The Case of the Two MS Gangs

More than 600 people with MS. Same diagnosis, wildly different trajectories. Some collect new lesions like parking tickets. Others lose brain volume quietly, efficiently, almost politely. The old drawers labeled “relapsing” and “progressive” help… but only up to a point.

So the question that instantly tickled my inner SherlockMS was this: Are there hidden MS types we’re missing and can MRI plus blood markers tell us which “gang” a patient belongs to?

That’s exactly what a new Brain paper investigated: combining MRI-derived features with the blood biomarker neurofilament light (sNfL), using a learning algorithm to uncover patterns without being told what to look for.

🔬 Evidence bag: MRI images and a blood trail

The case file is hefty:

• high-resolution MRI scans: brain volumes, lesion burden, subtle T1/T2 signal changes
• blood sNfL: a marker reflecting ongoing neuro-axonal injury
• an algorithm called SuStaIn—a kind of digital profiler that discovers subtypes and stages from raw data

Instead of sorting patients into “RRMS” or “SPMS,” the model was fed the actual evidence: limbic cortex, deep grey matter, parietal regions, corpus callosum, plus sNfL.

After multiple rounds of modeling, cross-validation, and statistical teeth-grinding, one thing became clear: In MS’s underworld, two distinct gangs emerged.

Naturally, I named them:

• the early-sNfL gang 😈
• the late-sNfL gang 🐍

😈 The early-sNfL gang: loud, aggressive, visibly restless

This first group stands out like a crew of street bruisers:

• sNfL spikes early—the blood trail appears right away 🩸
• early microstructural damage in the corpus callosum, reflected in altered T1/T2 signals
• new lesions accumulate faster

On average, these patients are younger and show a pattern that screams active inflammation plus parallel neurodegeneration: fire in the system, and plenty of it.

The model also suggests:

• a much higher risk of new contrast-enhancing lesions—roughly more than double compared with the other gang 🚨
• faster average brain volume loss

But here’s the key detective detail: Under treatment, both sNfL levels and active lesion activity drop particularly strongly. This gang causes early, loud trouble, but it’s also notably responsive when you intervene decisively.

🐍 The late-sNfL gang: quiet, covert, easy to underestimate

The second group operates more like a background syndicate:

• sNfL looks normal at first, the blood seems calm, almost boring
  
• yet early volume loss shows up in specific regions, especially limbic cortex and deep grey matter
  
• sNfL rises later, after structural damage has already been unfolding in the shadows

These patients are, on average, older. Early on, there’s less obvious inflammatory “noise,” but a more smoldering structural injury pattern.

Their risk of new lesions is lower than in the early gang, and atrophy progresses somewhat more slowly. Treatment helps here too, but the tempo is different. This is less “bank heist at noon” and more “slow, stubborn case that refuses to make headlines.”

📊 The algorithm beats the classic glasses

A key quantitative punchline: A model using MRI + sNfL describes disease severity (as measured by EDSS) better than an older approach relying on MRI alone.

In detective terms:

• MRI only = a blurry black-and-white photo 📷
  
• MRI + sNfL = a sharp color image with a live damage ledger overlaid

It also does better at predicting:

• who is likely to develop new lesions soon, and
• whose brain atrophy is likely to progress faster.

And crucially, it doesn’t just assign a gang—it also places each person on a stage along a disease timeline. Not only “which syndicate,” but also “how far up the ladder.”

💡 What does this mean for real-life MS care?

Before anyone panics: no, tomorrow’s MS clinic won’t be reorganized into “early-sNfL gang, second floor left” and “late-sNfL gang, follow the grey carpet.”

But the case makes a few things very clear:

• The classic labels “relapsing” vs “progressive” are too coarse to capture the underlying biology.
  
• Combining MRI patterns with blood sNfL can uncover hidden MS subtypes that differ in inflammation, brain shrinkage, and treatment responsiveness.
• The early-sNfL gang is particularly clinically interesting: high activity, higher risk, but also strong treatment impact—exactly the group where early, firm action matters most.
  

Long term, approaches like this could help:

• select therapies more precisely,
• stratify clinical trials more intelligently, and
• counsel patients more realistically about the pace and “style” of their disease course.
  

🧠 Closing notes from Baker Street

So I sit in my London flat. Outside, a taxi honks. Somewhere, a bus driver is swearing at a delivery van that has chosen chaos as a parking strategy.

While the city wrestles with visible traffic jams, we’ve started mapping the invisible jams inside the brain—between inflammation, axonal damage, and structural loss—into two major patterns.

For me, the conclusion is simple: Multiple sclerosis isn’t a single perpetrator. It’s at least two organized gangs, loud and early versus quiet and late, that we’re finally learning to identify properly.

And the more precisely we know who’s causing trouble in the nervous system, and where a patient sits on that gang’s career ladder, the sooner we can send the right response team.

The case isn’t closed. But with this work, the crime scene is far better mapped,
and that shifts the odds a little more in the patient’s favor.

Yours Truly,

SherlockMS