Joffe Dyslip DM Sa

OBSERVATIONS
Remitting Diabetes
A new genetic subgroup?
W
e describe a previously unre-
ported clinical scenario of remit-
ting diabetes in two young
brothers who do not t existing diagnos-
tic classications. They may represent a
new genetic subgroup of diabetes.
Case 1 presented at 3.6 years with
poor linear growth, polyuria, and poly-
dipsia. Diabetes was diagnosed based on a
blood glucose value of 11.5 mmol/l and
an elevated HbA
1c
of 7.0% (normal 3.8
6.0). During 3 months observation, hy-
perglycemia and an elevated HbA
1c
persisted. Working diagnosis was very
early type 1 diabetes, and insulin was
commenced (0.2 units kg
1
day
1
).
HbA
1c
improved from 7.0 to 5.3%. How-
ever, linear growth did not improve.
The family also tested asymptomatic
siblings and identied hyperglycemia in
the 18-month-old brother. Diabetes was
diagnosed on repeated blood glucose val-
ues 11 mmol/l and HbA
1c
9.0%. Al-
though thriving, there was concern of
early type 1 diabetes, and he commenced
insulin (0.2 units kg
1
day
1
). HbA
1c
normalized to 5.5%after 16 months. Two
additional family members had glucose
abnormalities: the 44-year-old father had
impaired glucose tolerance (IGT) (glu-
cose 6.4 mmol/l [0 min] and 10.4 mmol/l
[120 min] in an oral glucose tolerance test
[OGTT]; BMI 29 kg/m
2
), and the 74-year-
old paternal grandmother was diagnosed
with type 2 diabetes at age 60 years and is
on metformin (not overweight, no diabe-
tes complications). The mothers OGTT
was normal.
No evidence of autoimmunity was
found in either child (insulin, islet cell,
and GAD antibodies). After age 1.6 and
2.3 years, respectively, insulin was
ceased, as requirements had remained
low with normal HbA
1c
and blood glu-
cose. Both boys had OGTTs showing nor-
mal glucose tolerance. Interestingly, the
older boy had hypoglycemia (glucose 2.0
mmol/l) at 120 min of OGTT, suggesting
possible insulin secretion dysregulation.
After 2 years off insulin, HbA
1c
has re-
mained normal (5.45.5%).
The cause of diabetes resolution in
these boys remains unexplained. Tran-
sient hyperglycemia can occur during in-
tercurrent illness, is of very short
duration, and is not associated with ele-
vated HbA
1c
. Although type 1 diabetes
was the initial diagnosis, their subsequent
clinical course and absence of autoimmu-
nity markers make this unlikely. Type 1
diabetes may have an extended honey-
moon (i.e., partial remission), sometimes
up to 2 years, but normal HbA
1c
off treat-
ment 4 years after diagnosis is very un-
usual. IGT in the father and type 2
diabetes in the paternal grandmother is
consistent with autosomal-dominant in-
heritance suggesting maturity-onset dia-
betes of the young (MODY); however, no
MODY subgroups remit (1). While glu-
cokinase mutations could explain the
adults hyperglycemia, neither child had
fasting blood glucose (5 mmol/l) effec-
tively excluding MODY2. Case 1 tested
negative for hepatocyte nuclear factor-
(HNF-1) mutations. Transient neonatal
diabetes remits but is excluded as they
presented after age 3 months (1.5 and 3.6
years). A remitting formof atypical diabe-
tes is described in black adolescent Amer-
icans (2) but not in whites or young
children.
In summary, disappearance of diabe-
tes in these young boys is unusual and
does not t clinically recognized syn-
dromes. Two affected siblings suggest a
novel genetic syndrome probably altering
-cell function. This could be a recessive
condition with coincidental hyperglyce-
mia in adults. Alternatively, it may repre-
sent different stages in a dominant
disorder, with adults having undetected
hyperglycemia during childhood, sug-
gesting that the children may later re-
lapse. This cyclical pattern of diabetes
remission and relapse occurs in transient
neonatal diabetes, and we hypothesize a
novel genetic mutation causing a similar
process. We would welcome reports of
further cases of remitting diabetes, as they
could provide further insights into this
potential new genetic form of diabetes.
CHRISTINE P. BURREN, MBBS, MD,
FRACP, FRCPCH
1
ANDREW T. HATTERSLEY, DM, FRCP
2
From the
1
National Health Service Trust, Epsom
and St. Helier University Hospitals, Queen Marys
Hospital for Children, Epsomand St. Helier Univer-
sity Hospitals National Health Service Trust, Car-
shalton, Surrey, U.K.; and the
2
Department of
Diabetes, School of Postgraduate Medicine, Royal
Devon and Exeter, National Health Service Health-
care Trust, Exeter, U.K.
Address correspondence to Dr. Christine P. Bur-
ren, Queen Marys Hospital for Children, St. Helier
Hospital, Wrythe Lane, Carshalton, Surrey SM5
1AA, U. K. E- mai l : chr i s t i ne@cpbur r en.
freeserve.co.uk.
2004 by the American Diabetes Association.

References
1. Stride A, Hattersley AT: Different genes,
different diabetes. Annal Med 34:207
216, 2002
2. Winter WE, MaclarenNK, RileyWJ, Clarke
DW, Kappy MS, Spillar RP: Maturity-
onset diabetes of youth in black Ameri-
cans. N Engl J Med 316:285291, 1987
Seroconversion of
GAD Antibody in
Unclassied
Diabetes With Long
Duration of Disease
L
atent autoimmune diabetes in adults
(LADA) (1), or slowly progressive in-
sulin-dependent diabetes mellitus
(SPIDDM) (2), is a subtype of type 1 dia-
betes with a slowly progressive course.
LADA, or SPIDDM, is diagnosed by the
detection of islet-associated autoantibod-
ies such as islet cell antibody (ICA) or
GAD antibody (GADA) in the serum;
moreover, patients with LADAare usually
originally diagnosed as having type 2 di-
abetes. If islet-associated autoantibodies
are not detected in the serum, these pa-
tients who are originally diagnosed as
having type 2 diabetes cannot be diag-
nosed as having LADAand are followed as
unclassied diabetes at present.
In cl i ni cal si tuati ons, i f i sl et-
associated autoantibodies are not initially
detected in the serum, these cases are usu-
ally followed as type 2 diabetes or un-
classied diabetes without reevaluation
of autoantibodies because it is unknown
whether islet-associated autoantibodies
will appear later in the disease course. It
has been reported that islet-associated au-
toantibodies are detected within 1 year
after onset in 15% of cases of classical
type 1 diabetes without islet-associated
autoantibodies at the onset of disease (3).
L E T T E R S
1836 DIABETES CARE, VOLUME 27, NUMBER 7, JULY 2004
The following case patient, who was
conrmed as not having GADAs at 31
years after the onset of diabetes and was
followed as having unclassied diabe-
tes, is a rare case in whom GADA was
detected at 36 years.
The patient was diagnosed as having
type 2 diabetes at age 47 years, had an
HbA
1c
level of 7%, and was being
treated with a sulfonylurea (gliben-
clamide). However, glycemic control sub-
sequently worsened despite being treated
with 8.75 mg/day of glibenclamide, 150
mg/day of buformin, and 0.9 mg/day of
voglibose, and she was admitted to the
hospital at age 78. On admission, her
height was 155 cm and her body weight
was 38.8 kg (BMI 16.1 kg/m
2
), with no
history of obesity. According to labora-
tory ndings, her fasting plasma glucose
level was 338 mg/dl and her HbA
1c
level
was 10.1%. Her insulin secretion was also
low (serum C-peptide level 0.3 ng/ml on
fasting, 0.9 ng/ml at 2 h after breakfast,
and 24-h urine C-peptide level 10.5 g/
day), thus requiring insulin therapy (total
16 units/day at discharge). Based on these
ndings, it was possible that she had
SPIDDM; however, GADA was negative
(detection limit 0.4 units/ml; 100%
sensitivity and 100% specicity of the as-
say in the GADA prociency test [Immu-
nology of Diabetes Workshop], lab IDno.
305), resulting in the diagnosis of un-
classied diabetes.
At age 83, she was again admitted to
the hospital because of acute myocardial
infarction. On the second admission, her
body weight was 39.4 kg (BMI 16.4 kg/
m
2
). Laboratory results indicated that her
fasting plasma glucose level was 198 mg/
dl, her HbA
1c
level 7.2%, and her 24-h
urine C-peptide level 12.5 g/day. Sur-
prisingly, GADA, which was negative at
the time of her rst admission, was now
positive (144 units/ml), although insuli-
noma-associated protein 2 antibody was
negative. Furthermore, HLA typing de-
tected DR4, which is considered to be a
susceptible HLA type for type 1 diabetes
(other HLA types: A24, A26, B35, B55,
DR8). Based on these observations, she
was diagnosed as having SPIDDM.
The frequency of seroconversion of
GADA in diabetic patients who were orig-
inally diagnosed as having type 2 diabetes
is not known, and a more extensive large-
scale study is needed to clarify the fre-
quency of seroconversion in this type of
diabetes. Based on this case, however, we
would like to emphasize that it is essential
to measure islet-associated autoantibod-
ies such as GADA periodically for the pre-
cise diagnosis of diabetes, especially in
patients given the diagnosis of unclassied
diabetes, even if the patient has suffered
from diabetes for a long period of time.
TOSHIKATSU SHIGIHARA, MD
1
AKIRA SHIMADA, MD, PHD
1
YOSHIHUMI SAISHO, MD
1
TOSHIHIDE KAWAI, MD, PHD
1
HIROSHI MARUYAMA, MD, PHD
2
TAKAO SARUTA, MD, PHD
1
From the
1
Department of Internal Medicine, Keio
University School of Medicine, Tokyo, Japan; and
2
Sano-Kousei General Hospital, Tochigi, Japan.
Address correspondence to Akira Shimada, MD,
PhD, Department of Internal Medicine, Keio Univer-
sity School of Medicine, 35 Shinanomachi, Shin-
juku-ku, Tokyo 160-8582, Japan. E-mail: asmd@
sc.itc.keio.ac.jp.

