How Not to Validate a Biomarker In 14 vaccinated long COVID patients who were healthy enough for intense exercise

When 14 patients and 52 words of discussion meet logical analysis

https://link.springer.com/article/10.1007/s15010-025-02612-x

The Setup: Why This Matters

Three years into long COVID, millions of patients desperately need answers. They need diagnostic tests. They need treatments. They need hope grounded in solid science.

What they don’t need? Premature headlines about “breakthroughs” based on preliminary observations.

Today, we’re dissecting a new study (Abbasi et al., 2025) that will likely generate excited headlines about discovering a long COVID biomarker. We’ll show you exactly why those headlines are wrong – and why that matters.

The Study at a Glance

What they did: Studied 14 long COVID patients, looking for viral proteins in their blood
What they found: SARS-CoV-2 proteins in tiny packages called extracellular vesicles
What they claim: Potential biomarker for long COVID
What it actually shows: Sometimes, in some samples, from some patients, during exercise, we found some viral fragments

Let’s dig deeper.

🚩 Red Flag #1: The Exercise Testing Shell Game

What The Paper Says:

Samples were collected “at rest and peak exercise”

What The Paper Doesn’t Emphasize:

All sampling occurred during exercise testing visits
No samples from non-exercise days
No analysis comparing rest vs. peak detection rates

Why This Destroys Their Claims:

Exercise is like shaking a biological snow globe. It mobilizes proteins, cellular debris, and yes – potentially old viral fragments – from tissues throughout your body. When you claim proteins are “persistent” but only sample during exercise sessions, you’re not measuring persistence. You’re measuring what exercise shakes loose.

The Critical Omission: They never tested whether exercise increased detection. They had rest and peak samples but never compared them. Why? Did the analysis not support their narrative?

Think about it: If you only checked for dust after beating your carpets, would you conclude your house has persistent dust, or that beating carpets releases dust?

🚩 Red Flag #2: The Incredible Vanishing Proteins

The Mathematics of “Persistence”:

What they claimed: "Persistent" proteins, Found in "all" patients, Biomarker potential

What The Data Shows: Detected in only 39% of samples, Each patient had 4 samples; most were negative, 61% false negative rate?

If something is truly persistent, it doesn’t play hide-and-seek. The sporadic detection pattern suggests either:

These proteins aren’t actually persistent
The test is unreliable
Exercise randomly mobilizes them
All of the above

Statistical Reality Check: A biomarker that’s undetectable 61% of the time isn’t a biomarker – it’s a lottery ticket.

🚩 Red Flag #3: The Sample Size Catastrophe

Biomarker Development Requirements:

Discovery Phase:        30-50 patients minimum
Validation Phase:       100-200 patients
Clinical Validation:    500+ patients
FDA Approval:          1000+ patients

This Study:            14 patients 🤦

With 14 patients, you can’t validate a pizza topping preference, let alone a clinical biomarker.

Context: The FDA has never approved a biomarker based on 14 patients. Ever. For anything.

🚩 Red Flag #4: The Vaccine Paradox (With a Twist)

The Setup:

13 of 14 patients were vaccinated
Only 1 unvaccinated patient (no comparison possible)

The Plot Twist:

The protein they found (Pp1ab) is NOT in any COVID vaccine. Vaccines only contain spike protein. This actually HELPS their case – these proteins must be from infection, not vaccination.

The Problem:

They never emphasized this critical distinction!

Instead of clearly stating “we found non-vaccine proteins proving infection persistence,” they buried this crucial fact. Why? Either they didn’t realize its importance (concerning) or didn’t think it through (more concerning).

What’s Still Missing:

When were patients vaccinated?
How many doses?
Could vaccination affect viral clearance?
Do unvaccinated long COVID patients show different patterns?

Without this data, we’re flying blind.

🚩 Red Flag #5: The One-Hit Wonder Validation

The Cherry-Picking Championship:

Proteins Discovered:     65 different peptides
Proteins Validated:      1 peptide
Validation Rate:         1.5%
Scientific Credibility:  📉

Imagine a teacher grading only question #37 on a 65-question exam and declaring the student passed. That’s what happened here.

Standard Practice: Validate at least 3-5 candidates using orthogonal methods
This Study: Picked their favorite and called it a day

🚩 Red Flag #6: The Clinical Correlation Black Hole

Questions They Should Have Answered:

❌ Do sicker patients have more proteins?
❌ Do proteins decrease as symptoms improve?
❌ Which symptoms correlate with protein levels?
❌ Can protein levels predict outcomes?
❌ Do proteins change with treatment?

Questions They Did Answer:

[This section intentionally left blank because they answered none of them]

Remember: A biomarker that doesn’t correlate with disease is like a thermometer that doesn’t correlate with temperature – useless.

