September 3, 2009

Two New Studies Support Utility of Metformin

UPDATE: Sept 16, 2009: Mouse study finds metformin blocks breast cancer stem cells. This is a mouse study, which means it may be irrelevant to humans, but added to the other data linking metformin to reduced cancer incidence, it's worth noting.

Science Daily: Diabetes Drug Kills Cancer Stem Cells in Combination Treatment in Mice.


It's well known that people with diabetes have a higher risk of cancer. In some cases, undetected cancer may be causing the diabetes--diabetes is, rarely, an early symptom of pancreatic cancer. In others the many years of exposure to high blood sugar may feed baby cancer cells which do not apparently need insulin to be able to take in and metabolize glucose.

Until now, all we could advise people with family histories of diabetes to do, to prevent or slow cancer, was to strive for early diabetes or pre-diabetes diagnosis and to lower blood sugars aggressively once abnormal blood sugars were found.

Now new data from a British epidemiological study suggests that metformin may exert an anti-cancer effect. The study is:

New Users of Metformin Are at Low Risk of Incident Cancer: A cohort study among people with type 2 diabetes. Gillian Libby et al. Diabetes Care September 2009 vol. 32 no. 9 1620-1625.doi: 10.2337/dc08-2175

The study made use of medical records collected in Tayside, Scotland UK. The researchers compared 4085 people with type 2 diabetes who were new users of metformin in 1994–2003 to a group of people with diabetes diagnosed the same year who were not given metformin.

The result was that
Cancer was diagnosed among 7.3% of 4,085 metformin users compared with 11.6% of 4,085 comparators, with median times to cancer of 3.5 and 2.6 years, respectively (P < 0.001). The unadjusted hazard ratio (95% CI) for cancer was 0.46 (0.40–0.53).
This association held up even when adjusted for "sex, age, BMI, A1C, deprivation, smoking, and other drug use."

This result is intriguing, though it makes me want to ask, "What kept the doctors from prescribing metformin to the second group, the one that had so much more cancer?" Until we know the answer to that question, it is impossible to screen out the possibility that the same factor that kept doctors from prescribing the usual first line treatment for Type 2 diabetes might have also promoted cancer.

Doctors often avoided prescribing metformin to newly diagnosed patients if those patients' A1cs are between 7-8%--a level that correlates with very damaging post-meal blood sugars--high enough,perhaps, to actively promote tumor growh. That is because doctors erroneously believe an A1c near 8% to be close enough to "good control" (which they define as an A1c of 7%) that it can be managed with "diet and exercise." The diet prescribed in the UK, however, is a low fat diet full of carbohydrates which tends to push blood sugars up. So patients managed only on diet and exercise are likely to have blood sugars spiking into the very high range after every meal.

The results in this study were supposedly analyzed to adjust for A1c--but we know that the identical A1c can represent very different peak blood sugars and very different lengths of time of exposure to those peak blood sugars.

This was demonstrated very clearly in the Kumamoto study where patients with the identical 7% A1c to the patients in the UKPDS study had far fewer diabetic complications than their peers in the UKPDS study because they used fast-acting insulin to cap their post-meal blood sugar excursions to 180 mg/dl. In the UKPDS post meal sugars were not monitored and though the patients achieved the same 7% A1c as the Kumomoto study participants, they did so with much higher post-meal numbers. (Citations to these studies can be found HERE.)

Because of this, it is not possible to adequately adjust for blood sugar status, simply by adjusting for A1c. This means it is impossible to rule out the possibility that the Metformin group in the cancer study may have been prescribed metformin because their initially higher blood sugars impelled doctors to give them more aggressive treatment which meant, long term, they had less exposure to very high blood sugars and hence, less glucose feeding their baby cancers.

It is also possible the people in the no-metformin group were sicker. Doctors won't prescribe metformin to people who show evidence of significant kidney damage or liver disease. In that case, their higher rate of cancer might be related to their other health problems. Or the kidney damage that made metformin inappropriate might have been caused by years of undiagnosed diabetes and hence much more opportunity to feed glucose to baby tumors.

But we should not dismiss the possibility that metformin really does exert a protective effect.

Whatever the answer, it should be one more bit of data that supports the idea that, if you can take it, metformin is the very best diabetic drug to take.

Another piece of research that also supports the use of metformin is the finding, also reported in this month that liver fat not, as we have been told for years, visceral fat, appears to be the culprit in the metabolic disorders associated with obesity.

Intrahepatic fat, not visceral fat, is linked with metabolic complications of obesity. Elisa Fabbrini et al. PNAS Published online before print August 24, 2009, doi: 10.1073/pnas.0904944106

The abstract of the above study is pretty tough going. Fortunately, you can read a clearer explanation of what this study found in this report in Diabetes in Control

Diabetes in Control:Liver Fat Has Greater Impact on Health than Abdominal Fat

The lead researcher in that study was quoted as saying, ""We have found that excess fat in the liver, not visceral fat, is a key marker of metabolic dysfunction." The metabolic dysfunction in question was insulin resistance and the secretion rate of very low density triglycerides--the cholesterol fraction most associated with clogged arteries.

