoid arthritis (RA) (86, 87) and in gastric ulceration (88) and atherogenic vascular damage (89, 90). Statins have been reported to decrease production of MMPs 1, 3, and 9 in human macrophages and vascular smooth muscle cells (91), MMPs 1 and 9 in carotid plaques (85), and MMP-3 in IL-1β–stimulated chondrocytes (92). In our own studies on rheumatoid synovial fibroblasts, MMP-1 secretion was strongly inhibited by FTI but not GGTI or squalestatin, suggesting that statin inhibition of MMP secretion in these cells may be mediated via Ras, rather than Rho, proteins (Abeles AM, et al: unpublished observations).
Statins likely inhibit the expression of multiple inflammatory cytokines through one or several common mechanisms. The ability of statins to inhibit NF-κB activation in monocytes or ECs exposed to inflammatory stimuli suggests that this transcriptional regulator of >40 inflammatory genes may be an important statin target (93–95). The strength of statin NF-κB inhibition varies, however, according to drug, cell type, and stimulus (93–95). In vivo evidence for statin-induced NF-κB repression comes from a rabbit model in which administering atorvastatin to atherogenic rabbits reduces NF-κB activation in both arterial smooth muscle and macrophages (96). Down-regulation of Rho-related protein activation is one probable mechanism for this effect, since NF-κB is known to be activated via Rho GTPases (97). In vitro, NF-κB suppression by statins has been reversed by mevalonate, FPP, and GGPP (93). Given the centrality of NF-κB in inflammatory diseases, the ability of statins to inhibit NF-κB may support the clinical relevance of these agents as antiinflammatory drugs.
Statins also activate antiinflammatory transcription factors known as peroxisome proliferator–activated receptors (PPARs). PPARs are intracellular, ligand-activated transcription factors that interfere with NF-κB transcriptional activity (98–100). In vitro, statins induce the expression of PPARα and PPARγ mRNA and protein in stimulated ECs, macrophages, and hepatocytes (80, 101, 102). Statin-induced PPAR up-regulation is reversed by mevalonate and GGPP, but not by squalene, which implies a Rho target (80, 101). Concordantly, GGTI induces PPARα activity, as does transfection with DN RhoA, but not DN Cdc42 or DN Rac (101).
Other apparent statin targets that have been implicated in inflammatory and/or rheumatic diseases include the MAP kinase families of signal transduction molecules (ERK, JNK, and p38 families) (103–105), as well as the JAK/STAT signaling pathways (106).
Statins as immunomodulators.
Statins decrease T cell activation. In vitro, statins inhibit IFNγ-inducible class II major histocompatibility complex (MHC) expression in macrophages and ECs, an effect that is reversed by mevalonate (107). In contrast, statins have no effect on class I MHC expression in ECs. Statins' inhibition of stimulated class II MHC expression is accompanied by decreased class II MHC mRNA levels and lower levels of class II MHC transactivator mRNA (107). Decreased expression of class II MHC proteins may lead to reduced T cell activation during antigen presentation: in mixed lymphocyte reactions, atorvastatin decreases T cell activation and proliferation (107). In addition, simvastatin has been shown to inhibit T cell ERK activation through a Ras-dependent mechanism, and to decrease T cell p38 activation by a Rho-dependent mechanism; the former effect was reversed by FPP, and the latter by GGPP (108). That study also directly confirmed that statin treatment results in decreased Rho, Ras, and Rac association with lipid membranes.
Another mode by which statins inhibit T cell activation may be through inhibition of costimulatory molecules necessary during antigen presentation. The actions of statins on LFA-1, discussed above, may reduce not only leukocyte adhesion, but also costimulatory signaling. Statins also down-regulate expression of the costimulatory molecule CD40 (109, 110). The latter finding may be relevant to RA, since overexpression of CD40 and CD40 ligand (CD40L) in the rheumatoid synovium (synovial fibroblasts and lymphocytes, respectively) may play a role in RA pathogenesis (111). Indeed, interruption of CD40–CD40L interaction has been identified as a potential therapeutic target in RA (112, 113).
