Identification and Characterization of Neoplastic Cells in Keratin-Positive Giant Cell-Rich Tumors

This study utilizes single-nuclei RNA sequencing and spatial transcriptomics to identify the keratin-positive, HMGA2::NCOR2-fusion cells as the rare neoplastic drivers in KPGCT that activate the Hippo pathway to stimulate CSF1-mediated recruitment and differentiation of macrophages, thereby establishing KPGCT as a distinct entity from TGCT and supporting the therapeutic potential of CSF1R inhibitors.

The Gist

The Big Picture: Solving a Medical Mystery

Imagine a rare, strange tumor called KPGCT (Keratin-Positive Giant Cell-Rich Tumor). For a long time, doctors looked at these tumors under a microscope and saw a chaotic crowd of cells. They knew something was wrong, but they didn't know who the "bad guys" (the cancer cells) actually were, or why the tumor was growing.

This study acts like a high-tech detective squad. They used advanced "single-cell" technology to interview every single cell in the tumor individually, figuring out exactly who is running the show and how they are manipulating the crowd.

---

1. The "Invisible Boss" vs. The "Hired Crowd"

The Discovery:
In a typical tumor, you might think the cancer cells are the majority. But in KPGCT, the real "boss" cells (the neoplastic cells) are actually a tiny minority—only about 12% of the tumor.

The Analogy:
Think of the tumor as a massive concert stadium.

• The Bosses (Neoplastic Cells): There are only a few VIPs in the front row (the keratin-positive cells). They are the ones holding the megaphone.
• The Crowd (Macrophages): The rest of the stadium (88% of the cells) is filled with regular people (macrophages and giant cells).
• The Trick: The VIPs aren't fighting the crowd; they are hiring the crowd. The VIPs shout out instructions that make the crowd gather, multiply, and stay in the stadium.

The study confirmed that these "VIPs" are the only cells carrying a specific genetic mutation (the HMGA2::NCOR2 fusion) and expressing keratin (a protein usually found in skin, not soft tissue). This proves they are the true origin of the tumor.

2. The "Loudspeaker" Strategy (CSF1 Signaling)

The Discovery:
How do the VIPs get the crowd to stay? They shout out a chemical signal called CSF1. The crowd has special ears (receptors) that hear this signal, causing them to multiply and turn into giant cells.

The Analogy:
Imagine the VIPs are holding a Loudspeaker (CSF1).

• They blast a message: "Come here! Stay here! Grow bigger!"
• The crowd (macrophages) hears this and rushes toward the VIPs.
• Because the VIPs are shouting so loudly, the stadium fills up with people, making the tumor look huge, even though the VIPs themselves are few.

Why this matters for treatment:
Since the tumor grows because of this "Loudspeaker," scientists realized we can turn the volume down. Drugs that block the "ears" of the crowd (CSF1 inhibitors) have already worked on a similar tumor. This study suggests these drugs should work on KPGCT too, effectively silencing the VIPs and making the crowd leave.

3. The "Engine" Behind the Loudspeaker (Hippo Pathway)

The Discovery:
The researchers wanted to know: Why are the VIPs shouting so loud? They found that a specific internal engine inside the VIPs is running wild. This engine is called the Hippo signaling pathway.

The Analogy:
Think of the VIPs as a car.

• The Hippo pathway is the gas pedal.
• In a normal car, the gas pedal is controlled by a foot (regulation).
• In these tumor VIPs, the gas pedal is stuck to the floor.
• This stuck pedal forces the car to scream (produce CSF1) constantly.

The study found that a protein called YAP1 (the driver) is stuck in the "on" position inside the VIPs, keeping the gas pedal down. This suggests a new way to treat the tumor: instead of just silencing the crowd, we could try to fix the stuck gas pedal in the VIPs.

4. The "Imposter" Problem (KPGCT vs. TGCT)

The Discovery:
KPGCT looks very similar to another tumor called TGCT (Tenosynovial Giant Cell Tumor). They both have a crowd of giant cells and both use the CSF1 "Loudspeaker." Doctors often confuse them.

The Analogy:
Imagine two different bands playing the same song.

• Band A (TGCT): They are a "Jazz Band" (Synovial origin). They play jazz instruments (Synovial markers like PRG4).
• Band B (KPGCT): They are a "Rock Band" (Mesenchymal origin). They play rock instruments (Keratins and JUN protein).

