Thymic Microenvironment Remodeling in Cancer Cachexia as a Determinant of Checkpoint Inhibitor Efficacy and Toxicity

Abstract

Background: The discovery of immune checkpoints links autoimmunity and cancer, with thymus atrophy reportedly causing autoimmune multiorgan inflammation. The impact of cancer cachexia on thymic involution and its clinical significance remains unclear. This study aimed to investigate this effect and its association with immune checkpoint inhibitor (ICI) treatment.

Methods: Single-cell sequencing, immunofluorescence and flow cytometry analyses were conducted to explore changes in the thymus in orthotopic hepatocellular cancer (HCC) mice with cachexia. Patients with advanced and locally advanced cancers receiving anti-PD-1/L1 antibody treatment were followed up to investigate the relationship between the amount of serum autoantibodies and the efficacy of ICIs.

Results: Single-cell sequencing in cachexic HCC mice revealed thymic fibroblast maturity disorders characterized by elevated immature medullary fibroblasts, impaired antigen processing functions, reduced interaction with single-positive thymocytes and decreased expression of tissue-restricted antigen-related genes. The thymus of mice with cancer cachexia exhibited degradation of the thymic medulla and decreased expression of LtβR, Mmp9 and Ccl19 in thymus medullary fibroblasts (mFbs). Single-cell TCR sequencing showed that inflammatory-related V/J TCR genes were highly used in expanded thymocyte clonotypes in cachexic HCC mice, suggesting impaired T cell negative selection. Results from coculture and cell transfer assays suggest that cancer cachexic CD45⁺ erythroid progenitor cells (EPCs) induce the death of CD34⁺ progenitor cells and decrease the number of LtβR⁺, Mmp9⁺ and Ccl19⁺ mFbs in tumour-free mice. CD24⁺CD4⁺CD8^- single-positive thymocytes, typically eliminated in negative selection, did not decrease after the administration of anti-CD3 mAb. Serum autoantibodies were markedly produced in cachexic HCC mice, cachexic HCC mice administered with anti-PD1 and tumour-free mice that received cancer cachexic CD45⁺ EPCs. Autoantibodies against tumour-restricted antigens were found in patients with advanced and locally advanced cancer who received two cycles of ICI treatment. Univariate Cox regression analysis showed that patients with a low level of autoantibodies had a higher risk of disease progression (hazard ratio [HR]: 2.39, 95% CI [1.02-5.63], p = 0.046). Analysis of the receiver operating characteristic curve indicated that the number of autoantibodies against tumour tissues predicted treatment failure (area under the curve [AUC] 0.726, p = 0.021) and long-term duration of treatment response (AUC 0.697, p = 0.024). Patients with high levels of serum autoantibodies against tumours had favourable progression-free survival (HR, 0.389; 95% CI [0.158-0.960], p = 0.04).

Conclusions: Cancer cachexia disrupts mFbs maturity, affecting T cell negative selection and expanding the TCR repertoire against tissue-restricted antigens. This might mediate the adverse and favourable effects of ICIs as anticancer treatments.

Keywords: T cell negative selection; cancer cachexia; immune checkpoint inhibitors; thymus involution; thymus medullary fibroblast.

