Astroblog

Over at Respectful Insolence Orac has discussed the misrepresentation of a study about the effect of fructose on pancreatic cancer. This post is a slightly expanded version of that comment, with some added diagrams and a bit more on fructose concentrations. The point is still that the press release (that limiting fructose intake may disrupt cancer growth), and some of the conclusions of the paper, are not justifyable from the data.

First, a little background. Sugar to most people is sucrose, or cane sugar. But actually "sugar" covers a large number of similar chemical compounds. Ones commonly encountered apart from sucrose include glucose and fructose (also known as fruit sugar as it is found in high concentrations in fruit, it is also found in honey). Indeed sucrose is glucose and fructose linked by a chemical bond. When you eat sucrose, enzymes in your gastrointestinal tract break it down into glucose and fructose, which are then absorbed (your body can absorb intact sucrose).

There is some concern that consumption of high levels of fructose may have negative health consequences above that of simply consuming more calories. People in the United States are particularly concerned as many of their soft drinks and foods are sweetened with High Fructose Corn Syrup (HFCS), as opposed to sucrose in most of the rest of the world (like Australia, where sugar cane fields are part of the national psyche). It should be pointed out that the proportion of fructose in honey is about the same as that in HFCS and some fruits (eg apples and pears) have significantly more. However, the large amount of extra sugar added to processed foods and soft drinks means we are be exposed to much higher concentrations of fructose (and glucose) than we would eating a diet of fresh fruit and vegetables.

In the study Orac talks about (Cancer Res; 70(15) August 1, 2010) the researchers were looking at whether fructose could support cancer cell growth.

Now in the paper there are two sets of results, one set on cancer cell proliferation, and one set on metabolism (seeing whether metabolic pathways involved in cancer growth were selectively activated by fructose). The authors show reasonably convincingly that fructose is preferentially metabolised to nucleic acids in pancreatic cancer cells, pancreatic cell models and hepatic cancer cells (with some caveats I will cover later[1]). However, what they say about the rate of cancer proliferation is almost, but not completely, dead wrong.

Press release:

"These findings show that cancer cells can readily metabolize fructose to increase proliferation," Dr. Anthony Heaney of UCLA's Jonsson Cancer Center and colleagues wrote.

Abstract:

These findings show that cancer cells
can readily metabolize fructose to increase proliferation.

If you have access to the paper, look at Figure 1. If not, I've abstracted some of the key results in Figure 1 below, and added a line showing the control level of proliferation for clarity.


Figure 1, panels A, B, C, D and F from
Cancer Res; 70(15) August 1, 2010








The first important thing to note is that when they did the proliferation assays, they compared the results to cells cultured in 10% Foetal Bovine Serum alone - which contained *0.4 mmol/L glucose*[2] (FBS in the figure, open bar, the black bar is complete medium with 10% FCS and 18mmol/L gulcose) . This means at the lowest concentrations of additional sugar, they are comparing 400µM glucose + 5.5µM fructose vs 405.5µM glucose. Which kind of messes up their interpretation.

Looking at the graph (note the red line showing basal growth, and the error bars) Generally, glucose and fructose are the same...they don't do anything above and beyond the Foetal Bovine Serum control. Only in MiaPaCa-2's, HPAF and, possibly, HPDR6 do you see proliferation which MAY be statistically significant from the baseline proliferation at the highest concentrations used (400µM glucose + 5500µM fructose vs 5900µM glucose, they look higher, but the error bars are significantly bigger too). An important caveat is that for the MiaPaCa-2 cells, when proliferation was remeasured using a different assay (BrDU incorporation), fructose basically did nothing. So any apparent increase in proliferation may just be assay variation.

It's a bit hard to clearly work out what data points are statistically significant, if any, as their statistics are borked, badly borked. The legend to Figure 1 says "P = not significant, fructose versus glucose or normal medium (NM)." Which if correct, means that the proliferation rates at 5.5µM added sugar (glucose of fructose) is no different from the 18 mM glucose normal medium values, which makes their data uninterpretable (see Figure 1B and C particularly). No other stats tests on the proliferation data are presented anywhere.

So we should reformulate the statement of the abstract as "In some cancer cell line, 5 mM fructose and 5mM glucose might cause the same amount of proliferation above that found with 0.4 mmol/L glucose, if we could statistically distinguish them from the control" (actually, the best interpretation is that fructose and glucose did nothing to proliferation, if their presentation of the statistics is correct).

Now, the kicker, what is the plasma concentration of fructose in most people? It's 8 µM, even in diabetics it's only 12 µM, while plasma glucose in normal people is around 5000µM (doi: 10.2337/diacare.25.2.353 Diabetes Care February 2002 vol. 25 no. 2 353-357, other papers have found diabetics in ketoacidosis to have levels of around 88µM, which dropped to 11µM when their diabetes was controlled [Clin Biochem. 2010 Jan;43(1-2):198-207. Epub 2009 Sep 8.]).

This is probably due to fructose being mainly taken up and metabolised in the liver, but the bottom line is that pancreatic tissue will never see any more than micromolar concentrations of fructose. Even if plasma levels of fructose from dietary sources reached 55 µM (which would be insanely unlikely), it would, on the basis of the results presented in the article (Cancer Res; 70(15) August 1, 2010) do absolutely nothing to cell proliferation (see Fig 1 again). The authors report that pancreatic cancer patients have 2.5 times the plasma levels of fructose of non-cancer patients, but that is still is 100 times less than the 5mM needed to see any difference in proliferation rate (assuming that the results are statistically significant, which they may not be).

There is one tissue that could be exposed to high levels of fructose, and that is the liver, which receives blood flow direct from the gut. However, even high levels of fructose ingestion (4g/kg; about 10 times what you would get from a can of soft drink) produce concentrations of fructose of no more than 1-2 mM in the portal blood (J Am Coll Nutr. 1986;5(5):443-50., Nutr Metab 1971;13:331-338 (DOI: 10.1159/000175352)), still well below the levels where we see some increase in proliferation of pancreatic cancer cells (well, assuming the results were statistically significant, which they may not be; the hepatic cancer cells were not examined for proliferation effects). Even then, the cells exposure at these high levels is brief, at most 2 hours, whereas in the experiments the cells were exposed for 48 hours. So even if you do drink 10 cans of soft drink at one sitting, your liver cells will be unlikely to be exposed to the proliferative levels of fructose.

Put simply, their results do not support their conclusions on prliferation.

1. While the data on metabolism is clear, again these results are in cells exposed to unphysiological concentrations of fructose for 72 hours. This is completely different from what the cells in your body would be exposed to.

2. It is very hard to get rid of glucose from FCS, and if you did the cells wouldn't grow (and actually die at 0 glucose). And you can't get rid of the FCS, for it contains growth factors the cells need to grow as well. Well, there are serum-free media with added growth factors, but they have their own problems. So there are limitations in how they can study this question. Most of my work is done in cancer cell lines (PC-12 and SHY-5Y amongst others, so I am acutely aware of both the strengths and limitations of tissue culture.

Labels: science, science matters