That might sound silly – women don’t even have prostates, after all – but it seemed to be what the study concluded.
But as quite often happens, it’s not quite as simple a story as it seems. Let’s take a look at what the researchers actually did, and what they concluded.
Where did the data come from?
As we’ve said before, unlike other cancer types, the things in our environment that influence prostate cancer risk are still largely unknown.
In this new study, published in BMJ Open, researchers from the University of Toronto were looking for things that might be linked to the disease.
To do so, they looked at data on prostate cancer rates in various countries from across the world, and compared this to data on oral contraceptive use in those countries to see whether they were linked.
There are a couple of issues with this approach, though. In the UK, we’re lucky enough to have exceptionally high-quality cancer statistics, which cover almost the entire population. But in many countries in the study the data collection is less accurate, or it’s confined to a small area of the country, or even just the major cities.
And as far as contraceptive use is concerned, again there’s very little in the way of reliable worldwide data, so the researchers combined data from a range of sources including surveys of contraceptive use, the number of women who were married, and estimates of the number of women of reproductive age in different countries.
So the sources of data the researchers used were, although probably the most reliable available, still not up to the standards that allow them to say anything definitive.
Comparing across countries
This is an ‘ecological study’ – where researchers compare the rates of two apparently unrelated things across countries, to see whether they are linked.
But this is a notoriously unreliable way of finding out whether two things are truly related to each other, because so many things vary between different countries that it’s impossible to say whether one thing is causing the other, or if the whole thing’s a fluke.
One of these factors becomes apparent when you look at things like PSA testing.
The USA tends to carry out more PSA testing than the UK. And higher rates of PSA testing will mean more prostate cancers are diagnosed. So any comparison between the UK and the US needs to take this into account.
And although the authors concluded that rates of Pill use were related to prostate cancer rates, the data were by no means clear in that conclusion. If you’re looking for a relationship between two things, you’d expect to see the data clustering along a line – so it might look something like this:
But what they got was a bit more messy:
So although they did show a link between the two, it’s clear that the data are a bit ‘all over the place’.
How could there be a link?
In many cases, the first indication that something may be increasing the risk of cancer is from looking for statistical links between the two. A classic example is the finding in the 1950s that smokers had higher rates of lung cancer. But that doesn’t necessarily mean that the thing in question is causing an increase in cancer rates – you need more evidence to prove your case.
To try to establish whether one thing causes another, it’s important to have evidence from several types of research. And one important line of evidence is whether there’s an evidence-based, plausible mechanism to explain how the two things are linked.
For example, we know that cigarette smoke contains over 80 chemicals that can travel around the body, damaging DNA and increase the risk of cancer. We can even see the tell-tale signs of this type of damage in the genomes of cancer cells.
The researchers of this study have suggested two potential mechanisms.
One is that prostate cancer could be caused by an infection that’s spread sexually, and in some way linked to contraception. But although there have been some studies on this, the evidence is in no way strong enough to even say that the disease could be linked to an infection, let alone whether its spread would be affected by contraception.
The other mechanism the authors proposed is that hormones in the Pill enter into the water supply and affect men.
This is superficially more plausible. There’s growing laboratory evidence that oestrogen could drive the growth of certain subtypes of prostate cancer, and could be involved in the disease’s progression in some way.
We wrote about this back in 2008, when researchers discovered that a frequently mutated gene in prostate cancer can be activated by oestrogen. But as we concluded:
What [the new research] doesn’t suggest – immediately – is that oestrogen, or oestrogen-like chemicals, cause prostate cancer. We’re well aware of the anxieties people have about ‘hormones in the environment’ … but this research looks at how existing prostate cancers grow, not how they arise in the first place.
So that’s a long way from proving that hormones in water are related to prostate cancer risk – there’s no evidence that environmental hormone levels are related to Pill use in different countries, nor that men’s risk of prostate cancer changes when they’re exposed to different levels.
And research that has been done on prostate cancer and other chemicals that could potentially mimic hormones has been contradictory, with some studies finding an effect of certain chemicals and others no link.
So although this study’s findings seem to point in the same direction as some other results, they disagree with others. And given the limitations of the study’s design, we need to be careful before we point the finger at ‘things in the environment’, and worrying people unnecessarily, when the data doesn’t support it.
So it seems that some of today’s headlines don’t represent the conclusions that can reasonably be drawn from this study. In fact, the authors themselves say in the paper, “This study is an ecological study and thus has signiﬁcant limitations with respect to causal inference. It must be considered hypothesis generating, and thought provoking.”
Margel, D, & Fleshner, NE (2011). Oral contraceptive use is associated with prostate cancer: an ecological study BMJ Open DOI:10.1136/bmjopen-2011-000311