References
1. Zimmet PZ, Toumi T, Mackay R, Rowley
MJ, Knowles W, Cohen M, Lang DA: La-
tent autoimmune diabetes mellitus in
adults (LADA): the role of antibodies to
glutamic acid decarboxylase in diagnosis
and prediction of insulin dependency.
Diabet Med 11:299303, 1994
2. Kobayashi T, Itoh T, Kosaka K, Sato K,
Tsuji K: Time course of islet cell antibod-
ies and -cell function in noninsulin-de-
pendent stage of type 1 diabetes. Diabetes
36:510517, 1987
3. Landin-Olsson M, Arnqvist H, Blohme G,
Littorin B, Lithner F, Nystrom L, Schersten
B, Sundkvist G, Wibell L, Ostman J, Lern-
mark A: Appearance of islet cell
autoantibodies after clinical diagnosis of
diabetes mellitus. Autoimmunity 29:57
63, 1999
Normal -Cell
Function in
PostLiver
Transplantation
Diabetes Treated
With Tacrolimus
U
se of immunosuppressive agents is
mandatory after organ transplanta-
tions but may be complicated by
the development of hyperglycemia or di-
abetes. The prevalence of diabetes after
liver transplantation in adulthood ranges
from13.6 to 33%(1,2) with a progressive
increase in risk that parallels the time after
transplant (3). Tacrolimus, a calcineurin
inhibitor, is a recent immunosuppressive
agent largely used to prevent and treat
transplant rejections often following
other immunosuppressive failures (45).
The incidence of post-transplant diabetes
mellitus (PTDM) in tacrolimus-treated
patients was signicantly higher than in
those treated with cyclosporine (6,7) and
in pediatric-age patients (8). Usually
PTDM refers to renal recipients, and only
one pediatric case has been reported after
liver transplantation (9). PTDMin tacroli-
mus-treated patients has been related to
reduced pancreatic -cell function and is
generally reversed by dose reduction
(1011).
Here we describe the -cell function
in a 15-year-old male liver transplant re-
cipient treated with tacrolimus at the on-
set of PTDM and 1 year after remission.
He underwent liver transplantation at age
7 years for Alagille Syndrome and was
subsequently treated with steroids and
cyclosporine. Because of acute rejection at
age 15, treatment with cyclosporine was
switched to tacrolimus and mycopheno-
late mofetil. One month later, he pre-
sented symptomatic hyperglycemia (42
mmol/l) without ketoacidosis; his BMI
was 21.3 kg/m
2
, and his HbA
1c
was 9.4%
(normal values 3.36.0). Initially, the pa-
tient required 1.8 units kg
1
day
1
of
insulin. -Cell function was investigated
by glucagon stimulationbasal C-
peptide levels were 0.90 nmol/l (normal
values 0.1650.993) and 1.67 nmol/l af-
ter 6 min (relative increase 186%, normal
values 130377%). Tacrolimus was sub-
stituted by cyclosporine while continuing
other immunosuppressive agents. This al-
lowed a gradual decrease of insulin and its
withdrawal within 5 weeks. One year
later, an oral glucose tolerance test
showed normal glucose tolerance (basal
levels 4.5 mmol/l, peak 7.7). In addition,
stimulated C-peptide response was nor-
mal (relative increase 231%), fasting in-
sulin level was 54.6 pmol/l, and HbA
1c
was 5.6%.
The absence of ketoacidosis and the
presence of normal C-peptide levels dur-
ing tacrolimus treatment indicate that
-cell function was normal in our patient,
not conrming the -cell impairment as-
cribed to tacrolimus (9,10). The high in-
sulin dose required suggests that insulin
resistance, which was not due to steroids
Letters
DIABETES CARE, VOLUME 27, NUMBER 7, JULY 2004 1837
since they were never withdrawn from
therapy, may have played a role. Further-
more, because proneness to diabetes de-
pends on several genetic mechanisms, it is
possible that immunosuppressive agents
play only a triggering role in PTDM. In
addition, since most PTDMcases were de-
scribed after renal transplantations, we
cannot exclude a putative role of the
transplanted liver itself on the abnormal
peripheral insulin action. With the
spreading use of tacrolimus in liver trans-
plantation, pediatric diabetes practitio-
ners will probably face many newcases of
PTDM.
NICOLINA DI COSMO, MD
1
PIETRO VAJRO, MD
1
DOMINIQUE DEBRAY, MD
2
GIULIANA VALERIO, MD, PHD
3
MICHELA GIUGLIANO, MD
1
PIETRO BUONO, MD, PHD
1
ADRIANA FRANZESE, MD
1
From the
1
Department of Pediatrics, Federico II
University, Naples, Italy; the
2
Paediatric Hepatology
Unit, University Hospital of Bicentre, Le Kremlin
Bicetre, France; and the
3
Faculty of Movement Sci-
ences, Parthenope University, Naples, Italy.
Address correspondence to Adriana Franzese,
MD, Department of Pediatrics, via S. Pansini 5,
80131 Naples, Italy. E-mail: franzese@unina.it.

References
1. Tabasco-Minguillan J, Mieles L, Carroll P,
Gavaler J, VanThieldDH, Starzl TE: Long-
term insulin requirement after liver trans-
plantation with FK 506 in American vet-
erans. Transplant Proc 25:677678, 1993
2. Margarit C, Rimola A, Gonzalez-Pinto I,
Cuervas-Mons V, Edo A, Andreu H: Ef-
cacy and safety of oral low-dose tacroli-
mus treatment in liver transplantation.
Transplant Int 11:S260S266, 1998
3. Greenspan LC, Gitelman SE, Leung MA,
Glidden DV, Mathias RS: Increased inci-
dence in post-transplant diabetes mellitus
in children: a case-control analysis. Pedi-
atr Nephrol 17:15, 2002
4. Scott LJ, McKeage K, Keam SJ, Plosker
GL: Tacrolimus: a further update of its use
in the management of organ transplanta-
tion. Drugs 63:12471297, 2003
5. Debray D, Furlan V, Baudouin V, Houyel
L, Lacaille F, Chardot C: Therapy for acute
rejection in pediatric organ transplant re-
cipients. Paediatr Drugs 5:8193, 2003
6. The US Multicenter FK 506 Liver Study
Group: A comparison of tacrolimus (FK
506) and cyclosporine for immunosup-
pression in liver transplantation. N Engl
J Med 331:11101115, 1994
7. Cai TH, Esterl RM Jr, Nichols L, Cigarroa
F, Speeg KV, Halff GA: Improved immu-
nosuppression with combination tacroli-
mus (FK506) and mycophenolic acid in
orthotopic liver transplantation. Trans-
plant Proc 30:14131414, 1998
8. Reisaeter AV, Hartmann A: Risk factors
and incidence of posttransplant diabetes
mellitus. Transplant Proc 33:8S18S, 2001
9. Keshavarz R, Mousavi MA, Hassani C: Di-
abetic ketoacidosis in a child on FK506
immunosuppression after a liver trans-
plant. Pediatr Emerg Care 18:2224, 2002
10. Al-Uzri A, Stablein DM, Cohn RA: Post-
transplant diabetes mellitus in pediatric
renal transplant recipients: a report of the
North American Pediatric Renal Trans-
plant Cooperative Study (NAPRTCS).
Transplantation 72:10201024, 2001
11. Filler G, Neuschulz I, Vollmer I, Amendt
P, Hocher B: Tacrolimus reversibly re-
duces insulin secretion in paediatric renal
transplant recipients. Nephrol Dial Trans-
plant 15:867871, 2000
Denition of
Microalbuminuria in
First-Morning and
Random Spot Urine
in Diabetic Patients
M
easurement of albumin excretion
in a 24-h urine collection has long
been the gold standard for
quantitative evaluation of albuminuria in
diabetic patients; however, collection er-
rors due to improper timing and missed
samples may lead to signicant over- and
underestimation of albuminuria. For con-
venience and consistency, the American
Diabetes Association (1) and the National
Kidney Foundation (2) have recently rec-
ommended measurement of albumin-to-
creatinine ratio (ACR) in a random spot
urine collection for diagnosis of mi-
croalbuminuria. Microalbuminuria is di-
agnosed if ACR ranges between 30 and
300 mg/g creatinine. The guidelines rec-
ommended using a rst-morning sample
because of the potentially higher correla-
tion with 24-h albumin excretion, but a
random sample is considered acceptable
if a rst-morning specimen is not avail-
able. Measurement of ACR using a rst-
morning or random urine sample may
differ signicantly, as exercise stress, di-
urnal variation, and other factors may af-
fect urinary albumin excretion. We
provide data from a cross-sectional study
evaluating potential differences in ACR
obtained from rst-morning and random
spot urines collected on the same day.
A total of 717 adult diabetic patients
with and without nephropathy were re-
cruited from the outpatient clinic of the
Diabetes Center, Tokyo Womens Medi-
cal University Hospital, Tokyo, Japan. Pa-
tients were instructed to bring a rst-
morning urine specimen to the clinic and
then provide a random urine specimen
immediately upon arriving at the clinic on
the same day. ACR was calculated fromuri-
nary albuminandcreatinine concentrations
determined using radioimmunoassay and
Jaffes method, respectively. Paired samples
with a random urinary ACR of more than
1,000 mg/g were excluded to provide a
more accurate range for estimating the rela-
tionship between measurements.
Paired samples were analyzed from
668 patients (289 women and 379 men,
mean age 58 12 years, 95%with type 2
diabetes). The majority (75%) of random
spot urine was collected during a morning
visit (8:30 A.M. to 12:00 P.M.). There was a
strong relationship between ACRs mea-
sured from rst-morning and spot urine
samples, yielding a linear correlation on a
logarithmic scale: log
10
ACR (rst-
morning sample) 0.8589 log
10
ACR
(random spot sample) 0.0604 (r
0.871). Applying this equation, ACR val-
ues of 30300 mg/g in a rst-morning
urine specimen would correspond to val-
ues of 51391 mg/g in random spot col-
lection. Using the ACR cutoff value of
30300 mg/g, 135 patients (20%) would
receive a diagnosis of microalbuminuria
based on rst-morning urine, whereas 234
patients (35%) would receive this diagno-
sis based on a random spot collection.
This rather large discrepancy between
ACR fromrst-morning and randomspot
urine may have important implications in
the diagnosis of microalbuminuria in di-
abetes. We advocate strict adherence to
the use of rst-morning urine or possibly
an upward adjustment of the range for
diagnosis of microalbuminuria using ACR
from random spot urine. Further analysis
of the relationship of ACR in rst-
morning and random spot urines to 24-h
urinary albumin excretion, as well as their
value in predicting future development of
clinical proteinuria, is required.
TETSUYA BABAZONO, MD, PHD
1
CHIEKO TAKAHASHI, MD, PHD
1,2
YASUHIKO IWAMOTO, MD, PHD
3
Letters
From the
1
Division of Nephrology and Hyperten-
sion, Diabetes Center, Tokyo Womens Medical
University School of Medicine, Tokyo, Japan; the
2
Department of Food Science and Human Nutri-
tion, Tokyo Metropolitan College, Tokyo, Japan;
and the
3
Department of Medicine, Diabetes Center,
Tokyo Womens Medical University School of Med-
icine, Tokyo, Japan.
Address correspondence to Tetsuya Babazono,
MD, PhD, Division of Nephrology and Hyperten-
sion, Diabetes Center, Tokyo Womens Medical
University School of Medicine, 8-1 Kawadacho,
Shinjukuku, Tokyo 162-8666, Japan. E-mail:
babazono@dmc.twmu.ac.jp.

