The fasting insulin bands
| Fasting insulin, µIU/mL (pmol/L) | Band | Where the line comes from |
|---|---|---|
| Under 6 (under 36) | Optimal | Phoenix choice. No guideline sets a fasting insulin target. The app and the Blood Work Blueprint already use 6. |
| 6 to under 12 (36 to under 72) | Monitor | Phoenix band. Worth working on. |
| 12 or higher (72 or higher) | Talk to your doctor | Phoenix line. No society sets one either. Bring it to your next appointment with your glucose and HbA1c. It is not an emergency. |
The line rule. "Under 6" excludes 6, and "12 or higher" includes 12. So 6 is Monitor and 12 is Talk to your doctor.
Units. µIU/mL and mU/L are the same number. The pmol/L figures above use a factor of 6. Some labs use 6.945, which makes 6 µIU/mL about 42 pmol/L and 12 µIU/mL about 83 pmol/L. Check the unit and the factor on your report.
Confidence in the optimal line: low. There is no guideline, and insulin tests are not standardized between labs. I still measure it, because it moves before the numbers doctors do track.
Why measure fasting insulin?
Because insulin resistance shows up years before blood sugar does.
In 6,538 British civil servants followed for about 10 years, the people who went on to develop diabetes lost insulin sensitivity steeply over the 5 years before diagnosis, while their fasting glucose only shot up in the last 3. Their pancreas pushed out more insulin to compensate, then gave way (Tabák, 2009). A fasting insulin catches that compensation phase, when glucose and HbA1c still look normal.
It matters for the brain too. In 3,695 Finnish adults followed for 11 years, higher fasting insulin and insulin resistance predicted poorer verbal fluency and a steeper decline in it. Fasting glucose and hs-CRP did not (Ekblad, 2017).
What does APOE4 change?
APOE4 and insulin meet inside the brain.
- In mice and in neurons, APOE4 traps insulin receptors inside the cell and blunts insulin signalling. Age makes it worse, and a high-fat diet speeds it up (Zhao, 2017).
- In people, the Finnish study found insulin resistance predicted decline the same way with or without APOE4 (Ekblad, 2017).
So there is a clear mechanism and no carrier-specific number. Phoenix uses one insulin target for everyone, and treats insulin resistance as one of the most direct metabolic levers a carrier can pull.
What do the guidelines say?
Nothing. No medical society sets an optimal fasting insulin, and there are two reasons:
- The tests disagree. When an American Diabetes Association work group ran the same 39 blood samples through 10 commercial insulin tests, 7 of the 10 were off by more than 15.5% against the reference method in 36 to 100% of samples (Miller, 2009).
- No outcome trial has used it. Diabetes is diagnosed with glucose and HbA1c, so guidelines are written around those.
That is why both lines on this page are labelled as Phoenix lines.
Why did Phoenix choose under 6?
I kept under 6 µIU/mL, the line the app and the Blood Work Blueprint already agree on.
- It flags compensation early. Insulin rises before glucose does (Tabák, 2009), so a tight insulin line is where a carrier gets the earliest warning.
- Not under 5. Some earlier posts, including one of ours, said under 5. No study supports 5 over 6, and with lab-to-lab differences above 15% (Miller, 2009), a 5 and a 6 can be the same blood.
- No lower line. Phoenix sets no floor. An older app row started "optimal" at 2; that floor had no evidence and is being removed.
- The talk line, 12, is the Blueprint's existing line, twice the optimal line. Bring it to your doctor alongside your glucose and HbA1c.
How do you get fasting insulin tested?
Ask for "fasting insulin" on the same draw as your fasting glucose and HbA1c, after the same overnight fast. It is not on most routine panels, so ask by name. The Blood Work Blueprint lists it with the rest of the APOE4 panel.
Use the same lab every time. Because insulin tests are not standardized (Miller, 2009), a change of lab can shift your number with no change in you.
What about HOMA-IR? HOMA-IR is calculated from your fasting glucose and fasting insulin. It tracks the gold-standard clamp test closely (correlation 0.88) and correlates 0.81 with fasting insulin alone, but it is imprecise for one person, with about 31% variation (Matthews, 1985). Phoenix's HOMA-IR line is still under review, so this page gives no HOMA-IR number. Read insulin and glucose side by side.
