How Little Is Too Much?
How Do We Know What Really Causes Cancer?
A patient is rushed to the emergency department by ambulance after having a seizure at home, witnessed by several family members. In the ED, the patient, a 35-year-old man with no history of epilepsy, is lethargic and difficult to arouse.
Physical examination, blood and urine tests, and a CAT scan of the head are all obtained. While doctors are considering whether to perform a lumbar puncture (spinal tap) on the patient, the first batch of laboratory test results come in and there is a notable abnormality: the patient’s blood level of sodium is very low.
While adding extra sodium to his intravenous fluids, family members arrive in the ED and tell the doctors that their relative suffers from a severe mental illness and has been drinking a great deal of water for several days. A diagnosis of water intoxication, secondary to psychogenic polydipsia (the technical word for excessive fluid intake) is made. The patient is successfully treated for the dangerously low level of sodium, called “hyponatremia,” and transferred to the inpatient psychiatry unit of the hospital for further treatment.
Water intoxication? How could that be possible? Doesn’t every wellness guru admonish us to “hydrate, hydrate, hydrate?” Aren’t we warned not to imbibe sugary drinks or even diet soda, but to choose safe, healthy, uncomplicated H2O instead? So how could drinking too much water even be possible, let alone the cause of a near fatality?
Pathological water intoxication is, in fact, a real entity. To be sure, it is uncommon, and it takes a lot of water to cause it. Most cases occur in people with severe mental illness, but anyone who drinks too much water or is given too much intravenous fluid is at risk for water intoxication. Low sodium levels caused by the dilution of salt in the blood results in brain swelling and, if untreated, death.
We discuss this case to make an important point about health and quantity: anything taken to extremes can cause illness and even death. Even something as innocuous as water.
Which is why we all need to be very careful when blaring headlines tell us that some food, product, or common exposure in our environment causes cancer. Seemingly every day we are warned that scientists have found that some substance we use to kill weeds in our gardens, a constituent of plastic bottles or an artificial sweetener is a “carcinogen”, the technical term for a any substance that increases the risk for getting cancer. The stories are frightening, especially when they are about something we take for granted or don’t even realize we are being exposed to. And because we all fear that mysterious group of diseases we call “cancer,” we shudder to think that using our cell phones or living near power lines may be slowly mutating the cells in some organ of our bodies and causing cancer.
But do we stop and ask what it means when scientists are quoted as saying that their experiments indicate something formerly believed innocuous is in fact capable of causing cancer? And once a journalist publishes a story that something “might” increase the risk for cancer, does the journalist follow-up and tell us a year later when further studies fail to corroborate the initial warning?
The only absolutely definitive way to prove that something causes cancer would be to perform a controlled experiment with humans. In such a study design, human volunteers would be randomly assigned to either be exposed to a putative carcinogen or to some identically appearing placebo. Then, after many years of exposure and observation, scientists would count the number of cases of cancer in the group that received the real substance and if the incidence of new cancers exceeds the number in the control group, we would know for sure that the exposure caused the cancers.
But of course, that kind of experiment is never—and could never—be done. We can’t deliberately give people something that might cause cancer and we can’t follow up on large enough numbers of people for the many years it takes for something to cause cancer. So scientists must use three other methods to determine if something causes cancer.
The first method is an observational human study. In this design, you take people who have already developed a specific type of cancer and see if they have been exposed to something that is hypothesized to cause cancer more often than people who don’t have cancer. This is the way we first found out that cigarette smoking causes cancer. People with lung cancer are overwhelmingly more likely to be or have been smokers than people who don’t have lung cancer. Depending on the type of lung cancer, smoking increases the risk for lung cancer by as much as 100 times the rate for non-smokers. The difference is so large that there is absolutely no question that smoking causes lung cancer.
But this method has problems because most exposures are not as clear cut as smoking is for causing lung cancer. And when the difference is small, it becomes very hard to know for sure if it is the exposure that caused the cancer or something else. Let’s say that instead of increasing the risk for cancer by 100-fold, an exposure increases the risk by 20%. Scientists express risks like this with a statistic called the “odds ratio.” Cigarette smoking has an odds ratio for some kinds of lung cancer of 100 or more. An increase in risk of 20% translates to an odds ratio of 1.2. When newspaper headlines declared that red meat causes cancer, people were understandably alarmed. Americans eat a lot of red meat. But if you look at the studies that support this contention, the odds ratios are around 1.2. Given how much red meat we eat, even a 20% increase in risk will be statistically significant and maybe even legitimately worrisome. But it is still far less risky than smoking. Furthermore, people who eat a lot of red meat might also smoke and drink alcohol more and exercise less. Observational studies have a tough time separating out these various risks, making it very difficult to know for sure that it really is red meat doing the damage. Finally, whereas no studies have ever failed to show that cigarette smoking causes lung cancer, several studies found no association between red meat and cancer.
