The Jeremiah Metzger Lecture Cancer in the Twenty-First Century: An Inside View from an Outsider

Abstract

Cancer is not a single disease. The term refers to literally hundreds of illnesses sharing common features: inappropriate proliferation of imperfectly differentiated cell types, invasion of nearby vital structures, and spread to distant sites (metastasis). Invasiveness and metastasis distinguish cancers from benign tumors such as fibroids and meningiomas. Yet, each type is distinct, possessed of defining morphologic, histologic, biochemical, and genomic features that have allowed oncologists to develop a nosology that guides diagnosis and therapy. Cancer is thus a complex collection of disorders. That complexity is increasing exponentially as modern technologies allow us to dissect each form in ever greater detail. The notion of curing cancer with a “magic bullet” like the polio vaccine is no more realistic than using the same wrench to fix a bicycle, a car, and an airliner just because they are all vehicles.

Cancer is the second leading cause of death in the developed world, causing approximately 600,000 deaths annually in the United States alone; the global toll is approximately 10,000,000. This scourge persists despite massive “wars” on cancer that have accomplished only a small reduction in mortality. Yet, those of us working in this field are more optimistic than ever that we have finally turned a corner, that progress is tangible in the form of better patient outcomes, and that we now grasp cancer pathobiology sufficiently to accelerate that progress significantly. We are also chastened by a deepened recognition of the astonishing biological complexity of cancer, and the costs of new agents that are beginning to attain better outcomes.

My goal is to share this admixture of excitement and apprehension by framing my commentary in four parts. First, we shall review the basics of cancer — “Cancer 101” — what cancers are, how they originate, grow, invade and spread, and the properties that make eradicating them safely such a daunting challenge. Second, we will consider recent advances that place us at the threshold of “Cancer Precision Medicine.” This paradigm has the potential to match highly customized therapies to the specific tumors present in each individual patient by exploiting molecular profiling capabilities. For a few patients, this strategy already guides the clinician to administer “targeted” drugs attacking the precise genomic and immunological vulnerabilities of their tumors. Third, I will consider the realities, biological, clinical, and economic, attendant upon broad deployment of cancer precision medicine. Finally, I will offer some thoughts about how we might achieve widespread deployment in a sustainable manner.

CANCER 101: WHAT IS THE ROOT CAUSE OF CANCERS AND WHY ARE THEY SO DIFFICULT TO CURE?

Cancer deaths are declining despite the aging of the population. Prevention (e.g., tobacco control) and early detection (e.g., colonoscopy) account for much of this decline. While better treatment of established cancers has also contributed, curative therapies are available for only a few forms of cancer (e.g., childhood acute lymphoblastic leukemia). For most other forms, surgical extirpation of lesions detected before they have spread remains the best hope of a cure. Failing that, the best available therapies cure (unpredictably) a few patients, provide moderate prolongations of survival for some, and, for all too many, offer only short-term or no benefits.

Progress has been steady but slow, despite billions of dollars invested in research, thousands of clinical trials, and millions of years of work by extraordinarily talented and dedicated scientists and clinicians. Too many patients are still diagnosed and treated by strategies that were established decades ago. Their tumors are characterized by anatomic site (e.g., breast cancer, lung cancer) and descriptions of microscopic morphology (carcinoma, sarcoma, etc.). Their treatment still consists of some combination of surgery, radiation therapy, and cytotoxic chemotherapy, often with drugs that are derivatives of, or the exact same, drugs developed decades ago. These agents, most aimed bluntly at the normal processes of DNA metabolism and cell division, are highly toxic and modestly effective. Too many patients will thus endure months of treatment, major side effects, and huge expense, all in the hope that they will be among the slim majority who will be cured or at least granted significantly extended life. For some forms of cancer, this approach remains the best available even at advanced academic cancer centers.

The general public, the funders of cancer research, and policy makers are increasingly restive about the limited results of the “Wars on Cancer.” It is important to understand the root cause for slow progress, and to articulate it clearly, lest support diminish just as marvelous new tools to solve these obdurate problems are becoming widely available.

The root cause of the modest return on decades of effort lies in the root cause of human cancers themselves: enormously complex molecular alterations of critical growth controlling genes within each cancer cell. These alterations (mutations) are also dynamic. Cancers are constantly evolving, creating diverse, often drug-resistant, sub-populations of cells. This “shape shifting” heterogeneity and adaptability defeats therapies and constitutes the root cause of slow progress. Cancer will be conquered only if we address each patient’s array of sub-populations in an individualized way. This will require far better insights into cancer pathobiology and powerful technological capabilities that are just now emerging.

