henhaobio
Sign in

bioRxivImmunologyPreprint

C1Q-associated adaptive myeloid remodelling accompanies early response to BCMA CAR-T therapy in multiple myeloma

Shubing Wang, Qian Wang, Yan-Ruide LiiD, Song LiiD

PDFDOICodeData

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (25 Aug 2026) · CC BY 4.0

In a ten-patient multiple myeloma reanalysis, myeloid pseudobulk from day-30 BCMA CAR-T responders showed C1Q enrichment with interferon-γ signalling, not cholesterol efflux; patient-level composite scores stayed exploratory.

In diffuse large B-cell lymphoma, C1QB macrophages tracked progression and macrophage cholesterol efflux curbed CAR-T cytotoxicity in co-culture, yet those from complete responders expressed antigen-processing and interferon programmes, leaving C1Q's role unsettled. This reanalysis tests whether, about 30 days after BCMA CAR-T in multiple myeloma, response accompanies an adaptive rather than suppressive C1Q-associated myeloid state in blood and marrow, using patient-level pseudobulk.

Key results

  1. SuggestedAt day 30 in peripheral blood (PB), myeloid pseudobulk from complete responders was enriched for the C1Q, antigen-presentation and interferon-γ programmes and depleted for TGF-β signalling compared with nonresponders.

    Complete response (CR) vs nonCR, nine patients, normalized enrichment score (NES) with within-contrast false discovery rate (FDR): C1Q NES = 1.70 (FDR = 1.92 × 10^−5); antigen presentation NES = 1.99 (FDR = 3.96 × 10^−3); interferon-γ NES = 3.17 (FDR = 2.56 × 10^−30); TGF-β NES = −1.76 (FDR = 3.96 × 10^−3); hypoxia NES = −1.22 (FDR = 0.119). · Patient-level reanalysis resolves early blood…; Responding blood shows a C1Q-associated…

  2. SuggestedDay-30 bone marrow (BM) reproduced the responder C1Q and interferon-γ enrichment, with TGF-β and hypoxia programmes depleted; antigen presentation rose but did not pass correction.

    CR vs nonCR, six patients: C1Q NES = 1.70 (FDR = 1.33 × 10^−4); interferon-γ NES = 2.48 (FDR = 9.08 × 10^−14); TGF-β NES = −2.49 (FDR = 1.54 × 10^−7); hypoxia NES = −1.74 (FDR = 6.93 × 10^−4); antigen presentation NES = 1.49 (FDR = 0.0838). · Patient-level reanalysis resolves early blood…; Responding marrow recapitulates C1Q and…

  3. SuggestedResponders showed no positive enrichment of cholesterol-efflux genes (GO:0033344) in blood or marrow, and treatment did not raise them in paired blood, which argues against a joint C1Q and cholesterol-export state accompanying response.

    Cholesterol efflux NES = −1.08 (FDR = 0.365) in day-30 PB, NES = −0.91 (FDR = 0.586) in day-30 BM, and NES = −0.81 (FDR = 0.745) in the paired baseline-to-day-30 PB contrast. · Responding blood shows a C1Q-associated…; Responding marrow recapitulates C1Q and…; Longitudinal blood analysis supports…

  4. SuggestedIn paired blood samples, the C1Q, antigen-presentation and interferon-γ programmes rose from baseline to day 30, so the programme also reflects change after treatment and is not only a difference between response groups.

    Nine paired patients, edgeR with patient as blocking factor: C1Q NES = 1.72 (FDR = 3.57 × 10^−5); antigen presentation NES = 2.01 (FDR = 9.25 × 10^−4); interferon-γ NES = 2.79 (FDR = 1.72 × 10^−19); TGF-β and hypoxia not significant (FDR = 0.153 and 0.0804). · Patient-level reanalysis resolves early blood…; Longitudinal blood analysis supports…; Patient-level pseudobulk differential expression…

  5. SuggestedSample-level C1Q-cholesterol composite scores gave only nominal links to response and CD8 dysfunction, and none survived correction across the primary test family. Where a difference appeared, the score was higher in responders, the opposite of what a uniformly suppressive model predicts.

