Sunday, July 26, 2015

Black et al (Low Fat Diet for Skin Cancer)

Hooper et al included this trial by Black et al as a reduced fat trial in their 2012 meta-analysis and as a reduced SFA trial in their 2015 meta-analysis

Studies Associated with the Trial

Effect of a Low-Fat Diet on the Incidence of Actinic Keratosis (1994) [1]
Evidence that a low-fat diet reduces the occurrence of non-melanoma skin cancer (1995) [2]
General guidelines for a low-fat diet effective in the management and prevention of nonmelanoma skin cancer (1997) [3]

Purpose

Ecological studies at the time and animal studies found a relationship between fat intake and the incidence of skin cancers and actinic keratosis (potentially precancerous tissue) [1] [2]

Methods

133 people with a previous diagnosis of skin cancer (define) were randomised to a low fat group or a control group.

The low fat group attended 8 weekly classes where they were advised to reduce fat intake to 20%, keep protein at 15% and increase carbohydrate to 65% and to avoid calorie deficit.  Foods containing sufficient vitamins and minerals to meet the recommended dietary allowances for adults were emphasized. The patients were taught behavioral techniques such as stimulus control, self-monitoring, and cognitive behavioral self-management to help them comply with the dietary regimen” [1] [2] [3].  “The primary strategy recommended for increasing carbohydrate intake was to increase consumption of grains, fruits, and vegetables. This strategy introduces an additional potential variable; i.e., a statistically significant inverse relationship has been reported between nonmelanoma skin cancer and high vegetable intake”. [3]

“The diet of the patients assigned to the control group was not changed” [1]

Results

Fat intake was maintained well at a little over 20% and despite advice to increase complex carbohydrates to maintain weight (which they did: 264.5g vs. 303.5g), the low fat group had a significant decrease in calorie intake and significant weight loss [1] [2] [3]

  
Consistent with the advice to consume more fruits, vegetables and grains, the low fat group increased their fibre and vitamin C intake, but not by much [3]


The low fat group had a significantly lower incidence of actinic keratoses [1] and total skin cancers [2] [3]


Hooper et al included this trial as a reduced fat trial in their 2012 meta-analysis and as a reduced SFA trial in their 2015 meta-analysis.  Similar to the other low fat trials, while it’s very appropriate to include this as a reduced fat trial, it is less appropriate to include this as a reduced SFA trial due to the simultaneous decreases in MUFA and PUFA as well.  Hooper et al reports the following results in their meta-analyses.  The incidence of CVD events and CVD mortality was not published, but it was reported that 1 person died in the low fat group and 2 people died in the control group.  They also flagged the trial as having high risk of ‘free of systematic difference in care’ and low risk of ‘free of dietary differences other than fat’ 


Low Fat Group (N = 66)
Control Group (N = 67)
CVD Events
0
2
CVD Mortality
0
2
Total Mortality
1
2

Sunday, July 19, 2015

The MeDiet Project (Mediterranean Diet for Endogenous Estrogens)

The meta-analysis by Hooper, et al 2012 included 4 other fat modification trials, the last of which is this trial called MeDiet (short for Mediterranean Diet)

Studies Associated with the Trial

The Mediet Project (2002) [1]
A traditional Mediterranean diet decreases endogenous estrogens in healthy postmenopausal women (2006) [2]
Dietary enterolactone affects androgen and estrogen levels in healthy postmenopausal women (2009) [3]

Methods

The MeDiet project investigated whether a Mediterranean diet would reduce estrogen levels as: (1) previous research has found that estrogen levels are positively associated with the initiation and development of breast cancer; and (2) observational studies that found the incidence of breast cancer to be lower in southern Italy compared to northern Italy, which was related to dietary differences [1] [2]

106 postmenopausal women (aged 44-71) who met the eligibility criteria* were randomised to a Mediterranean diet group (the experimental group) or a control group for 1 year [1] [2]