All known cancers arise from mutations that pervert the behaviors of certain key genes and their RNA and protein products (Figure 1). In their normal state, these genes work together to ensure a harmonious and regulated balance of the proliferation, survival, and timely death of our cells. Disruption of this harmony causes cells to proliferate when they should not, forming the masses that we call tumors. A major insight has been the recognition that all of the “causes” of cancer such as tobacco exposure, ionizing radiation, ultraviolet radiation from sunlight, environmental carcinogens, etc., have in common the propensity to increase the rate of these mutations (Figure 1). In the nucleus of every cancer cell resides a complex array of these mutated, malfunctioning genes. They cause cells to divide when they shouldn’t, to invade normal tissues, and to metastasize.

Fig. 1.

Mutation of key growth controlling genes is the root cause of cancer. Etiological causes of cancer are all agents promoting DNA alterations (mutation). All of the internal and environmental agents shown in the left side of the figure damage DNA by chemical alterations (carcinogens in tobacco smoke, in environmental toxins, dietary carcinogens, etc.), by physical or oxidative damage (sunlight due to ultraviolet exposure, ionizing radiation), or by accumulation of replication errors inevitably occurring with each cell division (aging). Microbial infection can lead to cancer, disrupting genome integrity via insertion in the host genome, by causing chronic inflammation and its DNA damaging byproducts, and by immunosuppression of the host’s ability to eliminate mutant cells. For the most part, these mutations are acquired in somatic cells (somatic mutations). One can inherit genes, e.g., BRCA, Rb, Wilms, VHL, etc., that increase the propensity to cancer (germ line mutations) because these cells are already one step closer to malignant transformation. The final common pathway in every case is mutations in key genes responsible for governing the growth and/or survival of cells and the maintenance of the integrity of the genome.

Roughly 1,000 of the 25,000 to 30,000 genes in our genomes can, when mutated, lead to cancer. We can regard them as comprising four broad categories:

1. Proto-oncogenes code for proteins that promote cell proliferation (growth factor receptors, “signal transduction” proteins that convey growth signals from the receptor to the nucleus of the cell, or the transcription factors that are activated or repressed by signaling pathways, and thereby stimulate or subdue the expression of genes governing cell division. Oncogenes are mutationally activated forms of these genes that accelerate cell division.

2. Tumor suppressor genes code for proteins that shut down proliferation. They “put the brakes” on cell growth. In most cancers, they tend to be deleted or inactivated.

3. DNA repair and proofreading genes — During every cell division, 6 billion base pairs of DNA are replicated with an accuracy of approximately 99.9999% (one copying error per million); each daughter cell thus bears thousands of mutations. Fortunately, DNA proofreading and repair systems eliminate most of these. Most cancerous cells have acquired mutations that impair or derange these corrective mechanisms. In an established cancer, these mutations beget more mutations.

4. “Pro -“or “anti-apoptotic” genes — apoptosis (“programmed cell death”) is the process by which cells auto-suicide in an orderly manner if badly damaged, worn out, or no longer needed. DNA damage is a potent trigger; apoptosis is thus a potent anti-cancer defense. Mutations inactivating pro-apoptotic genes, or activating anti apoptotic genes, allow cancers to survive when they should not.

Fortunately, no single mutation is sufficient to make a cancer cell. Modern DNA sequencing methods have shown that a typical human cancer cell contains dozens of mutations. Cancer cells transform to their malignant behavior when they accumulate enough growth accelerating mutations and lose enough tumor suppressor functions to tip the balance in favor of inappropriate proliferation, invasion, and metastasis (Figure 2). These mutations accumulate over time, but only in those cells able to evade apoptosis and immune surveillance. This explains why cancers develop slowly, often years after exposure to potent carcinogens such as cigarette smoke. Most of these mutations are also somatic, that is, acquired after birth, as opposed to being inherited. Indeed, even inherited or congenitally acquired mutations such as BRCA, RB, etc., increase the propensity to get cancer. However, they do not by themselves cause cancer in everyone inheriting them.

Fig. 2.

