Monday, April 14, 2008

Artificial sweeteners, part deux

I just wanted to post an update about artificial sweeteners, because one of the papers I mentioned in my earlier post on the subject was at that time still in press and now it has been published.

This article (1) is by a pair of researchers in the Department of Psychology at Purdue University, Susan Swithers and Terry Davidson. In their first experiment, they compared the weight gain of rats fed non-fat yogurt sweetened with glucose (a sweetener that has the same sweetness and number of calories as table sugar) vs. rats fed yogurt sweetened with saccharin. Each group got sweetened yogurt three days a week, unsweetened yogurt three days a week and chow one day a week. The rats stayed on the diets for 5 weeks while the researchers monitored their food intake and body weight. At the end of the study, they also measured the animals' body fat content.

The researchers found that all of the animals gained quite a bit of weight (about 20% over their starting weight) . However, they found that the animals which got saccharin yogurt gained slightly more weight than rats that got glucose yogurt in three out of the five weeks of testing, for an overall statistically significant effect*. They also found that the saccharin group had slightly more body fat. However, the researchers did not find a statistically significant difference in total calorie intake between the groups.

For a second experiment, the researchers sought to do a more detailed analysis of the effects of artificial sweetener on food intake. During a "training" period, rats were given normal chow along with yogurt for 14 days. On half of the days, each animal was given plain yogurt, and on the other half of the days one group of animals got glucose-sweetened yogurt and the other group got saccharin yogurt.

After the 14 day training period, they fed the animals chow and water for 1 day (presumably to "flush out" any short term effects of their yogurt diets), followed by an overnight fast (presumably to make the rats hungry...rats are nocturnal so they normally do most of their eating at night). Then half the rats in each group were given a "premeal" of Chocolate Ensure Plus** for 30 minutes, and then normal lab chow was given back to all the rats and their food intake was measured.

The researchers found that the rats that got the glucose yogurt ate slightly less during the 14-day training period than the rats that got saccharin yogurt. As in the first experiment, both groups gained weight, about 10% of their starting weight, but the saccharin group gained slightly more. During the testing period, they found that the glucose rats who got the premeal ate slightly less than those that didn't, but the saccharin rats ate the same amount, regardless of whether they ate a premeal. However, there does not appear to be an overall difference in chow intake in the testing period between the glucose and saccharin groups.

For their third and final experiment, the researchers sought to establish a physiological basis for the differences between the glucose and saccharin groups. In this experiment they look at the well established phenomenon of "postprandial thermogenesis,": when you eat food, it causes a small, temporary increase in body temperature. Studies have shown that when food is tasted, but not swallowed, it tends to cause a bigger increase in body temperature than if the food were actually eaten (2,3), whereas if food is delivered directly to the gut via a tube and not tasted, the increase in body temperature is smaller (2,4). The authors hypothesized here that since the saccharin rats have been trained to eat something that tastes sweet, but without the calories of sugar, that they might also see decreased body temperature in response to a meal compared with the rats whose bodies "expect" the calories associated with sugar. (n.b. Thanks to anonymous, who pointed out that my original interpretation of their hypothesis made no sense, and I have corrected this post accordingly. Sometimes I read stuff too fast!)

For this experiment, all the rats were surgically implanted with thermometers that transmitted temperature readings to a receiver, so the researchers were able to continuously monitor the temperature of each rat. Then they did the same 14-day training with glucose- or saccharin-sweetened yogurt as in the previous experiment. Then all rats were fasted overnight, and then given a premeal of chocolate Ensure, followed by ad lib access to chow.

They found that during the first hour of the 14-day training period that both the glucose and saccharin groups increased their body temperature after eating the yogurt, but the glucose group's temperature came down more slowly after the meal. Then during testing, they found that the glucose group's temperature went up slightly more in response to the Ensure than the saccharin group's, although both groups increased their body temperature.

Anyway, the authors' interpretation of this all is that there is a Pavlovian conditioning phenomenon occurring here. They believe that when rats eat yogurt sweetened with sugar, their bodies learn that sugary taste is associated with the intake of calories, and they adjust their diet accordingly when presented with things that are sweet. In contrast, rats that eat saccharin yogurt do not develop this association, and so when they are presented with sweet foods their bodies don't expect them to have any calories. They believe that the temperature differences following glucose- and saccharin-sweetened foods are part of the mechanism by which animals make these adaptations, but things get a little hand-wavy at this point for me.

This paper is kind of intriguing, but I think there are a few big leaps between the findings of the paper and the conclusions the authors draw. I think the biggest caveat is that they are assuming that saccharin's effects are due to the fact that it is an artificial sweetener and not due to some other possible effect of this chemical on the rats' bodies. A good way to test for this would be to look and see if other types of artificial sweeteners have the same effect.

