ACCORD - Take a deep breath

We're trying ourselves to take some deep breaths after seeing some of the coverage of the ACCORD trial results, out this morning. We take specific exception to the New York Times, which opened its story with:

"For decades, researchers believed that if people with diabetes lowered their blood sugar to normal levels, they would no longer be at high risk of dying from heart disease."

First of all, it's untrue. The trial EDIC, published in 2005, has proven conclusively that better glycemic control results in less cardiovascular risk for those with type 1 diabetes  - that came out with GREAT acclaim and it's never been disputed. The DCCT was published in 1993 and UKPDS in 2000 ... those trials proved conclusively that tighter glycemic control would lead to lower microvascular risk - so, risk of eye disease, kidney disease, nerve disease.

But not cardiovascular disease. The EDIC was the DCCT follow up - that data was widely awaited and hugely celebrated. UKPDS - there hasn't been follow up and that trial showed microvascular benefits, not cardiovascular benefit. (Not because there aren't any - it may just be too early to know.)

So, the writer is wrong in saying "it's been thought for decades". But that's not the half of what is wrong with this piece, what is so irresponsible.

Why does that opening make me so mad? It implies that patients who are in good control should consider changing their minds on tight glycemic control, on good diabetes management. That's wrong - that is not what ACCORD shows.

Now, ACCORD results WERE stunning, no question. Many experts I read, many on the diaTribe advisory board, expected the trial would show that greater focus on earlier, more aggressive therapy, and getting to a low A1c, would show cardiovascular benefit. It hasn't. The stunning thing is that actually a greater focus on getting to an A1c of 6 actually, at least in this trial, prompted MORE cardiovascular deaths.

What!?!

So speculation began flying about why this was true in a high-risk population. I think we don't have enough information, which is never a very interesting thing to say, but true in this case. The researchers said it wasn't due to hypoglcyemia, and that seems very clear - although we all know some hypos we report, some we don't ... and some could have come in the "sudden death" reports but we will never know.

They also said it didn't seem to be linked to any drug in particular. This one is harder to assess - so we think that means that when you add up all the deaths, it isn't statistically significant that people were on one drug than another. Does this mean that the different drug combinations and the different drug doses were all studied? No one knows. Usually announcements like this come with publishing of data, but not in this case ... so there's a lot of questions. By mentioning doses, we're not coming down on Avandia, or insulin, or SFUs, or any drugs in particular - just noting that, for example, 8 mg of Avandia is more than 4 mg and just want to make sure all the combos are studied to try to see if anything looked odd.

I digress! The rest of the New York Times piece was equally disturbing because .. well, mainly because there's no reference to the "very high risk" population that the trial focuses on. The ACCORD trial is important, but it has nothing to do with type 1 diabetes, for example. It's important, but newly diagnosed type 2 patients are on their own when figuring out how to interpret it. Those with diabetes under control are not the focus of the trial. The focus of the trial is very out of control type 2 patients. We think the NYT just as a public service should've mentioned this! They could also write about how everyone with diabetes could improve control since that would also be a good public service.

Here's the next part we saw as irresponsible: "And there are troubling questions about what the study means for people who are younger and who do not have cardiovascular disease. Should they forgo the low blood sugar targets? No one knows."

Um, well, actually, we do know. DCCT and UKPDS both show less blindness, kidney disease, and nerve disease based on lower glcyemic levels. The ACCORD trial had ZERO to do with "younger people who do not have cardiovascular disease" - so nothing has changed for anyone else, except  those at big-time risk for cardiovascular disease.  Everyone else - tight glycemic control matters. 

Now granted, here's the twist. How do we know how to define "patient at high risk of cardiovascular disease"? Is this a patient who had the entry criteria?  Is this a person the average criteria at time of death? Unfortunately, researchers wouldn't say the average A1c at the time of death or give any details on that. So we predict there will be a lot of questions flying around for the next little while. NIH confidently advised patients today to "check with their doctor" before changing anything about their treatment - on one hand, of COURSE ... on the other hand, do they think all the doctors in the country have thought it all through and know how to advise patients? Not with a trial this complex.