References
1. American Diabetes Association: Nephro-
pathy in diabetes (Position Statement).
Diabetes Care 27 (Suppl. 1):S79S83, 2004
2. Keane WF, Eknoyan G: Proteinuria, al-
buminuria, risk, assessment, detection,
elimination (PARADE): a position paper
of the national kidney foundation. Am J
Kidney Dis 33:10041010, 1999
Lipid Levels in Black
South Africans With
Type 2 Diabetes
Dispelling misconceptions
D
yslipidemia in type 2 diabetes typi-
cally comprises hypertriglyceride-
mia and reduced HDL cholesterol,
usually associated with hypercholesterol-
emia (1). Relatively little is known about
its occurrence in developing African com-
munities, where the prevalence of type 2
diabetes may escalate dramatically (2).
This is important because dyslipidemia
constitutes a major risk factor for coro-
nary artery disease (CAD) in type 2 diabe-
tes (3). We therefore assessed serum
lipids of South African blacks with type 2
diabetes of differing socioeconomic status
and compared levels with those of pa-
tients two decades earlier. Our initial ex-
pectations were that dyslipidemia would
be substantially greater in the higher so-
cioeconomic group due to afuence and
that lipids would have risen considerably
over time because of increasing urbaniza-
tion.
Our low socioeconomic community
comprised 445 black African patients
(241 women and 204 men) with recently
diagnosed type 2 diabetes studied be-
tween 1994 and 1996. They attended the
diabetes clinic at Johannesburg Hospital
and were mostly domestic workers, labor-
ers, and pensioners who were overweight
and not undergoing lipid-lowering ther-
apy. The higher socioeconomic cohort
consisted of 82 patients (32 women and
50 men), also with recently diagnosed
type 2 diabetes studied during 2001
2003. They attended a private diabetes
clinic in the same city and were mainly
civil servants, clerks, and executives.
Again, the majority was overweight and
not receiving hypolipidemic drugs. Final-
ly, we reevaluated lipid data from 47 Af-
rican patients (mainly women, mean age
55 years) with type 2 diabetes who at-
tended Johannesburg Hospital complex
in 1976; they were predominantly do-
mestic workers.
Venous blood was collected at
0800, and although not formally fast-
ing, most subjects had not eaten over-
night. Serum was analyzed for total
cholesterol, HDL cholesterol, and triglyc-
erides by automated enzymatic methods
(4). LDL cholesterol was calculated using
the Friedewald formula. (These differed
from the 1976 automated techniques
[5].) Long-term diabetes control was as-
sessed by HbA
1c
concentrations. Statisti-
cal analysis of data between patient
groups utilized the unpaired t test, with
P 0.05 after Bonferonni adjustment be-
ing signicant.
Biochemical data (Table 1) showed
unimpressive dyslipidemia in women
with no signicant differences, although
total and LDL cholesterol tended to be
lower and triglycerides higher in the
higher socioeconomic cohort. For men,
similar trends emerged (P 0.05 for trig-
lycerides). In both sexes, diabetes control
was signicantly worse in the Johannes-
burg Hospital setting.
Comparing mean total cholesterol
and triglyceride concentrations in African
diabetic patients (three-quarters of them
women) from the 1976 survey with Afri-
can diabetic women attending the same
clinic two decades later, serum choles-
terol had risen from 4.8 1.1 to 5.3
1.3 mmol/l (P 0.01). Serum triglycer-
ides, however, had not changed signi-
cantly (1.2 0.5 to 1.5 1.1 mmol/l).
Dyslipidemia in the higher socioeco-
nomic cohort of South African blacks with
type 2 diabetes was unimpressive and not
substantially greater than in their less so-
phisticated counterparts, apart from mild
hypertriglyceridemia; importantly, total
and LDL cholesterol tended to be lower.
This was our rst misconception to be cor-
rected and is explainable by two factors:
tighter metabolic control and the likeli-
Table 1Demographic, anthropometric, and biochemical characteristics of black South
African patients with type 2 diabetes of differing socioeconomic status
Johannesburg hospital
19941996
(Low socioeconomic
community)
Private diabetes clinic
20012003
(Higher socioeconomic
cohort)
Women
n 241 32
Age (years) 53.6 5.4 53.9 10.8
BMI (kg/m
2
) 31.7 6.6 30.1 5.9
Total cholesterol (mmol/l) 5.3 1.3 4.9 1.0
LDL cholesterol (mmol/l) 3.3 1.4 3.1 1.0
HDL cholesterol (mmol/l) 1.3 0.4 1.4 0.5
Triglycerides (mmol/l) 1.5 1.1 1.9 2.4
HbA
1c
(%) 9.6 3.2* 7.9 1.5
Men
n 204 50
Age (years) 52.6 4.9* 49.5 9.0
BMI (kg/m
2
) 27.6 4.9 28.3 4.6
Total cholesterol (mmol/l) 4.9 1.3 4.6 1.2
LDL cholesterol (mmol/l) 3.0 1.1 2.8 1.0
HDL cholesterol (mmol/l) 1.2 0.4 1.2 0.7
Triglycerides (mmol/l) 1.6 1.2* 2.2 1.2
HbA
1c
(%) 9.3 3.3* 8.5 2.1
Data are means SD. *P 0.05 compared with private diabetes clinic.
Letters
hood that they had more affordable access
to a prudent diet containing reduced sat-
urated fat and rened carbohydrates.
Their higher triglyceride levels, sig-
nicant in men despite better glycemia,
could reect greater insulin resistance in
these subjects who likely were less phys-
ically active than the less afuent subjects.
By contrast total cholesterol levels may
have risen signicantly, if not dramati-
cally, over two decades, reecting a rela-
tively small impact of westernization.
Our protocol revealed methodologi-
cal limitations: the automated techniques
for lipid assays differed in the 1976 study,
the nonfasting collection of some serum
samples may have inuenced triglyceride
concentrations, and the absence of formal
dietary histories limited our interpreta-
tion of the lipid data. Nevertheless, these
ndings have implications for evolving
CAD in Africans with type 2 diabetes.
Their CAD prevalence remains low but
may be increasing among urban dwellers
(6). If subsequently conrmed, this sug-
gests that prolonged exposure to only
modest dyslipidemia, particularly when
combined with other risk factors such as
hypertension, smoking, and hypercoagu-
lability, is the key atherogenic stimulus.
BARRY I. JOFFE, DSC, FRCP
1
LARRY A. DISTILLER, BSC, FCP(SA), FACE
1
W. JOHN KALK, MB, FRCP
2
From the
1
Centre for Diabetes and Endocrinology,
Johannesburg, South Africa; and the
2
Derbyshire
Royal Inrmary, Derbyshire, U.K.
Address correspondence to Prof. Barry Joffe, Cen-
tre for Diabetes and Endocrinology, 81 Central St.,
Houghton 2198, South Africa. E-mail: barry@
cdecentr.co.za.
W.J.K. was formerly the head of the Diabetic
Clinic, Johannesburg Hospital, Johannesburg,
South Africa.
Acknowledgments We thank Greg Dis-
tiller for statistical help, Dr. Vanessa Panz for
initial data processing, and Terry Neweld for
secretarial assistance.

References
1. Durrington P: Dyslipidaemia. Lancet 362:
717731, 2003
2. Amos AF, McCarty DJ, Zimmet P: The ris-
ing global burden of diabetes and its com-
plications: estimates and projections to
the year 2010. Diabet Med 14 (Suppl. 5):
5784, 1997
3. Laakso M: Hyperglycaemia as a cardiovas-
cular risk factor. International Diabetes
Monitor 12:15, 2000
4. Deeg R, Ziegenhorn J: Kinetic enzymatic
method for automated determination of
total cholesterol in men. Clin Chem 29:
17981802, 1983
5. Anonymous: Simultaneous measure-
ments of cholesterol and triglycerides:
Auto Analyzer Manual AA11. Tarrytown,
NY, Technicon, 1970
6. Joubert J, McClean CA, Reid CM, Davel D,
Pilloy W, Delport R, Steyn L, Walker ARP:
Ischaemic heart disease in black South Af-
rican stroke patients. Stroke 31:1294
1305, 2000
Diabetes Is Still a
Risk Factor for
Restenosis After
Drug-Eluting Stent in
Coronary Arteries
W
e read with interest the meta-
analysis of the effect of diabetes
on restenosis rates among pa-
tients receiving coronary angioplasty
stenting (1). This meta-analysis of six
clinical trials showed that the odds ratio
(OR) of coronary artery restenosis associ-
ated with diabetes was 1.61 (95% CI
1.212.14, P 0.004) in univariate logis-
tic regression models, but it decreased to
1.30 (0.991.70, P 0.055) after age
was controlled in multivariate models. All
trials selected in this meta-analysis used
classical bare-metal stents. However,
drug-eluting stents (DESs) have been
shown to improve outcomes among pa-
tients undergoing percutaneous coronary
intervention by signicantly reducing re-
stenosis rates (2).
We performed a meta-analysis of the
results from four recently published trials
comparing bare-metal stents with DESs,
two using sirolimus (RAVEL and SIRIUS)
(3,4) and two using paclitaxel (TAXUS II
and TAXUS IV) (5,6) (Fig. 1). These four
trials comprised a signicant proportion
of diabetic patients (20%) and provided
gures that allowed us to recalculate the
rate of restenosis after a follow-up of at
least 6 months in both the diabetic and
nondiabetic populations. The OR of in-
stent restenosis in diabetic patients com-
pared with control subjects averaged 1.94
(95%CI 1.462.58) in the groups receiv-
ing bare-metal stents (P 0.00001). Un-
fortunately, none of the studies gave
details about the respective ages of the di-
abetic and nondiabetic patients, so cor-
rections for possible differences in age
were not feasible (1). A similar OR was
found when DESs were considered (OR
2.24 [1.393.61], P 0.0009), suggest-
ing that the use of new DESs does not
allow to suppress differences between di-
abetic and nondiabetic patients. In both
groups of patients, the rate of restenosis
was markedly and signicantly (P
0.00001) higher with bare-metal stents
compared with DESs, with an OR of 6.33
(4.578.76) in the nondiabetic popula-
tion and 5.27 (3.368.28) in the diabetic
population.
These results demonstrated the po-
tential interest of DESs in patients at high
risk of restenosis. Nevertheless, even
when using DESs, the diabetes status re-
mains a signicant risk factor. This was
recently conrmed in a large cohort of
patients who beneted from unrestricted
utilization of sirolimus-eluting stents
compared with conventional bare-metal
stent implantation in the real world (7).
This research registry focused on a 1-year
cumulative rate of major adverse cardiac
events rather than on restenosis. It
showed that the benet of DESs did not
reach statistical signicance in diabetic
patients (OR 0.72 [95% CI 0.301.77],
P 0.50). Furthermore, diabetes was a
signicant predictor of major adverse car-
diac events (OR 1.62 [1.092.43], P
0.02) and of clinically driven target vessel
revascularization (OR 1.81 [1.102.99],
P 0.02) (Fig. 1).
In conclusion, diabetes remains a ma-
jor risk factor for restenosis after both
bare-metal stents and DESs. Considering
the high prevalence and burden of coro-
nary heart disease in diabetic patients,
specic studies should be performed in
this population in order to test the possi-
bility of reducing the risk of restenosis
and major cardiac events, including the
use of more effective DESs (2).
ANDRE
J. SCHEEN, MD, PHD

1
FABIAN WARZE
E
1
From the
1
Division of Diabetes, Nutrition and Met-
abolic Disorders, Department of Medicine, Centre
Hospitalier Universitaire, Sart Tilman, Lie` ge, Bel-
gium.
Letters
Address correspondence to Andre J. Scheen, Di-
vision of Diabetes, Nutrition and Metabolic Disor-
ders, Department of Medicine, Centre Hospitalier
Universitaire, Sart Tilman (B35), B-4000 Lie` ge, Bel-
gium. E-mail: andre.scheen@chu.ulg.ac.be.

References
1. Gilbert J, Raboud J, Zinman B: Meta-anal-
ysis of the effect of diabetes on restenosis
rates among patients receiving coronary
angioplasty stenting. Diabetes Care 27:
990994, 2004
2. Sousa JE, Serruys PW, Costa MA: New
frontiers in cardiology: drug-eluting stents
(part I & part II). Circulation 107:2274
2279, 23832389, 2003
3. Morice M-C, Serruys PW, Sousa JE, Faja-
det J, Ban Hayashi E, Perin M, Colombo A,
Schuler G, Barragan P, Guagliumi G,
Molnar F, Falotico R, for the RAVEL
Study Group: A randomized comparison
of sirolimus-eluting stent with a stan-
dard stent for coronary revascularization.
N Engl J Med 346:17731780, 2002
4. Moses J, Leon MB, Popma JJ, Fitzgerald
PJ, Holmes DR, OShaughnessy C, Caputo
RP, Kereiakes DJ, Williams DO, Teirstein
PS, Jaeger JL, Kuntz RE, for the SIRIUS
Investigators: Sirolimus-eluting stents
versus standard stents in patients with ste-
nosis in a native coronary artery. N Engl
J Med 349:13151323, 2003
5. Colombo A, Drzewiecki J, Banning A,
Grube E, Hauptmann K, Silber S, Dudek
D, Fort S, Schiele F, Zmudka K, Gua-
gliumi G, Russell ME, for the TAXUS II
Study Group: Randomized study to
assess the effectiveness of slow- and
moderate-release polymer-based pacli-
taxel-eluting stents for coronary ar-
tery lesions. Circulation 108:788794,
2003
6. Stone GW, Ellis SG, Cox DA, Hermiller
J, OShaughness C, Tift Mann J, Turco
M, Caputo R, Bergin P, Greenberg J,
Popma JJ, Russell ME, for the TAXUS-IV
Investigators: A polymer-based, pacli-
taxel-eluting stent in patients with cor-
onary artery disease. N Engl J Med 350:
221231, 2004
7. Lemos PA, Serruys PW, van Domburg
RT, Saia F, Arampatzis CA, Hoye A,
Degertekin M, Tanabe K, Daemen J, Liu
TKK, McFadden E, Sianos G, Hofma
SH, Smits PC, van der Giessen WJ, de
Feyter PJ: Unrestricted utilization of
sirolimus-eluting stents compared with
conventional bare stent implantation in
the Real World: the Rapamycin Elut-
ing Stent Evaluated at Rotterdam Cardi-
ology Hospital (RESEARCH) registry.
Circulation 109:190195, 2004
Major Diabetes
Complications Have
an Impact on Total
Annual Medical Cost
of Type 2 Diabetes
B
randle et al. (1) found that the oc-
currence of diabetes-related comor-
bidities in type 2 diabetic patients
are associated with an increase in average
annual medical cost.
To investigate the medical cost attrib-
utable to type 2 diabetes, we conducted a
retrospective longitudinal cost-of-care
study in a diabetologic center in Italy. A
priori, we estimated to validly enroll 300
type 2 diabetic patients with at least 1 year
of follow-up. To this aim, we randomly
selected 315 type 2 diabetic patients from
a base of 2,000 diabetic patients attend-
ing the diabetologic center of Portogruaro
during the period from January 2001 to
August 2002.
Cost included hospitalizations, visits,
diagnostics, and pharmacological thera-
pies and were quantied in the perspective
of the National Health Service. We ex-
Figure 1Meta-analysis of four trials comparing the effects on restenosis of bare-metal stents and drug-eluting stents in diabetic and nondiabetic
patients.
Letters
tracted clinical and demographic informa-
tion from the electronic database and
performed extensive chart review, includ-
ing the comorbidities retinopathy, cardiop-
athy (coronary heart disease), vasculopathy
(other than coronary heart disease), and
nephropathy.
We analyzed the association between
diabetes-related comorbidities and aver-
age annual medical costs using univariate
and multiple linear regression analyses. In
the linear regression analysis, cost was
transformed using the square-root trans-
formation to better t a Gaussian
distribution.
Sixteen type 2 diabetic patients were
excluded because it was found that their
follow-up period was 1 year; the main
reasons were premature mortality and
loss to follow-up. A total of 299 type 2
diabetic patients were considered for this
analysis and followed-up for an average of
476 days, totaling 520 person-years of
observation. Their mean (SD) age was
68.6 8.8 years, and 201 (67.2%) were
men. The mean systolic blood pressure,
diastolic blood pressure, total cholesterol,
HDL cholesterol, HbA
1c
, and Hb levels
were 152.5 20.9 mmHg, 82.7 10.0
mmHg, 195.5 41.6 mg/dl, 50.2 20.3
mg/dl, 7.1 1.5%, and 13.9 1.5 g/dl,
respectively. The average annual cost of
care was 1,909.67 (i.e., $2,425 U.S.; ex-
change rate, 1 Euro $1.27 U.S.); 52%of
costs were attributable to drugs, 28% to
hospitalizations, 11% to diagnostics, and
9% to visits.
A total of 101 (33.8%) type 2 diabetic
patients were free of diabetes-related co-
morbidities, 117 (39.1%) had one com-
plication, and 81 (27.1%) had two or
more complications. The more frequent
complication was vasculopathy, which af-
fected 89 (29.8%) type 2 diabetic pa-
tients, followed by cardiopathy (79
[26.4%]), retinopathy (66 [22.1%]), and
nephropathy (65 [21.7%]).
The annual medical costs increased
with the number of complications from
1,039.59 ($1,320 U.S.) to 1,808.17
($2,296 U.S.) and to 3,141.21 ($3,989
U.S.) in type 2 diabetic patients with
none, one, and two and more complica-
tions, respectively, with the association
being statistically signicant in both uni-
variate (Kruskall-Wallis test, 73,035; P
0.0001) and multiple linear regression
analyses (R
2
0.21; F test 82.5, P
0.0001).
We could not assess the impact of di-
alysis on cost, since the care of type 2 di-
abetic patients developing end-stage renal
disease is not controlled by the diabeto-
logic center.
We did not consider the cost of sup-
plies for self-monitoring of blood glucose,
which is, at any rate, minimal (100
patient
1
year
1
) and is not related to
the type of complication.
Our study conrms the ndings of
Brandle et al. regarding the annual medi-
cal cost and its determinants in type 2
diabetic patients. Strategies aimed at pre-
venting the onset of diabetes complica-
tions are likely to reduce medical costs in
the long term, while improving patients
health.
ANDREA MORSANUTTO, PHARMD
1
PATRIZIA BERTO, PHARMD, MBA
2
STEFANIA LOPATRIELLO, PHARMD
2
DARIO VOINOVICH, PHD
3
RENZO GELISIO, MD
4
LORENZO G. MANTOVANI, MSC, DSC
1
From the
1
Center of Pharmacoeconomics, Depart-
ment of Pharmacological Sciences, University of Mi-
lan, Milan, Italy;
2
PBE Consulting, Verona, Italy; the
3
Department of Pharmacological Sciences, Univer-
sity of Trieste, Trieste, Italy; and the
4
Diabetologic
Center, ULSS 10 Portogruarese, Portogruaro, Italy.
Address correspondence to Dr. Lorenzo G. Man-
tovani, Center of Pharmacoeconomics, Department
of Pharmacological Sciences, University of Milan,
Via Balzaretti, 9 I-20133 Milan, Italy. E-mail:
lorenzo.mantovani@unimi.it.

