📄 genetic and genomic medicine

Metatranscriptomics-Derived Disease Risk Scores as a Preventive, Diagnostic, and Treatment Support Tool

This study presents and validates a metatranscriptomics-based Disease Risk Score (DRS) framework that integrates active microbial and human gene expression signals from stool, saliva, and blood to stratify risk for 15 chronic conditions, demonstrating its potential as an adjunct decision-support tool for preventive, diagnostic, and treatment guidance in clinical scenarios characterized by uncertainty or delayed diagnosis.

Original authors: Hu, L., Bass, M., Patridge, E., Molusky, M., Antoine, G., Vuyisich, M., Banavar, G.

Published 2026-06-06

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hu, L., Bass, M., Patridge, E., Molusky, M., Antoine, G., Vuyisich, M., Banavar, G.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine your body is a bustling, high-tech city. For years, doctors have mostly looked at the city's blueprints (your DNA) to guess what might go wrong in the future. But blueprints don't tell you what the city is doing right now. Is the traffic jammed? Are the power plants running hot? Is the sanitation crew working overtime?

This paper introduces a new tool called the Disease Risk Score (DRS) that acts like a real-time "city status report." Instead of just looking at the static blueprints, it listens to the active conversations happening in the city's neighborhoods to spot trouble before a full-blown disaster occurs.

Here is how the paper explains this tool in simple terms:

1. The Three Neighborhoods (The Samples)

To get a full picture of the city's health, the researchers look at three specific "neighborhoods":

The Gut (Stool): This is the city's recycling and food processing plant.
The Mouth (Saliva): This is the city's front gate and air filtration system.
The Bloodstream: This is the city's central communication network, carrying messages from all departments.

The researchers don't just count the "residents" (bacteria) in these neighborhoods; they listen to what the residents are saying and doing right now. They use a technology called metatranscriptomics, which is like recording the active radio broadcasts of the city's workers rather than just taking a census of who lives there.

2. The "Not Optimal" Signals (The Pathways)

The researchers identified 41 specific "jobs" or "processes" happening in these neighborhoods (like making gas, breaking down fats, or fighting inflammation). They call these pathways.

The Good News: Some pathways are running smoothly (like a well-oiled machine).
The Warning Signs: Other pathways are "not optimal." Maybe the gut is producing too much methane gas, or the mouth is struggling to balance its pH levels.

The DRS tool scans your samples and counts up all the "not optimal" signals. It then combines them into a single number for specific diseases, similar to how a weather app combines temperature, wind, and humidity to give you a "heat index."

3. The "Risk Score" (The Dashboard)

The tool takes these messy, complex signals and turns them into a simple Disease Risk Score (DRS) for 15 different conditions, including:

Digestive issues (like IBS, GERD, and IBD)
Metabolic issues (like Type 2 Diabetes, Obesity, and fatty liver)
Mental health and sleep (like Anxiety, Depression, ADHD, and Insomnia)
Immune issues (like Sjögren's syndrome)

If your score for a specific disease is 5 or higher, the tool flags you as "High Risk." This doesn't mean you definitely have the disease right now. It means your city's "active signals" look like a neighborhood that is on the verge of a problem.

4. How It Helps Doctors (The "Decision Support" Tool)

The paper is very clear: This is not a diagnostic test. It won't tell a doctor, "You have IBS." Instead, it acts as a co-pilot for the doctor to help solve tricky cases.

The paper gives five real-world examples of how this works:

The "Wrong Diagnosis" Fix: One patient thought they had diarrhea (IBS-D), but the tool showed their gut was actually producing too much methane gas (a sign of constipation/IBS-C). The gas was causing "overflow," making it look like diarrhea. The doctor changed the treatment, and the patient got better.
The "Convincing" Factor: Another patient had diarrhea but didn't believe they had IBS. The tool showed specific "bad" signals (like sulfide gas) that explained why they felt sick. This helped the patient trust the diagnosis and stick to the treatment plan.
The "Motivation" Boost: A patient with a fatty liver (MASLD) thought it was a minor issue and ignored advice to change their diet. The tool showed specific metabolic pathways failing, making the risk feel real and personal. This motivated them to finally make lifestyle changes.
The "Sleep & Stress" Link: For a patient with early Type 2 Diabetes, the tool highlighted gut bacteria signals linked to stress and sleep. This helped the doctor explain that fixing their sleep schedule was just as important as watching their sugar intake.
The "Atypical" Case: A patient had a chronic cough and hoarseness but no heartburn. The tool showed their mouth bacteria were creating acid. This helped the doctor realize it was actually acid reflux (GERD) acting strangely, leading to the right treatment.

5. The Bottom Line

The researchers tested this tool on over 15,000 people. They found that for 15 out of 20 diseases they looked at, people who said they had the disease also had much higher "Risk Scores" than those who didn't.

In short: The DRS is a new way to listen to the active conversations inside your body. It doesn't replace a doctor's diagnosis, but it gives them a powerful flashlight to find hidden problems, explain confusing symptoms, and help patients understand why they need to make changes before they get sick.

Note: The paper emphasizes that this is a "preprint" (a draft study) and is intended to support doctors, not replace them. It is a tool for risk evaluation, not a final medical verdict.

Technical Summary: Metatranscriptomics-Derived Disease Risk Scores as a Preventive, Diagnostic, and Treatment Support Tool

Problem Statement
Chronic diseases and symptom syndromes often develop following prolonged biological changes that precede formal clinical diagnosis. This creates a practical gap between "apparently healthy" individuals and those who are "clinically diagnosable," characterized by nonspecific symptoms, borderline laboratory results, or low patient acceptance of diagnoses when the biological basis is not apparent. While Polygenic Risk Scores (PRS) have established a framework for aggregating genetic effects to estimate risk, they rely on static inherited variation. Many chronic diseases are driven by dynamic, modifiable processes involving immune activation, metabolic regulation, and microbiome activity. There is a need for a risk stratification tool that captures these active, functional molecular signals to support earlier prevention, targeted evaluation, and improved patient engagement in routine care.

