The First Cell

by Azra Raza

  In addition to acquired genetic mutations, another issue with cultured cells relates to expression of genes as messenger RNAs. The sum of all transcripts representing expression at the RNA level is called the transcriptome. When gene expression profiles of various cell lines derived from different cancers were studied, the transcriptomes of the cell lines resembled each other more than they did the cells of organs from which they were derived.

  Compounding the issues was the discovery that some of the fastest-growing cultured cells regularly find their way into adjacent plates, even under the most stringent of lab protocols. The first hint of trouble came as early as the 1970s, when chromosomal studies of cell lines derived from a variety of cancers showed that all appeared contaminated with HeLa cells, which turned out to be the Mother of All Contaminants.

  Drugs tested on these cell lines could reliably predict response in the cell lines. The in vitro test showed no predictive value when brought to the bedside. HeLa cells accurately predicted the efficacy of drugs for HeLa cells. Not humans. Despite their utility for genetic and scientific experiments, cells cultured in vitro could not be relied on for drug development.

  At that point, it might have been logical to give up the idea of in vitro modeling attempts for drug development. Instead, more artificiality was introduced into the preclinical model. Although it appeared that cell lines grown in animal models instead of plastic dishes were more comparable to cancers thriving in humans, it was not clear what the precise in vivo requirements were for hospitable and—importantly—comparable growth. The infinite complexity of a human body was neither comprehensible nor reproducible. Instead, researchers sought to hijack the body of a surrogate to grow these tumor cell lines. Enter the mouse model.

  ON THE MORNING of May 3, 1998, my husband, Harvey, having been diagnosed with cancer in March, looked over his coffee mug and handed me the New York Times. HOPE IN THE LAB, a headline shouted. A CAUTIOUS AWE GREETS DRUGS THAT ERADICATE TUMORS IN MICE. The gobsmacking opening line of the article read: “Within a year, if all goes well, the first cancer patient will be injected with two new drugs that can eradicate any type of cancer, with no obvious side effects and no drug resistance—in mice. Some cancer researchers say the drugs are the most exciting treatment that they have ever seen.” Richard D. Klausner, the director of the National Cancer Institute, was quoted calling the work “the single most exciting thing on the horizon.” Jim Watson, the Nobelist for discovering the structure of DNA, said, “Judah is going to cure cancer in two years.” Judah Folkman himself, the researcher at the heart of the story, was more cautious; as the article’s author, Gina Kolata, put it, “All he knows, Dr. Folkman said, is that ‘If you have cancer and you are a mouse, we can take good care of you.’”

  Harvey and I had lived through many cycles of frenzy in our professional life caused by laboratory triumphs of drugs followed by dashed hopes in humans. Now our relationship was more personal. Harvey expressed skepticism, yet a cancer patient’s wistful anticipation had propelled him to ask me what I thought in the first place. The basic premise of the strategy was exciting and the animal data deeply compelling. Both drugs acted by cutting off the blood supply of tumors, causing starvation, growth arrest, and eventual regression without producing any toxicity. Thanks to the New York Times report, the sensational story leaped from the confines of a research laboratory in Boston to make headlines in newspapers and television broadcasts across the nation. Cancer patients pleaded with their oncologists, desperate to get the drugs, imploring to be selected for clinical trials, ready to travel anywhere needed. The stock price of the company EntreMed, which produced the drugs, shot up fivefold in one morning, soaring from twelve dollars to eighty-five dollars. I got in touch with Dr. Folkman, who was exceptionally responsive and kind. He invited me to a daylong scientific conference in Boston where all the data along with clinical trial plans were to be presented. I registered for the meeting and came back greatly encouraged about the possibility of rapid translational success. Within a short time, word got out: however spectacularly the drugs worked in mice, they failed spectacularly in humans.

  Although mice and human lineages diverged about eighty-five million years ago, humans have been recording observations related to physiologic traits in mice since the dawn of civilization. The systematic practice to understand human ontogeny through a study of anatomy and physiology in animal models dates back to ancient Greece, and as Aristotelian methodology traveled along the ancient trade routes, animal models became the preferred research tool of Arab and later European physicians.

  Domestication of a variety of mice as pets occurred in China and Japan in the eighteenth century, eventually leading to the development and creation of modern laboratory mice. While Victorian England was busy trading in “fancy” mice, the use of animal models had become the established method to conduct biologic studies by the beginning of the twentieth century. Theories of Mendelian inheritance were investigated through mating programs in mice, and genetic mapping was well under way as early as 1915. A variety of approaches was pursued in developing mouse models for cancer research, and as is true for every model, each had its advantages and its limitations. Approximately 97 percent of human genes have homologues in the mouse genome, for example—a clear advantage versus other laboratory organisms. But the nucleotide sequences of mouse and human genomes are only about 50 percent identical.

