DEAN BURK, Warburg’s closest friend after the war, had failed to convince him of the importance of insulin to cancer. It was too late to convince Warburg of a new idea; it had always been too late. But Lewis Cantley, who had grown up in West Virginia taking apart and rebuilding car engines, was a different type of scientist. He had no battles to fight with other scientists, no favored theories to champion.
Cantley had found that insulin activates particular molecules in the cell and that those same molecules, when mutated, cause cancer. His science had brought him to an inescapable question: Could too much insulin signaling overactivate the PI3K pathway in much the same way as a genetic mutation might, flooding the cell with more nutrients and growth signals than it should ever have? Diabetes was an insulin story. Could cancer be as well?
It was a tantalizing idea. And if there wasn’t yet good evidence that insulin could set off the signaling cascades linked to cancer, there was already a growing body of indirect evidence that supported Burk’s original hunch about the insulin-cancer connection. By the 1980s, scientists already knew that many tumors have an unusually large number of insulin receptors and that insulin, a growth factor, causes both healthy and cancerous cells to multiply.
But it all still sounded absurd. Insulin was a metabolic hormone that governed how cells eat. Like the Warburg effect, it was assumed to be, at most, peripheral to the emergence of cancer in an organism. Even Cantley had his doubts. To find out whether insulin could really be as important to cancer as his research seemed to suggest, Cantley would have to determine what happens inside a cell once insulin arrives on the scene. He began to do so in the early 1990s, and what he found was unambiguous. While other growth factors circulating in the blood can activate PI3K—the molecule Cantley had linked to cancer and the Warburg effect—none does it more effectively than insulin. Cantley calls insulin “the champion of all PI3K activators.”1
Even as Cantley was carrying out this research, cancer epidemiologists were making striking observations of their own. In 1995, Edward Giovannucci, of Harvard, found that people diagnosed with colon cancer also tended to have high insulin. Giovannucci was soon followed by Rudolf Kaaks, who published a groundbreaking paper titled “Nutrition, Hormones, and Breast Cancer: Is Insulin the Missing Link?” Yet it wasn’t just breast and colon cancer. Each year seemed to bring evidence of yet another cancer that could be connected to insulin. The list includes cancer of the pancreas, uterus, kidney, esophagus, and prostate. People who have high insulin when they are diagnosed with cancer also tend to have worse outcomes.2
It is sometimes said that too much insulin in our blood may be driving cancer only indirectly, by making us heavier. And yet elevated insulin is a risk factor for cancer even in the non-obese. Conversely, someone who is overweight but has normal insulin does not appear to be at greater risk for cancer. Excess fat might contribute to tumors by causing inflammation and the release of additional hormones, but even in this scenario, insulin can be blamed for causing that excess fat to accumulate in the first place.3
With the start of the twenty-first century, the evidence for insulin’s role in cancer would only build. Cantley’s work on insulin and PI3K turned out to fit perfectly the findings of Craig Thompson’s lab. Thompson had found that the body eliminates unwanted cells by starving them. Insulin gives cells the opposite message. It is a growth factor that tells cells to take up nutrients. AKT, the enzyme Thompson connected to the Warburg effect, is “downstream” of PI3K, part of the same pathway, or chain of reactions. Once PI3K is activated, AKT, too, will be activated. The cell will have more food than it ever should and begin to ferment much of it via the Warburg effect.
