Single-Cell Analysis Reveals Inflammatory-Immunosuppressive Niches in Daratumumab-Resistant Primary AL Amyloidosis

This single-cell analysis of daratumumab-treated primary AL amyloidosis patients reveals that suboptimal responders harbor amyloidogenic plasma cells with distinct stress signatures and foster an inflammatory-immunosuppressive bone marrow niche driven by prostaglandin-mediated myeloid expansion and non-classical MHC I interactions, which collectively impair immune function and drive treatment resistance.

The Gist

The Big Picture: A Battle in the Bone Marrow Factory

Imagine your body's bone marrow is a massive, bustling factory. In a healthy person, this factory produces a diverse workforce of cells (red blood cells, white blood cells, etc.) to keep you running.

In Primary AL Amyloidosis, a small group of "rogue workers" (called Plasma Cells) goes haywire. Instead of making useful products, they start churning out a toxic, misshapen glue called amyloid light chains. This glue clogs up the factory's machinery and leaks out into the body, damaging vital organs like the heart and kidneys.

The standard weapon to stop this is a drug called Daratumumab. Think of Daratumumab as a highly skilled security guard wearing a high-visibility vest. Its job is to find the rogue workers (who wear a specific badge called CD38) and remove them from the factory.

The Problem: In about 10–30% of patients, the security guard fails. The rogue workers hide, multiply, or the factory environment protects them. The glue keeps building up, and the organs keep getting damaged.

This study asked: Why does the security guard fail in some people but succeed in others? The researchers used a super-powerful microscope (Single-Cell Analysis) to look at the factory floor of 30 patients, taking snapshots before and after treatment.

---

The Two Main Reasons the Guard Fails

The researchers discovered that the failure happens for two main reasons: the Rogue Workers themselves are tricky, and the Factory Environment is rigged against the security guard.

1. The Rogue Workers' Secret Survival Kit (The Cells Themselves)

Some rogue workers have a specific "survival manual" (genetic code) that helps them hide.

• The Stress Factory: In patients who didn't respond well, the rogue workers were already under high stress (like a factory running on too much overtime). They had a special "stress shield" (Endoplasmic Reticulum Stress) that made them tough to kill.
• The Dormant Seeds: When the security guard attacked, the main group of rogues was knocked down, but a small, hidden group of "seed" cells (which were busy dividing and multiplying) survived. These seeds were like dormant weeds that the guard missed. Once the guard left, these seeds sprouted and grew back into a full field of weeds, causing the disease to return.

2. The Rigged Factory Floor (The Immune Environment)

Even if the security guard (Daratumumab) tries to do its job, the factory floor is filled with saboteurs that stop the guard from working. The researchers found two main types of sabotage:

Saboteur A: The "Smoke Screen" (Inflammation)

• The Analogy: Imagine the factory is filled with thick, smelly smoke (Prostaglandin E2).
• How it works: A specific type of worker (a Monocyte) starts pumping out this smoke. The smoke confuses the security guard's allies (Natural Killer cells and T-cells).
• The Result: The smoke makes the allies tired, confused, and unable to see the rogue workers clearly. Even though the guard is there, the allies can't help finish the job. In patients who didn't respond well, this smoke never cleared up; it actually got thicker.

Saboteur B: The "Invisibility Cloak" (Immune Suppression)

• The Analogy: The rogue workers put on special invisibility cloaks (Non-classical MHC I proteins).
• How it works: These cloaks send a signal to the security guard's allies saying, "Stop! This is a friend, don't attack!"
• The Result: The allies (T-cells and NK cells) are programmed to stop attacking when they see these cloaks. In the patients who failed treatment, the rogue workers wore these cloaks very effectively, tricking the immune system into standing down.

---

The "Good" vs. "Bad" Factory

The study compared two types of factories:

• The "Good" Factory (Good Responders):

  • The smoke clears up after the guard arrives.
  • The rogues don't wear the invisibility cloaks.
  • The security guard's allies (T-cells and NK cells) stay alert, sharp, and ready to fight.
  • Outcome: The factory is cleaned out, and the organs are saved.

