NS Summer Meeting 2012: Day Two –we’re all going large

Emergent ontological antireductionism across the scales

To Belfast for the NS Summer Conference for “Translational Nutrition: Integrating Research Practice and Policy” and a bit of general shindiggery. I use summer in the loosest sense. The two symposia which were held today were in the areas of “Innovation in diet and lifestyle assessment” and “intervention study design and personalised nutrition”, and we heard plenary lectures from Profs Bruce German and David Jacobs.

There was a stunning compare and contrast exercise presented by Phyllis Stumbo describing the history of diet assessment approaches, and updating with a series of programmes underway in the US to move towards pictorial capture and analysis of plated food. Inevitably this will have limitations, but the technology is developing extremely quickly and it is likely that one or more platforms will become the tools of choice (over, say, FFQ or recall) for imminent generations of nutritionist. However the NIH has invested over $40,000,000 in the various programmes and dr Stumbo contrasted this with a $2 app you can now buy for a smartphone that gives as not-bad an estimate as current state-of-the-art technologies. How does it work? Crowdsourcing and Mechanical Turks (don’t ask me, Wikipedia is your friend). Ashley Adamson raised the level of challenge further and highlighted the greater difficulty when working at extremes of age, and the greater importance of these sorts of platforms and innovations with very young and very old populations.

Of course, one question about nutrition assessment is “Why”? and as with so many areas of science, the answer will reflect the actual question being asked. The granularity of diet data obtained may reach ever greater levels of personalisation and accuracy, but the scope of the assessment period will reduce accordingly. Both Plenary speakers, Profs German and Jacobs, gave presentations suggesting that reductionist and fine grain models may not be the right approaches. In particular Prof Jacobs suggested that treating nutrients and analysing nutrition using the same approaches as would be used for a drug had been a disaster, with most nutrient interventions yielding either neutral or highly negative results. Likewise he, very rightly to my mind, suggested that viewing “food” or “diet” as only the reduced 100 or so essential elements, and viewing these as independent, was wrong thinking. From any kind of modern biology viewpoint this seems perfectly logical. His central thesis was that we should take a synergist or combinationist view of food and diet and create an entirely new infrastructure of nutrition research, free of its reductionist legacy, and embrace pattern analysis, food level analysis and food (or whole diet) RCTs, with all the difficulty, variation and imprecision entailed, is still more valuable an effort than a nutrient-led intervention. The current funding landscape must change and must embrace such approaches, suggests Jacobs.

Prof German offered an equally direct critique of how we dissociate composition from function in food. Using lactation as an example of how the major milk component is of no direct nutritional value to the neonate, but serves to promote growth of B infantis, and related examples of how we need more careful consideration of why a biomarker might be up or down (could be synthesis, excretion, intake or absorption, but one needs to know which competing process before trying, say, to treat hypercholesterolaemia).

Marga Ocke gave a very clear talk on the different emergent statistical approaches to analyse diet, and I came away with a far better understanding of the difference between a cluster and a pattern (I think). What is clear is that not only are approaches available, but that these should no longer be consigned to statistical eclecticists, but placed centre-stage in a suite of approaches to nutrition. The approaches seem to be not to be prospective, but nonetheless have the power to yield greater insight and more convincing results.

In a later part of the session Henk Hendricks and John Draper talked about the use of metabolomic approaches to the analysis of health biomarkers and as (objective, urine-based) diet assessment tools. The former talk indicated some progress being made on the very thorny issue of how close we’re getting to profiling normal health and early onset, albeit using a lengthy catalogue of markers at the moment, the second trawled through urinary markers to move towards some solid and reliable indices of intake, which may, in time, obviate traditional methods.

Over the last 1-2 years a few areas of my own research have troubled me, we are finding discord between cross-sectional and intervention studies in one area, and have found pattern analysis reveals behavioural changes that are simply not reflected in nutrient-level analysis of the same data. All this is early stage, submitted or in preparation, however it is clear that there are emerging and sound ontological reasons to take a much larger view of food, metabolism and health in order properly to understand nutrition.