, Volume 20, Issue 1, pp 1–18

Targeting leukocyte migration and adhesion in Crohn’s disease and ulcerative colitis


DOI: 10.1007/s10787-011-0104-6

Cite this article as:
Thomas, S. & Baumgart, D.C. Inflammopharmacol (2012) 20: 1. doi:10.1007/s10787-011-0104-6


Crohn’s disease and ulcerative colitis are two chronic inflammatory bowel diseases. Current biologic therapies are limited to blocking tumor necrosis factor alpha. However, some patients are primary non-responders, experience a loss of response, intolerance or side effects defining the urgent unmet need for novel treatments. The rapid recruitment and inappropriate retention of leukocytes is a hallmark of chronic inflammation and a potentially promising therapeutic target. We discuss the immunological mechanisms of leukocyte homing and adhesion in the gut mucosa. The interaction of lymphocytes (CD4+ T-cells, CD8+ T-cells, TREG, TH1, TH17, B-cells), monocytes, macrophages, dendritic cells and granulocytes with endothelial and epithelial cells through integrins [α4β7 (LPAM-1), αEβ7 (HML1 Human Mucosal Lymphocyte Antigen 1), α4β1 (VLA-4), αLβ7, (LFA-1)] and their ligands immunoglobulin superfamily cellular adhesion molecules (CAM) (MAdCAM-1 Mucosal Addressin Cellular Adhesion Molecule 1, ICAM-1 Intercellular Cell Adhesion Molecule, VCAM-1 Vascular Cell Adhesion Molecule), fibronectin as well as chemokine receptors (CCR2, CCR4, CCR5, CCR7, CCR9, CCR10, CXCR3, CX3CR1) and chemokines [CCL5, CCL25 (TECK Thymus Expressed Chemokine), CCL28, CX3CL1, CXCL10, CXCL12] in the process of gut homing is critically reviewed and summarized in scientific cartoons. Moreover, we discuss the clinical trial results of approved and investigational antibodies and small molecules including natalizumab (anti-α4, Tysabri®, Antegren®), AJM300 (anti-α4), etrolizumab (anti-β7, rhuMAb-Beta7), vedolizumab (anti-α4β7, LDP-02, MLN-02, MLN0002), PF-00547659 (anti-MAdCAM), Alicaforsen (anti-ICAM-1), and CCX282-B (anti-CCR9, GSK-1605786, Traficet-EN™) and their risks such as PML reported for natalizumab. Hopefully, the newer gut specific drug designs discussed in this article will have an impact on both efficacy and safety.


Crohn’s disease Ulcerative colitis Inflammatory bowel disease IBD Immunoglobulin superfamily cellular adhesion molecules Integrins Chemokines Adhesion Migration Recruitment Leukocytes T-cells Dendritic cells 



Extracellular matrix


Inflammatory bowel disease


Ulcerative colitis


Crohn’s disease


Dendritic cell(s)




Central nervous system


Intraepithelial lymphocytes


Lamina propria


High endothelial venule


Crohn’s disease and ulcerative colitis are chronic inflammatory bowel diseases (IBD) resulting from an inappropriate immune response, in genetically susceptible individuals, to microbial antigens of commensal microorganisms. This inappropriate response is promoted by certain environmental factors. Both diseases manifest themselves primarily in the gastrointestinal tract but can affect the entire human body (Baumgart and Carding 2007; Baumgart and Sandborn 2007).

The circulation of leukocytes including lymphocytes, monocytes, macrophages, dendritic cells, and granulocytes from the peripheral circulation (blood pool) to other tissues including the gut mucosa is an important aspect of immune surveillance (Delves and Roitt 2000b; Delves and Roitt 2000a; McIntyre et al. 2003). Endothelial cells represent a physical barrier between the blood and (mucosal) tissue, which must be crossed by blood leukocytes. To ensure a rapid movement of leukocytes through the endothelial barrier, endothelial cells interact with leukocytes after their respective activation by cytokines or other pro-inflammatory stimuli. This activation process is facilitated by the interaction of integrins as well as chemokine receptors with their respective endothelial and mucosal ligands immunoglobulin superfamily cellular adhesion molecules and chemokines (Charo and Ransohoff 2006; Hynes 2002).

A hallmark of all chronic inflammatory disorders including Crohn’s disease and ulcerative colitis is the rapid recruitment and often inappropriate retention of leukocytes, particularly T-cells to and at the site(s) of inflammation (Baumgart and Carding 2007; von Andrian and Mackay 2000). Recent basic and clinical research indicates that targeting molecules involved in leukocyte recruitment and retention may be a promising therapeutic strategy in chronic inflammatory disorders such as inflammatory bowel disease and multiple sclerosis (von Andrian and Engelhardt 2003). Here, we focus on the current understanding of leukocyte migration and adhesion in the gut mucosa relevant to inflammatory bowel disease and summarize the latest clinical trial data in this area.

Adhesion molecules

Integrins and integrin ligands


Integrins are a family of α, β heterodimeric transmembrane receptors that support cell–cell and cell–ECM (extracellular matrix) interactions (Hynes 2002; van der Flier and Sonnenberg 2001). They are transmembrane proteins that are constitutively expressed, but require activation in order to bind their ligand. The integrin family can form at least 24 different pairings existing of 18 α-subunits and 8 β-subunits, whose different combination result in a wide variety of adherence specificity (Hynes 2002; Shimaoka and Springer 2003). However, the associations between α and β subunits are restricted. The α subunit determines the specificity of the integrin ligand whereas the β subunit is connected to the cytoskeleton and affects multiple signaling pathways (Barczyk et al. 2010). Very important for the interaction between leukocytes and the endothelium is the β2 subfamily, especially leukocyte function associated antigen 1 (LFA-1, CD11a/CD18) and macrophage 1 antigen (Mac-1, CD11b/CD18). They are responsible for a stable adherence on the endothelium after they have achieved the high-affinity state by a change of conformation. CD11a/CD18 is normally expressed on the surface of most leukocytes and interact with intercellular adhesion molecules ICAM-1 and ICAM-2 (Panes et al. 1999). Another subfamily consists of the β1 (CD29) chain. One major integrin of this family is the α4β7 integrin (VLA-4) which is involved in the adhesion process as well. The vascular cell adhesion molecule (VCAM-1) is a ligand for α4β7 integrin.