  Many of these differences are directly owed to the dissimilarities in the environment in which the two species evolved. The major dissimilarities between mice and humans relate to factors such as the life cycle of mice. They reach sexual maturity at six to eight weeks, gestate a litter of five to eight pups in less than three weeks, and live only about three years. Mice have a metabolic rate seven times greater than humans. Since drugs in mouse models are very rapidly metabolized, the amount used in mice and humans is very different. The dosage of drugs is reduced drastically when used in clinical trials. The immune system in mice evolved to combat earthborn pathogens, whereas most of our challenges come from airborne pathogens. This stark difference in the immune systems is reflected in the cell types circulating in the blood of the two species. Humans have 70 percent neutrophils and 30 percent lymphocytes, while mice have 10 percent neutrophils and 90 percent lymphocytes in the blood. Besides these glaring differences, one of the biggest challenges in using mice as the in vivo host to human tumor cells is that, unlike a human with cancer, the target lab mice are healthy. To accept transplanted human cells without having a mouse’s body reject them as foreign bodies, the immune system of the recipient mouse has to be destroyed first. Such immunocompromised mice could hardly represent the in vivo environment of the human body in which cancer cells thrive. Yet scientists fully expected the behavior of these cells to help them identify useful drugs for patients.

  The idea of using an animal to provide the vital growth environment for tumor cells led to the birth of today’s most frequently used cell line–derived xenografts (CDX). Tissue culture cell lines were injected into mice with the intent of creating a more reliable model for cancer therapeutics. Use of animal models as preclinical platforms for cancer drug development began in earnest with the mouse-in-mouse grafted tumors during the 1960s. Such models produced by transplanting a given mouse tumor yielded early successes in that several cytotoxic chemotherapies like procarbazine and vincristine were identified and proved useful in the treatment of a host of cancers. That does not say much for the efficiency of the CDX model per se because cytotoxic drugs kill cells indiscriminately, be they normal or cancerous. This is why they are so toxic when administered to patients. The same results would likely be seen in less elaborately constructed, cheaper cell culture systems. Nevertheless, CDX became the model of choice for all kinds of drug development. Responses to cytotoxic drugs ranged between 25 and 70 percent among different cancer types. The NCI invested generously in producing between six and nine cell lines each, derived from a number of common tumor types, hoping that this would
cover the variability seen in efficacy. This led to the creation of the NCI-60 panel, comprising sixty cell lines derived from nine types of cancers, which was then handed over to investigators for the development of CDX models.

  They failed uniformly as far as drug development was concerned.

  In reality, such models for drug development represent an irresponsible and serious waste of shrinking research resources, and not just in oncology. Sepsis, burns, and trauma in animals were all investigated as models for the inflammatory changes associated with those phenomena in humans. There was no correlation. Indeed, every one of the 150 treatments for sepsis brought to the bedside of acutely ill humans because of their success in treating mice was a staggering catastrophe.

  Humans do not benefit but are harmed by misleading animal testing, especially when it comes to predicting the efficacy of targeted therapies. These are drugs developed to attack individual and specific cancer-driving proteins. The targeted therapies identified through CDX models have an abysmal success rate of 5–7 percent when brought into clinical practice. This includes the agents developed to target genetic mutations such as BRAF, EGFR, HER2, and a few others. When occasional drugs appear to work in both humans and the in vitro models, it is not because of similarities in the biology of the diseases but because the drugs happen to be general cytotoxic agents. Timothy Johnson, a physician, told the Boston Globe during the height of the enthusiasm for Folkman’s work that “my own medical perspective is that animal cancer research should be regarded as the scientific equivalent of gossip—with about the same chance of turning out to be true, i.e. truly effective in humans. Some gossip turns out to be true, but most of it does not… and gossip can cause great anguish for those affected, in this case millions of desperate cancer patients worldwide.” He was right.

  As various in vitro and CDX efforts failed, focus then turned to improving quality of the cancerous seed rather than the soil in which it was planted. Instead of using cultured cell lines as starting points for creating a preclinical in vivo CDX model, freshly obtained human tumors were implanted in animals, at times, even matching organ to organ; cancer cells from human pancreas implanted in mouse pancreas. These patient-derived xenografts (PDX) models could serve as “avatars” for individual patients to test a variety of drugs against their tumor cells directly as they grew in vivo in a mouse. Once again, the NCI invested large sums of money in producing and handing out one hundred PDXs to investigators for research.

  Unfortunately, the technique didn’t always work. In one instance, a laboratory company pursuing this research was able to culture tumors for only half of the 1,163 people who sought their help. The researchers ultimately found only 92 patients who received treatments based on testing in the PDX models, although they did find that the PDX predictions were accurate 87 percent of the time. How practical this approach would be is questionable since it can take six weeks or more for the tumor to grow in the mouse and be ready for appropriate testing against a series of drugs.