Cancers develop and evolve over extended periods of time, during which mutations accumulate until growth control is deranged and invasiveness and metastatic potential develop. (Figure adapted from National Cancer Institute Slide Set.)

Cancers are difficult to treat because, in most cases, their survival and behavior are driven by so many misbehaving genes. Efforts to identify a single molecular target and kill the tumor cell by attacking that target have had limited success. Many cancers suppress apoptosis, the “suicide” signal, so that even cells heavily damaged by our drugs may refuse to die. They are also especially difficult to eradicate because deranged DNA proofreading makes them genetically dynamic. Each cancer cell diverges from the others as it accumulates its own unique array of mutations, subverting the fundamental control system that regulates the appropriate growth and persistence of our cells. This causes both the malignant behavior of the cancer, and a “mutator” phenotype that renders most cancers highly heterogeneous and adaptable with respect to primary or acquired resistance to therapy.

Even when it presents at an early stage, a human cancer has been evolving for years, adapting so as to thrive in our bodies. The tumor that survives until it is large enough to be detected is one that has tricked the body into accepting it, rather than rejecting it. It does so by suppressing our immune systems and by co-opting cells in its immediate micro-environment to produce substances that actually sustain and nurture it, for example, by forming blood vessels to supply it with oxygen and nutrients. It has acquired these capabilities by progressive accumulation of mutations that accomplish these nefarious objectives and support subpopulations of cells whose mutations render them resistant to therapy. Once treatment is complete, these sub-clones merely grow back out again. This is one reason why some drugs can dramatically shrink tumors but do not prolong survival.

Given these realities, it might be hard for those outside our field to understand why optimism is so high. Fortunately, a new approach has emerged during the past 20 years that allows precise targeting of cancer’s root cause — mutations. Novel agents such as trastuzumab for a specific subtype of breast cancer (HER2-positive), imatinib for chronic myelogenous leukemia, and rituximab for certain B-cell lymphomas achieve better long-term remission and even some cures for patients with these subtypes of cancer because they all work by targeting molecular derangements that arise from mutations characteristic of each type. These isolated successes have encouraged the development of a promising new paradigm based on “smart bomb” drugs targeting precisely the molecular misbehaviors specific to the cancer occurring in individual patients. Applying this “Cancer Precision Medicine” strategy to as many patients as possible has become the overarching goal of the field.

CANCER RESEARCH AND CARE IN 2017: TARGETED THERAPIES, IMMUNOTHERAPY, AND PRECISION CANCER MEDICINE

The first molecularly targeted therapies arose from decades of painstaking research, using now obsolete methods of molecular biology to pinpoint and attack genetic vulnerabilities in individual forms of cancer. Those methods were only robust enough to discern meaningful targets in few forms of cancer. However, the clinical success resulting from application of that approach in those special cases constituted a proof of principle that inspired the belief that, with more robust tools, every cancer’s “driver mutations” could be identified and interdicted. The Human Genome Project generated those tools. Comparative analysis of a patient’s normal genome with that of her/his tumor genome(s) is now widely available. We can interrogate each tumor at the core of its malevolence, its mutated genome, to a precision of a single nucleotide code letter.

Adroit application of these tools has fostered delineation of many of the key mechanisms and elements responsible for the malignant behaviors of cancers. For an increasing minority of patients, genomic analysis is already inspiring the development truly novel drugs. These targeted agents attack specifically the mutated gene products driving that patient’s tumor. The best example is Gleevec (Novartis, Inc., Basel, Switzerland) (imatinib), which shuts down the inappropriate activation of the bcr-abl fusion protein that drives the proliferation of chronic myelogenous leukemia cells, but many other targeted therapies have entered clinical practice. Examples include tumors with activating mutations of the epidermal growth factor receptor in non–small cell lung cancer, of the b-raf kinase in melanoma, and of the “Bruton’s kinase” in chronic lymphocytic leukemia (Figure 3). These new drugs have proven to be more potent and less toxic than the generalized “carpet bombing” poisons that dominated conventional regimens.

Fig. 3.

The paradigm of precision medicine. As indicated, three patients with seemingly identical cancers by standard diagnostic criteria who actually have distinct tumors based on genomic profiling of the mutations in each tumor would be treated differently and more precisely with agents targeting the mutations in each of their individual cancers. (Figure provided by Professor Barrett Rollins, Dana Farber Cancer Institute.)