The one very interesting part of the article was speculation by Dr. Irl Hirsch of University of Washington who has studied people with diabetes for years and is one of the most highly thought of researchers around: " ...Or it may be that participants reduced their blood sugar too fast, Dr. Hirsch said. Years ago, researchers discovered that lowering blood sugar very quickly in diabetes could actually worsen blood vessel disease in the eyes, he said. But reducing levels more slowly protected those blood vessels." So basically he is saying maybe the speed of the drop resulted in trauma - he was referencing a similar study about retinopathy that showed this, though .. this is actually quite different from blood glucose control being bad for you.

Signing out now. Hoping to see better articles in the am. In the meantime, if you are type 1, you can read this piece with interest, but know that ACCORD has zero to do with you. If you are type 2 and newly diagnosed, keep moving on your plan - ACCORD has zero to do with you. If you have had type 2 for awhile but don't fall into the "high risk" profile, keep your goals intact. If you have type 2 and have heart disease or are at major risk of heart disease, I would read what the NIH website says, and then look out for broader information to emerge (should be very soon) - then you can look at it with your healthcare team and you can be very involved in all the decisions. And remember - no one is saying the typical goals of 6.5 - 7.0 aren't the right goals.

Although stopping the intensive arm of this trial is viewed as a reasonable thing to have done, it's not uniformly seen as the only thing to have done and many experts seem to feel it is prudent to wait and find out more before making any major changes.

And IF you know you fall into a high risk category for diabetes, definitely DO research this study - you can print it out and the FAQ at http://www.nih.gov/news/health/feb2008/nhlbi-06.htm. Please ask your doctor or educator what your A1c target should be in light of all the hubbub - you might suggest (okay, ask) if 7 is appropriate, if below 7 is right, if 6 should be avoided. This isn't that controversial to ask - it's your right - keeping in mind that your healthcare team doesn't have all the data either and that really, we're operating in a bit of a vacuum in terms of what patients at high risk of CVD should do until some details about the study come out - you can commiserate with your healthcare team for now that we don't yet have all the information.

In the meantime, if you have the time and inclination, whip out a quick letter to the mainstream media complaining about their implications that good diabetes management is questionable for people with diabetes - that's the biggest load of bollocks I've heard of in forever!  Very wrong for them to perpetuate the myth ... all just to sell some papers.

Comments

[Steve Freed, R.Ph.Diabetes Educator]

I had the opportunity to ask Dr. Richard K. Bernstein, author of the Diabetes Solution, Diabetes Diet and Audio CD's Secrets to Normal Blood Sugars, what are your impressions on the ACCORD Study:

The question is really, was the claim made by the NIH that “lowering blood sugar actually increased the risk of death” a valid claim and if not, what was really going on? Well, what I maintain is that the problem was in the treatment, not in the fact that blood sugars were lowered. When this study was started, the ADA very clearly mandated 60% to 65% dietary carbohydrate for diabetics. This high carbohydrate diet forced the use of industrial doses of insulin, large doses of sulfonylurea drugs and very possibly at the very least a hazard for hypoglycemia; and a great likelihood of rollercoaster blood sugars as we usually see in diabetics who are put on a high carbohydrate diet. So it looks like the likely high carbohydrate diet was at the core of the problem. Now, aside from the impossibility of real blood sugar control and indeed, although the Gina Colata article in the New York Times claims that the blood sugars were very low. They weren’t. They did not get below 6% on the A1C.

Six percent corresponds to an average blood sugar of 140 mg/dL. So the blood sugars were still quite high. What hazards were introduced? Well, first of all, we know from a number of studies that high carbohydrate diets tend to increase rather than reduce cardiac risk factors. But in this case, probably much more significant was the likelihood of weight gain. We use low carbohydrate diets to reduce weight. So if you’re going to give them high carbohydrate diets and large doses of insulin to off set the resulting high blood sugars, you’re going to be building fat. And doing that to someone who’s already at high cardiac risk is just adding to their burden, probably adding tremendously. In addition, there’s been a lot written about the adverse effects of high serum insulin levels, principally, on the vasculature. For example, high serum insulin levels increase vascular leakage, cause pathologic proliferation of the vascular endothelium, arterial stiffening, and hypertension. In addition, the obesity also exacerbates hypertension. So if you’re making people fatter, you’re making their hypertension worse.

