March 24, 2022

expert reaction to observational study of sweeteners and cancer risk

An observational study published in PLOS Medicine looks at artificial sweeteners and cancer risk.

Dr Michael Jones, Senior Staff Scientist in Genetics and Epidemiology at The Institute of Cancer Research, London, said:

“This large observational study of over 100,000 adults in France links the use of artificial sweeteners to cancer risk.  However, the current consensus, according to the US National Cancer Institute, is that there is no clear evidence that artificial sweeteners cause cancer in humans – although there is frequent re-evaluation of the available evidence by food safety health authorities.

“The link between artificial sweeteners and cancer reported in this study does not imply causation – it is not proof that artificial sweeteners cause cancer. The types of people who use artificial sweeteners may be different in many ways to those who do not, and these differences may partly or fully explain the association. The authors used statistical methods to try to make allowance for these differences but they admit there could be other reasons for the association between artificial sweeteners and cancer risk.

“Some limitations of the study are that the authors had to exclude 15% of participants from the analysis because of under-reporting, and artificial sweetener use was not measured directly, but came from 24-hour dietary records linking food brands with information from food content databases.

“In addition, the ‘dose-response’ relationship was not strong. For some results, risk of cancer was higher in the lower consumer group than the higher consumer group, despite higher consumers reporting 10 times the total artificial sweetener consumption than lower consumers. This also suggests that cancer risk may be raised in the type of person who uses artificial sweetener rather than the sweetener itself.”

Dr Duane Mellor, Registered Dietitian and Senior Teaching Fellow, Aston Medical School, Aston University, said:

“This is a large study which attempted to assess artificial sweetener intake over a period of about 8 years. It has carefully tried to look at how much of individual sweeteners people were consuming, however this French cohort was more educated and more likely to be female than the general population. Also it is difficult to measure intake of individual sweeteners in the food and drink we consume as often they are blended in combinations, with aspartame and acesulfame-K being the most commonly used blend and perhaps reflecting why the overall effect of sweeteners and risk of breast cancer and particularly weight related cancers was also seen with these two sweeteners.

“The authors did try to rule out other factors which could increase risk of these cancers, such as total calorie intake, weight and overall sugar intake. After accounting for these there was still an increased relative risk of developing cancer of 13%, but when looking at the number of cases that is approximately 3 more cases of all types of cancer per 10,000 people over an average of about 8 years. However, when looking at the differences between the groups the authors did not account in their analysis that higher consumers of artificial sweeteners also consumed more highly processed foods, drank more sugary drinks and reported more attempts to try to lose weight. This could suggest the risk of consuming artificial sweeteners could be partly linked to a poorer overall quality of their diets or that their diet may have been altered more to try and lose weight, something called reverse causality. 

“This study does not prove or even suggest that we should go back to sugar and turn our backs on artificial sweeteners or diet drinks.  It does however, suggest that artificial sweeteners are not a perfect replacement for sugar, they come with their own potential risks, as does sugar. The ideal answer is probably to move away from both, however that may be unappealing to many, who like a little sweetness in their life so ditching the regular or diet soft drink (soda) for water may not be a well received health message. So, maybe step-wise approach could be better, maybe looking to steadily reduce sugar and if need be use small amounts of sweeteners only as a short term measure or stepping stone to ultimately reducing intakes of both sugar and sweeteners. 

“This study, perhaps also reminds us, that when we try to improve our diet, and maybe lose weight, the easier option to switch to reduced sugar and artificially sweetened foods (which can be low fat too), may not be the best way. Instead it is probably much better to take a bit of time to plan a sustainable meal plan based on minimally processed foods and rich in vegetables, fruit, beans, peas lentils, wholegrain and nuts or seeds (or a Mediterranean type diet) is a far better way to maintain and improve health whilst reducing our risk of cancers and particularly weight related cancers.”

Prof Tom Sanders, Professor Emeritus of Nutrition and Dietetics, King’s College London, said:

“This study of French women purports to show a weak association between the consumption of artificial sweeteners and risk of breast cancer. Previous studies have shown alcohol intake in particular, and to a lesser extent obesity (in post menopausal women), are more strongly associated with risk breast cancer. US studies where artificial sweetener use is greater than in France have failed to demonstrate any relationship between sweetener use as breast cancer. Furthermore, there is no good evidence from animal studies to show that approved sweeteners increase risk of mammary tumours. It is well known that women who are obese or who have a tendency to gain weight are more likely to use artificial sweeteners and this limits the validity of the conclusions of this study because it is not possible to completely control for this in the statistical analysis.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This is, in my view definitely one of the better, more careful studies of its general kind. But it’s important to understand that it does not show that consuming more artificial sweeteners causes people’s risk of cancer to increase, and indeed the researchers are careful not to claim that it has shown there’s a cause-and-effect relationship. That’s for several reasons to do with the way the study was carried out, which are generally unavoidable. I’ll explain in more detail below – but I should also point out that, just because the study cannot prove that consuming artificial sweeteners causes an increased cancer risk, that doesn’t mean that it has shown that there is definitely no cause-and-effect relationship. There might be, and the findings are consistent with that – the issue is that they are consistent with other possibilities too. At the end of their report, the researchers do not claim any more than that their study suggests associations between consumption of some sweeteners and cancer risk. They rightly say that more research in other study cohorts, and experiments to cast light on what is going on within the body, are needed.

