3 ms·

Time for a history lesson. From 1922 until 1981, the only insulins available commercially were extracted from bovine and porcine pancreases -- a process which

Time for a history lesson. From 1922 until 1981, the only insulins available commercially were extracted from bovine and porcine pancreases -- a process which was both expensive and dangerous, since if the product was insufficiently purified it could cause life-threatening immune reactions or carry diseases from the animals used. (This is why insulin-using diabetics are to this day not allowed to donate blood in most of the world.) It took sixty years between discovering insulin and figuring out how to produce it biosynthetically (all modern insulin is now produced by genetically engineered yeast or E. coli). This yielded so-called "regular" insulin, which -- at the concentrations necessary to make it feasible for injection -- has the unfortunate property of hexamerizing. As a result, it has is peak activity about 3 hours after injection, and ends its activity around 6 hours after injection -- compared with endogenous insulin, which acts within 5-10 minutes, but is excreted 24 hours/day. In 1996, we finally had "rapid" insulin -- Insulin Lispro, which adjusts a couple amino acids in order to prevent hexamerization. This makes it act roughly twice as fast as regular insulin, roughly matching the time taken for a meal to be digested and to enter the bloodstream -- thus reducing the postprandial "peak" and cutting down on the chronic hyperglycaemia related complications of diabetes. Remember how I said that regular insulin stops working after around 6 hours? If you want to get 8 hours of sleep, that's not so good. In the 1940s and 50s some modified insulins were released which "slow down" the insulin so that it lasts for longer; but it wasn't until 2003 that the first "24 hour" insulin was released. So why is insulin still so expensive, 90 years after it was first introduced? Because the insulin we're using now is a heck of a lot better than the insulin 90 years ago. You might as well ask why a Tesla is so expensive when the Model T was launched over a century ago.