So given the difficulty getting accurate information about what causes cancer from large-scale observational studies of humans, a faster and more direct way to figure out if a substance is a carcinogen is to test it in the laboratory. There are two ways to do this. One is to use preparations of cells or bacteria and see if the substance causes mutations in their DNA that are consistent with the induction of cancer. The other is to give the substance to laboratory animals like mice and rats and see if they develop tumors or other forms of cancer. If these tests are negative, it is safe to assume that there is little risk that the substance in question will cause cancer in humans.
But what if these tests are positive and, for example, a rat develops brain cancer from being fed a common preservative in foods we eat? Does that mean we are being poisoned by that preservative?
An important problem with interpreting these studies is the big leap from animal to human biology. A good example of this is the supposed cancer risk from cell phones. Just as we thought this issue had sunk beneath the public’s radar, new studies in laboratory rats that have been widely reported in the media suggest a link between cancer and cell phone use. But there are at least three challenges in interpreting these studies. The first is biological plausibility: unlike X-rays and sunlight, the radiation from cell phones is of the non-ionizing type. While non-ionizing radiation can heat tissues, it does not cause the kind of DNA damage that leads to cancer.
The second issue is whether the results in rats are applicable to humans. In male rats, but not female rats or mice, cell phone-type radiation caused a statistically higher rate of a type of tumor of the heart called a schwannoma. But schwannoma’s of the heart are incredibly rare in humans. Certainly no one has noticed any increase in the incidence of them since we began using cell phones several decades ago.
Finally, we have the paradoxical finding in the studies that the radiation-exposed rats actually lived longer than the controls. In other words, there is a lot to consider before reading a headline that declares “New studies link cell phones to cancer” and throwing your phone away.
Another major challenge in deciding whether laboratory cancer studies are relevant to humans involves dose. This can be seen in the flap over a California judge’s recent ruling that coffee must be labelled as a carcinogen.
Carcinogenicity studies in the laboratory are done by increasing the amount of the substance under study to levels that are often far higher than humans could ever come close to consuming. At extremely high doses, laboratory rats seem capable of developing cancer to almost anything. And we have already noted at the beginning of this article that even in humans, benign things like water can be highly toxic when taken in massively excessive amounts.
Coffee naturally contains a chemical called acrylamide, which has been shown to increase the risk of cancer in rats when put into their drinking water. But that only happens when rats are given thousands of times the dose of acrylamide that humans can possibly consume. As Aaron E. Carroll wrote in the New York Times on April 23, 2018, “Cigarettes have a clear and easily measured negative impact on people’s health. Acrylamide, especially the acrylamide in coffee, isn’t even close”.
Bruce Ames, who developed a laboratory test called the Ames test that is frequently used to determine if something might cause cancer, himself wrote that “…at the low doses of most human exposures, where cell killing does not occur, the hazards of rodent carcinogens may be much lower than is commonly assumed.” In other words, the doses that it takes to give rats cancer are often analogous to the amount of water a person has to drink to cause water intoxication—far more than most of us come close to consuming.
A recent study showed that many people believe in “fake” causes of cancer, like stress, food additives, electromagnetic frequencies, GMOs, and microwave ovens. Will this distract people from focusing on things we know really do cause cancer, including smoking, sunlight, excessive alcohol consumption, and infection with viruses like HPV and hepatitis B and C? Does belief in fake causes of cancer alter the life styles of a significant number of people as they try to avoid things that are unavoidable, like stress and GMOs?
We want scientists and regulatory agencies to be vigorous in testing things in our environment that might cause cancer. Often, as in the case with smoking, this reveals common exposures that turn out to be deadly. But we also need to help people understand what is really risky. This means figuring out how to explain to people the following principles:
- An association between something and cancer in a large association study does not establish causation.
- There are risks and then there are risks and some “risks” reported by the media are actually very small. Scientists report on relative risks in their papers; the important thing for the public to know is absolute risks. That distinction is not a simple one to explain.
- Laboratory studies are important screening tools for detecting carcinogens, but there are many false positives. A finding in laboratory mice may have no biological translation to humans.
- Even when something is shown to cause cancer in rats, it is often at doses that are so high that humans could never get near the amount that causes cancer.
In interpreting media reports about things that are alleged to cause cancer, we must always ask how big the risk is and in whom or what was the study done. This often boils down to inquiring about how much it took to cause cancer. Or, to put it another way, how little of something is enough to cause cancer, hence the question “how little is too much?”
 Ames BN, Gold LS: Chemical carcinogenesis: too many rodent carcinogens. Proc Natl Acad Sci USA 1990;87:7772-7776.
 Shahab L, McGowan JA, Waller J, Smith SG: Prevalence of beliefs about actual and mythical causes of cancer and their association with socio-demographic and health-related characteristics: Findings from a cross-sectional survey in England. 2018.