  But, the above notwithstanding, there are strong signs that PDX is, generally, not going to be predictive, again owing to adaptations for the implanted tumor to its new environment. To study how the genome of the tumor changed through multiple rounds of transplantation in mice, more than one thousand PDXs representing twenty-four types of cancers were studied. Implanted tumors evolved differently from their parent cells. While glioblastomas gain extra copies of chromosome 7 in humans, the PDX model of the tumor lost them over time. The National Cancer Institute tested twelve anticancer drugs—that were already being successfully used to treat humans—on PDX mice growing forty-eight different kinds of human cancers. In 63 percent of cases, the drugs failed. Even worse, according to a report in Nature on the study, researchers at the NCI concluded that other compounds that might work in humans were never tested on the erroneous belief that if they couldn’t help PDX mice, they couldn’t help humans either. But from my perspective, even if the models worked as well as we had hoped they would, the fundamental problem would still remain—very few effective anticancer treatments exist, so the predictions made through these models are more likely to be useful in what to avoid rather than what to give the patient. I cannot stress this enough times; scientists need to stop making more and more artificial mouse models and tissue culture cell lines for cancer drug development. These resources can and should be invested in better pursuits.

  No one, however, willingly surrenders their pet projects, no matter how far they have drifted from the original intent, as long as they can maintain their grip on grants and power. A repetitive triangular pattern characterizes the scientific culture, similar to the kyklos, the recurrent cycles of government, described by the Greeks, of democracy, aristocracy, and monarchy along with their degenerate forms—ochlocracy, oligarchy, and tyranny. What begins as a perfectly sensible democratic state of affairs transmutes into an oligarchy when a small group of privileged individuals exercising control over institutions and organizations handing out perquisites succeed in dominating the field. The democratic-to-oligarchic shift gives rise over time to a “hereditary aristocracy” in which newly minted key opinion leaders, with the blessing of their scientific mentors, inherit the exclusive power to define rules, monopolize grant-funding powers, and reward each other with perks the field has to offer. Adding a final insult to injury, these little arrogant cliques manage to hijack the entire narrative in a field.

  I met a young male researcher recently whose ego was so dense, light would bend around him. He presented a seminar at Columbia University, where he described mouse models carrying a mutated gene associated with MDS. He also presented data that administering a drug that inhibited activity of the protein, not the mutated protein, was curing whatever disease he had inflicted upon the mice (it certainly was not anything even close to human MDS). When I asked him what gave him any confidence that the results of the drug therapy he showed in mice would have value for humans, he scoffed, “Sorry, Azra, mouse models are not going away.” That was already two years ago. I am sure he has cured a lot more mice since then. I am also sure he has received grants to continue this work. His coresearcher at the same institution started a recent lecture with a slide comparing survival curves of AML patients for each decade from 1970 to the present. The graph showed essentially zero improvement. He then used words I have been hearing for forty years describing how he was going to understand the intricate molecular mechanisms inside AML cells and then devise ways not to kill them but to modify their behavior so they no longer remained malignant. This is precisely the problem. It is as if the past forty years have not existed. Freshly minted brilliant young scientists arriving in waves, confidently proclaiming their plans to convert cancer into a chronic disease that patients can live with and not die from. On what basis? Indeed, there are unimagined novel technologies now that did not exist a few decades ago, but the complexity of cancer remains beyond their reach. To think otherwise is unrealistic and a victory of hope over experience.

  Clinical researchers are busy trying to open new experimental trials, and basic researchers are worrying over the next grant they need to write. The only way to unmask the magnitude of bizarreness is to find a new and improved way of doing things, a way not just marginally better but quantum leaps better. This is precisely what oncology needs right now. If we’d kept trying to improve upon the typewriter, we would never have invented the word processor. Toying with or repairing old models of treating cancer will yield incremental advances at best. The cancer problem requires a radically different approach. We should not be aiming for weeks of improved survival. Our goals should be higher. The public needs to see how far we have drifted from the original aims as oncologists and researchers and at what cost to the patient.

  Everyone needs to pause and think about what they are doing and why. Young researchers and all oncologists must think differently, to question dogma, to reject the deep-rooted archaic traditions, to discard the existing, inadequate research models and boldly use th
e emerging technologies to explore exciting new strategies to solve the cancer problem. Only a new way of thinking and doing will shift the paradigm and get the practitioners to discard their old ways. All researchers need to pay attention to technologic aspects emerging within and outside of their own disciplines, developing a broader strategy to address the complexity of cancer by using inclusive, pluralistic approaches rather than relying solely on reductionist strategies. Young researchers need to practice consilience, learning from and cooperating with experts in disparate fields to solve the biologic and technologic hurdles. The traditional strategy of treating cancer reached its maximum potential several decades ago. Dying for a Cure, a British advocacy group, bemoans that “at the current rate of progress it would take 1,778 years at least before we saw a 20-year survival improvement for all 200 types of cancer!”