dekhn 12y ago
That's not the only reason it's that expensive. Pharma costs are artificially created and amortized globally.
- cperciva 12y ago
  The costs of bringing new drugs to market are not simply artificial. Quite aside from the costs of developing the drugs in the first place (tweaking a couple amino acids is cheap; knowing which amino acids to tweak is a bit harder) and the costs of production (significant in the case of insulin, considering the lengthy process for biosynthesis and then purification), demonstrating that a new drug is both effective and safe costs billions of dollars. Novel insulins are harder than most drugs too, since insulin is very closely related to some oncogenic hormones. A lot of novel insulins have died in trials when animals were found to have increased rates of cancer.
  - dekhn 12y ago
    Sorry, I meant: pharma drug prices, not development costs. Prices are artifically created, rather than being purely based on supply and demand. I know plenty about pharma costs, having worked for one and studied the area of biophysics including protein engineering extensively. My point is that the prices of drugs are only weakly correlated with the costs of bringing a drug to market. More importantly (and kind of ignored by most) is that pharma R&D costs are shared between many drugs. For example, Genentech has a huge R&D division and there are people who work on general projects that benefit all their biologics. Also, there are ridiculous inefficiencies built into the process of bringing drugs to market, many of which are designed to avoid large-scale adverse outcomes, or to make it possible to track where the adverse outcome root cause is.
    - lotsofmangos 12y ago
      Perceived value added and range pricing is one of those invasive diseases that have managed to take root across such a wide range of human culture that it will take decades to find a cure, especially since it is doubtful that industry is working particularly hard on the problem.
    - cperciva 12y ago
      We're talking market forces here. Yes, the drug companies will try to squeeze as many dollars out of the market as they can; the counterbalancing force is that if they get too greedy, someone else will enter the market and undercut them. The fact that entering the market is so bloody expensive is why drug companies can set their prices so high without provoking a new competitor into existence.
    - raverbashing 12y ago
      " My point is that the prices of drugs are only weakly correlated with the costs of bringing a drug to market." You can replace drugs in the above phrase with pretty much anything. Prices are what the market can bear (until of course someone else comes with a cheapest solution and ruins the competition)
      - 12y ago
serf 12y ago
>So why is insulin still so expensive, 90 years after it was first introduced? Because the insulin we're using now is a heck of a lot better than the insulin 90 years ago. You might as well ask why a Tesla is so expensive when the Model T was launched over a century ago. That analogy would make more sense if I needed a car to continue living. It feels somewhat akin to being okay with amputees having to live with peg-leg style prosthesis in the first-world due to being under-insured. Some people think that healthcare should be subsidized by societal infrastructures (whatever government you're a part of) due to the inherent benefits that a healthy society provides for itself and its' citizens, not including whatever morality certain folks subscribe to. In short : People tend to feel as if it is particularly low-brow for pharmaceutical companies to engage in the same type of strategies that you see normal capitalism-centric companies engage in, even if they are driven by that same normal capitalism, economies, and market.
- joshuapants 12y ago
  Insulin would still be expensive if medicine were subsidized, you just wouldn't see the price tag yourself. I don't disagree with your point, but it doesn't seem entirely relevant.
gwern 12y ago
> From 1922 until 1981, the only insulins available commercially were extracted from bovine and porcine pancreases -- a process which was both expensive and dangerous From the medical paper being summarized in OP: "A series of innovations in the insulin manufacturing process in the early 1970s helped to improve purity and reduce these side effects. In short succession, Novo introduced “monocomponent” insulins and Lilly introduced “single-peak” insulins. These safety improvements extended insulin patents into the late 1980s...Although recombinant insulin was heavily advertised as a clinically superior agent in the 1980s (Fig. 1), almost no evidence was provided to demonstrate its superiority to the best available animal-extract insulins. 26" > This makes it act roughly twice as fast as regular insulin, roughly matching the time taken for a meal to be digested and to enter the bloodstream -- thus reducing the postprandial "peak" and cutting down on the chronic hyperglycaemia related complications of diabetes...but it wasn't until 2003 that the first "24 hour" insulin was released. From the medical paper: "Although long-acting analogues cause less hypoglycemia than NPH does, 27 it has yet to be shown that analogues lead to better long-term outcomes than standard recombinant human insulin does. 28" > Because the insulin we're using now is a heck of a lot better than the insulin 90 years ago. "On the whole, insulin today is demonstrably safer and more convenient to use than products available in 1923. But whether each incremental innovation is worth the price we pay, in a world where insulin remains unaffordable to many patients with diabetes, is less certain." Tesla vs Model T?
- cperciva 12y ago
  "Although long-acting analogues cause less hypoglycemia than NPH does, 27 it has yet to be shown that analogues lead to better long-term outcomes than standard recombinant human insulin does. 28" Doctors vs. patients. Regardless of what doctors may think, as a patient I can assure you that not randomly passing out due to hypoglycaemia is a better outcome. Similarly taking one shot of basal insulin per day is better than three shots per day. "On the whole, insulin today is demonstrably safer and more convenient to use than products available in 1923. But whether each incremental innovation is worth the price we pay, in a world where insulin remains unaffordable to many patients with diabetes, is less certain." Bovine and porcine insulins didn't go away simply because newer insulins were more profitable. They went away because 99.9% of patients didn't want them any more.
westoncb 12y ago
Interesting—had no idea the 24 hour insulins were so new. I think that also explains a certain trend I've seen with insulin usage guidelines (typically) given by doctors, versus by learned users. I found at first independently, then later had the finding corroborated, that using meal-time insulin (e.g. Humalog) was too unpredictable: too many factors are involved; blood sugar almost always go too high or too low—and, unpredictability varies with dose size (even while 'appropriately' matched by carbs). So, many people, including myself decide that since the lowest dose is most predictable, use zero (this is a Type 1 speaking)—zero meal-time insulin that is—and compensate by eating an extremely small number of carbs. This is only possible because I have the 24 hour insulin running in the background, however. My perplexity about it was that doctors seem unaware of this strategy (in my experience, which, while limited does extend to several doctors, and I hear this consistently from other diabetics). Instead, there's this myth that you just calculate your carbs, match with insulin—everything's good! But if it's only been around for 11 years or so, I guess it takes a while for the literature to catch up. The other side of the situation, at the risk of sounding unappreciative, is that it feels like I'm paying Tesla prices, but still opt to get out and walk most of the time since the car is so bad. Better blood sugar monitoring tech (e.g. continuous) and quicker meal-time insulins would seriously improve things. Quicker is better since carbs convert to blood sugar more rapidly than the insulin acts—unless I take a large insulin dose. That's another aspect of insulin's unpredictability: its processing rate is proportional to dose size, so there isn't just one ratio of insulin-to-carbs—though one could still infer their dose->(insulin units/gram of carbs) function :)
- cperciva 12y ago
  So, many people, including myself decide that since the lowest dose is most predictable, use zero (this is a Type 1 speaking)—zero meal-time insulin that is—and compensate by eating an extremely small number of carbs. Yes, a lot of T1s find that keto diets work well for them. (It doesn't work for me, unfortunately; if I drop below about 80g/day of carb my liver decides that I'm starving it and dumps crazy amounts of glucose into my bloodstream.) Better blood sugar monitoring tech (e.g. continuous) and quicker meal-time insulins would seriously improve things. We have continuous blood glucose monitoring. It's expensive and not very accurate right now, but it does exist. Faster insulins are one of the big targets for use in insulin pumps, because they would make a closed-loop system much easier (less need to predict where blood glucose is going if you can just wait 10 minutes, measure a new value, and give a new bolus), but there are two difficulties: 1. Diffusion from subcutaneous tissue into the bloodstream depends mostly on the molar mass, and insulin monomers are quantized; diffusing half of a monomer would be faster, but it wouldn't be useful. 2. People need to be able to "unplug" pumps from time to time without dying, and if ultra-ultrarapid insulins leave the body too fast the mere act of taking a long shower could become life-threatening.

