Abstract
Substantial evidence now exists to recognize hepatic stellate cells (HSCs) as the main matrix-producing cells in the process of liver fibrosis. Liver injury of any etiology will ultimately lead to activation of HSCs, which undergo transdifferentiation to fibrogenic myofibroblast-like cells. Quantitative analysis of HSC activation by immunohistochemistry has been shown to be useful in predicting the rate of progression of liver fibrosis in some clinical situations. In the activation process, transforming growth factor β is thought to be the main mediator of fibrogenesis and platelet-derived growth factor is the major inducer of HSC proliferation. Different platelet-derived growth factor and transforming growth factor β inhibitors have been shown to effectively prevent liver fibrosis in animal models and represent promising therapeutic agents for humans.
Fibrosis is the essential pathophysiologic consequence of chronic liver injury, and it represents the common underlying mechanism for hepatic insufficiency and for most clinical complications of end-stage liver disease. Initially thought to represent a passive process, whereby hepatocellular death led to condensation of the preexisting stroma,1 fibrosis was redefined by the World Health Organization in 1978 as “the presence of excess collagen due to new fiber formation.” 2(p411) Currently, fibrosis is known to be part of a dynamic process of continuous extracellular matrix (ECM) remodeling in the setting of chronic liver injury, which leads to an excessive accumulation of several extracellular proteins, proteoglycans, and carbohydrates.3,4 Although the process has only partially been elucidated, hepatic stellate cells (HSCs) have consistently been shown to play a key role in hepatic fibrogenesis, regardless of the underlying etiology.5–10 The understanding of the complex pathophysiologic mechanisms involving these cells is pivotal for the development of antifibrotic therapies.
Hepatic stellate cells have been known to pathologists for well over a century, since they were first described by von Kupffer in 1876. Formerly known as Ito cells, lipocytes, or perisinusoidal cells, the role of HSCs in storing lipid-soluble vitamin A in their quiescent state was well described decades ago.11 In the mid-1980s, Friedman and colleagues12 identified hepatic “lipocytes” as the main collagen-producing cells in rat livers. Experimental animal models of liver fibrosis as well as human HSC culture techniques were rapidly developed and standardized thereafter.12–15 A large number of subsequent studies uncovered some of the essential mechanisms of HSC activation in response to injury and described the steps involved in transdifferentiation of HSCs into myofibroblast-like cells. Currently, substantial evidence exists that HSCs are the main collagen- and ECM-producing cells in the process of liver fibrosis.3,4,16 The relatively recent recognition of HSCs as key elements in liver fibrosis led to an unprecedented interest in this cell (and its activated form) as a prognostic indicator of progression of liver fibrosis,17,18 as well as a potential target for therapeutic intervention to prevent the development of cirrhosis.19–24
THE HSC AND ITS ROLE IN THE NORMAL LIVER
The nature of stellate cells, traditionally considered cells of mesenchymal origin due to their morphologic features as well as their expression of molecules such as vimentin, desmin, and α-smooth muscle actin (α-SMA),25–30 has recently been questioned, as numerous studies have clearly demonstrated expression of neural and neuroendocrine markers, such as glial fibrillary acid protein,31 nestin, neurotrophin receptor,32 and synaptophysin.33 To add complexity to the subject, Lim et al34 have also reported expression of cytokeratins 18 and 19 in human HSC culture that is lost during activation. The authors suggest a possible epithelial origin of stellate cells and an epithelial to mesenchymal phenotype transdifferentiation upon activation. Therefore, the origin and nature of HSCs are still uncertain.
