I’m guilty of being over-invested in fibrosis. At first glance, it is an extremely compelling space in which to think about new drug development. There are few effective therapies, a large unmet need, and seemingly a very rich substrate of biology to unpack and identify improved therapeutic targets. However, as I’ve spent the last week reviewing fibrosis datasets, trying to identify the most promising mechanisms, I’m realizing that there is a good reason why few effective therapies exist.
During aging, people feel ‘stiff’. Frozen shoulder, the buildup of scars, and especially organ specific fibrosis (e.g. heart, kidney, liver, lungs) are all very common and often viewed as normal. The resulting effects can range from simply cosmetic to life threatening. It can be exciting to think that an ‘anti-fibrotic’ mechanism can thus have broad applicability across many age related diseases. Massive markets unlocked by a single or few mechanisms. After all, the disease should be relatively homogeneous: fibroblasts have rarely been reported to have somatic mutations outside of UV induced DNA damage in the skin, and unlike I&I, there are not established endotypes of disease (e.g. Th2, Th17, interferon, etc).
Fibrosis is an entirely natural process. The ‘natural’ role of fibroblasts is to support the recovery and homeostasis of nearby epithelial and endothelial cells. In many instances, this comes in the form of ‘filling in space’ from dead/dying cells by proliferating and secreting ECM, but fibroblasts can also play a role by putting out cytokines and other growth factors. Ultimately, their behavior is controlled by cellular crosstalk and communication, most importantly with local epithelial and myeloid cells. There was a seminal paper by Ruslan Medzhitov and Uri Alon showing that fibroblasts and myeloid cells exhibit reciprocal control over respective population sizes within tissue. Macrophages secrete PDGF to sustain proliferation of myofibroblasts and are maintained by secreted CSF1 from fibroblasts. Similarly, epithelial cells in healthy tissue are able to repopulate and regrow after damage, but through aging, lose their ability to regenerate properly. Aged epithelia that are unable to regenerate are more inflammatory and chronically secrete fibroblast growth factors and activate latent TGFb to support fibroblast proliferation. Fibrosis is not as much a problem of fibroblasts, but in my view primarily a problem of epithelia who have lost the ability to regenerate and chronic sterile inflammation that increases macrophage persistence in tissues.
Fibroblasts are some of the most resilient cell types. They derive from mesenchymal stem cells but are also capable of proliferating themselves. Not only are they capable of migratory behavior, but they can also secrete large amounts of protein and withstand significant mechanical stress. When I’ve cultured fibroblasts in vitro, they grow extremely fast (often faster than cancer cells), are very robust to handling, and often have a unique propensity to self organize. Fibroblasts are responsible for a disproportionately large amount of the biomass of a human due to the abundance of ECM that originates from them. While fibroblasts are susceptible to cell senescence just like any other cell type, it is not the case that the aging process automatically makes them more proliferative and more prone to secreting ECM.
One last thing, within all tissues, there is constant turnover of ECM. Matrix proteases degrade ECM, and fibroblasts secrete more. During fibrosis, there is a temporary shift in this balance to favor greater synthesis, but when fibrosis resolves, fibroblasts are programmed for apoptosis. Fibroblasts are told what to do and when they run out of instructions, they die. Very noble!
So all this to say, that there is often a very good reason why there is fibrosis. In systemic sclerosis, there are often anti-endothelial antigen autoantibodies. Continuous injury to microvasculature is bound to increase macrophage infiltration, accumulate myofibroblasts, and lead to deposition of ECM. Cardiac fibrosis does not happen randomly, but rather its a response to increased mechanical strain from being hypertensive, ischemic events, and potentially cardiac inflammation. Renal fibrosis arises from persistent kidney injury (e.g. AKIs, immune complex deposition, hypertension, etc.). Nephrons don’t repopulate and when they die, fibroblasts fill the space. In the liver, fibrosis starts when hepatocytes lose their regenerative capacity.
So how do you ‘solve’ fibrosis?
