Background: Visceral (VAT) and subcutaneous abdominal (SAAT) and trunk (STRAT) adipose tissue (AT) have been suggested to be differentially influenced by diet.
Objective: We investigated whether and to what extent usual patterns of nutrient intake are associated with VAT, SAAT, and STRAT compared with nondietary predictors in northern German adults (n = 583).
Design: AT volumes were quantified by magnetic resonance imaging. Nutrient intake was estimated by a 112-item food-frequency questionnaire linked to the German Food Code and Nutrient Database. Exploratory nutrient patterns were derived by principal components analysis (PCA) and partial least-squares regression (PLS) of 87 nutrients. Cross-sectional associations between nutrient patterns, single nutrients, or total energy intake and AT compartments were analyzed by multiple linear regression.
Results: Next to sex and age, respectively, which were important nondietary predictors and accounted for more of the variation in VAT (∼13% and ∼4%) than in SAAT or STRAT (both 4–7% and <1%), variation in VAT (16.8% or 17.6%) was explained to a greater extent by 9 or 2 nutrient patterns derived by principal components analysis or partial least-squares regression, respectively, than variation in SAAT (10.6% or 8.2%) and STRAT (11.5% or 8.6%). Whereas VAT (16.6%) was primarily explained by nutrient quality, SAAT (6.9%) and STRAT (7.4%) were mainly explained by total energy intake. VAT was positively associated with nutrients characteristic of animal (except for dairy) products, including arachidonic acid (standardized β: 0.25; 95% CI: 0.15, 0.34; P < 0.0001), but negatively with dietary fiber, including polypentoses (standardized β: −0.17; 95% CI: −0.24, −0.09; P < 0.0001), and nutrients found in milk. The direction and strength of many associations, however, depended strongly on sex and adjustment for BMI.
Conclusion: VAT may be particularly associated with sex-specific interplays of nutrients found in animal products and fiber, whereas SAAT and STRAT are associated with total energy intake.
Alex’s Notes: It is well known that visceral adipose tissue (VAT) is a highly metabolically active and inflammatory source of trouble in the realm of health. While a lot of research has been devoted to figuring out the role of VAT in human health, the puzzle of subcutaneous fat is far less complete. Regardless, something often wondered by health authorities and layman alike is to what extent, if any, dietary and non-dietary patterns are related to VAT and subcutaneous abdominal (SAAT) and trunk (STRAT) adipose tissues.
There are two principle methods for investigating this relationship: single-nutrient and food-pattern. What makes the current study unique is that it combines these two methods into “nutrient patterns,” which is an attempt to get the best of both worlds. by capturing the interrelations and synergy effects of multiple aspects of diet, the pattern approach integrates the complex mechanisms and physiologic pathways of the effect of diet on health. The difference between this and looking at a single-nutrient is obvious. The advantage over food patterns, which are more easily translated into public recommendations, however, is that nutrients are not functionally exchangeable (you cannot replace vitamin C with another vitamin like you can replace bread with a potato) and nutrients are globally consumed (the same cannot be said for Surströmming).
The researchers gathered data from 583 individuals of the PopGen control cohort of northern Germany. Usual food and nutrient intake over the previous year was determined from a 112-item food frequency questionnaire analyzed by the German Food Code and Nutrient Database. In total, the researchers had 133 single-nutrient or nutrient group items to use in statistical analysis against blood-borne markers of health and anthropometric measurements. This included whole-body MRI determined VAT, SAAT, and STRAT. Importantly, to help reduce any potential confounding, all statistical models were controlled for age, sex, physical activity, smoking status, and either BMI (model 1) or height (model 2).