dekhn 12y ago

That's not the only reason it's that expensive. Pharma costs are artificially created and amortized globally.

cperciva 12y ago
The costs of bringing new drugs to market are not simply artificial. Quite aside from the costs of developing the drugs in the first place (tweaking a couple amino acids is cheap; knowing which amino acids to tweak is a bit harder) and the costs of production (significant in the case of insulin, considering the lengthy process for biosynthesis and then purification), demonstrating that a new drug is both effective and safe costs billions of dollars. Novel insulins are harder than most drugs too, since insulin is very closely related to some oncogenic hormones. A lot of novel insulins have died in trials when animals were found to have increased rates of cancer.
- dekhn 12y ago
  Sorry, I meant: pharma drug prices, not development costs. Prices are artifically created, rather than being purely based on supply and demand. I know plenty about pharma costs, having worked for one and studied the area of biophysics including protein engineering extensively. My point is that the prices of drugs are only weakly correlated with the costs of bringing a drug to market. More importantly (and kind of ignored by most) is that pharma R&D costs are shared between many drugs. For example, Genentech has a huge R&D division and there are people who work on general projects that benefit all their biologics. Also, there are ridiculous inefficiencies built into the process of bringing drugs to market, many of which are designed to avoid large-scale adverse outcomes, or to make it possible to track where the adverse outcome root cause is.
  - lotsofmangos 12y ago
    Perceived value added and range pricing is one of those invasive diseases that have managed to take root across such a wide range of human culture that it will take decades to find a cure, especially since it is doubtful that industry is working particularly hard on the problem.
  - cperciva 12y ago
    We're talking market forces here. Yes, the drug companies will try to squeeze as many dollars out of the market as they can; the counterbalancing force is that if they get too greedy, someone else will enter the market and undercut them. The fact that entering the market is so bloody expensive is why drug companies can set their prices so high without provoking a new competitor into existence.
  - raverbashing 12y ago
    " My point is that the prices of drugs are only weakly correlated with the costs of bringing a drug to market." You can replace drugs in the above phrase with pretty much anything. Prices are what the market can bear (until of course someone else comes with a cheapest solution and ruins the competition)
    - 12y ago

serf 12y ago

>So why is insulin still so expensive, 90 years after it was first introduced? Because the insulin we're using now is a heck of a lot better than the insulin 90 years ago. You might as well ask why a Tesla is so expensive when the Model T was launched over a century ago. That analogy would make more sense if I needed a car to continue living. It feels somewhat akin to being okay with amputees having to live with peg-leg style prosthesis in the first-world due to being under-insured. Some people think that healthcare should be subsidized by societal infrastructures (whatever government you're a part of) due to the inherent benefits that a healthy society provides for itself and its' citizens, not including whatever morality certain folks subscribe to. In short : People tend to feel as if it is particularly low-brow for pharmaceutical companies to engage in the same type of strategies that you see normal capitalism-centric companies engage in, even if they are driven by that same normal capitalism, economies, and market.

joshuapants 12y ago
Insulin would still be expensive if medicine were subsidized, you just wouldn't see the price tag yourself. I don't disagree with your point, but it doesn't seem entirely relevant.