It’s complicated, because the root cause of the issue, and the targets that will have the most immediate benefit to symptoms, are somewhat independent. Below is a table listing most of the mechanisms I have found that are either compelling in preclinical work or have entered clinical trials.
| Fibrosis area | Mechanism | Company / Organization | Indication | Stage |
|---|---|---|---|---|
| Cardiac fibrosis | IL-11 | Academic (Stuart Cooke) | N/A | Academic |
| MD2 | Academic (Joseph Wu) | N/A | Academic | |
| ANTXR1 | Academic (NIH) | N/A | Academic | |
| Lung fibrosis | Prostacyclin | United Therapeutics | IPF | Approved |
| PDE4 | Boehringer Ingelheim | IPF | Approved | |
| Nintedanib (multi-TKI) | Boehringer Ingelheim | IPF | Approved | |
| Pirfenidone | Roche | IPF | Approved | |
| IL-6 | Roche | SSc-ILD | Approved | |
| CTGF | Pieris Pharma | IPF | Ph2 | |
| Hedgehog | Endeavor | IPF | Ph2 | |
| IL-11 | Boehringer Ingelheim | IPF | Ph2 | |
| S100A4 | Calluna Pharma | IPF | Ph2 | |
| STAT3 | Tvardi | IPF | Ph2 | |
| AREG | Pulmogene | IPF | Ph2 | |
| GARP/TGFβ1 | Henlius | IPF | Ph2 | |
| WISP1 | Mediar | IPF | Ph2 | |
| LPA1 | BMS / Contineum | IPF | Ph3 | |
| LPA1 | Contineum | IPF | Ph2 | |
| TNIK | Insilico Medicine | IPF | Ph2 | |
| ROCK2 | Sino Biopharmaceuticals | IPF | Ph2 | |
| RARβ/γ | GRI Bio | IPF | Ph2 | |
| Pirfenidone (next gen) | Multiple | IPF | Ph2 | |
| PTX2 | Roche | IPF | Fail | |
| CTGF | FibroGen | IPF | Fail | |
| ENPP2 | Galapagos | IPF | Fail | |
| αvβ6 / αvβ1 integrins | Pliant | IPF | Fail | |
| Trans-IL-6 | Academic | IPF | Academic | |
| FAP | Academic | IPF | Academic | |
| FGF21 | Academic | IPF | Academic | |
| Skin fibrosis | EphrinB2 | Mediar | SSc | Ph2 |
| FT011 | Certa Therapeutics | SSc | Ph2 | |
| BAFF / IL-17 | Zura | SSc | Ph2 | |
| BCMA | Candid | SSc | Ph1b | |
| CB2 agonist | Corbus Pharma | SSc | Fail | |
| GI fibrosis | CCL24 | Chemomab | PSC | Ph2 |
| PDE4 | Palisade | Crohn’s | Ph1b | |
| OSMR / TL1A | Mirador Therapeutics | Crohn’s | Preclinical | |
| LOXL2 | Gilead | PSC | Fail | |
| FXR agonist | Gilead | PSC | Fail | |
| Kidney fibrosis | SMOC2 | Mediar | CKD fibrosis | Preclinical |
| IL-11 | Academic | CKD fibrosis | Academic | |
| Latent TGFβ | Scholar Rock | Various | Preclinical | |
| Latent TGFβ | Chugai | Various | Preclinical | |
| Liver fibrosis | FGF21 | Multiple | NASH | Approved |
| THR-β partial agonist | Madrigal | NASH | Approved | |
| GLP-1 | Multiple | NASH | Approved | |
| MERTK | Academic | N/A | Academic | |
| CCR2/CCR5 | AbbVie | NASH | Fail | |
| LOXL2 | Gilead | NASH | Fail | |
| ASK1 | Gilead | NASH | Fail |
Most of our current therapies for fibrosis address the symptoms without touching the root cause. You can increase the rate of myofibroblast death (taladegib), or block signaling pathways for myofibroblast persistence (IL-11, CTGF, Nintedanib), or inhibit pathways that lead to myofibroblast proliferation (TGFb, AREG, Pirfenidone). Very few address non-fibroblast cell types, but maybe counterintuitively, the ones that do seem to have the largest effect sizes. FGF21 strongly improves MASH fibrosis scores by acting on hepatocytes. BCMA targeted T cell engagers show strong efficacy in systemic sclerosis by eliminating the root cause of pathogenic autoantibody secreting B cell clones.
