444 AppendixA

tions may at this moment represent more theory than fact, but they're
based on the latest information avaUable to us.

HOW DID THE COMMONLY PRESCRIBED

HIGH-CARBOHYDRATE DIET COME ABOUT?

If, like me, you've had diabetes for a whUe, you've probably been told
to cut waydown on your dietary intake of fat, protein, and salt, and to
eat lots of complexcarbohydrate.You mayeven still read this advice in
publications circulatedto diabeticpatients.
Why is such advice being promulgated, when the major cause of
such diabetic compUcations as heart disease, kidney disease, high
blood pressure,and blindness is high blood sugar?
When I firstdeveloped diabetes, in 1946, Uttle wasknown about why
this disease, even when treated, caused early death and such distressing
compUcations. Priorto the avaUabiUty of insulin, about twenty-five years
earUer, peoplewith type 1 diabetes usuaUy died within a few months of
diagnosis. TheirUves couldbe prolonged somewhat with a diet that was
verylowin carbohydrate and usuaUy highin fat. Mostsufferers from the
milder type 2 diabetessurvivedon this type of diet, without supplemen
tal medication. When I became diabetic, oral hypoglycemic agents
were not avaUable, and many people werestiU foUowing very low carbo
hydrate, high-fat diets. It was at about this time that diets very high in
saturated fats,with resultant high serum cholesterollevels, were experi-
mentaUy shown to correlate with blood vessel and heart disease in ani
mals. It wasprompdyassumed bymanyphysicians that the then-known
compUcations of diabetes, most of which related to abnormaUties of
large or smattbloodvessels, were caused bythe high-fat diets. I and many
other diabetics were therefore treatedwith a high-carbohydrate, low-fat
diet. This new diet was adopted in the mid-1940s by the American Dia
betes Association (ADA), the New York Heart Association, and eventu-
aUy by the American HeartAssociation (AHA) and other groupsaround
the world. On the new diet, most of us had much higher serum choles
teroland triglyceride levels, and still developed the grave long-termcom
pUcations of diabetes. Seemingly unaware of the importance of blood
sugar control, the ADA raisedthe recommended carbohydrate content
from 40 to 50 percent of calories, and then more recentiy to 60 percent.
The ADA's most recentguidelines havebackedoffbyvaguely stating that
some diabetics may do betterwith less carbohydrate.
What About Dietary Restrictionson Fat, Protein, and Salt? 445

RECENT DEVELOPMENTS REGARDING RISK

FACTORS FOR HEART DISEASE

In the past twenty years, researchstudies have generated considerable

new information about heart disease and vascular (blood vessel) dis
ease in general, and their relationship to diabetes in particular. Some
of this recent information is summarized here.
A number of substances have been found in the blood which relate
to riskof heart attacks and vascular disease. These includeHDL(high-
density Upoprotein), LDL (low-density Upoprotein), triglyceride,
fibrinogen, homocysteine, C-reactive protein, ferritin (iron), and Upo
protein^). High serum levels of dense, compact LDL particles, tri
glyceride, fibrinogen, homocysteine, C-reactive protein, ferritin, and
lipoprotein(a) tend to be associated with increasedcardiovascularrisk,
whUe high levels of HDL tend to protect from cardiovasculardisease.
Cholesterol is a component of both LDL and HDLparticles. The frac
tion of total cholesterol found in LDL particles isan indexof risk,whUe
the fraction of cholesterol found in HDL particles is an indexof pro
tection.Nowadays, whenwewant to estimate the effects of Upids (fatty
substances) upon the risk of coronaryartery disease, welook at the ra
tio of total cholesterol to HDL and also at fasting triglyceride levels.
Someone with high serum HDL can thus have a high total cholesterol
and yet be at lowstatistical risk for a heart attack. Conversely, a person
with lowtotal cholesterol and verylowHDLmaybe at high risk.
A large multicenter study (the LipidResearch ClinicsTrial) investi
gatedthe effects of a low-fat, high-carbohydrate diet versus a high-fat,
low-carbohydrate diet on nondiabeticmiddle-aged men with elevated
cholesterol levels. The study foUowed 1,900 people for seven years.
Throughout this period, total cholesterol had dropped only 5 percent
from baseline in the low-fat group, but serum triglyceride went up
about 10 percent! (Serum triglyceride rises very rapidly after a high-
carbohydrate meal in nondiabetics, and moves up and down with
blood sugarlevels in most diabetics.) As with prior studies, no signif
icant correlation was found between serum cholesterol levels and
mortality rates. Furthermore, a study reported in the Journal of the
American Medical Association in 1997 showed that a 20 percent in
crease in either saturated or monounsaturated dietary fat loweredthe
risk of stroke to one-eighth of what it was in individuals on lower-fat
diets. Unsaturated fats showed no such benefit.
On average, diabetics with chronicaUy high blood sugars have ele-
446 AppendixA

