Are New Insulins Better?

Insulin was first usefully injected into a diabetic in 1922.
By 1923, insulin was being produced and sold commercially for Eli Lilly and by Haegdorn Nordisk as a nonprofit. But the insulin they were producing was worlds away from the insulin that you and I inject today. For one thing, they got it from cows and pigs; for another, it contained significant impurities and was not at all concentrated. It took a little while to standardize strengths, so that a person didn't know how much ze was getting, and there weren't any options between insulin types.

In 1936, protamine zinc insulin was released from Nordisk. This insulin had been formulated to act more slowly, to reduce the number of daily injections needed. The protamine came from the semen of trout.

In 1946, Neutral Protamine Hagedorn insulin was developed, and it began to be sold in 1950. This was called neutral because it has a pH of 7.0 like water, protamine because it contained protamine, Hagedorn after Hans Christian Hagedorn (cofounder of Nordisk), and insulin because it contained insulin (from pigs). NPH Insulin is still sold today although it currently contains synthetic insulin and synthetic protamine.
NPH had and continues to have, two points in its favor. Point one is that it last a fairly long time- two or sometimes even just one, shot of it per day will provide some insulin coverage all day. Point two is that it can be mixed with other insulins. Mixing NPH with regular insulin (or, later, Novolog, Humalog, or Apidra) allows a person to, in just two shots per day, inject for breakfast (covered by morning regular), lunch/daytime basal (covered by morning NPH), supper (covered by evening regular), and bedtime snack/nighttime basal/dawn phenomenon (covered by NPH). Some people still wanted fewer injections, a smoother basal, and an insulin with less reputation for being inconsistant- which it was.

Of the insulins developed prior to 1990, NPH and regular are the only ones that are in any sense their same selves and modern insulins, proving that they work in FDA trials, have to show non-inferiority to NPH and regular. However, a number of other insulins were around in the latter half of the 20th century, and some were probably better than NPH as far as basal insulins go.

Anyways. Time went by. Pork insulin began being molecularly altered so that it was molecularly the same as human insuln instead of being off by a molecule. In the 1980s, insulin began being produced by genetically engineered bacteria. In 1983, Lilly started selling the stuff as Humulin (Humulin N, Humulin R, Humulin L, etc) and Nordisk followed suit five years later with the Novolins. Ironically, the new insulins were less pure. Also in the 1980s, the use of insulin pumps got really going. In the 1990s, the first insulin analogs appeared, claiming to have shorter durations of action and to therefore be more appropriate for mealtime injections as well as for pump wearers. These were Humalog and Novolog. In the first decade of the twenty first century, four new insulins- Lantus and Levemir for basal, Apidra and Exubera for bolus (Apidra claiming to be super fast and Exubera being inhaled) hit the US market. The last of the pork based insulins (and Exubera) left the market.
Currently, NPH and Regular sell for considerably less than Lantus, Levemir, Humalog, Novolog, or Apidra (oddly, Apidra is the cheapest and least often covered by insurance of the newer insulins). Most diabetics are started on the newer insulins.

So when I really read the package inserts for my insulins, looking to see how much people's A1cs had improved, I was in for a shock.

None of the newer insulins have shown themselves to improve A1cs by a significant amount in any study (I would consider it significant if it represented an A1c difference of 0.4 or greater, although none really did any better than 0.2 at the most), when compared to NPH and Regular. Many of them make the claim (some more convincingly than others) that they reduce significant episodes of hypoglycemia and/or that they improve after meal excursions. But they don't do it enough to alter the A1c or fructosamine scores in large studies of patients. I will include some links at the bottom, but here is my overall impression after reading studies looking at these insulins:

Novolog (insulin aspart). Some studies comparing Novolog to Regular found slightly improved A1cs in the Novolog group versus Regular; others found no difference. After looking at a lot of studies, it seems to me that Novolog wins- by a teeny tiny margin. Novolog users have A1cs about 0.1 lower than Regular users (difference between an A1c of 7.8% and 7.9%). I suspect this has something to do with the clearer need for Regular users to inject with time to spare before meals. Novolog users in open label trials consistantly preferred Novolog, probably because getting Novolog generally meant being told to inject right before eating instead of half an hour earlier. Novolog is associated with more stomach aches. Results on whether Novolog reduces hypoglycemic events are mixed but the overall answer seems to be that Novolog users have fewer really major hypoglycemic events- probably because they don't inject and then lose track of time and forget to eat.