Most of the drug development in the fibrosis space is in a disease called IPF. It’s technically a rare disease, but still affects up to 140k people in the U.S.. More than 60% of cases are in those who are >75 years old, and it is rarely diagnosed under age 60. Incidence is 50k annually, with median survival of just 3-5 years. I was initially shocked by how low this survival statistic is, but again these are very old people with likely incurable lung problems.
It is idiopathic and thus highly heterogeneous. GWAS implicates MUC5B (mucin protein) promoter variants and TERT (telomeres), both likely affecting epithelial cells. The genetic signal points to the same point I referred to above, which is that fibrosis is a consequence of dysfunctional epithelium. In this case, we just have accumulated epithelial damage that can no longer be compensated for in old age.
It is generally considered a ‘graveyard’ for drug development, but notably there are several mechanisms that have shown efficacy in clinical trials. Several drugs are approved, and early data from the hedgehog, AREG, and TNIK inhibitor classes are interesting. The hedgehog inhibitor seems to actually improve forced vital capacity (FVC), the endpoint used to assess lung function. Notably, these are all fibroblast targeted. Deuterated pirfenidone seems to have really strong data and I wonder about how much juice there really is left to squeeze once these various agents start being combined. I’m most looking forward to the IL-11 data which should arrive in mid 2027.
| Mechanism | Company | Asset | Phase | Timepoint | FVC on drug | FVC on placebo | Placebo-corrected delta | Other key effect size / note |
|---|---|---|---|---|---|---|---|---|
| Pirfenidone | Roche / Genentech | Pirfenidone | 3 | Week 52 | -235 mL | -428 mL | +193 mL | Also reduced proportion with ≥10% FVC decline or death; 48% slowing of worsening |
| PDE4 | Boehringer Ingelheim | Nerandolimast | 3 | Week 52 | -114.7 mL | -183.5 mL | +68.8 mL | ~38% slowing of worsening |
| Multi-TKI | Boehringer Ingelheim | Nintedanib | 3 | Week 52 | -114.7 mL/yr in INPULSIS-1; -113.6 mL/yr in INPULSIS-2 | -239.9 mL/yr; -207.3 mL/yr | +125.3 mL/yr and +93.7 mL/yr | ~50% slowing of worsening |
| Prostacyclin | United Therapeutics | Inhaled treprostinil | 3 | Week 52 | -49.9 mL | -136.4 mL | +95.6 mL | 63% slowing of worsening |
| CTGF | FibroGen | Pamrevlumab | 3 | Week 48 | -260 mL | -330 mL | +70 mL | Not statistically significant; 21% slowing of worsening |
| PTX2 | Roche / Genentech | Zinpentraxin alfa (PRM-151) | 3 | Week 52 | -235.72 mL | -214.89 mL | -20.83 mL | Ph3 stopped for futility |
| ENPP2 / autotaxin | Galapagos | Ziritaxestat | 3 | Week 52 | -173.8 mL | -176.6 mL | +2.8 mL | |
| Pirfenidone (next gen) | Multiple / PureTech | Deupirfenidone (LYT-100) | 2b | Week 26 | -21.5 mL | -112.5 mL | +91.0 mL | Time to IPF progression HR 0.439 for 825 mg vs placebo; 81% slowing of worsening |
| LPA1 | BMS | Admilparant (BMS-986278) | 2 | Week 26 | - | - | +45.5 mL | Improved rate of ppFVC decline by 1.4%; time to progression HR 0.54; 62% slowing of worsening |
| STAT3 | Tvardi | TTI-101 | 2 | Week 12 | -15 mL | -22 mL | +7 mL | |
| Hedgehog | Endeavor | Taladegib (ENV-101) | 2a | Week 12 | - | - | +107 mL | ppFVC between-group difference +3.95% at week 12; HRCT total lung capacity +257.0 mL vs placebo |
| TNIK | Insilico Medicine | Rentosertib (ISM001-055) | 2a | Week 12 | +98.4 mL | -20.3 mL | +118.7 mL | |
| AREG | Pulmogene | PMG1015 | 1b | Week 12 | - | - | +96.9 mL | CT lung volume +221.7 mL; ground-glass volume -1.02%, p=0.031; >98% receptor occupancy |
My opinion (if not already made clear) is that fibrosis is typically secondary to an epithelial or immune related issue. IPF is one example. Cystic fibrosis is another.
If you don’t want a scar, don’t get a cut.
