Shared multicellular injury programs of acute and chronic kidney disease enable mechanistic patient stratification

This study utilizes single-nucleus transcriptomics of over 150 kidney biopsies to identify shared multicellular injury programs in acute and chronic kidney disease that transcend traditional clinical categories, revealing non-invasive protein signatures predictive of long-term renal outcomes.

The Gist

The Big Picture: Two Different Types of "Kidney Trouble"

Imagine your kidneys are a highly sophisticated water filtration plant. They clean your blood, remove waste, and balance fluids.

For a long time, doctors have treated kidney problems in two separate boxes:

1. The "Sudden Crash" (Acute Kidney Injury - AKI): Like a pipe bursting or a power surge. It happens fast, often due to an infection, dehydration, or a bad reaction to medicine.
2. The "Slow Rust" (Chronic Kidney Disease - CKD): Like the pipes slowly corroding over years due to high blood pressure or diabetes. It's a slow, steady decline.

The Problem: Doctors usually look at how well the kidney is working right now (like checking the water pressure). But two people with the same "low pressure" reading might have completely different problems inside the machine. One might have a burst pipe that can be fixed; the other might have rusted pipes that are permanently damaged. This makes it hard to know who will get better and who will get worse.

The New Discovery: Listening to the "Workers" Inside

This study used a high-tech microscope (single-nucleus transcriptomics) to look at the individual cells inside kidney biopsies from over 150 patients. Instead of just looking at the "water pressure," they listened to the "workers" (the cells) to see what they were saying.

They discovered that the kidney doesn't just have "sudden" or "slow" problems. Instead, there are two universal "alarm programs" that run inside the cells, regardless of whether the patient was diagnosed with AKI or CKD.

1. The "Fire Alarm" Program (Acute Injury)

• What it is: This is the kidney's immediate panic response. It's like a fire alarm going off.
• What's happening: The cells are shouting, "We are under attack! Send help!" They are stressed, inflamed, and trying to adapt to survive a sudden shock.
• Who has it: Mostly people with sudden injuries (AKI), but surprisingly, many people with "slow rust" (CKD) have this alarm ringing too.

2. The "Structural Damage" Program (Chronic Damage)

• What it is: This is the sign of long-term wear and tear. It's like the building's foundation cracking and the walls turning to dust.
• What's happening: The cells are giving up on fixing themselves. They are entering a "failed repair" state. The tissue is becoming scarred (fibrosis) and losing its ability to function.
• Who has it: Mostly people with chronic disease (CKD), but many people with "sudden injuries" (AKI) have this damage starting to happen too.

The Big Insight: The study found that AKI and CKD are not totally different diseases. They are often the same two alarm programs running at the same time, just at different volumes. A patient might have a loud "Fire Alarm" (acute) and a growing "Structural Damage" signal (chronic) happening simultaneously.

The "Magic Crystal Ball" (Non-Invasive Testing)

The researchers wanted to know: Do we need to cut a piece of the kidney out (a biopsy) to see these alarms?

The Answer: No.

They found that the "Fire Alarm" and "Structural Damage" signals leak out of the kidney and show up in the blood and urine.

• They built a "decoder" using blood and urine proteins.
• This decoder can look at a simple urine or blood test and tell you: "Hey, your kidney's Fire Alarm is ringing loud," or "Your Structural Damage is getting worse."

The "Double Trouble" Warning

The most exciting part of the study is what happens when you combine these two signals. They tested this on a massive group of 44,000 people (the UK Biobank).

• Low Fire Alarm + Low Damage: You are safe.
• High Fire Alarm only: You are at risk of a sudden kidney crash.
• High Damage only: You are at risk of slow decline.
• The "Double Hit" (High Fire Alarm + High Damage): This is the danger zone. If a patient has both the acute stress and the chronic scarring active at the same time, they are most likely to have a major kidney failure event in the future.

Why This Matters

Think of it like a car dashboard.

