April 3, 2009

Latest Metformin Study Really Shows Impact of Improperly Dosed Insulin

I've had a chance to read the full text of the latest study to examine the impact of metformin on cardiac health and complications over a five year period. The study is:

Long-term Effects of Metformin on Metabolism and Microvascular and Macrovascular Disease in Patients With Type 2 Diabetes Mellitus. Adriaan Kooy et al., Arch Intern Med. 2009;169(6):616-625.

This is one of those studies where if you read the abstract rather than look at the actual statistics you come away with a very different idea of what the study really found.

This study involved 390 Dutch subjects with Type 2 diabetes. All were using Insulin. Half were given metformin and half were given a placebo. They were followed for a median 4.3 years.

The crucial information about this study does not appear in the abstract. It is this:
All subjects monitored their glucose levels at home every 2 weeks (ie, just before and roughly 90 minutes after breakfast, lunch, and dinner, and at bedtime) using the same monitoring device (Glucotouch; Lifescan, Beerse, Belgium).
You immediately notice that with subjects testing blood sugar once every two weeks, there could be no attempt to match insulin dose to carbs-eaten-per-meal. This is probably why the participants in both arms of this study never achieved A1cs close to even the ADA's anemic target 7%.

Another problem is that it is not clear how much metformin people were given. The only statement regarding this is this one:
Patients with a creatinine clearance in the range of 40 to 60 mL/min/1.73 m2 were allowed a maximum of two 850-mg tablets per day; those with a range of 30 to 40 mL/min/1.73 m2 were allowed one 850-mg tablet per day; and those below 30mL/min/1.73m2 were withdrawn from the study.
As many people will require a dose of 2550 mg a day to see the full effect of metformin, it sounds like quite a few people in the "metformin" arm were not taking an effective dose.

Cardiac Outcomes

So knowing that both groups had A1cs of at least 7.5% and knowing further that both groups ended up with fasting blood sugars higher than 140 mg/dl it should come as no surprise that whether or not they took metformin made almost no difference in their outcomes both for heart attack or complications.

At the outset of the study 21 people in the placebo group had had heart attacks (11% of all in the group) and at the end 25 or 13% of the placebo group had had heart attacks. In the group who took metformin 24 started out having had heart attacks (12% of the group) and at the end 28 had had heart attacks (14% of all)

Since the incidence of heart attack skyrockets as A1c rises from the 5% to 7% and both groups had A1cs well over 7%, this should be no surprise. The blood sugars in both groups, despite insulin use were terrible. In the placebo group the mean fasting blood sugar at the end of the study was 144 mg/dl. In the metformin group it was 146 mg/dl.

Post meal blood sugars (90 minutes after eating) in the placebo group at the end of the study averaged 162 mg/dl in the placebo group and 160 mg/dl in the metformin group but with those fasting blood sugars so high, these people were spending most of their day well over the blood sugar level (140 mg/dl) known to correlate with the development of complications.

The summary of this study while admitting that metformin "...did not improve the primary end point," adds, "Metformin did, however, reduce the risk of macrovascular disease." This makes it sound like metformin had a significant impact on macrovascular disease, but examining the actual event numbers provided in the study shows the "significance" here is statistical and very small.

The actual differences in the incidence of events like stroke and heart attack were tiny--in the range of 1 or 2% and because there were problems with the way the groups were randomized, it is hard to trust these statistics.

Most notably, 30% of the placebo group were smokers while only 19% of the metformin group were smokers. This alone would be likely to skew cardiovascular outcomes. The metformin group, conversely was on average 5 years older than the placebo group, but this is offset by the fact that the people put on metformin at the start of the study had lower serum insulin, slightly better blood sugars, and were using less insulin than the placebo group at the beginning of the study.

These tiny differences between the group taking metformin and those not, to me, suggest that the tiny 1 or 2% differences in the incidence of the cardiac outcomes were probably not as significant as the statistics might make them seem. Yes, you can adjust for smoking, etc, but with such tiny differences between the groups, such adjustments are suspect.

Impact on Weight

Over the 4.3 years were these: People in the placebo group, who started out heavier than those in the placebo group gained an average of 4 lbs. Those taking metformin gained an average of 2 lbs. These people had an average weight of 190 lbs, so the two pound difference in average final weight is not exactly miracle weight loss.

Impact on Blood Sugar

Over the 4.3 years the average A1c of the placebo group stayed unchanged at 7.9%. That of the metformin group dropped .5% from 7.9% to 7.5% but interestingly, the difference between mean fasting and postprandial blood sugars were only on the order of 2 mg/dl each and the metformin group had higher average fasting blood sugars (by 2 mg/dl) than the placebo group.

Impact on Insulin

There was a major difference between the two groups in the amount of insulin circulating in their blood streams and in the amount of insulin they were injecting--though obviously, given the very high blood sugars, the amount of insulin being injected was not anywhere near the amount needed to get real control.

The placebo group started out with average serum insulin level of 43.3 μIU/mL and ended up with a average levels of 75.2 μIU/mL. The metformin group started out with average serum insulin levels of 35.7 μIU/mL and ended up with average serum insulin levels of 46.5 μIU/mL.

The daily dose of insulin in the placebo group was 64 units a day at outset and 100 at the end of the study. In the metformin group it was 62 at outset and 75 at the end of the study.