vated levels of LDL (the "bad" cholesterol) and depressed levels of

HDL (the "good" cholesterol), even though the ADA low-fat diet has
now been in use for many years. Of great importanceis the recentdis
coverythat the forms of LDLthat harm arteries are smaU, dense LDL,
oxidized LDL, and glycated LDL. AU of these increase as blood sugar
increases. In addition, independendy of blood sugars, high serum in
sulin levelsdictatedby high-carbohydrate diets bring about increased
production of the potentiaUy hazardous smaU, dense LDL particles
and enlargement of the ceUs lining the arteries. We now can measure
the size distribution of these LDL subparticles as a routine laboratory
test. Most labs report the benevolent large, buoyant LDL subparticles
as"type A."
Under normal conditions, receptors in the liver remove LDL from
the bloodstreamand signal the Uver to reduce its manufactureof LDL
when serum levels rise even sUghdy. Glucose may bind to the surface
of the LDL particle and also to liver LDL receptors, so that LDL can
not be recognized by its receptors. In people with high blood sugars,
many LDLparticles become glycosylated, and are therefore not cleared
by the Uver. This glycosylation is reversible if blood sugar drops. After
about 24 hours, however, a rearrangement of electronbonds occursin
glycosylated proteins, so that the glucose can't be released even if
blood sugar drops. This irreversible glycosylation is caUed glycation,
and the affected protein molecules are said to be "glycated." They are
also referred to as AGEs, or advanced glycosylation end products.
These AGEs accumulate in the blood, where they can become incor
poratedinto the waUs of arteries, forming fatty deposits caUed atherotic
plaques. Since Uver LDL production cannot be turned off by the
glycosylated/glycated LDL (and also the presence of glycosylated/
glycated LDL receptors), the liver continues to manufacture more
LDL,even though serum levels may be elevated.
The proteins in the waUs of arteries can also become glycosylated/
glycated, rendering them sticky. Otherproteins in the bloodthen stick
to the arterial waUs, causing further buUdup of plaque.
Serum proteins glycosylate in the presence of glucose. White blood
ceUs caUed macrophages ingest glycosylated/glycated proteinsand gly
cosylated/glycated LDL. The loaded macrophages sweU up, becoming
very large. These transformed macrophages, loaded with fatty mater
ial, are catted foam ceUs. The foam ceUs penetrate the now sticky arte
rialwaUs, causing disruption of the orderly architecture of the artery,
and narrowthe channel through which blood can flow.
What About Dietary Restrictions on Fat, Protein, and Salt? 447