Insulin, according to this model, isn’t responsible for the first mutations that arise. Mutations appear all the time as our cells divide and age. Even healthy tissues are often rife with mutations, including mutations linked to cancer. Some cells with these mutations will be quickly eliminated; others might contribute to tiny, incipient cancers that remain harmless. Studies of the thyroid, breast, and prostate have shown that even healthy bodies often harbor these harmless cancers. “We have had a lot of men die with prostate cancer rather than from prostate cancer,” said Derek LeRoith, a pioneer of insulin and cancer research at Mount Sinai Medical Center.4
Insulin’s role is to whisper encouragement into the ear of an incipient tumor. As Michael Pollak, head of the Division of Cancer Prevention at McGill University, explained, when cells are “insulin stimulated” and there is “a lot of PI3K and AKT signaling,” they become “pro-survival.” In this state, they are “more likely to stay alive, even if they have genetic damage.” According to Vuk Stambolic, of the Princess Margaret Cancer Centre in Toronto, even “a smidgen of a survival signal” from insulin could allow a cell an “extra division or an extra step leading to more cancer.”5
Stambolic had always been at least vaguely aware that insulin helped cancer cells thrive. Scientists growing cancer cells in the laboratory place the cells in a culture, a mixture of food and growth factors that allows the cells to eat and flourish. For many cancer types, insulin is part of the standard recipe. Stambolic, like most researchers, had never thought much about why insulin was commonly used in cell cultures. That changed in 2006, when he learned that the vast majority of breast cancers overexpress the insulin receptor. “That just blew my mind,” Stambolic said.6
Cancer cells often have more insulin receptors, Stambolic explained, because in the Darwinian landscape of a growing cancer, the cells that are best equipped to take advantage of elevated insulin in the blood are more likely to multiply. As Stambolic thought about the many insulin receptors on breast cancer cells, he became “obsessed” with a question. If cancers of the breast were so sensitive to insulin, could lowering insulin levels in the body serve as an effective treatment? If so, the benefits might extend far beyond breast cancer. Insulin receptors are also overexpressed in a wide array of cancers, including cancers of the prostate, uterus, colon, and lung.
As a first step, Stambolic returned to the recipe he had long used to grow cancer cells in his lab. Although he knew it included insulin, he did not know if the insulin was absolutely necessary for the cancer cells to live. To find out, he gradually weaned the cancer cells in his cultures off insulin, as though the cells were addicts in a rehab program.
The experiments were simple, not very different from those Paul Ehrlich had conducted a century earlier to determine if bacteria could grow in a cell culture without hemoglobin. Stambolic’s results were startling: some types of cancer could survive without the insulin, but many others could not. The vulnerable cells had plenty of nutrients and other growth factors, but, without insulin, their demise was swift. “There’s actually a dependency,” Stambolic said.7
Insulin doesn’t act alone in telling cells to eat and grow. Another central player is called insulin-like growth factor 1 (IGF-1), a closely related hormone. There are hybrid receptors on many cells that respond to both molecules. But because excess insulin leads to higher levels of IGF-1, insulin can be blamed for excessive IGF-1 signaling. Like insulin itself, IGF-1 tells cells to grow and divide and is strongly associated with cancer in population studies. (Growth hormone works, in large part, by increasing IGF-1, which in turn tells cells to eat and proliferate.)
Cancer cells often have many more receptors for IGF-1 than the cells of the surrounding tissue, just as they have more receptors for insulin, and IGF-1 signaling can also prevent cells from dying. In studies of mice, at least, raising IGF-1 levels leads to tumor growth and the Warburg effect. Lowering IGF-1, by contrast, slows tumor growth.8
Perhaps the most remarkable evidence of IGF-1’s influence on our health was uncovered in the 1990s in a remote village of Ecuador. The village is home to a community of dwarfs who are believed to descend from Sephardic Jews who were forced to convert during the Spanish Inquisition. Researchers have traced their dwarfism, known as Laron syndrome, to an inherited genetic mutation that impairs their growth hormone receptors and thus the release of IGF-1. The Laron dwarfs are not known to live particularly healthy lifestyles, and yet, according to the scientists who study them, it makes no difference when it comes to cancer. Without IGF-1 signaling, they are virtually immune to cancer.9
The research on Laron dwarfs, like much of the research on insulin and cancer, has only produced indirect evidence. But more direct evidence is now emerging. In 2019, James Johnson, of the University of British Columbia, published a paper covering five years of experiments on insulin and pancreatic cancer in mice. The results were clear. Even a little less insulin led to a “significant reduction in the development of pancreatic cancer.” Though effects found in mice often fail to materialize in people, Johnson believes that the findings “will be clinically relevant to humans because of the strong known associations between hyperinsulinemia and a number of different human cancers.”10
It is perhaps no accident that both Johnson and Stambolic are studying insulin in Canada, where the hormone was first discovered. Stambolic’s current lab is in the same building where Frederick Banting and his colleagues made their world-changing discovery of “the internal secretion,” insulin. To inspire his new students, Stambolic takes them on tours of the historic space.