What moves fasting insulin?
Ranked by the size of the drop in the best trial data. Most trials enrolled people with overweight.
| Lever | What it did to fasting insulin | Who was studied | Source |
|---|---|---|---|
| Plant-based diet (vegan or vegetarian) | 4.13 µIU/mL lower | Adults with overweight, 8 trials, 716 people | Termannsen, 2024 |
| Regular exercise (aerobic and resistance worked best) | 2.40 µIU/mL lower | Adults with overweight, 35 trials, 2,752 people | Hejazi, 2023 |
| Combined aerobic and strength exercise | A large drop (standardized effect 1.02) | Inactive adults without diabetes, 24 trials | Silva, 2024 |
| Berberine (supplement) | HOMA-IR about 1 point lower | Adults with metabolic disorders, umbrella review | Nazari, 2024 |
| Low-carbohydrate diets | Large gains in insulin sensitivity at 6 months, faded by 12 | Type 2 diabetes, 23 trials, 1,357 people | Goldenberg, 2021 |
| Intermittent fasting, 6 months or more | No change in insulin against a control diet | Adults with overweight, 24 trials, 2,032 people | Khalafi, 2025 |
Weight loss drives it. In a controlled trial, losing just 5% of body weight improved insulin sensitivity in the liver, muscle and fat at once, and losing more improved it further (Magkos, 2016). Semaglutide, a prescription GLP-1 medicine, produced 14.9% weight loss in adults without diabetes (Wilding, 2021) and improved insulin resistance in people with prediabetes (Perreault, 2022). A GLP-1 medicine plus structured exercise improved insulin resistance more than either alone (Vandoni, 2026).
If you are lean, the closest trial data is exercise in inactive adults without diabetes: combined aerobic and strength exercise gave one of the largest insulin drops in the table (Silva, 2024).
What Phoenix members use. Among Phoenix members who log their stack (counted 2 Oct 2026): magnesium (156 members), berberine (38), and metformin, semaglutide or tirzepatide (38). Magnesium also improved insulin-sensitivity markers in people at risk of diabetes (Veronese, 2021).
How often should you retest?
No guideline sets an interval, because no guideline uses fasting insulin.
Every time you test: take fasting insulin with fasting glucose and HbA1c, at the same lab, after the same fast. After a change, retest with your next blood test and compare like with like. Read a trend over two or more results, not one number.
What Phoenix members do: among the 63 members who uploaded two or more insulin results, the median time from first to latest was 207 days.
What Phoenix members' results show
Where members stand. Of 144 Phoenix members who uploaded a fasting insulin result (counted 2 Oct 2026; as reported by the lab, since most uploads do not record whether the draw was fasting), 107 (74%) are under 6 and 31 (22%) are in Monitor. The median is 4.4 µIU/mL. Insulin is where Phoenix carriers look best: most were already under 6 at their first test.
Two members. In Phoenix Research Release 001:
- An APOE3/4 woman in her 60s lowered her fasting insulin from 12.6 to 6.5 µIU/mL in 162 days on metformin and tirzepatide: from the talk band to the edge of optimal. Nothing else was recorded in her stack.
- An APOE4/4 woman in her 60s went from 6.3 to 4.2 µIU/mL in 70 days, from Monitor to optimal. Release 001 credits tirzepatide; she also took a B complex, ezetimibe and magnesium glycinate.
Members chose their own changes, and there is no placebo group. How Phoenix reads member data
When should you talk to your doctor?
Talk to your doctor if your fasting insulin is 12 µIU/mL (72 pmol/L) or higher. This is a Phoenix line, not a guideline line, so bring it with your fasting glucose and HbA1c: your doctor will read the three together.
Raise it sooner if your fasting glucose is 100 mg/dL or higher, or your HbA1c is 5.7% or higher, whatever your insulin.
Between 6 and under 12, the result is worth working on. Confirm it at the same lab before you read much into it.
If you take metformin, a GLP-1 medicine or insulin, do not change it because of this page.