The middle layer ofthe waUs oflarge arteries contains smooth mus
cle ceUs that caninvade the fatty coating (plaque) that these ceUs cre
ate. They then prevent the plaque from breaking loose. When the
nerves that control this smooth muscle die, as in diabetic autonomic
neuropathy (caused by high blood sugars), the muscle layer dies and
calcifies. It then cannot prevent plaque rupture. When a piece of rup
tured plaque enters the blood it can block narrow vessels upstream
and cause a heart attack or stroke.
In recentyears, the tendency ofblood to clot has come into focus as
a major cause of heart attacks. People whose blood clots too readUy
are at very high risk.You may recaU that one of the medicalnames for
a heart attackis coronary thrombosis. A thrombus is a clot, and coro
nary thrombosis refers to the formation of alarge clot in oneof the ar
teries that feed the heart. People who have elevated levels of certain
clotting precursors or depressed levels of clotting inhibitors in their
blood are at high riskof dying from heart attacks. This risk probably
far exceeds that caused by high LDL or low HDL. Some of the blood
factors that enhance clotting include fibrinogen and factor VII. An
other factor, Upoprotein(a), inhibits the destruction of smaU thrombi
before they become large enough to cause a heart attack. AU of these
factors have been found to increase in people with chronicaUy high
blood sugars. Platelets, or thrombocytes, are particles in the blood that
playmajorroles in the blocking of arteries andthe formation of clots.
These have been shown to clump together and stick to arterial walls
much more aggressively in people with high blood sugars. What is ex
citingis that aU of these factors, including sticky platelets, tend to nor
malize aslong-term blood sugars improve.
Diabetics die from heart faUure atarate far exceeding that of people
with normal glucose tolerance. Heart faUure involves a weakening of
the cardiac muscleso that it cannot pump enoughblood. Most long-
term, poorly controlled diabetics have a condition called cardiomy
opathy. In diabetic cardiomyopathy, the muscle tissue of the heart is
slowlyreplaced by scar tissue overa periodof years. This weakens the
muscle so that it eventuaUy "faUs." There is no evidence linking car
diomyopathy with dietary fat intake or serumlipids.
A fifteen-year study of 7,038 French policemen in Paris reported
that "the earUest marker of a higher risk of coronary heart disease
mortality is an elevation of serum insulin level." A study of middle-
aged nondiabetic women at the University of Pittsburgh showed an
increasing risk ofheart disease asserum insulin levelsincreased. Other
448 AppendixA

studies in nondiabetics have shown strong correlations between ele

vated serum insuUn levels and other predictors of cardiac risk such as
hypertension, elevated triglyceride, and lowHDL. The importance of
elevated serum insulin levels (hyperinsulinemia) as a cause of heart
disease and hypertension hastaken on such importancethat a special
symposium on this subject was held at the end of the 1990 annual
meeting of the ADA.A report in a subsequent issue of the journal Di
abetes Care quite appropriately points out that"there are few avaUable
methods of treatingdiabetes that do not resultin systemichyperinsu
linemia [unless the patient is foUowing a low-carbohydrate diet]."
Furthermore, research pubUshed in the journal Diabetes in 1990 demon
strated that elevated serum insulin levels cause excessive leakage of
protein from smaU bloodvessels. This is a common factor in the etiol
ogy of blindness (via macular edema) and kidney disease in diabetics.
Although the AHA and the ADA have been recommending low-
fat, high-carbohydrate diets for diabetics for many decades, no one
had compared the effects on the same patients of low- versus high-
carbohydrate diets untilthe late 1980s. Independent studies performed
in Texas and CaUfornia demonstratedlowerlevels of blood sugarand
improved blood Upids when patients wereput on lower-carbohydrate,
high-fat diets. It was also shown that, on average, for every 1 percent
increase in HgbAlc (the test for average blood sugar over the prior
four months), total serum cholesterol rose 2.2 percent and triglyc
erides increased 8 percent. A long-term study of 7,321 "nondiabetics"
in 2006 showed that for every 1 percent increase in HgbAlc above 4.5
percent, the incidence of coronary artery disease increased 2.5-fold.
The same study also showed that for every 1 percent increase in
HgbAlc above 4.9 percent, mortality increased by 28 percent.
The National Health Examination FoUow-Up Survey, which fol
lowed 4,710 people, reported in 1990 that "in the instance of total
blood cholesterol, we found no evidencein any age-sexgroup of a risk
associated with elevatedvalues." That's right: they found no risk asso
ciated direcdy with elevated total cholesterol. On the same page, this
study lists diabetes as by far the single most important risk factor af
fecting mortality. In males aged 55-64, for example, diabetes wasasso
ciatedwith 60 percent greater mortality than smoking and double the
mortaUty associated with high blood pressure.
The evidence is now simply overwhelmingthat elevatedblood sugar
is the major causeof the high serum lipid levels among diabetics and,
more significantiy, the major factor in the high rates of various heart
What About Dietary Restrictions on Fat, Protein, and Salt? 449

