On my way to a meeting on cancer and personalized medicine a few weeks ago, I found myself thinking, improbably, of the Saul Steinberg New Yorker cover illustration “View From Ninth Avenue.” Steinberg’s drawing (yes, you’ve seen it — in undergraduate dorm rooms, in subway ads) depicts a mental map of the world viewed through the eyes of a typical New Yorker. We’re somewhere on Ninth Avenue, looking out toward the water. Tenth Avenue looms large, thrumming with pedestrians and traffic. The Hudson is a band of gray-blue. But the rest of the world is gone — irrelevant, inconsequential, specks of sesame falling off a bagel. Kansas City, Chicago, Las Vegas and Los Angeles are blips on the horizon. There’s a strip of water denoting the Pacific Ocean, and faraway blobs of rising land: Japan, China, Russia. The whole thing is a wry joke on self-obsession and navel gazing: A New Yorker’s world begins and ends in New York.
In the mid-2000s, it felt to me, at times, as if cancer medicine were viewing the world from its own Ninth Avenue. Our collective vision was dominated by genomics — by the newfound capacity to sequence the genomes of cells (a “genome” refers to the complete set of genetic material present in an organism or a cell). Cancer, of course, is typically a disease caused by mutant genes that drive abnormal cellular growth (other features of cellular physiology, like the cell’s metabolism and survival, are also affected). By identifying the mutant genes in cancer cells, the logic ran, we would devise new ways of killing the cells. And because the exact set of mutations was unique to an individual patient — one woman’s breast cancer might have mutations in 12 genes, while another breast cancer might have mutations in a different set of 16 — we would “personalize” cancer medicine to that patient, thereby vastly increasing the effectiveness of therapy.
This kind of thinking had an exhilarating track record. In the 2000s, a medicine called Herceptin was shown to be effective for women with breast cancer, but only if the cancer cells carried a genetic aberration in a gene called HER-2. Another drug, Gleevec, worked only if the tumor cells had a mutant gene called BCR-ABL, or a mutation in a gene called c-kit. In many of our genome-obsessed minds, the problem of cancer had become reduced to a rather simple, scalable algorithm: find the mutations in a patient, and match those mutations with a medicine. All the other variables — the cellular environment within which the cancer cell was inescapably lodged, the metabolic and hormonal milieu that surrounded the cancer or, for that matter, the human body that was wrapped around it — might as well have been irrelevant blobs receding in the distance: Japan, China, Russia.
To bring the promise of mutation-directed therapies to life, researchers began two kinds of trials. The first was called a “basket trial,” in which different forms of cancer (e.g., lung, breast and stomach) containing the same mutations were treated with the same drug — in essence, lumping genetically similar cancers into the same “basket.” The obverse of the basket trial was an “umbrella trial.” Here, one kind of cancer — say, lung cancer or melanoma — was divided into different subtypes based on genetic mutations, and each subtype was targeted by a different medicine. Under a seemingly common umbrella — lung cancer, say — genetically distinct tumors would be treated with therapeutically distinct drugs.
Basket trials worked — somewhat. In one landmark study published in 2015, 122 patients with several different types of cancer — lung, colon, thyroid — were found to have the same mutation in common, and thus treated with the same drug, vemurafenib. The drug worked in some cancers — there was a 42 percent response rate in lung cancer — but not at all in others: Colon cancers had a 0 percent response rate. More recent basket trials with newer drugs have demonstrated striking, even durable, response rates, although the mutations targeted by the drugs are relatively rare across all human cancers.
And the umbrella trials? The record here was also mixed — and, to some, disappointing. In the so-called BATTLE-2 study, patients with lung cancer were divided into different groups based on gene sequencing, and each group was treated with four different drug combinations. The hope was that patients with tumors that contained a mutant version of a gene called K-ras would be uniquely susceptible to one particular drug combination (preclinical data, gathered in mice, suggested that this combination would be potent in these patients). But the laborious strategy deployed in this study — biopsying the tumor, sequencing it and then dividing the patients into mutation-guided treatments — provided few novel therapeutic inroads. In general, patients carrying mutations in the K-ras gene, a key driver of cancer growth, did not survive longer when given the combined drug therapy. “Ultimately,” one reviewer commented, “the trial failed to identify any new promising treatments.” Sequencing, it seemed, had made us none the wiser about treatment.