Patients taking statin drugs exhibit decreased T cell activity. The plasma concentration of the Th1 cytokine IL-2 is lower in patients taking statins (114). Ex vivo, T cells of patients taking fluvastatin or simvastatin secrete less IFNγ and IL-2 than prior to treatment (115). The in vivo effects of T cell inhibition by statins are elaborated upon below, in a discussion on animal models of inflammation.
Much of the evidence that statins exert their antiinflammatory properties through small molecule GTPases is indirect, but direct evidence supporting the theory that statins act on inflammation by inhibiting these molecular switches continues to accumulate. Treating fibroblasts in vitro with simvastatin significantly decreased the concentrations of fully modified forms of RhoA, Cdc42, and Rac-1; a similar effect in vitro and in vivo was seen in monocytes treated with atorvastatin (116). Another study demonstrated that simvastatin decreased membrane translocation of Ras (completely reversed by coincubation with FPP) and Rho (completely reversed by coincubation with GGPP) in macrophages (105).
Statins as antiinflammatory agents in animal models
Various animal models have demonstrated that statins may act as antiinflammatory drugs (Table 1), from simple models of inflammation, such as carrageenan-induced footpad edema, to more complex ones, such as experimental autoimmune encephalomyelitis (EAE) and collagen-induced arthritis (CIA).
Table 1. Effects of statins on animal models of inflammatory and/or autoimmune disease
Decreased disease incidence, activity, and histologic scores; not replicated in followup study
Statins in simple animal models of inflammation.
In a comparison of simvastatin and indomethacin for treatment of carrageenan-induced footpad edema in mice, the agents had similar efficacy (117). One hour prior to induction of footpad edema, animals received placebo, indomethacin, or simvastatin via oral gavage. The resultant footpad swelling in the treatment groups was significantly less than in the control group (P = 0.0001), and there was no significant difference in swelling between indomethacin- and simvastatin-treated mice.
Statins are also effective in a mouse air-pouch model of inflammation (118). A subcutaneous dorsal pouch is created by injection of air; thereafter, an irritant (e.g., LPS, carrageenan) is injected into the pouch. Administration of the study drug is initiated before administration of the inflammatory stimulus. The air pouch is later excised and examined for leukocyte concentration and leukocyte products. In this model, lovastatin, pravastatin, and simvastatin significantly decreased leukocyte recruitment into pouches (P < 0.01, P < 0.01, and P < 0.05, respectively) injected with either LPS or carrageenan. The statin effects were comparable with those of indomethacin, and were reversed by coadministration of mevalonate. Moreover, all 3 statins decreased air-pouch levels of IL-6, and lovastatin decreased RANTES and MCP-1 levels (not investigated in the other 2 statins). Whereas lovastatin did not reduce serum cholesterol levels in these studies, the squalene synthase inhibitor squalestatin had no antiinflammatory effect despite significantly lowering serum cholesterol. Thus, the effects of statins on IL-6, RANTES, and MCP-1 expression were cholesterol independent.
Statins in animal models of nonrheumatic autoimmune and inflammatory diseases.
Statins are also effective in more complex animal models of autoimmune disease. For example, simvastatin ameliorates a murine model of allergic asthma (119). Mice treated with intraperitoneal (IP) simvastatin had significantly less lung inflammation as seen on histologic analysis. Mice treated with either enteral (40 mg/kg per day) or parenteral simvastatin (40 mg/kg per day) prior to intranasal ovalbumin challenge had significantly lower total cell counts and eosinophil counts as determined by bronchoalveolar lavage; only IP-administrated simvastatin lowered macrophage counts and IL-4 and IL-5 levels. Lymphocytes cultured from thoracic lymph nodes of killed mice to which simvastatin 40 mg/kg per day had been administered by either route produced significantly less IFNγ and IL-6 than did those of control groups.
