Excuse me for going on a rant, but I am concerned about the future of peptide-based therapeutics for treating diabetes and obesity. Perhaps I should add “my future”.
September 2023: I am 78 and a type 2 diabetic retiree, who is treated with basal insulin, metformin, and semaglutide (Ozempic). At least that was the idea. The trouble is I, and many other Ozempic patients, can no longer get a regular supply of this drug. I am tired of pharmacists telling me to try elsewhere. At my age, I would like to spend my time more usefully than visiting out-of-stock dispensaries. And apart from that, is it really my job to be the last link in the pharmaceutical industries’ supply chain?
What worries me more is that I feel this problem is about to get much worse as NovoNordisk’s semaglutide (Ozempic, Wegovy, Rybelsus) and Lilly’s tirzepatide (Mounjaro, Zepbound) and, to a lesser extent, NovoNordisk’s liraglutide (Saxenda, Victoza) free-fall into an expanding obesity and life-style market. No question that obesity is a disease in need of treatment. With similarities to alcohol or drug addiction – if you restrict access to the causative agent (food), the disease can be controlled. However, the typical two-pronged administrative strategy for tackling addiction, namely regulating the price (and collecting some tax revenue) and installing legal barriers, will not work here. Be realistic, no government agency is planning on placing restrictive taxation on food or legislating food intake. Although ideas for a sugar tax are out there, it is hardly likely to have any real effect. Unlike in chemistry, in the food labelling business, sugar is not necessarily sugar. How often do we read labels with “Sugars 0%, Dextrose 10%” or “sugar-free” raisin-rich Trail Mix? Apart from this aside, sugars are only part of the problem. The reality is that people will eat what they can afford and want, and many of them become obese. What is now different is that the affluent can tackle their eating addiction with approved drugs, of which the leading candidates are peptide-based, incretin agonists based on glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) as exemplified by Ozempic, Wegovy and Rybelsus (all semaglutide), Mounjaro and Zepbound (both tirzepatide). This is happening at the expense of the diabetic population.
Well, is this surprising? No, cash oils and influences supply chains. There are shortages out there for many other drugs too, most of the world lacks adequate medication. So, am I just grousing about me not getting my medication because semaglutide is following the money (a) out of the country (parallel trading) and/or (b) into obesity and want-to-stay-slim lifestyles rather than into diabetes? I am not alone, Belgium is considering temporarily banning the use of Ozempic for treatment of obesity (1), Germany is considering preventing its export, many other European countries have already imposed bans (2).
It would certainly help if the respective pharmaceutical manufacturers would accept their ethical responsibility for the drugs that they release into a particular domestic market by exerting greater control over their wholesale distributors.
But do not get me wrong. These are exciting times for incretin-based drugs and peptide therapeutics in general. As much as I am frustrated with the supply chain, I am a great admirer of both Lilly and Novo. These companies have done and are doing truly fantastic work on incretin agonists. And both are aware that, now the genie is out the bottle, there are some unsettling issues to deal with.
I have additional concerns. I was in the peptide business for over 30 years and have been peripherally involved with manufacturing GLP-1 and other incretins almost from the start, including some of the first batches of these peptides for the pharmaceutical industry and European endocrinologists in the late 1980s. I worry now about the massive doses, the associated costs and environmental impact of this type of drug, especially future incretin-based drugs that may be manufactured using standard solid phase peptide (SPPS) chemistry. The doses of these drugs are, for peptides of ~30 amino acids, exceptionally high. The maximum dose for Ozempic is 2.0 mg/week (higher doses are in PII trials), for Wegovy it is 2.4 mg, for Mounjaro and Zepbound it is 15 mg. Compare these with the weekly doses of two earlier, daily, short-acting GLP-1 receptor agonists, exenatide (7 x 20 µg) and lixisenatide (7 x 10 µg). We are talking about a factor >102. And what happens to the dose when these drugs become available in oral formulations? Rybelsus (oral semaglutide) can be dosed at 14 mg/day (~100 mg/week); 50 mg/day was the dose in recent Phase 3 trials. To my best knowledge, the bioavailability is less than 1.0%. No trials seem planned, but one can imagine what the dose regimen for an oral tirzepatide might look like.
