Does coffee increase cancer risk?

With millions of cups of coffee being quaffed down daily, we’d like to think that one of the world’s most popular beverages isn’t slowly killing us. And as a coffee drinkers at the Naked Scientists, we're more invested than most. So that’s why a paper in the journal Clinical Nutrition caught our eye this week: the authors had used the clever trick of linking genes that make people hooked on caffeine with cancer outcomes. Moreover, reading the author list, Chris Smith discovered that one of the authors works down the road from him, so he invited him along for a chat over… you’ve guessed it… a coffee…

Barista - What can I get for you?

Chris - What do you want, Steve?

Steve - Hey, can I get a medium latte, please?

Chris - Yeah, I'll have the same.

Barista - Anything else for you?

Chris - That's all. Thanks.

Barista - So that's 3 45, please. Thank you.

Chris - Should we go outside?

Steve - My name's Steven Burgess. I'm a program leader at the MRC, the Medical Research Council biostatistics unit at the University of Cambridge.

Chris - Now the irony won't have escaped you, Steven, that we've brought you to a coffee shop, and we're gonna talk to you about a paper that you've just written, linking the stuff in that cup with cancer.

Steve - Firstly, thank you very much for buying me the coffee. It's very much appreciated. So obviously a lot of people in the world drink coffee. It's quite a common drink. So we really want to know from a health point of view is coffee good for you? Particularly with relationship to cancer? There's been a lot of studies out there and we really wanted to bring some clarity to that situation.

Chris - I read somewhere that more than a billion kilos of coffee gets drunk every year. At that level of consumption, surely if it was killing people we'd know wouldn't we?

Steve - Yeah. So certainly if coffee was similar to cigarettes, then yeah we would learn about that pretty quickly. But it could well be that there is a small harmful effect of coffee, which would be really difficult to pick up. I mean, really when you want to do these studies in a scientific way, you'd really like to do a randomized trial, but to do a randomized trial of coffee consumption, particularly for an outcome, which takes a long time to develop like with cancer its gonna be pretty much impossible. We can't randomize ourselves but the idea is that we try to use naturally occurring factors, which act a little bit like randomization. So in our case, we were looking at genetic variants and there are some genes which influenced the amount of coffee that you drink.

Chris - So how did you use that information to marry up the likelihood of having cancer from drinking coffee then?

Steve - People went into a questionnaire. They were asked how many cups of coffee do you drink? So we link the coffee consumption to certain genetic variants and then we see 'do those same genetic variants associate with risk of diseases'. In this case, we were looking specifically at cancers.

Chris - How many people did you consider and where did you find these people?

Steve - So this is data from a study called UK Biobank with about half a million participants in.

Chris - And what did you find?

Steve - When we looked in the overall analysis, there was no association between having a genetic predisposition to drinking more coffee and having cancer, which obviously is aa good finding for me as a coffee drinker. But when we looked specifically at cancer subtypes, we found a particularly strong association for something called oesophageal cancer.

Chris - How could it be present when you look at the oesophagus but when you look at all cancers, you don't see an effect? Is it just that it drowns out the effect of the oesophagus and it's only when you look specifically at that one that you see this association?

Steve - So our theory and it's a theory, I mean we can talk about the evidence supporting it, but our theory is that the link with oesophageal cancer isn't due to the coffee per se, isn't due to anything in the coffee or the caffeine, but due to thermal injury, just from drinking hot liquids. I mean we know that hot temperature can, can cause cancer. We know about skin cancer and if you're putting 70, 80 degree liquid down your throat, then really is it that big surprise that that's leading to greater instance of cancer in the throat?

Chris - How did you dissect out the temperature effect then? Did you actually have data from the participants on whether they like it or like it cold?

Steve - We did actually have some data yes. So UK Biobank asked a whole load of questions and one one of those questions was 'what temperature do you prefer to drink your hot beverages at'. Coffee consumption was associated with a greater change in cancer risk for those people who had a preference for hot or very hot.

Chris - How much then has my risk of esophageal cancer changed by quaffing down a few coffees every day compared to if I didn't drink coffee at all?

Steve - Yeah, that really is the $64,000 question. So when we compared people with a genetic predisposition for drinking 50% more coffee, we saw, in the warm group, we saw about a two and a half fold increase and in the hot, very hot group, we saw a four, five fold increase.

Chris - So presumably then if you switch the study round and you said rather than coffee, what about tea? Since both are beverages usually drunk hot, you ought to see a similar trend then, or even hot chocolate.

Steve - Yeah, so we didn't have quite as granular data from the study on tea drinking as we did for coffee drinking. But certainly the genetic score, we know from other data, that the genetic predisposition to drink more coffee also gives you a predisposition to drinking more tea and when we restricted the analysis to those people who reported that they didn't drink any coffee, we still saw an association with cancer. So that's, again, that's one of the lines of evidence which led us to believe that this thermal injury hypothesis, that it's really the hot liquid and not something specific to coffee.