More than a decade of experience has confirmed the potential of this new paradigm and also revealed two major limitations (Table 1). First, most cancers treated with the appropriate targeted agent respond briskly with dramatic reductions in tumor burden, but durable responses that greatly prolong life are uncommon. In many patients, resistance develops, sometimes rapidly. Second, the percentage of patients for whom targetable driver mutations can be identified is increasing but remains low. Nonetheless, there is good reason to believe that the field has “turned the corner” because we now have a much firmer understanding of what we are up against. We have also learned that, while these basic mechanisms are widely shared across human cancers, the varied combinations in which they exist in individual tumors make the cancers in each patient distinctive and unique. There is reason to believe that customized combinations of drugs targeting multiple mutations could improve outcomes and broaden the base of treatable patients. This potential drives efforts to develop better means to deliver personalized cancer medicine: the right drugs in the right doses at the right time for the right tumor in the right patient, more widely (Figure 3).

TABLE 1.

Targeted Therapies Report Card

• Responses dramatic but usually short lived – many resistance mechanisms identified

• Target mutations can be matched to precision drug in only small, but growing, percentages of patients

• Long term survival improved at best in a subset of the subset of patients

• Experience has revealed that tumors behave in patients in ways far more complicated than anticipated from animals and in vitro models

• Using genomics to guide therapy is complicated and expensive because massive amounts of data must be accumulated and interpreted.

Cancer genomics has revealed that there are many more forms of cancer than is apparent by traditional classifications based on morphology and organ location. We recognize that there are more than a dozen forms of breast cancer and more than 50 forms of leukemia and lymphoma. Within each of these subtypes, there are individualized patterns of mutations (gene signatures). These often cross the boundaries of traditional classification schemes. For example, from the perspective of treatment, chronic myelogenous leukemia (CML) is more like gastrointestinal stromal tumor (GIST) sarcoma than it is like other forms of leukemia because CML and GIST share very similar molecular targets that make both susceptible to dramatic responses from the same drug (Figure 4).

Fig. 4.

Some examples of drugs targeting specific “driver” mutations found in some forms of cancer. The type of cancer is shown on the left, the mutate gene target(s) in the middle, and the drug(s) targeting them, on the right. The body figure on the bottom is meant to indicate melanoma skin cancer. (Figure provided by Professor Barrett Rollins, Dana Farber Cancer Institute.)

Molecular dissection of the tumor micro-environment has also made it clear that successful cancer treatment will require more than a simple attack on the cancer cell. That attack will still be vital; however, it must be supplemented by strategies that address the surrounding tissues that nurture the tumor and protect it from immune surveillance (Figure 5). Scientists have begun to pinpoint the mechanisms and molecules by which tumors pervert these normal physiologic processes to serve their own ends. These have revealed a new array of possibilities for drug development. Most promising at the moment has been the development of agents that block the immunosuppressive molecules that tumors express (e.g., CTLA-4, PD.1, PDL.1). These “checkpoint inhibitors” provoke robust immunological reactions that are yielding prolonged responses in significant numbers of patients. Their efficacy is confined to the minority of tumors expressing these targets, and autoimmune toxicities limit their use. Nonetheless, this a very promising approach that is substantially improving outcomes in patients with melanoma and lung cancer.

Fig. 5.

Molecular targeting of the immune response. The figure shows only a subset of the many molecules that act to stimulate or inhibit the immunological defense against tumors. A number of these are subverted by cancers in a way that protect them from immune attack. The cancer precision medicine paradigm applies to immunotherapy as it does to targeting tumor mutations, except that the goal here is to promote a tumoricidal immune response rather than direct stasis or killing of the tumor cells. To date, drugs targeting PD-1, PD-L1, and CTLA-4 are the only ones in clinical use, but this approach is in its infancy. (Figure provided by Professor Stephen Hodi, Dana Farber Cancer Institute.)

Even though cancer outcomes are improving slowly, the pace of progress in cancer science is more rapid and exciting than ever. The next decade holds great promise for making major inroads against the human suffering caused by cancer. Knowledge gained in just the past few years has yielded more insights about the aberrant behavior of cancer cells than has been gathered in the entire previous history of cancer research. Technological advances allow us to analyze human tumors in ways unimaginable just a decade ago. It is now possible to identify every mutation in each patient’s tumor, and to ascertain their effects on the behaviors of their gene products. These breathtaking new capabilities can be coupled with clinical research in real time to devise more precise and effective therapies for individual patients. Genomics, information science, cancer pathobiology, and clinical research are coalescing around the care of the individual at an accelerating pace. For a small but growing percentage of patients, measurements available only in the most cutting-edge basic science laboratories just a few years ago are now guiding clinical management decisions. For them, one can now devise care plans that are far more effective, less toxic, and better individualized to their needs.

