The 🚩case of Rubius Therapeutics and what we can learn from it. 🩸
Also published in full on LinkedIn.
In the last two months, I’ve been posting case studies for discussion on my personal LinkedIn feed when I find something we can learn from. They have become quite popular, at least to the LI algorithm 😉 and some serious, thoughtful biotech researchers, entrepreneurs, docs, business people, and investors like the readers of this newsletter.
I normally post these cases directly on my feed. However, sometimes the lessons to be learned are just too numerous to fit into the constraints of a normal post, so this case warrants an article in The Biotech Leadership Lab.
While my policy in the posts is to forego calling out the names of the companies since many have and will continue to get themselves into similar situations (although I’m trying to remedy that somewhat), this company, Rubius, is dead and buried and an egregious example of all that can go wrong in business, science, and medicine, so let’s talk turkey.
I started tackling this case study as part of my research for an upcoming book, The Brilliant Biotech Boss. For which, I’m developing a blueprint for executive leadership at the sweet spot of business, science, and medicine. That is because finding your sweet spot in trailblazing efforts is one of the trickiest yet most impactful things we can do for successful biotech.
Unfortunately, Rubius is an example of missteps in all three areas. Could it have created a different outcome if even one of these aspects had been better executed? You decide!
To be honest, this company bothered me from the moment I first learned of it, well before I envisioned writing a book and well before it imploded. My initial concern was based on the scientific premise.
Being a "pseudo-engineer" as my true engineer colleagues like to call me and an old immunochemist who dabbled in RBC cell fusion long ago, I wondered what the efficiency of the surface protein distribution would look like in RBCs that have a specific architecture in part due to an inner band of protein just under their cell membrane acting like a skeleton supporting their unique shape and resiliency. Would it be a limitation for efficient cell-cell interaction? Did they know? Also, I was concerned that the approach was bucking biological evolution, which would make it that much more difficult. But I digress...
🧐 Be assured that there is something for everyone here, so buckle up and let's examine the many lessons: scientific, operational, medical, and financial, by first imagining ourselves in their shoes.
Here are the details:
You start with a very novel concept in 2013, initiated by two prominent scientists from MIT. The idea is that you can genetically engineer donor red blood cells to serve as biologic devices for antigen delivery. You produce genetically altered mature RBCs (that lose their nucleus upon maturity) to result in red blood cells with interactive molecules on their surface (that are usually left to other types of blood cells to express and regulate) - creating a “new class of cellular therapeutics” to fool the other blood cells involved in immunity into reacting more than they usually do, for example. Envision a blood transfusion with immunity superpowers. Far out, right? All these small modified RBCs (mRBC) would go to the spleen and also circulate, stimulating the white blood cells to do a better job, say, fighting cancer.
You are backed by a high-flying incubator run by a prominent VC firm that prides itself in developing pioneering ideas in search of the next unicorn. However, they have a penchant for thinking all the best unicorns will come from top-tier academics, and that if a brilliant academic says it is so, then it likely is, or at least they can get a sufficient number of fellow investors to believe in it (with a “little” hype). The company is funded to the tune of $100M.
The details of all they'd planned were far-reaching and replete with grand claims: a “new era of cellular therapy”, “cellular medicines”, and the like. An interview of the CEO in Genetic Engineering News is an interesting read, given the outcome. https://www.genengnews.com/gen-edge/a-ruby-in-the-rough-rubius-therapeutics-transforms-red-blood-cells-into-cellular-medicines/
Presented with a wealth of possibilities, you focus on autoimmunity and cancer, the remnants of their aggressive clinical plans still present on the company’s LinkedIn page:
https://www.linkedin.com/company/rubius-therapeutics/
You work to demonstrate proof of concept in mice and develop your production protocols for an off-the-shelf allogeneic product. It is a tall order to accomplish in a few year’s time, yet by 2018, the VCs "foist" you onto the public markets, achieving a valuation of $2B, the largest that year. However, you are still in mice and there remains much to nail down. The gaps between academic science and translational science are cavernous, particularly given that the platform and science buck up against normal biology (one of my primary early gripes). Nevertheless, kudos go to the people R&D and regulatory who appear to be tasked with working at the speed of light.
