Kraft test results: still insulin resistant

Yes, you can be insulin resistant with perfectly cromulent glucose after years on a ketogenic diet.

As you’ve probably figured out, I love tests. Maybe cause I’m a nerd. I just want to find out what things are happening, and why.

So for my latest adventure, I spent 6 hours in the lab, getting blood drawn every hour.

Insulin

A lot of people, when confronted with the idea that insulin resistance or similar insulin-related issues are causing the obesity diabetic, will argue that only a small percentage of even obese people have very abnormal fasting glucose levels.

I watched a debate between Former Carnivore but Still on Team Insulin Paul Saladino and Renowned Hair Product Enthusiast and Cardiologist Mohammad Alo recently.

Saladino claims that practically every disease of civilization (obesity, diabetes, cardiovascular disease) is caused by insulin resistance. Alo replies that only 15-20% of people in the U.S. are shown to have insulin resistance.

But clearly, more than 15-20% of Americans have “something wrong with their metabolism,” right?

It’s just that the standard measure of fasting glucose is not nearly good enough to detect this. Even post-prandial glucose can be misleading.

For example, on ex150 I often times have fasting glucose levels in the 70s and 80s. I’ve even hit 51mg/dL a few times.

Over half a year of wearing a CGM, my glucose was quite good - even the few times I consumed large amounts of carbs.

Yet clearly, something is still wrong with my metabolism, as I’m still obese. So what could we measure instead?

Insulin

Of course it makes sense that, if testing for insulin resistance, you would test.. insulin instead of glucose. Fasting insulin is a pretty easy test to add to your blood panel, and I had it tested a few times this year:

March: 6.8
April: 10.1
July: 18.0

Rreference range: 2.6 - 24.9 uIU/mL

Huh! That trend is.. not exactly going in the right direction?

Yet despite that, even my worst measurement, in July, is still solidly within the reference range. I’d argue that the reference range is way too lenient.

The insulin levels of healthy, fit, lean people on Twitter seem to be around the 2-3 uIU/mL mark. Even a level of 5 is considered somewhat problematic.

But fasting insulin tests just that - your insulin before you’ve eaten anything. What if you wanted to graph your insulin response over time?

The Kraft test: insulin over time, after a meal

Invented by Joseph Kraft, this test was created to get a better picture of people’s metabolism than just the standard fasting glucose or Oral Glucose Tolerance Test.

Kraft believed that, by the time insulin resistance manifested as disturbed fasting glucose or even post-prandial glucose control, it was too late - the disease (“real” diabetes if you want) had already been going on, undisturbed, for probably decades.

It makes sense if you think about it: if only 15-20% of Americans have a dysfunctional metabolism (measured as e.g. fasting glucose), how come that 42% are obese, 74% (!) are overweight, and 11% have diabetes?

Clearly, we’re missing something important here.

Kraft’s goal was to diagnose people long before there would be a visible effect on glucose, hoping to catch them before it was too late.

The idea is simple: measure glucose AND insulin, and give people the same sweet drink used to measure oral glucose tolerance, a common test given to e.g. pregnant women. The drink contains 75g of glucose. It is INSANELY sweet. Let me tell you, disgustingly sweet. Like 3 bottles of coke in one.

The test measures glucose + insulin a total of 7 times. Fasted, 30 minutes after taking the glucose drink, and then at hours 1, 2, 3, 4, and 5.

Yes, you will be sitting in the lab for 6 hours. And you will have 7 holes in your arms. My arms looked like a battlefield after this.

It’s unfortunate that the test takes so long. Finding 6h of time to sit around in a lab is probably pretty prohibitive for most people. I brought a book and nearly finished it, plus did some emailing.

Alternatively, if there was a way to test insulin with a finger prick monitor at home like glucose or ketones, that would also make it much easier.

Anyway, here are my results:

Glucose is red, insulin is blue. Carbs are sweet, and heavy cream is too.

Here we see my glucose (in red) and insulin (in blue) over time. Fasted glucose is 84mg/dL, perfectly normal. Fasted insulin is 18.1, which is quite high. The discrepancy between my perfectly fine fasting glucose and my pretty high fasting insulin should raise concern already.

After drinking the disgusting glucose drink, there’s a huge spike, which peaks around 1h at 173mg/dL glucose and insulin of 107.

After 2 hours, both are back down significantly: glucose is at 95mg/dL, which would not even be pre-diabetic if fasted, let alone post-effing-75g-glucose-drink. My insulin is down to 36.8 at this point.

At 3 hours, I’m below baseline level with both glucose of 64mg/dL (technically hypoglycemic, since <70) and insulin of 6.9, which isn’t great, but it’s below 10 so I’ll take it.

After that, both glucose and insulin stabilize and I feel like I wasted the last 2 hours sitting in the lab.

So, besides the fasting insulin being quite high, how do we score this thing?

Scoring the Kraft test: patterns

The response is scored not on a single value, but on the pattern of the insulin curve.

Kraft test response patterns from an interview with Kraft

Kraft observed 4 typical patterns when putting thousands of people through his test:

• Pattern 1: Real non-diabetic. Insulin barely goes over 50, peaks early, and comes down quickly.

• Pattern 2: Slightly hyperinsulinimic. Insulin spikes slightly later and goes over 100, takes much longer to come down.

• Pattern 3: Quite hyperinsulinimic: Insulin peak very delayed, and takes even longer to come down.

• Pattern 4: Insulin rises to extreme levels (near 200), peaks late, takes forever to come down.

Obviously, pattern 1 is ideal, and the further you get away from it, the worse your metabolism handles insulin. Or should I say, the worse your insulin handles food intake? More on what exactly this could mean later.