IP pravastatin (1 mg/kg per day) has been shown to reduce the severity of Dextran sulfate–induced colitis, an animal model of inflammatory bowel disease (IBD) (120). Study mice treated with pravastatin, when compared with placebo-treated mice, maintained their body weight and had significantly lower disease activity indices. Pravastatin decreased colon inflammation and colon epithelium permeability, prevented shortening of the colon, and blocked changes in colon histology. Mechanisms by which the drug may have exerted its protective effects included down-regulation of the mucosal addressin cell adhesion molecule 1, and prevention of mucosal eNOS degradation. The protective effects of pravastatin in Dextran sulfate–induced colitis were not found in eNOS-deficient mice, suggesting that eNOS regulation was critical to the statin effect.
Although the latter study did not investigate the effects of pravastatin on Rho protein activation in the colon, increased activation of RhoA has been found both in inflamed intestinal mucosa of IBD patients and in the colons of rats with trinitrobenzenesulfonic acid–induced colitis (121). Use of Y-27632 to inhibit Rho kinase significantly reduces colonic inflammation in rats with experimental colitis by preventing NF-κB activation. Thus, it may be that pravastatin ameliorates experimental IBD secondary to inhibition of Rho proteins, and statins may prove efficacious in human IBD for the same reason. No clinical studies of statin treatment of IBD have been performed yet, but monocytes from patients with Crohn's disease treated in vitro with atorvastatin (10 μM) produce significantly reduced levels of TNFα (by 45%) and MCP-1 (by 42%) (122).
Experimental autoimmune myocarditis, an animal model in which myocarditis is stimulated by myosin immunization, is alleviated by oral fluvastatin treatment. Rats treated with high-dose fluvastatin had improved functional and histologic scores when compared with placebo-treated rats. Fluvastatin inhibits myocardial inflammation by decreasing the production of the Th1 cytokines IFNγ and IL-2, inhibiting expression of NF-κB in the myocardium, decreasing IL-4, IL-6, IL-10, IL-1β, and TNFα transcription in the myocardium, and preventing T helper cells from infiltrating the heart (123).
IP lovastatin (20 mg/kg per day), but not oral lovastatin or atorvastatin, ameliorates intraocular inflammatory disease in a mouse model (i.e., experimental autoimmune uveitis). IP lovastatin treatment decreased retinal vascular leak and clinical and histologic retinal pathology. This effect was reversed with mevalonate but not with squalene, suggesting an action on one or more prenylated proteins. The cell type(s) through which lovastatin mediated this effect is unclear, but lovastatin significantly suppressed in vitro transendothelial migration of mouse lymphocytes; lymphocyte transmigration was restored with the addition of mevalonate (124).
Statins dramatically altered the course of EAE, the animal model for MS, in a number of independent studies (125–128). EAE is a T cell–driven disease in which animals are immunized with myelin proteins and, upon a second injection of these proteins, CD4+ T cells recognizing myelin antigens are activated, which leads to relapsing paralysis and central nervous system demyelination (129). Lovastatin and atorvastatin administered prophylactically both have been reported to prevent EAE, and to reverse established EAE in affected animals. Atorvastatin dramatically inhibits class II MHC expression on central nervous system microglia in mice with EAE, which leads to decreased T cell activation (127). In addition, atorvastatin completely inhibits in vitro class II MHC expression in IFNγ-stimulated microglia; coincubation with mevalonate abrogates this effect. These findings have led to clinical trials of statins in the treatment of MS, which are addressed below.
Statins may hold therapeutic promise not only for autoimmune diseases, but also for other inflammatory conditions, such as sepsis. Emerging data on statins in animal models of sepsis are encouraging. Simvastatin provides a dramatic improvement in mortality in a murine model of sepsis. Mice pretreated with simvastatin prior to cecal ligation and perforation had a mean survival time ∼4 times that of untreated mice. The simvastatin-treated animals also did not experience drops in blood pressure or cardiac output while septic. In addition, monocytes obtained from the simvastatin-treated mice had significantly reduced adhesion to ECs under physiologic flow conditions compared with those from untreated mice (an effect reversed by mevalonate) (130). A human ex vivo study showed that oxidative stress, known to increase morbidity and mortality in sepsis, may be decreased by simvastatin in patients with sepsis. Simvastatin prevented Rac-2 activation in stimulated monocytes, and decreased superoxide anion production in whole blood by 40% in 14 patients with sepsis (131).