“Women in the intervention group adhered to a traditional, restricted Mediterranean diet for 6 mo, whereas women in the control group continued to follow their regular diet. In particular, intervention women attended a weekly “cooking course,” followed by a social dinner, and were trained by professional chefs in the correct use of natural ingredients of a traditional Mediterranean diet. Women were then instructed to consume the same food at home using a series of recipes based on a traditional Sicilian diet. This diet included whole cereals, legumes, seeds, fish, vegetables, and other Mediterranean seasonal food containing several biologically relevant anticancer agents. Furthermore, women were asked to avoid the use of refined carbohydrates and additional animal fat, and to limit the use of salt. The dietary intervention was intended to 1) reduce the intake of both refined sugar and saturated and total fat; 2) increase the consumption of mono- and polyunsaturated fat; 3) increase the intake of fruits and vegetables (notably cruciferous plants); and 4) increase the intake of food rich in phytoestrogens.” [2]

Macronutrient intake is presented below.  In summary, both groups significantly decreased their calorie intake though to a greater extent in the experimental group (302 vs. 164).  In the experimental group this was achieved by reducing intake of animal protein, animal fat and carbohydrates, whereas the control group just reduced intake of protein and carbohydrate.  In both groups the significant reduction in carbohydrates was due to reductions in both starch and sugar [2]

  
* “Eligibility criteria were: (1) postmenopausal for at least 2 years; (2) no history of bilateral ovariectomy; (3) no hormone replacement therapy for the previous 1 year; (4) no history of cancer; (5) no adherence to a vegetarian or macrobiotic diet or to any other medically prescribed diet; and (6) no treatment for diabetes, thyroid disease, and chronic bowel disease” [1]

Results

The experimental diet significantly reduced several estrogen metabolites in urine (except E2 which significantly increased) as well as total urinary estrogen (see table below) [2].  Total urinary androgens slightly increased (NS) [3].  There were also some positive correlations between androgens and enterolactone* intake [3]

* Lignin precursors of enterolactone are found in the fibre of plant foods and are metabolised to enterolactone by gut bacteria.  Enterolactone seems to work by inhibiting aromatase, which converts testosterone to estrogen, and therefore seems gets some attention, both in research regarding estrogen associated cancers such as breast and prostate cancer; and also as a potential anti-aging therapy for men


Neither of the three papers reported any data on endpoints.  What we have comes from Hooper, et al (who would have contacted the authors), which is presented below.  There weren’t many hard endpoints which is most likely due to the small size (N = 106), not strictly being a secondary prevention trial and its short duration (1 year)


Experimental (N = 51)
Control (N = 55)
Stroke
0
1
Combined CVD Events
0
3
Cancer Diagnosis
0
2

Hooper, et al included MeDiet as a reduced and modified fat trial, but wasn’t much of either.  Reduced fat: the experimental group reduced total fat intake from 82.2g to 67.9g, but due to the overall reduction in calories (from both protein and carbohydrate as well), the percentage of fat only decreased by 0.95% from 37.40% to 36.45%.  Modified fat: the experimental group reduced SFA intake from 27.7g to 18.4g, but there was no reciprocal increase in either MUFA or PUFA intake.  In fact, MUFA intake decreased from 40.7g to 37.2g and PUFA intake decreased from 9.1g to 8.6g.  The ‘fat modification’ was all relative to total calories.  Therefore, it’s inappropriate to include MeDiet in either a meta-analysis on reduced fat or modified fat trials, since it is neither.  Instead, it would be more appropriate to classify it as a ‘reduced calorie, reduced SFA trial with a highly multifactorial dietary intervention plus potential behaviour modification’, making it an ‘inadequately controlled trial’

Sunday, July 12, 2015

Sondergaard et al (Mediterranean Diet for Endothelial Function)

The meta-analysis by Hooper et al 2012 included 4 other trials, one of which was a trial by Sondergaard et al.


Methods

131 people with ischemic heart disease and total cholesterol ≥5 mmol/L (193 mg/dL) were randomized to receive Mediterranean dietary advice (n = 68) or no specific dietary advice (n = 63).  In addition, both groups were treated with Fluvastatin (40 mg once daily)

The Mediterranean group was advised to “eat at least 600 grams of fruits and vegetables daily, to modify the intake of fat, especially saturated fat from meat and dairy produce, to eat fatty fish at least once a week and preferably several times a week, to eat plenty of bread and cereals, and to replace refined, hard, animal margarine products with vegetable oils, preferably canola oil”

While the control group “was offered booklets about heart-healthy diets that are usually delivered to patients in the coronary care unit (CCU). They were also offered a single visit to a dietitian who was not participating in the study”

Food and macronutrient intake is presented in table 4 and the differences between the groups is summarised below.  Calorie intake wasn’t reported, but I calculated* it to be approximately 1518 and 1545 respectively

