Here’s something to think about, from João Pedro de Magalhães at Liverpool. Writing in Trends in Genetics, he notes that there are about 30 million publications in the PubMed database, and of those, about four million of them mention cancer. That list is growing at roughly 200,000 papers a year, at present. But the situation is even more striking when you get down to genes: there are 17,371 human genes that have at least one paper mentioning them in PubMed, and 15,233 (87%!) of them also have at least one paper mentioning them in the context of cancer. In the top genes with the most papers (over four thousand of them with at least 100 publications), only three of them have never been associated with cancer. If you’re wondering, those are SLC26A5 (which codes for prestin, a key protein in the hair cells of the inner ear), PRPH2 (which codes for peripherin 2, a protein in the light-sensitive cells of the retina that can be associated with retinitis pigmentosa), and CRYZ, coding for a quinone reductase that’s found in the lens of the eye. The less-published genes that are not associated with any cancer publications tend to be olfactory receptors, or bacterial genes and the like.
This leads to the title of the paper: “Every gene can (and possibly will) be associated with cancer”. You can see why that is. Cancer in one form or another is of course a huge medical problem, and getting larger every year as life expectancies increase. (Basically, the belief is that everyone who lives long enough will see some form or it or another, as you give your cells longer to pile up possible mutations and damage). And it’s comparatively easier to study than many other medical conditions, because a lot can be done at the cellular assay level without necessarily having to go to primary tissues, organs, or whole organisms. That’s not to say that you should ever look at those, of course, because conditions inside tumors and their interactions with other signaling networks in the rest of the body are very important indeed. But you can still do an awful lot of work with the (many) cancer-derived cell lines that are easily available.
And there’s another good reason, too, as de Magalhães notes: “The study of nearly any human gene can be justified (e.g., in grant applications) based on existing literature by its potential relevance to cancer. . .” All of these together make the medical literature very cancer-centric, and it’s important to keep that in mind. Given those numbers above, it’s especially important to do so when you’re talking about genetics, because I’ll bet that someone could add even those three genes above to the list with a little effort. But if everything’s associated with cancer in some way, that doesn’t do us much good, either, does it? That just turns into the background noise of molecular biology; all it can do at that point is mislead the unwary or the uninformed. The real task is to figure out which of these associations are actually important, and why. And that’s the hard part.
Those low-entry-barrier experiments mentioned above can still be useful in that work. The cells may not be the same as “real” tumor tissue, but there are crucial things wrong with them that are well worth figuring out. On the flip side, it’s widely believed (and I think with good reason) that if your oncology drug candidates can’t slow down or kill such cultured cells, or (in the usual next step) go on to do that to xenografts of tumor tissue in animals, then they are probably not going to be worth much in the real world. Both those assay systems are artificial, but they’re almost always artificial on the low side of the difficulty scale. So the number of papers is not going to be going down any time soon! But what should increase is our ability to put things in the right context.
