Increased mTOR activation in idiopathic multicentric Castleman disease

Abstract

Idiopathic multicentric Castleman disease (iMCD) is a rare and poorly understood hematologic disorder characterized by lymphadenopathy, systemic inflammation, cytopenias, and life-threatening multiorgan dysfunction. Interleukin-6 (IL-6) inhibition effectively treats approximately one-third of patients. Limited options exist for nonresponders, because the etiology, dysregulated cell types, and signaling pathways are unknown. We previously reported 3 anti-IL-6 nonresponders with increased mTOR activation who responded to mTOR inhibition with sirolimus. We investigated mTOR signaling in tissue and serum proteomes from iMCD patients and controls. mTOR activation was increased in the interfollicular space of iMCD lymph nodes (N = 26) compared with control lymph nodes by immunohistochemistry (IHC) for pS6, p4EBP1, and p70S6K, known effectors and readouts of mTORC1 activation. IHC for pS6 also revealed increased mTOR activation in iMCD compared with Hodgkin lymphoma, systemic lupus erythematosus, and reactive lymph nodes, suggesting that the mTOR activation in iMCD is not just a product of lymphoproliferation/inflammatory lymphadenopathy. Further, the degree of mTOR activation in iMCD was comparable to autoimmune lymphoproliferative syndrome, a disease driven by mTOR hyperactivation that responds to sirolimus treatment. Gene set enrichment analysis of serum proteomic data from iMCD patients (n = 88) and controls (n = 42) showed significantly enriched mTORC1 signaling. Finally, functional studies revealed increased baseline mTOR pathway activation in peripheral monocytes and T cells from iMCD remission samples compared with healthy controls. IL-6 stimulation augmented mTOR activation in iMCD patients, which was abrogated with JAK1/2 inhibition. These findings support mTOR activation as a novel therapeutic target for iMCD, which is being investigated through a trial of sirolimus (NCT03933904).

Graphical abstract

Figure 1.

pS6 staining in lymph node tissue. (A) Stained pS6 area proportion at different lymph node structures for the first cohort of iMCD-TAFRO patients (cohort 1, n = 10) compared with a control group of sentinel lymph nodes (n = 5). Statistical significance, denoted by an asterisk, was calculated using 2-compositional analysis of the centrometric log-rate transformation of the proportions. There were significant increases in pS6 staining in the mantle zone (P = .038) and interfollicular space (P = 3.3 × 10⁻⁴), and nonsignificant increases for the entire follicle (P = .050) and germinal center (P = .19). (B) Comparison of various staining intensity proportions of the interfollicular space for cohort 1 and the control group. The results indicate that the iMCD-TAFRO cases had significantly higher medium (P = 3 × 10⁻⁴) and strong (P = 6 × 10⁻⁴) staining. (C-D) Representative images of pS6 (brown) staining for a sentinel lymph node (C) and an iMCD-TAFRO lymph node (D). Hematoxylin counterstain provides a blue nuclear stain to assess cell and tissue morphology. Scale bars, 200 µm. *P < .05, **P < .01. GC, germinal center; Inter, interfollicular space; MZ, mantle zone.

Figure 2.

Comparison of pS6 staining across CD subtypes and other lymphoproliferative diseases. (A) pS6 across different CD subtypes. A second cohort of iMCD-TAFRO patients (cohort 2, n = 10) was compared with iMCD-NOS (n = 6), HHV-8–associated MCD (HHV8 MCD, n = 4), and UCD (n = 7). The proportion of pS6 staining was similar across both clinical subtypes of iMCD as well as HHV-8–associated MCD. Comparison between the combined iMCD cases (n = 16) and the UCD cases (n = 4) showed nonsignificant increase in pS6-staining of the germinal center (P = .17), mantle zone (P = .065), and interfollicular space (P = .089). Of note, a 1-tailed test comparing expression in the interfollicular space of iMCD to UCD would have been significant, but the a priori hypothesis involved testing for difference, not directional difference. (B) The second cohort of iMCD-TAFRO patients and the iMCD-NOS cases were combined and compared with a control group of sentinel lymph nodes, a second control group of reactive lymph nodes, and to 3 other diseases involving lymphoproliferation and inflammatory lymphadenopathy: SLE, HL, and ALPS. pS6 could only be assessed in the interfollicular space of the HL cases due to disruption of the remainder of lymph node architecture. ALPS and iMCD had similar pS6 staining; the iMCD group had significantly higher staining in the interfollicular space compared with SLE, HL, reactive nodes, and sentinel nodes (P = .001, .01, .032, and .019, respectively). (C-H) Representative images of pS6-stained (brown) lymph node tissue (with blue hematoxylin counterstain) for (C) reactive lymph nodes, (D) SLE, (E) HL, (F) iMCD-TAFRO, (G) ALPS, and (H) UCD. Scale bars, 200 µm. *P < .05, **P < .01.

Figure 3.