Higher in Experimental Group
Higher in Control Group
Fatty fish
Beef, veal and pork
Carbohydrate
Fat
PUFA
SFA (P = 0.06)

MUFA

There were a few differences between the groups at baseline (table 3).  The experimental group had fewer men (62% vs. 79%), more people with hypertension (35% vs. 16%) and less people on ‘long-acting nitrates’ (6% vs. 19%).  The experimental group may have had more people taking multivitamins (35% vs. 24%) as I suspect the p value of 0.3 may be a typo

* SFA, MUFA and PUFA are measured in grams and total fat is measured in % of energy.  Therefore: total calories ≈ (SFA + MUFA + PUFA) x 9 / (percentage of calories as fat / 100)

Results

As expected from the use of statins, total-C and LDL-C significantly decreased in both groups.  Triglycerides also significantly decreased to a lesser extent, and both groups experienced a minor, non-significant increase in HDL-C (table 2)

Flow mediated dilation (a measure of endothelial function) was similar at baseline and improved in the Mediterranean group to a much greater extent such that the difference at 12 months was significant (P < 0.01) (table 2)


Mediterranean
Control
Baseline
4.32
4.30
12 Months
8.62
5.72

The paper didn’t publish any data on endpoints.  What we have comes from Hooper et al (who would have contacted the authors), which is presented below.  There weren’t many hard endpoints, such as myocardial infarctions (heart attack), strokes and deaths, most likely due to the small size (N = 131) and short duration (12 months).

What’s surprising is the relatively large number of total CVD events, most of which are likely to be very soft, which was unaffected by the Mediterranean diet used in this study despite a substantial difference in flow mediated dilation


Experimental Group
(N = 68)
Control Group
(N = 63)
Myocardial Infarction
2
3
Stroke
1
2
Combined CVD Events
40
42
CVD Mortality
3
4
Total Mortality*
4
4
* The extra death in the experimental group came from cancer (which was the only cancer diagnosis)

Hooper et al classified this trial as a reduced and modified fat trial, but there were other differences as well and Hooper et al flagged the trials as not being free of differences besides fat:

·         The higher fish intake in experimental group (67 vs. 46, P = 0.03)
·         The slightly higher intake of fruits and vegetables (572 vs. 504, P = 0.13)
·         The advice to reduce “refined, hard, animal margarine products”, which may include major sources of trans fat
·         The likelihood that the experimental group replaced refined grains with whole grains, given the kind of dietary advice the Mediterranean group was given, although the study didn’t differentiate whole grains and refined grains in the sections on dietary advice and food intake

In addition, this wasn’t much of a modified or reduced fat diet.  The difference in total fat was significant, but fat intake only differed by 2.7% (26.2% vs. 28.9%).  Likewise, while SFA intake was almost significantly lower (P = 0.06) and PUFA intake was significantly higher in the Mediterranean group, SFA intake only differed by 3g (17.4g vs. 20.4g) and PUFA intake by 2.4g (13.0g vs. 10.6) (MUFA intake was 13.8g vs. 18.6g).

Sunday, July 5, 2015

Houtsmuller et al (Linoleic Acid for Diabetic Angiopathy)

Hooper et al included this trial by Houtsmuller et al in their 2012 meta-analysis as a fat modification trial and in their 2015 meta-analysis as a reduced SFA trial

Studies Associated with the Trial

Unsaturated fats and progression of diabetic retinopathy (1979) [1] (preliminary results)
Influence of different diets on the progression of diabetic retinopathy (1980) [2] (no access)
Favorable influences of linoleic acid on the progression of diabetic micro- and macroangiopathy in adult onset diabetes mellitus (1980) [3] (this post came entirely from this paper)

Methods

102 people who were recently (< 1 year) diagnosed with type 2 diabetes, but otherwise were healthy and didn’t have CHD, were stratified to one of two diet groups for 6 years (don’t know the age of participants).  Diet 1 was a control diet, high in SFA.  Diet 2 (the experimental diet) had a similar macronutrient composition but contained approximately 4 times more linoleic acid.  There is very limited dietary information, the totality of which is quoted below

“Diet I contained carbohydrate 50cal%, saturated fats 35 cal%, proteins 15 cal% (diet I), and diet II contained carbohydrate 45 cal%, fats 40 cal% (1/3 linoleic acid) and proteins 15 cal%. The linoleic acid content of diet II was 4 times that of diet I, being 20.4 gr/1000 kcal for group II and 5.3 gr/1000 Kcal for group I. The fiber content was similar in both groups as was the cholesterol content, being 88 mg/1000 Kcal in both groups except for 4 patients of group I who preferred butter over saturated margarines. The cholesterol content of the diet for these 4 patients was 148 mg/1000 Kcal.”

