#41–Jake Kushner, M.D.: How to thrive with T1D and the valuable insights that can benefit anyone

Kushner describes his own family's transition: four years ago he and his wife cleared the pantry of all enriched carbohydrates. He now suspects his own beta-hydroxybutyrate is routinely above 1 mM. His best-controlled patients — including a man on 5–6 units/day who walks 6–10 miles daily — demonstrate the outer edge of what low-carb plus exercise can achieve in T1D.

Kushner explains the error pattern he sees in patients who try low-carb naively: they eliminate pasta and eat a large steak, take no insulin because there are 'no carbs,' and then watch a massive slow-wave glucose excursion build over 3–4 hours on their CGM. The fix is to anticipate the wave with Regular insulin.

Laurie Goodyear at the Joslin Diabetes Center has studied this pathway for decades. Kushner's patient on 5–6 units/day has the lowest body fat and leptin he has ever measured — a walking demonstration that insulin-independent glucose disposal via exercise can nearly normalize the metabolic profile of T1D.

Kushner traces the CGM history through the Dexcom G4 as the 'transformative device' that, combined with Bernstein's low-carb approach, produced near-normal A1cs of 5.7–5.9% in his best patients. Prior to CGM, A1cs of 9% were normal — not because that was physiologically optimal, but because the tools to do better safely did not exist.

Kushner points out a paradox: the trial showed that tight control saves lives, but it simultaneously revealed that the medical system lacked the tools to deliver it. The NIH spent >$100M over 7 years for 1,400 patients — an unsustainable cost. The combination of CGM + low-carb + Bernstein-style protocols finally gives individual patients and providers the tools to approach DCCT-level outcomes without near-infinite resources.

Kushner restructured his practice to see patients starting at 4:30 PM with no room turnover, so each patient gets as long as needed. He has seen A1cs drop from 14 to under 7 — transformative outcomes — primarily through this approach of radical empathy before protocol delivery. The critical realization: many teenagers with A1cs of 14 are thinking about their diabetes constantly and are overwhelmed by shame, not indifferent.

Kushner describes patients with A1cs of 14 who are, in his words, 'on a superhighway to death' — their first heart attack projected for age 35. When he spends an hour with them asking open-ended questions rather than running through a medication list, many break down because no one has ever asked how they actually feel. The ego-syntonic nature of the distress is especially insidious: patients just think 'this is life' and cannot imagine feeling different. He has seen patients who drop their A1c from 14 to under 7 describe the experience as transformative — they can think, feel, and perceive love they couldn't previously recognize.

The trial was halted because it was deemed unethical to continue depriving the control arm of what had been learned. The intervention required near-infinite resources: over $100 million for ~1,400 patients in 7 years, including coordinated conference calls where centers crowd-sourced best practices. The subsequent 23-year EDIC follow-up showed that the cardiovascular hazard ratio approximately one-third lower in the former tight-control arm — a finding Attia calls 'pretty amazing' given that both groups had been metabolically identical for most of the follow-up period.

Bernstein uses Regular (human) insulin rather than rapid-acting analogs (Humalog/Novolog) to cover protein, because Regular peaks at 1–2 hours and lasts 6–8 hours, which roughly matches the delayed gluconeogenic conversion of protein (10g protein → ~6g glucose-equivalent, over several hours). Rapid-acting analogs cause faster glucose swings that are harder to chase. Kushner describes watching his colleague Mary Ann pull croutons out of a Caesar salad and realizing she had independently arrived at the same core insight: avoid the foods that produce volatile glucose excursions.

Kushner emphasizes that the kinetics are delayed — gluconeogenesis from protein does not produce a rapid glucose spike the way starch does, so the mismatch with rapid-acting insulin analogs is even worse. Regular insulin's 6–8 hour tail roughly matches the protein wave. However, the ratio is not fixed: a person who just finished lifting weights will use more protein for muscle repair and produce less glucose, whereas a sedentary person at homeostasis produces proportionally more glucose from the same protein load.

Kushner illustrates the magnitude with a clinical anecdote: children with stable insulin doses who go on ski vacations can require continuous carbohydrate supplementation because sustained aerobic activity obliterates their insulin requirements — parents eventually learn to reduce doses by two-thirds. His leanest, most-active patient walks 6–10 miles daily, consumes no enriched carbohydrates, and takes only 5–6 units of insulin per day — a 10-fold reduction versus the population median.

Kushner frames this as 'cognitive dissonance' — patients look at a normal CGM number and have no reason to suspect DKA is developing. Beta-hydroxybutyrate may already be elevated above baseline in SGLT2-treated T1D patients even at normal glucose. The phenomenon appears most prominently in pump users with catheter occlusions, but also in injecting patients who miss a dose. Kushner emphasizes this is all off-label in T1D and urges listeners not to self-experiment without clinical trial supervision.

Attia and Kushner discuss the C. elegans work cautiously: in the worm, the effect required both reduced daf-2 and caloric restriction. The gestalt — that the T1D patient on 6 units/day and 6 miles of walking daily is 'telling us something' about what circulating insulin load does to longevity — is the clinical principle Kushner takes from it.