• The "Bad" Factory (Suboptimal Responders):

  • The smoke (inflammation) gets thicker, driven by the saboteur Monocytes.
  • The rogues wear heavy invisibility cloaks.
  • The security guard's allies get exhausted and tired (like a soldier who has been fighting too long without rest).
  • Outcome: The guard leaves, but the weeds grow back, and the factory remains clogged with toxic glue.

Why This Matters (The Takeaway)

This research is like finding the blueprint for the factory's security flaws.

1. New Predictors: Doctors can now look at a patient's "factory blueprint" (their specific cell types and stress levels) before starting treatment to predict if the standard security guard will work.
2. New Weapons: If we know the factory is full of "smoke" (Prostaglandin), we can add a smoke machine (anti-inflammatory drugs) to clear the air. If the rogues are wearing "invisibility cloaks," we can develop a new type of guard that can see through them.

In short: The study explains that Daratumumab resistance isn't just about the bad cells being strong; it's about the whole factory environment being rigged to protect them. By understanding the smoke and the cloaks, scientists can design better strategies to clean the factory once and for all.

Technical Summary

1. Problem Statement

Primary light-chain amyloidosis (pAL) is a systemic disorder caused by clonal plasma cells (PCs) secreting misfolded immunoglobulin light chains that deposit in organs, causing failure. While the anti-CD38 monoclonal antibody daratumumab is the established first-line therapy (often combined with bortezomib and dexamethasone), approximately 10–30% of patients exhibit suboptimal responses (failing to achieve Very Good Partial Response, VGPR).

• The Gap: Current baseline predictors (e.g., 1q21 gain) explain only a fraction of non-responders. The specific mechanisms driving resistance, particularly the interplay between intrinsic PC gene programs and the bone marrow (BM) microenvironment, remain poorly understood.
• Objective: To define the molecular mechanisms of daratumumab resistance in pAL by characterizing the single-cell transcriptomic landscape of amyloidogenic PCs and their immune microenvironment.

2. Methodology

The study employed a comprehensive single-cell multi-omics approach on a prospective cohort of 30 newly diagnosed pAL patients treated with daratumumab-bortezomib-dexamethasone (D-Dex).

• Cohort & Sampling:
  • 30 patients (27 evaluable for response; 10 suboptimal, 17 good responders).
  • 78 single-cell RNA sequencing (scRNA-seq) samples generated, including paired pre- and post-treatment samples for a subset.
  • 2 healthy donors served as controls.
• Sequencing Strategy:
  • 3' scRNA-seq: Performed on whole bone marrow aspirates to profile the entire immune microenvironment.
  • 5' scRNA-seq with V(D)J: Performed on CD138-enriched PC fractions to capture full-length B-cell receptor (BCR) sequences and identify clonotypes.
  • Validation:
    • LC-MS: Liquid chromatography-mass spectrometry on serum to validate amyloidogenic light chain sequences against scV(D)J findings.
    • Multiplex Immunofluorescence (mIF): To validate protein expression of key pathway components (PTGS2, PTGER2/4, CD14, CD138) in bone marrow sections.
• Bioinformatics Pipeline (PRISM Framework):
  • Clonotype Identification: Used scV(D)J to define dominant amyloidogenic clones vs. polyclonal PCs.
  • Gene Expression Programs (GEPs): Applied Consensus Non-negative Matrix Factorization (cNMF) to identify distinct transcriptional programs in PCs.
  • State Projection: Used Non-Negative Least Squares (NNLS) to project these GEPs onto all PCs (including those without V(D)J data).
  • Clonal Dynamics: Applied CoSpar (Coherent, Sparse Optimization) to infer transition probabilities between PC states (e.g., from pre- to post-treatment) using light chain sequences as static barcodes.
  • Cell-Cell Communication: Used CellChat to analyze ligand-receptor interactions, specifically focusing on PC-centered signaling.
  • Functional Analysis: ssGSEA and PROGENy to quantify pathway activities (e.g., IFN-γ, exhaustion, phagocytosis).