PERSONALIZED CANCER MEDICINE: OPPORTUNITIES AND CHALLENGES

The emerging paradigm that is variously described as personalized cancer medicine, high-precision cancer medicine, or individualized cancer medicine offers the best hope for making major improvements in cancer survival (Figure 3). Yet, nearly 15 years after the first proof of principle applications, its clinical utility is still confined to a few special circumstances. The excitement of the first successful uses of cancer genomics has been tempered by hard experience that has taught us to respect even more the enormous complexities of human cancers as they occur in each patient, and the logistical and economic challenges attendant on broad application of this model.

Progress has been impeded primarily by our incomplete mastery of the pathobiology of cancers. Our ability to overcome the root causes of cancer’s lethality remains woefully limited. We can identify the many mutated genes in cancer cells, but major gaps remain in our basic understanding of exactly how they interact to create, drive, and sustain their miscreant behaviors. Much remains to be learned about their ability to conscript other normal cell types to form the cancerous tissues (tumors) that nourish them and protect them from host defenses. We remain unable to interdict their lethal ability to spread (metastasize). We are only beginning to understand how they evolve, and how they exploit mutations that destabilize their genomes and render them resistant to all known forms of therapy. Absent this knowledge, we lack what is needed to design truly effective anti-tumor drugs.

Major initiatives in basic cancer research will be necessary to apply cancer precision medicine robustly in most patients. This, however, will be far from sufficient. Experience in the few major cancer centers using this approach has made it clear that technological analysis is the most straightforward and cost-effective element of cancer precision medicine (Figure 6). Genomic analyses generate enormous amounts of data that must be correlated with large amounts of clinical, laboratory, imaging, and administrative data to place the results into an interpretable clinical context. Highly subspecialized experts in clinical cancer care, cancer pathobiology, and information science must then confer to interpret the data in a way leading to practical benefit for the patient. These individuals are scarce. Another consequence of personalizing therapies is that targeted drugs attacking a specific vulnerability will be useful only in the small percentages of patients whose tumors have that vulnerability. For example, crizotinib effectively treats lung cancers only in the 3% to 5% of patients whose tumors carry a rearrangement of the ALK gene. The smaller market space for this drug makes it hugely expensive to those patients and their insurers.

Fig. 6.

Complexity of delivering cancer precision medicine in a practical clinical setting. The figure shows the process used at Dana Farber Cancer Institute to make decisions about treatment or potential enrollment in a clinical trial on the basis of information generated by genome sequencing and/or other high throughput technologies (e.g., gene expression data, epigenomic mapping, proteomics, etc.). (Figure provided by Professor Barrett Rollins, Dana Farber Cancer Institute.)

In its present form, precision cancer medicine requires scarce expertise, a complex infrastructure, and the use of many highly expensive targeted agents, each useful in small subsets of patients. If it becomes applicable to many more patients, as present progress suggests that it will, we could encounter widening disparities in access to the best therapeutic options.

Despite these challenges, we believe that the basic stratagems of precision cancer medicine will ultimately benefit most patients. To deliver on that promise, comprehensive cancer centers must foster an environment that harmonizes an array of established and new approaches converging on the most refractory problems in cancer prevention, early detection, and therapy. This will require novel applications of molecular and classical epidemiology, better targeted therapies, and manipulation of the immune system. It will require the development of better agents that can modulate angiogenesis, inflammation, cancer cell metabolism, and the epigenetic disorganization of the cancer genome. Success will require highly focused and interdigitating research, clinical research, and clinical care initiatives (Figure 6). These efforts must integrate molecular, clinical, pathologic, genomic, pharmacological, epidemiological, and clinical outcomes data in a concerted way and bring insights derived therefrom to bear on patient care. Achieving this will be accomplished only if coherent information is gleaned from the data and made readily accessible to scientists and clinicians.

