The son of an English merchant and, himself a surgeon, James Lind must have wondered how he ended up being tossed around on the HMS Salisbury just off the coast of France. After entering his medical apprenticeship in 1731, Lind eventually served in the Royal Navy as a surgeon-in-training, ultimately sailing as far as the West Indies. The combination of poor ventilation, constant dampness, food and water rations, and lack of personal space onboard these floating cesspools would have served to make Lind further question his life decisions. Instead, he channeled his efforts to improve hygienic ship conditions by promoting ventilation for passengers, along with cleanliness and fumigation of the corridors below deck – epidemics of cholera, typhus, and dysentery were common onboard. His efforts to distill sea water would provide passengers with a clean drinking source, a successful measure to decrease rates of dysentery and other water-born illnesses.
Yet, a rampant condition aboard Britain’s navy vessels continued to stymie Lind and other physicians. It had been known for nearly a century that sailors were commonly plagued by a condition known as scurvy, yet no treatment was in sight. A typical barrage of symptoms would strike its victim, starting with fatigue, weakness, and soreness within the arms and legs, eventually progressing to bleeding, nonhealing wounds, and finally, death. Lind was to have said that scurvy was killing more sailors than opposing armies. The cause remained elusive, and Lind and others believed it was from the rotting of food within the body, a process known as putrefaction. The natural cure, Lind felt, was to increase acids within the body to help break down the rotting organic matter as a remedy for the putrefaction.
The hypothesis of curing scurvy with acids was not unique to Lind. Over a century before, the surgeon for the East India Company, John Woodall, had recommended fresh foods and citrus fruits for similar symptoms. Prior to this, Portuguese explorers were consuming citrus fruits as a scurvy remedy in the early 1500’s. Unlike his predecessors, Lind set out to prove his hypothesis scientifically. During a voyage throughout the Bay of Biscay, he took 12 sailors with weak legs, putrid gums, and lassitude – all symptoms pathognomonic for scurvy – and divided them into six groups of two sailors each.1 All 12 men subsisted on the same diet of water and gruel sweetened with sugar for breakfast, mutton-broth or pudding for dinner with occasional boiled biscuits with sugar, barely and rice, or wine. Lind then supplied these epicurean meals with several potentially medicinal foods, with each group receiving either a daily quart of cider, 25 drops of sulfuric acid, six spoonfuls of vinegar, one-half pint of seawater, two oranges and a lemon, or barley water with a spice paste, respectively. Within six days, Lind had run out of oranges and lemons, cutting the experiment short, however, he had already noticed that one of the men in the fifth group, partaking in the citrus fruits, was no longer experiencing his scurvy symptoms. The first group, with the added cider, appeared to be improved physically as well. However, by the time they reached port at Plymouth, roughly four weeks later, the man in the citrus group was so healthy, he was now serving as the nurse for the other men. The other groups were unaffected by the acidic compounds.
As the HMS Salisbury navigated around the Bay of Biscay, Lind’s acidic theory of the cure for scurvy was disproven by what is now considered the first documented controlled clinical study. Simply viewing the associated benefit of citrus fruits for scurvy may have led to the incorrect assumption that acidic foods cured scurvy. Lind instead compared the different substances head-to-head, some simply with acid, and others with more robust ingredients. Lind’s trial would eventually prove that the ascorbic acid within the citrus fruits, known as vitamin C, was responsible for the treatment of scurvy (scorbutic means “related to scurvy”). While Lind’s results were ignored (an often common theme within the history of medicine), what he may not have realized was that he had performed the first clinical trial, serving as the basis for experimental research over the next several centuries.
A drawing of scurvy by British ship surgeon Henry Mahon over a century after Lind’s study.
Cancer Nutrition Research: From Lind to Epidemiology Studies
The first randomized controlled trial would not take place until 1946. Austin Bradford Hill, the boss and colleague of the famous physician and epidemiologist, Sir Richard Doll, would test the efficacy of streptomycin for tuberculosis treatment. While Lind served as the randomization technique in his earlier studies, Hill used a statistical determination based on random sampling to decide which subjects of the trial received the experimental treatment of the antibiotic streptomycin versus the standard of care – bedrest. Streptomycin was found to be significantly more effective at combating tuberculosis. This technique remains the gold standard in medicine and several arms of scientific research.
Shortly after Hill’s monumental randomized study, Doll’s warning on jumping to conclusions from epidemiological findings may have been considered nothing more than an afterthought at the time. As Doll cautioned, associations can suggest relationships between certain behaviors and cancer, but this suggestion forms nothing more than a hypothesis – further research is essential to confirm or dispute relationships before they become the “basis for preventative action.” Finding coincidences within data are not uncommon, especially when the coincidence is related to an unseen factor, and is in fact no coincidence at all. Four decades later, however, his final words may seem more germane as recent studies find conflicting relationships so often that it remains difficult to keep these associations straight. Like any association, mechanistic evidence – in other words human or animal studies that provide a physical or physiologic rational explanation for the association – must provide an explanation.