In the normal liver, HSCs comprise approximately 1.4% of total liver volume and are present at a ratio of about 3.6 to 6 cells per 100 hepatocytes (or 1:20). Hepatic stellate cells are typically located in the perisinusoidal space of Disse, a recess between endothelial cells of sinusoids and hepatocytes. Stellate cells (from the Latin stella, which means star) typically have a starlike configuration due to their denditric cytoplasmic processes that partially embrace adjacent endothelial cells3 in a manner somewhat analogous to astrocytes around terminal cerebral vessels or podocytes around renal capillaries, extending into the spaces in between hepatocytes and reaching the cytoplasmic processes of other stellate cells.35
In its quiescent state, HSCs show large perinuclear lipid droplets, which serve as the main storage site for vitamin A and are essential in the regulation of retinoic acid homeostasis.36 Stellate cells also seem to play a key role in the maintenance of steady-state levels of basement membrane–like matrix (mostly types IV and VI collagen) in normal hepatic sinusoids,13 as well as in regulation of hepatic blood flow and portal venous pressure.35,37
PATHOPHYSIOLOGY OF LIVER FIBROSIS
The ECM
Fibrosis is characterized histologically and biochemically by a several-fold elevation in the total ECM content of the liver.4 All major constituents of normal ECM are represented, to some extent, in the newly formed matrix during the fibrogenic process. As in normal ECM, collagen (especially types I and III) and elastin are the most abundant proteins, but glycoproteins and pure carbohydrates are also present. When compared to normal matrix, scar tissue produced in liver fibrosis has a significantly higher percentage of type I collagen.3 Extracellular matrix deposition occurs as a result of both excessive production and decreased fiber degradation. In a normal liver, matrix metalloproteinases (MMPs) have a well-described ECM-degrading function. The activity of MMPs, however, is suppressed in the setting of liver injury as a result of overexpression of tissue inhibitor of metalloproteinases (TIMPs) by activated HSCs.4
HSC Activation
Hepatic stellate cells are unequivocally the main cells involved in the production of excessive ECM seen in liver fibrosis. As a result of injury, HSCs undergo “activation” or transdifferentiation, from a quiescent, vitamin A–storing cell to a myofibroblast-like cell, with several new phenotypic characteristics, such as enhanced cell migration and adhesion, expression of α-SMA, increased proliferation, production of chemotactic substances capable of recruiting inflammatory cells as well as other HSCs, contractibility, loss of normal retinoid-storing capacity, increased rough endoplasmic reticulum, changes in cytoskeletal organization and cellular morphology and, most importantly, acquisition of fibrogenic capacity (Figures 1, and 2, A and B).4,20,38–40 The process of activation has classically been divided into two distinct phases. In the first phase, or the initiation phase, HSCs undergo the initial changes toward a myofibroblast-like cell differentiation and become more responsive to proliferative and fibrogenic cytokines by up-regulation of membrane receptors. Once HSCs have been induced to overexpress cytokine receptors, increased fibrogenesis, cellular proliferation, and other features of the “activated” phenotype will be perpetuated by the continued release of mediators from chronically inflamed and injured tissue. This second stage of the activation process has been termed the perpetuation phase. Several cell types present in the normal liver, such as hepatocytes, Kupffer cells, sinusoidal endothelial cells, platelets, and activated HSCs themselves have been implicated in production of cytokines and other mediators thought to play a role in this latter part of the process.3,40 Once activated, HSCs will up-regulate gene expression of ECM components, matrix-degrading enzymes and their respective inhibitors, resulting in matrix remodeling and accumulation at sites with abundant activated HSCs.4,41
Phenotypic changes of hepatic stellate cells upon activation during liver injury. PDGF indicates platelet-derived growth factor; ET, endothelin; TGF-β1, transforming growth factor beta1; MMP, matrix metalloproteinase; MCP, monocyte chemoattractant protein; and WBC, white blood cell. Reprinted with permission from Friedman.72 Copyright 2000, The American Society for Biochemistry and Molecular Biology
Phenotypic changes of hepatic stellate cells upon activation during liver injury. PDGF indicates platelet-derived growth factor; ET, endothelin; TGF-β1, transforming growth factor beta1; MMP, matrix metalloproteinase; MCP, monocyte chemoattractant protein; and WBC, white blood cell. Reprinted with permission from Friedman.72 Copyright 2000, The American Society for Biochemistry and Molecular Biology
Typical morphology of quiescent hepatic stellate cells. A, Note the presence of short cytoplasmic dendritic processes and perinuclear vacuoles containing retinoids (cell culture, original magnification ×400). B, Upon activation, stellate cells undergo transdifferentiation into myofibroblast-like cells, with changes in cytoskeletal organization and loss of retinoid-storing capacity (cell culture, original magnification ×400)
Typical morphology of quiescent hepatic stellate cells. A, Note the presence of short cytoplasmic dendritic processes and perinuclear vacuoles containing retinoids (cell culture, original magnification ×400). B, Upon activation, stellate cells undergo transdifferentiation into myofibroblast-like cells, with changes in cytoskeletal organization and loss of retinoid-storing capacity (cell culture, original magnification ×400)