My point, however, is not the dose per se (after all metformin can be dosed daily at 500 mg or more), but that these high doses mandate significantly larger scale manufacturing than has been the previous norm for peptides. Go back 35 years, champagne flowed the day the first SPPS peptide at a kg-scale was achieved. Go back 15 years, only 3 peptide therapeutics were synthesized chemically at >100 kg and it was almost universally assumed that recombinant processes cut in for economic and logistic reasons between 20 kg and 50 kg. Today, we are talking about SPPS at the multi-metric ton scale. It is primarily the obesity market world-wide (with about 1 billion potential patients) that is driving this growth and advances in SPPS. A recent report (3) predicts that there will be 30 million GLP-1 users in the USA by 2030, splitting fairly evenly between diabetic and obese patients. Back-calculating from Lilly’s expected sales of Mounjaro and Zepbound, the current pricing and their published yield (8), we are already looking at manufacturing at a 1 kMol-scale, which – with an estimated overall yield of ~30% – translates into ≥10 tons of starting materials. This could potentially increase several times more now the drug is approved for treating obesity.
I do not believe any single company currently has the capacity to fulfill these manufacturing requirements; it would almost certainly be gobbling up much of the world’s available SPPS space. In the next 5 to 10 years, demand for large SPPS reactor space, as well as downstream capacity and starting materials, could well outstrip availability. That is not just a problem for the existing manufacturers; they must – as they are doing right now – live with the shortfalls while still enjoying the profits. But for those biotech and pharmaceutical companies already, or in the future, developing new, and maybe better or more economical peptide-based obesity drugs, it may become existential. Right now, innovators tend to think manufacturing capacity is their CMDO’s problem, but if there is no capacity, then it becomes their problem too.
“Don’t change a winning horse”, has long been the pharmaceutical industry’s philosophy towards new technologies. The problem here is that when the horse gets old and is retired, although we may not be short of new blood, we could well be short of stable stalls, either because they are occupied or because they are held in reserve for other clients. A medium-term cartel-like weaponizing of available SPPS capacity could loom on the horizon. If we do not rethink our manufacturing strategies, we are going to be in trouble.
And maybe we are already in trouble. Ozempic and Mounjaro are rotating in and out of the FDA Drug Shortage List allowing compounding pharmacies to legally enter the supply chain (4). However, it is uncertain where these pharmacies are sourcing their active ingredients and whether they are strictly observing Section 503A of the FD&C Act governing compounding (5).
There are currently no approved generic equivalents of semaglutide or tirzepatide, so where is this material coming from? Do pens even contain the correct ingredients? Almost beyond belief, insulin has turned up packaged as Ozempic in Austria (6). Even when the pen contains semaglutide, is the drug substance truly equivalent in (salt) form and purity (7)? How much of the supply is coming from manufacturers who do not qualify as “reputable sources“? This scenario is all too reminiscent of the demand for Peptide T for treating HIV back in the 80’s. However, that was to an extent excusable because those Peptide T users had no other survival strategy. Obese people do.
Unsurprisingly, peptidic incretin-based active ingredients manufactured by SPPS are expensive when compared to small molecule drugs. This is reflected in the cost-to-patient. The list prices of tirzepatide-based (similar to recombinant semaglutide-based) drugs are – without insurance – close to $1000 per month in the USA. This begs the question as to whether health service organizations can continue to carry these costs as the patient base expands; about 1 in 10 US citizens are diabetic, over 1 in 3 are obese, and – in terms of material supply – who knows what the life-style market looks like?
And consider the environmental impact. In 2019 Pawlas et al. (9) calculated the carbon intensity (CI) of a published SPPS manufacturing process for exenatide, an incretin mimetic of similar sequence length to tirzepatide or semaglutide at a 1-kg scale to be approximately 117 tons of carbon dioxide per kg peptide. That is not a typo. Over 100 tons carbon dioxide per kg peptide product with 90% of this arising from solvent production and disposal. We are talking about manufacturing processes, which when scaled-up, are vying with the annual CO2 output of a mid-sized city. Given that the SPPS reference process cited by Pawlas for exenatide is probably obsolete and solvents were not recycled, CI values for current processes are almost certainly not as high, but it does give us some idea how environmentally detrimental SPPS can be. And this only reflects the carbon footprint, it does not consider general wastage or the hazardous nature of some of the chemicals and solvents used.
Is there a way out?
If we are going to treat obesity with peptide therapeutics such as GLP-1, GIP, or glucagon or amylin analogs, either as standalone’s or as dual/triple agonists, we cannot easily get away from molecules of ~30 amino acids length or more. The known binding sites of these peptides cannot be easily addressed by small molecules because of their spatial conformation. Although a class of small molecule drugs exemplified by danuglipron (10) has been shown to exert GLP-1 agonistic action by independent, unique interactions with the receptors, it remains to be seen whether the latter will become approved drugs, let alone whether any small molecule will mimic the current dual/triple agonists currently in development.