The disappointments of these early studies fueled public criticisms of precision medicine. Perhaps we had been seduced by the technology of gene sequencing — by the sheer wizardry of being able to look at a cancer’s genetic core and the irresistible desire to pierce that core with targeted drugs. “We biomedical scientists are addicted to data, like alcoholics are addicted to cheap booze,” Michael Yaffe, a cancer biologist from M.I.T., wrote in the journal Science Signaling. “As in the old joke about the drunk looking under the lamppost for his lost wallet, biomedical scientists tend to look under the sequencing lamppost where the ‘light is brightest’ — that is, where the most data can be obtained as quickly as possible. Like data junkies, we continue to look to genome sequencing when the really clinically useful information may lie someplace else.”
It’s that vision of “someplace else” — a view of the world beyond Ninth Avenue — that oncologists and patients are now seeking. Mutations within a cancer cell certainly carry information about its physiology — its propensity for growth, its vulnerabilities, its potential to cause lethal disease — but there’s a world of information beyond mutations. To grow and flourish within its human host, the cancer cell must co-opt dozens, or even thousands, of nonmutant genes to its purpose — turning these genes “on” and “off,” like a pathological commander who has hijacked a ship and is now using all its normal gears and levers to take a new, malignant course. And the cell must live in a particular context within its host — dodging the immune system, colonizing some tissues and not others, metastasizing to very particular sites: bones but not kidneys for some cancers; liver but not the adjacent spleen for others. What if the “really clinically useful information” lies within these domains — in the networks of normal genes co-opted by cancer cells, in the mechanisms by which they engage with their host’s immune system or in the metabolic inputs that a cell needs to integrate in order to grow?
At the annual meeting of the American Society of Clinical Oncology (ASCO) in Chicago this year, it was this altered — and more expansive — vision of precision cancer medicine that was on display. Perhaps the most significant among the presented studies was a very large clinical trial that identified breast cancers that were unlikely to benefit from chemotherapy based on information carried by patterns of gene expression — not single gene mutations — in cancer cells. By identifying tumors that carry these “safer” genetic fingerprints, the study hopes to reduce the use of toxic, expensive — and ineffective — chemo for tens of thousands of women every year. This, too, is precision medicine: Our capacity to find women who should not be lumped into the basket of standard chemotherapy must rank among one of the most worthwhile goals of personalized cancer therapy. Other teams at ASCO reported responses to new generations of drugs that enable the immune system to attack certain cancers, beginning an intensive search for biological markers on cancer cells that predict which tumors are likely to respond (hint: It may not be a single gene mutation).
The point is that precision medicine is not just precision mutant-hunting. It may be decidedly low-tech and may apply to conditions other than cancer. In orthopedics, precision medicine might involve finding an anatomical variant in some shoulders that have sustained fractures, say, that predicts that conventional shoulder surgery will not succeed for those patients. It might invoke gene sequencing again — but this time with computational algorithms that use combinations of genes to predict outcomes (Does A plus B without C predict a response to a drug?). Or it might skip gene sequencing altogether: In my own laboratory, a postdoctoral researcher is trying to grow cancer cells from individual patients in the form of tiny “organoids” — three-dimensional cellular structures that recapitulate living tumors — and testing thousands of drugs to find ones that might work in these organoids before deploying them on patients.
These strategies must, of course, be tested in randomized clinical trials to see if they provide benefit. Can they be deployed at reasonable costs? Will the benefits have an impact on a public scale? But the reinvention of cancer therapy needs time, patience and diligence — and, yes, skepticism. By narrowing our definition of precision medicine too much, we almost narrowed our ambition to deliver precise, thoughtful therapy — or, at times, no therapy — to our patients. It would be a shame to view cancer through such narrow lenses again.
Siddhartha Mukherjee is the author of “The Emperor of All Maladies: A Biography of Cancer” and, more recently, “The Gene: An Intimate History.”