On the importance of luck

That paper that proposed that most cancers were due to bad luck, that is, that they were a consequence of biological factors that could not be controlled, has been surprisingly controversial. I thought it was a fairly unsurprising paper that confirmed what we already suspected, but wow, the furious pushback has been something to behold.

Today, though, a couple of MDs have responded to the paper and reinforce what I said.

Steven Novella thinks the general logic is sound.

This is an interesting study and it will be interesting to look at replications and other methods, if they are available, of making the same sort of estimation. What this study suggests is that at least 2/3 of all all cancers are due to random mutations – bad luck. The figure may be higher once breast and prostate cancers are included. Of the remaining third it is not clear how much is due to inherited genes vs lifestyle factors.

The logic of the study is sound, in my opinion. The authors assume that lifestyle and genetic factors affect the risk of tissue specific cancers, but not cancer in general. This study would miss, however, lifestyle or genetic factors that affected the risk of all cancers (regardless of tissue type) equally. One might argue, therefore, that it overestimates the role of random mutations, but that is only if you accept that there are universal risk factors out there.

David Gorski emphasizes that the result actually fits well with prior estimates of the relative contribution of environmental/genetic factors and a probabilistic component.

That cancer is due to a combination of random probabilistic processes, environmental exposures, and heredity is a non-controversial statement. What is controversial are estimates of the relative contribution of environment, given that the percentage of cancers due to inherited cancer-causing mutations is known and low. Take the example of breast cancer, which is a cancer for which environmental and lifestyle contributions are not particularly high, with perhaps 27% of breast cancers being due primarily to environment (which includes diet and exercise, as well as hormone replacement therapy). The vast majority of those environmental contributions come from obesity and alcohol consumption, neither of which reaches the double digits, percentage-wise. Yet there are organizations that promote the idea that “chemicals” in our environment are a major cause of breast cancer. Unfortunately, about 5-10% of breast cancer is inherited, while perhaps up to 27% has a strong environmental component. That leaves around 60% of breast cancer (or, even using higher estimates, at least 50%) as falling into the “we don’t know” or “stochastic” category, with, sadly, nothing that we know of right now that can be done to prevent these cases, while the 10% of hereditary cases can only be prevented by aggressive means, such as chemoprevention or prophylactic surgery.

I still find it interesting, though, that so many people have complained to me about ascribing phenomena to chance -- there are some serious misconceptions floating about out there, and I've always taken these ideas for granted. I guess I can't. David Colquhoun and I spent an enlightening afternoon yesterday trying to get through to a few people on Twitter who could not believe we actually thought chance was a reasonable explanation for anything. I did not have the impression that these were anti-science people, or creationists, or anything absurd like that -- they just had a striking psychological antipathy to the whole idea of random effects.


no not chance. There must be factors of which we are unaware. U can't abandon sci. method!


Luck, by definition, has no cause therefore unscientific.


random does not mean without a cause - which is what luck means :-)

Fascinating. Random events are not even scientific? Where did this idea come from? Everything has to have a "cause" of some sort? Weird.

Maybe it's because a lot of my early training in biology was in genetics, and there you acquire a strong appreciation for the importance of chance events. Genetic gene mapping, for instance, is done by looking at recombination frequencies -- the probability that a meiotic crossover event will occur between two genes on the same chromosome, which is a factor of the physical distance separating them. We understand the physical basis of this event, which involves a protein, Spo11, that binds to a random location on the chromosome and induces a double strand break. Why does it land on a particular spot? It's all about the higgelty-piggelty jiggling of proteins in the cellular environment -- there isn't a magic finger telling Spo11 to go to a pre-defined place on the chromosome, it simply does its job wherever it happens to find itself.

You could say the recombination event has a specific cause, the protein complex that cuts and swaps strands of DNA, but the question at hand -- why does it recombine at a particular spot in a particular chromosome? -- is not specified by any causal agent in the cell. It's random.

It's not unscientific. We can study chance processes statistically, no problem. If we threw out all study of chance as unscientific hocus pocus, well, there goes genetics. And epidemiology. And chemistry. And any science that uses statistics. Dang.

Why do they reject chance? One idea that emerged is that they have an excessive faith in causality, and paradoxically, too much trust in the ability of science to give complete, exhaustive explanations for everything.


A deeper and better understanding of DNA will erase the concept of luck/random.

No, it won't. The more I learn about chemistry and biology, the stronger the value of understanding chance becomes.

I think the idea is that all we have to do is catalog all of the efficient causes to work out every step of an event. Your cancer was caused by a cosmic ray striking and damaging the short arm of your 12th chromosome, creating a defective RAS oncogene. That cosmic ray originated in a supernova 15,000 light years away. That exploding star condensed from a cloud of matter that originated in the Big Bang, so all we have to do is map how every atom, from the beginning of the universe to that detonation in a distant star, and further, every molecular event in the evolution of that RAS oncogene that put it in that particular location on the chromosome, and then every event in your life that led to that cell and your body to be in that specific location to intercept that cosmic ray, we'll finally understand why you have cancer.