Only one study comparing Humalog to Regular found that Humalog reduced A1cs compared to Regular; that one was a small (30 people, 19 on Humalog and 11 on Regular) study of adult pumpers that found a difference of 0.34 (difference in average A1c was 7.66 v 8.0). Many large studies found no difference whatsoever; the one study is either a fluke or it takes a pump to see a difference (I vote for fluke- that one study didn't look well done).

Lantus does not lower A1c compared to NPH in any studies, not even studies where NPH is being given only once per day, and people know they're trying Lantus. Lantus also doesn't have the 24 hour duration it's advertised to have (it lasts longer than 24 hours in most people) and it has more site reactions- it's not neutral, it's acidic, and it stings. It does seem to cause less hypoglycemia, probably because it doesn't have that middle of the night peak. It also is associated with lower wake up numbers- I am not sure I see the value in that.

Apidra is mostly compared to Humalog and Novolog instead of to Regular. It doesn't produce lower A1cs. Nuff said.

Levemir has been tried in twice a day shots compared to NPH twice per day and sometimes did a little better, sometimes not. Overall, a meta analysis seems to say that Levemir improves A1c by 0.05 percentage points- meaning that if the A1c tests were totally accurate to the degree of the reading (that is, an A1c of 6.5 always meant your A1c was betwee 6.45 and 6.54) then half of all A1cs would go down by the last number- say, from 7.9 to 7.8. Not a huge difference.

All of the insulins mentioned had the benefit that in some cases where a person was having weird reactions to another insulin, they were able to switch to another insulin and did better.

The major differences between the insulins were ease of use- one shot of Lantus per day vs two of NPH, injecting right before eating vs half an hour earlier- and price.

Sources:

