Biology in the News Explained

Does coffee prevent prostate cancer?

We are just beginning to scratch the surface of how genes and environment interact to affect susceptibility to various cancers, because it is a complex relationship. We are learning the most about sex-hormone driven cancers, because there is a more obvious pathway that has been identified. It is fairly well known that these cancers include uterine, ovarian, breast, and prostate cancer, but less known that they also can include lung and colon cancer.

Prostate cancer is the topic of a well publicized recent paper about its association with coffee-drinking (Wilson et al., 2011). The media delighted in passing along results that suggest a correlation between high coffee consumption with lower lethal prostate cancer risk. Although it is surely possible that a lot of studies taken together may show reasonable support for this correlation, by itself this is one of those “ehh, maybe” papers. Yes, they show a significant trend from less than one cup of coffee per day to greater than 6 for a 0 to 8 year time lag in advanced cancer, but data can be massaged a lot to produce significance.

For the most basic example, their two significant values had a p-value of 0.008. This seems pretty good until you see they did analyses of twelve different trends – four categories of time lag by three cancer tests – all prostate cancer, advanced, and nonadvanced, and a Bonferroni correction would result in a much more conservative significance level of 0.004, which would render their two strongest results nonsignificant (and ignores other comparisons they made of relative cancer risk). It’s important to realize that how the data are broken up is a decision that should be made by researchers a priori in order to produce more confidence in results. Alas, in the competitive world of publication, grants, and promotion, it is less likely this has happened. The categories for coffee consumption seem valid enough, but the truth is those categories are not set in stone, but can be adjusted (for example in a related paper on coffee consumption and breast cancer, Li et al [2011] use slightly different categories of consumption levels). There is certainly no way to know if categories were set after the fact to manipulate significance, but the possibility exists, and unfortunately anyone actively involved in scientific research can attest to the prevalence of such a strategy.

It is important to note too that the authors themselves admit other caveats in their discussion, so Bioblog does not have to:

This study also has some limitations. First, we relied on self-reported diet, which will inevitably be imperfect. Although coffee is well reported, we assessed usual intake only every 4 years, thus missing shorter-term fluctuations in intakes. In addition, we do not have coffee intake information from earlier periods of life, limiting our ability to determine the most relevant time periods of exposure. Finally, although reverse causation does not appear to explain our findings, we cannot rule it out as a possible source of bias.

And at the risk of sounding like a broken record, sample size is always an issue. Most researchers are aware that too small a sample size means real biological effects won’t show up. But a lot fewer realize that too large a sample size means that statistical effects show up that may not be biologically meaningful. The sample size of 48,000 men in this study thus actually gives less support to its biological significance than a similar statistical result with a much smaller sample size would have.

But none of this is to say that the results are necessarily wrong, because there have been dozens of papers now showing correlations between coffee-drinking and a variety of health effects, and the study authors did have a priori physiological reasons to expect that coffee would indeed have effects. Coffee consumption has been correlated with lower risk for diabetes as well. Coffee seems to have the opposite effect on sex hormones as sugar consumption (and in fact can improve glucose metabolism). Both effects are mediated via the protein called sex-hormone binding globulin (SHBG), which goes up in association with coffee drinking, and down in association with high sugar consumption. Originally it was thought that it was increased insulin which lowered SHBG levels in the body, but now it is understood that this effect is mediated by higher liver fat produced by excessive sugar consumption. High insulin, however, can promote growth in cancers which have insulin receptors – such as at least some prostate cancers.

Below is a quick schematic which greatly oversimplifies what seems to be known now about probable chains of physiological effects:

a putative relationship between coffee drinking and cancer risk
A putative physiological relationship between coffee intake and cancer risk.
Possible relationships between sugar intake, diabetes, and cancer risk.

Possible relationships between sugar intake, diabetes, and cancer risk.

This is a possible mechanism for the fact that overweight people have a higher cancer risk, at the same time as a higher diabetes risk.

