In the panel discussion at the end of the first BioMed Central conference on Metabolism, diet and disease, the panellists confronted the overwhelming evidence for a link between obesity and cancer. The panel discussion at the second picked up where the first left off – Can cancer be prevented by diet?

The only categorical answer came from Stephen O’Keefe, starting from the epidemiology that shows a 100-fold difference in colon cancer risk between African Americans (high) and rural Africans (low). If you switch their diets – and he has done the experiment – the gut microbiota, he reports, switches within two weeks, with known carcinogens going up in the guts of the rural Africans and conversely down in African Americans.

But diet, he acknowledged, has direct effects on the environment of the cells of the gut – In other cancers, the environmental influences on the development of a tumor are dominated by its tissue of origin and the regulatory physiology of its host.

Can you feed or starve a tumor?

Craig Thompson, a spearhead in the field, set the tone early in the discussion in response to Eugene Fine, who challenged Gregory Petsko (chairing) on the legitimacy of calling cancer ‘a’ thing. It’s not, Thompson emphatically confirmed. Different tumors have different metabolic imperatives that reflect those of their tissues of origin – prostate tumors furiously produce citrate reflecting the need for prostate cells to allow sperm to save their ATP, glial cells ravenously scavenge glutamate from the extracelluar environment to prevent it from disrupting synaptic transmission between the neural cells they support – and so on for all three or four hundred types of tumors.

The metabolic distinctions of tumor tissue origins indeed, as Lewis Cantley chimed in to add, trump the metabolic distinctions conferred by the different dysregulating mutations that characterize them – the metabolic profiles of KRAS mutant lung, colorectal and pancreatic tumors have more in common with those of normal lung, colorectal and pancreatic tissue, respectively, than they do with one anothers’. And they all exist in critically different environments which determine their progress in the organism.

On top of that (Craig Thompson again), any effects of dietary adjustments may be negated by the regulatory physiology of the organism, and the tumor cell thereby protected from manipulations aimed at its particular metabolic needs.

Do we have the right models?

Are cell lines, then, Petsko asked, up to the job of providing models?

Resoundingly, no (Cantley again): they cannot represent the tissue environment or the endocrine or immune influences of the organism – indeed, they arguably can’t even represent the tumor itself: strikingly, there are only six prostate cancer cell lines from thousands of attempts to grow them – this means we know nothing about what makes all the other thousands grow.

The increasing recognition that cell lines cannot tell the story was reflected at the conference in David Sabatini’s presentation on the development of organoid cultures from intestinal crypt cells, where the stem cells that escape from regulatory constraints to grow into tumors can be studied and manipulated in something that more closely resembles the environment in which they naturally grow than does culture medium in a Petri dish.

Detailed metabolomic investigation of human tumors is another answer, represented at the meeting by Ralph deBerardinis and Joshua Rabinowitz, in search of singular points of vulnerability to which tumor cells expose themselves through the metabolic demands of rapid growth. – Though mouse models, also imperfect, have again served to illustrate the point that we cannot expect global solutions: tumors with different mutational backgrounds produce different responses to the same metabolic intervention.

And a vast untapped resource of potential human insight lies in human cancer patients in remission – a point made by Craig Thompson in response to a question from the audience (specifically in connection with breast cancer in obese women) – Once a tumor has formed, is there anything you can do with diet to discourage it?

Not known, Thompson said. But with 16 million cancer survivors as the single identified population at highest risk of cancer, already linked in to the oncology community and with every incentive to participate in trials, this is an issue that ought to be being studied.

Here however the discussion had strayed largely into intervention, or prevention of recurrence, which, though of huge practical importance, were not the issues the panel had set out to address.

Where now?

Mike Bishop disclosed the elephant. You can’t know how to prevent a cancer until you know its cause: It was the incrimination of cigarette smoke that brought down lung cancers, and the incrimination of papillomavirus (that and the vaccine) is now expected to do the same for cervical cancer. At present, Bishop argued, we don’t know if dietary differences are causal, or if they are mitigating the effects of the real causes. His answer – epidemiological data to suggest the causes and inform the questions you ask of model systems.

Craig Thompson, acknowledging the contribution of molecular biology and genomics to the understanding of cancer, argues the importance of marrying that understanding to a much deeper understanding of metabolism – the signalling pathways incriminated by molecular genetics all intersect with metabolic pathways, it’s imperative to understand both. And though the data on the major dietary components – carbohydrates, fats, proteins – may not have given clear answers, there may be important effects of micronutrients that we may not even be aware of.

Lewis Cantley, returning to the issue of cause, pointed to the potential of recent efforts by Michael Stratton and colleagues to identify the mutational landscape of 10,000 human tumors in search of signatures of the causal event – in principle it’s possible to distinguish mutations caused by ultraviolet light, for example, from those caused by chemical carcinogens or reactive oxygen species.

Next question?

And when we know the causes of cancers, what then?

The route to prevention may not be simple. Epidemiologist Elio Riboli vividly made the point. Many breast cancers could be prevented if all women had their first child before the age of 20 and then had another four, and breast fed all of them for three years. This is not going to happen on a large scale in the developed world.

But the control of diet must be susceptible to much more practical and socially acceptable solutions, and with the contribution of obesity (if not its mechanism) undisputed, the success of the anti-smoking interventions shines as a beacon of hope for the triumph of good sense over both folly and vested interest. The conquest of obesity is indeed essential in the face of the already intolerable and still rising toll it exacts on human health; but it will not prevent all cancers – not even all of the breast, colorectal, and endometrial cancers in which it is most strongly implicated.

For the rest, the panel agreed, we must look for a deeper understanding of causes and of metabolic mechanisms – and think very hard about how to do properly controlled trials of dietary modification on our own species.