REVIEW
SA JOURNAL OF DIABETES & VASCULAR DISEASE
90
VOLUME 10 NUMBER 3 • SEPTEMBER 2013
a person with diabetes prescribed an energy restricted high protein
diet has an increased or decreased risk of cardiovascular mortality. A
long-term randomised controlled trial, taking into account the type
of carbohydrate, glycaemic load, source of protein and amount of
fibre, is required to address this.
Very low-carbohydrate, high-fat diets
There has been a resurgence in interest over the last decade on
reducing carbohydrate intake. This stems to some extent from the
perceived lack of effect of the low-fat high-carbohydrate approach.
However, it ignores that central to a high carbohydrate diet is a
focus on dietary fibre and avoiding large amounts of refined
carbohydrate, neither of which are achieved in most westernised
diets. In the more extreme versions such as the “Atkins diet”, the
aim of the very low-carbohydrate, high-fat diet is to reduce initial
carbohydrate content to 20 g per day and thus induce ketosis.
43
The theory follows that this allows an individual to use fat stores
for energy instead of glucose. Variations in fat and protein content
abound, but generally the aim is 10% from carbohydrate, and
50% or more from fat, with protein intake making the up the
balance. The literature is complicated by inconsistent definitions of
low carbohydrate diet, very low carbohydrate diet and inadequate
reporting of actual intakes.
Weight loss
Studies in those with diabetes are limited, with few over a duration
longer than 12 weeks, but indicate potentially initial greater weight
loss but no difference over the longer term.
44
One non-randomised
study comparing a ketogenic very low-carbohydrate diet with a
low-calorie diet over 24 weeks did demonstrate greater weight loss,
improvements in glycaemic control and lipid profile with the very
low-carbohydrate approach.
45
However no long-term or randomised
controlled trials have demonstrated any superiority.
46
It has been
hypothesised that the generation of ketosis is central to facilitating
weight loss with a low-carbohydrate diet, however this is not
supported by Boden
47
or Krebs
et al.
48
who showed that weight loss
was directly related to the reduction of total energy intake achieved
by removing one macronutrient from the diet without appreciable
compensatory increase in actual fat or protein intake.
Glycaemic control
Significantly restricting carbohydrate in the diet has profound
effects on glucose metabolism. Several small and very short studies
ranging from seven days to five weeks have variably demonstrated
reduced glucose and insulin concentrations and improved glycaemic
control.
34,47,49,50
In short-term studies people with T2DM require
less antidiabetic agents to control glucose levels and achieve
improvements in HbA
1c
.
45,46,48
In some studies these improvements
are greater with a very low-carbohydrate diet compared with other
approaches,
51
but data are not consistent and there are no long
term studies convincingly demonstrating any greater benefit. Davis
et al.
demonstrated similar modest weight loss but no significant
improvement in glycaemic control in those with type 2 diabetes
on a low-fat or low-carbohydrate diet after one year.
52
In the
longest very low-carbohydrate diet trial currently reported, a very
low-carbohydrate diet achieved a –0.9% reduction in HbA
1c
at 2
years compared with –0.5% in a Mediterranean diet and –0.4 %
in a low fat diet, however the difference between groups was not
significant.
53
CVD risk factors
One of the concerns raised about very low carbohydrate diets is the
potential for increases in dietary saturated fat that may accompany
the foods chosen as part of this diet, will adversely affect lipid
profile. However this has not been generally observed when weight
loss is achieved, though can be seen in isolated individuals who do
not lose weight.
48
Hussein
et al.
demonstrate significant reduction
in triglycerides, total and LDL cholesterol and an increase in HDL
cholesterol over 24 weeks.
45
This global benefit is not observed in
all trials with increases in HDL cholesterol being the most consistent
observation.
52
Blood pressure has been shown to be reduced more
with very low carbohydrate diets compared with alternatives over 3
months, but no difference after 12 months.
52
Once again the only data on long-term cardiovascular outcomes
are from observational studies rather than randomised controlled
trials. These are reciprocal analyses of the protein intake as
described above.
38,39,41
One study showed impairment of flow-
mediated dilatation, but improvements in other endothelial
function markers, after 12 months on a very low-carbohydrate diet
in overweight and obese patients, but this was not specifically in
those with diabetes.
54
The Mediterranean diet
There is increasing interest in the Mediterranean diet in type 2
diabetes as results of studies such as PREDIMED become available.
The Mediterranean diet is characterised by replacing most red meat
with fish and poultry, including wine in moderation and plenty of
vegetables, legumes, grains, fruit, nuts and olive oil.
55
Impressive
evidence from the PREDIMED-Reus trial in diabetes prevention
demonstrated a diabetes incidence of 10.1% (5.1–15.1%) in the
diet plus olive oil group and 11.0% (5.9–16.1) in the diet plus nuts
group compared to 17.9% (11.4–24.4%) in the control group.
55
The EPIC-Interact project also demonstrated a positive effect, where
high adherence to the Mediterranean diet was associated with a
hazard ratio of 0.88 (0.79–0.97), compared with low adherence, in
participants over 50 years of age.
56
Weight loss
There have only been two studies examining the use of a
Mediterranean diet for weight loss in participants with type 2
diabetes.
57,58
Both studies favoured the Mediterranean diet for
greater weight loss but in one study the Mediterranean diet also
had a reduced carbohydrate content which may have had a
confounding effect.
Glycaemic control
Two cross-sectional studies analysing adherence to a Mediterranean
diet and Hba
1c
level have been conducted in Mediterranean
populations. In the Campanian Postprandial Hyperglycemia Study
(
n
= 901), high diet adherence was associated with significantly
lower HbA
1c
and 2-hour post-meal glucose concentrations
(difference: HbA
1c
0.9%, CI 0.5–1.2%,
p
< 0.001; 2-hour glucose
2.2 mmol/l, 95% CI 0.8–2.9 mmol/l,
p
< 0.001).
59
In the PREDIMED
study (
n
= 262), while a trend toward an inverse relationship
between diet adherence score and HbA
1c
was identified, it was not
statistically significant.
60
A small cross-over study (
n
= 27) compared
12 weeks on a Mediterranean diet (key foods provided) with 12
weeks on their usual diet. Compared with usual diet, the
ad libitum
Mediterranean diet resulted in a fall in HbA
1c
from 7.1% (6.5–7.7)