Blurb of study comparing Novolog to Regular, claiming Novolog is better with unchanged fructosamine scores. http://www.ncbi.nlm.nih.gov/pubmed/10332685
Abstract of study of 29 patients on pumps comparing Novolog and Regular. Fructosamine scores were better with regular: http://www.ncbi.nlm.nih.gov/pubmed/11194244
Abstract of large (n=1070) open label study comparing Novolog and Regular; difference in A1c between Novolog and Regular was 0.12, in favor of Novolog- confidence interval 0.03 to 0.22
Large (n=882) open label study comparing Novolog and Regular; difference in A1c was 0.15 in favor of Novolog. Novolog was injected right before eating and Regular was injected half an hour before eating. http://care.diabetesjournals.org/content/23/5/583.long
Study (n=294) comparing premixed Novolog with NPH to premixed Regular with NPH; difference in A1c was 0.10 with a confidence interval that making it statistically insignificant .http://www.ncbi.nlm.nih.gov/pubmed/12027927
Small study comparing Apidra to Novolog and Humalog used in pumps. No difference in A1cs :http://www.ncbi.nlm.nih.gov/pubmed/21457066
Small open label cross over study (n=107) comparing Regular and Humalog- Humalog made no difference in A1c http://www.ncbi.nlm.nih.gov/pubmed/8750567
Small double blind study of pumpers (n=30) on Regular or Humalog- Humalog users had A1cs .34 lower: http://www.ncbi.nlm.nih.gov/pubmed/9032100
The abstract to this large (n=1008) three month study of Regular users and Humalog users indicates that A1cs were taken but does not report changes in A1c in conclusion, suggesting that Humalog did not perform better than Regular. http://www.ncbi.nlm.nih.gov/pubmed/9000704
In another paper on the same group, they say that the A1c decreased in both the Regular and Humalog users over the study period, by the same amount: http://www.ncbi.nlm.nih.gov/pubmed/9183237
Another paper written by people who took part in a part of the above study, they note that after the big study ended, the A1cs of the people on Humalog rebounded- to higher than before the study, despite staying on Humalog: http://www.ncbi.nlm.nih.gov/pubmed/9418839
The abstract to this study of roughly 500 type 1 and type 2 diabetics says that insulin lispro users averaged A1cs of 8.1 compard to 8.3 in Regular users: http://www.ncbi.nlm.nih.gov/pubmed/9083709
This meta-analysis did not find Humalog to be better than Regular in terms of long term blood sugar control: http://www.ncbi.nlm.nih.gov/pubmed/9377611
This study of kids with type 1 showed an insignificant difference in A1c between Humalog and Apidra users. http://www.ncbi.nlm.nih.gov/pubmed/21291333
This exchange of letters includes a good summary of studies comparing Humalog and Regular, showing that only one- the one with 30 people on pumps- showed a statistically significant difference in A1c and arguing that Humalog didn't really improve hypoglycemia either: http://www.cmaj.ca/content/159/11/1353.reprint
Study (n=534 type 1s) comparing NPH and Lantus showed non-significantly better A1cs in NPH users. http://www.ncbi.nlm.nih.gov/pubmed/10834423
Study (n=518) of type 2 diabetics using NPH or Lantus showed non-significantly better A1cs in NPH users (although Lantus users didn't gain quite as much weight): http://www.ncbi.nlm.nih.gov/pubmed/11315821
Study (n=619) of type 1 NPH users who did or didn't switch to once daily Lantus. Switchers had a statistically insignificant decrease in A1c. http://care.diabetesjournals.org/content/23/11/1666.long
Study (n=349) of kids with type 1 who had been taking NPH once or sometimes twice a day who did or didn't switch to Lantus once daily. Those who switched had insignificantly higher A1cs: http://care.diabetesjournals.org/content/24/11/2005.long
Very small (n=19) study comparing Levemir and NPH. No difference.
Review of large numbers of studies comparing NPH and Levemir finds Levemir reduces A1c by statistically significant amounts- 0.05. That is, about half of people with an A1c reading of 7.7 would get a 7.6 instead. http://www.ncbi.nlm.nih.gov/pubmed/22046606
Insulin was first usefully injected into a diabetic in 1922.
By 1923, insulin was being produced and sold commercially for Eli Lilly and by Haegdorn Nordisk as a nonprofit. But the insulin they were producing was worlds away from the insulin that you and I inject today. For one thing, they got it from cows and pigs; for another, it contained significant impurities and was not at all concentrated. It took a little while to standardize strengths, so that a person didn't know how much ze was getting, and there weren't any options between insulin types.
In 1936, protamine zinc insulin was released from Nordisk. This insulin had been formulated to act more slowly, to reduce the number of daily injections needed. The protamine came from the semen of trout.
In 1946, Neutral Protamine Hagedorn insulin was developed, and it began to be sold in 1950. This was called neutral because it has a pH of 7.0 like water, protamine because it contained protamine, Hagedorn after Hans Christian Hagedorn (cofounder of Nordisk), and insulin because it contained insulin (from pigs). NPH Insulin is still sold today although it currently contains synthetic insulin and synthetic protamine.