A retrospective study of 388 human patients with bacteremic infections suggested that statins decreased mortality from sepsis. Deaths attributable to infection occurred in only 3% of patients taking statins (upon presentation to and during their hospital stay) versus 20% of those patients not taking statins. Multivariate analysis further revealed that only statin use was associated with a decreased mortality rate (132). In a prospective observational cohort study of 361 patients with bacterial infection, statin use was independently associated with a significantly lower rate of severe sepsis (P = 0.001), in addition to a lower relative risk of death (0.43) (133). The results of the latter 2 studies are not yet conclusive; a randomized controlled trial of statin use in sepsis may be warranted.
Statins in animal models of rheumatic diseases.
A limited number of animal studies have begun to address whether statins can alter the outcomes of rheumatic diseases. Fluvastatin decreased anti–β2-glycoprotein I–mediated endothelial activation in vitro (but not in the presence of mevalonate) (134), and prevented large thrombi from forming in an animal model of antiphospholipid antibody syndrome (infusion of mice with human IgG antibodies from patients with antiphospholipid syndrome) (135). Statin-treated animals developed thrombi no larger than those formed in control mice infused with normal human IgG (135). In separate studies, fluvastatin has been shown to decrease leukocyte–endothelial adhesion in postcapillary venules (135). Fluvastatin also significantly decreases expression of tissue factor by HUVECs exposed to human antiphospholipid antibodies (136).
Atorvastatin decreases disease activity in an animal model of lupus. (NZB × NZW)F1 (NZB/NZW) mice develop spontaneous autoimmune disease similar to lupus. Administering atorvastatin (30 mg/kg per day IP) to these animals resulted in lower anti–double-stranded DNA antibody levels, reduced proteinuria, lower serum urea levels, and delayed glomerular injury relative to untreated NZB/NZW mice. In addition, atorvastatin decreased class II MHC expression on B cells and monocytes, B cell and T cell activation, and T cell proliferation in the treated mice (137).
In one study, simvastatin proved effective in ameliorating CIA, an animal model of RA (138). This study had a prophylactic and a therapeutic arm (statin treatment before and after induction of disease). Mice in the prophylactic group were randomized to 4 dosage schedules: placebo, or simvastatin at 10 mg/kg, 20 mg/kg, or 40 mg/kg per day. Mice in the therapeutic arm received either control treatment or simvastatin at 40 mg/kg per day. In both arms of the study, mice receiving simvastatin at 40 mg/kg per day had significant improvement versus controls, with a decreased disease incidence and articular index in the prophylactic group, and a decreased articular index and fewer arthritic paws developing in the therapeutic group. In the therapeutic arm, histologic scores of joints dramatically improved with simvastatin (P < 0.01). Lymphocyte cultures from lymph nodes of statin-treated mice with CIA produced less TNFα and IFNγ than did lymphocyte cultures from control mice. T cell proliferation was also significantly suppressed in lymphocyte cultures from simvastatin-treated animals. However, a more recent study did not replicate these findings using atorvastatin, rosuvastatin, or simvastatin (139).
While these results are intriguing, it must be noted that the statin doses used in animal models of inflammatory disease have typically been higher than those used in human therapy. Whereas statins are typically prescribed in a range of 0.1–1.0 mg/kg per day, the doses used in animal experiments have been as high as 40 mg/kg per day.
Statin trials in human inflammatory and autoimmune diseases
Animal models of human disease may yield insight into disease pathogenesis and treatment, but cannot substitute for actual human data; statin efficacy in animal models has therefore led to clinical trials in inflammatory and/or autoimmune diseases.
Statins in human trials of nonrheumatic autoimmune diseases.
Two small initial randomized controlled trials (RCTs) investigating statins for prevention of allograft kidney rejection showed significantly lower rejection rates in statin-treated patients (140, 141). However, 3 subsequent RCTs demonstrated no significant differences between short-term rejection rates between patients in the intervention and control arms (142–144). It is unclear whether these studies failed to replicate the earlier findings because of drug inefficacy or because of drug choice and dosage. One study, for example, used only 10 mg of simvastatin per day in the treatment arm, which was 12.5% of the daily maximum recommended dose (142).
