Friday, February 08, 2008

The ACCORD study: what does it mean?

A safety analysis and early halt to the intensive glycemic control arm of the ACCORD study was reported yesterday in an NIH press release. It says intensive glycemic control (it was really what I’d call ultra-intensive glycemic control, more stringent than what current guidelines call for) was associated with increased cardiovascular mortality (essentially macrovascular) in patients with type 2 diabetes.

That’s the main, plain message I want to put forth at the beginning of this post before I dismantle the New York Times distortion of the study. The Times article not only over hypes the results but fails to mention the primary source anywhere in the article. That would at least have pointed readers to reliable information.

The opening sentence, patently absurd to anyone who understands epidemiologic issues related to type 2 diabetes, might be a bit confusing to the lay public:


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.

That might have been the popular hype but it is absolutely not what researchers believed for decades (or at any time, for that matter). Although a decade or so ago we learned that intensive glycemic control reduces microvascular complications in DM-1 (DCCT) and in DM-2 (UKPDS) there’s been nary a shred of evidence that it helps macrovascular disease (heart attacks and strokes) at least in DM-2. (A very long follow up of DCCT suggested late macrovascular benefit in DM-1 but that’s a different disease entirely from DM-2, and not the focus of this discussion).

Although evidence of macrovascular benefit in DM-2 with intensive control was lacking, as early as 1998 an analysis of UKPDS hinted at macrovascular benefits (stroke only) attributable to a unique drug (metformin) though not to intensive control, and only in a very narrowly defined population (patients with DM-2 who were obese and on metformin as initial monotherapy, not as add on). Hardly clean and convincing evidence, but it was a hint.

A stronger hint that any drug treatment might improve macrovascular outcomes in DM-2 did not come until quite recently with the release of the PROACTIVE study. So, not until late 2005 did we begin to see evidence that treatment might offer macrovascular benefits. That was unprecedented, and I blogged about it at the time as something of a breakthrough. But it was not evidence, nor in my opinion was it even hypothesis generating, that the benefits were a result of glycemic control. Far more plausible was a unique effect of pioglitazone via non-glycemic mechanisms such as favorable lipid effects, improvement in insulin resistance and improvement in endothelial function. (For a more detailed analysis on the non-evidence for the glycemic control-macrovascular notion see my posts here, here and here).

The New York Times article quoted several medical experts who expressed concern and disappointment in the results, over characterizing their reaction (in a one sentence paragraph, as if to add to the dramatic effect) as “stunned.”

There was nothing stunning or even surprising about it. So what was going on? Ironically, the clue might be found in a quote from the NYT article itself, although the article missed the significance of the clue entirely:

Dr. John Buse, the vice-chairman of the study’s steering committee and the president of medicine and science at the American Diabetes Association, described what was required to get blood sugar levels low, as measured by a protein, hemoglobin A1C, which was supposed to be at 6 percent or less.

“Many were taking four or five shots of insulin a day,” he said. “Some were using insulin pumps. Some were monitoring their blood sugar seven or eight times a day.”


Therein may lie the answer. Patients with type 2 diabetes, many of whom were no doubt obese, were given intensive insulin treatment. Some of then were even pumping! I suspect most of the patients were on insulin, given the aggressive glycemic targets that were achieved. Insulin resistant to start with, the patients treated intensively with insulin likely required high doses which promoted weight gain, sure to beget worsened resistance followed by up-ramping of the dose as they drifted out of control, and on the vicious cycle went.

Anyone who treats obese type-2 diabetics with insulin has seen the cycle. The patient is out of control, so up goes the insulin dose. After transient improvement the inevitable weight gain leads to worsened insulin resistance and worsened hyperglycemia. Up goes the dose again and the cycle repeats. That can’t be good for one’s macrovascular health because insulin resistance is in and of itself a potent macrovascular risk factor. Moreover, it fuels the metabolic syndrome, the principal dyslipidemia of type 2 diabetes.

The suggestion that intensive insulin treatment drove macrovascular disease should surprise no one. It adds to our understanding of the treatment of DM-2 by confirming what we already suspected: 1) that intensive glycemic control does not prevent macrovascular disease and 2) pharmacologic agents for DM-2, including insulin, with the possible exception of those with unique vascular protective actions, have the potential for macrovascular harm.

Other blog reactions to the study came from Med Rants and Health Care Renewal, which I may comment on in future posts as time permits. Their posts cover issues relating to the appropriateness of current guidelines and whether guideline concordant glycemic control for DM-2 really produces meaningful microvascular ourcomes.

