We Ranked Sirolimus #4 Out of 1,170 Drugs for HGPS.
Here's What Clinical Science Found After.

The Problem With Rare Disease Drug Discovery

Hutchinson-Gilford Progeria Syndrome (HGPS) affects approximately 1 in 20 million children worldwide. There are fewer than 400 known patients alive at any given time. For diseases this rare, the traditional drug discovery pipeline — billions of dollars, a decade of development — simply doesn't make economic sense.

Drug repurposing offers a different path: instead of building a new molecule from scratch, you ask whether an existing, approved drug might work for a disease it was never designed to treat.

The challenge is knowing where to look.

What Our Platform Found

At ORPHERA, we run computational screens that compare disease-specific cellular signatures against the known effects of thousands of approved compounds. The underlying logic: if a drug shifts a cell's molecular state in the opposite direction of a disease, it may reverse that disease's pathology.

When we screened HGPS using this approach:

That's a strong signal. But a computational rank is only a hypothesis. The real question is: does the biology hold up?

What the Literature Says

We cross-referenced our result against the independent scientific literature. What we found was striking.

The mechanism makes sense. Sirolimus inhibits mTOR (mechanistic target of rapamycin), a central regulator of cellular aging and protein clearance. In HGPS, cells accumulate a toxic protein called progerin — a mutant form of lamin A that distorts the nuclear membrane and drives premature aging. Rapamycin activates autophagy, the cell's internal recycling system, which clears progerin and reverses several hallmarks of HGPS at the cellular level. This mechanism was described in a 2011 review in Rejuvenation Research (Mendelsohn & Larrick, DOI: 10.1089/rej.2011.1238) and confirmed in HGPS fibroblast models in a 2012 Autophagy paper co-authored by Francis Collins of the NIH (Graziotto et al., DOI: 10.4161/auto.8.1.18331).

The effect extends beyond HGPS. A 2018 paper in PNAS tested everolimus — a close structural analog of sirolimus and fellow mTOR inhibitor — across six different laminopathy cell lines, including atypical Werner syndrome and Emery-Dreifuss muscular dystrophy. Every single cell line showed improvement: delayed cellular senescence, reduced nuclear blebbing, restored proliferative capacity (DuBose et al., DOI: 10.1073/pnas.1802811115).

It's now in clinical trials. As of 2023, lonafarnib (the first FDA-approved HGPS drug) and everolimus are being tested in combination in a Phase I/II clinical trial using iPSC-derived tissue-engineered blood vessel models from HGPS patients. The combination showed additive benefits over either drug alone (Abutaleb et al., Scientific Reports, 2023, DOI: 10.1038/s41598-023-32035-3).

What This Means for Computational Drug Repurposing

Let's be clear about what we're not claiming. We did not discover sirolimus as an HGPS treatment — researchers at NIH and elsewhere did that work through years of careful experimentation. What we're demonstrating is something different:

This is the validation logic that matters for drug repurposing platforms: not whether the algorithm can explain results it was trained on, but whether its predictions align with evidence generated independently, by different teams, using different methods.

In this case, they do.

Why It Matters for Your Pipeline

If your organization is developing therapies for a rare disease — whether in the laminopathy space or elsewhere — the question isn't whether computational screening can find real signals. The question is whether the data you're screening against is rich enough, and specific enough, to find your disease's signal.

Literature-mined datasets aggregate what's already published. Cell-level imaging data captures what's actually happening in disease-relevant cells — including signals that haven't been published yet.

That's the difference we're building on.

What's Next

In our next post, we'll look at what the same mTOR inhibitor class predicted for six rare diseases simultaneously — and what a 2018 PNAS paper tells us about cross-disease drug prediction.