In more recent trials, statins have begun to look promising. In 2004, a single-arm, open-label trial involving 30 individuals taking 80 mg of simvastatin per day for relapsing–remitting MS was reported (9). After 6 months of treatment, the number of gadolinium-enhancing brain lesions decreased by 44% (P < 0.0001) and the volume of the lesions decreased by 41% (P = 0.0018) when compared with the lesions noted on pretreatment magnetic resonance imaging. A randomized, double-blind, placebo-controlled study investigating atorvastatin for the treatment of MS is now under way.
Statins may also play a role in preventing cancer, including (but not limited to) colorectal cancer (145), and they may hold promise for treating established cancer (146). Although the role of inflammation in cancer is increasingly recognized (146–149), it is not yet known whether statins can modulate malignancy directly via their antiinflammatory effects.
Statins in human trials of rheumatic diseases.
Despite a great deal of excitement about the antiinflammatory potential of statins in rheumatic diseases, only a small number of studies have actually been carried out to evaluate the efficacy of statins in these settings (Table 2).
Table 2. Effects of statins in rheumatic disease clinical trials*
RA = rheumatoid arthritis; ACR50 = American College of Rheumatology 50% criteria for improvement; DAS28 = 28-joint Disease Activity Score; OA = osteoarthritis; OR = odds ratio; SLE = systemic lupus erythematosus; GCA = giant cell arteritis.
Retrospective, unblinded; 54 patients; 17 taking statins, most at low doses
Two small preliminary studies, 1 carried out in Japan and 1 conducted in Mexico, revealed dramatic RA disease improvement in statin-treated patients (150, 151). In the Japanese study, a 12-week, open-label, single-arm study of 24 patients receiving 10 mg of simvastatin daily, 39% of the treatment group met the American College of Rheumatology (ACR) 50% improvement criteria (achieved an ACR50 response) (152). The group in Mexico conducted an 8-week, open-label study of simvastatin at a dosage of 40 mg daily; after 4 weeks, 9 of 10 patients achieved an ACR50 response; by the end of the study, 7 of 10 patients achieved an ACR70 response. These exceptional response rates should be interpreted with caution, however. These studies were very small, their design allowed for observer bias, and their reported ACR response rates were equal to or better than those found with currently approved biologic agents. A recent cross-sectional study of patients in the National Databank for Rheumatic Diseases (153) revealed that statin use was independently associated with modestly reduced Health Assessment Questionnaire (HAQ) scores (154).
A larger randomized placebo-controlled study investigating atorvastatin as a disease-modifying antirheumatic drug (DMARD) in RA (Trial of Atorvastatin in Rheumatoid Arthritis) also showed modest effects (155). In this 116-patient study, patients received either 40 mg of atorvastatin per day or placebo in addition to current DMARD therapy; DMARDs were not allowed to be changed during the 6-month study. At the end of 6 months, the group receiving atorvastatin showed statistically significant improvements in the 28-joint Disease Activity Score (DAS28) (156).
Before adding statins to the DMARD arsenal, however, further studies need to be conducted, because there were limitations to that study. In particular, a potentially confounding difference existed between the control and study groups: 50% of patients in the atorvastatin group were taking methotrexate versus 26% of patients in the placebo group. Whether the response rates between the groups was actually due to more methotrexate use rather than statin use is an important question. Putting aside these randomization issues, the increased response in the statin group was actually modest. The improvement in DAS28 may have reached the level of statistical significance, but the patients demonstrated no subjective improvement. That is, the study group showed improvements in CRP, erythrocyte sedimentation rate, and the number of swollen joints by physician examination, but had no improvement in early-morning stiffness, tender joint count, visual analog score for pain, patient global assessment, or HAQ score.