🚩 Red Flag #7: The Time-Machine Control Group

Their Controls:

Blood from 2019 (pre-COVID era)
From ex-smokers
Different storage conditions
Different world, literally

Proper Controls Would Be:

Current samples
From recovered COVID patients without long COVID
Same storage and processing
Matched for age, sex, vaccination status

Using pre-2019 samples as controls is like comparing 2025 teenagers to 1950s teenagers and concluding smartphones cause anxiety. The contexts are incomparable.

🔴 The Ultimate Red Flag: The “Discussion” That Wasn’t

Word Count Analysis:

Typical Discussion Section:     500-1,500 words
This Paper's Discussion:        52 words
Percentage of Normal:           3.5%
Red Flags Raised:              🚩🚩🚩🚩🚩

Their Entire Discussion:

“Overall, our targeted analysis confirmed the presence of the Pp1ab peptide in at least one sample in all long COVID subjects, while absent in controls. This finding will require confirmation in other long COVID studies with appropriate controls to establish the durability and sensitivity/specificity of this peptide biomarker.”

What’s Missing From Their 52-Word “Discussion”:

✗ Why detection was sporadic
✗ How exercise affects results
✗ Sample size limitations
✗ Vaccine considerations
✗ Clinical relevance
✗ Comparison with other studies
✗ Biological interpretation
✗ Alternative explanations
✗ Statistical limitations
✗ Future directions beyond “needs confirmation”

This is academic malpractice. Publishing biomarker claims without discussing limitations is like selling a car without mentioning it has no brakes.

🎯 What This Study Actually Found

Let’s translate from science-speak to reality:

They claim: “Potential long COVID biomarker discovered”

Reality: “In 14 vaccinated long COVID patients who were healthy enough for intense exercise, we sometimes found viral protein fragments during exercise testing, using a method that worked 39% of the time, with no idea what it means clinically, compared to samples from before COVID existed, and we couldn’t be bothered to discuss any of this.”

📋 How Real Biomarker Development Works

The Proper Pipeline:

Phase 1: Discovery (This study is barely here)

Initial observation in small cohort
Multiple candidate identification
Proof of concept

Phase 2: Validation (Missing entirely)

100+ patient cohort
Orthogonal confirmation methods
Establish detection parameters

Phase 3: Clinical Correlation (Not attempted)

Link to symptoms/severity
Prognostic value assessment
Treatment response correlation

Phase 4: Clinical Validation (Years away)

Multi-site studies
500+ patients
Regulatory approval process

This study is at step 1 of phase 1, claiming they’re ready for phase 4.

💔 Why This Matters: The Human Cost

Who Gets Hurt:

Desperate Patients
- False hope from premature headlines
- Potential inappropriate testing
- Delayed focus on real solutions
Legitimate Researchers
- Funding diverted to dead ends
- Credibility undermined
- Higher bar for future claims
Clinicians
- Confusion about what to tell patients
- Pressure to order unvalidated tests
- Time wasted on non-solutions
Science Itself
- Public trust eroded
- Peer review questioned
- Standards lowered

✅ What Should Happen Next

For Researchers:

Recruit 100+ patients minimum
Include proper contemporary controls
Test resting (non-exercise) samples
Validate multiple peptides
Correlate with clinical symptoms
Write a real discussion section

For Journals:

Reject biomarker claims without proper validation
Require minimum sample sizes
Demand comprehensive discussion sections
Don’t allow major claims in “correspondence” format

For Patients:

Ask about sample sizes in studies
Look for clinical correlation
Check if there’s a real discussion section
Be skeptical of “breakthrough” headlines
Demand better science – you deserve it

For Media:

Stop writing “breakthrough” headlines for n=14 studies
Interview independent experts
Read the discussion section
If it’s under 100 words, it’s not ready for headlines

🔬 The Path Forward

This study found something potentially interesting: non-vaccine viral proteins in some long COVID patients during exercise. That’s worth investigating. But it’s not a biomarker, not a breakthrough, and not ready for clinical use.

Good science builds slowly. It validates thoroughly. It discusses limitations honestly. It replicates reliably.

This paper does none of those things.

Long COVID patients deserve better than premature hope based on 14 patients and 52 words of discussion. They deserve rigorous science that acknowledges uncertainty, validates thoroughly, and builds knowledge brick by careful brick.

📝 Final Thoughts

Science is hard. Biomarker development is harder. But cutting corners helps no one.

When researchers claim breakthroughs without proper validation, journals publish without adequate review, and media amplifies without verification, we all lose.

The solution isn’t to stop searching for long COVID biomarkers. It’s to search properly.

14 patients and 52 words don’t make a breakthrough. They make a preliminary observation that needs years of careful work to validate.

Let’s demand that work gets done before we declare victory.

Remember the key questions for any “breakthrough” study:

How many patients? (Under 30? Be skeptical)
What controls? (Wrong era? Red flag)
Clinical correlation? (None? Not a biomarker)
Discussion section? (Under 500 words? Not ready)

If it fails these tests, it’s not a breakthrough – it’s a work in progress being oversold.

Dedication: To the millions living with long COVID: You deserve rigorous science, not rushed headlines. Keep demanding better. Your health depends on it.