Why, you might ask, am I linking this second study with metformin? Because there is some interesting animal research that suggests that metformin may reduce liver fat. This research has not yet been repeated in humans, though it has been found that metformin improves liver enzymes.

One recent six month long Norwegian study failed to find evidence that metformin reduced liver fat. However, I can't help but wonder if the problem in repeating the animal result might not have something to do with the stress of the very high carbohydrate diets eaten by the human subjects here. We know the medical establishment still believes, in the face of a lot of data, that a low fat diet will lower body fat levels--even though we know to the contrary that high carbohydrate levels raise triglycerides, the basic building block of fats. Metformin lowers triglycerides, and did so in this study, but it is possible it can't lower them enough in the face of the on onslaught of 300 grams a day of carbohydrate.

This makes me wonder if combining metformin with a low carb diet in a manner that produces a normal level of triglycerides--one well under 150 mg/dl, may produce the same effect in humans as has been seen in animal studies where metformin did reduce the liver fat burden.

It's worth a try, especially as there is no negative finding in the study of people taking metformin who have high levels of liver fat. It lowered their triglycerides, blood sugar, and body fat, significantly, and there is no evidence the metformin made the condition worse.

If you decide to try metformin, take it in its cheap, generic form. Many pharmacies will sell it to you for $4-8 a month. Don't take your metformin mixed into a $4/a day pill that also includes another newer drug whose safety profile is a) already known to be dangerous (Actos or Avandia) or b)Unknown but already showing areas of concern (Januvia or Onglyza).

The extended release form (Metformin ER or XR) is easier on the digestive tract.

If you have been diagnosed with Non-alcoholic Fatty Liver Disease (NAFLD) and have been taking metformin in association with a lowered carbohydrate intake, let me hear from you about your experience.



Dodger said...

Dear Jenny,

The "intensive" cohort in UKPDS had median HbA1c scores of 6.2% at the start and 8.2% at year 10. They then "averaged" the yearly median scores to come up with 7%. Is this done in other studies?

Regards Dodger

Jenny said...


The use of statistical techniques is so stupid in most medical studies, I wouldn't be at all surprised if similar techniques were used elsewhere, though they usually use means rather than medians, which gives even less representative results.

The goal in most studies is to come up with a single number to represent some complex process and create a "result" rather than to understand the underlying processes.

Gretchen said...

One reason some patients weren't taking metformin might be because they are in the "I don't want to take any drugs" camp.

I get a lot of communications from people who say their BG goes over 200 after meals but they don't want to take drugs.

Many physicians are uninformed about diabetes, but we can't blame them for everything.

Jenny said...

The "I don't want to take drugs" folks would definitely make up a part of this group.

I hear from a surprising number of people who write, "I want to take metformin but my doctor won't prescribe it," too. Usually it is because of their kidney or liver function. For others it is that the doctor doesn't think their blood sugars are bad enough. Grrrr.

Susanne said...

I just ordered 90 tablets of 500 mg Metformin ER from a British pharmacy (because I don't have a physician/insurance or an RX) because my FBS are consistently 125-138, and those numbers will just keep going up (to 180 sometimes) if I don't eat a large serving of protien after my morning coffee. I'm hoping the 500 mg will take those numbers down a bit. I've all but cut out carbs from my diet, so I'm going to try medication, exercise and smaller portions and see if that doesn't help.

Jenny said...


Most people find they have to take at least 1000 mg a day, or 1500 mg to see any impact. I believe the maximum recommended daily dose for Metformin is 2000 mg.

Susanne said...

Thank you Jenny for your clarification.

v/vmary said...

Hi Jenny,
I recently got an Oral Glucose Tolerance Test using 75 grams of I believe dextrose. My results were 90 fasting (slightly lower than Iusually get, so maybe my meter is off). 164 a half hour in. 128 one hour in and 128 2 hours in. The nurse practitioner working at the endocrinologist's office said this was normal and that I should just get my A1c checked every year. But aren't these numbers pre-diabetic? If so, how do I find someone who will prescribe metformin for me? Just call around to different endocrinologists in my area and ask the secretaries to ask the endocrinologists if they will prescribe metformin for someone with my numbers? Thanks.

Jenny said...

Those numbers are indeed normal, though towards the upper end of the normal range. Prediabetes by definition would be fasting over 100 and 2 hr reading over 140 mg/dl. Just go easy on the carbs and you should be fine.

v/vmary said...

Thanks, Jenny!! You just made me feel a ton better.