A recent study compared the effect of taking the very cheap generic sulfonylurea drug glipizide to the new extremely expensive drug Januvia. The abstract reports the results as if they meant that Januvia was far superior to glipizide in preserving insulin secretion, reducing weight, and avoiding hypos.
Fortunately, the entire study is available online for free, and I was able to see the actual statistics. What they showed, very clearly, was that the differences here were statistically significant--i.e. they probably weren't attributable to random variation--but not significant in terms of the impact on patient health.
Given that Januvia costs $190 a month and glipizide costs $4 a month (on many pharmacy generic drug savings plans) it's worth taking a closer look at the statistics.
Here's the full text article:
Safety and Efficacy of Treatment with Sitagliptin or Glipizide in Patients with Type 2 Diabetes Inadequately Controlled on Metformin: A 2-year Study. T. Seck et al,. Int J Clin Pract. 2010;65(5):562-576.
The study involved 1172 people with starting A1cs ranging from 5.8 to 9.1%, divided
into two groups. All took metformin and half each took Januvia or glipizide. As is common with randomization, there were subtle differences between the groups. Most significantly the glipizide group had 5.6% more males than the Januvia group and the glipizide group as a whole started out with a greater BMI, a slightly higher fasting glucose, and more members with very high A1cs.
I mention this because the statistically significant differences between the two groups turn out to be very small which would make you take into consideration the statistically significant differences at the start of the study in the two test groups.
All study participants were eating the extremely high carbohydrate diet prescribed by the American Diabetes Association for the two years of the study.
Only 44% of those who started the study were still in it at the end of the study. The main reason for discontinuation was blood sugar rising to dangerously high levels. 52% of the discontinuations which means 29% of all people in the study were excluded from the study results because they developed extremely high blood sugars.
How high? Here are the criteria used to drop people from the study:
Throughout the study, patients were discontinued if they failed to meet prespecified, progressively stricter glycaemic control criteria. From randomisation through week 6, patients were discontinued for fasting plasma glucose (FPG) > 15.0 mmol/l (270 mg/dl); after week 6 through week 12, FPG > 13.3 mmol/l (240 mg/dl); after week 12 through week 18, FPG > 12.2 mmol/l (220 mg/dl); after week 18 through week 30, FPG > 11.1 mmol/l (200 mg/dl); after week 30 through week 52, HbA1c > 8.0%; and after week 52 to week 104,While I'm pleased to see that ethics were used in this study, so that people whose fasting blood sugars remained high enough to cause blindness and amputation were eventually removed from the study, the researchers gave very little attention to the differences between the two groups in this regard.
In fact far more people were dropped from the study after a year because Januvia was ineffective for them--84 compared to 58 in the glipizide group.
So all the results reported in the abstract were for the minority whose blood sugar remained at what was still dangerously high levels--waking daily with fasting blood sugars as high as 199 mg/dl.
Now lets look at the people who stayed in the study. Rather than quote all the many numbers here, I'm going to refer you to the graphs you will find on this page:
Medscape: Study Results with Graphs
As you can see the differences between A1c in the two groups never exceeded .2% and that difference only occurred early in the study and disappeared at its end. Fasting plasma glucose was virtually identical in both groups throughout the study and was showing an upward trend that is minimized because those one in three people with more strongly rising fasting blood sugars were dropped from the study in stages as it progressed.
Most importantly, the meal tolerance test result graphs show that the glipizide group was producing significantly LESS insulin and C-peptide at the outset of the study and had higher blood sugars at outset than the Januvia group, which calls into question the randomization.
Even so, the differences in insulin and blood sugar during the meal test, while statistically significant are functionally unimportant.The Januvia group started out with blood sugars that rose to about 234 mg/dl (13 mmol/L) 90 minutes after eating and after 2 years those post meal blood sugars were rising to about 230 mg/dl (12.7 mmol/L). This is a level high enough to guarantee complications.
In the glipizide group, at outset the 90 minute post meal blood sugar was rising to 252 mg/dl and it didn't change after two years.
How significant the difference is between the two groups is unknown given that one started out so much worse than the other.
The differences in insulin production over the course of the study were small, but the Januvia group produced more insulin than at baseline, though it obviously had very little effect on blood sugar.
Significantly, the glipizide group did not see any deterioration in insulin production which should be reassuring to those like Dr. Bernstein who believe that the insulin stimulating drugs cause beta cell burnout.
Though the abstract reports more hypos in the glipizide group, in practice the difference between the groups was, again, very small in absolute numbers because only 1/3 of the glipizide subjects who had hypos had more than 1 hypo over 2 years.
There was a tiny difference in weight in the two groups. The Januvia group lost an average of 3.5 pounds (starting from a baseline average weight of 194.7 lbs for the Januvia group.) The glipzide group gained an average of 1.54 pounds starting from an average of 198.7 pounds.)
These weight changes are negligible in functional or for that matter, even cosmetic terms.
The abstract suggests that other side effects were pretty much the same between the two groups, but here I have to point out that they missed something important.
As I have blogged many times before, Januvia works by suppressing the DPP-4 gene which produces a peptide that is used for many different things. It helps regulate glucose secretion, yes, but it also plays an important part in how the immune system functions.
The side effects that occurred more frequently in patients taking Januvia were those that probably reflect changes in immune function: cystitis, urinary tract infection, pain in extremity and asthma. "Weight decrease" was also listed as an adverse effect which probably has to do with Januvia's ability to induce anorexia, a side effect I personally experienced when I took it for 3 months which is related to the impact of GLP-1 on the brain.
Arthritis, sinus pain, sciatica and contact dermatitis--i.e. rash--were also higher in the Januvia group. The FDA received many reports of serious drug related rashes among people taking Januvia which makes me think that the rashes reported here might not have been "contact dermatitis" at all, but a reaction to the drug.
Bottom Line: The abstract of this study will be used by Merck sales reps to convince doctors that Januvia is far superior to glipizide for their patients because it causes weight loss and increases beta cell function. But now that you've seen the actual results, you can see this simply isn't true.
Januvia and glipizide failed for 54% of those who tried them over two years and over one years far more people found it worthless compared to those who tried glipizide. The impact on blood sugar was functionally insignificant for both. The tiny amount that Januvia lowered blood sugar on average was not enough to make any difference in health. Nor was the change in weight.
At the end of the study, the average patient taking either of these drugs was experiencing extremely high blood sugars both fasting and after meals--blood sugars that we know from a huge amount of research are high enough to cause all the nightmare complications of diabetes.
So what were patients taking Januvia getting for the extra $186 a month they were paying to take Januvia rather than glipizide?
Beats me. The satisfaction of knowing they were providing shareholders of Merck with great profits?
Given that we know that inhibiting DPP-4 may encourage the growth of melanoma, ovarian cancer, lung cancer and prostate cancer, the tiny "gains" achieved on average by Januvia are hardly worth the expense or risk
But don't expect your doctor to know this, because all he or she will ever hear about is the abstract which ignores the high drop out rate, the study design that eliminated those who failed, and the fact that none of the results, whatever their statistical significance ever reached functional significance.