andvascular diseases associated with diabetes. Many diabetics were put

on low-fat diets for so many years, and yet these problems didn't stop.
It is only logical to look to elevated blood sugar and hyperinsulinemia
for the causes of what kills and disables so many of us.
My personal experience with diabetic patientsis very simple.When
we reduce dietary carbohydrate, blood sugars improve dramaticaUy.
After several months of improved blood sugars, we repeat our studies
of Upid profiles and thrombotic risk factors. In the great majority of
cases, I see normalization or improvement of abnormaUties.* This
paraUels what happenedto me more than thirty-fiveyears ago, when I
abandonedthe high-carbohydrate, low-fatdiet that I had been foUow
ing since 1946.
Sometimes, months to years aftera patienthasexperienced normal
or near-normal blood sugars and improvements in the cardiac risk
profile,we wiU seedeteriorationin the resultsof such tests as those for
LDL, HDL,homocysteine,and fibrinogen. AU too often, the patient or
his physician wiU blame our diet. Inevitably, however, we find upon
further testingthat his thyroid activity hasdeclined. Hypothyroidism
is an autoimmune disorder, like type 1 diabetes, and is frequendy in
heritedby diabetics andtheirclose relatives. It canappear years before
or after the development of diabetes and is not caused by high blood
sugars. In fact, hypothyroidism can cause a greater likelihood of ab
normaUties of the cardiac risk profile than canblood sugar elevation.
The treatment of a low thyroid condition is oral replacement of the
deficient hormone(s) — usuaUy one piU daUy. The best screening test
is free T3, tested by tracer dialysis. If this islow, then a fuU thyroid risk
profile should be performed. Correction of the thyroid deficiency in
evitably corrects the abnormaUties of cardiac risk factors that it
caused.

* If your physician finds aU of this hard to beUeve, he or she mightbenefit from

readingthe seventy articles and abstracts on this subject contained in the Pro
ceedings of the Fifteenth International Diabetes Foundation Satellite Sympo
siumon "Diabetes and Macrovascular CompUcations " Diabetes 45,Supplement
3, July 1996. Also worth reading is "Effects of Varying Carbohydrate Content
of Diet in Patients with Non-Insulin Dependent Diabetes MelUtus," by Garg et
al., Inl Amer Med Assoc 1994; 271:1421-1428. Many studies comparing low-
carbohydrate and low-fat diets arepresented each year at meetings of the Metab
olism and Nutrition Society. The low-carbohydrate dietsinvariably haveshown
reduced cardiac risk.
450 AppendixA

WHY IS PROTEIN RESTRICTION

SO COMMON?

About 30 percent of diabetics develop kidney disease (nephropa

thy). Diabetes is the greatest single cause of kidney faUure in the
United States. Earlykidney changes can be found within two to three
years of the onset of high blood sugars. As we discussed briefly in
Chapter 9, the common restrictions on protein intake by diabetic pa
tients derive from fear regarding this problem, and ignorance of the
actual causes of diabetic kidney disease.
Bylooking at how the kidney functions,one can better understand
the relative roles of glucose and protein in the kidney faUure of dia
betes. The kidney filters wastes, glucose, drugs, and other potentiaUy
toxic materials from the blood and deposits them into the urine. It is
the urine-making organ. A normal kidney contains about 1 miUion
microscopic blood filters, caUed glomeruli. FigureA-1 iUustrates how
blood enters a glomerulus through a tiny artery caUed the incoming
arteriole. The arteriole feeds a bundle or tuft of tiny vessels caUed cap-
Ularies. The capUlaries contain tiny holesor pores that carry a negative
electrical charge. The downstream ends of the capUlaries merge into

Glomerulus
High Pressure
Incoming Arteriole Outgoing Arteriole
Blood In Blood Out
(Low Pressure)

CapillaryTuft

Capsule

•Mesangium

•Pore

Negative
Electrical
Charge

Terry Eppndge
Fig.A-1. The microscopicfiltration
unitof the kidneys.
WhatAboutDietaryRestrictions on Fat, Protein, andSalt? 451

an outgoing arteriole, which is narrower than the incoming arteriole.