However, what to do in otherwise healthy people with unwanted fibrosis? Cosmetic scars? Keloids? Crohn’s disease strictures? Surgery is always an option, but there must be medicines to be discovered as well.
For cutaneous scars and potentially also keloids, I’m pretty excited about siRNA STAR* particle technology that is being pioneered by Alys Pharmaceuticals to deliver siRNA to the skin.
Two studies have shown that CTGF siRNA/ASO can improve scar appearance. Both medicines, PF-06473871 and RXI-109 are discontinued, probably due to business reasons or lack of profound phenotype. These approaches only getting ~40% knockdown, but couldn’t we get more than this? CTGF antibodies were underdosed in IPF studies and I continue to think this is an active mechanism in humans despite the Ph3 pamrevlumab fail. A more potent antibody is sitting with Astellas right now, waiting to be developed. There is too much mechanistic support and preclinical model data for it to be inert and even those working at Fibrogen think its a drug specific issue!
I continue to be bullish on IL-11. Interestingly, olamkicept which is a soluble gp130-Fc protein with activity against IL-11 had efficacy data in UC and across many many settings in the literature. The IL-11 data from Stuart Cook and other groups is very compelling. Both genetic and pharmacologic targeting with high efficacy across many mouse models.
PDE4B has already shown efficacy in IPF, and Palisade has shown preliminary data in a handful of Chron’s patients. Meaningful anti-fibrotic AND anti-inflammatory activity is very nice.
Lastly, I think TL1A is going to continue to impress in upcoming clinical studies. High potency Spyre antibodies, TL1A/OSMR bispecifics (Mirador), TL1A coformulations, etc. ATHENA-SSc-ILD trial should read out later this month.
Here is a good review for additional trials in the cutaneous fibrosis space.
Where to look in terms of discovery? One interesting study published earlier this year investigated scar formation differences between the face and other peripheral tissues. The skin evolutionarily has less scarring and heals faster than other tissues. They wound EP300 inhibitors as effective scar prevention agents. Interestingly, 5-FU is used off label for resistant scars (I wonder why so many agents are effective in the fibrosis mouse models!). This paper looked for antifibrotics in the heart setting by screening on cardiac iPSCs. They found artesunate to be an MD2 inhibitor, which somehow has some anti-fibrotic effects through modulating TLR signaling. Who knows.
What would a DepMap for fibrosis look like? Fibroblasts grow pretty fast. If your goal is simply to kill them, its not that bad of an idea to do CRISPR screens on them with a factor driving their proliferation. Or, maybe do the drug repurposing/CRISPR screens with an aSMA or collagen reporter with many different patient derived cell lines from various organs and types of fibrosis. Again, root cause first, but assuming we can’t repair aged epithelial cells, this might be the next best option.
Unfortunately, the preclinical models are pretty bad. Below is a list from ChatGPT. I spent a decent amount of time trying to compile effect sizes in the bleomycin lung fibrosis model (which seems to be the most common), and I couldn’t correlate with clinical efficacy whatsoever. Its definitely true that if there are no effects whatsoever in these models, your drug/mechanism is probably a dud, but success in these models otherwise means little.
The reason is that mouse models necessitate feasibility. You need to be able to actually execute these experiments, meaning that the results have to be observable over a relatively short time frame and the fibrosis phenotype needs to be inducible in a relatively short period of time also. This is incongruent with how fibrosis typically develops in disease, which is a long chronic process.
Instead, I think the best ways of speeding up drug development are improving the value and predictability of Ph1b/2 trials. IPF is noteworthy for poor Ph2 -> Ph3 replicability. Heterogeneity of background medication certainly plays a role, but in the case of CTGF, dosing also did. Due to significant heterogeneity, there is often high study effect variance and slow developing effect sizes. Improved biomarkers to help with dose selection I think would be quite impactful for reducing development costs.
Imaging based fibrosis detection scans I think are the future here. Just like imaging tests have increased diagnostic yields for ATTR amyloidosis, I think they would be broadly useful for fibrosis. There are already high-resolution CT (HRCT) scans that can do this and companies (e.g. Imvaria) that use these diagnostics to diagnose IPF and other lung diseases. There are already imaging based detection tests for fibrosis for the liver (e.g. Fibroscan).