Computational discovery of precision therapeutics for hidradenitis suppurativa

By integrating transcriptomics, drug perturbation data, and translational immunology, this study identifies sirolimus, pioglitazone, and fulvestrant as novel precision therapeutics that reverse the disease signature and suppress inflammation in hidradenitis suppurativa.

The Gist

Imagine your body is a bustling city. Usually, the security guards (your immune system) patrol the streets, keeping things peaceful and fixing small problems before they become disasters. But in a condition called Hidradenitis Suppurativa (HS), the city's security system goes haywire. It starts attacking its own neighborhoods, causing painful, deep-seated abscesses, tunnels under the skin, and chronic inflammation. It's like a riot that won't stop, and unfortunately, the current "police" (existing drugs) only calm down about half the rioters, leaving many patients in pain.

This paper is like a team of digital detectives who decided to solve this mystery not by guessing, but by using a massive library of data to find the perfect "peacekeepers" already sitting on the shelf.

Here is the story of how they did it, broken down into simple steps:

1. The Crime Scene Investigation (The Data)

First, the researchers looked at the "crime scene." They gathered thousands of genetic reports (like DNA blueprints) from the skin and blood of people with HS and compared them to healthy people.

• The Analogy: Imagine taking a photo of a messy room (HS skin) and a clean room (healthy skin) and using a super-computer to highlight exactly what's different. They found that in HS, the "mess" involves specific genes screaming in anger (inflammation) while the genes that usually keep things calm (like those dealing with hormones and metabolism) are silent.

2. The "Wanted" Poster (The Signature)

Once they knew what the "mess" looked like, they created a digital "Wanted Poster." This poster listed the specific genes that were acting up.

• The Analogy: Think of this as a fingerprint of the disease. The researchers said, "We are looking for a drug that can take this fingerprint and flip it upside down, turning the angry genes back to normal."

3. The Great Library Search (Computational Screening)

Instead of testing thousands of new chemicals in a lab (which takes years and costs millions), they used a giant digital library called the Connectivity Map. This library contains the genetic "fingerprints" of thousands of drugs already approved by the FDA for other diseases.

• The Analogy: Imagine you have a broken toy, and you need a tool to fix it. Instead of building a new tool from scratch, you walk into a massive hardware store (the library) and ask the computer: "Show me every tool in the store that, when used, makes the broken toy look exactly like a brand-new one."
• The computer scanned millions of drug profiles and found three "tools" that were predicted to fix the HS fingerprint perfectly.

4. The Three New "Peacekeepers"

The computer pointed to three existing drugs that are already safe for humans but were being used for totally different problems:

1. Sirolimus: Usually used to stop organ transplant rejection. Think of it as a "brake pedal" for the immune system.
2. Pioglitazone: Usually used for diabetes. Think of it as a "metabolic tune-up" that also calms down inflammation.
3. Fulvestrant: Usually used for breast cancer. Think of it as a "hormone blocker" that stops the body from overreacting to estrogen signals.

The researchers realized these drugs might be the missing keys to unlocking HS treatment because they target the specific "angry" pathways the computer found.

5. The Real-World Test (The Lab)

Computers are great, but you have to prove it works in real life. Since mice don't get HS the same way humans do, the researchers created a unique test: they took actual skin tissue from HS patients who were having surgery and kept it alive in a dish (an ex vivo model).

• The Analogy: It's like taking a piece of a burning building and putting it in a controlled room to see if a specific fire extinguisher works, without having to burn down a whole city first.
• The Result: When they added these three drugs to the skin tissue, the "fire" went out. The immune cells stopped panicking, stopped multiplying wildly, and stopped pumping out the chemicals that cause pain and swelling.

Why This Matters

This paper is a game-changer because it uses AI and big data to skip the long, expensive process of inventing new drugs from scratch.

• The Metaphor: Instead of trying to invent a new type of car engine to fix a traffic jam, they looked at the map, realized a bus (an existing drug) could clear the traffic faster, and just told the bus to go there.

The Bottom Line:
The researchers found that drugs already approved for cancer, diabetes, and transplants might be the secret weapons to finally stop the painful "riots" of Hidradenitis Suppurativa. They have a blueprint, a computer prediction, and a lab test to prove it. The next step is to try these drugs on real patients to see if they can finally bring peace to the city.

Technical Summary

1. Problem Statement

Hidradenitis Suppurativa (HS) is a chronic, immune-mediated inflammatory skin disease characterized by painful nodules, abscesses, and scarring. Despite its significant physical and psychosocial burden, HS remains underdiagnosed with a median diagnostic delay of 7–10 years.

• Therapeutic Gap: Current treatment options are limited. Only three FDA-approved biologics exist (adalimumab, secukinumab, bimekizumab), targeting TNF-α or IL-17. These show modest efficacy (~50% response rate in ~50% of patients) and often fail in advanced disease or lead to relapse.
• Scientific Challenge: The pathophysiology of HS is heterogeneous and multifactorial, involving complex interactions between immune cells, metabolic pathways, and hormonal signaling. Traditional targeted therapies often fail to address individual genetic variability or the full spectrum of dysregulated molecular pathways.
• Modeling Limitation: There is no validated mouse model for HS that accurately recapitulates human skin immunology, hindering preclinical drug testing.

2. Methodology

The authors employed a data-driven precision medicine framework integrating multi-omics analysis, computational drug repurposing, and translational preclinical validation.

