It would be nice if we could predict who will respond to which anti-cancer drugs.

Excerpts from an article at Popular Mechanics by Michael Rowe:

Cancer is not a single disease. In concept it is the growth of abnormal cells that spread around the body, overwhelming its systems like a weed in a garden. But what kind of weed it is and how fast it spreads are different in each case. Pancreatic cancer will kill you much faster, on average, than breast cancer. A type of lung cancer that is caused by a mutation in the gene for a protein called KRAS is far more lethal than the same type of lung cancer caused by a mutation in the gene for EGFR. What makes cancer so difficult to beat is that the mechanisms are unique to each person and, indeed, to each cancer. A drug that causes remission in one case may do nothing in another.

Because a tumor's genetic background can direct potential treatments, researchers have been keen to genetically map tumors since the technology became possible in 2000. If oncologists could map a tumor's entire genome, goes the theory, they could compare it with a reference sequence made up of normal DNA. Mismatches might explain what caused the tumor to grow. Correcting the mismatch could cure the cancer.

Rowe tells the story of one doctor named David Solit who said,

"It would be nice if we could predict who will respond to which drugs."

Back in 2010 Solit was in a meeting on the fourth floor of the Kimmel Center, a treatment space for prostate, bladder, and related cancers at Memorial Sloan Kettering Cancer Center in New York City, where he works. In a windowless conference space, some fifteen doctors, researchers, and nurses sat around a U-shaped table in white coats, discussing the hospital's clinical drug trials. The meeting was a weekly status update for a drug called everolimus, which blocks a protein called mTOR that makes cells grow and divide. Forty-five bladder cancer patients had received everolimus. Nearly all of them—forty-three—had no response at all.

The doctors considered scrapping the drug entirely. But then Solit asked a critical question: What happened to the other two? One patient had had a partial response. But the other had gone into complete remission. She had remained cancer-free for two years. "I knew from that one patient that enough of the drug got into her tumor for it to work," Solit says. "For me, that meant it was a good drug. We just didn't know who to give it to."

Read more here.