🩸Remarkably, by 2022, you are in three Phase I/II clinical trials. The first target is aimed at stimulating 4-1BB, a member of the TNFR superfamily, and the pro-inflammatory cytokine IL-15, established drivers of T cell and NK activation. Your RBCs express 4-1BB ligand (4-1BBL) and IL-15 on their surface by genetically engineering RBC progenitor cells and then maturing them into anucleated mRBCs. The mRBCs are meant to stimulate the white blood cells within the spleen and circulation, where they'll then activate and migrate to the tumor (refer to the diagrams taken from the company's March 2022 promotional article (advertisement) in Nature's Biopharma Dealmakers (Biopharma Deal) online publication. https://www.nature.com/articles/d43747-022-00074-w
What gives you the confidence to enter the clinic? The 4-1BB/IL-15 combination seems to reduce melanoma tumor growth in mice using mouse mRBCs; however, the published data from both in vitro testing and animal studies is highly academic, a “nice” way of saying that the differences in treatment vs. control vs. antibody alternatives were weak, even if statistically significant. As we’ll see, robustness of the mechanism and response matters greatly. It is a red flag if you can’t knock it out of the park with your best-case scenario (in vitro data) for proof of mechanism and the anti-tumor data isn’t compelling from a translational perspective even in tiny mouse tumors (yes, size matters here). 🚩🚩🩸Undaunted, you rationalize that your approach will be better because of a problem of liver toxicity with 4-1BB agonist antibody efforts.
‘Because of the restricted biodistribution of RCTs, we believe we can stimulate the immune system against cancers without causing toxicity due to extravascular target engagement,’ your CSO explains. It also means that your super RBCs must act within the spleen and bloodstream.
Your second candidate is 4-1BBL plus IL-12, a different pro-inflammatory cytokine and you again base it on mouse data showing anti-tumor activity.
🚩On the process development end, you’ve started clinical trials with a "liquid" formulation that is only stable for 50 days, vowing that future clinical trials will be with a frozen option that you are working on, also citing production yields and COGs as things you are working on. You are currently in 50L bioreactors as you enter the clinic for multiple trials. You estimate that each batch of mRBCs will eventually be able to treat 100 to potentially 1,000 patients but for now you feel it is working well enough to get it into humans. It appears you are getting desperate to demonstrate a human response to validate your concept. The problem is that humans will be the most complex lab animal, and these are patients with real, advanced disease.
The staff is well over 200 by now, and you have 60 people in manufacturing alone. R&D continues at a feverish pace, to publish preclinical "proof of concept" (references can be found in the GenEng piece). Your stock has steadily declined from its high (refer to stock chart) and you’ll need to raise more cash.
You run your first candidate as both a monotherapy and in combination with Keytruda. Patients treated have different types of solid tumors, from sarcoma to various carcinomas.
Here are some of my takeaways; however, there is so much here to weigh in on (almost TNTC) that I am hoping, dear readers, that you will help us out and add your expert insights to the learning experience in the comments.
From my perspective:
🚩 The concept suffered from naivety, creating an uphill battle against native immune mechanisms. Simply put, RBCs and white blood cells evolved the way they did for a reason. The platform hinged on white blood cells being sufficiently stimulated by the mRBCs, entering the tissue, and maintaining this stimulation for a long enough time to affect a difference. That is both a tall and indirect order. Based on the marginal preclinical data, did they really believe they could achieve an informative readout with the mRBCs alone? Should they have backed up and worked harder on better preclinical work or could they have found that the approach might be a no-go? I believe so. Fast-fail wasn’t in their vocabulary.
🚩 The approach, rationale, and ongoing justification for the platform was overly academic and remained that way. They were either unduly rushed to the clinic, or unaware of the serious gaps that needed to be filled.
🚩 Did them going public so early seal their fate? Certainly the second public offering indicates that they were desperate to show clinical proof of concept. The design of their “PhI/II” trials seemed more like a fishing expedition than a deliberate test of something they had confidence in.
🚩 Academic science is not the same as industrial science. The fact that they chose to publish the promotional piece in a Nature affiliate publication is telling to me. Good academic science does not translate to a good product without a heck of lot more than what they had.
🚩 The approach’s objective was to augment an anti-tumor response, yet there are other more controllable, well-documented ways to achieve that. What more would this platform be able to provide over alternatives?
🚩 The production process would be difficult and costly for a host of reasons. Kudos to the team that got it as far as they did, but the question is for what purpose?
🚩 Entry into the clinic with a premature process is a red flag. They claim to have had enough control over reproducibility to deliver the identical product to patients. However, they were working with multiple candidates at the same time, batches with a relatively short shelf-life. We have to remember that while the FDA will stop you from doing something potentially harmful with out good supporting data, they will not stop you from doing something stupid, that was up to the C-suite. Think of the decades of cell therapy applications that have entered the clinic never to emerge. It is why I’m such a curmudgeon about robustness in process and mechanism.
👉 I’ll leave my comments at that in hopes that others will offer their wisdom. The goal, as always with these case studies, is to learn from them.
What would you have done differently? Can you envision a different outcome?