Like in music, science has its own version of one-hit wonders. But one-hit wonders are not necessarily musical oddities, some of them are great songs. What makes a scientific one-hit wonder and how do we avoid throwing away the great information?
Exposome Perspectives Blog by Robert O. Wright, MD, MPH
I rank one-hit wonders among my favorite songs. Maybe it’s the Icarus effect of a band or artist who puts everything into one song and then burns up (or more commonly fades away). One-hit wonders may actually be the rule in popular music, and not the exception. I imagine it’s difficult to provide an exact percentage, but clearly a large number of artists have had just one hit song, unable to replicate that success again. Analyzing this phenomena in the language of science, there may be many causal factors that contribute to long vs. short term success in music. It’s not just talent, artists also need marketing, work ethic, innovation and luck. It’s a complex system, like biology, and even a talented artist may not achieve long-term success if they don’t have the right team and resources backing them up. After all, look at the enormous success of the Monkees who had a whole television network creating and marketing them. For every Beatles, there are likely a dozen Five Stairsteps, and these days, well, Gotye is just somebody that I used to know.
Science has several versions of one-hit wonders. “Big” findings that get a lot of attention but eventually fizzle because they can’t be replicated. Often these are studies that get massive press coverage when they first get published. “Coffee causes cancer” might be a major headline one day, or “Gene for Parkinson’s disease discovered” may garner bold headlines another day. These headlines may be followed by unread follow up studies buried in the small print that can’t show the same result. Or the study just gets forgotten, fading in our collective memory, with the headlines remaining in the corners of our brains living on as a kind of fuzzy urban legend—aluminum exposure and the development of Alzheimer’s disease is an example. It gained headlines in the 1960s and 70s, but subsequent studies have failed to find a convincing causal association. Many times the authors rush to announce their findings as groundbreaking without considering what it would take to be a truly groundbreaking study. Like a one-hit wonder, they just fail to replicate.
I often google one-hit wonders so I can download them onto my phone—it is a hobby. One day while on a run in my neighborhood, I was listening to Redbone’s “Come and Get Your Love” when something occurred to me. Running is my way of preventing diabetes—even if I were to carry genes that promote obesity, I can prevent it if I eat well and exercise regardless of my genetics. Obesity is reaching epidemic proportions in the last 30 years so why might that be?
There is a theory that obesity has become more common because our ancestors experienced famine frequently relative to today. In response, gene variants were selected to this harsher distant past environment that could overcome periods of famine by promoting the storing excess energy in fat. In the present day, when the U.S. food supply is steady and famine is very rare, these genes might cause an inappropriate increase in fat storage, obesity and other adverse outcomes that might explain high obesity rates today. This is called the “thrifty genotype” theory. This theory proposes that the obesity causing genetic variants are common and also implies that our diets and exercise patterns are random. In fact, it was once applied to Native Americans, who have suffered from high obesity rates. My concern with the “thrifty genotype” theory is that it used genetics alone to explain obesity and left out the impact of mass forced migration into reservations in which the food rationed by the military was mostly lard and flour. Centuries of cultural practices were made inaccessible. When it comes to obesity, diet, culture and behavior are more important than genetics. Redbone, BTW, was the first Native American band to have a top 10 single, which is likely why this all came up in my mind while listening to their song.
Another problem with the thrifty genotype theory is that food has been plentiful in the U.S. for a long time and so our obesity promoting genes used to live in harmony with our food supply. So why did obesity rates only start to rise in the last 30 years? I don’t think genetics causes obesity in most Americans—I think eating poorly and limited exercise does, and genetics plays a limited role. But as I was running I began to realize that a different genetic theory has some good lessons for exposomics.
The importance of “rare” in exposomics
The “rare variant” theory of genetics posits that most diseases are caused by the accumulated effects of rare, deleterious genetic variants. These variants occur at a low frequency in the general population, which makes sense evolutionarily, as natural selection should reduce their prevalence over time. The alternative theory that common variants cause common diseases was once popular, but has not panned out. Rare variants fits the data from genomic studies much better than common variants, and explains why we are not all suffering from genetic diseases. I am not sure we’ve learned this lesson about the importance of being rare yet in exposomics.
I should pause here and acknowledge that the meaning of “rare” is different in exposomics than in genetics. Rare in genetics is <1% of the population being a carrier of a genetic variant, while 30% detection rates may be considered “rare” in exposomics. In fact, what I am calling rare in this blog, are chemicals that are measurable in millions of people. It’s only “rarer” than what we set out to look for and not actually rare. If fewer than 30% of people in a study have a detectable level of a chemical, we may think it not worth the effort to pursue it—even though 30% of the U.S. population would be 90 million people.
Some of this thinking may be a remnant of metabolomics—the field that gave birth to exposomic tests. Metabolomics is interested in food metabolites, which are ubiquitous for obvious reasons, as we all eat some combination of carbohydrates, fat and protein that get broken down to very similar metabolites. Metabolomics focuses on the overall patterns of chemicals that come from food. The goals are different and it makes sense to weed out uncommon metabolites to save time and effort. But the exposome is about more than just diet. Metabolomics often wants detection rates of 70% or greater for its metabolites. Exogenous chemicals may be found in food, but they are typically present at lower levels than metabolites making them harder to detect than metabolites. The majority of an apple is not pesticide, and a slice of cheese doesn’t grow from the plastic wrap. In fact, we may want to ask ourselves whether we think highly toxic chemicals should be commonly measurable in everyone. After all, if >70% of the population were exposed to a very toxic chemical might we all be sick?
Perhaps it makes more sense for highly toxic chemicals to be “rarer.” This doesn’t mean that common chemicals are nontoxic—only that their toxicity might be less acute or overt on average than “rarer” chemicals. Dose will always matter, but the rarer chemicals found at low doses may be what is making a larger portion of us sick. Maybe we are missing some of the more important players in the toxic exposome, because we think that being found in less than half the population means it’s an outlier, which means it’s unimportant, which means we shouldn’t bother measuring it. Our workflows and processing algorithms favor common chemicals. Why put in all that effort to find something that is only in 10% of the population?
Discovering the next Bob Dylan
These “rarer” chemicals may be very important- just as many one-hit wonders are not just musical oddities. Many are great songs. Take “Tainted Love” by Soft Cell from the 1980s. Maybe you can’t name another Soft Cell song, but that doesn’t mean “Tainted Love” is not great. Exposomics will have its share of one hit wonders, and it will have its Rolling Stones and Stevie Wonders too, i.e. common and important exposures that replicate findings (e.g. prolonged heat waves, hurricanes, plastics, “forever chemicals” etc). All I am saying is that we should put some thought into whether we throw out a lot of useful information on rare exposures in our work simply because we want to discover the next Bob Dylan, but maybe finding a bunch of one-hit wonders instead will give us a really good playlist.