If we develop incretin peptides that contain non-canonical amino acids (which includes semaglutide and tirzepatide), we typically cannot use recombinant approaches, or we at least need to be able to chemically modify a recombinant backbone, such is the case for liraglutide (introduction of a fatty acid side-chain) and semaglutide (introduction of aminoisobutyric acid and a fatty acid side-chain). However, many new obesity drugs may not be amenable to such modification protocols, which require a single lysine in the amino acid sequence to enable mono-acylation. It is also very unlikely that any innovator without in-house fermentation experience and capacity would choose a recombinant manufacturing route without first going down the well-trod generic SPPS path.
Is chemical synthesis or recombinant manufacture the most economical process for incretins? The cost structures and starting materials are radically different. Medicinal chemistry typically uses chemistry for the design of novel entities, but – as production scales up – common sense dictates that although chemical synthesis must be more economical for low kg-quantities, a recombinant approach (where possible) is the method of choice for metric tons. The cross-over point (if any) depends on the structure of the peptide, the scale that needs to be achieved, as well as the historical expertise and individual preferences of the manufacturer.
So right now, no escape, but can we start to mediate some of the excesses?
One way out of the reactor capacity issue is to assemble peptides from fragments. Lilly is already using this strategy in a (semi-)continuous process for tirzepatide (8). It has many theoretical advantages. Shorter fragments that are manufactured in smaller reactors typically use proportionally less solvents than full-length stepwise SPPS, and – from a logistics perspective – spread batch failure risks and can be stockpiled until needed for the final assembly. The final assembly is performed in solution, not requiring complex, expensive, custom-built SPPS reactors.
In contrast to the situation for SPPS, there is no foreseeable shortage of standard solution (LPPS) phase reactors. It is not clear whether fragment approaches which inherently involve more steps are more attractive economically than stepwise SPPS procedures; much depends on scale, ligation efficiency and logistics.
Depending on the peptide fragments needed, it may be possible to use solution phase (LPPS) or tag-assisted LPPS technologies (molecular hiving, Agiphase, GAP, etc.,) to manufacture the fragments. These do not require special reactors and would remove pressure on SPPS capacity. In addition, the reaction intermediates of the latter – when esters – may be useful substrates for some ligation technologies, thereby reducing the number of process steps.
The ostensible draw-back of tag-assisted LPPS and several other innovative technologies, is that these are proprietary, and their licensing fees (and accessibility) need to be factored into the cost equation. The maths needs to be done on a product-specific basis; savings can be significant. The reticence of the industry to embrace these technologies is resulting in their loss to larger corporations that can then monopolize the technology. This comes at a cost to the peptide community at large.
Since some of these technologies are unique, this trend may eventually shift the balance of who-can-do-what in the CMDO and peptide therapeutic market.
In addition to standard solution procedures, fragments from SPPS, LPPS or recombinant procedures can be ligated by enzymes. The use of enzymes expands the number of potential fragment structures because chemo-enzymatic peptide synthesis (CEPS) has different flanking amino acid requirements to other ligation methods. Where multiple fragments are involved, the selectivity of enzymes opens the door for one-pot procedures. They also permit the incorporation of acylated and non-canonical amino acids. Since enzymes use aqueous solvents, the ligations could theoretically be performed in single-use plastic reactors.
With reactor capacity secured, the focus should turn to reducing costs. The choice of manufacturing process and conditions contribute to cost reduction, but ultimately it is the final yield of the process (overall yield and yield per batch) and the CMO’s manufacturing logistics which govern the cost of the drug substance; this may mean developing a new or optimizing an existing manufacturing process.
In the context of incretins, enzymes are useful process aids for improving yields in fragment processes. Timo Nuijens first described a CEPS-based process for the manufacture of semaglutide at the Peptide Therapeutics Foundation Symposium in 2018; a patent was recently submitted (11). The 2019 paper of Jan Pawlas (9) demonstrated that CEPS – at a scale suitable for commercial production – increased the overall yield of exenatide by ~100% and significantly reduced the estimated cost of the final product (to less than $60/gram API), while shrinking both the CI and Complete E-Factor (cEF) by over 80%. cEF and the now more commonly used metric Process Mass Intensity (PMI) are effectively the same at this order of magnitude.