Another big caveat is that the results in this study are mostly pretty subtle, and these studies were all done in rats that had been raised on a diet of lab chow...in other words, these experiments were the rats' first experience with sweet-tasting food. This makes me wonder about the applicability of the study in humans, since most people eat a combination of natural and artificial sweeteners, so we are not being "trained" in the same way as the rats.

A third caveat is the fact that their proposed mechanism is pretty vague. They don't identify an neural pathways or hormones or brain areas that might be implicated in this Pavlovian response. They merely show a difference in the effects on body temperature, which again, could be due to some specific effect of saccharin. I think a really interesting test of their hypothesis would be to train rats on glucose yogurt and then give them a pre-meal sweetened with saccharin and see if that caused them to eat less subsequently. Maybe that will be in their next paper.

In conclusion, I think this paper has some significant flaws (among which is the fact that they issued a press release about the paper before it was published so the results were reported in the media without any scrutiny). But I also think there might be something there. I think it's just not quite clear yet exactly what that is.

*These researchers do something I've never seen before with their statistics, which is that for the measures they say are statistically significant, they just say that the p-value is less than 0.05, they don't tell you what the p-value actually is. For the nonsignificant measures, they do tell you the actual p-value. Has anyone else ever heard of this? Is there any reason to do it other than my suspicion that these p-values were all very close to 0.05, and they wanted to downplay this by just called 0.05 the cutoff and then claiming that their data fit that criterion?

**Ensure is often used to fatten laboratory rats, and I used to work next door to a lab that did this a lot. As you can imagine, it definitely colored my view of those TV commercials for Ensure where people talk about how they give their senior citizen parents Ensure because they care about them so much.




1. Swithers SE, Davidson TL. A Role for Sweet Taste: Calorie Predictive Relations in Energy Regulation by Rats. Behavioral Neuroscience 122: 161-173, 2008.

2. Diamond P, Brondel L, LeBlanc J. Palatability and postprandial thermogenesis in dogs. American Journal of Physiology 248: E75-E79, 1985.

3. LeBlanc J, Cabanac M. Cephalic postprandial thermogenesis in human subjects. Physiology and Behavior 46: 479-482, 1989.

4. LeBlanc J, Cabanac M, Samson P. Reduced postprandial heat production with gavage as compared with meal feeding in human subjects. American Journal of Physiology 246: E95-E101, 1984.

6 comments:

Anonymous said...

This part confused me:
"Studies have shown that when food is tasted, but not swallowed, it tends to cause a bigger increase in body temperature than if the food were actually eaten (2,3), whereas if food is delivered directly to the gut via a tube and not tasted, the increase in body temperature is smaller (2,4). The authors hypothesized here that since the saccharin rats are eating something that tastes sweet, but without the calories of sugar, that they might also see decreased body temperature in response to a meal here, too."

Why is ingesting an artificial sweetener like ingesting something without tasting it? I would think it would be like tasting something without ingesting it (since you taste sweet, but there are no calories ingested). And then it would be have a larger thermic effect (which obviously it doesn't). I don't understand how this hypothesis applies.

Dr. LaWade said...

Anonymous, you're right! I totally misread that. I guess what they're actually saying is that since the saccharin rats' bodies haven't been trained to expect calories in response to a sweet taste, they wouldn't have the thermic response that the glucose-trained rats would. And indeed, that is the case. I'll correct that in my post right now. Thanks again!

Anonymous said...

I think the difference in glucose and saccharin is in their action on incretin hormones (GIP, GLP-1) in the gut. If they gave one group yogurt sweetened with glucose and one group plain yogurt and gave them an identical amount of intravenous glucose there would also be a difference in appetite and weight.

Anonymous said...

So as far as the APA (psychology assoc) goes, significance is p <= 0.05. Since it is a memory and learning paper, my guess is this is old psych training coming out (it used to be okay to do this before computers and sometimes is still used).

Also, what about the calories from the yogurt itself? Don't those count?

Weetabix said...

I've seen ROI and business cases where they don't reveal the actual P value less than .05 but I'm really surprised to see that in scientific methodology. You'd think you'd report the findings, just for the sake of repeatability?

j said...

Yeah, in my (APA style) papers I don't have to include anything other than the p significance level (<= .05, or <= .01, whatever). When I report non-significant results, I can do it one of two ways, saying all p values are > whatever level (e.g., .10), or all F values < 1.

If the results are less than .05, then it is significant, and the degree to which it is below .05 isn't really relevant. (This of course begs the question of why you'd ever describe something as marginal.) The p value doesn't tell you anything about the size of the effect, which is a different measure. In fact, some of my journals are moving away from standard F/t statistics and moving to confidence intervals/effect size measures.