SHBG binds with sex hormones (estrogen and testosterone), reducing the free amount of the hormones in the blood stream. This can change hormone balances in the body, which affects susceptibility to hormone-mediated cancers. Some research has suggested that variation in the gene for SHBG is associated with variation in risk for prostate cancer. Thus, an interesting (and certainly more informative) study in the future would sort out SHBG genotypes and then look at the interaction of behaviors such as coffee-drinking and genotype on cancer risk.

Another paper hot off the press (Li et al., 2011) supports Wilson et al.’s finding by showing that high coffee consumption lowers post-menopausal breast cancer risk. Interestingly, though, the greatest benefit was for hormone-negative cancer. This study though is weakened by the fact that it was also questionnaire-based and did not employ a Bonferroni correction for its many statistical comparisons; however when hormone receptor status was correlated with coffee drinking, the effect on ER-negative breast cancers was strong. ¬†Interestingly, there was not a strong effect on hormone-positive breast cancer, so the picture above is clearly woefully incomplete if these results are accurate.

Because previous findings suggest that high caffeine intake may be correlated with higher cancer risk, the presumption now is that the effect is correlated with other compounds in coffee. The chemical makeup of coffee is complex, however, and so at this point authors only speculate as to protective effects of any specific compounds. Hopefully future research will keep in mind the emerging consensus that few compounds found in plants, whether they be vitamins, anti-oxidants, etc., work the same way in our bodies in isolation. Humans evolved in the environment of natural, unmodified products, which have complex mixes of compounds; we did not evolve in the world of single-compound supplements. Thus, anyone should be skeptical of the value of any supplement taken in isolation (except in the special case where it is prescribed to make up for a clear physiological deficiency).

And, finally, it is always important to understand that correlation does not equal causation. Although authors of studies such as these always claim that they have factored out all the other variables and still found a significant effect, there is always the possibility of a covariate that is the real cause of the effect. For example, susceptibility to cancer has definite genetic variation. What if certain genotypes affecting cancer risk are linked with taste preferences, which are known to be highly genetically variable? This is not so farfetched as may appear; seemingly unrelated genes can be associated, for instance genes controlling pigmentation and behavior.

The flip side of these interactions involving hormones is as is oftentimes the case with cancer treatment; it may help suppress the cancer, but it can be bad for your health otherwise. For example, hormonal treatment of prostate cancer using gonadotropin-releasing hormone (GnRH) agonists (drugs which ultimately decrease testosterone production) unfortunately appears to increase risk for diabetes and cardiovascular disease by decreasing insulin sensitivity (Smith et al., 2006). Similarly, aromatase inhibitors, which are used to block estrogen production in women with hormone-positive breast cancer, also increase risk for cardiovascular disease, among other problems. So it is very important for those with early-stage cancer receiving treatment to have the conversation with their doctors about the risks (and not just benefits) of treatment, which need to be balanced with risk of the cancer becoming more advanced. For prostate cancer, this risk is often low.



Li J, Seibold P, Chang-Claude J, Flesch-Janys D, Liu J, Czene K, Humphreys K, Hall P. 2011. Coffee consumption modifies risk of estrogen-receptor negative breast cancer. Breast Cancer Res. May 14;13(3):R49. [Epub ahead of print]

Smith MR, Lee H, Nathan DM. 2006. Insulin sensitivity during combined androgen blockade for prostate cancer. J Clin Endocrinol Metab. 91(4):1305-8.

Wilson KM, Kasperzyk JL, Rider JR, Kenfield S, van Dam RM, Stampfer MJ, Giovannucci E, Mucci LA. 2011. Coffee Consumption and Prostate Cancer Risk and Progression in the Health Professionals Follow-up Study. J Natl Cancer Inst. 103:1-9.


Leave a Reply

Your email address will not be published. Required fields are marked *

5 + nine =

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>