7 comments:

Anonymous said...

Thanks for this. Your explanation makes a lot of sense. Do you think that rather than use insulin which induces weight gain and further insulin resistance, it makes sense to eliminate sugar, flour products and below-ground vegetables like potatoes in order to reduce the need for insulin in the first place?

Bernard said...

The NHLBI site has more details about the study and what happened. And the ACCORD trial page gives details on how the various groups were organized.

It does seem strange that the aimed for such a low A1C target. What was wrong with aiming for below 7%? That would have been challenging enough, even for type 2.

I do agree that the media is blowing this out of proportion. And I do wish there was a better explanation for the additional deaths. But in the end, that may never be available.

Anonymous said...

I have a lot of trouble getting Type 1 patients to take insulin 4 times a day, much less Type 2.
Maybe these mass studies suggest something terribly sinister: individuals need individualized regimens, rather than protocols !

I have eased up on glucose control in several elderly Type 1 patients, whose hypoglycemic reactions were getting a little scary. They had really nice HgbA1c values before that easing.

We should all recall the DCCT in one of its earliest publications in Diabetes Care (1985?) showing that in the "treatment group" they had 26% of people requiring help from hypoglycemic reactions. These are the people that wind up in the practice of the endocrinologist; i.e., the difficult ones.

We have to go easy on those 26% people and simply watch as they have HgbA1c levels that make us squirm. Of course, we will now use the excuse that more people died from "tight control" and reduce the vigor of glucose control.

But that leaves the 74% of people who are not "labile" (or "brittle" as we said decades ago). That group need to get their BGs down as low as they can go--without having hypoglycemia.

I wonder what the discussion will say on the ACCORD paper--will it acknowledge that hypoglycemia is dangerous in some (about 25%) or will it attempt to generalize to everyone? Does "first do no harm" mean nothing will be done to anyone that cannot be done to everyone? That will be a tough call. If NOTHING CAN BE DONE TO ANYONE, THAT CANNOT BE DONE TO EVERYONE, then Senator Clinton's idea of more nurse practitioners will make a lot of sense. We won't individualize, just follow the national protocol. Interesting.

Anonymous said...

The Times' opening is not as absurd as you make it out to be. If researchers didn't believe that tight control could reduce cardiovascular mortality, why would they have wasted millions of dollars on this huge study designed to test this premise? It only stands to reason that some of the top docs in the field believed this would be beneficial.

Robert W Donnell said...

Why do the study? Because they didn't know and wanted to find out. There was no prior evidence that it improved macrovascular outcomes and some evidence of harm with sulfonylureas and possibly insulin

Anonymous said...

In 1995 the DCCT study reported that patients with type 1 diabetes treated with intensive insulin therapy showed a reduction in diabetic retinopathy by 73%. This study has been cited by everyone (including this write up) as evidence to support that intensive treatment with insulin reduces complications in both type 1 and type 2 diabetes. The April issue of the journal Diabetes (Lachin JM et al, Diabetes 2008; 57(4): 995-1001) has reported that a repeat analysis of the same study reported in 1995 was a statistical error, and that the conclusion that intensive treatment reduces microvascular complications by 73% is no longer true. So, the most cited evidence that insulin treatment reduces diabetic microvascular complications is no longer true. Overall, treatment with insulin has neither been shown to reduce microvascular nor macrovascular complications in any randomized study so far. The results of the ACCORD study, suggest that intensive insulin treatment increases moratlity.
There is no evidence that supports a beneficial role for insulin in treatment of diabetes (except lower blood sugar, but so what!!), yet doctors still stick to the belief that insulin saves lives.

Anonymous said...

Did you know that there are certain organs in the body that rely completely on glucose for survival and function? These are: retina, nervous tissue, kidney cells, immune cells and heart muscle cells (during ischemia). What is interesting is that all these cells are involved in diabetic complications. What is even more interesting that insulin does not drive glucose into any of these cells. There are some other unknown factors that do this. I believe that diabetic complications occur because of too little glucose entering inside these 'glucose critical cells'. These cells therefore remain starved of glucose and lose function as manifested by diabetic complications. Not because there is too much of glucose entering inside these cells and causing harm. It is important to realize that glucose transport inside the cell is a highly regulated process, and merely having high glucose outside the cell, does not mean that it all enters inside. What are your thoughts on this?