FIGURE 1

scRNA‐seq analysis revealed differences in the fibroblast composition in the thymus between sham and cachexic HCC mice. (A) A two‐dimensional representation of cells, separated by group using UMAP, and the ratio of cell types in each group displayed via a bar chart. The cells are coloured based on their type identity in the thymus of sham and cachexic HCC mice. The cell types include double‐negative thymocytes (DN), double‐positive thymocytes (DP), double‐positive blast thymocytes (DPblast), double‐positive thymocytes undergoing rearrangement (DPres), double‐positive thymocytes undergoing selection (DPsels), CD4 and CD8 single‐positive thymocytes (CD4SP and CD8SP), natural killer T cells (NKT), regulatory T cells (Treg), conventional dendritic cells (cDC), plasmacytoid dendritic cells (pDC), migratory dendritic cells (Migration DC) and macrophages and fibroblasts. (B) A two‐dimensional representation of cells, separated by group using UMAP, and the ratio of cell types in each group is shown via a bar chart. The cells are coloured based on their type identity in CD45‐thymus cells. The cell types include capsular fibroblasts (capFbs), medullary fibroblasts (mFbs), thymic epithelial cells (TECs), endothelial cells, mesothelial cells, pericytes and immune cells. (C) An interaction analysis among CD4SP, CD8SP, immature mFbs and mature mFbs in sham and cachexic HCC mice using the CellChat package (Version 1.6.1) in the CCL signalling pathway, along with the expression of pathway‐related genes among cell types. (D–F) Gene Ontology (BP), Kyoto Encyclopedia of Genes and Genomes (KEGG) (E), and ReactomePA (F) analysis of immature and mature mFbs in the thymus of sham and cachexic HCC mice. Items associated with antigen processing and presentation functions are marked in red. cHCC, cachexic hepatocellular cancer; ImFb, immature thymus medullary fibroblasts; MmFb, mature thymus medullary fibroblasts.

FIGURE 2

Cachexic HCC mice showed an increased ratio of immature mFbs in the thymic medulla and decreased expression of TRAs. (A) Flow cytometry analysis of the mFb/capFb ratio in cachexic HCC and sham mice. (B) Immunofluorescence analysis comparing the ratios of LtβR⁺ mFb, Mmp9⁺ mFb and Ccl19⁺ mFb in the thymic medulla of cachexic HCC and sham mice. Statistical analyses and immunofluorescence images are presented. (C, E) Flow cytometry analysis of the mean fluorescence intensity of LtβR in mFbs (C), the ratio of Mmp9⁺ mFbs (D) and the median fluorescence intensity of CCL19 (E) in mFbs in cachexic HCC and sham mice. (F) Changes in the expression of mature mFb‐specific TRA genes and Aire‐controlled TRA genes in mature and immature mFbs, presented via box plots and heatmaps. Data are presented as mean ± SD Student's t‐test was used to determine statistical significance. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001.

FIGURE 3

TCR analysis of thymocytes showed more frequent usage of v/j genes related to inflammatory diseases in cachexic HCC mice. (A) A line chart illustrating the number of clonotypes of v/j genes with certain abundance (clone numbers > 1) for β and α chains in the thymocytes of cachexic HCC and sham mice. (B) The number of v/j genes in the top 10% repertoires of the β and α chains in thymocytes of cachexic HCC and sham mice. (C) Diversity analysis of TCR‐β and TCR‐α repertoires in thymocytes, evaluated by the Chao1 value (left) and rarefaction analysis (right) of TCR‐β and TCR‐α repertoires in thymocytes of cachexic HCC and sham mice. (D) The frequency (usage) of expanded clonotypes of v/j genes of the β and α chains related to inflammatory diseases in thymocytes of cachexic HCC and sham mice, shown using a bar plot. (E, F) CD4 SP thymocytes (E) and CD8 SP thymocytes (F), illustrating the expression of inflammatory disease‐related v/j genes of the β and α chains in the thymocytes of cachexic HCC and sham mice in dot plots, showing the mean and percentage of expression, conducted using the Seurat package (4.4.0) in R.