Cells and Cytokines Interacting With HSCs
An increasingly complex interplay between various cell types in the liver and stellate cells has now become apparent. Among the cells known to interact with stellate cells are hepatocytes, Kupffer cells, platelets, epithelial cells, sinusoidal endothelial cells, and neutrophils. Each of these cell types will release a subset of mediators that will have diverse effects on HSCs. Platelet-derived growth factor (PDGF) and transforming growth factor beta (TGF-β) are the two best-characterized cytokines in stellate cell activation. A large number of studies identify PDGF as the main mediator for proliferation and TGF-β as the most important cytokine stimulating fibrogenesis in stellate cells.3,4,21,23,24 A variety of mediators, however, have been shown to promote HSC activation and fibrogenesis, including monocyte chemotactic protein type 1,42 endothelin 1,43 angiotensin II,4 and such adipokynes as leptin.9 In addition, several molecules, including tumor necrosis factor α (TNF-α), TGF-β, TIMP-1, collagen 1, and integrins have been shown to be fibrogenic by causing inhibition of HSC apoptosis, thereby contributing to the increased number of these cells in the setting of hepatic injury.24 Reactive oxygen species produced by Kupffer cells and damaged hepatocytes also have a role in HSC activation as well as in the recruitment of inflammatory cells. Since activated HSCs themselves secrete inflammatory chemokines, a vicious cycle is formed, whereby fibrogenic and inflammatory cells stimulate each other and perpetuate a process of liver damage and repair.2,44
Finally, a recent study has also demonstrated that integrin-linked kinase, an intracytoplasmic integrin-associated signaling molecule, has an important role in fibrogenesis both in vitro and in vivo in experimental models.20 These findings reinforce the hypothesis that activation of HSCs may be triggered not only by circulating and locally released mediators, but also by alterations in the ECM itself, due to direct interaction between these cells and adjacent matrix fiber molecules.
HSCs and Clinical Liver Dysfunction
Extracellular matrix deposition in the space of Disse leads to disruption of the normal, fenestrated microanatomy of the liver sinusoids in a process referred to as capillarization of these vessels (Figure 3).45 Capillarization of sinusoids impairs the normal bidirectional exchange between the portal venous blood and hepatocytes, causing substances that would otherwise be degraded or metabolized by hepatocytes to bypass the liver and reach the systemic circulation (portosystemic shunting), and preventing substances that are produced in the liver to reach the blood. This process further complicates problems primarily caused by portal hypertension and decreased hepatocellular synthetic function, such as hyperbilirrubinemia, hepatic encephalopathy, hypoalbuminemia, and deficiency of coagulation factors.3 Accumulating evidence exists that stellate cells are also involved in the regulation of portal venous blood flow and have a role in the development of portal hypertension and its complications. Previously thought to result exclusively from hemodynamic changes associated with increased intravascular volume, increased splancnic blood flow, and decreased portal venous compliance secondary to fibrosis and architectural distortion by regenerative nodules, there is now strong evidence to support that portal hypertension may be in part caused by modulation of the contractile activity of stellate cells in the perisinusoidal space, which function as liver-specific pericytes, under the influence of vasoactive substances, such as endothelin 1 and nitric oxide. Studies in experimental models have shown that intrahepatic vascular resistance can be decreased by up to 20% to 30% with pharmacologic agents, reinforcing the concept that a “variable” or “reversible” component regulating portal pressure exists.46
Sinusoidal events during liver injury. Normal liver architecture with quiescent hepatic stellate cells within space of Disse (left). Injured liver with activated hepatic stellate cells producing extracellular matrix (fibrogenesis). Also note the closure of fenestrae or “capillarization” of sinusoids (right). Reprinted with permission from Friedman.72 Copyright 2000, The American Society for Biochemistry and Molecular Biology
Sinusoidal events during liver injury. Normal liver architecture with quiescent hepatic stellate cells within space of Disse (left). Injured liver with activated hepatic stellate cells producing extracellular matrix (fibrogenesis). Also note the closure of fenestrae or “capillarization” of sinusoids (right). Reprinted with permission from Friedman.72 Copyright 2000, The American Society for Biochemistry and Molecular Biology
HSC Activation and Morphologic Features of Fibrosis
Several notable differences exist in the molecular events leading to HSC activation in the different underlying etiologies of liver disease. Some of these differences are also reflected in distinct histopathologic features associated with specific etiologies. In ethanol-induced fibrosis, for example, the deposition of ECM typically starts in the perisinusoidal space of Disse of zone 3 of the liver, in a perivenular distribution, later evolving to fibrosis around lobular hepatocytes, in a “chicken-wire” pattern. Conversely, fibrosis secondary to hepatitis C virus and chronic cholestatic diseases (primary biliary cirrhosis and primary sclerosing cholangitis) typically starts in the periportal region rather than around central veins, progressing to bridging fibrosis at later stages of the disease.4