What must also happen in concert with the mastery of big science and big data is the crafting of smarter ways to use them to study the disease and its response, or lack thereof, to therapies in individual living patients. In other words, clinical trials must become smarter. Until very recently, the study of a new treatment regimen has depended on comparing the new regimen to an established one in progressively larger groups of patients. The patients were often eligible for a particular trial on the basis of broad tissue diagnoses — breast cancer, lung cancer, etc. Once the comparative treatments had been administered, months or even years would pass before certain end points were or were not achieved — regression of the tumor, delay in progression, adverse side effects, or prolongation of overall survival.

The problems with this approach, the best available until recently, are several. First, little was learned if the trial was unsuccessful. Why did the new regimen fail even though it was anticipated that it should work? Why did the few who responded do so while the majority did not? If unanticipated toxicities doomed the trial, what was behind them? Second, for the same reasons, too little is learned about a successful trial. Since response and remission rates are rarely 100%, how could we predict better who should or should not incur the risks of unproven therapy? How were the responders and non-responders or their cancers different? Third, the costs of these studies are huge; the failure rate is highest at the final most costly steps. Fourth, the timeline is tortuously long. The heroes and heroines of these studies are the patient volunteers. We owe them better and quicker results in return for their willingness to accept the risk of the unknown, the extra time and effort involved in participation, and, all too often, the added personal expense of additional treatments and procedures that payers may deny as experimental.

Science will be generating a flood of new drugs predicted to be beneficial on the basis of laboratory analysis of experimental animal models. Only a few of these will prove to be successful in clinical practice. How then to find the “winners” and eliminate the “losers” earlier, more safely, and less expensively? How do we learn more from every study, so that the next one has a better chance of working? How do we match the new agent to the patients most likely to benefit and least likely to suffer side effects?

The emerging paradigm for clinical trials, one whose mastery will determine whether cancer precision medicine will achieve widespread impact, is the “smart” clinical trial, having the following characteristics.

First, the patients entering a trial must be better matched to the therapy being tested. Classifying patients based only on the anatomic origin of their tumors no longer suffices. Newer agents are designed to attack specific molecular vulnerabilities. The relevant target is typically found in a few percent of patients with renal cancer, a few percent with colon cancer, etc. Moreover, the efficacy of attacking that target can be accentuated or abrogated by other mutations. Thus, smart trials will increasingly depend on patient groups who can be matched by nuanced molecular profiling of their tumors (Figure 6). By eliminating patients from the study for whom the drug would be irrelevant, this customization should generate definitive answers about efficacy and toxicity from smaller groups of patients.

Second, every trial must learn more from each patient studied. Simply waiting for a clinical result wastes time and learning opportunities. Methods are being developed to measure in real time what the drug is doing — is it penetrating the tumor? Is it hitting the metabolic target? Is it injuring the target cell? Is it provoking an appropriate immune response? Applying these will require an array of concerted resources that can support the intense “correlative biology” that will make each trial more informative. It will also require operational platforms and reimbursement mechanisms facilitating the performance of cutting edge imaging and re-biopsy during and after treatment.

Third, the trials must be faster and more cost effective by providing real-time information predictive of success or failure. Clinical testing of a new drug typically lasts nearly 5 to 10 years, only to result in failure more than half the time, at a cost of approximately $100,000 per day. Hence, the high cost of cancer drugs. Intense efforts are now underway to use correlative biology to define biomarkers that can guide the decision to “kill” a drug unlikely to succeed early in the process and expend resources only on those predicted to succeed.

The scientific intensity of the smart trials will raise the cost per patient studied and the costs of the early phase trials. On balance, though, smart trials, well-executed, should shorten time to approval, eliminate costs due to futile testing of agents doomed to fail, and lead to smarter drug design. Cost efficiency and higher predictability might also incent the biotech and pharmaceutical industry to take “more shots on goal” at a wider variety of targets, even those that seem riskier based on preclinical models.

FINAL THOUGHTS: WILL CANCER PRECISION MEDICINE MAKE A DIFFERENCE. CAN IT BE WIDELY APPLIED IN A SUSTAINABLE MANNER?