It takes a very deterministic attitude to find that explanation at all satisfying.

No matter how hard we work, we will never have a sufficiently detailed explanation of every feature of the universe to negate the importance of chance. I think quantum physics is also drilling down deep into the nature of how the universe works, and finding that chance plays a role; but even if it were found that the universe is completely deterministic, the complexity of the phenomena and the number of parameters means that those kinds of causes are unknowable, and randomness is a good higher-level description of what is going on.

So get used to it. Why did you get cancer? Bad luck. Chance. But of course, the odds might also have been skewed by inheriting a gene that predisposes you to cancer, or by a poor diet, or by your odd habit of spiking your morning tea with N-ethyl-N-nitrosourea.


More like this

Physicists, since quantum physics, have had no problem distinguising random - in the sense of not predictable- and causation.


By test@test.com (not verified) on 05 Jan 2015 #permalink

Well of course citing random processes as a cause is going to raise hackles. A whole lot of people base their entire lives upon the assumption that they have control and that that control, if exercised well, will interact with a fair and just universe to give outcomes that are under their control and justified by their actions.

Anything hinting that steering wheel and levers on the busy-box of life are not strongly connected to much of anything may be traumatic.

Folks might argue there isn't really a probability for "heads" on the next coin toss - it's actually determined by various causes. What do they do to ask if the coin is "fair" (or has different p than another coin) - say the question is not interesting or not scientific?
We make models with random variables for a reason - they have uses. They help us make decisions for example.
Please tell me these twits weren't scientists.

First, I don't understand your rebuttal in the other blog to the critics who said that correlation of cancer rates with number of cell divisions did not prove that errors in DNA replication had no environmental causes. If a cell has one important gene damaged by an external cause, whether by a cosmic ray or UV light or cigarette tar, and that cell normally will never divide again, the chance that it will subsequently have damaged all the other necessary genes for it to turn into a cancer that grows out of control will be very small. If it divides repeatedly, so that there are ultimately a large population of daughter cells with the same error, there's a greater chance that one of them will, by chance replication errors OR by exposure to carcinogens, develop the additional needed errors. Thus, even if you presumed as a thought experiment that ALL non-genetic cancers were caused by environmental exposures rather than random replication errors - which certainly is not true - you would expect to see a strong correlation between cell division rates and cancer rates.

Second, it's odd that Novella and Gorski are so confident that breast and prostate cancer will prove to be even more attributable to random, unavoidable fate than the two-thirds estimate derived from other cancers, since in fact age-adjusted incidence rates of these cancers vary among cultures by well over one-third. (For that matter, overdiagnosis alone may raise rates by a third - where does that fit in the distribution of causes proposed?) Of course Western cultures and particularly our culture have displayed some of the highest rates of these diseases, and these men are both known for having cultural axes to grind, e.g., those who question the safety of just about any modern commercial technology, such as the widespread use of endocrine disruptors in consumer products, may be accused of hating Science.

First of all, we need to get rid of the word "luck" in science except in social sciences studies of relevant attitudes and beliefs.

Here's a reasonable and scientifically viable definition of "luck": "The belief that random events are influenced for or against oneself based on the qualitative aspects of one's personality or character."

In other words, belief in "the personalization of randomicity," as in, belief in the Sun God is "the personalization of the Sun." Attribution of personalization where it does not exist.

Reputable science journalists should NOT use the word "luck" in stories of this kind, any more than they should use phrases like "the Particle Faerie" when writing about the outcome of physics experiments.

What we're left with is the plain-vanilla statement that the majority of cancers are caused by random factors that are not presently predictable. OK, if that's empirically correct, then so be it, and our job is to adjust our attitudes to reality and then deal with it in the best way possible.

"Ma'am, your cancer was a random occurrence, and has nothing to do with your personality or character or behavior. But we're going to fight this thing together: I can give you the best treatment that science has to offer, you can build up your strength and courage for the fight ahead, and there's a good chance you'll get better." Offer science-based medicine plus emotional support, work with the patient's beliefs (e.g. if they're religious, reinforce the support they receive from their congregation), and proceed.

I'm instantly suspicious of "scientifically viable" definitions of words, if they are not the dictionary definitions, that segue into "of course this does not exist so the word should never be used again". Sometimes we do "have good luck" or "have bad luck" in an ordinary dictionary usage of the word, that of experiencing unusual good or bad events that we did not obviously cause or bring about, and my feeling is that most people do not think that their "personality or character" is responsible for those events. If you did, you might call it karma, or say that God or the gods were rewarding or punishing your good or bad behavior. That wouldn't be luck; it would be viewed as a non-obvious cause and effect relationship. "Bad luck" is the ruinous random event that may happen even to the person of delightful personality and excellent character. Sometimes a perfectly nice person has one devastating loss or harm after another after another for years, all undeserved, and to say that he "has bad luck" may not mean anything from the perspective of physics, but from the perspective of a human life as a narrative story, it's a meaningful statement.