Even though they play the same song (CSF1 signaling), the instruments are different. The study found that KPGCT is not a Jazz band; it's a Rock band. It doesn't come from the joint lining (synovium) like TGCT does. This is crucial because it means KPGCT is a unique disease, not just a weird version of TGCT.

Summary: What Does This Mean for Patients?

1. We know who the bad guys are: They are the rare, keratin-positive cells hiding in the crowd.
2. We know how they grow: They use a "Loudspeaker" (CSF1) to recruit a massive army of helper cells.
3. We have a treatment plan:
   • Immediate: Use existing drugs that block the "Loudspeaker" (CSF1 inhibitors) to shrink the tumor.
   • Future: Develop drugs that fix the "stuck gas pedal" (Hippo pathway inhibitors) to stop the bad guys from shouting in the first place.
4. Better Diagnosis: Doctors now have a checklist to tell the difference between this tumor and its "look-alike" twin (TGCT), ensuring patients get the right diagnosis.

In short, this paper took a confusing, messy tumor and organized it into a clear story: A few bad bosses are shouting at a crowd to build a fortress. If we stop the shouting, the fortress falls.

Technical Summary

1. Problem Statement

Keratin-positive giant cell-rich tumor (KPGCT) is a rare, newly described soft tissue and bone tumor characterized by scattered keratin-positive cells, the presence of HMGA2::NCOR2 fusions, and a background rich in macrophages and osteoclast-like giant cells.

• Key Unknowns: It was previously unclear which specific cell population harbored the neoplastic driver (HMGA2::NCOR2 fusion), whether the keratin-positive cells were the neoplastic component, and the molecular mechanisms driving the tumor's growth and macrophage recruitment.
• Diagnostic Challenge: KPGCT shares histological features with Tenosynovial Giant Cell Tumor (TGCT), a tumor driven by CSF1 rearrangements. Distinguishing between the two is critical for diagnosis and therapy, yet KPGCT lacks a clear molecular definition for its neoplastic cells compared to TGCT.
• Therapeutic Gap: While anecdotal evidence suggests KPGCT responds to CSF1 inhibitors (used for TGCT), the mechanistic rationale and the specific signaling axis driving the tumor microenvironment (TME) were not fully elucidated.

2. Methodology

The study employed a multi-modal single-cell and spatial transcriptomics approach to deconstruct the tumor architecture and signaling networks:

• Single-nucleus RNA Sequencing (snRNA-seq): Performed on formalin-fixed paraffin-embedded (FFPE) tissue from 5 KPGCT cases and 5 TGCT cases (as a comparative control).
  • Pipeline: Nuclei isolation (10x Genomics), sequencing, and analysis using Cell Ranger, Seurat (v5.3.0), and DESeq2.
  • Analysis: Clustering, dimensionality reduction (UMAP), differential gene expression, and Gene Set Enrichment Analysis (GSEA) for pathways (KEGG, GOBP) and transcription factor targets.
• Spatial Transcriptomics (Xenium): Performed on a KPGCT tissue microarray (TMA) core to validate cell-cell interactions and spatial niches.
  • Tools: BANKSY for spatial niche identification; CellChat (v2.1) for ligand-receptor interaction analysis.
• Validation:
  • Immunohistochemistry (IHC): Co-staining for Keratin and CSF1.
  • Immunofluorescence (IF): Staining for YAP1 (Hippo pathway), phosphorylated C-JUN, GFPT2 (neoplastic marker), and keratin to confirm protein expression and nuclear localization.
  • Quantification: ImageJ analysis of nuclear signal intensity for YAP1 and C-JUN.

3. Key Contributions & Results

A. Identification of the Neoplastic Cell Population

• Minority Population: snRNA-seq revealed that neoplastic cells constitute only ~12% of the total tumor cellularity, while macrophages make up ~55%.
• Molecular Identity: The neoplastic cluster was identified as fibroblast-like cells that co-express:
  • Keratins: High expression of KRT16, KRT14, KRT8, KRT10, and KRT7.
  • HMGA2: Expression was restricted exclusively to this keratin-positive cluster, confirming these cells harbor the HMGA2::NCOR2 fusion.
  • GFPT2: A marker previously associated with TGCT neoplastic cells, also found here.
• Developmental Origin: GSEA indicated enrichment for mesodermal developmental processes (specifically intermediate mesoderm/kidney development), suggesting a primitive origin rather than a synovial one.