This narrowing results in high fluid pressure when blood flows
through the capUlary tuft. The high pressure forces some of the water
in the blood through the pores of the capUlaries. This water dribbles
into the capsule surrounding the capUlary tuft. The capsule, acting like
a funnel, empties the waterinto a pipelike structurecaUed the tubule.
The pores of the capUlaries are of such a size that smaU molecules in
the blood, such as glucose and urea, can pass through with the water
to form urine. In a normal kidney, large molecules, such as proteins,
cannot readUy get through the pores. Since most blood proteins carry
negative electrical charges, even the smaUer proteinsin the blood can
not easUy get through the pores, because they are repeUed by the neg
ativecharge on each pore.
The glomerular filtration rate (GFR) is a measure of how much fil
tering the kidneys perform in a given period of time. Many diabetics
with frequent high blood sugars and normalkidneys wiU initiaUy have
an excessively high GFR. This is in part because blood glucose draws
waterinto the bloodstream from the surroundingtissues, thus increas
ing bloodvolume,blood pressure, andblood flow through the kidneys.
A GFR that is one-and-a-half to two times normal is not uncommon in
diabetics with high blood sugars priorto the onset of permanent injury
to their kidneys.These peoplemay typicaUy have asmuch glucose in a
24-hour urine coUection as the weight of 5 to 50 packets of sugar. Ac
cordingto an ItaUan study,an increase in blood sugarfrom 80 mg/dl to
272mg/dl resulted in an average GFR increase of40 percent even in di
abetics with kidneys that were not fuUy functional. Before we knew
about glycosylation of proteins and the other toxic effects of glucose
upon blood vessels, it was speculated that the cause of diabetic kidney
disease (nephropathy) wasthis excessive filtration (hyperfiltration).
The metabolism of dietary protein produces waste products such as
urea and ammonium, which contain nitrogen.* It therefore had been
speculated that in order to clear these wastes from the blood, people
eatinglarge amounts of protein would have elevated GFRs. As a result,
diabetics have been urged to reduce their protein intake to low levels.
Studies by an IsraeU group, however, of nondiabetic people on high-
protein (meat-eating) and very low protein (vegetarian) diets, disclosed

* A 1995 article in the journal Nutritional Biochemistry, 6:411-437, demon

strated that a higher protein diet enables the kidneys to increase their capacity
for net acid secretion as ammonium.
452 Appendix A