FIGURE 4

Thymocyte subtype analysis revealed more frequent usage of v/j genes associated with inflammatory diseases in cachexic HCC mice. (A) Two‐dimensional representation via UMAP/t‐SNE of cell subtypes of DN, consisting of DN1 (multipotent CD44⁺, c‐Kit⁺ and CD25⁻ early thymic progenitor cells), DN2 (stages of T lineage commitment), DN3a (γδ T cell lineage commitment), DN3b (β‐selection) and DN4 (αβ lineage–committed thymocytes that undergo cell cycle progression, initiate proliferation, and then mature into DP blasts). (B) Two‐dimensional representation via UMAP/t‐SNE of the cell subtypes of DP blasts, classified into G1/S, G2M and M subgroups according to the average expression level of all cell cycle‐associated genes. (C) Two‐dimensional representation via UMAP/t‐SNE of the cell subtypes of DPres, including DPres1, DPres2 and DPres3. (D) Two‐dimensional representation via UMAP/t‐SNE of the cell subtypes of DPsels, based on cell trajectory analysis. The trajectory of the DPsels subtypes was shown via trajectory analysis using the Monocle2 (Version 2.26.2) package. (E) Two‐dimensional representation of CD4 SP thymocyte cell subtypes via UMAP/t‐SNE, split into CD4SP1 (CD24⁺CCR7⁻), CD4SP2 (CD24⁺CCR7⁺) and CD4SP3 (CD24⁻CCR7⁺). (F) Two‐dimensional representation via UMAP/t‐SNE of the cell subtypes of CD8 SP thymocytes, classified into CD8SP1 (CD24⁺CCR7⁺) and CD8SP2 (CD24–CCR7⁺). The bar charts show the ratios of cell subtypes in each group, with cell types coloured based on their type identity.

FIGURE 5

Cachexic HCC CD45⁺ EPC impairing the viability of CD34⁺ progenitors leads to mFb maturation disorder. (A, B) Effect of CD45⁺ EPCs on the viability of CD34⁺ progenitor cells via the coculture system assay, using trypan blue staining, in humans with cancer cachexia (A) and cachexic HCC mice (B). (C) Flow cytometry analysis of the mFb/capFb ratio in mice administered CD45⁺ EPCs or PBS via tail vein injection. (D, E) Immunofluorescence analysis of the ratios of LtβR⁺ mFb, Mmp9⁺ mFb and Ccl19⁺ mFb in the thymic medulla of mice administered CD45⁺ EPCs or PBS via tail vein injection. Statistical analysis (D) and immunofluorescence images (E) are shown.

FIGURE 6

Cachexic HCC mice exhibit dysfunction in T cell negative selection in the thymus. (A) Flow cytometric analysis of the ratio of CD24⁺CD4⁺CD8–SP thymocytes in cachexic HCC and sham mice following intraperitoneal injection of anti‐CD3 mAb (10 mg/kg, clone 17A2, BioLegend) or isotype controls. (B) The amount of serum Mmp9 and Hmgcs2 autoantibodies in cachexic HCC and sham mice was measured via ELISA. (C) Immunofluorescence analysis of autoantibodies in multiple organ sections of Rag1^−/− mice incubated with serum from sham or cachexic HCC mice. Statistical analysis (bottom) and representative immunofluorescence images (top) are shown. (D) Immunofluorescence analysis of autoantibodies in multiple organ sections from Rag1^−/− mice incubated with serum from tumour‐free mice, mice with early‐stage HCC (esHCC), and cachexic HCC (cHCC) mice administered anti‐PD‐1 (10–12.5 mg/kg, clone RMP1‐14, Bio X Cell) via intraperitoneal injection. Representative immunofluorescence images and statistical analyses are shown. (E) ELISA of Mmp9 autoantibody levels in tumour‐free mice administered CD45⁺ EPCs from cachexic HCC mice or PBS via tail vein injection. (F) Immunofluorescence analysis of autoantibodies in multiple organ sections from Rag1^−/− mice incubated with serum from tumour‐free mice, which were administered CD45⁺ EPCs from cachexic HCC mice or PBS via tail vein injection. Representative immunofluorescence images and statistical analyses are shown.

FIGURE 7

Serum autoantibodies against tumours predict the antitumour efficacy of ICIs among patients with cancer. (A) Immunofluorescence analysis of autoantibodies against tumour tissues in the serum of patients with cancer cachexia. (B, C) ROC analysis of the amount of serum autoantibodies against tumour tissues to predict treatment failure (progressive disease) (B) and a duration of treatment response > 180 days (C). (D) PFS of patients with low or high levels of serum autoantibodies. (E) ROC analysis of the amount of serum autoantibodies against tumour tissues to predict the overall response.