On the other hand, there have been articles pointing out that for insulin deficient individuals, if you put them on insulin, you facilitate vascular repair. So there’s a balance. Small physiologic blood levels of insulin help the vasculature whereas these industrial blood levels of insulin are harmful for the vasculature. So simultaneously, you’re increasing hypertension, you’re making them fat and you’re causing vascular damage in a number of different ways just from the high carbohydrate and the high insulin.

So the conclusion that the study should be terminated perhaps should have been replaced with a revamping of the study so that the treatment would be changed to mandate very low carbohydrate diets and physiologic instead of industrial doses of insulin. One other point to remember is that high doses of insulin combined with high carbohydrate cause the rollercoaster effect. If you take a type 2 diabetic who already has cardiac disease, and you put them on the rollercoaster blood sugars, up and down, and up and down, you’re not doing his heart any good.

A similar situation occurred in the DCCT trial in the intensive arm, more hypoglycemia. I wrote an article that was published in the American Journal of Medicine, after the DCCT. They were complaining that they had so much hypoglycemia, even at A1C’s of 6% and they didn’t get below 6%, but approaching 6% they got more and more severe hypos. I pointed out the laws of small numbers wherein high carbohydrate plus high insulin causes unpredictable blood sugars..

Items missing in the NIH press release included what sort of diet they were on and which oral medications they were using. I’d be very foolish if they were put on a low carbohydrate diet. I’m willing to bet that they were put on an ADA diet and it was probably very high in carbohydrate.
I am assuming that they used high doses of insulin and sulfonylureas because it’s very hard to offset carbohydrate with metformin, Actos® or Avandia®; whereas sulfonylureas, are almost as potent as insulin but they work by boosting your insulin levels. If you can make insulin and you get a sulfonylurea, you’ll have very high serum insulin levels.

Dr. Richard K. Bernstein is the Author of Diabetes Solution, www.diabetes911.net

[Eugene Louis Siscoe]

I was disturbed by the ACCORD press releases. One was posted thru the National Heart, Lung, and Blood Institute and the other was thru the American Diabetes Association. Then came the two articles in the New York Times. This blog lambasted the New York Times. But, you failed to note that John Buse, MD of the ADA was a source quoted in the material. You also failed to note that the ACCORD Project participants ARE NOT STATISTICALLY TYPICAL. They represent the 55 to 74 year-old segment of the adult US population who have had Type-2 diabetes for at least 10 years AND who have medically significant heart issues. It should also be noted that the Project ACCORD definition of Type-2 Diabetes is closer to Metabolic Syndrome Insulin Resistance than it is to Adult Onset Insulin Dependence.

I do not understand how the people at Project ACCORD were able to declare categorically that HYPO-glycemic episodes WERE NOT the cause in any of the 257 deaths in the Intensive Insulin Treatment segment of the trials. I did not find any remarks in the Project ACCORD Protocols which would indicate that the decedents were given specialized autopsies. In an overworked medical examiner's office, an eldery patient with obvious signs of heart damage will be logged as yet another heart failure. No one will question the Cause of Death. The decedent is just one more of the 870 MILLION people who died of some form of heart failure in the US. The Diabetes issue may be a footnote; but, it will not be a primary focus.

Everyone toots the Intense Insulin Treatment program. People even speak about tight glucose control PREVENTING diabetic complications. The STATISTICAL BEST CASE is that tight glucose control only REDUCES THE SEVERITY of diabetic complications. But, the prospect of an improved quality of life must be weighed against the risk of a fatal HYPO-glycemic incident when the diabetic patient is elderly and when the diabetic patient has other significant medical issues.

Very few medical practitioners EDUCATE their patients on how to achieve the sub-6% A1c numbers safely. When I went full-bore and was able to achieve an A1c 0f 5.5%, it was a trial-and-error process INVOLVING MORE INSULIN. I experienced HYPO-glycemia daily (PG<50 mg/dL) and I experienced medically significant HYPO-glycemia (PG<30 mg/dL) at least twice each week. My current A1c target is only 6.7%. My HYPO-glycemia has dropped to perhaps once a month. My insulin usage is much closer to the 0.3µ insulin/kg body weight/day dosage factor found in the various product information sheets.