no differencein GFRs. Furthermore, overmany years on these diets,kid

ney function was unchanged between the two groups. A report from
Denmarkdescribed astudyin whichtype 1diabetics without discernible
kidney disease were put on protein-restricted diets, and experienced a
verysmaU reduction in GFR andno change in other measures of kidney
function. As long ago as 1984, a study appeared in the journal Diabetic
Nephropathy demonstrating thatelevated GFR isneither anecessary nor
a sufficient condition forthe development of diabetic kidney disease.
This evidence would suggest that the currendy prevaUing admoni
tion to aU diabetics to reduce protein intake is unjustified.
Studies on diabetic rats haveshown the foUowing: Ratswith blood
sugars maintainedat 250mg/dl rapidly develop diabetic nephropathy
(kidney disease). If their dietary protein is increased, kidney destruc
tion accelerates. At the samelaboratory, diabetic ratswith blood sugars
maintained at 100 mg/dl Uve fuU Uves and neverdevelopnephropathy,
no matter how much protein they consume. Diabetic rats with high
blood sugars and significant nephropathy have shown total reversal of
their kidney disease after blood sugars were normalized for several
months.
Other studies have enabledresearchers to piecetogether a scenario
for the causes of diabetic nephropathy, where glycosylation of pro
teins, abnormal clotting factors, abnormalplatelets, antibodiesto gly
cosylated proteins, and so on, join together to injure glomerular
capUlaries. Early injury may only cause reduction of electrical charge
on the pores. As a result, negatively charged proteins such as albumin
leak through the pores and appear in the urine. Glycosylated proteins
leak through pores much earUer than normal proteins. High blood
pressure, and especiaUy high serum insuUn levels, can increase GFR
and force even more proteinto leakthrough the pores. If some of these
proteinsare glycosylated or glycated, they wiU stick to the mesangium,
the tissue between the capUlaries. Examination of diabetic glomeruU
indeed discloses large deposits of glycated proteins and antibodies to
glycated proteins in capUlary waUs and mesangium. As these deposits
increase, the mesangium compresses the capUlaries, causing pressure
in the capUlaries to increase (enlarging the pores) and larger proteins
to leak from the pores. This leads to more thickening of the mesan
gium, more compression of the capUlaries, and acceleration of de
struction. EventuaUy the mesangium and capUlaries become a mass of
scar tissue. Independentiyof this,both high blood sugars and glycated
proteins cause mesangial ceUs to produce type IV coUagen, a fibrous
What About Dietary Restrictions onFat, Protein, andSalt? 453

material that further increases their bulk. Increase in mesangial vol

ume hasbeen found to be commonplace in poorlycontroUed diabetes
even before albumin or other proteins appear in the urine.
Many studies performed on humans show that when blood sugars
improve, GFR improves and less protein leaks into the urine. When
blood sugars remain high, however, there is further deterioration.
There is a point of no return, where a glomerulus has been so injured
that no amount of blood sugar improvement can revive it. Although
this seemsto be true for humans,blood sugar normaUzation hasactu-
aUy brought about the appearance of new glomeruU in rats.
Nowadays many diabetics who have lost aU kidney function are
treated by artificial kidneys (dialysis machines) that remove nitroge
nous wastes from the blood. In order to reduce the weekly number of
dialysis treatments, which are cosdy and unpleasant, patients are se
verely restrictedin their consumption of both water and dietary pro
tein. Instead ofusing large amounts of carbohydrate to replace the lost
calories, many dialysis centers now recommend oUve oU to their dia
betics.OliveoU is high in monounsaturated fats, which arebeUeved by
some to lower the risk of heart disease.
Becausethe survival rate of diabetics on dialysis is so much lower
than that of nondiabetics, some dialysis centers are now using low-
carbohydrate, high-protein diets for their diabetic patients.
In summary: Diabetic nephropathy does not appear if bloodsugar is
kept normal. Dietary protein does not causediabeticnephropathy, but
can possibly (stiU uncertain) slightly accelerate the process once there
hasbeen major, irreversible kidney damage. Dietary proteinhasno sub
stantial effect upon the GFR of healthy kidneys, certainlynot in com
parison to the GFR increase caused by elevated blood sugar levels.*

*Your physician might find informative the following articles on this subject:
"Molecular and Physiological Aspects of Nephropathy in Type 1 Diabetes MeUi-
tus," by Raskin and Tamborlane, Jnl Diabetes and Its Complications, 1996,
10:31-37; "The Effects of Dietary Protein Restrictionand Blood PressureControl
on the Progression of Chronic Renal Disease," byS.Klahr et al.,New England Jnl
Med, 1994,330:877-884; also,in the same issueof NewEngland Jnl Med, the ed
itorial "The Role of Dietary Protein Restriction in Progressive Azotemia" (pp.
929-930). Another study, in the journal Diabetes Care, 25:425-430, in the year
2000, showed that obese people on a high-protein diet lost more fat and less
muscle mass than those on a low-fat diet. They also showed more than double
the reduction in LDL (the "bad" cholesterol).