References
1. Brandle M, Zhou H, Smith BR, Marriott D,
Burke R, Tabaei BP, Brown MB, Herman
WH: The direct medical cost of type 2
diabetes. Diabetes Care 26:23002304,
2003
Normal Insulin
Sensitivity During
the Late Preclinical
Stage of Type 1
Diabetes
I
n most cases, type 1 diabetes is the late
consequence of a -cell autoimmune
destruction leading to absolute insulin
deciency (1). At onset of clinical diabe-
tes, -cell mass is thought to be reduced
by 8090%(2), and 73%of adult patients
have ketosis or ketoacidosis at diagnosis
(3). Patients with type 1 diabetes have
major defects in insulin sensitivity at di-
agnosis of overt diabetes (48). However,
the fact that pre-diabetic patients can
maintain normal blood glucose levels in
spite of dramatically lowinsulin secretory
capacities suggests that their insulin sen-
sitivity is normal.
Here we report that insulin sensitiv-
ity, measured by the glucose clamp
method in nine patients, remains normal
even during the very late preclinical stage
of type 1 diabetes. None of these patients
had clinical symptoms of overt diabetes,
spontaneous weight loss, or ketosis before
or at the time of the study. Mean age at
entry in the study was 27 years (range
2041). The mean BMI was 20.9 kg/m
2
(range 16.323.2). All patients were islet
cell antibody positive, and seven of eight
tested had at least one susceptibility HLA
haplotype. Mean fasting glucose level was
7.2 1.3 mmol/l (range 59.6). Of the
eight patients who met the criteria for di-
abetes, four had fasting blood glucose
level 7 mmol/l; in the ninth patient, oral
glucose tolerance test disclosed impaired
glucose tolerance. The mean HbA
1c
level
was 6.4 0.8% (range 5.37.8; normal
values 5.0 0.5%). The sum of 1- and 3-
min plasma insulin levels after the intra-
venous glucose tolerance test was 15.3
5.8 mU/l (range 1125).
Four patients were treated with insu-
lin immediately after the metabolic explo-
rations reported here. Four others have
initiated permanent insulin therapy
within 24 months of follow-up. The last
patient was still noninsulin dependent
when last seen 1 year after the study.
During clamp studies, the insulin
dose-response curve of preclinical type 1
diabetic subjects was superimposable to
that of 20 healthy control subjects. Maxi-
mal glucose infusion rate, at a plasma in-
sulin level of 1,400 120 mU/l, was
17.2 1.6 mg kg
1
min
1
(Fig. 1). It
was 16.4 0.6 mg kg
1
min
1
in 20
control subjects at a similar plasma insu-
lin level: 1,500 100 mU/l. By contrast,
the maximal glucose infusion rate was sig-
nicantly lower in 15 patients with symp-
tomatic diabetes of recent onset (11 1.4
mg kg
1
min
1
, P 0.003 by Mann-
Whitney Utest). These results conrm, at
an even later stage of the natural history of
the disease, those observed in a substudy
of the Diabetes Prevention Trial-1 (9).
Thus, the onset of overt clinical diabetes
may be triggered by the addition of insu-
Letters
lin resistance to insulin deciency; the
triggers may be external or internal fac-
tors. In this respect, the seasonal peaks of
the incidence of diabetes (10) suggest that
viral infections may play a role in the very
nal act of the preclinical phase by trig-
gering insulin resistance and metabolic
storm. The respective parts of insulin de-
ciency and insulin resistance may be de-
terminants for the onset and duration of
the honeymoon period.
ETIENNE LARGER, MD, PHD
1
BENJAMIN RAKOTOAMININA, MD
2
MOHAMMED EDDOUKS, PHD
1
JOSE TIMSIT, MD
3
CHRISTIAN BOITARD, MD
3
ROGER ASSAN, MD
1
RE MY BURCELIN, PHD
1
JEAN-JACQUES ROBERT, MD, PHD
2
From the
1
Service dendocrinologie-diabe tologie,
Ho pital Bichat, Assistance Publique-Ho pitaux de
Paris, Paris, France; the
2
Unite Diabe` te de lEnfant et
de lAdolescent, Ho pital Necker-Enfants Malades,
Assistance Publique-Ho pitaux de Paris, Paris,
France; and the
3
Service dimmunologie clinique,
Ho pital Necker-Enfants Malades, Assistance Pub-
lique-Ho pitaux de Paris, Paris, France
Address correspondence to Dr. Etienne Larger,
Service dendocrinologie-diabe tologie, Ho pital Bi-
chat, 46 rue H. Huchard, 75877 Paris cedex 18,
France. E-mail: etienne.larger@bch.ap-hop-paris.fr.
M.E. is currently afliated with the UFR de Physi-
ologie, Faculte des sciences, nutrition et endocrinolo-
gie, Errachidia, Morocco; J.T. and C.B. are currently
afliatedwiththe Service dimmunologie-diabetologie,
ho pital Cochin, Assistance Publique-Ho pitaux de
Paris, Paris, France; R.B. is currently afliated with
UMR 5018 CHU Rangueil, Toulouse, France.
Deceased in August 2003.

References
1. The Expert Committee on the Diagnosis
and Classication of Diabetes Mellitus:
Report of the Expert Committee on the Di-
agnosis and Classication of Diabetes Mel-
litus. Diabetes Care 20:11831197, 1997
2. Pipeleers D, Ling Z: Pancreatic Beta cells
in insulin-dependent diabetes. Diabetes
Metab Rev 8:209227, 1992
3. Karjalainen J, Salmela P, Ilonen J, Surcel
H-M, Knip M: Acomparison of childhood
and adult type 1 diabetes mellitus. N Engl
J Med 320:881886, 1989
4. Yki-Jarvinen H, Koivisto V: Natural
course of insulin resistance in type 1 dia-
betes. N Engl J Med 315:224230, 1986
5. Gray R, Cowan P, Duncan L, Clarke B:
Reversal of insulin resistance in type 1 di-
abetes following initiation of insulin treat-
ment. Diabet Med 3:1823, 1986
6. Nijs H, Radder J, Fro lich M, Krans H: The
course and determinants of insulin action
in type 1 (insulin-dependent) diabetes
mellitus. Diabetologia 32:2027, 1989
7. Burcelin R, Eddouks M, Beylot M, Nor-
mand S, Boitard C, Feutren G, Landais
P, Riou JP, Girard JR, Bach JF, Assan RI:
Hypersensitivity to insulin during re-
missions in cyclosporin-treated IDDM
patients. Diabetes Care 16:880895,
1993
8. DeFronzo RA, Matsuda M, Barrett E: Di-
abetic ketoacidosis: a combined meta-
bolic-nephrologic approach. Diabetes
Rev 2:209238, 1994
9. Greenbaum CJ, Prigeon RL, DAlessio DA:
Impaired -cell function, incretin effect,
and glucagon suppression in patients with
type 1 diabetes who have normal fasting
glucose. Diabetes 51:951957, 2002
10. Le vy-Marchal C, Patterson C, Green A:
Variation by age group and seasonality at
diagnosis of chilhood IDDM in Europe.
Diabetologia 38:823830, 1995
COMMENTS AND
RESPONSES
Comparative Study
of Prognostic Value
for Coronary Disease
Risk Between the
U.K. Prospective
Diabetes Study and
Framingham Models
Response to Protopsaltis et al.
T
he comparison by Protopsaltis et al.
(1) of the Framinghamrisk equations
and the U.K. Prospective Diabetes
Study (UKPDS) risk engine as predictors of
coronary risk in diabetes is of interest, as
previous analyses have shown that the Fra-
mingham equations can underestimate ab-
solute coronary heart disease (CHD) risk in
diabetic subjects by a factor of 2 or more
(2,3). The sensitivity and specicity analy-
ses presented by the authors, however, are
difcult to interpret because of method-
ological concerns. They examined the inci-
dence of coronary angiographically
determined CHD but not hard CHD (de-
ned as fatal or nonfatal myocardial infarc-
tion) as estimatedby the UKPDSriskengine
(4). Their use of a retrospective survivor co-
hort introduces bias, as patients with fatal
CHDwill have beenexcluded. This may ex-
plain the apparent poor performance of
HbA
1c
as a risk predictor, since myocardial
infarction is more often fatal in those with
higher HbA
1c
(5). The use of a mixedcohort
Figure 1Glucose clamp study. Dose-response curves obtained during the euglycemic-
hyperinsulinemic clamps in patients with preclinical diabetes (E). For the purpose of comparison,
results obtained in control subjects (I) and in patients with symptomatic diabetes studied within
13 week after diagnosis () are shown. The latter two are from ref. 7.
Letters
of type 1 and type 2 diabetic subjects is
problematic because the UKPDS risk en-
gine, a type 2 diabetesspecic risk calcula-
tor, has not been evaluated in subjects with
type 1 diabetes. The UKPDS analysis (6)
cited by the authors shows that a 1%decre-
ment in HbA
1c
was associated with a 37%
risk reduction in microvascular disease, not
a 10-fold reduction (90%).
A full validation of a risk model re-
quires a prospective study of a cohort to
which the model is applicable. In the case
of the risk engine, a cohort with type 2
diabetes and suitable demographic char-
acteristics (3) is needed. Covariates
should be measured at the beginning of
follow-up, and the cohort should be mon-
itored for the end points addressed by the
model. In the case of the risk engine, these
end points would be fatal and nonfatal
myocardial infarction and sudden cardiac
death. Publication of the observed rate of
CHD in the cohort and of the mean pre-
dicted rate according to the model would
then allow an evaluation of risk tools
against true rates of heart disease. An ideal
study would also adjust for assay differ-
ences, if appropriate, and for regression
dilution and competing risks if necessary.
RICHARD J. STEVENS, PHD,
1
RURY HOLMAN, FRCP
1
From the
1
Diabetes Trials Unit, Oxford Centre for
Diabetes, Endocrinology and Metabolism, Univer-
sity of Oxford, Oxford, U.K.
Address correspondence to Dr. Richard Stevens,
Diabetes Trials Unit, Oxford Centre for Diabetes,
Endocrinology and Metabolism, Churchill Hospital,
Ol d Road, Oxford, OX3 7LJ U. K. E-mai l :
risk.engine@dtu.ox.ac.uk.

