A 34-year-old man presented with a 1.5-year history of fever, night sweats, rash, and myalgia. Laboratory evaluation was unremarkable, including normal levels of hemoglobin, white blood cell count, liver function tests, and tumor markers (alpha-fetoprotein, carcinoembryonic antigen, and CA 19-9). Viral hepatitis and human immunodeficiency virus serologies were negative. Serum protein electrophoresis, immunoglobulin concentrations, and erythrocyte sedimentation rate were within normal limits. C-reactive protein (CRP) level was not determined preoperatively. Magnetic resonance imaging (MRI) demonstrated masses in the left retroperitoneum and right liver, and biopsies were consistent with retroperitoneal Castleman's tumor and hepatocellular adenoma (Fig. 1A,B).

thumbnail image

Figure 1. (A) Axial view of arterial-phase, contrast-enhanced MRI. The right liver contained a well-circumscribed, hyperintense mass with a central area of poor enhancement. (B) Left retroperitoneal mass with avid early enhancement. (C) Histologic, IHC, and genetic analysis. Hematoxylin and eosin stained sections of hyaline-vascular variant of Castleman's disease and (D) well-differentiated HCC (×200 magnification, inset at ×400). (E) Messenger RNA levels in HCC (black) and Castleman's tumor (CT; gray). IHC demonstrating (F) nuclear accumulation of β-catenin and (G) CRP overexpression in HCC (×400 magnification).

Download figure to PowerPoint

The patient underwent partial right hepatectomy and resection of the left retroperitoneal mass. Postoperatively, his inflammatory symptoms resolved, and he remains disease free after 10 months. Surgical pathology of the left retroperitoneal mass demonstrated hyaline-vascular variant of Castleman's disease, and the liver revealed a conglomerate mass composed of multiple, Edmonson grade I hepatocellular carcinomas (HCCs) with microvascular invasion (Fig. 1C,D). Surrounding nontumorous liver was normal.

DNA sequencing of the HCC revealed the absence of mutations in STAT3, but the presence of somatic activating mutations of CTNNB1 (c.121A>G; p.T41A) and IL6ST (c.556_576delinsGTG; p.Tyr186_Phe191del), which encode β-catenin and gp130, the interleukin-6 (IL-6) transducer of signal, respectively. The Castleman's tumor did not harbor mutations in CTNNB1 or IL6ST. Quantitative reverse transcriptase polymerase chain reaction demonstrated high expression of IL6 in the Castleman's tumor, but not in the HCC (Fig. 1E). IL-6-mediated inflammatory response genes (SAA2 and CRP) and β-catenin target genes (GLUL and LGR5) were overexpressed by the HCC, relative to a panel of healthy liver tissues. These results were confirmed by immunohistochemistry (IHC), showing β-catenin nuclear staining, homogeneous overexpression of glutamine synthase, the protein encoded by GLUL, and CRP and serum amyloid A overexpression (Fig. 1F,G). Immunostains for human herpesvirus-8 were negative in both the Castleman's tumor and HCC.

Recently, somatic mutations in the gp130 gene, IL6ST, were observed in 60% of inflammatory hepatocellular adenomas (HCAs).1 These mutations cause ligand-independent activation of the IL-6 pathway and its downstream effectors, including Janus kinase (JAK) and signal transducer and activator of transcription 3 (STAT3), resulting in inflammatory signaling and hepatocyte proliferation. Inflammatory HCAs are associated with inflammatory infiltrates, overexpression of acute-phase reactants by hepatocytes, and systemic inflammatory symptoms.2 Independent of IL6ST mutations, 10% of inflammatory HCAs mutated for IL6ST also carry activating mutations in CTNNB1, leading to induction of the Wnt/β-catenin pathway, which is implicated in hepatocarcinogenesis. IL6ST mutations are rarely observed in HCC (<2% of cases), and all cases of IL6ST-mutated HCC are associated with CTNNB1 mutations, suggesting that activation of STAT3 can cooperate with the Wnt/β-catenin pathway for malignant transformation of hepatocytes.

In Castleman's disease, IL-6 oversecretion by germinal center B cells leads to proliferation of lymphocytes and plasma cells, as well as systemic inflammatory symptoms. In our patient, an intriguing question is whether the Castleman's disease contributed to the development of the HCC or vice versa. Double transgenic mice with high levels of IL-6 and the soluble form of its receptor, sIL-6R, develop hepatocellular hyperplasia, which can progress to HCA.3 This hyperplasia occurs in double transgenics, but not in single IL-6 transgenics, suggesting that a certain threshold of IL-6 stimulation is necessary for the development of hepatocellular hyperplasia. Similar to the double transgenic mouse model, in our patient, simultaneous overstimulation of the IL-6-signaling pathway by both the elevated IL-6 produced by the Castleman's disease and activated gp130 may have accelerated the growth and proliferation of an inflammatory HCA, whereas the CTNNB1 mutation may have provided the second hit, leading to complete malignant transformation.

In conclusion, we describe the first case in the literature of the synchronous presentation of retroperitoneal Castleman's disease and HCC in a healthy 34-year-old man. Molecular analysis suggests the development of HCC from a transformed inflammatory HCA. Mutations activating the IL-6- and Wnt/β-catenin–signaling pathways in hepatocytes could have exerted synergistic effects with IL-6 overproduction by the retroperitoneal Castleman's disease to promote tumor growth and malignant transformation to HCC.


  1. Top of page
  2. Acknowledgements
  3. References

The authors thank Drs. Harry Cooper and Valentin Robu for their pathologic analysis and review of the manuscript for this article.


  1. Top of page
  2. Acknowledgements
  3. References