Discovery and validation of a novel subgroup and therapeutic target in idiopathic multicentric Castleman disease

Abstract

Idiopathic multicentric Castleman disease (iMCD) is a poorly understood hematologic disorder involving cytokine-induced polyclonal lymphoproliferation, systemic inflammation, and potentially fatal multiorgan failure. Although the etiology of iMCD is unknown, interleukin-6 (IL-6) is an established disease driver in approximately one-third of patients. Anti-IL-6 therapy, siltuximab, is the only US Food and Drug Administration-approved treatment. Few options exist for siltuximab nonresponders, and no validated tests are available to predict likelihood of response. We procured and analyzed the largest-to-date cohort of iMCD samples, which enabled classification of iMCD into disease categories, discovery of siltuximab response biomarkers, and identification of therapeutic targets for siltuximab nonresponders. Proteomic quantification of 1178 analytes was performed on serum of 88 iMCD patients, 60 patients with clinico-pathologically overlapping diseases (human herpesvirus-8-associated MCD, N = 20; Hodgkin lymphoma, N = 20; rheumatoid arthritis, N = 20), and 42 healthy controls. Unsupervised clustering revealed iMCD patients have heterogeneous serum proteomes that did not cluster with clinico-pathologically overlapping diseases. Clustering of iMCD patients identified a novel subgroup with superior response to siltuximab, which was validated using a 7-analyte panel (apolipoprotein E, amphiregulin, serum amyloid P-component, inactivated complement C3b, immunoglobulin E, IL-6, erythropoietin) in an independent cohort. Enrichment analyses and immunohistochemistry identified Janus kinase (JAK)/signal transducer and activator of transcription 3 signaling as a candidate therapeutic target that could potentially be targeted with JAK inhibitors in siltuximab nonresponders. Our discoveries demonstrate the potential for accelerating discoveries for rare diseases through multistakeholder collaboration.

Conflict of interest statement

Conflict-of-interest disclosure: D.C.F. has received research funding from Janssen Pharmaceuticals and EUSA Pharma for the ACCELERATE Natural History Registry (financial sponsorship of the ACCELERATE study has now been transferred to EUSA Pharma), received study drug from Pfizer for a clinical trial of sirolimus in iMCD, and holds two provisional pending patents related to the diagnosis and treatment of iMCD. S.S. and J.M. provide consultancy for Medidata Solutions. A.B.O., L.K., P.B., and D.L. are employed by and/or receive equity ownership from Medidata Solutions. M.G., M.R., and C.T. are employed by and receive equity ownership by Janssen Pharmaceuticals. M.W. provides consultancy and/or receives research funding from Amgen, Bristol Myers Squibb, Crescendo Bioscience, Sanofi/Regeneron, and UCB Pharmaceuticals. M.B. receives honoraria from Merck, Gilead, ViiV, and Janssen Pharmaceuticals. Y.M. receives research funding from Kyowa Hakko Kirin, Astellas, Ono, Eisai, and Pfizer. A.F. receives honoraria from Janssen Pharmaceuticals. F.v.R. provides consultancy for EUSA Pharma, Takeda, Sanofi Genzyme, Adicet Bio, Kite Pharma, and Karyopharm Therapeutics. The remaining authors declare no competing financial interests.

© 2021 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

Advocacy-industry-academic collaboration using multiple technologies and platforms to perform precision medicine science on a collection of iMCD samples.

Figure 2.

Clustering analysis of serum proteomes of iMCD and related diseases reveals notable heterogeneity. (A) A t-SNE plot visualizing serum proteomes of iMCD, Hodgkin lymphoma (lymphoma), HHV8-associated MCD (HHV8+MCD), and rheumatoid arthritis (RA) patients during active disease. Among the iMCD patients, siltuximab responders (partial response or complete response, per durable clinical and tumor [radiologic lymph node] response criteria as determined in NCT01024036) are indicated with open triangles, nonresponders with closed triangles, and patients for which siltuximab was not given as a monotherapy or response was not assessed by independent clinical trial review are represented by closed circles. Colored lines are drawn around clusters as determined by elastic net with fivefold cross-validation. (B) A revisualization of the t-SNE plot of iMCD and related disease serum proteomes to better indicate iMCD in relation to all other samples. Colored lines are drawn around clusters A to E as determined by elastic net with fivefold cross-validation, and samples within each cluster are colored accordingly. iMCD patients are indicated by a closed upward-facing triangle, and non-iMCD patients are visualized by an open downward-facing triangle. (C) Top 40 serum analytes that best distinguish between clusters A to E, as selected by elastic net with fivefold cross-validation, across iMCD and related disease samples.

Figure 3.

Clustering analysis of iMCD serum proteomes demonstrates 6 distinct clusters with differential response to siltuximab. (A) Subtyping of iMCD patients into 6 clusters by elastic net clustering of iMCD samples using serum analyte levels, as measured by SOMAscan. Siltuximab responders (partial response or complete response, per durable clinical and tumor (radiologic lymph node) response criteria as determined in NCT01024036) are indicated with open triangles, nonresponders with closed triangles, and patients for which siltuximab was not given as a monotherapy or response was not assessed by independent clinical trial review are represented by closed circles. Lines are drawn around clusters as determined by elastic net with fivefold cross-validation. (B) Top 40 serum analytes that best distinguish between clusters 1 to 6, as selected by elastic net with fivefold cross-validation, across iMCD samples. (C) Proportion of patients within each cluster that demonstrated a partial or complete response to anti–IL-6 therapy when administered during active disease (as determined in NCT01024036). Cluster 1 patients demonstrated significantly higher response to siltuximab than the other patients (P = 8.94 × 10−4).

Figure 4.

Validation of a novel proteomically definable iMCD subgroup that has superior response to siltuximab, increased disease activity, and elevated IL-6 levels. (A) A heat map of the 7 serum analytes that best distinguish cluster 1 vs other clusters, as selected by elastic net with fivefold cross-validation in the discovery dataset. (B) Correlation analysis between cluster 1 score and response, disease activity, and IL-6 levels in the discovery cohort (2-sided P values). (C) A heat map of the 7 serum analytes tested in an independent validation dataset. (D) Correlation analysis between cluster 1 score and response, disease activity, and IL-6 levels in the validation cohort (1-sided P values). Box plots show center median, first and third quartile, and whiskers extend to 1.5 × interquartile range. Cluster 1 scores are scaled from 0 to 1 for each cohort.

Figure 5.

Consistent with enrichment analyses, immunohistochemistry reveals increased pSTAT3 expression in the interfollicular space of iMCD lymph node tissue. (A) iMCD demonstrated significantly more positive staining for pSTAT3 in the interfollicular space compared with normal lymph nodes (P = .0037). No significant differences were observed in germinal centers (P = .2610), secondary follicles (P = .4119), and mantle zones (P = .552). (B) Within the interfollicular space, iMCD lymph nodes demonstrated significantly higher weak (P = .014) and medium (P = .0066) with no difference in strong staining intensity. Representative images of a normal lymph node (C) and an iMCD lymph node (D) (40× magnification) are provided.