[Richard]

Unfortunately the newspapers are deplorable when it comes to conveying any sort of balanced medical information. A fair discussion would use up too much print and would not sell as well as a slanted black and white depiction that is far from the reality. The fact remains that epidemiological studies and clinical trials have failed to make the case that targeting dysglycemia (whether tight control or just improved control) reduces macrovascular disease. On the contrary, there is supportive evidence from cohort studies, pharmacoepidemilogy studies and clinical trials that current diabetes therapies may in some patients increase cardiovascular risk (depends on how you define the outcome). We should not confuse the issue of cardiovascular risk reduction with the results from the DCCT as it really does not apply to patients with type 2 diabetes. Diabetes drugs are developed to reduce blood sugar. The benefit is improved quality of life and reduction in microvascular complications. The problem is that most patients with diabetes will die from macrovascular complications. ACCORD, VA Diabetes Trial, ADVANCE were designed to hopefully answer (finally) whether reducing glucose will result in a substantial macrovascular benefit. However we should remember that targeting dysglycemia (whether it be fasting, daily mean, HbA1c) is only targeting the end game. Loss of glucose homeostasis is evidence of the body’s failed attempt to maintain that which is of much importance to survival.

[SueW]

What everyone seems to be missing is that the combining of fenofibrate with a statin drug - (zocor) may well have been a factor in the increased deaths. First, zocor and the statin drugs have a rather impressive set of risks of their own and are of questionable help in reducing heart attack risk anyway (since many studies have suggested that heart attacks happen in people of ALL cholesterol levels). And secondly, the warnings on fenofibrate (Tricor) state rather plainly that it is NOT to be used with any of the statin drugs so combining those at higher doses as was done in the ACCORD study was definitely questionable. It's a no brainer that a lower A1C is the way to lower risks of all types in diabetes type II but what everyone seems to miss is that this can be done with no medication or a low dose of metformin and DAILY cardio and dietary changes. I think they would rather sell medications than advocate what REALLY makes people "healthier" (all people, not just diabetics)

[Doc Brooks]

I have never had that much respect for the NYT, Washington Post, or just about any newspaper, tabloid, or television news outlet. I listen & watch as many as I can, then do the math to figure out the consensus, THEN do my own research. I'm keeping my A1C under 6. Because I have to watch my food as a type II diabetic, I eat better quality meals than most everyone else on the planet. I also believe we would do much better by focusing on diabetes prevention with better nutrition training much earlier in life.

[kelly]

Thanks Bernard, yes, the media gave a lot of interpretations that I would really question as someone with diabetes that has benefited hugely from improving my management from tighter control. I also don't think they gave the right information to all the doctors out there, especially primary care, who don't have time to read the fine print of studies (not that this one is out yet) but who do read the NYT.

I love that word too and am glad you didn't think it was too impolite! My maiden name is Shaughnessy : >

[kelly]

hi Chris The researchers said they had done sub-analysis for ACCORD that said that no single drug or drug combinations were responsible for the deaths, nor was hypoglycemia. I'm not sure how they reached this conclusion (half the deaths were "sudden deaths" so that's not necessarily so easy to put a cause to, for example) but there is expected to be a study out in a peer-reviewed journal, likely New England Journal of Medicine, that will give a lot more data so we can get key questions answered like: 1) what was the underlying cardiovascular disease in the entire group and in the part of the intensively managed group; 2) how many risk factors did the group that die have on average at death - or at least more info on what that group looked like. To get into the trial, you needed underlying cardiovascular disease or two of four risk factors for cardiovascular disease - obesity, high blood pressure, high cholesterol, or to be a smoker. I don't know if the write up will have all this info but we would hope so. Another question is the doses of the medicines taken - that could relate to your thinking but there isn't data yet. There is a lot of "standing aside" until there is more information. I think in a way the data safety monitoring board did a big disservice by stopping this arm of the trial before they had data to share with everyone. It is very common to see a study published the same day trial results are announced and although it's clearly a LOT of work to pull together the manuscript, if there were more information, the media might avoid jumping to the wrong conclusions, like was seen this week. There's equally as much or far more damage that can be done by getting the wrong info to 17.5 million of us. I understand for safety reasons they felt they needed to stop the trial, but it had been going four years - another two weeks to get out the relevant info about WHY they were doing it seems reasonable, no!?