What if you had cancer? Would you think You have bad luck eh?

right i agree its not bad luck istent all it it is in ur blood some people dont know

By reallly love (not verified) on 07 Jan 2015 #permalink

Second, it’s odd that Novella and Gorski are so confident that breast and prostate cancer will prove to be even more attributable to random, unavoidable fate than the two-thirds estimate derived from other cancers, since in fact age-adjusted incidence rates of these cancers vary among cultures by well over one-third.

I see that Jane's trip retains it's distinctively bland incoherence when away from trying to play Washing Machine Charlie at RI.

I see that Narad still thinks ad hominems are adequate substitute for logical, numerical or fact-based arguments, a tactic that is welcomed in Gorski's playpens but, I have been happy to get the impression, is considered less definitive by our current host. Perhaps he objects to the use of "fate" for something that the universe dishes out at random and might not have dished out; I figure in common speech it is a meaningful way of referring to an event that you could not by making any alternate choices have avoided, which is the claim here.

Now to give an example of a factual argument: People who know anything about biology know that when you attempt to examine how much of the variation in a trait is due to genetics and how much to environment, this is ALWAYS, inherently and unavoidably, context-dependent. If all individuals are kept in identical conditions from conception onwards, then all observed variation must be due to genes or to interaction between genes and environment. The more variable the environment becomes, the more possible variation due to environment there will be. Got it?

Thus, the percentage of cancers of a specific type that are due to "bad luck", versus genetics or potentially identifiable environmental factors, will vary geographically and temporally depending upon local variation in genetic risks (Ashkenazi Jewish women are more likely to carry genes that give a high risk) and environmental risks (American women are more likely than women in many nations to have children late or not at all, to be obese and inactive, or to be marinated in endocrine disruptors).

Gorski would have us believe barely over a quarter of breast cancer is environmental, but several lines of evidence suggest otherwise. (1) The recent variance in age-adjusted incidence among nations is greater than that. (2) Girls from Japan, say, who move to the U.S. and start living American-style end up with American-style risks. Hence the traditionally much lower incidence in Japan can't all be attributed to genetic differences (which Gorski thinks is smaller anyway). Is it geographically distributed variation in bad luck? (3.) Total population incidence in America has varied more than that within our lifetimes, during which time one wouldn't expect either our genes or our luck to have changed much. The large increase in the late 20th century, followed by a noticeable decline, is attributed in part to the rise and fall in popularity of cancer-causing HRT (as Gorski admits). If a weirdly specific 27% of breast cancer is environmental now, what was the percentage in 1990 when there were many more iatrogenic cases? (4.) A number of epidemiological studies have shown greater than 27% variation in risk within Western cohorts associated with single risk factors, for example, HRT, but also diet (one pattern in the UK Women's Cohort study associated with a 40% reduced risk, which phrased otherwise means that those with the unhealthiest pattern had a 66% increased risk), obesity (recent meta-analysis suggests 26% increased risk relative to normal-weight people) and various others. Some of those factors are not independent, but when you consider them all together and add in other emerging risks, such as unwitting exposures to various industrial chemicals, it's hard to believe that only 27% of breast cancer is neither genetic doom nor totally random.

I see that Narad still thinks ad hominems are adequate substitute for logical, numerical or fact-based arguments, a tactic that is welcomed in Gorski’s playpens

Jane, that's a lot of bloviating to demonstrate that you didn't understand what the boldface was for:

Fate, by definition, isn't random.

Got it?

There's some extra ambivalence in the term "luck" that the discussion probably doesn't need. There's enough problem explaining that "random" doesn't mean "without a cause," but rather "without a cause we can [in some cases practically, in others as a matter of principle] predict."

"Luck" brings in yet another layer of confusion, since it is sometimes used to mean fortunate beyond rational explanation, and sometimes randomly fortunate. (E.g., we all thought my aunt was extremely lucky - she won a car in a sweepstakes once - more so than the laws of probability would dictate. And I was once lucky to escape with my life from a potential industrial accident. Nothing to shake the laws of probability there.) So if "random" rather than "bad luck" was used, this might have gone down *slightly* easier. If "stochastic" had been the term, few enough people would have known or cared what that term meant that the paper would likely have sunk without a trace of a comment from the wider media. :-)

"Your cancer was caused by a cosmic ray striking and damaging the short arm..."
I was under the impression that everyone develops cancer cells occasionally, with the immune system normally destroying those cells. Cancer develops when the cells somehow escape destruction. So the cause would be a compromised immune system or maybe just a random failure of it. Still plenty of room for chance of course.

I dint think so.