Evidence on coffee consumption and pancreatic cancer: not great, not terrible

Abstract

In his correspondence, Dr. Kawada expresses doubts regarding the risk reduction of pancreatic cancer with increasing filtered coffee consumption found in our paper [1, 2]. He also raises a concern about differences in the pooled-results from previously published meta-analyses that aimed to summarize the evidence regarding coffee consumption and the pancreatic cancer risk. Dr. Kawada correctly points out that the heterogeneity in some of the meta-analyses was moderate to high [3–5]. As a reminder, heterogeneity is a measure of any kind of variability between the studies that are included in a metaanalysis [6]. In the case of high heterogeneity, the Cochrane Collaboration indeed suggests that the results from the meta-analysis should be interpreted with great caution, and even suggests that pooling of the result should not be attempted in the first place [6]. The sources of variability might be due to diversity of the study participants, or methodological diversity between the studies. In the meta-analyses on coffee intake and cancer risk, the additional source of heterogeneity could be a difference in coffee brewing methods across studied populations. In Norway, as well as in other Scandinavian countries, filtered coffee is now considered by far the most popular brewing method surpassing both boiled and instant coffee as the alternatives [7]. Other countries might have a more even distribution of different brewing methods. Dr. Kadawa further points to the evidence that chronic pancreatitis is a precursor of pancreatic cancer and suggests that chronic pancreatitis is an epidemiological confounder. Unfortunately, we do not have information on chronic pancreatitis in the Norwegian Women and Cancer (NOWAC) study. Even though according to the World Cancer Research Fund there are only limited evidence regarding alcohol consumption and pancreatic cancer risk [8], it is possible that adjusting for alcohol intake might, given that alcohol is correlated to coffee consumption (which is not the case in the NOWAC cohort), could to a certain extent diminish the effect of residual confounding by chronic pancreatitis. To this day, there is strong evidence that coffee is associated with the reduced risk of liver and endometrial cancer [8]. It goes without saying that more studies are needed to determine if and how coffee consumption is associated with pancreatic cancer risk. As Dr. Kawada notes, the studies should aim to minimize bias as much as possible, something that is far beyond an easy task when it comes to nutritional epidemiology [9]. This holds even more true when it comes to coffee consumption and the risk of pancreatic cancer, as some studies that aimed to quantify this association are now used in classrooms as a “tool” to teach students about selection bias [10]. Finally, it should be pointed out that the aim of our study was to assess the effect of heavy coffee consumption on cancer risk. In that sense, we aimed to evaluate the safety of excessive coffee drinking. Our conclusion, which is in line with all but one meta-analysis [11], that even in high quantities, coffee seems to be safe when it comes to pancreatic cancer risk. On a side note, even if coffee was indeed reducing the risk of (pancreatic) cancer, the question remains on how to communicate this finding to the public. Many laymen might adopt heavy coffee drinking as a preventive measure against pancreatic cancer, potentially ignoring other factors lurking around that are attributable to a substantial number of (pancreatic) cancer cases.