    Day-30 BM composite vs CD8 dysfunction: Spearman ρ = 0.943, permutation P = 0.0168, FDR = 0.247 (six patients); PB ρ = 0.167, P = 0.679 (nine patients). Composite CR minus nonCR: PB difference 0.954 (P = 0.0213, FDR = 0.247); BM 0.921 (P = 0.182, FDR = 0.528). · Patient-level composite scores remain…; Patient-level association tests

Methods, data and reproducibility

Secondary reanalysis of public scRNA-seq (GSE234261): per-sample QC, Scrublet and CellTypist annotation; counts summed by patient, specimen and compartment; edgeR quasi-likelihood tests (day-30 CR vs nonCR, blood and marrow separately; paired baseline vs day 30 in blood); gene-set enrichment analysis (GSEA) on six prespecified programmes; and 10,000-permutation tests on sample-level scores.

  • GSE234261 scRNA-seq of peripheral-blood and bone-marrow mononuclear cells from multiple myeloma patients before and about 30 days after BCMA CAR-T therapy · human · n = 25 specimens from ten patients; 171,971 of 179,876 cells retained after QC · GSE234261
  • Day-30 CR vs nonCR comparisons, analysed separately by tissue · human · n = nine patients (PB); six patients (BM) · GSE234261
  • Paired baseline vs day-30 peripheral blood · human · n = nine paired patients · GSE234261
  • Myeloid and CD8 T cells from broad CellTypist annotation · human · n = 86,830 myeloid cells; 31,653 CD8 T cells · GSE234261
  • Reference myeloma cohorts discussed but not used as independent clinical validation · GSE210079; GSE271915
Methods & tools
Scrublet, CellTypist (Immune_All_Low model), edgeR (trimmed mean of M-values normalization, robust quasi-likelihood GLM, filterByExpr), multilevel GSEA, MSigDB Hallmark gene sets, Gene Ontology (GO:0033344), NCBI GEO, BioPathfinder, Biomni
Reproducibility gaps
The GitHub link covers the general BioPathfinder workflow only. Analysis-specific scripts, the locked gene-set definitions and patient-level result tables are not yet public; the authors plan to deposit them in an archival repository before formal submission. No software versions are stated.
Limitations (6)
  • Ten patients in all: nine have day-30 blood profiles and six have day-30 marrow profiles. No power calculation was done, because every eligible public specimen was reused. · stated by the authors
  • The three-gene C1Q set can give extreme enrichment statistics when its genes rank together, which shows coordinated expression, not a new cell type. GSEA FDR was controlled within each contrast only. · stated by the authors
  • Pseudobulk over the whole myeloid compartment mixes monocytes, macrophages and dendritic cells, so the programme cannot be assigned to a C1Q-high macrophage subset. · stated by the authors
  • Response was early CR versus nonCR, not durable remission. The reference myeloma cohorts could not be mapped to the same endpoint, so there is no independent outcome-labelled validation. · stated by the authors
  • The design is observational: the data cannot tell a state that promotes response from one that follows falling tumour burden or simply reflects treatment exposure. · stated by the authors
  • The blood-marrow concordance comes from one ten-patient cohort with shared response labels, so it shows internal consistency, not independent replication. · noted in this summary

Read this if you study myeloid remodelling after CAR-T in myeloma, or want a template for patient-level pseudobulk testing of prespecified gene programmes in a small public scRNA-seq cohort.

Citations
1

Abstract

BCMA-directed chimeric antigen receptor T cells induce deep responses in multiple myeloma, yet the immune ecology accompanying early response remains incompletely resolved. We reanalysed 171,971 single-cell transcriptomes from 25 peripheral-blood and bone-marrow specimens from ten patients. At day 30, responders showed concordant enrichment of C1Q and IFN-{gamma} programmes in blood and marrow myeloid pseudobulk profiles. Antigen-presentation genes were enriched in blood, whereas TGF-{beta} and hypoxia programmes were depleted in responding marrow. Cholesterol-efflux genes were not enriched in responders or after treatment. A composite C1Q-cholesterol score showed nominal associations with response and CD8 dysfunction in selected compartments, but none survived study-wide correction. The pathway results support an adaptive, antigen-presenting C1Q-associated programme rather than a uniformly suppressive C1Q macrophage model. This state-contingent interpretation of early myeloid remodelling requires prospective, patient-level validation before biomarker or causal claims are warranted.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISCamviel '22Deredec '26Lieberman '24Maura '25Ohlstrom '25Zah '20
70% semantic matchImmunology
N. Camviel, B. Wolf, G. Croce+3 more·bioRxiv·Jul 23, 2022