Cancer epidemiology today: Not strengthening the value proposition
Click Here to Manage Email Alerts
Editor’s note: To read a response to this editorial from Amanda I. Phipps, MPH, PhD, see her commentary. To read HemOnc Today’s coverage of the research by Phipps and colleagues, click here.
It is not always easy, for this series of editorials, to come up with a topic that will be relevant to the readership, and that might make a difference in the world of cancer research and treatment.
One of the interesting — and sometimes arduous — roles that my co-Chief Medical Editor, John Sweetenham, MD, and I have taken on is to review vast numbers of articles prior to publication in HemOnc Today.
This is interesting because we have the chance to learn about real progress at an early stage, and arduous because we sometimes have to try to digest some strange studies that have been presented or published, including those with questionable statistics, and occasionally research that simply does not make sense.
Epidemiology studies
One pet peeve of mine is the amount of potentially pointless research that seems to be emerging in the world of oncology, and more recently in cancer epidemiology.
I have known for years — from my periods of service on study sections of the NIH and other granting organizations — that cancer epidemiology seems generally to be well funded, and grants can sometimes be huge. However, I have not always been sure that adequate return on investment has truly existed in recent times.
In the halcyon days of the past — when luminaries like Sir Richard Doll, Malcolm Pike, PhD, and the late Ron Ross,MD, wrote amazing grants that sought to explain oncological conundrums, to clarify the genesis of cancer and to influence our public health approaches to cancer avoidance — the work was meritorious and truly important, and it generated hallmark publications that changed lifestyles and patterns of care.
That is how we learned about the relationship between asbestos and cancer, smoking and cancer, viruses and genital malignancy, and so many other important topics.
That said, it is important to remember that influencing the public in its health behavior patterns — even when the data are extreme and incontrovertible (eg, smoking or asbestos and cancer) — is not that easy, and one always needs to be cognizant of the likely true impact of any epidemiological study.
Alcohol and cancer
These days, I have an emerging concern that runs along the lines of: “What were they thinking?”
This refers both to the numbers of investigators who seem to propose truly pointless — or duplicative — studies, and funding agencies that appear happy to support them.
For example, I recall reading studies in the 1970s that suggested, with many caveats, that alcohol (wine, beer, spirits and even sake) may be associated with the genesis of cancers, and perhaps with altered treatment outcomes. In the intervening 30 years, there have been numerous studies, reviews and meta-analyses that have concluded that there may be some type — either positive or negative — of modest association between alcohol and cancer genesis or treatment outcomes.
In the Jan. 10 edition of HemOnc Today, we reported outcomes of a study by Rivera and colleagues that showed alcohol — and particularly white wine — may increase melanoma risk. This was yet another possible statistical association, drawn from yet another large set of accumulated demographic data.
I asked myself why the investigators sought to carry out yet another alcohol association study, why a funding agency supported it, and why an editor — and reviewers — chose to publish it in a good journal. What is one to do with this information? Based on a modest statistical association, what should one advise the public to actually do?
For several decades, most of the papers on alcohol and cancer — with the exception of hepatoma — have consistently shown only mild to moderate possible statistical associations, and the outcome data have been quite inconsistent. Who knows whether an evening glass of wine will make things better or worse for the patient on treatment or after treatment is completed?
In this edition of HemOnc Today, we have covered a study from Phipps and colleagues, who have shown that prediagnostic alcohol consumption may slightly improve colorectal cancer outcomes. Parenthetically, these researchers also have studied patients treated in a North Central Cancer Group Trial, and reported that mild to moderate red wine consumption was associated with slightly improved prognosis.
Who cares?
What was the underlying biological hypothesis — presumably a rehash of the old resveratrol story? Why did the researchers do these studies and why were they published?
I ask myself that question particularly because Phipps and colleagues published data in Cancer in 2011 suggesting an absence of obvious outcome association between alcohol intake and colorectal cancer outcomes. Did they not believe their own data? Most puzzling is that, 5 years later — after these researchers have consumed salaries, grants and time — they have again demonstrated another similarly weak association without a strong governing biological hypothesis and have managed to get it published.
I do not really intend to pick a fight with the anti- (or is it pro-) alcohol lobby group. I have become an equal opportunity curmudgeon, and am equally troubled by the litany of papers reporting weak associations — positive and negative — between cancer and sleep patterns, metformin and other oral diabetic agents, variable nutritional habits and diet, coffee, and a range of much wackier epidemiological targets.
Basic questions
I fully endorse the outstanding work still being done by many luminaries in the field — many excellent studies will influence definitively our approach to cancer avoidance, change of lifestyles, surveillance of long-term survivors, etc — but those paradigm-shifting studies are not the focus of this editorial.
However, at a time when the national health care well is running dry, perhaps the world of cancer epidemiology might consider policing itself more effectively, and pose some basic questions before initiating projects, funding them and then reporting their datasets:
Maybe this could begin at the NIH with a review of the composition of the grant review sections that focus on epidemiology and related topics, and perhaps a structured analysis of the return on investment from the past decade of funded cancer epidemiology studies.
That may yield surprising results, and might even prove — once again — that my thinking is wrong!
References:
Phipps AI, et al. Cancer. 2011;doi:10.1002/cncr.26114.
Phipps AI, et al. Cancer. 2016;doi:10.1002/cncr.30446.
Phipps AI, et al. Int J Cancer. 2016;doi: 10.1002/ijc.30135.
Rivera A, et al. Cancer Epidemiol Biomarkers Prev. 2016:doi:10.1158/1055-9965.EPI-16-0303.
For more information:
Derek Raghavan, MD, PhD, FACP, FRACP, FASCO, is HemOnc Today’s Chief Medical Editor for Oncology. He also is president of Levine Cancer Institute at Carolinas HealthCare System. He can be reached at derek.raghavan@carolinashealthcare.org.
Disclosure: Raghavan reports no relevant financial disclosures.