Of course, it remains to be seen how the introduction of a more economical manufacturing process would be reflected in the retail prices of these drugs. Nor should we expect the retail prices of the respective generic versions, when they appear, to drop into the easily affordable range. These are not trivial molecules to manufacture, especially when SPPS is used for molecules where the innovator used a recombinant process (semaglutide); achieving the innovator’s impurity profile is exceptionally challenging. Short-term price relief could come immediately by lowering the retail cost. Given that the over-the-counter prices of current incretin drugs are 3 times higher in the USA than in some other countries, there is obviously room for downward movement.
The problem in the USA is not just opportunistic exploitation of the market, but the surrounding health insurance system structure.
Reducing the environmental impact implies (a) specifically reducing the carbon footprint (CI), (b) reducing waste per se (PMI), and (c) eliminating hazardous chemicals (mainly hazardous solvents).
Item (a) and item (b) are tightly linked to economic considerations. Increasing potency (new peptides) and yield (new processes) will accordingly decrease carbon footprint and overall waste, as will recycling of solvents and continuous (versus batch) processes. Eliminating hazardous solvents, item (c), is an ongoing goal within the industry, especially as regional restrictions on solvents such as dichloromethane and dimethyl formamide become tighter. The challenge is not so much the choice of alternative solvents, but the associated cost.
We should also not forget that, in the context of SPPS processes, the starting materials themselves, Fmoc-amino acids, are highly complex, high waste (on average ~50% weight is the protecting group), high carbon intensity commodities.
For long peptides containing unusual amino acids, there is much to advocate using hybrid approaches, taking recombinant peptide fragments (produced with no or low toxic solvent use) and then fusing these to short non-canonical sequences using chemical or enzymatic procedures. Whatever, if we are heading in the direction of 100 tons or more of longer peptides, and we want to help our environment, we need to re-think our manufacturing strategies. As much as possible, think aqueous.
Why is this not already happening? It would seem so obvious. The answer is sitting in the saddle of that “winning horse”. Most innovators convey a willingness to test alternative, more green manufacturing procedures, but understandably they take the proven, no-thought-needed stepwise SPPS path forward first, with the promise to look at “bigger and better” alternatives once the drug is approved and generating income. Welcome to greenwashing. When the time comes, “bigger” and “better” become uncoupled. Alternative approaches require sequence-specific design, time-consuming experimental assessment, and risk. And there are already next generation candidates to focus on. If the patient can pay, why bother to change? The benefits of changing would have to be multi-10-million massive to alter the mindset of CMC and regulatory departments, and management in general. It is a combination of reluctance to presently invest extra work and money to develop and register a new process – even when the long-term advantages are palpable and compelling – and failure to recognize that an unbiased assessment of alternative approaches and/or of an alternative CDMO is worth performing before succumbing to the omnipresent NIH syndrome (NIH = not invented here). Unprecedented demand demands unprecedented approaches.
Can we change this? I think we might be able to. It is a question of finding the right carrot. Currently, drugs are approved based on their efficacy. The manufacturing process, if it has no hazardous implications for the patient, is irrelevant.
We cannot impose restrictive cost caps on new drugs because this would kill incentive to develop new modalities. But we can reward those companies that can demonstrate environmental responsibility by reducing the carbon footprint of the manufacturing process either in the approval process (e.g., accepting NCEs with similar efficacy to approved medications but with better carbon footprints) or by lengthening the wait period for generic equivalents if the manufacturing process does not have a better carbon footprint than the innovator’s product. This latter would also ensure that pressure remains on the innovator to develop alternative processes in parallel or continue improving their existing process. This, in turn, would require a more pragmatic approach to process changes and re-registration, but this is long overdue anyway.
Of course, calculating the carbon footprint or PMI of a process is not a simple task. The CI values for many solvents have not yet been calculated, and for protected amino acids, welcome to the jungle. Somewhere up that road it becomes very foggy. But we can approach this in a rational manner by making assumptions for standard starting materials and process aids and applying these across the industry to create a more level playing field.
Wishful thinking?
Do not expect changes soon, the industry resists change. But maybe we can start a serious and consequential dialog about the future of large-scale peptide manufacturing.
