There’s a lot of talk right now about diet optimization for metabolic health. Bryan Johnson, carnivore vs. low-carb, intermittent fasting, all of it. Some of it is solid science. But I think there’s a gap between optimizing your diet and actually reversing pre-diabetes. I was eating well before my A1c hit 6.1. Actually well. So when people say “diet is the foundation,” I don’t disagree. But i also know it wasn’t enough for me. The problem wasn’t that I was eating badly; it was insulin resistance that food choices alone couldn’t fix. What actually moved my numbers was medication plus the diet. My fasting glucose dropped from 119 to 108 in three months. Real progress. But diet without addressing the underlying metabolic issue would’ve kept me spinning my wheels. The takeaway: diet matters. But if you’re already pre-diabetic, it might not be the only thing that matters. Has anyone else found that true?
one thing the diet-only crowd undersells is the role of liver fat in fasting glucose specifically. hepatic insulin resistance is what drives that morning number more than peripheral muscle resistance does, and DNL (de novo lipogenesis from fructose and refined carbs) can keep liver fat elevated even on what looks like a clean diet, depending on body comp and genetics. there’s a body of work going back to Roy Taylor’s twin cycle stuff suggesting fasting glucose tracks liver fat pretty tightly, which is why aggressive caloric restriction sometimes drops it fast while “eating well” at maintenance does very little. the 119 to 108 drop is meaningful but worth noting fasting glucose has an intra-individual CV around 5-7%, so single draws are noisier than people treat them. trend over 3-4 measurements > any one number imo.
edit: forgot to add
Your point about the noise window matches my experience exactly - daily logs swing from 105 to 112 without me knowing why, and I only caught the 119-to-108 improvement by plotting the full three months together. Single readings are practically decorative without the context of trend. idk, That’s where the story actually is - not in any individual morning.
“the problem wasn’t that I was eating badly; it was insulin resistance that food choices alone couldn’t fix” is the part I’d sit with, because that framing is right but it’s also where the diet-only crowd loses the thread. insulin resistance has a behavioral component that diet can touch and a physiological component that often won’t move on diet alone past a certain threshold, and sorting which one is doing the work in any given person isn’t something you can do by feel, you have to look at fasting insulin trending alongside the glucose, not just the glucose. where I’d push back a little: “diet was already good before the A1c hit 6.1” is the part that’s hard to evaluate from the outside, because “eating well” is doing a lot of work in that sentence. for some people genuinely well-constructed diets still aren’t enough because the underlying beta cell stress and hepatic insulin resistance are too far along, and for others “eating well” turns out to mean “no obvious junk but a lot of stealth carbs and not enough protein,” which is a different problem with a different fix. I’m not assuming yours was the second one, I’m just saying the only way to know is fasting insulin and trig:HDL alongside the A1c, not the food log. the fasting glucose dropping 119 to 108 in three months is solid and consistent with what the medication does mechanically, but I’d want to see the fasting insulin alongside it before I’d call it “the medication did what diet couldn’t.” sometimes the glucose moves first because the medication is suppressing hepatic glucose output and slowing gastric emptying, while the insulin resistance itself is on a slower curve. it doesn’t change your conclusion necessarily, it just means the picture is a bit more layered than diet vs. drug. so yes, agree diet alone wasn’t going to do it for you. just be a little careful with the framing that diet was already optimised, because that’s the piece that’s hardest to verify from the inside.
the intra-individual CV on a fasting glucose draw is something like 5-7% in non-diabetics and creeps higher as you move into IFG/IGT territory, which means a single 105 to 112 swing is sitting almost entirely inside assay plus biological noise before you’ve even talked about what you ate at 9pm or how you slept. the “practically decorative” line is doing real work, and the part of it that doesn’t get said often enough is that the noise floor isn’t constant across the dysglycemic range. as fasting glucose drifts up into the 110-125 zone, the day-to-day variability widens, partly because hepatic glucose output gets twitchier and partly because the cortisol awakening response starts loading more variance onto the morning draw. so the same person who looks reasonably stable at an avg fasting of 95 will look like a chaos generator at an avg of 115, and a lot of what people read as “my diet stopped working” is really just the variance fan opening up. the related thing nobody talks about in these threads is that fasting glucose and A1c don’t degrade in lockstep, which matters for how you read your trend. A1c is essentially a 90-day weighted average that biases toward the most recent 30 days (something like 50% of the signal comes from the last month iirc), while fasting glucose is mostly catching hepatic insulin sensitivity at one specific time of day. you can have a fasting glucose that improves meaningfully while A1c barely moves, or vice versa, because they’re sampling different things. one paper i’ve seen cited put the correlation between fasting glucose and A1c at around r=0.5 to 0.6 in pre-diabetic populations, which sounds high until you remember that means roughly 65-75% of the variance in one isn’t explained by the other. practical version: if you’re going to plot fasting glucose, also plot a 7-day rolling mean alongside the raw points, because the rolling mean is closer to what your A1c is actually integrating. the raw daily points are real measurements but they’re not the variable you’re trying to move. ymmv on whether your meter is calibrated for that level of inference, finger-stick CV is its own conversation.
fasting glucose 119 to 108 in three months is real but worth flagging that’s still a glycemic-endpoint move, not a remission move, and the pre-diabetic bucket probably has more reversibility headroom than full T2D bc beta cell dysfunction isn’t as entrenched yet. the “spinning my wheels” framing tracks for me too, my fasting insulin was the number that exposed it, A1c looked fine for years while insulin climbed quietly underneath.
the case for “diet is the foundation” holds up, but you’re right that it assumes a working insulin response to build on. where I’d gently push back: A1c 6.1 with fasting glucose of 119 isn’t always insulin resistance alone, sometimes it’s phenotype. fwiw asking what ur fasting insulin actually was before medication might have changed the framing.