On a molecular level, the different morphologic features of fibrosis associated with various etiologies are probably related to the distribution of the primary sites of HSC activation. In alcoholic liver disease, the initial perivenular distribution of ethanol-induced fibrosis is thought to be related to the predominantly centrilobular expression of cytochrome P4502E1, a P-450 isoenzyme that participates in ethanol oxidation and is also induced by this agent. Even with chronic exposure to alcohol, when this isoenzyme is also overexpressed in the midlobular and, eventually, in the periportal region, a gradient is maintained, with greater expression around the centrilobular vein, corresponding to the morphologic progression of alcoholic liver fibrosis.47,48 Alcohol metabolism by the P450 enzymes is associated with production of mediators such as reactive oxygen species, which will locally activate HSCs and induce fibrogenesis at the sites of more intense enzymatic activity. Conversely, for most other etiologies of chronic liver injury, such as viral hepatitides and chronic cholestatic diseases, the inflammatory response and HSC activation occur predominantly within and around the portal tracts, where fibrosis typically initiates.4
HSC ACTIVATION AS A PROGNOSTIC MARKER
Besides mesenchymal markers such as α-SMA, vimentin, fibroblast activation protein,49,50 and desmin,25–27 HSCs have been shown to express several neural/neuroendocrine markers, including glial fibrillary acid protein,31 nestin,51 synaptophysin,33 and HMB-45,52 as discussed earlier. α-SMA, however, is by far the most commonly used marker in both experimental and clinical studies, and it is widely accepted as a reliable indicator of HSC activation (Figure 4, A and B). Multiple studies have demonstrated a correlation between the severity of hepatic fibrosis (staging) and the number of activated stellate cells present in the liver.17,18,35 Hepatic stellate cell activation, as assessed by α-SMA immunoreactivity, has also been well documented in patients with chronic viral hepatitis without cirrhosis, but correlation with necroinflammatory activity is poor.17,53,54
A, Inconspicuous, quiescent stellate cells in perisinusoidal spaces of a normal liver (α-smooth muscle actin immunohistochemistry, original magnification ×400). B, Activated stellate cells with intense α-smooth muscle actin cytoplasmic staining in a liver biopsy of a hepatitis C–positive patient with lobular inflammation (immunohistochemistry, original magnification ×400)
A, Inconspicuous, quiescent stellate cells in perisinusoidal spaces of a normal liver (α-smooth muscle actin immunohistochemistry, original magnification ×400). B, Activated stellate cells with intense α-smooth muscle actin cytoplasmic staining in a liver biopsy of a hepatitis C–positive patient with lobular inflammation (immunohistochemistry, original magnification ×400)
While necroinflammatory activity is generally regarded as a pathologic finding with some value in predicting the rate of progression of liver fibrosis, it would be reasonable to hypothesize that a quantitative assessment of HSC activation could be a better prognostic tool for development of fibrosis, since stellate cell activation represents a more terminal event in the fibrogenesis process. In fact, two recent studies have shown that in patients with hepatitis C, HSC activation scores in liver biopsies performed 4 months after transplantation were predictive of development of fibrosis within 2 years. In these studies, necroinflammatory activity was found to be less useful in predicting development of fibrosis. Interestingly, the distribution of activated HSCs may also be relevant in this setting. Activated HSCs within portal tracts and fibrous septa (“mesenchymal” HSCs) were found to have a positive predictive value of 100% for development of fibrosis within 2 years of transplantation, although the sensitivity of this finding was low.17,18
Nonalcoholic fatty liver disease, likewise, represents a clinical situation in which a subset of patients will develop progressive liver fibrosis. However, no clinical or pathologic finding can reliably identify individual patients at increased risk for development of cirrhosis. Using α-SMA immunohistochemistry, HSC activation scores have recently been found to be of value in this condition. In one study, patients who developed fibrosis (mean follow-up of 22 months) had significantly higher activation scores compared with patients with stable disease. Again, although the sensitivity of high HSC activation scores was relatively low, the reported positive predictive value and specificity were 90% and 94%, respectively.55
Quantitative analysis of activated HSCs, therefore, has been shown to be a useful tool in predicting development and progression of liver fibrosis in patients with recurrent hepatitis C following liver transplantation as well as in patients with nonalcoholic steatohepatitis. The potential of HSC activation analysis as a prognostic indicator in liver fibrosis, however, is yet to be fully explored, as different immunohistochemical markers for HSC activation, such as glial fibrillary acid protein and fibroblast activation protein, as well as analysis of different patterns of HSC activation may increase our ability to predict the development of fibrosis in various clinical scenarios.
