OK, let’s start
I am going to speak about type 2 diabetes. Type 1 is a completely different disease – different genetics, demographics, biochemistry, clinical issues, prognosis – it is also much less common
The fundamental problem in the treatment of people with type 2 diabetes is that we really don’t know what the cause of the complications of the disease is. And so, we really don’t know exactly what thing or things to target with treatment, what the targets should be, and how to tell who will benefit from what. It is a protean condition, probably several different entities that all get bundled into one definition, and so has been very difficult to even study.
It is the complications of the disease that are the issue. Very few people with this are symptomatic, and the symptoms are easily treated and eliminated – this does not address the issue of complications – if there were no complications, which would care what their blood sugar was?
There is very little hard evidence about what is effective in the prevention of the problems, despite decades of work and literally millions of published articles. There are reasons for this.
1) It is extremely difficult to design a good study for this disease. Just selecting who would be the subjects is difficult, since you have a definitional disease (fasting bs >124 x2, or 200 once) – The spectrum is thus from people who have fasting sugars of 128, and nothing else, to people who run blood sugars of 300-400 on huge does of insulin, to people who get ketoacidosis if stressed, to people who develop blood sugars in the >1000 range (think syrup). All of these people have type 2 diabetes
So, if you are planning a study, you have 2 choices – either you enter everybody who fits the definition, or you select a particular sub-group you think is most appropriate to whatever it is you’re studying. If you do the first, and you have a study that does not demonstrate any difference, you might have missed a sub group that did benefit, lost in the noise of the study. On the other hand, if you choose a single subgroup, a study showing improvement may not generalize to everybody.
You can get around the first problem by randomizing in each subgroup before you start – the problem there, is you would need a huge number of people to get enough in each group to be able to make a statistical argument(looking at sub groups afterward, called post-hoc analysis, is what you do when your study does not show what you want – you then slice and dice the results until you find some group that showed results, say, left handed green eyed Scandinavians – since the groups were not randomized to start with, you cannot say this was a valid effect, but can request another grant to do another study of that group)
It is hard to say how you could surmount the second problem – the best you could do is to select a group either of very high risk, or a very large portion of the total group, so that a positive result would have some real meaning
Complications(1) – the only reason we care about diabetes is the issue of complications. The number of people who are actually symptomatic with type 2 diabetes is minimal, and these are easily relieved. If it wasn’t for the complications, nobody would care what their blood sugar was
There are two types of complications of diabetes – direct, things that generally don’t happen to those without it, and indirect, those that do happen to others, but happen more or more seriously in people with diabetes
Direct:
First I would say is neuropathy, disease of the nerves directly, primarily peripheral sensory nerves, worst in the feet and lower legs, but frequently involves the autonomic system (which governs things like blood pressure, gut function, etc) and less frequently other nerves including motor nerves – there are a lot of neuropathies, and many people with them but diabetes is by far the most common cause, in my practice – this will frequently result in complete inability to sense things in the feet (I have removed carpet tacks, sewing needles, small nails, glass, small pebbles from the feet of people who had no idea they were there) – fairly commonly there is significant pain associated with, paradoxically, the inability to sense foreign objects or weight – this is a bad thing
Kidney disease is next – it is hard to tell exactly what the risk is of bad kidney disease in type 2 – partially, this is because of the change in definition – about 8 years ago , with the stroke of a word processor, the definition was changed to include now a huge number, millions, of people who were not considered diabetic before – since it takes about 20-30 years to develop consequences of diabetes, these new diabetics will not have had a chance yet to develop problems, if they ever do, and so the denominator of risk is greatly expanded – on the other hand, the group of people whose incidence of going onto dialysis right now is greatest is the over 80 population, and they virtually never have a kidney biopsy to determine what the actual cause of the kidney failure is, if they have ever had a fasting blood sugar over 125, and almost all of them have, diabetes is blamed for the kidney failure – of those kidneys I have seen at autopsy in this group, none of them have had the classical findings of diabetic disease, but that was the diagnosis they had – this swells the numerator – the risk prior to the definition change was about 3% over 30 years, which seems about right – this is a small percentage, but of a very large number, and so results in a lot of people with kidney failure
Eye disease is mentioned next – this is a terrible problem in type 1 diabetes, but not so much in type 2 – the worst form of this,called proliferative retinopathy, is pretty rare in the type 2′s – background retinopathy, much less destructive, is the most common form – since they are older, the type 2′s get all of the other problems non-diabetics get as well
vascular disease is a common problem – there are two types, microvascular and macro – the microvascular disease affects vessels like capillaries – these are microns wide, but are the business end of the circulatory system, where all the vital exchanges take place- these are selectively damaged - it is not uncommon to see someone with good palpable pulses, but terrible microvascular disease with skin ulcer and worse – type 2 are also more susceptible to disease of medium sized arteries, think coronarys, leg vessels, cerebral vessels, and even larger – the legs are involved more than the arms, for which there is no clear reason(at least when you are thinking of the big, proximal arteries – the aorta itself is not involved(go figure) – when you combine microvascular disease, macrovascular disease, and neuropathy(can’t feel the feet), you can see why foot ulcer, infections, and loss of limbs occurs