That young cancer researchers, rather than diabetes researchers, now pay tribute to Banting is fitting. Banting had moved on to cancer research in the late 1920s, with the hope of doing for cancer patients what he had done for diabetes patients. For a number of years, he put all of his energies into cancer, filling notebook after notebook with ideas. “He fished and fished for the answer to cancer,” according to Banting’s biographer, Michael Bliss. “Nothing took the hook.” Banting died in a plane crash in 1941, unaware that his great diabetes breakthrough might itself have been the very cancer breakthrough he had sought.11
THERE IS STILL more evidence of the importance of insulin. In studies of mice, researchers can make tumors shrink or grow simply by enhancing or diminishing insulin signaling through genetic alterations. Then there is the metformin story. Metformin is a popular diabetes drug that helps control blood glucose levels. As it became ubiquitous in the 1990s, it wasn’t thought to have any connection to cancer. But studies of cancer trends found that metformin appeared to have a remarkable side effect. Diabetes patients taking metformin were anywhere from 25 to 40 percent less likely to get cancer than patients taking other diabetes drugs. Cantley has said that metformin “may have already saved more people from cancer deaths than any drug in history.”12
A number of different ideas have been proposed to explain how metformin might help prevent cancer. But the simplest explanation requires no special knowledge of molecular biology or biochemistry. Metformin lowers glucose levels in type 2 diabetes patients, and in so doing lowers insulin as well.
Elevated insulin appears to be especially hazardous to the epithelial cells that form a protective layer around our organs. Under normal conditions, Cantley explained, many such epithelial cells “rarely see insulin.” But when hyperinsulinemia (the condition of continuously elevated insulin) sets in, the situation is very different. Someone with hyperinsulinemia might have “50 times more insulin” than normal circulating in the blood all day and night.13
Some researchers, including Cantley, believe that elevated insulin may also be driving a process that leads to dangerous new mutations. The mechanisms involved, like so much of our current understanding of biochemistry, can be traced directly back to Warburg’s work in the 1920s and 1930s. Warburg elucidated both the first and last steps of respiration, but as the Cambridge researcher David Keilin made clear in the 1920s, electrons travel along a chain of molecules, now known as the electron transport chain, before reaching oxygen at Warburg’s respiratory ferment. (Warburg never forgave Keilin for changing the name of his “respiratory ferment” to “cytochrome c oxidase.”)14
Electron transport chains, found throughout the inner membranes of our mitochondria, are essentially tiny electrical wires containing copper and iron. (We need to eat, first and foremost, because we need electrons to supply the electricity for this system.) The wires are one of evolution’s most remarkable inventions, but they also create one of life’s great vulnerabilities: as electrons are passed along the chain, some will inevitably leak, leading to the formation of molecules that are commonly—and somewhat inaccurately—known as “free radicals” or “reactive oxygen species.” Such molecules are unstable and hungry for electrons. They will take them from whatever nearby molecules they can, including from our DNA.
Reactive oxygen species aren’t inherently bad for us. They are components of a healthy signaling system, and other molecules, the antioxidants, have evolved to sweep them up. Problems tend to arise only when the electron processing machinery malfunctions. Sometimes a damaged mitochondrion will lack the appropriate components to handle the electron flow. But other times, such as when a cell has more glucose than it should, there will be more electrons than the system can handle. More of the electrons will leak and more reactive oxygen species will form.
In the castle analogy, it is as if the guards are sending wood into the castle at all hours. To keep a chaotic situation under control, some of the desperate workers inside the castle have tossed piles of wood to the side, but the short-term solution comes at a price. The discarded wood is catching fire and everyone inside the castle is being sickened by the smoke. The excess fuel that a cell takes up and burns “actually creates a way to mutagenize itself,” Thompson said.15
The biochemist Nick Lane compares reactive oxygen species to muggers stealing a handbag. But the analogy, as Lane points out, is not perfect: the molecules that are mugged are now short an electron and become highly reactive. As Lane writes, “It is as if having your handbag snatched deranges your mind and turns you into a mugger yourself, restless until you have snatched someone else’s bag.”16
That reactive oxygen species can cause cancer is well understood. Radiation drives mutations by splitting the water molecules inside an organism. The splitting gives rise to the same reactive oxygen molecules and the same chain reaction of electron snatching. And the reactive molecules don’t have to make it all the way to the nucleus of the cell to attack our DNA. The radicals form in the mitochondria, which has its own genes that might now become damaged. There is a fire burning in the castle, and some of the most precious objects are in the very same room.