Integrin receptors

Integrin receptors belong to the immunoglobulin (Ig) superfamily of adhesion molecules. They are distinguished by possessing multiple Ig-like domains. These Ig superfamily CAMs (IgSF CAMs) are a class of cell adhesion receptors and they are either homophilic or heterophilic. Most integrin receptors have the ability to bind a wide variety of ligands. Moreover, many extracellular matrix and cell surface adhesion proteins bind to multiple integrin receptors (Humphries 1990; Plow et al. 2000; van der Flier and Sonnenberg 2001). Some of the members of this IgSF CAMs like ICAM-1, VCAM-1 and the mucosal addressin MAdCAM-1 are involved in leukocyte–endothelial cell interactions.

Immunoglobulin superfamily adhesion molecules are evolutionarily ancient and widely expressed. One or more repeats of Ig fold 60–100 amino acids and form sites of adhesion. At the Ig domain, no somatic hypermutations are found. Moreover, sandwiches of two β sheets are held together by hydrophobic interactions (Table 1).
Table 1

Major human integrin subunits and important ligands



Cellular expression

Important ligands


CD49a, VLA1




CD49b, VLA2




CD49c, VLA3




CD49d, VLA4


Fibronectin VCAM-1 (CD106)


CD49e, VLA5




CD49f, VLA6








Fibronectin, vitronectin, nephronectin



Tenascin-C, VEGF-C, VEGF-D

















ICAM-1 (CD56), ICAM-2 (CD102), ICAM-3, ICAM-5




ICAM-1, fibrinogen




iC3b, fibrinogen + more


CD41, GpIIb


Fibrinogen, fibronectin



Vitronectin, fibronectin, fibrinogen












Fibronectin, TGF-β-LAP




MAdCAM-1, VCAM-1 fibronectin


CD103, HML-1






Vitronectin, TGF-β-LAP

Characteristics of the members of the integrin family and how they combine to form the 24 heterodimeric integrins. To form a heterodimer, integrin subunits bind to each other by distinct ligand-binding specificity, tissue and cell distribution

ICAM intercellular adhesion molecule, VCAM vascular cell adhesion molecule, VEGF vascular endothelial growth factor, VLA very late activation factor

Bold formatting denotes integrins and ligands relevant for gut mucosal leucocyte migration and adhesion in inflammatory bowel disease. Adapted from (Hynes 2002)

Chemokines and chemokine receptors


Chemokines are small secreted polypeptides that direct the movement of circulating leukocytes to the sites of inflammation. They are released from a wide variety of cell types. More than 50 chemokines have been identified that are divided into 4 families based on structural and functional differences (CC chemokines, CXC chemokines, CX3C chemokines, XCL1 chemokine) (Cyster 2005; Gerard and Rollins 2001; Handel and Domaille 1996; Luster 1998; Rot and von Andrian 2004; Zlotnik and Yoshie 2000). CC chemokines consisting of two adjacent cysteines attract mononuclear cells to the site of inflammation. In CXC chemokines two cysteines are separated by another amino acid, called X. For CX3C chemokines only one member (CX3CL1) is known in which the chemokine domain is fused to a mucin-like stalk. The fourth group consists of lymphotactin (XCL1) shows a single cysteine residue. Chemokines mediate their function by binding to a subfamily of seven transmembrane, G protein-coupled receptors, which are expressed at the surface of the plasma membrane of leukocytes (Charo and Ransohoff 2006).

Chemokine receptors

Another important mechanism for T-cell homing to the intestine is the expression of chemokine receptors on the surface of lymphocytes and the chemoattraction offered by these chemokines which is an important process for the protection against infectious agents (Mora 2008; von Andrian and Mackay 2000). However, chemokines might also affect the body by maintaining and amplifying chronic inflammation due to a late removal or neutralization of invading agents or sustaining chronic responses against self-antigens which are responsible for the development of autoimmune diseases. More than 18 chemokine receptors have been identified so far (Kim and Broxmeyer 1999; Zlotnik and Yoshie 2000) (Table 2).
Table 2

Major human chemokines and chemokine receptors


Cellular expression


Chemokine receptor(s)

C chemokines









CC chemokines






Site of acute inflammation




Site of acute inflammation




Site of acute inflammation




Site of acute inflammation
















































Lymph nodes








Lymph nodes
















Small bowel lamina propria












Small bowel lamina propria, colon lamina propria



CXC chemokines






































Site of acute inflammation




Site of acute inflammation




Site of acute inflammation




Lymph nodes, Peyer’s patches




Peyer’s patches










CX3C chemokines





Adapted from (Charo and Ransohoff 2006)

Adhesion molecule expression and function in the normal gut mucosa

A crucial feature of any inflammatory response is a rapid recruitment of leukocytes from the blood to the site of inflammation (Delves and Roitt 2000a, b). This is facilitated through post-capillary high endothelial venules (HEV) and it requires leukocytes to migrate through the blood vessels and enter tissues by a multistep mechanism which is known as extravasation (Muller 2003; Yadav et al. 2003). Pivotal steps during extravasation include the initial attachment and rolling of the leukocytes along the activated endothelium, chemokine-mediated activation of special integrins on the surface of leukocytes, stable adherence of the activated leukocytes to the endothelium, degradation of the subendothelial basement membrane and finally the migration of leukocytes along chemokine gradients into the inflamed compartment (Parish 2006) (Figs. 1, 2).
Fig. 1