References
1. Protopsaltis ID, Konstantinopoulos PA,
Kamaratos AV, Melidonis AI: Compara-
tive study of prognostic value for coronary
disease risk between the U. K. Prospective
Diabetes Study and Framingham models
(Letter). Diabetes Care 27:277278, 2004
2. Yeo WW, Rowland Yeo K: Predicting
CHD risk in patients with diabetes melli-
tus. Diabet Med 18:341344, 2001
3. Stevens RJ, Adler AI, Stratton IM, Manley
SE, Holman RR: The UKPDS in Wiscon-
sin: comparing cardiovascular models
from the UK to data from the US (Ab-
stract). Diabetes 50 (Suppl. 1):A204, 2001
4. UKPDS Group: The UKPDS Risk Engine:
a model for the risk of coronary heart dis-
ease in type 2 diabetes (UKPDS 56). Clin
Sci 101:671679, 2001
5. Stevens RJ, Coleman RL, Adler AI, Strat-
ton IM, Matthews DR, Holman RR: Risk
factors for myocardial infarction case fa-
tality and stroke case fatality in type 2 di-
abetes (UKPDS 66). Diabetes Care
27:201207, 2004
6. UKPDS Group: Association of glycaemia
with macrovascular and microvascular
complications of type 2 diabetes (UKPDS
35). BMJ 321:405412, 2000
Comparative Study
of Prognostic Value
for Coronary Disease
Risk Between the
U.K. Prospective
Diabetes Study and
Framingham Models
Response to Stevens and Holman
W
e appreciate the comments of
Stevens and Holman (1) regard-
ing our analysis comparing the
accuracy of the U.K. Prospective Diabetes
Study (UKPDS) (2) and Framingham
models in the prediction of 10-year risk
for coronary artery disearse (CAD) in di-
abetic patients. In our letter, we men-
tioned that the diagnosis of CAD was
established by coronary angiography. In
addition, this study was performed in a
cohort of patients with type 2 diabetes
and included a 10-year follow-up. In the
sample of cases with myocardial infarc-
tion who were examined during this anal-
ysis, fatal and nonfatal cases were
included.
We agree that this analysis was retro-
spective with its disadvantages, and it is
absolutely equitable that the variables
used by both of these models should have
been determined at the beginning of fol-
low-up. However, it is obvious that this
study was designed as a retrospective,
since the UKPDS prediction model has
recently constituted a useful cardiovascu-
lar disease (CVD) risk prognostic model.
In the epidemiologic analysis of the UK-
PDS, Stratton et al. (3) showed that the ef-
fect of HbA
1c
values on microvascular
complications over a range of 5.511%was
nearly 10-fold, whereas the effect of HbA
1c
on myocardial infarction incidence, over
the same range of HbA
1c
, was 2-fold.
Regarding the effect of HbA
1c
on
myocardial infarction incidence, the
range of HbA
1c
was from 5.5 to 11%, so
we can say that 11% (upper limit) is two
times 5.5% or 1 more time 5.5% (lower
limit). Therefore 1 refers to one time and
not 1%. The percent symbol was added
by mistake.
Moreover, as the UKPDS study
showed, improved glycemic control had
no signicant impact in cardiovascular
outcomes in patients with type 2 diabetes.
The Veterans Affairs Cooperative Study
(4) also showed a nonstatistical signi-
cant deterioration of CVDevents in inten-
sively treated patients compared with
those receiving standard treatment.
The exception for reducing mortality
from myocardial infarction was in the
overweight UKPDS cohort patients
treated with metformin (5). However,
metformin use was not associated with
lower blood glucose levels compared with
insulin or sulfonylureas, so it can be as-
sumed that the cardioprotective effects of
metformin could be interpreted by its
well-known actions in the atherothrom-
botic risk prole (6) and blood pressure
levels through nonglycemic pathways.
Summarizing the previously epide-
miologic data, since coronary heart dis-
ease is a multifactorial disease and
glycemic control is not signicantly asso-
ciated (7,8) with reduced CVD risk in di-
abetic patients, it can be assumed that the
small contribution of chronic hyperglyce-
mia (HbA
1c
) at the incidence of macrovas-
cular complications has a respectively
weak contribution on the calculation of
coronary heart disease risk by the UKPDS
mathematical model.
Finally, we absolutely agree with
Stevens and Holman that prospective
studies will be needed to determine the
validity of the predictive value for both of
these models.
IOANNIS D. PROTOPSALTIS, PHD
1
GEORGE NIKOLOPOULOS, MD
2
ANDREAS MELIDONIS, PHD
1
From the
1
Diabetes Center, General Hospital Tza-
neio of Piraeus, Piraeus, Greece; and the
2
Hellenic
Center for Infection Disease Control, Athens,
Greece.
Address correspndence to Dr. Andreas Melido-
nis, General Hospital Tzaneio of Piraeus, Zanni &
Afentouli 1, 18536 Piraeus, Greece. E-mail:
tzaniodiabetes@yahoo.com.
L E T T E R S

References
1. Stevens RJ, Holman R: Comparative study
of prognostic value for coronary disease
risk between the U.K. prospective diabe-
tes study and Framingham models (Let-
ter). Diabetes Care 27:18431844, 2004
2. Protopsaltis ID, Konstantinopoulos PA,
Kamaratos AV, Melidonis AI: Compara-
tive study of prognostic value for coronary
disease risk between the U. K. Prospective
Diabetes Study and Framingham models
3. Stratton IM, Adler AI, Neil HA, Mathews
DR, Manley SE, Cull CA, Hadden D,
Turner RC, Holman RR: Association of
glycaemia with macrovascular and micro-
vascular complications of type 2 diabetes
(UKPDS 35): prospective observational
study. BMJ 321:405412, 2000
4. Abraira C, Colwell JA, Nuttall F, Sawin
CT, Henderson W, Comstock JP, Eman-
uele NV, Levin SR, Pacold I, Lee HS: Car-
diovascular events and correlates in the
Veterans Affairs Diabetes Feasibility trial.
Arch Intern Med 157:181188, 1997
5. UK Prospective Diabetes Study (UKPDS)
Group: Effect of intensive blood glucose
control withmetforminoncomplications in
overweight patients with type 2 diabetes
(UKPDS 34). Lancet 352:854865, 1998
6. Mamputu JC, Wiernsperger NF, Renier G:
Antiatherogenic properties of metformin:
the experimental evidence (Review). Dia-
betes Metab 29:6S716S76, 2003
7. Wild SH, Dunn CJ, McKeigue PM, Comte
S: Glycemic control and cardiovascular
disease in type 2 diabetes: a review. Dia-
betes Metab Res Rev 15:197204, 1999
8. Huang ES, Meigs JB, Singer DE: The effect
of interventions to prevent cardiovascular
disease in patients with type 2 diabetes
mellitus. Am J Med 111:633642, 2001
Dietary Antioxidant
Intake and Risk of
Type 2 Diabetes
Response to Montonen et al.
T
he results by Montonen et al. (1) re-
garding the potential role of dietary
antioxidants in the prevention of type
2 diabetes are of considerable interest. The
authors identify -cryptoxanthin as a pre-
ventive factor, regardless of adjustment for
other potential confounding factors, and
point out that data support the hypothesis
that a sufcient intake of antioxidants plays
a role in type 2 diabetes prevention.
Although the follow-up time and
number of subjects are considerable, the
nutritional and public health relevance on
a population basis is uncertain, at least
concerning -cryptoxanthin. The authors
report that dietary intake is on average 4
g -cryptoxanthin/day in both groups,
with the lowest and highest quartiles
0.3 to 4.2 g/day. However, regard-
less of the value obtained for the risk ratio
(0.58 [95% CI 0.440.78]), these data
should be interpreted in practical terms.
In developed countries, -cryptoxan-
thin is mostly provided by citrus fruits,
(2,3) and, after these major contributors
are considered, dietary -cryptoxanthin
intake may be easily assessed, making
misclassication of individuals unlikely,
especially when considering extreme cen-
tiles. Also, -cryptoxanthin content in
fresh orange and mandarin is 1201,300
g/100 g (3,4). Thus, assuming the low-
est end of this range, intake of -crypto-
xanthin in the highest quartile in
Montonen et al.s study (5 g/day)
would be equivalent to the consumption
of 5 g fresh orange or mandarin per day,
which is less than one-twentieth of a stan-
dard portion.
In other European countries, median
-cryptoxanthin intake is 0.451.36 mg/
day (2,3); the southern countries have
higher intakes than those in the north, a
difference that is also observed in their se-
rum levels (5). Because there is a biologi-
cal correlation between -cryptoxanthin
intake and serumconcentrations, it is rea-
sonable to assume that the lower the in-
take the lower the serum concentrations.
Considering these facts, in our opin-
ion, it is very difcult to imagine any po-
tential benecial effects associated with
the intake of such lowamounts of -cryp-
toxanthin as those reported by Montonen
et al. (1), even when the statistical evi-
dence is strong. Where nutritionally rele-
vant population-based recommendations
are concerned, we should not forget the
multifactorial nature of the disease and
that people, in the real world, consume
foods and diets.
FERNANDO GRANADO-LORENCIO, PHD
BEGONA OLMEDILLA-ALONSO, PHARMD, PHD
From the Unidad de Vitaminas, Hospital Universi-
tario Puerta de Hierro, Madrid, Spain; and the
Redes Tema ticas de Investigacio n del Instituto de
Salud Carlos III, Madrid, Spain. RCMN C03/08 and
RGMBE G03/090 (B.O.-A).
Address correspondence to Fernando Granado-
Lorencio, PhD, Hospital Universitario Puerta de
Hierro, Unidad de Vitaminas, 28035 Madrid, Spain.
E-mail: bolmedilla.hpth@salud.madrid.org.

