Systematic Review/Meta-analysis Shows Coffee May Protect from Oral and Pharyngeal Cancers
Coffee (Coffea spp., Rubiaceae) is, after water, the most-consumed drink worldwide. Coffee compounds, besides caffeine, include phenolic derivatives and diterpenes. Studies suggest that these and other coffee compounds have anticancer effects. Inverse associations of higher levels of coffee have been found for liver, endometrial, and colorectal cancers. The International Agency for Research on Cancer (IARC) classified coffee as possibly carcinogenic in 1991, a conclusion that has since been reconsidered. The IARC has ruled that evidence out but has not addressed potential protective effects. Oral and pharyngeal cancer (OPC), the sixth-most-common cancer type, has high mortality and poor outcomes for survivors. Many studies have investigated OPCs’ multi-etiology, including the use of tobacco (Nicotiana tabacum, Solanaceae) or alcohol, human papillomavirus (HPV), and dietary factors. Studies of coffee’s contribution to OPC risks have reported conflicting results. The authors performed a meta-analysis of qualifying studies.
A search of electronic databases up to August 2016 yielded 22,515 references in English, Portuguese, French, and Spanish. Of these, 22,470 were duplicates or irrelevant to the study. Of 25 articles screened,* eight did not meet inclusion criteria, leaving 17 for analysis. They were prospective cohort (n=4) and case-control (n=13) studies reporting risk ratios (RRs) or odds ratios (ORs) and 95% confidence intervals (CIs) for highest vs. lowest coffee intake.** Data were extracted using a standardized form. Study quality was assessed via the Newcastle-Ottawa scale; all articles selected scored ≥5 (out of a maximum of 9). Heterogeneity was measured using Cochran’s Q test and I2 statistic; sensitivity, via the “leave-one-out” method. RRs/ORs were pooled for summary RRs—first for all OPCs, then for oral cavity cancer (OC) and pharyngeal cancer (PC) separately.
Quantitative analysis showed significant association of coffee consumption and all OPCs (P<0.001), with RRs for individuals with the most coffee intake 1.45 times less than for those with the lowest intake (pooled RR=0.69; 95% CI, 0.57-0.84). Significant (P=0.009) and moderate (I2=50.3%) heterogeneity was found across studies. In case-control studies considered alone, there was a small decrease in heterogeneity (I2=46.5%, P=0.033), but the level of effect remained significant (pooled RR=0.67; 95% CI, 0.53-0.84; P<0.001). However, in cohort studies alone, heterogeneity increased (I2=54.7%) and the level of coffee intake became unrelated to the presence or absence of OPCs (pooled RR=0.86; 95% CI, 0.69-1.08; P=0.202). Eight studies were conducted in Europe; five in North and South America; and four in Asia. There was significant and moderate heterogeneity among studies from North and South America (I2=58.1%, P=0.049); less among those conducted in Europe (I2=46.2%, P=0.072) and Asia (I2=47.3%, P=0.128). Studies with >300 subjects had lower (I2=35.3%) and nonsignificant (P=0.159) heterogeneity compared to those with <300 subjects, which showed moderate (I2=60.8%) and significant (P=0.006) heterogeneity. Sensitivity analysis found no significant change when any study was omitted. Duval and Tweedie’s nonparametric “trim and fill” method revealed no missed studies, and neither funnel plot examination nor Begg and Mazumdar’s rank correlation found any significant publication bias in the studies included in this meta-analysis.
In OC reports considered alone, there was no significant association between coffee intake and risk. Pooled RR for highest vs. lowest intake was 0.82; 95% CI, 0.58-1.16; P=0.257, with nonsignificant (P=0.056) and moderate (I2=53.6%) heterogeneity. However, sensitivity analysis found that omitting a particular study had a significant impact, revealing a significant association between coffee intake and OC risk (pooled RR=0.77; 95% CI, 0.61-0.95; P=0.017), with lower (I2=26.7%) and nonsignificant (P=0.244) heterogeneity. Omitting one other study also yielded a lesser but significant association (not reported).
In PC reports considered alone, there was a significant association between coffee consumption and risk. Pooled RR for highest vs. lowest intake was 0.72; 95% CI, 0.54-0.95; P=0.019, with moderate (I2=37.3%) and nonsignificant (P=0.188) heterogeneity. The leave-one-out analysis did not significantly affect results for PC risk.
The inverse association between coffee and OPCs in this meta-analysis confirms results of earlier systematic reviews and suggests a protective effect, especially against PCs. Limitations of many studies used include lack of data on caffeine and other coffee compounds’ concentration, type of coffee, and size of cups. Different categories of coffee intake did not allow for a dose-response analysis. The coffee temperature was uncontrolled and data on it lacking in almost every study. Beverage temperature may be linked with some cancers and should be controlled in future studies. Both alcohol and yerba maté (Ilex paraguariensis, Aquifoliaceae) consumption are linked with higher risk of OPCs. Whether more coffee use is linked with less alcohol intake should be explored. Users of alcohol or tobacco, harmful to the oral mucosa, may not have been equitably distributed in study samples. Other variables that may affect coffee’s benefits include sex, ethnic or genetic differences in coffee metabolism, health status, and whether coffee used has caffeine or is decaffeinated.
* What happened to the other 20 references retrieved is not disclosed.
** For three studies in which 95% CIs were not included, it was estimated based on data provided and sample sizes.
Miranda J, Monteiro L, Albuquerque R, et al. Coffee is protective against oral and pharyngeal cancer: a systematic review and meta-analysis. Med Oral Patol Oral Cir Bucal. September 2017;22(5):e554-e561. doi: 10.4317/medoral.21829.