A study published just recently in the Proceedings of the
National Academy of Sciences describes a novel technique for
diagnosing cancer and other diseases that, if it pans out, could
conceivably diagnose hundreds of different disease types with
very high accuracy (i.e. if it says you have the disease, you
probably do) and very high sensitivity (i.e., if you have the
disease, it will probably show up on the test) from what is,
effectively, a single drop of blood.
Here's a link to the abstract:
http://www.pnas.org/content/111/30/E3072
The technique is as follows:
Artificially synthesize a large (10,000+) random sample of
possible antigens. These are short strings of amino acids
that might, possibly, bind to antibodies in the blood.
Put all of these possible antigens on a "microarray". It's a
flat surface.
Cover the microarray with a solution made of water plus the
drop of blood. The dilution used was 500 drops of water to 1
drop of blood. (There's a reason for the 500:1 that I can
explain if anyone is interested.)
Look at the pattern of what "sticks", and compare it to
patterns made from the blood of known patients of the
disease.
It turns out, according to the researchers, that if you have,
e.g. lung or prostate or breast cancer, the pattern of antibody /
antigen interactions from your drop of blood will match the
pattern of interactions from the blood of a patient with one of
those diseases.
Amazingly, the scientists don't need to know what antigens are
produced by prostate cancer or lung cancer or any of the other
cancers. They don't need to know about PSA or any other
particular chemical "biomarker" that the disease produces. All
they need to know is that the pattern of your antibody
interactions with a random sample of chemical targets matches the
pattern for patients with a particular disease.
Give one drop of blood. Send it to the lab. Get back a report
on all the diseases you have. What could be simpler or cheaper
for the patient and the health care system?
I don't know if this will work out or not. It has to be
replicated by other labs. It has to be "tuned" to get the best
results. It has to be mass produced. But it sure looks
promising.
For the record, the authors of the study are: Phillip Stafford,
Zbigniew Cichacz, Neal W. Woodbury, and Stephen Albert Johnston.