References
1. Montonen J, Ja rvinen R, Knekt P, Reun-
anen A: Dietary antioxidant intake and
risk of type 2 diabetes. Diabetes Care 27:
362366, 2004
2. Granado F, Olmedilla B, Blanco I, Rojas-
Hidalgo E: Major fruits and vegetables
contributors to the main serum carote-
noids in the Spanish diet. Eur J Clin Nutr
50:246250, 1996
3. ONeill ME, Carroll Y, Corridan B,
Olmedilla B, Granado F, Blanco I, Van
den Berg H, Hininger I, Rousell AM, Cho-
pra M, Southon S, Thurnham DI: A Euro-
pean Carotenoid Database to assess
carotenoid intakes and its use in a ve-
country comparative study. Brit J Nutr 85:
499507, 2001
4. Poorvliet EJ, West CE: The carotenoid
content of foods with special reference to
developing countries. Vitamin A Field
Support Project (VITAL). Arlington, VA,
International Science and Technology In-
stitute, 1993
5. Olmedilla B, Granado F, Southon S, Wright
AJA, Blanco I, Gil-Martinez E, Berg HVD,
Corridan B, Roussel AM, Chopra I, Thurn-
ham DI: Serum concentrations of carote-
noids and vitamins A, E, and C in control
subjects fromve Europeancountries. Brit J
Nutr 85:227238, 2001
Dietary Antioxidant
Intake and Risk of
Type 2 Diabetes
Response to Granado-Lorencio and
Olmedilla-Alonso
W
e would like to express our warm
thanks to Granado-Lorencio and
Olmedilla-Alonso for carefully
reading our article (1) and for their com-
ments regarding the inverse association
between -cryptoxanthin intake and dia-
betes risk (2). Granado-Lorencio and
Olmedilla-Alonso pointed out that the
-cryptoxanthin intake in our study pop-
ulation was exceptionally low; therefore,
any potential benecial effect associated
with such low amounts of -cryptoxan-
thin is difcult to imagine.
In epidemiologic studies, it is difcult
to draw conclusions about the absolute
amount of a specic nutritional com-
Letters
pound necessary for a benecial health
effect. One of the mainreasons for this is the
fact that nutrient compositiontables are un-
der continuingdevelopment. Acommonal-
ternative technique, which we used in the
present study, is to rank the study popula-
tion according to the intakes. In such an
approach, the validityof the results depends
on the accuracy of the ranking, so absolute
intake levels are less important.
In our analyses, the assessment of
-cryptoxanthin intake was based on an-
alyzed values of Finnish foods available in
the late 1980s. However, more recent es-
timates of the -cryptoxanthin content of
oranges have been shown to be much
higher than those used in the present
study (average orange intake 29 g/day).
To ensure the accuracy of our analyses,
we recalculated the -cryptoxanthin in-
take using more recent published values
(3,4). We calculated the coefcient be-
tween quartiles of the recalculated and the
original variables and noted a high level of
agreement between these two variables
( 0.9), suggesting that the study par-
ticipants are very similarly ranked using
either the original or the recalculated
-cryptoxanthin intake values. Thus, the
inverse association between -cryptoxan-
thin intake and diabetes risk presented in
our article is justied.
In conclusion, we suggest that the sig-
nicant inverse association between the
intake of -cryptoxanthin and the risk of
type 2 diabetes observed in our study is a
valid nding. The importance of this nd-
ing in the prevention of type 2 diabetes,
however, remains to be established.
JUKKA MONTONEN, MSC
1
PAUL KNEKT, PHD
1,2
RITVA J ARVINEN, PHD
3
ANTTI REUNANEN, MD, PHD
1
Fromthe
1
Department of Health and Functional Ca-
pacity, National Public Health Institute, Helsinki,
Finland; the
2
Social Insurance Institution, Helsinki
and Turku, Finland; and the
3
Department of Clini-
cal Nutrition, University of Kuopio, Kuopio, Fin-
land.
Address correspondence to Jukka Montonen, Na-
tional Public Health Institute, Department of Health
and Functional Capacity, Mannerheimintie 166,
00300 Hel si nki , Fi nl and. E- mai l : j ukka.
montonen@ktl..

References
1. Montonen J, Ja rvinen R, Knekt P, Reun-
anen A: Dietary antioxidant intake and
risk of type 2 diabetes. Diabetes Care 27:
362366, 2004
2. Granado-Lorencio F, Olmedilla-Alonso B:
Dietary antioxidant intake andrisk of type 2
diabetes (Letter). Diabetes Care 27:1845,
2004
3. ONeill ME, Carroll Y, Corridan B, Olm-
edilla B, GranandoF, BlancoI, VandenBerg
H, Hininger I, Rousell AM, Chopra M, Sou-
thon S, Thurman DI: A European carot-
enoid database to assess carotenoid intakes
and its use in a ve-country comparative
study. Br J Nutr 85:499507, 2001
4. Brugrd Konde , Staffas A, Dahl P,
Becker W: Carotenoids in Foodstuff in Swe-
den (in Swedish with an English sum-
mary). Uppsala, Sweden, National Food
Administration, 1996 (Reports of the Na-
tional Food Administration 12)
Twice-Daily NPH or
Mixture Insulins
Versus Triple
Therapy: Apples
Versus Oranges
Response to Poulsen et al.
P
oulsen et al. (1) recently demon-
s t r a t e d t ha t t r i pl e t he r a py
(metformin, rosiglitazone, and
preprandial insulin aspart) lowered
HbA
1c
levels much better than twice-
daily NPH or MIX insulin in type 2 di-
abetic patients. They then claim that the
reason for this improvement, compared
with the patients treated with insulin
alone, is the superiority of specically
treating the three pathophysiological
components of type 2 diabetes (periph-
eral insulin resistance, hepatic insulin
resi stance, and i mpai red gl ucose-
stimulated insulin secretion). Given the
design of the study, this conclusion is
suspect for two reasons.
First, the glycemic goals for the two
groups were different. The goals for pa-
tients receiving only insulin were a pre-
prandial value of 57 mmo1/l. The goal
for patients receiving triple therapy was a
postprandial value of 57 mmol/l.
Achieving a postprandial goal of 57
mmol/l will necessarily lead to better con-
trol than achieving the same goal prepran-
dially (since in the latter situation the
postprandial glucose concentrations will
obviously be higher).
Second, twice-daily NPH injections
are a poor insulin regimen to achieve near
euglycemia because there is no short- or
rapid-acting insulin to blunt postprandial
hyperglycemia. (The authors acknowl-
edge this in their discussion.) How many
of the eight patients in the control group
were on twice-daily NPH insulin? Even
the ones on MIX insulin (I assume this is a
premixed insulin) are not on an optimal in-
sulin regimen for achieving near euglyce-
mia. One example illustrates this point.
How does one adjust the MIX insulin dose
in a patient whose preprandial glucose con-
centrations before supper are in the lower
part of the goal range but whose preprandial
lunch values exceed the goal range?
For both of these reasons one would
expect higher postprandial glucose con-
centrations in the control group receiving
only insulin (which is borne out in Fig.
2A) and consequently higher HbA
1c
lev-
els. Hopefully these issues will not be ig-
nored in subsequent long-term studies.
MAYER B. DAVIDSON, MD
From the Clinical Trials Unit, Charles R. Drew Uni-
versity, Los Angeles, California.
Address correspondence to Dr. Mayer B. David-
son, MD, Director, Clinical Trials Unit, Charles R.
Drew University, 1731 East 120th St., Los Angeles,
CA 90059. E-mail: madavids@cdrewu.edu.

References
1. Poulsen MK, Henriksen JE, Hother-
Nielsen O, Beck-Nielsen H: The com-
bined effect of triple therapy with
rosiglitazone, metformin, and insulin as-
part in type 2 diabetic patients. Diabetes
Care 26:32733279, 2003
The Combined Effect
of Triple Therapy
with Rosiglitazone,
Metformin, and
Insulin Aspart in
Type 2 Diabetic
Patients
Response to Poulsen et al.
I
have some concerns regarding the re-
sults of Poulsen et al.s (1) study, which
suggested that triple therapy with insu-
lin aspart, metformin, and rosiglitazone
was superior to continuing treatment
Letters
with NPH insulin alone or NPH plus reg-
ular insulin (referred to as the control
group). These results should be inter-
preted with caution because of several
limitations in the studys design. First, I
disagree with the authors for applying
identical goals for metabolic control in
both study groups. Although the glycemic
target was similar in both arms, the timing
of self-monitored blood glucose was dif-
ferent. Thus, the adjustment of the insulin
dosage in the control group aimed at
blood glucose levels between 5 and 7
mmol/l in the preprandial period,
whereas in the triple therapy group, insu-
lin adjustment targeted the same range of
blood glucose but in the postprandial pe-
riod. Clearly, setting a glycemic goal of
57 mmol/l for postprandial blood glu-
cose was quite aggressive and substan-
tially lower than the recommendations of
the American Diabetes Association
(10.0 mmol/l) (2). This more stringent
glycemic target in the triple therapy group
compared with the control group may ex-
plain, at least in part, the superior meta-
bolic control observed with the triple
therapy regimen. In addition, the bias that
favored the triple therapy group (e.g.,
providing closer follow-up and care)
could not be excluded since the study was
not blinded. Moreover, the investigators
may have been reluctant to further in-
crease insulin doses in the control group
to achieve the studys preprandial glyce-
mic goal because of the concern over hy-
poglycemia. Second, some patients in the
control group received NPH twice daily
and others received a mixture of NPH
plus regular insulin. These two treatment
regimens are different and should not be
included in a single group because dia-
betic subjects receiving NPH alone may
have inadequate plasma levels of post-
prandial insulin. Third, one limitation of
triple therapy is the high cost of treatment
and related laboratory tests, a factor that
was not claried in the study presumably
because the three drugs were provided at
no cost by the respective companies.
Despite these limitations, this pilot
investigation addressed an attractive ther-
apeutic approach that targeted the three
main defects in the pathophysiology of
type 2 diabetes. In this respect, the studys
ndings suggested that insulin resistance
was probably the hardest abnormality to
correct. Indeed, the combination of the
two insulin-sensitizing agents, metformin
and rosiglitazone, in maximumtherapeu-
tic doses resulted in only partial amelio-
ration of insulin sensitivity after 6 months
(1). It would be interesting to evaluate the
long-termeffects and cost-effectiveness of
this form of triple therapy in type 2 dia-
betes in well-designed trials.
NASSER MIKHAIL, MD, MSC
From the Olive View-UCLA Medical Center, UCLA
School of Medicine, Sylmar, California.
Address correspondence to Dr. Nasser Mikhail,
Olive View-UCLA Medical Center, Department of
Medicine, 14445 Olive View Dr., Sylmar, CA
91342. E-mail: nasser.mikhail@gte.net.

References
Care 26:32733279, 2003
2. American Diabetes Association: Stan-
dards of medical care for patients with di-
abetes mellitus (Position Statement).
Diabetes Care 26 (Suppl. 1):S33S50,
2003
The Combined Effect
of Triple Therapy
with Rosiglitazone,
Metformin, and
Insulin Aspart in
Type 2 Diabetic
Patients
Response to Davidson and Mikhail
W
e were not able to normalize
HbA
1c
values completely during
triple therapy probably due to
the fact that insulin action was not com-
pletely normalized by rosiglitazone and
metformin, as stated by Davidson (1).
Still, insulin-mediated glucose uptake
was only 60% normal during triple ther-
apy; however, a longer treatment period
or a more potent insulin synthesizer may
help in that respect. Another way to fur-
ther reduce HbA
1c
values is to use a more
aggressive algorithm, increasing the No-
vorapid dose further before meals. How-
ever, this may induce more hypoglycemic
attacks during the day time. The South
Danish Diabetes Study, a newly initiated
2-year follow-up study that includes
400 subjects, will hopefully answer this
question.
Davidson (1) and Mikhail (2) focus
on the different ways of monitoring blood
glucose values when using the two ap-
proaches. They claim that a focus on the
postprandial values in the triple therapy
group compared with preprandrial values
in the control group favors the triple ther-
apy group. However, self-monitored glu-
cose values showed that the preprandial
values were identical in the triple therapy
group and the control group. Hence, the
adjustment of the insulin dosage would
have been identical only if the prepran-
drial blood glucose values were also used
in the triple therapy group (as in the con-
trol group). The curve for self-monitored
glucose in the triple therapy group is at
(no postprandial rise in blood glucose val-
ues), with a geometric mean of 7
mmol/l, which is close to the ndings in
nondiabetic subjects. However, monitor-
ing higher postprandial blood glucose
values in the control group (NPH insulin
alone) may have resulted in a further rise
of the insulin dose, as proposed by
Mikhail. We do not expect, however, to
see a further effect of this on blood glu-
cose values, since the 50% increase we
already performed did not inuence
HbA
1c
values (3), but this postulate is still
not tested. The last argument against the
criticism raised is that we aimed for the
same HbA
1c
values in both groups, which
means that we did not rely on blood glu-
cose values only. Thus, we cannot prove
that monitoring the postprandial values
in the control group also would have im-
proved the glucose control in that group,
but we showed that triple therapy was
able to nearly normalize HbA
1c
values in
type 2 diabetic patients, which has never
been the case when using NPH insulin
treatment alone.
We conclude that this trial conrms
our hypothesis that the key to the treat-
ment of type 2 diabetic patients is to con-
trol postprandial blood glucose values by
reconstructing the insulin peaks after
meals and by improving insulin action in
skeletal muscle, fat, and liver cells. Triple
therapy seems to be a safe and effective
treatment of hyperglycemia in type 2 dia-
betic patients, but further studies, espe-
cially of a longer duration, are needed.
HENNING BECK-NIELSEN, MD, MSCI
Letters
From the Department of Endocrinology, Odense
University Hospital, University of Southern Den-
mark, Odense C, Denmark.
Address correspondence to Dr. Henning Beck-
Nielsen, Odense University Hospital, Department of
Endocrinology, Kloevervaenget 6,4, 5000 Odense
C, Denmark. E-mail: henning.beck-nielsen@ouh.
fyns-amt.dk.