Adhesion molecule distribution and function in the healthy mucosa. There is great diversity in the expression of migration, adhesion and homing relevant receptors and ligands. Mucosal epithelial cells express the chemokine CCL25 in decreasing frequency from the proximal to the distal small intestine. CXCL12 and CCL20 are preferably expressed in Peyer’s patches. Moreover, the expression of the chemokines CX3CL1, CXCL10 and CXCL12 have been described. Unlike in the small intestine colonic mucosal epithelial cells express the chemokine CCL28 and CXCL12 throughout the large intestine. Moreover, the expression of the chemokines CX3CL1 and CXCL10 have been described. The immunoglobulin superfamily addressins MAdCAM-1, ICAM-1 and to a lesser degree VCAM-1 are expressed throughout the entire small intestine, while in the colon only MAdCAM-1 expression has been reported so far. Different mucosal leukocyte populations use different integrins and are chemoattracted along different chemokine gradients and pathways. Most small bowel Lamina propria T-cells (TH17, TREG, CD8+) are known to express the integrin α4β7 and chemokine receptor CCR9 and their migration and homing depends mainly on the interaction with its ligands MAdCAM-1 (expressed on venules) and ICAM-1 (on endothelial and epithelial cells as well as leukocytes) and CCL25, respectively. In the colon however, Lamina propria T-cells express the chemokine receptor CCR10, but their migration and homing depends mainly on the interaction with its ligands MAdCAM-1, ICAM-1 and probably CCL28, respectively. Some of these cells change their phenotype from α4β7 to αEβ7 when leaving the high endothelial venules to facilitate adhesion to E-cadherin expressing mucosal epithelial cells. CD4+ T-cells are largely independent of CCR9–CCL25 (CCR10–CCL28 in the colon) regulated homing. They may use other chemokine receptors CXCR3, CXCR4, or CCR5 and follow CXCL10, CXCL12, CCL5 gradients. Alternative homing mechanisms involving CCR4 and CCR7 have also been described for TREG. Intraepithelial lymphocytes (IEL) have been reported to express the integrin αEβ7 the chemokine receptors CCR2, CXCR3 and CX3CR1 that could impact on their homing. αEβ7, CX3CR1, CCR7 may also be used by certain dendritic cells known to be critically involved in mucosal innate and adaptive immune regulation

Fig. 2

Adhesion molecule distribution and function and drug targets in the inflamed mucosa. The tightly regulated interplay between receptors and ligands involved in leukocyte migration, adhesion and homing is disturbed in Crohn’s disease and ulcerative colitis. The inflamed mucosa expresses greater amounts of E-selectin, ICAM-1, VCAM-1 and MAdCAM-1 and extracellular matrix proteins such as fibronectin, driven by pro-inflammatory cytokines such as TNF-α and/or LPS. Leucocyte retention is mediated by binding of α4β7 expressing lymphocytes to MAdCAM-1 and VCAM-1 via the β7 subunits or fibronectin as well as α4β1 expressing leucocytes to VCAM-1 or fibronectin. Pharmacological compounds targeting α4, α4β7, ICAM-1, and CCR9 have been evaluated in human Crohn’s disease and β7, α4β7, MAdCAM-1 and ICAM-1 in ulcerative colitis, respectively

Small bowel

There is great diversity in the expression of migration, adhesion and homing relevant receptors and ligands. The IgSF CAMs MAdCAM-1, VCAM-1 and ICAM-1 are expressed throughout the entire small intestine and are shown to be involved in leukocyte-endothelial cell interactions. Most Lamina propria T-cells (TH17, TREG, CD8+) are known to express the integrin α4β7 and chemokine receptor CCR9 and CCL25, respectively.


The interaction of the α4β7 integrin which mediates the infiltration of the GI tract by memory T-cells and the mucosal addressin cell adhesion molecule (MAdCAM-1), an endothelial cell receptor, is believed to contribute to the chronic bowel inflammation (Feagan et al. 2005, 2008; Hesterberg et al. 1996; Picarella et al. 1997). MAdCAM-1 is primarily expressed in high endothelial venules in the small intestine, in the Peyer’s patches and the colon. Its interaction with β7 mediates lymphocyte recirculation under normal conditions. Finally, it is critical for lymphocyte homing to lymphatic tissues in the Peyer’s patches and the recruitment of these cells into the intestine during the process of inflammation. Patients suffering from CD show an increase expression of addressin at the sites of active inflammation of the bowel. Moreover, the α4β7 integrin also binds to VCAM-1 and fibronectin (Berlin et al. 1995). MAdCAM-1 is a ligand for L-selectin as well as α4β7 integrin. The contribution of β7 and VCAM-1 interactions to lymphocyte recruitment may only occur during inflammation (Panes and Granger 1998).

VCAM-1 vascular cell adhesion molecule

The vascular cell adhesion molecule VCAM-1 is an important regulator of lymphocyte trafficking. It binds α4β1 integrin as well as the α4β7 integrin. In unstimulated human umbilical vein endothelial cells (HUVEC) the adhesion molecule VCAM-1 is absent. However, a transcription-dependent up-regulation by LPS or specific cytokines can be initiated in these cell types. Animal work demonstrated that the expression of VCAM-1 is lower compared with ICAM-1 (Henninger et al. 1997).

Intercellular cell adhesion molecule (ICAM-1)

The intercellular adhesion molecule ICAM-1 is expressed on leukocytes, dendritic cells, fibroblasts, epithelial cells and endothelial cells and it recognizes CD11a/CD18 and CD11b/CD18. Activation of these cells with cytokines or LPS induces an increased expression of ICAM-1. Studies by the groups of Henninger and Panés showed that organs in the GI tract possess a higher endothelial expression of ICAM-1 after stimulation with LPS or TNF-α, respectively (Henninger et al. 1997; Panes et al. 1995; Panes and Granger 1996).


Rivera-Nieves et al. demonstrated that CCR9 plays an important role in the inflamed intestine. The authors were able to reduce early chronic ileitis in mice by blocking CCL25/CCR9 (Rivera-Nieves et al. 2006). Small bowel is mediated through chemokine ligand CCL25 which is also known as thymus-expressed chemokine (TECK) (Kunkel et al. 2003). Ligation of CCR9 by CCL25 triggers conformational changes in the α4β7 integrins and finally a stable adhesion of MAdCAM-1. Overall, lymphocytes in the small bowel express α4β7 over 90%. That high amount of α4β7 expression reflects the expression of CCR9 (Eksteen et al. 2004).


Other important chemokine receptors that are expressed during the process of homing in the small intestine include CCR2, CCR4, CCR5, and CCR7, as well as the presence of CXCR3 and CX3CR1. Their expression is not mainly important for lymphocyte homing. In fact, they are responsible to specify an activated pro-inflammatory phenotype, i.e. CCR5 and CXCR3 are associated with Th1 immune responses.


The expression of adhesion molecules in the colon is characterized to a lesser extent compared to the well-characterized adhesion molecule profile in the small bowel.

MAdCAM-1, ICAM-1, and VCAM-1

In the colon, the expression of MAdCAM-1 is much lower compared with the small bowel. Therefore, it is convincing that adhesion molecules like ICAM-1 and VCAM-1 are also cooperating in this process. In mice, Kato et al. (2000) were able to demonstrate that VCAM-1 mediates adhesion blockade and significantly improve inflammation. On the other hand, the blockade of MAdCAM-1 was less successful.