But, animal studies were expensive, messy, time-consuming, and imperfect. This messiness, combined with the propensity of certain countries and hospitals to begin more closely tracking their patient data, ushered in an era when researchers were able to perform massive scientific studies while sitting at the desk in their office – Walter Willett could analyze the data from thousands of his nurses without ever setting foot outside during the cold Boston winters. Furthermore, these studies sidestepped the ethical issues of purposefully altering individuals’ habits by simply observing their behaviors and the eventual consequences of those behaviors. As a result, these studies provided limitations, even when compared to James Lind’s rudimentary maritime clinical trial. These same limitations, in a sense, proved to be these studies’ strength, as the ease of data acquisition would fuel the generation of countless scientific reports, and thus served as the birth of the modern-day population study.
Cancer nutrition research is a little more complicated these days, but we would be wise to remember Lind’s lemons.
Many associations have been unearthed by population studies, and many of us have fallen victim to on over-reliance on these links to support our own personal biases and lifestyle habits. News and media sources constantly batter us with study results, summed up in quick and catchy headlines. The latest and greatest population or epidemiologic study reveals that coffee “prevents” cancer, only to be replaced by a similar study days later claiming that coffee causes cancer. In reality, neither study soundly demonstrates that coffee causes nor prevents cancer, but rather that it may be associated with an increased or decreased risk, coincidentally or not. Often, these news releases get used to support our lifestyle behaviors and ignored when they oppose them.
In medical school, I would track which substances were found to provide benefit or harm based on these studies, with the ultimate goal of creating a massive online database open to the public. This could have potentially provided some insight into which foods and lifestyle habits can lower our risk of cancer. The back-and-forth of studies, wide and rampant spread of conflicting information, and realization that most foods were found to both cause and prevent cancer, led me to quickly abandon the practice. Instead of providing concrete answers, the spread of results in these population studies, and especially those studies assessing risk of cancer, seemed to be consistently widening in their results. With each conflicting result, confusion within the public’s eye has begun to grow. The expansion of the internet’s reach, coupled with the media’s propensity of promoting “latest and greatest” studies, has further promulgated these ever-growing issues.
While the endless back and forth of results of these population studies has created well-deserved skepticism, they remain important. Yet, according to a Gallup poll in September 2016, only a third of viewers feel confident in the news reports from the media. For many, the nightly news is their source of health information, and between politics and health scares, the back and forth mentality only works to further enhance public dismay. Furthermore, while population studies serve to find connections in our activities and health outcomes, the majority reveal infinitesimally small associations. For instance, when Doll and Hill analyzed the risk of lung cancer and its association with smoking, they found an undeniable connection between the two. Those individuals who smoked 35 or more cigarettes per day were experiencing a death rate from lung cancer that was 40 times higher than nonsmokers.2 The strong relationship between the two led to the view that “beyond a reasonable doubt” smoking was not only associated with lung cancer, but it caused lung cancer. The study findings were immediately incorporated into public health measures. We were only midway through the twentieth century and one of the largest studies in the history of mankind had produced an undeniable relationship between a social habit and cancer. Doll and Hill were eventually knighted, and their impact on public health would be immeasurable.
What other findings remain unearthed? The future was bright for the budding field of epidemiology. Doll’s next goal was to find which foods were associated with cancer. However, the scientific field was unaware of just how high the bar was raised by Doll’s initial knightly work. Any comparison of more recent population studies with the poster child of population studies raises several issues. This initial work strongly implicated smoking as a cause of lung cancer, revealing a 40 times higher risk of contracting lung cancer in those who smoke (assuming lung cancer was uniformly fatal at the time). In actuality, these numbers reveal that the absolute risk of a lung cancer diagnosis, which occurs in 16 out of 100 smokers, is 16%. The absolute risk certainly seems less impressive when compared to the relative risk of 40. However, context must be applied, as the normal rate of lung cancer in the average nonsmoker is miniscule. Thus, that number only rises to 16%. However, the lethal nature of lung cancer still leaves us frightened with a 16% risk, albeit less impressive overall. Further cementing the association is the stepwise increase in lung cancer death rates with increased cigarette usage, eventually approaching 3,000% for heavy smokers.
To this day, the link between smoking and lung cancer remains one of the strongest ever perceived in epidemiologic studies. Fast forwarding several decades, the research field is now littered with population studies attempting to mimic these findings with other chemicals and foods in which humans are often exposed. For example, in 2012, researchers from the Division of Nutritional Epidemiology at the Karolinska Institutet in Stockholm, Sweden, published a tempestuous study that quickly led to a media frenzy when they linked processed meat with pancreatic cancer. Within their report, the researchers found that the consumption of 50 grams of processed meat – just a few slices per day – was associated with a 19% increased risk of the often-deadly pancreatic cancer.3 They did caution, much like Doll, that their findings may have been influenced by those hidden coincidences known as confounders. Regardless of these concerns, The British Broadcasting Corporation, or BBC, reported that consuming just a couple of slices of bacon per day could increase our pancreatic risk by 20%, while CBS news reported that only three sausages or six strips of bacon could increase our pancreatic risk by 57%.