NPH had and continues to have, two points in its favor. Point one is that it last a fairly long time- two or sometimes even just one, shot of it per day will provide some insulin coverage all day. Point two is that it can be mixed with other insulins. Mixing NPH with regular insulin (or, later, Novolog, Humalog, or Apidra) allows a person to, in just two shots per day, inject for breakfast (covered by morning regular), lunch/daytime basal (covered by morning NPH), supper (covered by evening regular), and bedtime snack/nighttime basal/dawn phenomenon (covered by NPH). Some people still wanted fewer injections, a smoother basal, and an insulin with less reputation for being inconsistant- which it was.
Of the insulins developed prior to 1990, NPH and regular are the only ones that are in any sense their same selves and modern insulins, proving that they work in FDA trials, have to show non-inferiority to NPH and regular. However, a number of other insulins were around in the latter half of the 20th century, and some were probably better than NPH as far as bolus insulins go.
Anyways. Time went by. Pork insulin began being molecularly altered so that it was molecularly the same as human insuln instead of being off by a molecule. In the 1980s, insulin began being produced by genetically engineered bacteria. In 1983, Lilly started selling the stuff as Humulin (Humulin N, Humulin R, Humulin L, etc) and Nordisk followed suit five years later with the Novolins. Ironically, the new insulins were less pure. Also in the 1980s, the use of insulin pumps got really going. In the 1990s, the first insulin analogs appeared, claiming to have shorter durations of action and to therefore be more appropriate for mealtime injections as well as for pump wearers. These were Humalog and Novolog. In the first decade of the twenty first century, four new insulins- Lantus and Levemir for basal, Apidra and Exubera for bolus (Apidra claiming to be super fast and Exubera being inhaled) hit the US market. The last of the pork based insulins (and Exubera) left the market.
Currently, NPH and Regular sell for considerably less than Lantus, Levemir, Humalog, Novolog, or Apidra (oddly, Apidra is the cheapest and least often covered by insurance of the newer insulins). Most diabetics are started on the newer insulins. So when I really read the package inserts for my insulins, looking to see how much people's A1cs had improved, I was in for a shock.
None of the newer insulins have shown themselves to improve A1cs by a significant amount in any study (I would consider it significant if it represented an A1c difference of 0.4 or greater, although none really did any better than 0.2 at the most), when compared to NPH and Regular. Many of them make the claim (some more convincingly than others) that they reduce significant episodes of hypoglycemia and/or that they improve after meal excursions. But they don't do it enough to alter the A1c or fructosamine scores in large studies of patients. I will include some links at the bottom, but here is my overall impression after reading studies looking at these insulins:
Novolog (insulin aspart). Some studies comparing Novolog to Regular found slightly improved A1cs in the Novolog group versus Regular; others found no difference. After looking at a lot of studies, it seems to me that Regular wins- by a teeny tiny margin. Novolog users have A1cs about 0.1 lower than Regular users (difference between an A1c of 7.8% and 7.9%). I suspect this has something to do with the clearer need for Regular users to inject with time to spare before meals. Novolog users in open label trials consistantly preferred Novolog, probably because getting Novolog generally meant being told to inject right before eating instead of half an hour earlier. Novolog is associated with more stomach aches. Results on whether Novolog reduces hypoglycemic events are mixed but the overall answer seems to be that Novolog users have fewer really major hypoglycemic events- probably because they don't inject and then lose track of time and forget to eat.
Only one study comparing Humalog to Regular found that Humalog reduced A1cs compared to Regular; that one was a small (30 people, 19 on Humalog and 11 on Regular) study of adult pumpers that found a difference of 0.34 (difference in average A1c was 7.66 v 8.0). Many large studies found no difference whatsoever.
Lantus does not lower A1c compared to NPH in any studies, not even studies where NPH is being given only once per day, and people know they're trying Lantus. Lantus also doesn't have the 24 hour duration it's advertised to have (it lasts longer than 24 hours in most people) and it has more site reactions- it's not neutral, it's acidic, and it stings. It does seem to cause less hypoglycemia, probably because it doesn't have that middle of the night peak. It also is associated with lower wake up numbers- I am not sure I see the value in that.