A. Integrative Transcriptomic Meta-Analysis

• Data Source: Aggregated microarray data from the Gene Expression Omnibus (GEO), comprising 89 HS samples (skin and blood) and 42 healthy controls across 5 studies.
• Analysis: Used MetaIntegrator to correct for batch effects and calculate standardized mean differences (SMD).
• Outcome: Defined a robust, tissue-specific disease signature:
  • Skin Signature: 3,165 upregulated and 3,499 downregulated genes.
  • Blood Signature: 809 upregulated and 581 downregulated genes.
• Pathway & Cell Deconvolution: Applied FGSEA for pathway enrichment and ImmunoStates for cell-type deconvolution. Identified upregulation of IFN-α/γ, JAK/STAT, and TNF-α pathways, and downregulation of estrogen response and metabolic pathways. Deconvolution revealed increased abundance of monocytes, neutrophils, plasma cells, and B cells in HS skin.

B. In Silico Drug Repurposing (Connectivity Map)

• Approach: Utilized the Connectivity Map (CMap) database, which contains gene expression profiles of ~5,000 small molecules.
• Algorithm: A non-parametric rank-based pattern-matching strategy compared the HS disease signatures (up/down-regulated genes) against drug-induced signatures.
• Metric: Calculated a "reversal score" where a negative score indicates a drug's ability to reverse the disease signature toward a healthy state.
• Filtering: Identified drugs with a False Discovery Rate (FDR) < 0.05 that reversed signatures in both skin and blood.

C. Single-Cell RNA Sequencing (scRNA-seq) Validation

• Dataset: Analyzed scRNA-seq data (GSE155850) from >20,000 sorted immune cells from HS and healthy skin.
• Goal: To determine if candidate drugs could reverse disease signatures in specific cell types (e.g., T cells, macrophages, dendritic cells).
• Method: Identified 13 distinct immune cell clusters and performed cell-type-specific differential expression analysis using Seurat and DESeq2.

D. Experimental Validation (Ex Vivo HS Skin Model)

• Model Development: Created a novel ex vivo human HS skin culture model using enzymatically digested surgical specimens from HS patients. This bypasses the lack of a mouse model.
• Assay: Cultured single-cell suspensions with candidate drugs (0–30 µM) for 2–3 days.
• Readouts:
  • Flow Cytometry: Measured viability, proliferation (Ki67), and activation of CD4+ T cells, CD8+ T cells, and Tregs.
  • Luminex Cytokine Assay: Quantified pro-inflammatory cytokines (IL-17A/F, IFN-γ, TNF-α, CXCL10, etc.) in supernatants.

3. Key Contributions

1. Identification of Novel Therapeutics: Discovered three FDA-approved drugs—Sirolimus (mTOR inhibitor), Pioglitazone (PPAR-γ agonist), and Fulvestrant (Estrogen receptor antagonist)—as potent candidates for HS.
2. Mechanistic Insight: Demonstrated that HS pathogenesis involves dysregulation not just of cytokines, but also of metabolic pathways (PPAR-γ, fatty acid metabolism) and hormonal pathways (estrogen response), suggesting a broader therapeutic target than current biologics.
3. Novel Preclinical Platform: Developed and validated an ex vivo human skin model that allows for high-throughput testing of drug effects on human immune cells and cytokine production, addressing the critical gap in HS animal modeling.
4. Cell-Type Specificity: Showed that the top candidates (Sirolimus and Fulvestrant) were predicted to reverse disease signatures across all 13 identified immune cell types, highlighting their broad immunomodulatory potential.

4. Key Results

• Computational Screening:
  • Identified 109 drugs reversing the skin signature and 260 for blood; 27 drugs overlapped.
  • Sirolimus, Pioglitazone, and Fulvestrant were selected as top hits due to strong reversal scores in both tissues and broad target coverage (Sirolimus and Fulvestrant hit all 13 cell types in scRNA-seq analysis).
  • Network analysis revealed these drugs modulate targets beyond their canonical mechanisms (e.g., Fulvestrant modulating 13 human protein targets).
• Ex Vivo Validation:
  • Cell Viability: All four tested compounds (including the control Imipenem) showed no significant toxicity to HS skin cells.
  • Proliferation: Sirolimus, Pioglitazone, and Fulvestrant significantly inhibited the proliferation of CD4+ T cells, CD8+ T cells, and Tregs in a dose-dependent manner (p < 0.05). Imipenem had no effect.
  • Cytokine Suppression: The three selected drugs significantly suppressed the production of key HS-associated pro-inflammatory cytokines (IL-17A/F, IFN-γ, CXCL10, G-CSF, TNF-α) compared to vehicle controls. Imipenem did not suppress cytokine production.
• Pathway Correlation: The efficacy of Pioglitazone (metabolic) and Fulvestrant (hormonal) supports the meta-analysis findings of downregulated metabolic and estrogen pathways in HS, validating the hypothesis that correcting these dysregulations is therapeutic.

5. Significance

• Precision Medicine Paradigm: This study validates a systematic, unbiased computational approach for drug discovery that leverages human transcriptomic data to identify therapies targeting the specific molecular "fingerprint" of a disease, rather than just a single cytokine.
• Clinical Translation: The identified drugs are already FDA-approved for other indications (transplant rejection, diabetes, breast cancer), meaning they have established safety profiles. This significantly accelerates the timeline for clinical trials in HS (drug repurposing).
• Broader Applicability: The framework (Meta-analysis $\rightarrow$ CMap screening $\rightarrow$ scRNA-seq refinement $\rightarrow$ Ex vivo validation) serves as a scalable model for discovering treatments for other complex, heterogeneous chronic inflammatory diseases (e.g., SLE, IBD, scleroderma) where animal models are inadequate.
• New Therapeutic Avenues: By highlighting the role of metabolic and hormonal pathways in HS, the study opens new avenues for combination therapies that could potentially achieve higher efficacy and durability than current monotherapies.