HSCs AS A POTENTIAL THERAPEUTIC TARGET IN LIVER FIBROSIS
Cirrhosis is currently ranked the 10th leading cause of mortality in the United States, affecting more than 1.5 million people. In spite of its relevance as a major cause of morbidity and mortality worldwide, removal of the underlying cause of chronic liver injury and liver transplantation are the only currently available therapeutic interventions capable of modifying the natural history of liver fibrosis.4 Elimination of the causative agent is often not possible, and liver transplantation has various drawbacks, including shortage of donors, costs, procedure-associated risks, and complications of immunosuppression. To date, no antifibrotic medication has been approved for clinical use.
The enormous amount of research done in the last 2 decades, however, has greatly refined our understanding of the molecular mechanisms involved in liver fibrosis, most of which are intimately connected to hepatic stellate cells. As a result, several key steps in the process of stellate cell activation and fibrogenesis have been identified as potential therapeutic targets, which may be clinically useful in preventing or treating liver fibrosis.
Among the mediators involved in fibrogenesis, TGF-β and PDGF are thought to play a central role. Consequently, several studies have been published looking at the effect of inhibitors of these substances in experimental models of liver fibrosis. Several substances with direct and indirect TGF-β inhibitory activity have been shown to significantly reduce liver fibrosis in animal models.4,56 Studies have also evaluated the effectiveness of genetically engineered soluble TGF-β and PDGF-β receptor obtained by fusing the Fc domain of human immunoglobulin G with the extracellular domain of the TGF-β type II and PDGF-β receptors, respectively, in rat models of liver injury, and have demonstrated significant reduction in bile duct ligation–induced liver fibrosis with these agents.23,24
Pirfenidone, an antifibrotic agent previously reported to be effective in experimental models of lung fibrosis, kidney fibrosis, and bronchiolitis obliterans,57 has now also been reported to be an effective liver antifibrotic drug in animals. Two studies have demonstrated that its antifibrotic properties are due, at least in part, to inhibition of TGF-β and PDGF.57,58 Imatimib mesylate (STI-571, or Gleevec), a Bcr-Abl and c-Kit tyrosine kinase inhibitor that also inhibits the PDGFR tyrosine kinase, has been tested in a rat model and shown to effectively prevent development of liver fibrosis.22 A recent study has also evaluated dual inhibition of PDGF and TGF-β with a combination of imatinib mesylate and an ACE inhibitor (perindopril) and found that either one of the substances was able to reduce CCl4-induced liver fibrosis in rats, and that combination therapy was more effective than either agent alone.21 It is interesting to note that unlike many other potential antifibrotic agents, both imatimib mesylate and ACE inhibitors are widely used in clinical practice with proven long-term safety, and they have been tested in doses comparable to those already being used for other indications. For this reason, imatinib mesylate and ACE inhibitors are likely to be among the first drugs to be tested in humans as hepatic antifibrotic agents.
Induction of HSC apoptosis has recently been explored as a possible antifibrotic strategy. Gliotoxin, a toxic fungal metabolite, was successfully used to stimulate HSC apoptosis in vivo and to inhibit fibrogenesis. Similarly, in vitro–activated HSCs infected with adenoviruses expressing p53 or retinoblastoma protein were shown to undergo apoptosis.59 Obviously, major technical obstacles still exist for this approach to be successfully used in vivo, but the concept of using targeted HSC apoptosis induction to prevent or treat liver fibrosis is certainly worthy of further investigation.