The chemokine receptor CCR10 is expressed on homing lymphocytes in the colon. Furthermore, CCL28, also known as the mucosal-expressed chemokine (MEC) which is a ligand for CCR10, is verifiable on colonic vessels and epithelial cells. CCL28 is present during physiological as well as inflammatory states of events (Pan et al. 2000).

Data from animal models of inflammatory bowel disease and from human studies in Crohn’s disease and ulcerative colitis patients

One of the well-documented pathogenic mechanisms in IBD is the infiltration of the gastrointestinal tract by T-lymphocytes and the molecular mechanism by which these cells enter the gut. These processes are distinct from those in the skin or the central nervous system (Agace 2006; Butcher and Picker 1996; Engelhardt et al. 1998; Engelhardt 1998; Salmi and Jalkanen 2005). Leukocyte infiltration requires a coordinated interaction of several adhesion and signaling molecules on the surface of T-cells and their corresponding ligands on the endothelium.

It has been postulated that the αEβ7 integrin locates and retains T-lymphocytes within the epithelium of numerous tissues by binding E-cadherin on the basolateral surface of epithelial cells (Hynes 2002). The expression of αEβ7 is known to be increased during inflammation, especially when T-cells infiltrate the epithelial tissue. Elewaut et al. (1998) demonstrated that patients suffering from Crohn’s disease shown a reduced expression of αEβ7 on intraepithelial lymphocytes (IELs) in the ileum. This change is seen in inflamed and non-inflamed regions, respectively, and might be an early sign of incipient disease. In addition, it could be shown in mouse models of colitis, that injection of CD4+CD45RBhi cells into severe combined immunodeficient (SCID) mice leads to an infiltration of T-cells in the epithelium and that these cells are αEβ7 positive (Aranda et al. 1997). In other autoimmune diseases like Sjögren’s syndrome effector CD8+ T-cells that are found in close association with the acinar and ductal epithelium express αEβ7 (Kroneld et al. 1998). A boost in the expression of αEβ7 is also seen in cultured peripheral blood T-cells in patients suffering from systemic lupus erythematosus (SLE) with epithelial involvement, and in cutaneous T-cell lymphomas (Dietz et al. 1996; Pang et al. 1998). Furthermore, the expression of αEβ7, particularly on CD4+ cells, increases dramatically in specific types of pulmonary inflammation, e.g. sarcoidosis, hypersensitivity pneumonitis, and idiopathic pulmonary fibrosis (Rihs et al. 1996). αEβ7 has also been detected in the inflamed synovium. It could be shown that E-cadherin is strongly expressed by synovocytes (Trollmo et al. 1996). αEβ7+ T-cells that infiltrate non-mucosal organs are sometimes linked with the gut. A more plausible explanation would be that the expression of αEβ7 is induced in situ by the influence of TGF-β in the local microenvironment. Another study has shown that a blocking antibody to the αE called M290, prevents or ameliorates immunization of induced colitis in IL-2 deficient mice (Ludviksson et al. 1999). This effect was traced back to the reduced maintenance of CD4+ cells in the mucosa. Another possibility is a direct damage of the epithelium mediated by T-cells where an inhibition of the interaction between αEβ7 and its ligand could be beneficial.

Moreover, a subset of dendritic cells in the intestine or associated lymphoid tissue express αEβ7 (CD103+CD11c+) (Iliev et al. 2007; Schlickum et al. 2008). It could be shown in mice that CD25+CD4+ expressing αEβ7 have an increased protective phenotype (Lehmann et al. 2002). Uss et al. (2006) were able to describe a regulatory role for CD8+CD103high T-cells in human (Uss et al. 2006).

It is unknown whether αEβ7 has a regulatory effect of these cells, or if it represents a marker for a cell subset. Work with αE-deficient mice has shown that expression of αEβ7 T cells is not required for their accumulation to the intestine and it is not necessary for mesenteric lymph node dendritic cells to induce gut-tropic receptors on T cells, e.g. α4β7 or CCR9 (Feagan et al. 2005; Jaensson et al. 2008; Lefrancois et al. 1999). Altogether, it appears that blocking αEβ7 results in a potential target of IEL, LPL, tolerogenic dendritic cells or Treg. However, the consequences of this blocking step need to be investigated in further trials. Ludviksson et al. (1999) have shown that blocking of αEβ7 in IL2−/− colitis mice that were immunized with TNP-OVA before inhibits the localization of αEβ7-expressing CD4+ cells in the LP. This prevents and ameliorates an already established colitis in these mice. However, Annacker et al. (2005) showed in a transfer model, that CD4+CD45RBhi naïve T cells from αE−/− mice induce colitis in SCID mice when compare to wild type mice. Furthermore, the complete role of CD4+αEβ7+ as regulatory T cells totally depends on the model used. It could be shown by several authors that these cells might protect from disease induced by CD4+CD45RBhi naïve T cells transfer into SCID mice (Annacker et al. 2005; Lehmann et al. 2002). However, this will not work when the cells are transferred into SAMP1/YitFc ileitis mice (Olson et al. 2004).

Natalizumab may block the interaction between the α4β7 integrin and MAdCAM-1 at the sites of inflammation. In animal models (TNBS-, dextran sulphate- and peptidoglycan/ polysaccharide-induced colitis and IL-10 knockout mice) researchers were able to demonstrate that mice with inflammatory bowel disease show higher levels of VCAM-1 expression (Conner et al. 1997; Kawachi et al. 2000; Sans et al. 1999; Soriano et al. 2000). In these animal models, ICAM-1 expression was not increased. However, they all showed an equal increase in their expression level of MAdCAM-1. In the severe combined immunodeficient (SCID) mouse model as well as in IL-2 knockout mice, a higher expression level of MAdCAM-1 could be detected in the colonic submucosa as well as in the lamina propria venules (McDonald et al. 1997; Picarella et al. 1997).

Pooley et al. (1995) have demonstrated that the culture supernatant of colonic biopsies taken from patients suffering from ulcerative colitis or Crohn’s disease induce an up-regulation of E-selectin and ICAM-1 in endothelial cells. However, there is a difference between the two diseases.