“The overall, inevitable, problem with the study is that it’s observational. The researchers didn’t make the participants eat particular things – they recorded what they ate over a period of time, and followed them up and recorded which people had a new cancer diagnosis. They then split up the participants into three groups – those who consumed no artificial sweeteners at all, those who did consume some but consumed less than average, and those who consumed more than the average. (They used the median consumption as the average for this purpose, separately for men and women.) For some cancers and some sweeteners, they found that the risk of cancer was higher in those who consumed sweeteners, particularly, in some cases, those who consumed greater amounts of sweeteners.

“The snag is that the three groups of people, that they compared, differed in many other ways than just their consumption of artificial sweeteners. For instance, those who consumed sweeteners were more likely to be women, to smoke, to be younger, and to have diabetes than those who didn’t consume sweeteners, and other aspects of their diets were different too. Any of those other factors could have been the cause of the differences in cancer risk, in whole or in part, rather than the sweetener consumption. It’s possible to make statistical adjustments to try to allow for differences in other factors, and these researchers did make many such adjustments. Many of the associations between artificial sweetener consumption and cancer risk were still there after these adjustments. But one can never be confident that every relevant factor has been adjusted for, and the adjustment involves a statistical model that may not correctly describe what is going on. So a single study like this just can’t establish that there’s a cause-and-effect relationship between sweetened consumption and cancer risk. This is what’s meant by the possibility of residual confounding, mentioned in the press release.

“How big is the difference in risk anyway, between people who consumed no artificial sweeteners and those who were classed as higher consumers of sweeteners? In the people studied, about two-thirds said they did not consume any artificial sweeteners at all. Out of every 1,000 of those who said they consumed no sweeteners, 31 had a new cancer diagnosis during the follow-up period of about 8 years (on average). That’s quite a low rate, but it reflects the age of the participants (average just over 40) and, probably, the fact that they are more health-conscious than average. What I can’t say is how many of that 1,000 would have got a cancer if, instead, they had consumed the higher amounts of sweeteners instead. That’s because the study hasn’t told us whether the association is one of cause and effect. But I can give a figure for how many out of 1,000 would get a cancer diagnosis during follow-up, if those 1,000 were similar to the 1,000 who consumed no sweeteners in terms of all the factors that the researchers adjusted for, but they all consumed higher amounts of sweeteners rather than none at all. The answer is that about 35 of them would have a cancer diagnosis, so  about four more than the 1,000 who consumed no sweeteners. There’s a statistical margin of error around that figure – it could be just one more cancer case, or eight more. And to repeat myself, that difference could anyway be caused by something other than the difference in sweetener consumption. The difference isn’t negligible, but perhaps it’s not very large.

“The press release also mentions the possibility of reverse causality. This means that cause and effect might, to some extent anyway, run in the opposite direction to what you’d expect. The possible scenario is that people may start to be ill, in some way, for a time before their cancer is diagnosed, and because of that change in their health, they may change their diet to consume more artificial sweeteners. But because their cancer diagnosis came after the diet change, it looks as if the sweetener consumption might have caused the cancer, while in fact it was the other way round.  It’s unclear how likely that might be, and the researchers did carry out some other analyses that make it look unlikely, but the possibility can’t be ruled out.

“That does indirectly draw attention to another feature of the research. The participants recorded in detail what they ate during 24-hour periods, on repeated occasions. The fact that these records were detailed and that the measurements were repeated about every 6 months is an advantage over some studies of food intakes, which look at consumption at just one time point. But the statistical analysis in this study looked, I believe, only at the average consumption over the whole period that people were in the study. So if their consumption of sweeteners changed a lot over time, the researchers couldn’t relate that directly to when any cancer diagnoses occurred. It’s difficult to see what else they could have done, given that cancers take a varying and often long time to develop.

“But, for instance, about a third of the participants who had higher consumption of sweeteners went on a weight-loss diet within the first two years of follow-up, while only about one in every eight of the participants who didn’t consume any sweeteners went on a weight-loss diet in that period. Going on a diet, that might involve avoiding sugar, would tend to increase average sweetener consumption during the study, but maybe the increased cancer risk has something to do with going on diets rather than directly with sweeteners. I’m definitely not saying that that is what happened. But I think it’s a possibility, and again it’s an issue in understanding what causes what, and it’s another reason that I agree with the researchers that more research is needed.

“The difficulty of research like this is illustrated also by the fact that, despite the careful and repeated recording of consumption, they had to exclude the data from almost one in every six of the original list of participants, on the grounds that the calorie content of their recorded food consumption was too low to be feasible. It makes sense to exclude data from those people, but it does make me wonder exactly how accurate the remaining self-reported measurements of consumption were.

“Finally on issues with the study methodology, there’s a problem that the participants came from a cohort of people, called NutriNet-Santé, that volunteered to take part in a long term study of nutrition and health. Again, using a group something like this is pretty well inevitable – you couldn’t just get anyone in the population to take part in a study that is time-consuming and needs people to continue involvement for many years. But it does mean that those involved are not typical of the French population in many ways – more women, better educated on average, and with some untypical health-related behaviours. So findings from this cohort may or may not also be the case in the general population of France (or indeed of the UK). And how the health-consciousness of this cohort might relate directly to what they eat, their views on sweeteners, and how they might change their behaviour or sweetener consumption if they get ill, is really not known.”

‘Artificial sweeteners and cancer risk: Results from the NutriNet-Sante´ population-based cohort study’ by Charlotte Debras et al. was published in PLOS Medicine at 6pm UK TIME on Thursday 24 March 2022.

DOI: https://doi.org/10.1371/journal.pmed.1003950

Declared interests

Prof Kevin McConway: “I am a Trustee of the SMC and a member of its Advisory Committee.  My quote above is in my capacity as an independent professional statistician.”

Dr Mellor: “has previously worked with the International Sweetener Association.”

No others received.