The damage to the mitochondria that ensues can have far-reaching effects, including making it harder for a cell to use oxygen—which is very much in keeping with Warburg’s vision of how cancer arises. It is a bad situation for a cell, and as it progresses, it only gets worse. When it is tightly wound, the DNA in the nucleus is hard for the electron snatchers to attack. But signals from distressed mitochondria and the excess nutrients in the cell can both lead to new gene expression, meaning the DNA will unspool and become still more vulnerable to mutations.
In this genetically unstable state, many of the mutations that arise will be irrelevant to cancer. But some of the changes will matter. KRAS is among the best-known cancer-linked genes. Once activated, it will activate PI3K and speed up glucose metabolism. As the genetic switches flip, one by one, proteins known as transcription factors will dramatically change which genes are being expressed in the nucleus and thereby reprogram a cell. In addition to driving the Warburg effect, MYC will now direct the cell to scavenge for glutamine, a source of nitrogen, and other key ingredients needed for dividing and growing. “All you have to do is make one mistake out of that more vulnerable state,” Chi Van Dang explained, “and then, all of a sudden, things start cascading from there.”17
As mutations accumulate, the cells that eat more and grow faster will triumph in the Darwinian competition. This explains why so many cancer cells end up not only with more insulin receptors on their surfaces but also with mutations in the insulin signaling pathway inside the cell. Both the receptors and the mutations make cancer cells more sensitive to insulin. Other mutations might be as likely to arise, but they are not as likely to help a cancer cell survive and flourish. The cells that are most responsive to insulin will take up more glucose and use it to build new parts, express new genes, and produce still more reactive molecules. A deadly cycle will run faster and faster.
INSULIN CANNOT EXPLAIN everything. Foodborne illnesses contributed to a rapid increase in stomach cancers in the late nineteenth century. More sophisticated diagnostic techniques and aging populations made real increases in cancer deaths appear all the more dramatic. As the twentieth century progressed, smoking, sun exposure, and cancer-associated viruses would each contribute to the growing death tolls. The interplay of genetics and bad luck will always lead to some portion of cancers.
Even among the cancers most closely tied to insulin, the story requires nuance. Insulin and IGF-1 interact with other hormones and growth factors; the PI3K pathway intersects with other signaling networks. The precise role of obesity and inflammation in the progression of cancer has yet to be fully worked out. A hundred other factors might be listed as well. The question is not whether too much insulin explains our modern epidemic of cancer, but whether it is the crucial factor we have long overlooked, the missing piece of the puzzle that best explains why prevention efforts continue to fail.
Research on insulin and cancer has long faced a significant obstacle: a siloed medical and research edifice. The people who know insulin best, the endocrinologists, are not thinking about cancer, Lewis Cantley explained. Likewise, oncologists are not typically thinking about insulin.
But then, it did not take the scientific world nearly as long to accept that other hormones and growth factors can be carcinogenic. “Everybody knows about what’s called the paradigm of hormone dependency of cancer,” said Michael Pollak, of McGill University. “And people think about that as androgens in prostate cancer and estrogens in breast cancer. You just have to extend that paradigm to insulin and many, many cancers.” Indeed, the paradigm hardly needs to be extended at all, given that high levels of insulin will also lead to increases in sex hormones—meaning that insulin can be both directly and indirectly implicated in cancers of the breast, prostate, and other organs.
For Pollak, insulin’s connection to diet is what makes it so interesting to study and so promising in the context of cancer prevention. While some cancer risk factors are largely out of our control, insulin levels are determined by what we choose to eat. And yet what makes insulin an opportunity in Pollak’s mind may be precisely what prevents other scientists from taking insulin’s role in cancer more seriously. Suggesting that insulin could be connected to hundreds of thousands of cancer deaths each year requires cancer specialists to overcome an often reflexively skeptical position on diet. That skepticism is understandable. Diet has been the realm of quackery and miracle cures since the earliest days of medicine. But that dietary approaches to cancer are commonly promoted by unscientific thinkers does not make the relationship between cancer and diet any less fundamental or important.18
Among the most important contributions of the Warburg revival is that it has changed the attitudes of many medical professionals about diet and cancer. Colin Champ, a radiologist and cancer researcher at Duke University, recounts that when he first started giving talks on cancer and diet, doctors would walk out of the room. Now the response is often the exact opposite. “Everyone is totally intrigued by the idea,” he said.