3. Key Contributions & Results

A. Intrinsic PC Resistance Programs

• Amyloidogenic Signatures: Amyloidogenic PCs showed upregulated protein translation, ribosome biogenesis, and ER stress pathways compared to polyclonal PCs.
• Predictors of Resistance (t(11;14) Subgroup):
  • Suboptimal Responders: At baseline, these patients' PCs exhibited lower protein translation (GEP1) and cell-cell adhesion (GEP6) but higher Endoplasmic Reticulum (ER) stress (GEP4) programs.
  • Treatment Response: During therapy, suboptimal responders showed an increase in protein translation activity, whereas good responders showed a decrease.
• Mitotic Reservoir: A distinct mitotic PC state (high GEP5) emerged in residual disease. In suboptimal responders, these mitotic clones acted as a "reservoir," transitioning into other pathogenic states rather than normalizing, suggesting they drive relapse.

B. The Inflammatory–Immunosuppressive Niche

The study identified two dominant cell-cell communication axes that define the resistant niche:

1. Prostaglandin (PG) Axis (Inflammatory Driver):

   • Mechanism: An axis driven by PTGES2/3 (producing PGE2) in PCs and PTGS2 (COX-2) in specific myeloid cells, signaling through PTGER2/4 receptors on immune cells.
   • Key Cell Type: Expansion of CD38⁻ CD14⁺ monocytes with a Myeloid-Derived Suppressor Cell (MDSC)-like phenotype (neutrophil-like, HLA-DR low, PTGS2 high).
   • Consequence: These MDSC-like cells expanded during treatment in suboptimal responders, fueling a persistent inflammatory niche that impairs myeloid effector functions (phagocytosis/endocytosis).

2. Non-Classical MHC I Axis (Immunosuppressive Driver):

   • Mechanism: PCs expressed inhibitory non-classical MHC I molecules (HLA-E, HLA-F, HLA-G) which engaged inhibitory receptors (KLRK1/NKG2D on NK cells, LILRB1 on myeloid/T cells).
   • Contrast: Good responders showed activating stress signals (MICB–KLRK1), while suboptimal responders showed dominant inhibitory signaling.

C. Immune Cell Dysfunction

• NK Cells: In suboptimal responders, cytotoxic NK cells were depleted and showed impaired FcγR-mediated ADCC (Antibody-Dependent Cellular Cytotoxicity) despite high IFN-γ signatures.
• T Cells: Exhibited profound exhaustion (upregulation of CTLA4, LAG3, PDCD1) and heightened IFN-γ responses that failed to translate into effective cytotoxicity.
• Myeloid Cells: Showed downregulation of apoptotic cell clearance and amyloid-β clearance pathways, further compromising the therapeutic efficacy of daratumumab.

4. Significance and Implications

• Mechanistic Insight: The study moves beyond simple biomarker identification to define a dual-mechanism model of resistance: (1) intrinsic PC survival programs (ER stress/mitosis) and (2) an extrinsic inflammatory-immunosuppressive niche driven by prostaglandins and non-classical MHC I.
• Novel Biomarkers:
  • Baseline ER stress programs and low protein translation in t(11;14) patients predict resistance.
  • The ratio of MDSC-like CD38⁻ CD14⁺ monocytes to surveillance CD16⁺ monocytes serves as a potential predictor of poor outcome.
• Therapeutic Opportunities:
  • Targeting the Niche: The findings suggest that combining daratumumab with COX-2 inhibitors (to block PTGS2) or PGE2 receptor antagonists could disrupt the inflammatory niche.
  • Immune Checkpoint Modulation: Blocking non-classical MHC I interactions (e.g., anti-LILRB1 or anti-HLA-G) or enhancing NKG2D activation could restore NK/T cell function.
• Clinical Translation: The study provides a molecular profile to identify patients at risk of suboptimal response before treatment begins, allowing for early redirection to alternative therapies or combination strategies.

Conclusion

This single-cell atlas reveals that daratumumab resistance in pAL is not merely a failure of drug binding but a complex failure of the immune microenvironment. Resistance is sustained by a prostaglandin-driven inflammatory loop involving MDSC-like monocytes and an immunosuppressive shield formed by non-classical MHC I signaling, which collectively exhausts effector cells and protects amyloidogenic PC clones. These findings offer a roadmap for precision medicine in AL amyloidosis.