In immunohistochemistry studies, an increase of adhesion molecules in IBD patients was shown. These patients show an increase of P- and E-selectin expression in venules or capillaries in the inflamed tissue (Cellier et al. 1997; Koizumi et al. 1992; Nakamura et al. 1993; Oshitani et al. 1995). Investigations regarding ICAM-1 expression in human IBD are showing discrepant results. Initial work showing an increase of ICAM-1 could not be confirmed by later examinations (Cellier et al. 1997; Koizumi et al. 1992; Nakamura et al. 1993; Oshitani et al. 1995). Further it could be shown, that there is an up-regulation of MAdCAM-1 in IBD patients compared to normal tissue (Briskin et al. 1997). The increase of VCAM-1 expression, which could be seen in various animal models, could not be confirmed in the intestinal mucosa of IBD patients. Here the patients show a similar expression pattern of VCAM-1 compared to that of controls (Cellier et al. 1997; Koizumi et al. 1992; Oshitani et al. 1995).

Hence, the α4β7 integrin looks like an ideal therapeutic target for IBD. This integrin is currently targeted through three different strategies. Two of these strategies target either the α4 or the β7 chain which are not specific for the α4β7 integrin and bind to other integrins containing these chains. The α4β1 integrin mediates extravasation of lymphocytes, monocytes, and eosinophils into numerous types of tissues. Thus, they bind to VCAM-1 which is expressed on the luminal surface of the endothelium, and to fibronectin within the extracellular matrix (Gonzalez-Amaro et al. 2005). The humanized natalizumab induces effects in multiple tissues, including leukocytosis, the mobilization of hematopoietic stem cells, and the inhibition of leukocyte trafficking into the CNS (Bonig et al. 2008; del Pilar et al. 2008; Ghosh et al. 2003; Sandborn et al. 2005; Targan et al. 2007; Zohren et al. 2008). Another strategy is exclusive targeting of the α4β7 integrin, which is used by vedolizumab. As mentioned, vedolizumab is a humanized version of Act-1, a mouse antibody (Lazarovits et al. 1984). It binds to a conformational epitope which is unique to the heterodimerization of the human α4 chain with the β7 chain (Schweighoffer et al. 1993; Tidswell et al. 1997). Act-1 binds specifically to the α4β7 integrin. It has been shown by Hesterberg et al. (1996) that the administration to colitic cotton-top tamarins leads to a resolution of disease Hesterberg et al. (1996). Vedolizumab is able to bind to the α4β7 integrin on peripheral blood lymphocytes and finally inhibits adhesion of the lymphocyte to MAdCAM-1 (Feagan et al. 2005, 2008).

Pharmacological targeting of leukocyte recruitment

Current approved biologic therapies in most jurisdictions are limited to monoclonal antibodies directed at tumor necrosis factor alpha (TNF-α) (Baumgart and Sandborn 2007). While these agents have substantially improved and expanded treatment of patients particularly those with Crohn’s disease, a considerable proportion of patients are primary non-responders, develop neutralizing antibodies during their administration, do not tolerate these therapies or experience dangerous side effects (Allez et al. 2010; Lee and Fedorak 2010; Peyrin-Biroulet et al. 2008). Thus, there is still an unmet need for new therapies. While we will be reporting detailed outcomes in the next section, there is no universally agreed and validated outcome criteria and the statistically significant differences in endpoints may not always be clinically meaningful or relevant. This makes a comparison between trials and compounds very difficult.

Antibodies targeting Integrins

Natalizumab (anti-α4)

Natalizumab (Tysabri®, Antegren® or anti-α4) is a recombinant humanized monoclonal IgG4 antibody directed at the integrin subunit α4. Natalizumab systemically blocks the integrins α4β7 and α4β1. It is the first and currently only approved drug developed in the class of selective adhesion molecule inhibitors for Crohn’s disease respectively (Sandborn and Yednock 2003).

First Gordon et al. (2001) investigated the administration of natalizumab in 30 patients with active CD with a Crohn’s Disease Activity Index (CDAI) of ≥151 and ≤450 in a double-blind randomized trial. The patients received a 3-mg/kg infusion of the drug or a placebo that was well tolerated by CD patients. After 2 weeks, remission was induced and the CDAI decreased significantly from baseline after infusion of natalizumab (mean 45 points) but not in the placebo group (mean 11 points). Seven patients (37%) from the natalizumab-treated group achieved remission at week 2, compared with 1 patient (8%) treated with the placebo. Interestingly, significant increases in circulating B and T lymphocytes were detected only after the administration of natalizumab which suggests an interrupted lymphocyte trafficking. The frequency of commonly reported adverse events did not differ significantly between groups.

In another double-blind, placebo-controlled trial of natalizumab, 248 patients with moderate-to-severe CD were investigated (Ghosh et al. 2003). The patients for this study were randomly assigned to receive one of four treatments: two infusions of placebo; one infusion of natalizumab 3 mg/kg, followed by placebo treatment; two infusions of natalizumab 3 mg/kg; two infusions of natalizumab 6 mg/kg. Infusions were given 4 weeks apart. At week 6, the estimated defined primary end point in the efficacy analysis, the group obtained two infusions of 6 mg/kg of natalizumab did not have a significantly higher rate of clinical remission (defined by a score of less than 150 on the CDAI) than the placebo group. However, both patient groups that received two infusions of the drug showed higher remission rates than the placebo group at multiple time points. Natalizumab also produced a significant improvement in response rates. The highest remission rate was 44%. However, the highest clinical response rate was 71% and it was documented at week 6 in the group given two infusions of 3 mg/kg. In this study, a treatment of natalizumab increased the rates of clinical remission and response, improved the quality of life and C-reactive protein levels, and was well tolerated in patients with active CD.

In another controlled trial natalizumab was administered as an induction and maintenance therapy in patients with active CD (Sandborn et al. 2005). In this large first trial, 905 patients with moderate or severe active CD were randomly assigned. The patients received either 300 mg of natalizumab (724 patients) or a placebo (181 patients) at weeks 0, 4, and 8. At week 10, the primary outcome was response that was defined by a decrease in the CDAI score of at least 70 points. In a second trial, 339 patients who had shown a response to the drug during the first trial were randomly reassigned to receive 300 mg of natalizumab or placebo every four weeks through week 56. Now the primary outcome was a sustained response through week 36 and a secondary outcome in both trials was disease remission (CDAI score of less than 150). During the first trial, the response and remission rates were 56 and 37% in the natalizumab-treated group compared with 37 and 30% in the group that received the placebo at week 10 (P = 0.05 and P = 0.12). Continuing the administration of natalizumab in the second trial resulted in higher rates of sustained response (61 vs. 28%, P < 0.001) and remission (44 vs. 26%, P = 0.003) through week 36 than did a switch to the placebo. 7% of the patients developed serious adverse events in the first trial and in 10% of the placebo group and 8% of the natalizumab group in the second trial.

