This research service has written before about new kinds of small molecules (whether that be molecular glues, PROTACs, or covalent inhibitors/activators) as well as novel approaches to targeting already tractable drug targets. The most recent note focused on molecular glue degraders, which leverage cells’ own protein degradation systems while (mostly) staying within standard small molecule design rules. The note on Merck’s ~5.8bn acquisition of Terns Pharmaceuticals focused on the current landscape of tyrosine-kinase inhibitors used to treat chronic myeloid leukemia (CML), and why TERN-701 was a potentially exciting option despite a number of already approved therapies.
Triana Biomedicines straddles the molecular glue degrader and improved TKI worlds. Its lead-asset, TRI-611, is a CNS-penetrant molecular glue degrader in development for treating ALK positive non-small cell lung cancer. ALK positive NSCLC refers to cases where the ALK gene fuses with another gene (most often EML4), in turn leading to uncontrolled cell growth. These patients comprise a small amount of overall non-small cell lung cancer patients (4-5% of cases), or ~8800 US cases per year. It tends to show up in younger patients with no history of smoking.
Like CML, the prognosis for ALK+ NSCLC is really quite good, and survival rates have improved dramatically as newer generation TKIs have been developed. Lorlatinib, Pfizer’s CNS-penetrant third-generation TKI for NSCLC, didn’t reach a median PFS value after a follow up period of over 5 years; crizotinib, the first FDA approved TKI for this disease, had a median PFS of 9.1 months. Big pharma tends to like assets playing in spaces such as these: Merck bought Terns, GSK bought Nuvalent for ~10.8bn, BMS spent ~4.1bn on Turning Point Tx, and Roche gave Nurix Therapeutics 700mm in cash as part of the recent agreement to co-develop/commercialize NX-5948.
A principal limitation of earlier generation kinase inhibitors for ALK positive NSCLC specifically is the degree to which they were CNS-penetrant/able to control brain metastases. >20% of ALK positive NSCLC cases have CNS metastases at the time of diagnosis, and almost 60% will have CNS metastases after three years.1 Crizotinib demonstrated very poor CNS penetrance, and so wasn’t helpful here.2 Second-gen TKIs did a better job, but lorlatinib was even further improved, with intracranial complete response rates (% of patients whose brain metastases disappear) of 72% vs alectinib’s (the dominant second-gen TKI) 38%.
Unfortunately, lorlatinib has a pretty non-ideal side-effect profile, including elevated triglycerides, weight-gain, and CNS effects (that in some cases can be as severe as psychosis). In other words, there’s a pretty clear market for a drug with similar efficacy but more tolerable side-effects.
This side-effect profile stems from a challenge innate to active-site kinase inhibitors: a kinase active site always bears some similarity to other kinases’ active sites, and this leads to off-target effects. In the case of lorlatinib, it also inhibits tyrosine receptor kinase B (TRKB), which is thought to cause the psychiatric effects.
The second-challenge kinase-inhibitors run into is cancers are adaptable, and so will eventually start to progress. This treatment resistance can have a few different causes:
1) A mutation within the kinase-active site. This mutation prevents the TKI from being able to bind effectively to the ALK-fusion protein.
2) Leveraging of other pathways to drive cancer growth (perhaps the cancer can take advantage of a mutation somewhere along the MAPK pathway rather than relying on the ALK-fusion)
3) Histologic transformation – the cancer actually changes cancer types, in this case going from non-small cell lung cancer to small-cell lung cancer.
Given (1), the TKI that a patient starts on is unlikely to be used forever. A lot of next-generation TKI design is about coming up with a new small molecule that can still bind to the active-site in the presence of specific mutations (as an example, Nuvalent made sure its compound could still bind to the kinase even if patients had a G1202R solvent-front mutation, a common occurrence in patients on second-generation TKIs).