Investigation of other mTOR effectors. (A) Simplified diagram of 3 key effectors of mTORC1. (B) Stained p4EBP1 area proportion in different lymph node regions for the first cohort of iMCD-TAFRO patients compared with 2 control groups: reactive and sentinel lymph nodes. The stained area proportion at the interfollicular space was significantly higher than the reactive and sentinel lymph nodes (P = .0034 and .0013, respectively). (C) Comparison of p70S6K stained area proportions for iMCD (n = 6) vs reactive lymph nodes. There was a nonsignificant elevation for iMCD across the follicle, germinal center, and the interfollicular spaces (P = .17 for all 3 comparisons). (D) Effect size comparison among the 3 mTORC1 effectors. Hedges’ g, an effect size utilizing the mean difference standardized by the deviation, was calculated for the iMCD-TAFRO to reactive lymph node comparison for the 3 stains in the interfollicular space. The 3 results were combined by random effects (RE) model, as we are testing effects related to mTOR activation. Synthesis by RE model yielded a combined significant effect size of 1.68 [95% CI, 0.52-2.84, Q(df=2) = 4.7, I² = 56%]. For comparison, the Hedges’ g for ALPS in the individual pS6 experiment was 1.64 [0.18, 3.09]. (E-F) Representative images are shown of p4EBP1 staining (brown) for reactive lymph node (E) and iMCD-TAFRO (F). (G-H) Representative images are provided for p70S6K staining (brown) for reactive lymph nodes (G) and iMCD-TAFRO (H). All representative images have hematoxylin counterstain (blue) as a nuclear stain. Bar = 200 µm. *P < .05, **P < .01.

Figure 4.

Identification of cell types with increased mTOR activation. (A) Co-IF for CD45, MUM1, CD138, CD3, CD20, and CD68 was performed to identify the cell types expressing pS6 in 4 iMCD-TAFRO cases (CD20 staining was only performed in 3 cases). Representative images (×20) are shown in panels B-E, where pS6 was stained red and the cell specific marker green: (B) CD138, (C) CD3, (D) CD20, (E) CD68. Labeled cells were manually counted to assess for percentage of pS6⁺ cells. A majority of pS6⁺ cells were of the hematopoietic lineage (CD45⁺). Only a small fraction (2% to 10%) of pS6⁺ cells were T cells (CD3⁺). A large proportion of cells expressed the transcription factor MUM1, which is expressed on activated B cells, T cells, and plasma cells. Plasma cells (CD138⁺) and macrophages (CD68⁺) also made up a large proportion of pS6⁺ cells.

Figure 5.

mTORC1 gene set enrichment in serum proteomics (A) GSEA of serum proteomic data from 88 iMCD patients compared with 42 healthy controls. In iMCD, genes involved in the mTORC1 signaling pathway were significantly enriched (FDR, 0.243) below the FDR set at 0.25. This targeted GSEA of mTORC1 signaling was performed in addition to and separately from a comprehensive proteomics analysis carried out by coauthors (S.K.P. and D.C.F., manuscript submitted January 2020) that did not include GSEA analysis comparing all iMCD patients to healthy controls for the mTORC1 signaling pathway or others. (B) Volcano plot for iMCD vs controls of the mTORC1 genes/proteins quantified with the SOMAscan platform. The cutoff for significance is indicated by the dotted line (FDR ≤0.05). The 7 significantly different genes are all labeled and increased in iMCD compared with healthy controls. All positive fold-change values belong to the proteins that are increased in the iMCD group compared with healthy controls.

Figure 6.

Increased mTOR signaling at baseline and upon stimulation with IL-6. PBMCs from 8 healthy donors and 8 iMCD-TAFRO patients in remission were left untreated, treated with 25 ng/mL IL-6, or treated with IL-6 and the JAK inhibitor ruxolitinib (1μM). (A-C) Kinetics of IL-6–mediated phosphorylation of S6 protein in CD14⁺ monocytes (A), CD4⁺ T cells (B), and CD8⁺ T cells (C) from healthy donors (blue) and iMCD patients in remission (red). Mixed analysis of variance analysis of the center-log-transformed proportions (compositional analysis) yielded that the proportion of pS6⁺ cells is higher for CD14⁺ monocytes (P = 6.0 × 10⁻⁸), CD4⁺ T cells (P = 4.6 × 10⁻⁷), and CD8⁺ T cells (P = 4.2 × 10⁻⁶) across the time measurements in iMCD compared with healthy controls. Compositional analysis specific to each time was also calculated using unpaired 1-tailed Mann-Whitney U tests, and the significance is denoted in panels A-C by asterisks. (D-F) Paired comparison of the frequency of pS6⁺ monocytes, CD4⁺ T cells, and CD8⁺ T cells at 0 minutes (black) and 120 minutes of stimulation with IL-6 for iMCD patient (red) and healthy donor samples (blue). P values from Wilcoxon signed-rank test on the transformed proportions. Since there is a <1% probability that 2 events with <.05 (individual type I error) occurs twice in only the subjects and not the controls, the results were considered statistically significant without a need for P-value adjustment approximations. (G-I) Paired comparison of the frequency of pS6⁺ monocytes (G), CD4⁺ T cells (H), and CD8⁺ T cells (I) following treatment with IL-6 alone or following treatment with IL-6 and the JAK inhibitor, ruxolitinib, for iMCD patient and healthy donor samples. *P < .05, **P < .01 by Wilcoxon signed-rank test. ns, not significant.