Methodology
The study developed and validated a Disease Risk Score (DRS) framework based on RNA-based metatranscriptomics, which captures active gene expression rather than genomic potential.

Data Sources: The framework integrates data from three sample types: stool and saliva (for microbial functions and taxa) and blood (for human gene expression).
Cohorts:
- Development Cohort ( $N=22,369$ ): Used to curate disease-pathway associations and calculate odds ratios. Participants were adults (18+ years) with high-quality metatranscriptomic data.
- Validation Cohort ( $N=15,908$ ): An independent cohort used to validate performance against self-reported disease labels.
Pathway Scoring: The model utilizes 41 pathway scores derived from KEGG Orthologs (microbial functions), microbial species, and human genes. Scores range from 0 to 100, standardized to a Gaussian-like distribution. Scores are categorized as "not optimal" (bottom 25%, $\le 35$ ), "average" (middle 50%), or "good" (top 25%, $\ge 65$ ).
DRS Calculation:
- For each of 20 initially evaluated chronic diseases, specific "not optimal" pathway scores were associated with the disease based on literature and data analysis.
- Logistic regression was used in the Development Cohort to calculate odds ratios (OR) for the association between "not optimal" scores and self-reported disease status, adjusting for age, sex, and BMI.
- A Cumulative Odds Ratio (cOR) was computed for each individual and disease by aggregating the weighted contributions of relevant "not optimal" pathway scores.
- A cOR $\ge 5$ is defined as "high risk."
Validation Criteria: To be included in the final risk panel, a disease had to meet two criteria in the Validation Cohort:
1. Distributional Shift: A statistically significant difference in cOR between self-reported cases and non-cases with a Cohen's $d \ge 0.2$ .
2. Risk Enrichment: A Risk Ratio (RR) of self-reported disease prevalence in the "high risk" group versus the non-high-risk group where the lower bound of the 95% confidence interval (CI) is $> 1$ .

Key Results

Validation Performance: Of the 20 initially evaluated diseases, 15 met both prespecified validation criteria. These include: ADHD, anxiety, chronic fatigue syndrome, depression, GERD, hypertension, inflammatory bowel disease (IBD), IBS-C, IBS-D, insomnia, MASLD (metabolic dysfunction-associated steatotic liver disease), obesity, obstructive sleep apnea, Sjögren's syndrome, and type 2 diabetes (T2D).
Statistical Significance: For all 15 validated diseases, self-reported cases exhibited significantly higher mean cOR values than non-cases. The lower bounds of the 95% CIs for the mean differences did not cross zero.
Risk Enrichment: Participants classified as "high risk" (cOR $\ge 5$ ) showed a consistent enrichment of self-reported disease prevalence compared to those not classified as high risk. The lower bounds of the 95% CIs for the Risk Ratios exceeded 1 for all 15 diseases.
Clinical Utility Scenarios: The authors presented five illustrative cases demonstrating how DRS can support clinical decision-making:
1. IBS Reclassification: A patient presenting with IBS-D symptoms was reclassified as IBS-C based on a high-risk DRS and "not optimal" methane gas production, leading to appropriate imaging and therapy.
2. Diagnostic Acceptance: A patient skeptical of an IBS-D diagnosis accepted the diagnosis and targeted treatment (rifaximin) after DRS revealed "not optimal" sulfide gas production and bile acid metabolism pathways.
3. MASLD Adherence: A patient with low adherence to lifestyle changes for MASLD engaged in personalized dietary modifications after DRS highlighted specific "not optimal" bile acid and oxalate metabolism pathways.
4. Early T2D Prevention: A patient with borderline T2D markers adopted targeted behavioral changes (e.g., reducing sugar-sweetened beverages) after DRS linked their risk to specific GABA and uric acid production pathways.
5. Atypical GERD: A patient with atypical GERD symptoms accepted a treatment trial after DRS linked symptoms to oral cariogenic and urease activity pathways.

Significance and Claims
The paper claims that the DRS framework successfully translates metatranscriptomic data into interpretable, disease-specific risk estimates that function as an adjunct to clinical evaluation, not as a standalone diagnostic tool.

Dynamic Risk Stratification: Unlike PRS, which reflects static genetic risk, DRS captures the current functional state of the microbiome and host, offering a dynamic view of disease risk that may precede formal diagnosis.
Bridging the Clinical Gap: The framework is designed to address situations where routine care encounters uncertainty, delay, or low patient engagement. By providing biologically interpretable pathway signals (e.g., specific gas production or metabolic pathways), DRS aims to improve patient understanding, facilitate diagnostic acceptance, and enable more personalized preventive interventions.
Scope: The study validates the framework across a broad spectrum of conditions, including gastrointestinal, cardiometabolic, neurobehavioral, sleep, and immune-mediated diseases, demonstrating the utility of multi-sample (stool, saliva, blood) integration.

Limitations and Future Directions
The authors explicitly note that the validation relies on self-reported disease labels rather than clinically adjudicated diagnoses, which may introduce recall bias or misclassification. The study is cross-sectional, meaning it cannot establish causality or whether DRS signals precede disease onset. The clinical cases presented are illustrative examples of potential utility, not definitive evidence of improved outcomes. The authors conclude that future prospective studies with clinically adjudicated endpoints are necessary to assess calibration, clinical outcomes, and the impact of DRS-guided care on patient adherence and disease progression.