Data links 22 chemicals to increased incidence of the disease

This week, I’ve been practicing saying words like trifluralin, thifensulfuron and linuron and no – I’m not working on a bootleg audiobook recording of the Silmarillion, I am learning about pesticides and herbicides and their potential relationship to prostate cancer.

Before we dig in, I’m going to “both sides” this for a minute. Yes, pesticides and herbicides – the stuff we use to grow our crops - are chemicals. Many of them are chemicals that can not be found in nature. But that being said, these are fabulously successful chemicals. Their development is arguably responsible for the world we have today. Estimates suggest that without the use of herbicides or pesticides, crop yields would be reduced around 30%. And crop stability – our ability to expect certain yields year in and year out, without worrying about blights and mass die-offs – is critical for societal stability. If I snapped my fingers and made all herbicides and pesticides disappear, we would quickly be in the midst of a worldwide famine – not to mention an economic crisis the likes of which we’ve never seen.

But… we might have less prostate cancer.

When it comes to the epidemiology of cancer, variation is really interesting. If cancer were truly random – afflicting individuals like a lightning strike – or more accurately like an errant cosmic ray damaging some DNA – we would see a relatively uniform distribution of cases around the country, proportional to the local population sizes.

But that’s not what we see. We see hot spots, clusters, areas where there are more diagnoses and more deaths. And that’s where epidemiology becomes interesting. Explaining the sources of that variation.

And this week, a new article appearing in the journal Cancer uses that variation – across the entire contiguous United States to link prostate cancer with various pesticides and herbicides.

What I’m going to describe to you has a name in epidemiology. It’s called an “ecological study”. And that sounds like we are studying the impact of a new solar farm on the local bees or something, but in this context it refers to using aggregated data sources to make inferences about disease.

I’ll walk you through it.

The authors broke the United States down into a bit more than 3000 counties. Using data from the CDC, they mapped cases of prostate cancer and deaths from prostate cancer across all of those counties, standardized for population size.  That’s the variation we were just talking about.

Now… pesticides.  This is trickier. Most of the data came from the National Water Quality Assessment Project from the US Geological Survey. The survey catalogued 295 different pesticides, and the authors tallied up how many kilograms of each were used in each of those 3000+ counties. Here’s the map for trifluralin, for example.

They did this in two time periods: 1997-2001 and 2002-2006. Given that prostate cancer takes a long time to develop and be diagnosed, they explored the relationship between the phase 1 pesticides  with prostate cancer in 2011-2015, and the phase 2 pesticides with prostate cancer cases from 2016-2020.

Of course, there are a lot of things that happen at the county level that might be related to prostate cancer that have nothing to do with pesticides. The authors adjusted for race, poverty levels, population density, agricultural product types, machinery utilization, farm valuation, and more.

But basically the question they asked… 295 times – one for each pesticide – is do counties with higher exposure to this pesticide have higher rates of prostate cancer.

Now – when you do 295 statistical tests, you’re bound to get some false positives. To avoid this issue, the authors used that Phase 1 pesticide time as a discovery cohort – a place to find potential carcinogens, but understanding that some false positives would sneak in. They then checked those potential carcinogens in phase 2 – the validation cohort.  And they present to us only the pesticides that were significantly associated with prostate cancer in both cohorts.

There were twenty-two substances that fit this bill, which I’m showing you here. Lines farther to the right imply a stronger link to prostate cancer.

Obviously, we can’t go through all of these but I’ll highlight a few points.  First – there is not a really consistent theme across these chemicals. Some are pesticides, some are herbicides. Some are fungicides. Chloropicrin – towards the bottom there - is an “all purpose soil fumigant” whatever that means – and, according to the State Department – was used as a chemical warfare agent by the Russians in Ukraine.

Four of these were also associated with prostate cancer mortality, as you can see here. One of those, trifluralin, is already classed by the EPA as a possible human carcinogen. Though some other studies have not corroborated these results. The other three are considered “not likely to be carcinogenic” or to have “evidence of non-carcinogenicity”.

This highlights how difficult this type of research is. It’s clear that these substances aren’t carcinogenic the way, say, asbestos is carcinogenic, but that doesn’t mean they are completely safe. And picking up small signals of harm is just really difficult in a world full of millions of chemical exposures every day.

I would never use a study like this to call for a ban on a certain chemical. One of the limitations of any ecological study is that we have no idea whether the people who got prostate cancer in a county are the same people who were working with the pesticides. We have no blood or urine or hair samples, no biopsy tissue, no individual-level data at all. But a study like this informs the need for those deeper studies – and – if we’re being honest – provides support for funding those more difficult, individual-level studies we require.

As I implied at the beginning, there are two sides to this issue. Herbicides and pesticides have changed the world, and, honestly, for the better. Overall. But like all technology, there are benefits and harms. And if those harms can be mitigated by good science – so much the better. That’s what I call progress.

A version of this commentary first appeared on Medscape.com.