Indirect:
Coronary artery disease is very common, and although it has been declining as a cause of death(dementia and its complications are rising), it still is the leading cause of death in this country – diabetes greatly increases the risk of lethal coronary disease – the stats show that diabetes(again, remember there are many diseases under this rubric) confers the same risk of a coronary event within 5 years as having had a previous event by itself
Stroke – this is about the same – diabetes raises the stroke risk on standard assessments by about 1/4, impressive since there are only 4 states
Infection – this is interesting – diabetics get foot infections that are unlike those anyone else gets: they are more persistent, and more destructive and harder to eradicate – this does not really have anything to do with whatever the persons A1C is – it is more related to the presence of neuropathy and ischaemia (blood flow) – this may not have anything to do with diabetes per se, but is a complication of other complications – there does seem to be an increased risk of TB, and aggressive TB at that, but I really don’t think diabetics do worse with ordinary bacterial or viral infections than anyone else, provided they are in areas of adequate perfusion – some of the view that they are at more risk has to do with decreased killing function of white blood cells in the laboratory when they are in high glucose environments, but not necessarily documented in live human beings – anyway, influenza isn’t good for anybody, and so these folks should be sure to be immunized (the antivirals are truly bogus)
There are other problems, but these are the most common and severe. One difficulty is that they take decades to develop, and so if you are going to study some type of intervention, you are going to need a study that goes over decades, involving thousands of people (see the first part of this) – How would you actually do such a thing? What researchers would still be around in 2o years? how would you keep track of thousands of people for that time? where would you put the data bank ? it would be an entirely different generation interpreting the results – very hard to contemplate
So, on to #2
There is an old story where you are walking down a city street at night, and see a man on his knees under a street light, obviously looking for something. You offer to help, and he tells you he has lost his wallet, and is looking for it – you ask him, roughly, where did he lose it, and he points to a dark alley across the street and half a block away – when asked why he is looking for it here, then, he tells you the light is better here
The only reason for treating diabetes is to prevent complications, but we really don’t know what causes the complications, and if it is different in different subsets of people. If it was a simple as blood sugar, we would know that by now, and yet we do not - It is not clear what the relationship of blood sugar to complications is in diabetes. Consider the recent ACCORD trial, very well done, showing increased mortality, short term but enough to stop the study in tight blood sugar control (which was not actually that tight); also ADVANTAGE, no difference; also NICE-SUGAR, increased mortality in ICU setting of tight control. Yet, as described above, it is virtually impossible to do a good study looking at the primary endpoints, the ones we really care about
So, we use “surrogate endpoints” – we decide, fairly arbitrarily and a priori, that we will use something else as a definition of success of our treatments – so, we state as an axiom(doesn’t need to be proven), that reduction of the average blood sugar, the A1C, is an end point defining success. We state that the presence of microalbuminuria is a marker of diabetic kidney disease(it might be), and that reducing it improves that outcome(no evidence for this). Fairly unrealistic blood pressure targets (less than 120 for fanatics, until ACCORD showed that although this greatly increased pharmaceutical use, it showed no improvement in anything, and more hospitalizations for complications of treatment), or under 130 – again, no evidence for this, expert opinion requiring more people to have more meds
To date, we have no solid evidence that attaining any of these surrogate end points actually changes outcome..
So, what do you do?
I do not have a magic flashlight, so I have to look under the streetlight too, although I think I do it with some humility, knowing that I really don’t know whether all of what I advise people is truly going to help them – I know it is going to cause them considerable inconvenience, perhaps a bit of pain, side effects of meds, expense, and some degree of real risk, varying degrees, and so do my best to keep things as simple as possible, use meds that have a long history so I know what bad things they can do, do not get all excited about the newest new thing that can decrease A1C by 0.9%(gasp) in trials, at the cost of nausea, vomiting, hypoglycemia (e.g. rosiglitazone, Avendia, in all analyses is now found to increase risk of heart attack significantly – its predecessor, troglitazone, wiped out livers, the last one standing, pioglitazone, Actos, is the youngest – I expect to see it’s clay feet in the next couple of years – all for a 0.9% decrease in a surrogate endpoint – I did not prescribe any of these drugs, although I did not stop them necessarily in people whose other physicans had given them) – So, pending further data, I tend to look for an a1c of 7-8(based on the ACCORD trial) for my younger people(45-80), and asymptomatic and less than 9 in my 80+ people, with no hypoglycemia – starting at least about age 65, hypoglycemia is a much greater threat than high blood sugars(within reason, say, less than 300) – you can die in any of several ways from hypoglycemia, but I am not worried about my 90 year old developing a renal complication in 30 years(if it is even related)