While the previous studies have shown evidence for the efficacy of natalizumab in the treatment of active CD, all trials failed to show statistically significant differences at the defined end points of the studies. Another study was conducted in which 509 patients with moderate to severe active CD and an active inflammation were characterized by elevated C-reactive protein concentrations (Sandborn et al. 2005; Targan et al. 2007). The patients in this study by Targan and colleagues were randomized (1:1) to receive natalizumab 300 mg or a placebo intravenously at weeks 0, 4, and 8. The primary end point was induction of response that was defined as >70-point decrease from baseline in CDAI at week 8 sustained through week 12. A response at week 8 sustained through week 12 occurred in 48% of all natalizumab-treated patients and in 32% of patients receiving placebo (P < 0.001). Sustained remission occurred in 26% of natalizumab-treated patients and 16% of patients receiving placebo (P = 0.002). The response rates at week 4 were 51% for natalizumab and 37% for placebo (P = 0.001). All responses remained significantly higher at subsequent assessments (P < 0.001) in natalizumab-treated patients compared to the placebo group. Furthermore, natalizumab-treated patients also had significantly higher remission rates at weeks 4, 8, and 12 (P < 0.009).

In the previous trials, the efficacy of natalizumab treatment was investigated for patients suffering from CD. Data evaluating the effectiveness in ulcerative colitis (UC) is rare. However, Gordon et al were also investigating this topic in a trial with ten patients with active UC defined by a Powell-Tuck activity score (>4) to assess the safety and effect of Antegren. The patients received a single infusion of natalizumab (3 mg/kg). The median Powell–Tuck score showed a significant decrease at 2 and 4 weeks of post-infusion (7.5 and 6) compared to the median baseline score (Bekker et al. 2007). Five patients showed a good clinical response at weeks 2 and another one at week 4. Moreover, the median C-reactive protein at 2 weeks (6 mg/L) was lower compared to the pre-treatment (16 mg/L). One patient remained in remission at 12 weeks. During the trial, side effects were infrequent and probably not related to the drug treatment. The authors showed that a single infusion of natalizumab was well tolerated in UC patients. However, the positive efficacy of this study needs further investigation by randomized, placebo-controlled trials (Gordon et al. 2002).

AJM300 (anti-α4)

AJM300 (anti-α4) is an orally active small molecule directed at the α4 integrin subunit. It was evaluated in the TNBS (Ito et al. 2005), DSS (Maruyama et al. 2007) and adoptive transfer animal models (Stefanich et al. 2011; Sugiura et al. 2009) of inflammatory bowel disease. AJM300 was also evaluated in one randomized, double-blind, placebo controlled trial in patients with active Crohn’s disease in Japan (Takazoe et al. 2009). In this study Takazoe and colleagues randomized 71 Crohn`s disease patients to receive placebo, AJM300 40 mg TID, 120 mg TID, or 240 mg TID orally for 8 weeks. Taken together, in patients receiving AJM300, the decrease of CDAI (40 mg: 19.9 ± 74.1, 120 mg: 25.5 ± 61.3, 240 mg: 21.6 ± 84.9; mean ± SD) was higher compared to the placebo group (5.2 ± 71.0). The patients who showed a CDAI ≥ 200 at week 0, present a decrease of CDAI from baseline with 41.5 ± 57.5 in the 120 mg group (P = 0.0485) and 41.6 ± 94.1 in the 240 mg group. The clinical response rate was 50% for the 240 mg group at the final evaluation. Furthermore, in the 240 mg group the CRP showed a decrease from 1.87 mg/dL at week 0 to 0.96 mg/dL at week 8 (P = 0.0220). Altogether, the group was able to show that AJM300 was effective in active Crohn’s disease at doses 120 mg and 240 mg TID. Moreover, AJM300 is safe and well tolerated in this clinical trial.

Etrolizumab (anti-β7)

Etrolizumab (rhuMAb-Beta7 or anti-β7) is a humanized monoclonal IgG1 antibody targeting the integrin subunit β7. It blocks both α4β7 and αEβ7 integrins. The effect of murine anti-αβ7 on lymphocyte homing in mouse models of autoimmune disease as well as it’s effects on circulating mucosal-homing versus peripheral-homing T cells in naive non-human primates were recently evaluated. In cynomolgus monkeys, occupancy of β7 integrin receptors by anti-β7 correlated with an increase in circulating β7+ mucosal-homing lymphocytes, with no apparent effect on levels of circulating β7 peripheral-homing lymphocytes. It also inhibited lymphocyte homing to the inflamed colons in severe combined immunodeficient (SCID) and CD45RBhighCD4+ T-cell transfer models of inflammatory bowel disease. Consistent with a lack of effect on peripheral homing, in a mouse model of experimental autoimmune encephalomyelitis, anti- anti-β7 treatment resulted in no amelioration of CNS inflammation (Stefanich et al. 2011).

The safety and pharmacology of etrolizumab, was evaluated in 48 patients with moderate to severe UC in a double-blind randomized within cohort, placebo-controlled study included a single-ascending dose (SAD) stage followed by a multidose (MD) stage and enrolled outpatients with a Mayo Clinic Score (MCS) ≥ 5. A single dose of etrolizumab (0.3, 1.0, 3.0, 10 mg/kg IV, 3.0 mg/kg SC or placebo), was given in a 4:1 ratio to each group. This was followed by a MD stage in a different group of patients in which 3 doses of etrolizumab 0.3, 1.5, 3.0 mg/kg SC, 4.0 mg/kg IV or placebo were given monthly in a 4:1 ratio in each group. Eligible patients receiving concomitant 5-ASA +/− steroids +/− immunosuppressants at entry stayed on stable doses. Concomitant steroids could be tapered 6 weeks after the first dose of study drug. MCS was evaluated 28 days after study drug administration in the SAD and at day 43 and day 71 in the MD stage. In the SAD arm, there were no dose limiting toxicities, no infusion or injection site reactions. Anti-Etrolizumab antibodies were detected at day 29 in a single patient in the lowest dose group. There were two SAEs related to impaired wound healing in two patients receiving active treatment who underwent urgent colectomy for exacerbation of UC, however, there were significant confounding factors. In the MD arm, 65% of patients had a history of treatment with anti-TNFs. There were no dose limiting toxicities, no infusion or injection site reactions. Headache was the most common adverse event occurring more often in actively treated patients in the higher dose groups. Clinical response (a decrease in MCS of 3 points + 30% reduction from baseline and a ≥1 point decrease in rectal bleeding or absolute bleeding score of 0/1) and clinical remission (an absolute MCS ≤ 2 + no individual subscore >1) at Day 71 was observed in 10/15 patients and 3/15 patients treated with etrolizumab respectively compared with 2/3 and 0/3 patients treated with placebo (Rutgeerts et al. 2011). This study suggests an early signs of clinical activity in moderate to severe UC and etrolizumab will be further developed in inflammatory bowel disease.