A small but rapidly growing body of research now suggests that following a very low carbohydrate (ketogenic) diet—which eliminates carbohydrates and so makes less glucose available to tumors—might make traditional therapies more effective in fighting some cancers. “It makes perfect sense that the nutrients outside of the cell will influence the tumor’s metabolism,” said Heather Christofk, a pioneering cancer metabolism researcher at UCLA who did her graduate work in Cantley’s lab. “And so it makes perfect sense that changing your diet will influence those circulating nutrients, which will, in turn, affect the tumor metabolism.”19
Siddhartha Mukherjee, the oncologist and cancer historian who did not mention the Warburg effect in his book The Emperor of All Maladies, is now investigating how a ketogenic diet (which keeps insulin levels very low) might make drugs that block insulin signaling more potent. He is also calling for more research on diet as a therapy that might be used in concert with cancer drugs. In his own lab at Columbia University, he is working with Cantley’s Weill Cornell lab to test whether drugs that block PI3K signaling might be more effective against cancer when combined with a diet that lowers glucose and insulin levels. “A careful scientific examination of diet as medicine is now long overdue in oncology, and in most fields of medicine,” Mukherjee wrote in 2018.
Mukherjee, to be sure, is not giving up on the promise of targeting specific mutations with specific chemicals. But he suggests that it is time for researchers to think more broadly. “Perhaps,” he wrote, “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.”20
The search for such “magic bullets” against cancer will continue, as it should. But no one, perhaps not even Ehrlich himself, seems to have remembered that in The Marksman, the nineteenth-century opera that inspired him, the longing for “magic bullets” was not something to be celebrated. The opera is a Faustian morality tale about the dangers of overreaching. The magic bullet is controlled by the devil. When Max fires the bullet, he nearly murders his beloved.
In the case of cancer science, the danger of seeking magic bullets isn’t so much that we can’t control which targets the bullets hit—although that is often true as well—but that the quest for magic bullets can distract us from less sophisticated, but perhaps more effective weapons. In an article in Science in 1956, Warburg wrote that the “struggle against cancer” was being undermined by the “continual discovery of miscellaneous cancer agents and cancer viruses.” The problem, Warburg maintained, wasn’t that these discoveries were false but rather that they obscured “underlying phenomena” and so stood in the way of the “necessary preventive measures.” This is the paradox of the ongoing fight against cancer. We still don’t know nearly enough about the disease, and yet the onslaught of new discoveries can sometimes be more dizzying than clarifying.
The renewed focus on diet in cancer research is not limited to diet’s impact on insulin levels. But if elevated insulin is as hazardous as many researchers now believe, we may still not know the full scope of the threat. In the early stages of type 2 diabetes, higher insulin levels will overcome the resistance of cells and successfully clear glucose from the blood—meaning that standard blood panels, which measure glucose but not insulin levels, will not indicate a problem, even as insulin resistance is setting in. This raises the question of how many might be at risk of insulin-related cancers without knowing it. A 2019 University of North Carolina study that looked at a number of markers of insulin resistance, such as elevated triglycerides in the blood, concluded that the “prevalence of metabolic health in American adults is alarmingly low, even in normal weight individuals.” In total, only 12 percent could be deemed “metabolically healthy.”21
IF MANY OF the most dangerous cancers can be linked to hyperinsulinemia, then another, still more fundamental, question inevitably follows: what causes insulin resistance and the subsequent elevation of insulin levels? Carbohydrates raise insulin more than protein or fat and so are the most obvious suspects. Low-carbohydrate diets were commonly prescribed to treat both obesity and diabetes at the turn of the twentieth century. But blaming carbohydrates for today’s epidemics of obesity, diabetes, and insulin-linked cancers leaves important questions unanswered. There are many examples of societies eating carbohydrate-rich diets without developing any of the visible signs of insulin resistance. If the answer was as simple as “carbohydrates,” Asian cultures, where rice is a central part of the diet, should have had epidemics of obesity and diabetes long before their Western counterparts. The reality is almost the opposite. Up until the end of the nineteenth century, diabetes in China was almost unheard of.
For the insulin-cancer hypothesis to make sense, it would have to explain both what changed in the Western diet in the nineteenth century and how that change led to higher levels of insulin in our blood. Once again, Lewis Cantley believed he knew the answer.