Immune-mediated liver injury of the cancer therapeutic antibody catumaxomab targeting EpCAM, CD3 and Fcγ receptors

Abstract

The immunotherapeutic catumaxomab targets EpCAM positive cancers and is approved for the treatment of peritoneal carcinomatosis. To assess the safety of intravenous applications a phase 1 clinical trial was initiated. Treatment of EpCAM positive tumor patients with catumaxomab caused dose dependent hepatitis as evidenced by significant elevations in serum alanine- and aspartate aminotransferases, bilirubin, γGT and induction of the acute phase C-reactive protein (CRP) and the cytokines IL6 and IL8. The first patient receiving 10μg catumaxomab experienced fatal acute liver failure which led to the termination of the study. Immmunopathology revealed catumaxomab to bind via its Fc-fragment to FcγR-positive Kupffer cells to stimulate CRP, chemokine and cytokine release. The observed CD3+T-cell margination at activated hepatic macrophages exacerbated T-cell mediated cytotoxicity. Strikingly, the combined Kupffer/T-cell responses against liver cells did not require hepatocytes to be EpCAM-positive. Catumaxomab's off-target activity involved T-cell mediated lysis of the granzyme B cell death pathway and the molecular interaction of hepatic sinusoidal macrophages with T-cells induced cytolytic hepatitis. Although the bile ducts were surrounded by densely packed lymphocytes these rarely infiltrated the ducts to suggest an intrahepatic cholestasis as the cause of hyperbilirubinaemia. Lastly, evidence for the programming of memory T-cells was observed with one patient that succumbed to his cancer six weeks after the last catumaxomab infusion. In conclusion, our study exemplifies off-target hepatotoxicity with molecularly targeted therapy and highlights the complexities in the clinical development of immunotherapeutic antibodies.

Keywords: acute liver failure (ALF); catumaxomab; epithelial cell adhesion molecule EpCAM; idiosyncratic drug hepatotoxicity.

Figure 1. Clinical liver function tests and serum cytokine profiling in patients receiving different doses of catumaxomab infusions

Serum liver function tests were assayed 24h after the 2nd catumaxomb infusion, while those for cytokines were assayed 6h after the first infusion, i.e. peak concentrations. Data are given as mean and standard deviation and the statistical significance was determined with the GraphPad software Prism version 6.0. Individual outliers are marked by a diamond symbol. * p<0.02 ** p<0.01.

Figure 2. Western Blotting of serum HMGB1

# 1-3. Depicted are immunoblots of HMGB1 of three healthy individuals. # 4. Depicted is the immunoblot of HMGB1 of a female patient with fulminant fatal acute liver failure at catumaxomab Cmax concentration. Note, shown are repeats of the same serum sample that was taken at the end of the infusion, i.e. 6h (tmax).

Figure 3. Histopathology and immunohistochemistry of liver tissue from a fulminant fatal case of acute liver failure after a single infusion of 10μg Catumaxomab

A1. (H&E) Overview of congestive hepatopathy and high grade steatosis (x 50). A2. (H&E) Severe macrovesicular steatosis with marked ballooning degeneration (x 200). A3. (H&E) Acute hepatic necrosis with extended cellular debris and nuclear fragments, isolated granulocytes intermingled among debris, severe hemorrhage and collapsed central vein. No apoptotic bodies, clear signs of cytolytic damage (x 200). B1. Hepatocytes are EpCAM negative; a direct molecular interaction between catumaxomab with hepatocytes is unlikely (x100). B2. The lung tumor is EpCAM positive and therefore a bona fide target of catumxaomab; no evidence for T-cell infiltrates and/or antibody dependent cellular toxicity of the tumor (x100). B3. Infiltration of the portal tract by CD3 positive lymphocytes (x100). C1. Marked portal-periportal infiltration of CD4 positive T-cell (x200). C2. Marked portal-periportal infiltration of CD8 positive T-cell concentrated at the portal-hepatocellular interface (x200). C3. Immunohistochemistry of cytokeratin 7. Evidence for focal ductular cholangiocellular proliferates within initial septal extension of the portal tract possibly due to a preexisting lesion (x100). D1. (H&E) Marked hemorrhage extended to the portal tract; severe hepatocellular damage with small trabecular cell remnants; increased T-cell margination at remnants of the sinusoids; portal infiltration of lymphocytes (x100). D2. Staining of portal-periportal vascular structures for CD34. Significant parenchyma damage and destruction results in frequently ruptured sinusoids; the perisinusoidal space of Disse is extensively dilated and blood filled as demonstrated by staining for CD34 of the vascular bed. Small remnants of trabeculae are seen (x100). D3. Immunohistochemistry of CD68 reveals activated Kupffer cells (swelling); phagocytosis of cell debris. Besides, signs of sinusoidal rupture and collapse (x200). E1. Immunohistochemistry of Granzyme B. Positive lymphocytes are seen and surrounded by a dust of exocytosed granula, particularly at the border of the portal tract (x 200). E2. Immunohistochemistry of TIA1; basically no cytoplasmic expression (200x). E3. Immunohistochemistry of High mobility group box chromosomal protein 1. Note the significant cytosolic staining of HMGB1 by centrilobular damaged hepatocytes.