As previously mentioned, in the setting of several types of liver injury, oxidative stress is also known to play a role in hepatic fibrosis by means of activation of HSCs. Many agents with antioxidant activity, such as N-acetylcystein, resveratrol, quercetin, glutathione, and α-tocopherol, have been studied and have been shown to interfere with fibrogenesis in vitro.3
REVERSIBILITY OF LIVER FIBROSIS
Liver fibrosis has traditionally been regarded as an irreversible process. However, accumulating evidence indicates that even advanced fibrosis may, in fact, be a reversible condition.60–62 Pathophysiologically, liver fibrosis is now thought to be a dynamic process in which fibrogenic and fibrolytic pathways co-exist and interact. During liver injury, HSCs undergo activation and proliferate under the influence of activating mediators and antiapoptotic factors, respectively, as previously discussed. In this setting, production of excessive ECM takes place, and increased activity of TIMP prevents its degradation, leading to a fibrogenic state. Upon removal of the underlying cause of liver injury, HSCs undergo apoptosis and the activity of TIMP decreases, allowing MMPs to exert their fibrolytic activity, initiating a process known as remodeling.4,60
In animal models of liver injury, regression of fibrosis has been well documented after removal of the insulting agent (ie, bile duct ligation and carbon tetrachloride infusion).63,64 Likewise, resolution of liver fibrosis has also been well described in humans after successful treatment of chronic liver diseases, such as alcoholic hepatitis, iron and copper overload, autoimmune hepatitis, and viral hepatitides.4,65–68 Of these, hepatitis C virus infection is the most extensively studied condition. After successful treatment with interferon α plus ribavirin, significant improvement of liver fibrosis has been reported in nearly half of the patients.60,66
In spite of these promising observations, however, several fundamental questions remain to be answered about the reversibility of liver fibrosis. Several studies suggest, for example, that “mature” fibrosis may be more resistant to degradation than newly formed fibrotic tissue, which may in part be due to more extensive cross-linking of collagen fibers69 and different composition of the ECM.60,70,71 Moreover, for unclear reasons, TIMP overexpression and HSC activation may persist even after eradication of the underlying cause of liver injury, preventing resolution of fibrosis.60
Therefore, the process of regression of liver fibrosis is yet to be fully elucidated. Nevertheless, there is strong clinical and pathophysiologic evidence to support the concept that liver fibrosis is a potentially reversible condition.
SUMMARY
Liver fibrosis is one of the leading causes of morbidity and mortality worldwide, but very limited therapeutic options are currently available for this condition. Indisputable evidence now exists that HSCs play a central role in hepatic fibrogenesis secondary to virtually all types of liver injury. In spite of the different forms of initial insult to the liver, affecting different cell populations through diverse mechanisms, and inducing the release of various combinations of mediators, activation of stellate cells represents the unifying pathophysiologic consequence of all forms of chronic liver injury: the convergence point of multiple pathways that ultimately lead to liver fibrosis.
Quantitative analysis of HSC activation is a promising prognostic indicator in chronic liver injury and may be useful in identifying patients at increased risk of developing end-stage liver disease in selected clinical situations in which some form of therapeutic intervention is available. Unfortunately, no drug has yet been approved for the treatment of liver fibrosis in humans. However, our understanding of the cellular and molecular mechanisms involved in hepatic fibrogenesis has increased exponentially in recent years and, as a result, some of the key steps in this process were described, and several new strategies capable of blocking essential pathways in liver fibrosis were selectively studied. Consequently, there is now strong evidence in animal models to support the effectiveness of a large number of agents as antifibrotic therapy. In addition, significant evidence now exists to support the concept that liver fibrosis is a potentially reversible condition. As our understanding of HSCs and their complex pathophysiologic interactions is further refined, treatment of liver fibrosis comes closer to becoming a reality.
Acknowledgments
I thank S. Sharma, MD (Department of Pathology and Laboratory Medicine, Emory University School of Medicine), for her guidance and advice in the preparation of this manuscript. I also thank S. L. Friedman, MD (Mount Sinai School of Medicine), X. Ding, PhD, and F. Anania, MD (Division of Digestive Diseases, Emory University School of Medicine), for providing illustrations.
References
The author has no relevant financial interest in the products or companies described in this article.
Author notes
Reprints: Roger Klein Moreira, MD, Department of Pathology and Laboratory Medicine, Emory University School of Medicine, 1364 Clifton Rd NE, Atlanta, GA 30322 ([email protected])