Vedolizumab (anti-α4β7)

Vedolizumab (MLN-02, LDP-02, MLN0002, or anti-α4β7) is a humanized monoclonal IgG1 antibody directed at integrin α4β7. Thus, its activity is focused on α4β7 MAdCAM-1 binding. Due to MAdCAM-1s virtually exclusive expression in the gastrointestinal tract it may have advantages over natalizumab systemic immunosuppressive effects (Soler et al. 2009). Two phase II trials evaluating the compound in patients with Crohn’s disease and ulcerative colitis have been published (Feagan et al. 2005, 2008).

The first phase II trial was a multicenter, double blind, placebo controlled study in 181 UC patients (Feagan et al. 2005). It looked adult UC patients with naïve to biologic therapy. Eligible patients were assigned randomly in a 1:1:1 ratio to receive vedolizumab 2.0 mg/kg, vedolizumab 0.5 mg/kg, or placebo. The randomization procedure was stratified by concomitant mesalamine use. Each patient received an intravenous infusion of study drug on days 1 and 29. The primary outcome measure was clinical remission at week 6. Secondary outcomes were the changes in the ulcerative colitis clinical scores, the modified Baron scores, the Riley scores, and the scores on the inflammatory bowel disease questionnaire. The study also evaluated the proportion of patients with clinical and endoscopic response at week 4 and week 6. Of 181 patients who underwent randomization, 58 were assigned to receive 0.5 mg of study drug per kilogram, 60 to receive 2.0 mg of study drug per kilogram, and 63 to receive placebo. After 6 weeks, the proportion of patients in clinical remission differed significantly among the three groups. In the group receiving 0.5 mg/kg of vedolizumab, 19 of 58 patients (33%) achieved remission, as compared with 19 of 60 (32 %) in the group receiving 2.0 mg/kg and 9 of 63 (14%) in the placebo group (overall P = 0.03). 28% of patients receiving 0.5 mg/kg and 12% of those receiving 2.0 mg/kg had endoscopically evident remission, as compared with 8% of those receiving placebo (P = 0.007). For the minority of patients in whom a vedolizumab antibody titer greater than 1:125 developed, incomplete saturation of the α4β7 receptor on circulating lymphocytes was observed and no benefit of treatment was identifiable. In this short-term study, vedolizumab was more effective than placebo for the induction of clinical and endoscopic remission in patients with active UC. There were no substantial differences among the three groups in the prevalence of adverse events.

The second phase II trial was also a multicenter, double blind, placebo controlled study in 185 CD patients (Feagan et al. 2008). It looked at adult CD patients with CD of the ileum and/or colon, naïve to biologic therapy. This study comprised three periods: screening, treatment, and observation. Moreover, safety data were gathered during a post-study follow-up period of up to 2 years. Eligible patients were assigned randomly in a 1:1:1 ratio to receive vedolizumab 2.0 mg/kg, vedolizumab 0.5 mg/kg, or placebo. The randomization procedure was stratified by concomitant mesalamine use. Each patient received an intravenous infusion of study drug on days 1 and 29. At clinic visits, the CDAI and inflammatory bowel disease questionnaire (IBDQ) scores were determined. The primary end point was the rate of clinical response at day 57. Secondary efficacy end points were the rate of clinical remission, the time to clinical response and clinical remission, the change in mean CDAI and IBDQ scores, the change in C-reactive protein (CRP), and the rate of treatment failure. Of 185 patients randomized, 65, 62, and 58 received vedolizumab 2.0 mg/kg, vedolizumab 0.5 mg/kg, and placebo, respectively. Vedolizumab failed to meet the primary endpoint: the proportions of patients with a clinical response at day 57 were 53, 49, and 41% in the vedolizumab 2.0 mg/kg, vedolizumab 0.5-mg/kg, and placebo groups (differences not statistically significant). Secondary endpoints achieved included the proportions of patients achieving enhanced clinical response at day 57 (P = 0.05 for the 2.0 mg/kg vs. placebo) and the proportion of patients with a normal CRP at day 57 (P = 0.018 for the percentage change in the vedolizumab 2.0 mg/kg vs. placebo). There were no statistically significant differences in the IBDQ between treatment groups. In total, 92% of vedolizumab treated and 86% placebo-treated patients experienced at least one adverse event; 43 and 38%, respectively, were judged to be possibly study-drug-related. Severe adverse events were experienced by 29% vedolizumab-treated and 33% placebo-treated patients. No opportunistic infections [progressive multifocal leukoencephalopathy (PML) or otherwise] occurred during the study. Vedolizumab appears particularly promising in UC and is currently evaluated for both UC and CD in ongoing phase III trials (GEMINI) (Millennium Pharmaceuticals 2008, 2010a, b, 2011a, b, c, d).

Compounds targeting integrin receptors

PF-00547659 (anti-MAdCAM)

PF-00547659 (anti-MAdCAM) is a monoclonal IgG2 antibody directed at MAdCAM. Functional adhesion assays and surface plasmon resonance were used to characterize, in vitro, the pharmacological properties of PF-00547659. The in vivo effects of PF-00547659 on restriction of β7+ memory T-lymphocytes were determined in mice and macaques, respectively, over the pharmacological dose range to confirm PK/PD relationships. PF-00547659 was bound with high affinity to mouse and human MAdCAM. It blocked the adhesion of α4β7+ leukocytes to MAdCAM with similar potency. PF-00547659 induced a similar, dose-dependent two- to threefold increase in circulating populations of β7+ memory T-lymphocytes in both animal models; without affecting the β7 memory T-lymphocytes populations (Pullen et al. 2009).