To date, ALK-fusion TKI design has focused on inhibition via the active-site. Like CML, however, there’s a clear appeal to other approaches that bind elsewhere: this gets around active-site mutations, avoids side-effects that stem from binding to off-target kinases, and enables combination usage with inhibitors that do bind via the active site. The potential for improved side-effects and combination therapies is especially helpful. The off-target effects of existing inhibitors limit how high dosing can go due to tolerability/safety concerns; minimizing these means one could take dosing, and thus efficacy, higher. The combination therapies piece is helpful for a similar reason. Today, when a patient has intolerable side-effects, one either has to lower the dose or switch inhibitors. Neither are ideal options, as the result is a patient isn’t on the most efficacious medication at the most efficacious dose. An inhibitor that binds elsewhere would mean the active-site TKI dose could be lowered and an adjunct medication could be added to boost efficacy.
Unlike Terns Pharmaceuticals in CML, Scorpion Tx (which Lilly acquired) in breast cancer, or Relay Tx also in breast cancer, Triana Biomedicines is trying to leverage protein degradation rather than allostery. Protein degradation doesn’t necessarily require that one bind somewhere other than the kinase domain: in the case of PROTACs one just needs some sort of binding site, whether functional or otherwise. However, in this case binding to the kinase isn’t all that helpful: efficacy could go higher because the ALK-fusion is degraded in its entirety rather than only inhibited, but there’s still the same worry about off-target kinase effects and maneuvering around active-site mutations. Consequently, TRI-611 is a molecular glue degrader that binds to a novel degron on the ALK-fusion protein, in turn leading to ALK-fusion degradation.
The hope with this therapeutic is twofold: it avoids off-target kinase effects (as Nuvalent’s neladalkib aims to do), and can be an effective option for those with increasingly complex active site mutations. That said, the most interesting part of TRI-611 is the novel degron it leverages. We’re quite early in understanding how protein-degradation works, and today are severely constrained in our knowledge of both E3 ligases (which facilitate ubiquitination of a target protein, thus tagging it for degradation) and degrons (a structural motif on the target protein that facilitates interactions said protein and the E3 ligase; molecular glue degraders aim to modify either the E3 ligase or target protein to enable recognition of a degron that’s already there). While there are over 600 E3 ligases, for therapeutic purposes we only try to leverage Cereblon or Von Hippel-Lindau (VHL). Our knowledge of degrons is even more limited; we don’t know how many there are, and most therapeutic candidates try to leverage the same G-loop degron that thalidomide does. This limited knowledge is especially unhelpful in the molecular glue degrader context, where rational discovery is already a significant challenge. TRI-611 uses cereblon as the E3 ligase, but makes use of a previously unknown degron:
“We next explored the degron motif on ALK responsible for recruitment by TRI-611–CRBN. The first described CRBN degrons for MGDs were defined β-hairpin structural motifs with a positional glycine within the loop (G-loop) that have since expanded to G-loop mimics….. a distinguishing feature of TRI-611 is the mode of action proceeding through a distinct turn-helix degron. While G-loops and G-loop mimics primarily rely on several key CRBN residues to promote neosubstrate binding, TRI-611 instead engages in a broad network of ligand–protein and protein–protein contacts that culminate in high ternary complex affinity and contribute to selectivity over commonly recruited neosubstrates.” - Source
Put differently, what the Triana team has done is (1) demonstrate an ability to discover novel degrons, and thus take a step towards legitimizing the claim that they can rationally design molecular glues, and (2) done so for a highly relevant target protein, where if the clinical trials data looks good the company becomes a straightforward buyout candidate.
Of course, the ALK-fusion protein isn’t exactly undruggable, and so a cynic might claim that Triana has spent time and resources on a less pressing disease in order to make itself more legible to big pharma. I think this misses the point a bit: figuring out rational molecular glue degrader design is a challenging, expensive, and uncertain endeavour. The additional hurdle of targeting an undruggable candidate results in an extra layer of existential risk that can mean the company’s accumulated body of expertise ends up going nowhere. Targeting ALK means you’ve minimized a portion of the science risk unrelated to designing molecular glues, and, if successful, can put the proceeds from that asset into chasing less conventionally druggable targets.
This makes CNS-penetrance much more important than in, say, CLL, where Nurix Therapeutics is developing a PROTAC. CNS-involvement only occurs in <2% of CLL patients, and so NX-5948’s ability to cross the blood-brain barrier is fascinating because it’s a large PROTAC, but doesn’t necessarily have implications for patients.
Instead, crizotinib was a great substrate for the P-Glycoprotein, quite an important protein to avoid when you’re trying to cross the blood-brain barrier!