One of the biopharma industry’s marvels of the 2020s has been the enormous medical and financial success of the GLP-1 class of drugs. But the path to commercialization of these therapies was far more complex than most understand, and some of the earliest history has never been presented.
Beginning in 1988, I was part of what I believe was the earliest commercial effort to develop GLP-1 as a metabolic therapy. Despite extremely promising results, Pfizer, the major funder of the work in alliance with California Biotechnology, abandoned the effort around 1991 after mistakenly concluding that the GLP-1 therapeutic approach was not worth continuing. To be clear, I’m not writing this to claim credit for discovery or development of GLP-1s or anything like that. Rather, I think the story offers valuable lessons on drug development for pharmaceutical companies, researchers, and the general public alike.
In 1987, John Baxter — a professor of medicine at UCSF, leader in molecular endocrinology, and founder of a biotech firm named California Biotechnology — asked me to work with him to create a start-up focused on metabolic disease, and I enthusiastically agreed. I enlisted two Harvard colleagues, Ron Kahn and Bruce Spiegelman, and together we outlined an approach that included a search for novel insulin analogues and insulin sensitizing agents, identifying the protein responsible for obesity in genetically obese mice (six years before identification of the ob gene), and exploration of gut factors regulating metabolism, such as the “incretins” that enhanced insulin secretion.
The company was formed as a “wholly owned subsidiary” of CalBio based in Mountain View, California. My colleagues and I held founders’ shares and chaired its scientific advisory board, while continuing our primary employment and research at Harvard. This structure allowed us to start work quickly, as Calbio was already operating, but it may also have contributed to the disappointing denouement.
Soon after, CalBio shopped our portfolio of projects to several large pharmas, and Pfizer was the first to respond. They sought shared rights to our entire portfolio in exchange for $30 million over five years; they would lead and conduct the clinical studies. This seemed like a great deal at the time.
The research on gut peptides was initially our least focused effort, but that changed after I attended a lecture by a fellow from the lab of Joel Habener of Mass General and HMS, who presented the first evidence that a gut peptide they named GLP-1 (or insulinotropin) strongly potentiated insulin release from rat islet cells. After the talk, I told Joel of the new company, and he expressed interest in following up. Within a few weeks, Calbio (and MetaBio, our subsidiary) had licensed the worldwide rights to GLP-1.
We made progress in several therapeutic areas, but GLP-1 was the most impressive, with work proceeding along two streams. There were clinical studies in healthy subjects and those with type 2 diabetes, carried out in Pfizer’s Groton, Connecticut, facility, and molecular studies performed in Mountain View, California. The clinical studies rapidly demonstrated glucose dependent enhancement of insulin secretion in healthy subjects, and soon thereafter showed substantial lowering of hyperglycemia after several days of infusion in type 2 diabetics. The studies also demonstrated slowing of gastric emptying and reduced hunger, suggesting a broader biology for GLP-1.
In parallel, the MetaBio team determined that GLP-1 had an extremely short half-life of only several minutes and set about to identify the responsible protease. They appeared on the brink of identifying what we now know to be DPP4, inhibition of which led to the clinically successful gliptin class of diabetes drugs. None of our results were ever published, so apart from this account, they’re only known to surviving participants, including me.
Through regular trips to Mountain View and Groton (this was before the Zoom era), I kept up with progress. We were all optimistic that we were on a path to impactful new drugs, either through more stable GLP-1 analogues or by raising GLP-1 levels via inhibiting the protease responsible for its degradation.
Of course, we were aware how little was known about the physiology and pathophysiology of the GLP-1 pathway, and that ultimate success was hardly assured. But we were confident in our ability to advance the work and could hardly imagine that Pfizer, our partner and primary funder, would soon abandon it.
But that is exactly what they did. Despite our emerging results, the Pfizer executives in charge of research and external alliances told us the company did not want to develop a new diabetes therapy that required injection, a space held exclusively by insulin since 1922. They gave us a year to find a way to deliver GLP-1 via transnasal, transcutaneous, or oral administration. Effective delivery by any of these approaches would have been great, but we knew success was unlikely in the year they gave us. Our effort was predictably unsuccessful, and after four years, Pfizer terminated our agreement as permitted under the alliance contract.