References
1. Davidson MB: Twice-daily NPH or mix-
ture insulins versus triple therapy: apples
versus oranges (Letter). Diabetes Care 27:
1846, 2004
2. Mikhail N: The combined effect of triple
therapy with rosiglitazone, metformin,
and insulin aspart in type 2 diabetic pa-
tients (Letter). Diabetes Care 26:1846
1847, 2003
Care 26:32733279, 2003
A Randomized Trial
Evaluating a
Predominantly Fetal
GrowthBased
Strategy to Guide
Management of
Gestational Diabetes
in Caucasian Women
Response to Schaefer-Graf et al.
S
chaefer-Graf et al. (1) rightly suggest
that the management of women
with gestational diabetes mellitus
(GDM) should be based on more than gly-
cemic factors alone. However, the study
in which they conclude that strict glyce-
mic control is not useful in the absence of
the measurements of fetal overgrowth has
two unusual features.
The authors criteria for diagnosing
GDM are lower than those of the World
Health Organization or the American Di-
abetes Association, and they may be in-
cluding women who could be considered
normal. This is supported by the fact that
there is little difference between the rate
of large-for-gestional-age (LGA) babies in
both their groups (12.1% above the 90th
percentile) and that of the dening nor-
mal population (10%). These women
may not have been equally represented in
the two groups, as 30% more women in
the ultrasound-dened group met their
criteria for insulin therapy.
More importantly, the women with
larger babies were treated more aggres-
sively. Because their outcome was not dif-
ferent from less aggressively treated
women without initially large babies,
there is the possibility that the latter might
have done better if they had been using
the same glycemic targets as the women
with large babies. They certainly had
higher fasting glucose concentrations,
and more of the LGA babies were born to
women who did not receive insulin in ei-
ther group. At the very least, the protocol
favors the authors hypothesis that insulin
intervention before the onset of fetal over-
growth is not helpful.
Jovanovic (2) draws attention to the
discrepancy in glucose targets but worries
that her own policies, based on strict gly-
cemic control and resulting in a lower
macrosomia rate in her GDM population
than in the local background population,
may have achieved this at the cost of an
increase in small-for-gestational-age
(SGA) babies. For us, in a clinic of pre-
dominantly African and Carribean back-
ground, SGA is not an issue. Using
Jovanovics protocols in women diag-
nosed by World Health Organization cri-
teria, our SGA rate is 5%. Because SGA
is dened as babies in the lowest 10th
percentile, our current rate is lower than
that of the background population.
Schaefer-Graf et al.s study supports
the use of strict glycemic control in
women with demonstrably large babies.
This is valuable evidence that such prob-
lems may respond to intervention with
insulin even after the onset of fetal over-
growth. Reassuringly, none of the differ-
ences in the rates of SGA in the study
approached signicance. We would not,
therefore, support the authors suggestion
that we abandon glycemic indications for
insulin therapy in our population until a
randomized prospective trial using simi-
larly strict glucose targets in all groups
shows that it does more harm than good.
STEPHANIE A. AMIEL, BSC, MD, FRCP
1
MAGGIE BLOTT, FRCOG
2
From the
1
Department of Diabetes, Kings College
Hospital, London, U.K.; and the
2
Royal Victoria In-
rmary, Newcastle upon Tyne, U.K.
Address correspondence to Prof. S.A. Amiel, BSc,
MD, FRCP, Kings College, Denmark Hill Campus,
New Medical School Building, Bessemer Road, Lon-
don SE5 9PJ, U.K. E-mail: stephanie.amiel@kcl.
ac.uk.

References
1. Schaefer-Graf UM, Kjos SL, Fauzan OH,
Buhling KJ, Siebert G, Buhrer C, Laden-
dorf B, Dudenhausen JW, Vetter K: A ran-
domized trial evaluating a predominantly
fetal growthbased strategy to guide
management of gestational diabetes in
Caucasian women. Diabetes Care 27:297
302, 2004
2. Jovanovic L: Never say never in medicine:
confessions of an old dog (Editorial). Di-
abetes Care 27:610612, 2001
A Randomized Trial
Evaluating a
Predominantly Fetal
GrowthBased
Strategy to Guide
Management of
Gestational Diabetes
in Caucasian Women
Response to Amiel and Blott
W
e appreciate Amiel and Blotts (1)
interest in our study (2) and
thank them for their helpful dis-
cussion. They rightly pointed out that the
diagnostic criteria used in Germany at the
time of our study had been lower than
those of the American Diabetes Associa-
tion. We were also aware that we treated
some women who would have been con-
sidered healthy in other countries. To
demonstrate that the fetal growthbased
approach is also safe for women with a
more severe glucose intolerance, we re-
peated the analysis for a subgroup of
women who fullled the diagnostic crite-
ria of Carpenter and Coustan (3) (80% of
our population). Again, there was no ad-
verse outcome in the ultrasound group.
The limitation to women who qualied
for gestational diabetes mellitus (GDM)
based on Carpenter and Coustan criteria
diminished the difference in the rate of
insulin therapy. Because of the higher de-
gree of glucose intolerance in these
women, more women qualied for insu-
lin therapy in the standard group. We also
Letters
realized that in the whole population, the
rate of women who met the criteria for
insulin was higher in the ultrasound than
in the standard group. However, this was
not because of an unequal representation
of fetal macrosomia, as suggested by the
authors. The rate of an abdominal cir-
cumference of 75th percentile at entry
was not different between both groups. It
was more likely due to our lowdiagnostic
criteria, which included women with a
low degree of glucose intolerance and,
therefore, less frequent need of insulin.
The low rate of large-for-gestational-age
(LGA) babies was an effect of intensive
treatment. The studies of Langer et al. (4)
and Buchanan et al. (5) demonstrate that a
normal LGA rate can be achieved in GDM
when we aim for a very tight control.
We were very grateful that Lois Jo-
vanovics comment (6) emphasized the
aspect of an increased risk for intrauterine
growth retardation in some women with
GDM. When we discuss treatment strate-
gies for GDM, we always focus on how to
reduce neonatal macrosomia. When we
did a subanalysis comparing the outcome
of women with hyperglycemia but nor-
mal fetal growth treated with (standard
group) or without (ultrasound group) in-
sulin, we also attempted to determine
whether withholding insulin would in-
crease the LGA and cesaerean section
rates. We were surprised when we real-
ized the high rate of SGA in the standard
group. The overall rate of SGAwas similar
in both study groups. We did not con-
clude that the policy of strict glycemic
control is harmful for the general popula-
tion with GDM. We suggested the use of
an additional test, the measurement of the
fetal abdominal circumference, to decide
who will benet and for whom a tight
glucose control might be harmful due to a
reduction of the maternal fuel supply to
the fetus. Although the differences be-
tween the groups were obvious, when we
looked at the outcome of the women who
were treated differently according to the
study protocol, the small sample sizes re-
sulting fromthe subanalysis impaired our
possibilities of an adequate statistical
analysis. In the future, we hope to initiate
and cooperate in multicenter studies to
gather more evidence to prove our results.
UTE M. SCHAEFER-GRAF, MD, PHD
1
From the
1
Department of Obstetrics, Vivantes Med-
ical Center, Berlin, Germany.
Address correspondence to Dr. Ute M. Schaefer-
Graf, Department of Obstetrics, Vivantes Medical
Center, Mariendorfer Weg 28, 12051 Berlin, Ger-
many. E-mail: ute.schaefer-graf@vivantes.de.

References
1. Amiel S, Blott M: A randomized trial eval-
uating a predominantly fetal growth
based strategy to guide management of
gestational diabetes in Caucasian women
2. Schaefer-Graf UM, Kjos SL, Fauzan OH,
Buhling KJ, Siebert G, Buhrer C, Laden-
dorf B, Dudenhausen JW, Vetter K: A ran-
domized trial evaluating a predominantly
fetal growthbased strategy to guide
management of gestational diabetes in
Caucasian women. Diabetes Care 27:297
302, 2004
3. Carpenter MW, Coustan DR: Criteria for
screening tests for gestational diabetes.
Am J Obstet Gynecol 144:768773, 1982
4. Langer O, Levy J, Brustman L, Anyaegbu-
nam A, Merkatz R, Divon M: Glycemic
control in gestational diabetes mellitus:
how tight is tight enough: small for gesta-
tional age versus large for gestational age?
Am J Obstet Gynecol 161:646653, 1989
5. Buchanan TA, Kjos SL, Montoro MN, Wu
PY, Madrilejo NG, Gonzalez M, Nunez V,
Pantoja PM, Xiang A: Use of fetal ultra-
sound to select metabolic therapy for
pregnancies complicated by mild gesta-
tional diabetes. Diabetes Care 17:275
283, 1994
6. Jovanovic L: Never say never in medicine:
confessions of an old dog (Editorial). Di-
abetes Care 27:610612, 2001
Multiple Symmetric
Lipomatosis: A
Paradigm of
Metabolically
Innocent Obesity?
Response to Haap et al.
T
he study of Haap et al. (1) on two
patients with multiple symmetric li-
pomatosis (MSL), also known as the
Launois-Bensaude Syndrome, adds inter-
esting new insights into a condition still
poorly understood.
As for the conclusions drawn con-
cerning insulin sensitivity in patient 1, we
would like to suggest some caution. Our
study group has also measured the insulin
sensitivity index (ISI) in three subjects
wi t h MSL wi t h t he eugl ycemi c-
hyperinsulinemic clamp and found the
ISI to be higher in one and lower in two
male subjects in comparison with sex-
and BMI-matched control subjects (2).
The good insulin sensitivity in patient 1 in
the study of Haap et al. may indeed be
related to the low visceral fat mass in this
very patient, but it is rather questionable
whether this nding is typical in MSL.
From ve patients with MSL identied in
our clinic, computed tomographies of the
abdomen had been performed in two case
subjects, and the degree of visceral fat was
described as comparable with that of
healthy subjects. In a systematic evalua-
tion of the fat mass and deposition in 18
patients with MSL by computed tomogra-
phy, Enzi et al. (3) also did not describe a
distinct reduction of visceral fat in these
patients. Given this and the considerable
phenotypical and clinical variability in
patients with MSL (4), the conclusion of a
good insulin sensitivity in MSL patients
should not be generalized. Studies in
larger populations are mandatory, al-
though not easily performed due to the
rareness of the disease. In a literature
search of our own, we identied about
400 cases published since the rst de-
scription by Brodie in 1846 (5).
One further aspect on MSL seems
worth mentioning: three of the ve pa-
tients with MSL in our clinic had obstruc-
tive sleep apnea syndrome (OSAS). We
recently identied (5) the presence of
OSAS as an independent risk factor con-
tributing to insulin resistance. Thus, the
investigation for OSAS in MSL patients
might also be useful in the discussion of
insulin sensitivity and of other metabolic
aspects of MSL.
IGOR A. HARSCH, MD
1
SIMIN POUR SCHAHIN, MD
1
RODERICH WIEDMANN, MD
2
From the
1
Department of Medicine I, Friedrich-
Alexander University Erlangen-Nuremberg, Erlan-
gen, Germany; and the
2
Department of Surgery,
Friedrich-Alexander University Erlangen-Nuremberg,
Erlangen, Germany.
Address correspondence to Igor Alexander Harsch,
MD, Department of Medicine I, Friedrich-Alexander
University, Division of Endocrinology and Metabo-
lism, Ulmenweg 18, 91054 Erlangen, Germany. E-
mail: igor.harsch@med1.imed.uni-erlangen.de.
Letters