Apidra is mostly compared to Humalog and Novolog instead of to Regular. It doesn't produce lower A1cs. Nuff said.
Levemir has been tried in twice a day shots compared to NPH twice per day and sometimes did a little better, sometimes not. Overall, a meta analysis seems to say that Levemir improves A1c by 0.05 percentage points- meaning that if the A1c tests were totally accurate to the degree of the reading (that is, an A1c of 6.5 always meant your A1c was betwee 6.45 and 6.54) then half of all A1cs would go down by the last number- say, from 7.9 to 7.8. Not a huge difference.
All of the insulins mentioned had the benefit that in some cases where a person was having weird reactions to another insulin, they were able to switch to another insulin and did better.
The major differences between the insulins were ease of use- one shot of Lantus per day vs two of NPH, injecting right before eating vs half an hour earlier- and price.
Humalog (insulin lispro)
Blurb of study comparing Novolog to Regular, claiming Novolog is better with unchanged fructosamine scores. http://www.ncbi.nlm.nih.gov/pubmed/10332685
Abstract of study of 29 patients on pumps comparing Novolog and Regular. Fructosamine scores were better with regular: http://www.ncbi.nlm.nih.gov/pubmed/11194244
Abstract of large (n=1070) open label study comparing Novolog and Regular; difference in A1c between Novolog and Regular was 0.12, in favor of Novolog- confidence interval 0.03 to 0.22
Large (n=882) open label study comparing Novolog and Regular; difference in A1c was 0.15 in favor of Novolog. Novolog was injected right before eating and Regular was injected half an hour before eating. http://care.diabetesjournals.org/content/23/5/583.long
Study (n=294) comparing premixed Novolog with NPH to premixed Regular with NPH; difference in A1c was 0.10 with a confidence interval that making it statistically insignificant .http://www.ncbi.nlm.nih.gov/pubmed/12027927
Small study comparing Apidra to Novolog and Humalog used in pumps. No difference in A1cs :http://www.ncbi.nlm.nih.gov/pubmed/21457066
Small open label cross over study (n=107) comparing Regular and Humalog- Humalog made no difference in A1c http://www.ncbi.nlm.nih.gov/pubmed/8750567
Small double blind study of pumpers (n=30) on Regular or Humalog- Humalog users had A1cs .34 lower: http://www.ncbi.nlm.nih.gov/pubmed/9032100
The abstract to this large (n=1008) three month study of Regular users and Humalog users indicates that A1cs were taken but does not report changes in A1c in conclusion, suggesting that Humalog did not perform better than Regular. http://www.ncbi.nlm.nih.gov/pubmed/9000704
In another paper on the same group, they say that the A1c decreased in both the Regular and Humalog users over the study period, by the same amount: http://www.ncbi.nlm.nih.gov/pubmed/9183237
Another paper written by people who took part in a part of the above study, they note that after the big study ended, the A1cs of the people on Humalog rebounded- to higher than before the study, despite staying on Humalog: http://www.ncbi.nlm.nih.gov/pubmed/9418839
The abstract to this study of roughly 500 type 1 and type 2 diabetics says that insulin lispro users averaged A1cs of 8.1 compard to 8.3 in Regular users: http://www.ncbi.nlm.nih.gov/pubmed/9083709
This meta-analysis did not find Humalog to be better than Regular in terms of long term blood sugar control: http://www.ncbi.nlm.nih.gov/pubmed/9377611
This study of kids with type 1 showed an insignificant difference in A1c between Humalog and Apidra users. http://www.ncbi.nlm.nih.gov/pubmed/21291333
This exchange of letters includes a good summary of studies comparing Humalog and Regular, showing that only one- the one with 30 people on pumps- showed a statistically significant difference in A1c and arguing that Humalog didn't really improve hypoglycemia either: http://www.cmaj.ca/content/159/11/1353.reprint
Study (n=534 type 1s) comparing NPH and Lantus showed non-significantly better A1cs in NPH users. http://www.ncbi.nlm.nih.gov/pubmed/10834423
Study (n=518) of type 2 diabetics using NPH or Lantus showed non-significantly better A1cs in NPH users (although Lantus users didn't gain quite as much weight): http://www.ncbi.nlm.nih.gov/pubmed/11315821
Study (n=619) of type 1 NPH users who did or didn't switch to once daily Lantus. Switchers had a statistically insignificant decrease in A1c. http://care.diabetesjournals.org/content/23/11/1666.long
Study (n=349) of kids with type 1 who had been taking NPH once or sometimes twice a day who did or didn't switch to Lantus once daily. Those who switched had insignificantly higher A1cs: http://care.diabetesjournals.org/content/24/11/2005.long
Very small (n=19) study comparing Levemir and NPH. No difference.
Review of large numbers of studies comparing NPH and Levemir finds Levemir reduces A1c by statistically significant amounts- 0.05. That is, about half of people with an A1c reading of 7.7 would get a 7.6 instead. http://www.ncbi.nlm.nih.gov/pubmed/22046606