• Old Way: The dashboard just says "Engine Light On." It doesn't tell you if it's a loose gas cap (fixable) or a blown engine (terminal).
• New Way: This study gives you a dashboard that says, "You have a loose gas cap (Acute) AND your transmission is wearing out (Chronic)."

The Takeaway:
By measuring these two specific "programs" using a simple blood or urine test, doctors can:

1. Predict the future: Know who is at high risk of kidney failure before it happens.
2. Personalize treatment: Treat the "Fire Alarm" with anti-inflammatory drugs and the "Structural Damage" with anti-scarring drugs, rather than using a "one-size-fits-all" approach.
3. Bridge the gap: Understand that acute and chronic kidney diseases are part of the same continuous story, not two separate chapters.

This research moves us from guessing based on symptoms to seeing the actual molecular story of the kidney, allowing for much smarter and earlier care.

Technical Summary

1. Problem Statement

Acute Kidney Injury (AKI) and Chronic Kidney Disease (CKD) are traditionally viewed as distinct clinical entities, yet they are deeply interconnected: AKI accelerates CKD progression, and pre-existing CKD increases susceptibility to AKI. Current clinical management relies on functional metrics (e.g., serum creatinine, eGFR) that measure the consequence of injury rather than its cellular cause. Furthermore, histopathological scoring is often invasive, late-stage, and incomplete.

• The Gap: There is a lack of an unbiased molecular framework that bridges cellular mechanisms with clinical diversity. Existing single-cell studies often rely on supervised clinical definitions or mouse models, failing to capture the continuous, intertwined nature of AKI and CKD in human patients.
• The Goal: To identify shared, continuous molecular programs of injury across AKI and CKD using unsupervised analysis of human kidney biopsies, and to translate these tissue-level findings into non-invasive biomarkers predictive of long-term outcomes.

2. Methodology

The study employed a multi-omics, patient-centric approach leveraging data from the Kidney Precision Medicine Project (KPMP) and the UK Biobank.

• Data Sources:

  • KPMP Cohort: Single-nucleus RNA sequencing (snRNA-seq) and single-cell RNA sequencing (scRNA-seq) data from 132 living participants (101 snRNA, 67 scRNA) undergoing research kidney biopsies. Participants included AKI, CKD (Diabetic/Hypertensive), and Diabetes Mellitus-Resistors (DM-R).
  • Proteomics: Paired plasma and urine proteomics (SomaScan platform) for a subset of KPMP participants.
  • External Validation: Plasma proteomics (Olink platform) and longitudinal clinical outcomes from the UK Biobank (n ≈ 45,000).

• Analytical Pipeline:

  1. Multicellular Factor Analysis (MCFA): An unsupervised dimensionality reduction technique (using MOFA+) was applied to pseudobulked gene expression profiles across multiple cell types (tubular epithelial, immune, fibroblast, endothelial, etc.). This decomposed variance into latent factors representing coordinated gene programs.
  2. Factor Annotation: Factor scores were correlated with clinical metadata (enrollment category, adjudicated diagnosis, histopathology scores like TIV scoring) to biologically interpret the factors.
  3. Mechanistic Dissection:
     • Compositional vs. Molecular: The study distinguished between programs driven by changes in cell type abundance (compositional) versus changes in gene expression within cell states (molecular).
     • Regulatory Inference: Transcription Factor (TF) activity was inferred using CollecTRI and MSigDB to identify upstream regulators (e.g., NFkB, AP-1, STAT, PPARA).
  4. Non-Invasive Translation: Linear models (LASSO regression) were trained to predict tissue-derived program scores using plasma and urine protein abundances.
  5. External Validation: The trained models were applied to the UK Biobank to assess associations with baseline eGFR and incident renal events (AKI/CKD) over ~14 years.