Statins have also been investigated in other rheumatic conditions. Statins did not exert a steroid-sparing effect in giant cell arteritis in one retrospective study, although more than half of these patients received only low-dose statin therapy, which rendered the data inconclusive (157). In an in vitro study, statins inhibited neutrophil activation in response to antineutrophil cytoplasmic antibodies (ANCAs) (103); whether this observation suggests potential utility of statins in ANCA-associated vasculitides remains to be determined. On the other hand, the authors of a very small study reported that 3 of 3 patients who had been treated unsuccessfully with cyclophosphamide and prednisone for lupus glomerulonephritis responded dramatically to 8-day treatment with high-dose simvastatin (80 mg per day) (151).
As mentioned above, statins have been associated with decreased rates of osteoporosis (of the hip) (10). In contrast, a recent prospective observational cohort study indicated that statin use may be associated with the development of hip osteoarthritis (OA) in elderly women, possibly related to increased bone density. However, several caveats regarding this latter study must be noted: 1) patients taking statins were significantly more obese than those not taking statins at baseline; 2) of 5 criteria used to diagnose radiographic hip OA, only 1 showed an increase in the statin group; and 3) statin use did not worsen existing hip OA during the duration of the study (158). A trial of statin prophylaxis for steroid-induced avascular necrosis in lupus patients is ongoing, with results pending (Belmont HM: personal communication).
In selecting a cholesterol-lowering agent for a patient with rheumatic disease, rheumatologists may wish to consider the antiinflammatory effects of statins. In doing so, the practicing clinician should bear in mind that individual statins may differ in their antiinflammatory potential. In one study, statins varied by as much as 10-fold in the degree to which they inhibited NF-κB activation in stimulated monocytes (cerivastatin > atorvastatin > simvastatin > pravastatin > lovastatin > fluvastatin). Given the current limitations of the data, however, it would be premature to make specific formal recommendations about the use of any particular statin as an antiinflammatory agent (94).
Statins as cardioprotective agents in rheumatic disease.
Systemic inflammatory diseases such as RA and systemic lupus erythematosus (SLE) are associated with accelerated atherosclerosis, and both RA and SLE patients have a significantly increased risk of myocardial infarction and death (159–161). Since this increased risk is not accounted for by traditional risk factors (162), it has been postulated that systemic inflammation itself may participate in accelerated atherosclerosis. Moreover, controlling systemic inflammation in patients with atherosclerotic heart disease may independently improve cardiovascular risk (163). Statins may therefore be indicated for cardiovascular prophylaxis in some rheumatic diseases, even in the absence of hypercholesterolemia.
Elevated CRP levels correlate with accelerated atherosclerosis in RA patients (164). Statins reduce CRP production in response to stimuli such as IL-6 in vitro (165) and reduce CRP levels in vivo, correlating with low-density lipoprotein–independent improvement in cardiovascular outcome (163, 166, 167). Whether CRP is itself pathogenic in atherosclerosis, or whether elevated CRP levels merely reflect the presence of inflammatory mediators such as IL-6, remains to be determined (168, 169). Statins may also ameliorate accelerated atherosclerosis in rheumatic disease via effects on endothelium. Even young RA patients with low disease activity have significant endothelial dysfunction (170, 171), and statins improve endothelial function in patients with RA. Studies to test the effect of statins on cardiac outcomes in lupus and RA are ongoing.
Over the last decade, it has become increasingly clear that statins have antiinflammatory properties, independent of their lipid-lowering effects. What is less clear is whether this class of drugs will prove to be useful as antiinflammatory agents for “high-grade” inflammatory diseases such as RA, Crohn's disease, and lupus. Preliminary data from open-label studies of statin treatment in inflammatory diseases have been impressive, but must be interpreted with caution, because data emerging from double-blind placebo-controlled trials have so far been less definitive. Even if statins prove only mildly effective in reducing inflammation and/or autoimmunity in rheumatic diseases, their relative safety, together with their potential for reducing the inflammatory and lipid-mediated processes of accelerated atherosclerosis, suggest that statins may at least prove to be useful adjunctive therapy in patients with rheumatic disease.
The authors wish to thank Steven B. Abramson for helpful suggestions, and Nada Marjanovic for performing statin experiments described in this report.