[kelly]

Scott, Thanks so much as always! A couple thoughts: 1) I wish the FDA would turn the corner and think about including glycemic variability, for example. Until there are long term studies showing the impact of glycemic variability, I'm worried they won't recognize other measures beyond A1c. Everyone who cares about it should put pressure on them to consider this, since Drs. Michael Brownlee and Irl Hirsch have demonstrated the importance of it. I think of variability like "quality of A1c" - i.e., gives some context for your A1c. 2) Hypoglycemia is key, you and I both agree on that as we have discussed. I think unfortunately that a lot of hypoglcyemia causes more glycemic variability because it's so hard not to overtreat hypos, then I'm high, then I am upset about being high, and I take one of Kerri's infamous "rage boluses." So I think the hypoglcyemia days of DCCT are different today because then, we didn't have insulin analogs and the instability of insulin contributed a lot to hypoglycemia. Today, of course, hypoglycemia is still a big problem but I think that's because everyone is trying to get to tighter control, to avoid the complications. Tough circle there but we do have proof that tighter control is fewer complications. 3) Continuous monitoring is what I look to for help in hypoglycemia. The technology is still in its infancy but I think the benefits will continue to mount AND that reimbursement will emerge. CGM catches many, many of my hypos and has really enabled me to have tighter control while not being scared I'll tumble too low. This is a technology all of us deserve to use. It was hard for me at the start at many have written vividly about the challenges - but one you turn the corner with it, wow. I write my payor every week to improve reimbursement as they still don't cover it - but studies will come out late this year that will start to make a difference.

[Chris]

Perhaps these results can best be explain by insulin. The people who were most aggresive at controlling blood sugar were taking 5-6 insulin shots a day. Insulin is known to have negative effects on enothilial function. This whole thing might be retitled - Group with highest insulin levels experienced highest rate of coranary artery disease.

If true, this would suggest optimum health could be obtained on a low carb diet that minimized blood sugar levels, and required that least amount of insulin.

[Bernard Farrell]

Kelly

Thanks for a great write-up. I'm also steamed about how the media has covered this story. And thanks for using bollocks! It's a word I miss from Ireland. I don't hear it too much over here.

[Scott]

Good write-up, but of course, a comment!

You note "Only the trial EDIC, published in 2005, has proven conclusively that better glycemic control results in less cardiovascular risk for those with type 1 diabetes." but the fact is that NO study, not the DCCT (or its follow-up the EDIC), nor the UK Prospective Diabetes Study has proven conclusively that intensive glycemic control will ELIMINATE these issues, rather they will only reduce the likelihood, suggesting that there is more behind the results than glycemic control alone.

I would also add that the results from EDIC was somewhat of a double-edged sword for patients, and that is increased risk for hypoglycemia. 40% of the cohort of more than 1,000 patients reported having had at least 1 incident of hypoglycemic coma or seizure, a significant finding that tends to be overlooked. The DCCT itself reported that the incidence of severe hypoglycemia increased threefold in intensively treated patients. Less frequently acknowledged is the fact that the DCCT began in 1983 with only 278 participants and the first 2 years were devoted to planning and feasibility studies. Of the original 278 participants in the DCCT, 8 dropped out (3%) and 11 died (4%) caused in large part by severe hypoglycemia. Changes were subsequently made to the eligibility criteria for the full-scale trial to exclude anyone with this very common short-term issue with today's insulin replacement therapy, which raises questions about exactly how "random" the DCCT really was.

You're right about the ACCORD study having zero to do with patients with type 1, and patients with type 2 and don't fall into the "high risk" profile should keep their goals intact. But, the fixation on glycemic control as measured by glycosated hemoglobin (A1c) tests alone without considering the bigger issues, including quality-of-life issues, incidence of complications and other adverse events should be a big red flag to the medical establishment and especially the Food and Drug Administration that its time to consider issues besides glycemic control alone, something that has not happened to date!

[Kerri.]

GREAT write-up, Kelly. Excellent summary - I'll be quoting you this morning!