No Effect on BP, LIpids, or Microvascular Outcomes

The researchers point out that there were no difference in the two groups in terms of lowering blood pressure or improving microvascular outcomes. With blood sugars as high as these people were running, microvascular damage was certain. We know you have to lower post meal blood sugars below 140 mg/dl to eliminate neuropathy, retinopathy and kidney failure. Since average fasting blood sugar in both groups was over 140 mg/dl and almost identical, it is no surprise they developed the same amount of microvascular complications.

What This Study Really Shows

This study is being interpreted as showing that metformin fights weight gain in people who take insulin and that it has a positive effect on some cardiovascular endpoints.

It does indeed show that using metformin makes a tiny difference in weight gain and that it reduces both the amount of circulating insulin and how much insulin need be injected to achieve the same blood sugar level.

But what this study really proves, if you read it carefully, is that if you dose insulin without regard to how many grams of carbohydrates people are eating at each meal, you will not be able to get anything approaching good control and that without good control you will end up with patients whose blood sugar is so high they are guaranteed outcomes--cardiovascular "events" and classic diabetic complications.

Until people with Type 2 are taught to use insulin in a way that matches insulin dose to the amount of carbohydrate they eat, per meal, people with advanced Type 2 requiring insulin will never be able to attain the lower blood sugars that have been associated with better outcomes, no matter what oral drugs they take.

Did These Patients Deteriorate Because Type 2 is a Degenerative Condition Or Because of Those High Blood Sugars?

The researchers as is usual, plot graphs showing that their patients needed more and more insulin as the study continued and use this to conclude that people with Type 2 always deteriorate because Type 2 diabetes is a degenerative condition.

However, all they have really proven is that people maintaining blood sugars that have been shown to be high enough to kill beta cells are likely to lose beta cells and see their blood sugar deteriorate. If the subject's blood sugars had been kept in the range advocated by the American Association of Clinical Endocrinologists, using modern insulin dosing techniques that match dose to carb intake, the outcome probably would have been quite different.

But don't expect your busy family doctor to understand that. Or, for that matter, your busy endo who doesn't have time to read anything but the abstract and who may not realize just how poorly insulin was dosed in this study.

A Silver Lining?

One last thing this data suggests to me that total insulin dose does not have a lot of impact on cardiac or complications outcome, which is probably good news. Since the metformin group was using significantly less insulin with almost identical outcomes in terms of heart attack and mortality, it seems we can stop worrying about whether high doses of insulin itself are promoting heart disease. The tiny differences in heart disease outcomes between the two groups suggest it doesn't.

The final point I must make here is this: in people who use metformin along with low carb diets with or without insulin, we usually do see improved lipids and slightly better blood sugars with the same level of dietary carbohydrate intake. Metformin is very helpful when you use it to cover a small amount of carbohydrate. This study suggests, as do so many studies, that metformin isn't able to counteract the impact of putting 300 grams of carbohydrates a day in your system with inadequate amounts of insulin. No oral drug will do that.


Anonymous said...

it is encouraging to see you mention the deleterious effects sustained moderately high sugar has on the beta cell itself. Also I am glad you brought this article out as I would never had know its existence, but how are you able to arbitrate result significance and then not even provide any p values for your fans?

Trinkwasser said...

I met someone else the other day who'd been turned down for a study on the grounds that her A1c (5.7 I think) was too low.

So long as they fail to study people who have good control all the research grants will continue to be wasted on studies like this which don't really add any information.

renegadediabetic said...

I'm not sure that the notion of insulin not causing heart disease always holds true. These folks were on a high carb diet and both groups had to inject "industrial strength doses" of insulin as Dr. Bernstein would say. The metformin group just had to inject a little less insulin. At those levels, it obviously doesn't make any difference. However, as Trinkwasser pointed out, they aren't studying people with lower A1Cs. For someone on a low carb diet who only injects small amounts of insulin, you might see a reduction in heart disease. Of course, we don't know at this point and there would be a debate as to whether it was the lower insulin dose, lower A1C, or a combination of both that caused the reduction.

Thanks for your dilligence in debunking these studies. The medical establishment is way off base with its research and is leading people down a dangerous path.

Trinkwasser said...

Mark Sissons pointed out this study which unfortunately is subscription only


A1c even in "nondiabetics" was correlated with cardiovascular outcomes, which correlates nicely with the well known previous studies in diabetics. Would love to get a hold of the data especially if insulin levels or response were measured

Adriaan Kooy, MD PhD said...

Dear reader, scientist,

Unfortunately, the appreciated comment above incorrectly states that no precise data on the metformin doses used have been reported. Therefore, I like to refer to the relevant text in the original article: Archives of Internal Medicine 2009; 169, page 619, left column, lines 15-16 from below: "The actual mean daily doses in the metformin group in the short- and long-term treatment phases were 2163 and 2050 mg, respectively." These doses are in accordance with the usual maximal / optimal doses in common practice.

Most kind regards,

Adriaan Kooy, MD PhD

Unknown said...

Don Moss
Average Type 2 diabetic human attempting to understand the intracacies of medicinal regimen including insulin and Metformian.
I appreciate your article and following comments.
Thank you