The results of a first-in-human study aimed to explore the safety and preliminary efficacy in UC (Vermeire et al. 2011). In a multicenter, randomized, double-blind placebo-controlled study, 80 patients with active UC received single or multiple (3 doses, 4-week intervals) doses of PF-00547659 0.03-10&emsp14; mg/kg IV/SC, or placebo. Exploratory efficacy analyses were based on Mayo score and endoscopic responder rates at weeks 4 and 12. Fecal calprotectin was quantified as a measure of disease activity, and the number of α4β7+ lymphocytes was measured as a surrogate to demonstrate drug activity. No obvious drug-related side effects were observed in the PF-00547659 group. However, patient numbers, especially those fully exposed, were small. Overall responder/remission rates at 4 and 12 weeks were 52%/13% and 42%/22%, respectively with combined PF-00547659 doses compared with 32%/11% and 21%/0%, respectively with placebo. Equivalent endoscopic responder rates were 50 and 42% versus 26 and 29%, respectively. Fecal calprotectin levels decreased to a greater extent with PF-00547659 than placebo (week 4: 63 vs. 18%). Despite variability, there was a trend for an increase in α4β7+ lymphocytes in patients receiving PF-00547659. It is too early to determine the role of this agent in UC yet (Vermeire et al. 2011). The short and long-term safety are currently evaluated in two ongoing trials (Pfizer 2009; b, c; Pfizer 2011a).

ISIS 2302 (anti-ICAM-1)

ISIS 2302 (Alicaforsen, INXC ICAM-1, anti-ICAM-1) ISIS 2302, a phosphorothioate antisense oligonucleotide specific for human intracellular adhesion molecule-1 (Glover et al. 1997), initially showed some promise in a placebo controlled trial in Crohn’s disease (Yacyshyn et al. 1998), but failed to meet more definitive primary endpoints in subsequent (randomized, double-blind, placebo controlled, masked) trials in this indication (Schreiber et al. 2001; Yacyshyn et al. 1998, 2002). An enema formulation was evaluated for left sided ulcerative colitis and showed efficacy comparable to mesalamine (Miner, Jr. et al. 2006a, b; van Deventer et al. 2006). Phase III trials were never conducted and it is unknown if the compound will be further developed for IBD.

Compounds targeting chemokines and chemokine receptors

CCX282-B (anti-CCR9)

CCX282-B (GSK-1605786, Traficet-EN™ or anti-CCR9) is an orally bioavailable small molecule that selectively blocks the human CCR9 receptor. In early pharmacological studies CCX282-B inhibited CCR9-mediated Ca++ mobilization and chemotaxis on Molt-4 cells. In the presence of 100% human serum, CCX282-B inhibited CCR-9-mediated chemotaxis and the addition of α1-acid glycoprotein did not affect its potency. CCX282-B inhibited chemotaxis of primary CCR9 expressing cells to CCL25. CCX282-B was an equipotent inhibitor of CCL25-directed chemotaxis of both splice forms of CCR9 (CCR9A and CCR9B). CCX282-B also inhibited mouse and rat CCR9-mediated chemotaxis. Inhibition of CCR9 with CCX282-B resulted in normalization of experimental Crohn’s disease such as histopathology associated with the TNF(ΔARE) mice (Eksteen and Adams 2010; Ungashe et al. 2008; Walters et al. 2010).

To date CCX282-B has been studied in a pilot phase II (Bekker et al. 2007; Hetzel et al. 2009; Keshav et al. 2007a, b) and a multicenter, double-blind, placebo controlled, parallel group study (PROTECT-1) (Bekker et al. 2008, 2009; Keshav et al. 2007a, b, 2009; Williamson et al. 2010) in patients with moderate to severe Crohn’s disease. The PROTECT-1 study is currently followed-up by a phase III open label extension study (SHIELD-3). After the acquisition of ChemoCentryx by GlaxoSmithKline (GSK) a phase I trial investigated four new formulations of CCX282 now named GSK-1605786 A, B, C or D to assess bioavailability in healthy male and female subjects was completed (GlaxoSmithKline 2010b). GSK1605786A was selected for ongoing trials. Two phase III randomized, double-blind, placebo-controlled studies to investigate the efficacy and safety of induction (GlaxoSmithKline 2010a, 2011) of GSK1605786A in patients with moderate to severe Crohn’s disease are currently recruiting.

The results of a first multicenter, randomized, double-blind, placebo-controlled trial of CCX-282 in patients with CD (PROTECT-1) have not been fully published yet. Here, 3 different doses were tested in 4,436 patients with moderately severe CD. CCX-282 at a daily dose of 500 mg showed a significantly higher effect in achieving the primary end point of CDAI reduction of ≥70 compared to placebo.

The extension study was carried out with 241 patients receiving a dose of 250 mg twice a day or a placebo for 36 weeks. Here the proportion of patients in remission (CDAI ≤ 150) are stable at 50% over the whole trial period in the CCX-282 group (P = 0.01). However, a drop was recognized from 50 to 31% in the placebo group.

The problem with this compound is that none of these studies was fully published yet despite a registration of the first trial in 2005. Another concern is that efficacy was reported at medical meetings both in patients with small and large intestinal Crohn’s disease. Considering the absence of the compounds ligand—CCL25—in the colon this raises questions about the specificity of the compound or validity and applicability of the reported preliminary results.

Risk of therapies targeting leukocyte migration

There is marked enthusiasm in the inflammatory bowel disease community about the potential of these new treatment strategies. However, when the first three cases of JC virus infection and PML were reported the manufacturers halted the program. The drug is now only available in some jurisdictions and under a special restrictive schedule (Kleinschmidt-DeMasters and Tyler 2005; Langer-Gould et al. 2005; Tan and Koralnik 2010; Van Assche et al. 2005). Recently another adhesion molecule antibody targeting αLβ2, efalizumab, was reported to be associated with PML and pulled from the market (Kothary et al. 2011).

The hope is that the newer, gut specific designs discussed in this article will have an impact on both efficacy and safety. So far, this hope seems to hold up. However, none of the current trials add ultimately convincing safety data for the anti-adhesion concept due their small patient numbers compared with the natalizumab program and considering how many exposed patients it took to uncover the JC virus problem. JC peripheral blood viremia screening does not help with risk reduction either, since it is not an appropriate screening method (Yousry et al. 2006).

Copyright information

© Springer Basel AG 2011

Authors and Affiliations

  1. 1.Division of Gastroenterology and Hepatology, Department of Medicine, Charité Medical Center, Virchow HospitalMedical School of the Humboldt University of BerlinBerlinGermany

Personalised recommendations