A body that stores fat and spends glucose

In the 1960s George Cahill and his colleagues at Harvard studied volunteers who went without food for days and, in some cases, weeks, under close medical supervision. They measured what went into and out of the blood, and from those measurements they drew the first complete picture of how a human body feeds itself when no food arrives.

The picture begins with an imbalance. The body stores almost all of its spare energy as fat, but the brain, the red blood cells and a few other tissues depend on glucose, and glucose is the fuel the body stores least.

reserves = {     # fuel stored in a 70 kg man, after Cahill (1970): kg and kcal
    "triglyceride (fat)":  (15.0,  141_000),
    "protein (muscle)":    (6.0,    24_000),
    "glycogen (muscle)":   (0.150,     600),
    "glycogen (liver)":    (0.075,     300),
    "glucose in fluids":   (0.020,      80),
}
need = 1800                       # kcal per day at rest, a round figure
total = sum(kcal for _, kcal in reserves.values())
print(f"{'store':22s} {'kg':>6s} {'kcal':>8s} {'share':>6s}  days if used alone")
for name, (kg, kcal) in reserves.items():
    print(f"{name:22s} {kg:6.3f} {kcal:8,d} {kcal / total:6.1%}  {kcal / need:7.2f}")
print(f"{'total':22s} {'':6s} {total:8,d}")
# store                      kg     kcal  share  days if used alone
# triglyceride (fat)     15.000  141,000  84.9%    78.33
# protein (muscle)        6.000   24,000  14.5%    13.33
# glycogen (muscle)       0.150      600   0.4%     0.33
# glycogen (liver)        0.075      300   0.2%     0.17
# glucose in fluids       0.020       80   0.0%     0.04
# total                          165,980

The liver’s glycogen, the store that keeps blood glucose steady between meals, holds about 300 kcal, a few hours to a day of the body’s needs. Muscle holds twice as much glycogen, but muscle lacks the enzyme that releases free glucose into the blood, so it keeps its glycogen for itself. Fat holds enough for weeks. The problem of fasting is therefore not energy in general but glucose in particular.

The first day: glycogen

After a meal, glucose and insulin are high, and the liver stores glycogen. A few hours later, as the meal is absorbed, insulin falls and the liver begins to break glycogen down and release glucose. At the same time it begins to make new glucose from lactate, glycerol and amino acids, a process called gluconeogenesis.

The common picture of “glycogen first, then fat” is therefore too simple. Even in the first hours, the liver is making new glucose; by the second and third day it is making almost all of it. What changes over the first day is the balance.

The second and third days: fat and the liver’s ketones

As insulin falls, fat cells release fatty acids into the blood. Most tissues, including the heart and resting muscle, burn fatty acids readily. The brain does not: fatty acids are carried in the blood bound to albumin, they enter the brain slowly, and the brain oxidises them only to a small extent. A body that ran on fat alone would still have to make glucose for the brain, and the only large source of new glucose is protein.

The liver solves this by turning fatty acids into something the brain can use. In the liver’s mitochondria, fatty acids are cut into two-carbon acetyl-CoA units, and when these arrive faster than the liver can burn them, they are joined into acetoacetate and then reduced to β-hydroxybutyrate (BHB). These ketone bodies are small, soluble in water, and carried into the brain by the same transporters that carry lactate.

Weeks: the brain on ketones, and the protection of protein

The point of the switch is to save protein. Every gram of glucose the brain uses in a fast must be made, and the main raw material is the protein of muscle and organs.

g_glucose_per_g_protein = 0.56    # roughly what gluconeogenesis makes from 1 g of mixed protein
protein_kg = 6.0                  # mobilisable protein (Cahill 1970)

for label, per_day in (("early: brain on glucose", 75), ("adapted: brain on ketones", 20)):
    glucose = per_day * g_glucose_per_g_protein
    days = protein_kg * 1000 / per_day
    print(f"{label:26s} protein {per_day:3d} g/day -> glucose {glucose:3.0f} g/day; "
          f"lasts {days:3.0f} days")
# early: brain on glucose    protein  75 g/day -> glucose  42 g/day; lasts  80 days
# adapted: brain on ketones  protein  20 g/day -> glucose  11 g/day; lasts 300 days

The second line is the reason a person of normal weight can survive a fast of two months, and an obese person much longer. It is also the reason ketones matter for the rest of this book. They are not just a by-product of burning fat; they are the body’s way of feeding the brain without eating the muscles.

Cahill’s curve, in one paragraph

Put together, the measurements give a curve that every later chapter will use. In the first hours after a meal, the brain runs on glucose from the meal and then from liver glycogen. Over the first day, glycogen falls and new glucose takes over. During the second and third days, fatty acids rise in the blood, the liver makes ketones, and blood BHB climbs from about 0.1 mM to 1–2 mM. Over the following weeks ketones rise further, the brain takes more of its fuel from them, and protein breakdown falls. The schedules of Chapter 6 are, in effect, different ways of cutting this curve short: a 16:8 day stops in the first part of the second phase, a 48-hour fast reaches the third.

What this chapter does not say

It does not say that the deeper parts of the curve are good for health. Prolonged fasting carries real risks, including dangerous shifts of salt and water, low blood pressure, gout attacks and, when eating resumes after a long fast, the refeeding syndrome, in which sudden shifts of phosphate, potassium and magnesium can affect the heart (Mehanna et al. 2008). The curve describes what the body does when it must, not what it should be asked to do.