My insulin: first pattern 2, then pattern 1

My insulin curve (green) vs. patterns 1-4

Above I’ve (roughly) transcribed the data points from Kraft’s reference patterns, with the same colors. My own insulin curve is green.

In the beginning my insulin follows a textbook pattern 2 (yellow) curve. But at the 1h mark, it starts to go down rapidly. By hour 3, it’s almost exactly a pattern 1 (blue) curve, following that trajectory.

I suppose this means I am kind of a pattern 1.5? My insulin response seems delayed, and my body definitely pours out too much of it, perhaps because it got a late start. But at hour 2, I’m back to normal.

If you look at the area under the curve, it would be about halfway between patterns 1 and 2.

Given that I used to weigh nearly 300lbs a year ago, you could maybe argue that I’m an insulin resister in recovery? But clearly, everything is not alright just yet.

Does keto affect the first-phase insulin response?

One thought I had: it is advised that people who take the Oral Glucose Tolerance Test (same 75g glucose drink, but only measures blood glucose, and only for 2 hours) eat at least 150g of carbs for 3 days before the test, if coming from a low-carb diet.

Suffice it to say I am coming from a VERY low carb diet. ex150 is ketogenic af. And that does not stand for animal fats. Although there are also plenty of animal fats.

So maybe my body got used to not having to dump a huge amount of insulin during the first-phase response? And maybe, if I were to carb-up for 3 days before the Kraft test, I would have an earlier response, and an earlier and lower peak?

On the other hand, I am still obese and VERY easily gain fat given the right circumstances. So something is clearly not quite right yet.

If only this test was a bit easier to take, instead of requiring 6 hours of sitting in a lab and getting your veins wrecked. Then I could try it again and A/B test these things.

Another thing I’d like to try: Kraft test with different foods. What if I ate a steak instead of drinking 75g of glucose? I predict the peak would be lower and more drawn out, but how long would it really take? What would the area under the curve be?

One of my main hypotheses about fat loss is, of course, that I react very unfavorably to even somewhat moderate amounts of protein. If my insulin response to protein is exaggerated, that would be a huge hint.

A thought, if I am to repeat this with the glucose drink: since everything was back to baseline after 3 hours, I would probably just cut it short right there. If anything, I expect a carbed-up insulin response to be better (due to the first-phase response), not worse. And 3 hours seems a much easier window to fit in than 6 hours. For a steak, I might have to do the full 6 hours, since it will digest much more slowly than pure, liquid glucose.

In fact, I think you could easily tell the Kraft reference patterns apart even at 2 hours. I’m not sure why he insists on 3 more draws after that. Sure, the entire curve is kind of cool. But unless you have people living in the hospital anyway, why not just do 2h like the OGTT?

A three-hour lab visit is slightly annoying. A six-hour lab visit is impossible for most people with.. you know, jobs, families, that sort of thing. I think the test would be VASTLY more accessible and frequently used with a 3 hour length.

By the way, the employees at the lab had never even heard of the Kraft test before. And they said they’ve never taken as much blood from someone in one day. It was nearly a pint, so basically a blood donation.

What does the relationship between glucose and insulin curve show us?

I think of it this way: How hard did the metabolism have to work to get the glucose intake under control?

An analogy could be your car’s engine. You can see two cars drive side by side doing 70mph. In a way, you could say, both are equivalent - they are doing the same speed.

But what if the first car is cruising comfortably at 2,000rpm, whereas the other car’s engine is redlined the whole time, just to keep up the 70mph highway speed?

The second engine is working much harder to achieve the same thing. If it’s really redlined all the time, that’s surely bad for the engine long term, and will cause increased wear and tear, eventually leading to engine failure.

A metabolically healthy person should be able to ingest a “normal” meal (simulated by the 75g glucose drink) and be able to deal with it swiftly and efficiently, without an exaggerated insulin response.

A metabolically unhealthy person might still have the glucose under control, but use vastly more insulin to do so.

By the time the car stops being able to reach 70mph, it’s much too late. We should’ve intervened years ago, when it was revving much too high while driving on the highway. Now we probably have years or even decades of increased wear & tear on our hands, and engine failure might be around the corner.

But… shouldn’t keto fix insulin resistance?

Riddle me this: if the naive Carbohydrate Insulin Model was correct, how come I still have a pretty concerning fasting insulin and a pattern 1.5 insulin curve after nearly 8 years of keto and 3 years of paleo before that?

Clearly, it’s not just the carbs. I am more and more convinced that excess carbs are the proximate cause of diabetes (and in some, obesity), not the root cause. And because of that, low-carb or keto is merely fixing the symptoms.

Now I love fixing symptoms as much as the next guy: it’s much smarter to wait out the research with your diabetes in remission than losing your eyesight or your legs and wrecking your entire body.

But let’s also not pretend that the CIM explains this sufficiently. Carbs are clearly downstream from insulin resistance. Maybe they can exacerbate it, but there must be something else.

Is it excess protein/BCAAs? Is it seed oils? Is it the sum of carbs + protein? Something completely different?

I don’t know. All I know is that if, after 8 years of strict keto, you keep telling me that carbs caused my insulin resistance, I chortle in your general direction.

Conclusion

I think I’ll repeat this test in a few months. I’d like to do it both carb-prepped and with protein. Unfortunately, if you don’t do it within a few days of each other, you’re comparing apples to oranges - did the carb prep change anything, or did I improve my insulin sensitivity in the meantime?

Still, the fasting insulin of 18.1 is a pretty good indicator of its own, especially combined with the low fasting glucose: I’m revving quite high to achieve a modest highway speed. Not quite out of reference range yet, but WAY higher than is healthy.

In case you haven’t, I definitely recommend adding fasting insulin to your routine blood panels.