References
1. Haap M, Siewecke C, Thamer C,
Machann J, Schick F, Ha ring H-U, Szei-
mies RM, Stumvoll M: Multiple symmet-
ric lipomatosis: a paradigmof metabolically
innocent obesity? (Brief Report). Diabetes
Care 27:794795, 2004
2. Harsch IA, Pour Schahin S, Fuchs FS,
Hahn EG, Lohmann T, Konturek PC,
Ficker JH: Insulin resistance, hyperlep-
tinemia, and obstructive sleep apnea in
Launois-Bensaude syndrome. Obes Res
10:625632, 2002
3. Enzi G, Biondetti PR, Fiore D, Mazzoleni
F: Computed tomography of deep fat
masses in multiple symmetrical lipomato-
sis. Radiology 144:121124, 1982
4. Enzi G, Busetto I, Ceschin E, Coin A,
Digito M, Pigozzo S: Multiple symmetrical
lipomatosis: clinical aspects and outcome
in a long-term longitudinal study. Int J
Obes 26:253261, 2002
5. Brodie BC: Lectures Illustrative of Various
Subjects in Pathology and Surgery. London,
Longman, Brown, Green, and Longman,
1846, p. 275282
6. Harsch IA, Pour Schahin S, Radespiel-
Tro ger M, Weintz O, Jahreiss H, Fuchs FS,
Wiest GH, Hahn EG, Lohmann T, Kon-
turek PC, Ficker JH: Continuous positive
airway pressure treatment rapidly improves
insulin sensitivity in patients with obstruc-
tive sleep apnea syndrome. Am J Respir Crit
Care Med 169:156162, 2004
Multiple Symmetric
Lipomatosis: A
Paradigm of
Metabolically
Innocent Obesity?
Response to Harsch, Schahin, and
Wiedmann
W
e have recently suggested (1) that
patients with multiple symmet-
ric lipomatosis (MSL) may repre-
sent a paradigm for metabolically
innocent fat accumulation. Harsch and
colleagues (2,3) raise some justied
words of caution with respect to the gen-
eralization of this statement. It is probably
true that we cannot make general meta-
bolic inferences on the syndrome of MSL
as such based on our rather limited sam-
ple size. But this was not our intention,
and we welcome the opportunity to clar-
ify our line of argument.
The original idea was to present a sit-
uation where a substantial increase in
body fat mass was not accompanied by
decreased insulin sensitivity, which is in
much contrast to common belief. And in
fact, fat accumulation, if conned to the
subcutaneous compartment, was not as-
sociated with much insulin resistance. By
selecting these two specic patients for
arguments sake, we may have introduced
a bias. But this is beside the point because
even if everything we said held true for
only these two subjects and no other pa-
tient with MSL, our ndings would be no
less interesting. And this is the reason why
we specically put paradigm in the title
rather than the syndrome alone. Never-
theless, Harsch et al. (2) appropriately
pointed out that fat accumulation in MSL
need not necessarily be conned to the
subcutaneous compartment and that,
without doubt, if it included the visceral
compartment, insulin resistance would
be present.
Since our observation was accepted
for publication, we were able to recruit
two more patients with the clinical ap-
pearance of MSL and very little visceral
fat. The whole-body fat volume of the two
subjects was 14 and 20 l, respectively.
The subcutaneous-to-visceral abdominal
fat ratio as measured by magnetic reso-
nance imaging was 3.3 and 5.9, respec-
tively, compared with BMI-, age-, and
sex-matched control groups (1.6 and 1.4,
respectively). Consistent with the small
visceral fat mass, liver fat was much lower
in both patients. Intramyocellular lipids
in soleus and tibialis anterior muscles
were also lower than those in the control
groups. And again, both MSL patients
were substantially more insulin sensitive
than their respective control groups. For
the sake of completeness we also came
across a subject with clinical MSL and
type 2 diabetes. But there are a number of
reasons, which have little to do with our
line of argument, for such a patient to ac-
quire secondary types of diabetes, for ex-
ample, via alcohol-induced pancreatitis.
Finally, the comment of Harsch et al.
regarding a higher prevalence of obstruc-
tive sleep apnea syndrome (OSAS) among
patients with MSL is interesting and valu-
able. OSAS seems to be an independent
risk factor for insulin resistance (4). Upon
reexamination of our small cohort, we in-
deed identied a man with documented
OSAS (not among the ones reported).
This patient appeared to be insulin sensi-
tive and otherwise healthy, but due to his
extreme obesity (BMI 64 kg/m
2
) would
not t into the magnetom, and we were
unable to assess his visceral fat mass. This
aspect is not necessarily in conict with
our original hypothesis but just another
mechanismto render someone insulin re-
sistant. Probably, among two equally
obese subjects with OSAS, the one with
only subcutaneous fat accumulation is
more insulin sensitive.
In conclusion, at some point the rea-
soning becomes circular and our way of
presenting these selected patients with
MSL may be just another way of demon-
strating that subcutaneous fat is metabol-
ically more innocent than visceral fat (5).
MICHAEL HAAP, MD
1
CLAUS THAMER, MD
1
HANS-ULRICH H ARING, MD
1
ROLF MARKUS SZEIMIES, MD
2
MICHAEL STUMVOLL, MD
1,3
From the
1
Department of Endocrinology, Metabo-
lism and Pathobiochemistry, University of Tu -
bingen, Tu bingen, Germany; the
2
Department of
Dermatology, University of Regensburg, Regens-
burg, Germany; and the
3
Third Medical Depart-
ment, University of Leipzig, Leipzig, Germany.
Address correspondence to Michael Stumvoll,
MD, Third Medical Department, University of
Leipzig, Philipp-Rosenthal-Str. 27, 04301 Leipzig,
Germany. E-mail: michael.stumvoll@medizin.uni-
leipzig.de.

References
1. Haap M, Siewecke C, Thamer C,
Machann J, Schick F, Haring HU, Szeim-
ies RM, Stumvoll M: Multiple symmetric
lipomatosis: a paradigm of metabolically
innocent obesity? (Brief Report). Diabetes
Care 27:794795, 2004
2. Harsch IA, Michaeli P, Hahn EG, Ficker
JH, Konturek PC: Launois-Bensaude syn-
drome in a female with type 2 diabetes.
Med Sci Monit 9:CS5CS8, 2003
3. Harsch IA, Schahin SP, Wiedman R: Mul-
tiple symmetric lipomatosis: a paradigm
of metabolically innocent obesity? (Let-
ter). Diabetes Care 27:18491850, 2004
4. Harsch IA, Schahin SP, Radespiel-Troger
M, Weintz O, Jahreiss H, Fuchs FS, Wiest
GH, Hahn EG, Lohmann T, Konturek PC,
Ficker JH: Continuous positive airway
pressure treatment rapidly improves insu-
lin sensitivity in patients with obstructive
sleep apnea syndrome. Am J Respir Crit
Care Med 169:156162, 2004
5. Montague CT, ORahilly S: The perils of
portliness: causes and consequences of vis-
ceral adiposity. Diabetes 49:883888, 2000
Letters
Has RoboCop Got
Diabetes?
W
e received a picture series from
Becton Dickinson (Fig. 1A, here
transformed to black and white
and translated from Swedish) showing
how syringes look after one, three, and
ve injections. We were astonished at
how deformed the syringes were after a
few injections.
To test the reproducibility of the sy-
ringe deformation, one physician and two
nurses, all with ordinary builds and with-
out scars or metal implants, tested Becton
Dickinsons insulin pen syringes Micro-
Fine 31 gauge 8 mm. We used the
syringes 1, 3, 5, and 10 times for injec-
tions in the abdominal fat. One syringe
was thereafter used to penetrate an ordi-
nary rubber mousepad 100 times, then to
cut into a wooden computer table, and,
nally, to cut a metal lamp foot. Pictures
were taken of the syringes with a Zeiss
Axioscope 2.5 lens with a lateral light
source and a Canon 10D camera with 6.2
mpx resolution. The only syringe that
looked similar to the used syringes in Bec-
ton Dickinsons picture series was the one
that cut into a metal lamp foot. The sy-
ringes used 0, 1, 3, 5, 10, and 100 times
are without distortion of the tip (Fig. 1B).
Hence, Becton Dickinson has im-
proved the quality of their insulin sy-
ringes dramatically since the pictures
were published, the syringes have been
manipulated to produce false evidence of
syringe defects, or Robocop was used as a
test subject.
BO BERGER, MD
PAVEL BURIAN, MD
KRISTINA NILSSON
MONA KARLEN
ELISABETH RYLANDER
From the Department of Medicine, Skovde Central
Hospital, Skovde, Sweden.
Address correspondence to Dr. Bo Berger, Med.
klin, KSS, S-54185 Sko vde, Sweden. E-mail:
bo.berger@vgregion.se.
Has RoboCop Got
Diabetes?
Response to Berger et al.
W
e thank Berger et al. (1) for rais-
ing questions about the serious
issue of reusing insulin syringes
and pen needles. The source of the pho-
tographs used by Becton Dickinson is an
independent study conducted in 1997 by
Dr. Jacques Garden, Chairman of the De-
partment of Metallurgy at the University
of Grenoble, France (unpublished obser-
vation).
In Gardens study, 80 diabetic pa-
tients fromthe Diabetes Service of Dr. Die-
ter Look returned their used pen needles
and indicated the number of times each
specic needle had been used. All needles
were used in a customary fashion and
spanned the range of manufacturer types,
gauges, and needle lengths. Needles were
used from 2 to 38 times. Needle tips were
then resterilized for safety reasons and ex-
amined by electron microscopy (2).
The tip damage found by Garden
ranged from mild bending of the extreme
tip to a hook-like distortion of the entire
distal shaft of the needle. Loss of the tip
was seen on one occasion. Some needles
appeared undamaged, even after many re-
uses. (In this regard Gardens ndings
agree with those of Berger et al.) However,
a clear association was seen between the
severity of tip damage and the number of
needle reuses.
Among the needles, some looked just
like the ones provided by Berger et al.
Some were much worse than the ones we
have chosen to show the public. We are
unsure why the results of Berger et al. dif-
fer from our own. Among the possible
causes are 1) the magnication used was
Figure 1The picture series distributed by Becton Dickinson (A), and our own pictures of used syringes (B).
Letters
at least an order of magnitude less than
that of Garden (300), 2) the health
care professionals in their study may be
more careful than patients when perform-
ing injections or may use different tech-
niques, or 3) they viewed fewer samples,
and sampling error may have resulted.
We would welcome a full discussion of
this study with Berger et al.
We advise against pen needle and in-
sulin syringe reuse for a number of rea-
sons. First, the syringes may become
hooked, as we have previously described.
The hook at the tip of the syringe may
break off and become lodged in the skin,
and the pain of injection with a syringe
increases with each reuse.
Second, studies conducted by a series
of nurses in Europe have shown signi-
cant increases in lipodystrophy in pa-
tients who reuse insulin syringes and pen
needles (3). Third, there are reports of in-
creased infections from reuse of insulin
syringes (4). Finally, reusing syringes can
lead to suboptimal therapy. With sy-
ringes, the patient can bring signicant
amounts of NPHinto the regular or lispro
vial. With pens, leaving the needle on be-
tween uses can lead to incomplete injec-
tions, which may cause the patient to
underdose themselves and receive as little
as 35% of the expected dose (5).
We believe the issue of insulin syringe
and pen needle reuse to be serious, with
real safety concerns that have not been
fully explored in the past. We welcome a
more complete scientic discussion of
this issue.
BARRY H. GINSBERG, MD, PHD
KENNETH STRAUSS, MD
From Becton Dickinson Medical Systems, Diabetes
Care, Franklin Lakes, New Jersey.
Address correspondence to Dr. Barry H. Gins-
berg, BD Medical Systems, VP Medical Affairs, Dia-
betes Care, 1 Becton Dr., Franklin Lakes, NJ 07481.
E-mail: barry_ginsberg@bd.com.

References
1. Berger B, Burian P, Nilsson K, Karlen M,
Rylander E: Has RoboCop got diabetes?
2. Look D, Strauss K: [Nadeln mehrfach ver-
wenden?] Diabetes Journal 10:S31S34,
1998
3. Strauss K, De Gols H, Hannet I, Partanen
TM, Frid A: A pan-European epidemio-
logic study of insulin injection technique
in patients with diabetes. Pract Diab Int
19:7176, 2002
4. Paily R: Perinephric abscess from insulin
syringe reuse. Am J Med Sci 327:4748,
2004
5. Parkes JL, Slatin SL, Pardo S, Ginsberg
BH: A new consensus error grid to evalu-
ate the clinical signicance of inaccuracies
in the measurement of blood glucose. Di-
abetes Care 23:11431148, 2000
Letters

Joffe Dyslip DM Sa

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Joffe Dyslip DM Sa

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Joffe Dyslip DM Sa

Uploaded by

Copyright:

Available Formats

OBSERVATIONS

J. SCHEEN, MD, PHD

You might also like