Nine major principal components representative of independent behavior-related nutrient patterns were identified, with no differences between men and women
These nine PCAs are listed in the table that follows. Importantly, each nutrient item was adjusted for total energy intake of the individual so as to give measure of nutrient density.
|PCA-1||Short-, medium, and long-chained fatty acids in the diet, especially in milk fat or other animal (except for fish) fat.|
|PCA-2||Nutrients found particularly in meat and egg products, including arachidonic acid, amino acids, and numerous vitamins and minerals|
|PCA-3||Fatty acids from fish oil and vitamin D|
|PCA-4||Key nutrients of vegetables and fruits such as cellulose, vitamin C, folate, vitamin K, and potassium|
|PCA-5||Long-chained fatty acids present in vegetable oils such as linoleic acid|
|PCA-6||Carbohydrates and fiber found in cereal grains and starchy tubers|
|PCA-7||Glucose, fructose, sucrose, and organic acids|
|PCA-8||Lactose & calcium|
|PCA-9||Alcohol and carbohydrates found in beer (maltose)|
Of the total variation in nutrient intake among the cohort, 84.2% could be explained by differences among the PCA intakes, and the PCAs are listed (1à9) in order of decreasing magnitude. This is actually quite interesting when you consider that those groups higher on the list are the ones with the most variation between people. In other words, alcohol intake (PCA-9) is relatively stable across persons, while dietary fat (PCA-1) can vary greatly.
PLS variables were then formed that combined the PCAs with the 3 markers of fatness in order to determine what dietary pattern maximally explained differences in VAT, SAAT, & STRAT.
The first PLS pattern (PLS-1) was positively correlated with all nutrient items and explained 45.9% of the variation in nutrient intake. However, it only explained 1% of the variation in VAT while explaining 6.4% and 6.7% of the variation in SAAT and STAT, respectively.
By contrast, the second PLS pattern (PLS-2) correlated positively with nutrients found in meat and egg products, fish, and beer, but negatively with nutrients found in plant and dairy products, and only explained only 7.1% of the variation in nutrient intake but 16.6% of the variation in VAT and only 1.9% of the variation in both SAAT and STRAT.
I know the above is somewhat confusing, so to put it in other words, the nutrients present in meat and egg products, fish, and beer were positively predictive of VAT, and to a minor extent, nutrients found in meat and egg products were positively predictive of SAAT and STRAT. By contrast, nutrients found in plant and dairy products were negatively predictive of VAT and, to a lesser extent, also of SAAT and STRAT.
Model 2 adjusted for height so that nutrient intakes are associated only with weight
Of all the dietary predictors, PCA-2 and PLS-2 were the only ones to be positively associated with VAT, SAAT, and STRAT, indicating that as nutrients from these areas increased, so did the tendency for greater fat. Sex was another strong predictor, explaining 14% of the variation in VAT alone. Men had more VAT and less SAAT and STRAT than did women. Collectively, the nine PCAs and two PLSs explained 16.8% & 17.6% the variation in VAT, 10.6% and 8.2% the variation in SAAT, and 11.5% and 8.6% the variation in STRAT, suggesting that diet plays a greater role in the accumulation of VAT than SAAT or STRAT.
Model 1 adjusted for BMI, removing weight status from the influencing variables
Things look quite different under these conditions. For instance, VAT was negatively associated with PCA-8 and positively with PLS-2, the latter explaining 16.8% of the variance in VAT. Sex also played a role as PCA-6 was negatively associated with VAT in men but not women. VAT was explained to a greater extent by nutrient quality (16.6%) than by total energy intake (1.4%), whereas SAAT (6.9%) and STRAT (7.4%) were explained to a larger proportion by total energy intake than by nutrient quality (both 1.9%).
Overall, about one-third of the 133 single-nutrient or nutrient group items explored were related to VAT, SAAT, or STRAT in BMI- or height-adjusted models.
I know that this article is somewhat confusing, and I completely agree. I had to read it about ten times, go look up certain statistical crap on Google, and sit in thought to try and wrap my head around it. After all that, the real takeaway is somewhat obvious and disappointing. That is, VAT is explained largely by nutrient quality, whereas SAAT and STRAT are explained largely by total energy intake. Moreover, all three AT variables are positively associated with animal products and their nutrients (except for dairy).
When adjustment for general obesity and weight status as determined by BMI was made, many of the nutrient items were no longer associated with AT variables, but those that were had associations more specific to either VAT or SAAT and STRAT. The direction of these associations was dependent on sex, age, and BMI.