The MetaBio founders tried hard to convince Calbio to continue the extremely promising work on GLP-1 and other areas, which required new funding or spinout of a metabolic company. Importantly, at that time Calbio was at a pivotal point with their lead asset, a peptide named auriculin. Related to atrial natriuretic peptide, it was ultimately marketed for heart failure as Natrecor, after the company, by then named SCIOS, was bought in 2004 by Johnson & Johnson for $2.4 billion. Perhaps given their intense focus on their lead asset, the effort to find new funding was unsuccessful. MetaBio was shuttered, the “founders’ interest” was bought out — providing a modest balm to our despair — and the Habener patent was reverted to MGH and HHMI. Game over.
The founders continued our academic careers, but I kept my eye on the GLP-1 field, which advanced slowly for at least a decade before its explosive rise.
Since then, three major paths to exploiting GLP-1 related therapeutics emerged. Exendin-4, a peptide isolated from Gila Monster venom in 1990, was shown to stimulate insulin secretion and be an analogue of GLP-1 with a longer half-life. After years of frustrating efforts by its discoverer, John Eng, to find a commercial partner, it was eventually marketed by AstraZeneca in 2005 as Exenatide (Byetta) for the treatment of type 2 diabetes by twice daily injections, and then Bydureon with once weekly injections. Many obese diabetic patients lost weight, suggesting (but not proving) the immense obesity market to come.
The enzyme DPP4 was initially discovered in 1966 and 30 years later was found to be the protease degrading GLP-1 and GIP. The first agent of the class referred to as “gliptins” — sitagliptin (Januvia) — was approved by the FDA in 2006 and marketed by Merck. At least three other gliptins are now on the market.
What about GLP-1 analogues? The CalBio patent was returned to MGH and then licensed to Novo Nordisk, and my recent paper describes the halting but eventually highly successful process to develop a stable analogue for daily injection, resulting in approval of liraglutide (Victoza) by the FDA in 2010 for treatment of diabetes, and Saxenda in 2014 for treatment of obesity followed by additional highly effective analogues. Today, Novo Nordisk and Eli Lilly share annual global worldwide GLP-1–related revenues of more than $50 billion, expected to grow by several fold in the coming years, with many new entrants to the field.
But despite our early promising data, it’s far from certain that Pfizer and CalBio would have brought GLP-1s to market much earlier than actually happened. The path to develop, approve, and market these drugs was difficult for many reasons, including the physiology and pathophysiology of GLP-1 being largely unknown (there are still many unknowns), and a challenging path to creating stable analogues.
And yet, it’s hard for me to understand, for either scientific or commercial reasons, Pfizer’s decision in 1990 to walk away from both efforts. We were studying a newly discovered hormone that acted by a unique mechanism to improve diabetic pathophysiology and reduce blood sugar. (The anti-obesity efficacy was only a hint at that time.) Why would a pharma giant like Pfizer with a big metabolic research program boldly invest in our program, and then — in the face of exciting data — abandon it?
Biotech and pharma companies make decisions to advance or kill programs every day, as they seek to balance their portfolios, improve risk/reward profiles, manage costs, and attempt with variable success to predict the future. Given the many unknowns and uncertainties, we can be sure that many impactful mistakes are being made today.
Why share this story now? Much will be written about this class of drugs, and I’d like to ensure that its early history isn’t lost. In addition, my experience highlights how drug discovery and development can fail. Success requires a mixture of scientific and medical insight, internal champions, capital resources, excellent judgment by decisionmakers, and a substantial measure of good fortune. Even the biggest future blockbusters can be prematurely abandoned when one or more of these ingredients is lacking, as occurred in the CalBio/Pfizer effort.
Drug discovery and development are indeed a very tough business, as all involved in the industry know. But, for both (the former) Calbio, and the present Pfizer, the magnitude of this mistaken judgment was truly enormous.
Thirty-four years after their fateful MetaBio decision, Pfizer will soon be seeking approval for an oral GLP-1 agonist — danuglipron — a new entrant to a crowded obesity therapeutic landscape. If approved, it will certainly be a case of profoundly delayed gratification.
Jeffrey Flier is a physician-scientist in metabolic diseases, Harvard University distinguished service professor, and the former dean of Harvard Medical School.