gwern 12y ago

> From 1922 until 1981, the only insulins available commercially were extracted from bovine and porcine pancreases -- a process which was both expensive and dangerous From the medical paper being summarized in OP: "A series of innovations in the insulin manufacturing process in the early 1970s helped to improve purity and reduce these side effects. In short succession, Novo introduced “monocomponent” insulins and Lilly introduced “single-peak” insulins. These safety improvements extended insulin patents into the late 1980s...Although recombinant insulin was heavily advertised as a clinically superior agent in the 1980s (Fig. 1), almost no evidence was provided to demonstrate its superiority to the best available animal-extract insulins. 26" > This makes it act roughly twice as fast as regular insulin, roughly matching the time taken for a meal to be digested and to enter the bloodstream -- thus reducing the postprandial "peak" and cutting down on the chronic hyperglycaemia related complications of diabetes...but it wasn't until 2003 that the first "24 hour" insulin was released. From the medical paper: "Although long-acting analogues cause less hypoglycemia than NPH does, 27 it has yet to be shown that analogues lead to better long-term outcomes than standard recombinant human insulin does. 28" > Because the insulin we're using now is a heck of a lot better than the insulin 90 years ago. "On the whole, insulin today is demonstrably safer and more convenient to use than products available in 1923. But whether each incremental innovation is worth the price we pay, in a world where insulin remains unaffordable to many patients with diabetes, is less certain." Tesla vs Model T?

cperciva 12y ago
"Although long-acting analogues cause less hypoglycemia than NPH does, 27 it has yet to be shown that analogues lead to better long-term outcomes than standard recombinant human insulin does. 28" Doctors vs. patients. Regardless of what doctors may think, as a patient I can assure you that not randomly passing out due to hypoglycaemia is a better outcome. Similarly taking one shot of basal insulin per day is better than three shots per day. "On the whole, insulin today is demonstrably safer and more convenient to use than products available in 1923. But whether each incremental innovation is worth the price we pay, in a world where insulin remains unaffordable to many patients with diabetes, is less certain." Bovine and porcine insulins didn't go away simply because newer insulins were more profitable. They went away because 99.9% of patients didn't want them any more.

westoncb 12y ago

Interesting—had no idea the 24 hour insulins were so new. I think that also explains a certain trend I've seen with insulin usage guidelines (typically) given by doctors, versus by learned users. I found at first independently, then later had the finding corroborated, that using meal-time insulin (e.g. Humalog) was too unpredictable: too many factors are involved; blood sugar almost always go too high or too low—and, unpredictability varies with dose size (even while 'appropriately' matched by carbs). So, many people, including myself decide that since the lowest dose is most predictable, use zero (this is a Type 1 speaking)—zero meal-time insulin that is—and compensate by eating an extremely small number of carbs. This is only possible because I have the 24 hour insulin running in the background, however. My perplexity about it was that doctors seem unaware of this strategy (in my experience, which, while limited does extend to several doctors, and I hear this consistently from other diabetics). Instead, there's this myth that you just calculate your carbs, match with insulin—everything's good! But if it's only been around for 11 years or so, I guess it takes a while for the literature to catch up. The other side of the situation, at the risk of sounding unappreciative, is that it feels like I'm paying Tesla prices, but still opt to get out and walk most of the time since the car is so bad. Better blood sugar monitoring tech (e.g. continuous) and quicker meal-time insulins would seriously improve things. Quicker is better since carbs convert to blood sugar more rapidly than the insulin acts—unless I take a large insulin dose. That's another aspect of insulin's unpredictability: its processing rate is proportional to dose size, so there isn't just one ratio of insulin-to-carbs—though one could still infer their dose->(insulin units/gram of carbs) function :)

cperciva 12y ago
So, many people, including myself decide that since the lowest dose is most predictable, use zero (this is a Type 1 speaking)—zero meal-time insulin that is—and compensate by eating an extremely small number of carbs. Yes, a lot of T1s find that keto diets work well for them. (It doesn't work for me, unfortunately; if I drop below about 80g/day of carb my liver decides that I'm starving it and dumps crazy amounts of glucose into my bloodstream.) Better blood sugar monitoring tech (e.g. continuous) and quicker meal-time insulins would seriously improve things. We have continuous blood glucose monitoring. It's expensive and not very accurate right now, but it does exist. Faster insulins are one of the big targets for use in insulin pumps, because they would make a closed-loop system much easier (less need to predict where blood glucose is going if you can just wait 10 minutes, measure a new value, and give a new bolus), but there are two difficulties: 1. Diffusion from subcutaneous tissue into the bloodstream depends mostly on the molar mass, and insulin monomers are quantized; diffusing half of a monomer would be faster, but it wouldn't be useful. 2. People need to be able to "unplug" pumps from time to time without dying, and if ultra-ultrarapid insulins leave the body too fast the mere act of taking a long shower could become life-threatening.