Fear ensued, as the public began questioning their physicians regarding this killer that could be found on the shelves of every grocery store. Yet, upon peering closer at the results, this example illustrates just how far population studies have regressed since Doll’s landmark study. The study revealed a 19% higher risk of pancreatic cancer, while Doll revealed that heavy smokers had a 40 times higher risk of dying from lung cancer. To put these numbers into perspective, we must account for the baseline risk of pancreatic cancer, which affects 1 in 67 individuals throughout their lifetime. Therefore, the overall risk of pancreatic cancer is 1.5%, and with an increase based on the 19% provided in this study, this risk would increase to 1.8%. If this risk was in fact more than a coincidence – which remains unresolved – comparing this 0.3% increase to Doll’s 16% increase in lung cancer paints a vastly different picture of concern. Doll had immediately stopped smoking upon exposing his lung cancer data, but whether this would have prompted him to remove his morning sausage is doubtful. Furthermore, whether this warrants media headlines questioning whether processed meats are as bad as smoking – which actually occurred – or if we have reached a point where fear mongering has overwhelmed the actual science, remains unclear.
Cancer Nutrition Research: Have Epidemiology Studies Set us Back by Decades?
The current issues with using population studies to exaggerate small, coincidental, or perhaps more often incorrect, association have been brought into the spotlight recently. Even Doll’s worldly food study from the 1970’s4 yielded results that were incomparable to the association of lung cancer and smoking, likely leading to his concern with taking away any solid conclusions from his findings. Fast forwarding a half century, current population studies rarely even approach a small fraction of this risk. Perhaps most enlightening, was the humbling realization, now almost 70 years later, that our exaggerated enthusiasm was more hopeful than reliable. The relationship that Doll and Hill had discovered was enormous compared to studies that would follow, and the realization began to set in that the gravity of the results of their epic study would be hard to reproduce. Per Willett’s experience, not only were many associations small, several were starting to present in the opposite direction. After providing us with an insurmountable precedence, what followed the initial study was a deflating realization that correlation does not necessarily mean causation, and the clear majority of studies that followed Sir Richard Doll’s work exposed a dwindling correlation. The sobering phrase “correlation does not equal causation” now serves to put those coffee drinkers in check, when they gallivant around the office exclaiming to their coworkers that their excessive coffee habit is defending them from cancer according to the latest and greatest study.
This lesson was put in the spotlight in 2013, when Jonathan Schoenfeld, a radiation oncologist and Assistant Professor at Harvard Medical School, and John Ioannidis, Professor of Medicine, Health Research, and Policy at Stanford University School of Medicine, joined forces to assess the impact of ingredients from random recipes in The Boston Cooking-School Cook Book. To add some humor and much insight into the spiraling world of population studies, they published Is everything we eat associated with cancer? A systematic cookbook review, in the American Journal of Clinical Nutrition.5 Taking the ingredients from the various recipes within, and performing an array of statistical calculation, they found studies providing associations with both an increase and decrease in cancer risk. Furthermore, much like the processed meat and pancreatic cancer study mentioned before, the overall impact of these ingredients and foods was miniscule. Additionally, many of these findings became negligible when multiple studies were evaluated in what is known as a meta-analysis, again supporting the drastic deviation from the colossal impact Doll and Hill found with lung cancer and tobacco. Ioannidis, also a Professor of Statistics at Stanford University, has since produced countless other studies exposing issues with the many similar reports within the research world, especially those pertaining to health and nutrition. Many of these have helped fuel the massive controversy of what the general population should be told to eat – if told anything at all – to follow a healthy lifestyle.
While I think the last century of studies assessing the link between fasting, carbohydrate restriction, periodic ketosis and cancer prevention are encouraging, we are now playing catch up due to several decades of a reliance on epidemiologic studies. We need to take some of the encouraging results that exist, and instead turn towards clinical trials like Lind. In other words, nutrition research in the cancer world could use a good dose of vitamin C.
Cancer Nutrition Research References:
- Bhatt, A. Evolution of clinical research: a history before and beyond james lind. Perspect. Clin. Res. 1, 6–10 (2010).
- Proctor, R. N. The history of the discovery of the cigarette-lung cancer link: evidentiary traditions, corporate denial, global toll. Tob. Control 21, 87–91 (2012).
- Larsson, S. C. & Wolk, A. Red and processed meat consumption and risk of pancreatic cancer: meta-analysis of prospective studies. Br. J. Cancer 106, 603–607 (2012).
- Armstrong, B. & Doll, R. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practices. Int. J. Cancer 15, 617–631 (1975).
- Schoenfeld, J. D. & Ioannidis, J. P. A. Is everything we eat associated with cancer? A systematic cookbook review. Am. J. Clin. Nutr. 97, 127–34 (2013).
© 2019 CDR Health and Nutrition, LLC. All Rights Reserved.
Pingback: Nutritional Orthodoxy: When Science becomes a Belief System - Colin Champ