As any oncologist would tell you –carpet bombing cancer cells with chemotherapeutics is less than satisfactory. To borrow an example from warfare, it would be like indiscriminately bombing the whole country of Afghanistan with “dumb” bombs, killing thousands of civilians in the hope of getting a few terrorists. A much more effective solution would be to send a few drones, armed with “smart” bombs with an accuracy of 1 meter, constantly patrolling the skies, and ready to strike at the first sign of danger.
There ia major hurdle on the way to achieving this oncological dream: a precise target has to be identified. For that, the genome of a cancerous cell has to be sequenced, and compared with the genome of cells in a normal tissue. Until a few short years ago this was a pipe-dream. Not only would such an undertaking have cost millions of dollars, but it would have taken years to complete.
Things have changed at a dizzying rate since the beginnings of the human genome project. Faster sequencing machines and new analytical methods brought the price of sequencing a person’s genome from $9 million in July 2007 to $10,500 in July 2011, and in a couple of years from now the price is expected to drop to below $1000 per patient. How fast is this decline? Here is a point of reference. Computing costs dropped by a factor of about 4 since July ’07; DNA sequencing costs dropped by a factor of 850! Not only the machines are faster, the analytical methods for identifying the 8 billion bases that make up the genome do it at a blazing speed.
The other part of the hurdle to personalized medicine is identifying the targets, or the mutations, that caused a normal cell to become cancerous. You might think –that should be easy, just sequence a cancerous cell and compare it with a normal one. Not so fast,; there are still some formidable problems to surmount. First, cancer cells are notorious for their heterogeneity, which makes a lot of sense; cancer cells are well, cancerous because they lost their internal controls of cell division due to mutations. an unregulated, “wild” division rate allows random mutations to arise. How do you separate the important, cancer-causing mutations from the “noise” of less consequential mutations? One approach is to compare the genomes of many cells from the same patient; Also, one could compare genomes of different patients with the same disease, with the hope that the “constant”, disease-relevant mutations will become evident. Now consider that each cell contains 8 billion bases, and you’d realize that no human, or thousands of humans, can manually analyze this deluge of data.
In the last decade a new technique became available because of the new powers of DNA analysis: GWAS, which stands for Genome-Wide-Association Study. Basically, it produces a “heat map” of thousands of genes at a time, with genes that are significantly over-expressed showing up in red (by convention). So what you get is a profile of the genetic makeup of certain cancerous cells, with their mutated and over-expressed genes highlighted.
But now starts the arduous task of identifying those mutations and their respective functions, out of the 10,000 known genes in a human cell. Without getting into the mind-numbing details you can already appreciate the enormity of the task.
A relatively new field, bioinformatics, provides techniques to mine useful information out of the flood of data. This tsunami of information is so immense, that even the most advanced bioinformatics techniques cannot cope with the waves of data coming out daily out of the sequencing labs.
Is it hopeless?
This is actually a “high quality” problem to have. We used to have too few data to make any rational analysis. Now we have too much of it. But new statistical techniques of sorting out information are being developed every day. In fact, doesn’t it sound like a Google-ish problem? Of course! Google’s mission is “to organize the world’s knowledge”, and it already is marshaling its enormous statistical firepower and is wading into the massive amount of DNA sequence data.
When I trained in science and medicine, the world moved relatively slowly. A problem such as the one we are facing today would have been literally unsolvable in one person’s lifetime. But the recent onset of exponential acceleration in acquisition of scientific knowledge gives me hope that the first personalized cancer treatments can become routine in 3-5 years. I find it at once mind-boggling and exhilarating. These are indeed interesting times we are living in.