In the results section the authors mention that “differences in body weight between several groups were never found”, suggesting that changes in calorie intake were most likely very similar


Pro:Fat:Carb
Linoleic Acid*
Experimental
15:40:45
20.4g/1000kcal
Control
15:35:50
5.3g/1000kcal

The major source of fat for most of the participants in the control group is “saturated margarines”, which is almost certainly not of animal origin as the cholesterol intake was identical to the control group and 41% less than the participants who ate butter instead, therefore it’s very likely that the control group in this study had a much greater intake of TFA (unless they used unhydrogenated palm or coconut oil, or cocoa butter), which is also the most likely explanation for the trial’s pretty incredible results

The description of the diets not only lacks detail, but is also poor and inconsistent.  Earlier they said linoleic acid was 1/3 of total fat intake in the experimental group, which makes it 13.3% of total calories.  However, 20.4g/1000kcal is equal to 18.36% of total calories (compared to 4.77% in the control group).  The control diet is described as containing “saturated fats 35 cal%”, but this obviously can’t mean 35% of calories from SFA as 35% of calories came from fat and approximately 5% of calories came from linoleic acid.

Results

The experimental group had significantly fewer people develop mild diabetic retinopathy and ‘signs of cardiac ischaemia’ (which seems like it would be angina)


Experimental (M/F)
N = 26/22
Control (M/F)
N = 26/22
Mild Diabetic Retinopathy at Baseline
6/6 (23%/27%)
6/6 (23%/27%)
Mild Diabetic Retinopathy at 6 Years
9/7 (35%/32%)
17/12 (66%/55%)


Experimental (M/F)
N = 51
Control (M/F)
N = 51
“Developed signs of cardiac ischaemia”
4/4
14/10
Myocardial Infarction
0/0
5/1
CHD Mortality
0/0
4/1

Hooper, et al reported 8/51 vs. 30/51 for ‘combined CVD events’, which includes cardiac ischemia and myocardial infarction and reported 0/51 vs. 6/51 for myocardial infarction.  This is exactly the same as the figures I’ve got


Experimental (N = 51)
Control (N = 51)
Myocardial Infarction
0
6
Combined CVD Events
8
30

The glucose and insulin response to a glucose tolerance test improved due to some unspecified ‘treatment’/‘therapy’* in both groups.  There wasn’t any difference between the two diet groups in men, whereas in women the ‘treatment’ only slightly improved glucose levels and didn’t improve insulin levels in the control group.  It’s interesting that among men, fewer people developed mild diabetic retinopathy and angina in the experimental group, despite glucose tolerance and insulin resistance being equal


After the ‘initial therapeutic improvement’, serum lipids and/or cholesterol increased in both groups, but to a greater extent in the control group and mainly among women


In addition, there was no correlation between “platelet agrregation and linoleic acid, progression of retinopathy, sex or age, serum glucose, insulin or lipids”

In their ‘characteristics of included studies’, Hooper et al drew attention to the many unknowns in this trial, particularly regarding the dietary advice and their execution, as well as the fact that neither the participants nor the researchers were blinded, leading to a ‘high risk of bias’.

Due to the likelihood that the “saturated margarines” were rich in trans fat and may have been a major source of fat in the control diet, the control group likely had a very high intake of trans fats.  If I had to include this trial in a meta-analysis it would probably fall in the ‘adequately controlled’ category because there’s no solid evidence of this, but it would definitely be excluded alongside the Sydney Diet Heart Study as a trial that could be heavily confounded by differences in TFA intake.  After all, the results seem too good to be true: ‘eat 13.3% or 18.36% linoleic acid and reduce your risk of CHD events by 73.3%’ (very believable).  The diet heart advocates can’t have it both ways

* I’m not sure what the treatment is.  It could be the two diets, which may be calorie restricted, or perhaps it could be insulin or an insulin-sensitising drug