B. The CSF1-CSF1R Signaling Axis

• Driver Mechanism: The neoplastic cells produce high levels of CSF1 (Colony Stimulating Factor 1).
• Microenvironment Interaction: The majority of the tumor (macrophages and osteoclast-like giant cells) expresses CSF1R.
  • Spatial Validation: Xenium data showed macrophages are significantly closer to neoplastic cells (mean distance 11.72µm) than to other non-neoplastic cells, confirming a chemoattractive paracrine loop.
• Differentiation: This axis drives the recruitment, proliferation, and differentiation of macrophages into osteoclast-like giant cells.
• Therapeutic Implication: This provides a strong molecular rationale for using CSF1R inhibitors (e.g., pexidartinib, imatinib) in KPGCT, similar to their use in TGCT.
• Contrast with Bone Tumors: Unlike Giant Cell Tumor of Bone, KPGCT neoplastic cells do not express TNFSF11 (RANKL), explaining the limited efficacy of RANKL inhibitors (denosumab) in KPGCT.

C. Pathway Activation: Hippo and C-JUN

• Hippo Pathway: The neoplastic cells show significant activation of the Hippo signaling pathway.
  • Evidence: Upregulation of CCN4 (an inhibitor of YAP1 degradation) and TEAD1 target genes.
  • Validation: Immunofluorescence confirmed nuclear localization of YAP1 specifically in GFPT2-positive neoplastic cells (p < 0.05), indicating active signaling.
  • Hypothesis: The HMGA2::NCOR2 fusion likely drives Hippo pathway activation, which in turn upregulates CSF1 expression.
• C-JUN Activation:
  • Keratin Regulation: Unlike TGCT, KPGCT neoplastic cells highly express JUN and its phosphorylated form (p-S73).
  • Significance: C-JUN is a known regulator of keratin genes, explaining the unique keratin-positive phenotype of KPGCT, which is rare in soft tissue tumors.

D. Distinction from Tenosynovial Giant Cell Tumor (TGCT)

Despite histological similarities, KPGCT and TGCT are distinct entities:

• Synovial Markers: KPGCT neoplastic cells lack synovial markers (CLU, PRG4/Lubricin, CD55) that are characteristic of TGCT.
• Gene Expression: KPGCT is defined by HMGA2, JUN, and KRT16 upregulation, whereas TGCT is defined by CLU, PRG4, and CD55.
• Diagnostic Utility: The study suggests Clusterin (CLU) IHC as a potential tool to distinguish TGCT from KPGCT (especially keratin-negative KPGCT cases).

E. Tumor Microenvironment Niches

• Spatial analysis identified two distinct macrophage niches:
  1. Foamy Macrophage Niche: Enriched with large foamy macrophages (markers: CCL18, CYP27A1) and FOLR2+ cells.
  2. SPP1+ Niche: Enriched with SPP1+, CXCL8+, and interferon-stimulated macrophages.
• This heterogeneity suggests complex local interactions within the tumor mass.

4. Significance

• Definitive Characterization: The study definitively identifies the keratin-positive, HMGA2-expressing fibroblast-like cells as the neoplastic driver of KPGCT, resolving previous uncertainty about the tumor's cellular origin.
• Therapeutic Rationale: It validates the use of CSF1R inhibitors for KPGCT by demonstrating that the tumor mass is largely composed of CSF1R-expressing cells recruited by neoplastic CSF1 production.
• Novel Targets: Identification of Hippo pathway (YAP1/TEAD1) activation offers a potential novel therapeutic target (e.g., YAP/TEAD inhibitors) that directly targets the neoplastic clone, potentially offering a more efficient strategy than targeting the reactive stroma.
• Diagnostic Clarity: The study clarifies that KPGCT is not a variant of TGCT but a distinct entity with a mesodermal (non-synovial) origin. It provides a molecular framework (presence of Keratin/JUN vs. absence of CLU/PRG4) to differentiate these tumors, which is crucial for clinical management.
• Pathogenesis Model: The paper proposes a model where HMGA2::NCOR2 fusion activates the Hippo pathway $\rightarrow$ upregulates CSF1 $\rightarrow$ recruits macrophages $\rightarrow$ drives tumor growth, while C-JUN activation drives the aberrant keratin expression.