There's been a lot of work in the last four or five years using nanoparticles of gold in biological systems. When they're are brought down to this size their electronic properties get quite unusual - in gold's case, the particles become huge absorbers and scatterers of certain wavelengths of light. They're thousands of times better then the kinds of dyes that organic chemists like me can crank out, which gives you potentially huge signal-to-noise in microscopy applications and assays.
Another exotic property these things have is that they convert the energy they absorb very efficiently into heat, and it didn't take long for people to have the idea of using this effect for cancer therapy. Heating cells, after all, kills them. Of course, you'd want to have some way to get the metal particles to stick only on to cancerous cells, and this has been realized by linking antibodies to the surfaces of both hollow and solid gold nanospheres.
The latest advance in this area has come from changing shapes. Rods are predicted to absorb more efficiently and at much different wavelengths than other shapes, and a joint Georgia Tech/UCSF team (who had worked earlier on the solid gold nanospheres) has verified these effects. They were able to grow rods of various sizes and aspect ratios and to conjugate them to anti-EGFR antibodies. EGFR, of course, is a well-known cancer target (via inhibition of angiogenesis), which is hit by several small molecules as well as antibodies like Imclone's Erbitux.
Each of these types of particle has its advantages. The gold nanosphere/antibody conjugates actually absorb at slightly different wavelengths when they interact with EGFR-expressing cancerous cells compared to noncancerous ones, which could make for a useful diagnostic assay. This can really only be done ex vivo, though, in thin preparations, because the wavelengths of light needed are also absorbed by the tissue itself.
The rods don't manage to show the differences between cells, although (as with the spheres) you can see the differences (scroll down on that page) qualitatively in how many particles are bound to the cell surfaces. But they do have a property that's potentially even more useful: their absorbing wavelengths are shifted to the near-infrared, which penetrates tissue much better (up to four inches)! You need green 520nm light for the spheres (with a convenient argon laser wavelength nearby at 514nm), but the rods need red 800nm light from a titanium/sapphire laser. When cell cultures are hit with that wavelength, the heating of the gold nanorods kills them off - and the EGFR-expressing cancerous cells can be killed by laser light of only half the strength needed for normal cells.
That's probably just barely enough of a gap to be therapeutically useful, for several reasons, not least because for tumors inside the body, I think that you'd be dosing the outer skin layers with too much wattage in order to hit the deeper tissues. No doubt work is already underway on widening the window between the two effects. I can certainly imagine some possible next steps as well: simultaneous treatment with conjugates of different antibodies, for example. Since many cancerous cell lines overexpress more than one type of cell surface protein, you might be able to hit them in several ways at the same time.
As much as I love small molecules (and the organic chemistry used to make them) I have to admit: they may not be able to hold their position against ideas like this. We can try to target things like EGFR that are overexpressed in many cancers, but we don't have much of a guarantee of success, because overexpression doesn't make a pathway crucial enough by itself. But overexpression alone is all you need with this technique, and the cellular pathways downstream don't matter a bit. It's a liberating thought. . .