3. Key Contributions

• Discovery of Shared Programs: Identified two distinct, continuous multicellular programs ("Acute" and "Chronic") that cut across traditional disease boundaries (AKI vs. CKD).
• Mechanistic Stratification: Demonstrated that patient heterogeneity is better explained by the degree of acute and chronic program activation rather than discrete clinical labels.
• Non-Invasive Surrogates: Successfully mapped complex tissue-level transcriptomic programs to specific plasma and urine protein signatures, creating a "liquid biopsy" equivalent.
• Longitudinal Prediction: Validated that these molecular signatures predict future renal outcomes (incident AKI and CKD) in a massive, independent population cohort.

4. Key Results

A. Identification of Acute and Chronic Programs

• Acute Program (Factor 2):
  • Associations: Strongly correlated with AKI enrollment, Acute Tubular Injury (ATI), acute interstitial nephritis, and higher AKI stages. Linked to eGFR decline and urinary albumin.
  • Cellular Basis: Characterized by an increase in immune cells and a shift in tubular epithelial cells (PT, TAL, DCT, CNT) from healthy states to adaptive states (aPT, cycPT).
  • Molecular Drivers: Upregulation of NFkB, AP-1, and STAT family transcription factors. Pathways include interferon response, angiogenesis, and apoptosis.
• Chronic Program (Factor 5):
  • Associations: Correlated with CKD enrollment, diabetic nephropathy, hypertensive kidney disease, and histologic markers of chronic damage (global glomerulosclerosis, interstitial fibrosis, tubular atrophy).
  • Cellular Basis: Associated with an expansion of fibroblasts and immune cells, and a shift in tubular cells toward failed-repair states (frPT, frTAL).
  • Molecular Drivers: Upregulation of MYC and TP53; downregulation of metabolic regulators PPARA, PPARG, HNF1A, and HNF4A. Notably, NFkB activity inversely correlates with PPARA, suggesting a mechanism where acute inflammation suppresses metabolic homeostasis, driving chronic fibrosis.

B. Compositional vs. Molecular Dynamics

• The Acute Program captures a mix of compositional shifts (more adaptive cells) and widespread molecular stress responses across all cell states.
• The Chronic Program is more specific to compositional shifts toward failed-repair states and fibroblast expansion, with a distinct molecular signature of metabolic suppression.

C. Non-Invasive Biomarker Signatures

• Acute Signatures: Plasma proteins (SPP1, KIM-1/HAVCR1, LRCH4) and urine proteins (TNC, CCL2) strongly predicted the acute program score.
• Chronic Signatures: Plasma proteins (CNPY3, CD46, SPOCK2) and urine proteins (ARL5B, EGF) predicted the chronic program score.
• Validation: These protein-based models successfully stratified KPMP participants by disease status and separated healthy controls from those with kidney stones or CKD.

D. Clinical Outcomes in UK Biobank

• Baseline Association: Higher acute and chronic scores were significantly associated with lower baseline eGFR.
• Longitudinal Risk:
  • Acute Score: Predicted incident AKI, with the strongest risk in the first 6 months (HR ~2.14).
  • Chronic Score: Predicted incident CKD, with risk persisting over years.
  • "Double Hit" Effect: Participants with high scores in both programs ("Double Hit") had the highest risk for both incident AKI (HR 2.13) and CKD (HR 2.23), supporting an additive risk model.

5. Significance and Implications

• Redefining Disease Classification: The study challenges the binary view of AKI vs. CKD, proposing a continuum where patients possess varying degrees of acute and chronic molecular injury.
• Mechanistic Insight: It elucidates the regulatory network (NFkB/PPARA antagonism) linking acute inflammation to chronic metabolic failure and fibrosis.
• Clinical Translation: The identification of plasma/urine protein signatures allows for non-invasive patient stratification. Clinicians could potentially assess a patient's position on the AKI-CKD continuum and predict long-term risk without repeated biopsies.
• Therapeutic Targets: The specific transcription factors (NFkB, AP-1, PPARA) and cell states (failed-repair) identified offer precise targets for developing therapies aimed at halting the transition from adaptive to failed-repair states.

In summary, this work provides a robust, unsupervised molecular framework that unifies the understanding of kidney injury, bridging the gap between single-cell biology, non-invasive biomarkers, and long-term clinical outcomes.