Figure 4. Histopathology and immunohistochemistry of liver tissue from a male patient after 11 consecutive infusions of 7μg Catumaxomab

A1. (H&E) Moderately altered liver microarchitecture. Signs of low grade reticular fibrosis and partially fragmented trabeculae. Hepatocytes appeared small and slightly basophilic. Note the primarily centrolobular macrovesicular steatosis of about 10 to 20% (x50). A2. (H&E) Same image as before but at higher magnification (x100). Moderate portal fibrogenesis with initial septal extensions; partial destruction of the limiting plate and ongoing interface-hepatitis. Focal aggregates of inflammatory cells (open arrow) and bile duct regenerates (small arrows) at the portal edge. Bilayered trabeculae and increased liver cell polyploidization as well as binuclear hepatocytes. Intranuclear glycogen inclusions of hepatocytes. A3. (H&E) Pronounced centrolobular macrovesicular steatosis. Signs of enhanced sinusoidal margination of lymphocytes (x100). B1. Immunohistochemistry of CD31. Some of the sinusoidal endothelial cells appear enlarged with signs of activation. The counterstain with hematoxylin highlights the frequently bilayered trabeculae (x200). B2. Immunohistochemistry of CD34. Hepatic artery and portal vein branches are CD34 positive while sinusoids were CD34 negative. The stain revealed a slightly extended periportal vascular network (x100). B3. Immunohistochemistry of CD68. Resident macrophages were moderately activated and enlarged. T-cell margination in proximity to activated Kupffer cells (x200). C1. Immunohistochemistry of EpCAM. Hepatocytes were negative for EpCAM. Note the faint membranous EpCAM staining of the bile duct epithelium (x200). C2. Immunohistochemistry of CD3. Marked portal T-cell infiltrates. Also visible is the sinusoidal margination of CD3 positive lymphocytes in close proximity to Kupffer cells at the adjacent liver parenchyma (left side, x200). C3. Immunohistochemistry of CD4. Marked portal T-cell infiltrates. Bile ducts are rarely infiltrated by lymphocytes; occasionally moderate swelling of ductual epithelium was observed (x200). D1. Immunohistochemistry of CD8. Unlike the case of fatal ALF (see Figure 3 C2) only very few lymphocytes are CD8 positive (x200). D2. Immunohistochemistry of Granzyme B. Some of the T-lymphocytes are positive and are surrounded by a delicate dust of exocytosed granula, especially at the interphase (x200). D3. Immunohistochemistry of TIA1; basically there is no cytoplasmic expression (x200). E1. Immunohistochemistry of CD3. T-cell infiltrates particularly at the periportal interphase and the septal extensions; the bile duct is surrounded by T-cells without considerable ductal infiltration (x100). E2. Immunohistochemistry of CK7. Bile duct proliferations at the edge of the portal tract and within a septal extension (x100). E3. (H&E) An overview of slight interface-hepatitis (x200).

Figure 5. Possible mechanism of catumaxomab induced immune-mediated liver injury