At the time of writing (November 2023), my Ozempic prescription has been pending for over 2 months. My fasting blood sugar levels are getting harder to control and HbA1c jumped a full one point. My diabetologist says he would like to give me Victoza (liraglutide) instead of the absent Ozempic, but guess what, that is not available either. So, take more insulin. I do not want to do that. Maybe I need to walk off those extra calories looking for an Ozempic-carrying pharmacy at the end of a rainbow, or just eat less.
“A further twist to the semaglutide compounding pharmacy story: The amino acid backbone of semaglutide is manufactured recombinantly. Compounding pharmacies are prohibited from formulating and selling biologics (syn., recombinant products), so in theory one might think that they cannot formulate and sell semaglutide. However, (a) the FDA does not consider peptides with ≤40 amino acids to be biologics (40 or less amino acids is a peptide, 41 or more is a protein), (b) it is highly unlikely that any semaglutide currently being compounded is manufactured recombinantly, (c) there is considerable likelihood that the impurity profile of the compounded synthetic semaglutide does not match that of the innovator’s product, and (d) do compounding companies care (enough to root out their own black sheep)?
Concerning the FDA’s “≤40> rule”, organizations have been trying to draw a definitive line between peptides and proteins for over 100 years. It is an ineffective exercise because there is no demarcation point. This has been obvious ever since insulin was sequenced in 1951 and it was proven for the first time that protein backbones are composed of contiguous amino acid residues in the same manner as peptides. However, an artificial border does serve two purposes. For many years it has enabled market analysts to follow the rise and fall of the peptide therapeutic sector without disproportionate interference of animal-sourced and recombinant insulins; sequence lengths of synthetic and recombinant peptides now overlap so seriously that this categorization strategy has become highly questionable. The more important and original purpose was to allow regulatory bodies to provide definitive guidance to manufacturers on the separate regulations for the manufacture and characterization of less complex (typically chemically synthesized) and more complex (typically recombinant) peptides. Recombinant manufacture and chemical synthesis are such different technologies that distinction should not be subject to a ≤40> line in the sand. Semaglutide seems to have a foot firmly rooted on each side of that line. ”
References and notes
Most of the information regarding drug dosage regimens, fraudulent products and compound pharmacy formulation regulations can be easily found on the internet, but here are a few pointers and notes.
- Belgium’s Health Minister Sparks Debate Over Ozempic Prescription Limitations. Belgium’s Ozempic Controversy: Testing the Bounds of Therapeutic Freedom (bnn.network): Belgium bans use of Ozempic for weight loss until summer. Belgium bans use of Ozempic for weight loss until summer | Reuters.
- German watchdog considers Ozempic export ban amid Europe shortages. German watchdog considers Ozempic export ban amid Europe shortages – WNEWS247.
- The increase in appetite for obesity drugs. The increase in appetite for obesity drugs | J.P. Morgan Research (jpmorgan.com). Note that 90% type 2 diabetics are overweight or obese, so making the split between diabetic and obesity markets has its challenges. “GLP-1” in this report refers to all approved GLP-1 analogs and incretin-based drugs.
- Drug Compounding and Drug Shortages. Drug Compounding and Drug Shortages | FDA.
- The Potential Risks Associated with Compounded Semaglutide: What Medical Professionals Should Know. Compounded Semaglutide — New Drug Loft & VLS Pharmacy.
- Fake Ozempic pens found to contain insulin concerning health officials. Fake Ozempic pens found to contain insulin concerning health officials (abcactionnews.com).
- There’s No Such Thing as Generic Ozempic or Wegovy. Generic Ozempic and Wegovy Don’t Exist. The FDA Is Right to Warn Patients. – Bloomberg.
There are four original citations. These are:
8. Michael O. Frederick et al. (2021) OPR&D 25, 1628 – 1636. Kilogram-Scale GMP Manufacture of Tirzepatide Using a Hybrid SPPS/LPPS Approach with Continuous Manufacturing.
9. Jan Pawlas et al. (2019) Green Chemistry, 21, 6451 – 6467. Sustainable, cost-efficient manufacturing of therapeutic peptides using chemo-enzymatic peptide synthesis (CEPS).
10. David Griffith et al. (2022) Med. Chem., 65, 8208–8226. A Small-Molecule Oral Agonist of the Human Glucagon-like Peptide-1 Receptor.
11. Timo Nuijens et al. (2022) WO 2022/171667 A1. Subtilisin variants and their use.
*Finally, the title of this essay is a play on words of the George Gershwin song, “Nice work if you can get it”. Fittingly, in the context of semaglutide, the last line of the song reads “And if you get it, won’t you tell me how?”