Discoveries made since the dawn of this century have fundamentally altered our understanding of cancer. Discovery research and technological advances have opened our eyes to a myriad of potential vulnerabilities within tumor cells. We are also now able to peer beyond the confines of the tumor cell and unearth the adaptive strategies used by tumors to create protective ecological niches for themselves. Fledgling efforts to reverse just one of these — immunosuppression via co-option of immune checkpoint mediators — are yielding gratifyingly dramatic clinical remissions. New technologies should permit us to use structural biology, combinatorial chemistry, and genetic engineering of proteins, RNAs, genes, and effector cells to create the more robust and safer agents needed to defeat every cancer. These advances embolden us to believe that, at last, the goal of ensuring longevity with high quality life for cancer patients is within reach.

Recent research has also sobered us by indicating that cancers are highly complex, adaptable foes, able to evade even the most modern smart drugs. Nonetheless, prudent optimism is justified because we have just begun to exploit our new insights therapeutically. Two decades ago, almost all available cancer drugs attacked only the processes of DNA replication and mitotic cell division. It is remarkable that these drugs accomplished anything. Yet, there are millions of cancer survivors for whom these agents were life-saving. Achieving even that much progress with such a limited repertoire of weapons should hearten us, because the plethora of strategies and agents coming into our grasp will yield far greater benefits.

Precision cancer medicine will have a deep and broad impact at a public health level only if can be applied to all patients. To realize this potential, we must approach the development and deployment of new therapies far differently. Each patient’s tumor will need to be interrogated to ensure that it is susceptible to exploitation of that vulnerability and has not evolved mechanisms that diminish the efficacy of the drug. Similarly, the tumor microenvironment and the patient’s immune profile must be characterized to determine whether immune augmentation therapies should be added or used instead. But, as we have pointed out, delivering the exact right drugs to the right tumor in the right patients at the right dose and time will be an incredibly resource-intensive and expensive undertaking. The expense could be so high that precision medicine, no matter how desirable, may be unsustainable.

If some form of care approaching the ideals of cancer precision medicine is our best hope to make great leaps in cancer survival with a high quality of life, then the pivotal challenge becomes how to achieve sustainability. I do not purport to have any better answers than the many experts who are pondering this question. Nonetheless, what follows are some personal thoughts about what might or might not work.

Many observers say that the high per-patient prices of new smart drugs are the culprit and that prices must be slashed even if by price controls. Discussions surrounding this topic are often tinged by political and social engineering agendas from both the right and left wing, but there is general agreement that, at the very least, the increase in smart drug prices must moderate or even be reversed if they are to be widely available. I am, however, pessimistic that major decreases in price will occur anytime soon, or that sustainability will be achieved even if they did. The number of new agents that must be developed, their small target market, and the short lifetime they enjoy as approved agents under patent protection (because of the prolonged development time with the patent clock ticking) means that even minimally profitable agents will carry high prices. Nevertheless, some moderation of drug prices must contribute to ensuring sustainability.

Speeding drug development with lower costs can also contribute. Smart clinical trials, better biomarkers, better analytics, and more sharing of “big data” gathered from the patients in these studies offer the means to accomplish this. Focusing resources on the best candidates and getting them into practice more quickly should cut development costs; pricing should follow. Additionally, far more patients must enroll in clinical trials so that more robust results can be obtained more quickly. Some trials addressing rare mutations now take years to accrue enough patients to get a result. Insurance coverage and reimbursement policies must change to remove disincentives to enrolling or even having one’s tumor profiled.

Less often mentioned as a source of expense is the massing of expertise, technologies, and infrastructure needed to perform the profiling, analyze the data generated, interpret it biologically and clinically, and translate the output into usable form for the clinician and patient. These highly expensive assets, currently accessible at only a few major cancer centers with a strong research base, are currently subsidized by research grants, philanthropy, research contracts with industry, and, ironically, the net revenues from clinical care, some generated from the mark up on cancer drug reimbursements. Who will provide the money to support this delivery system when precision medicine is no longer a research activity but standard clinical practice? No reimbursement mechanism currently exists to support these activities except below cost payments for time spent by physicians in actual face-to-face care encounters with a patient.

One can imagine ways to address the delivery system issue. Using real-time electronic communications and telemedicine, we could develop precision medicine networks anchored by hubs massing the needed expertise and technologies. One major center could potentially support tumor profiling and decision making across broad geographic areas and different care settings. To do this, reimbursement algorithms, medical liability practices, credentialing and licensing practices and many other system updates and policies must change, so that best clinical practice aligns well with legal, financial, and regulatory realities.

Even if all the foregoing changes were accomplished, precision cancer medicine would remain too costly. This leads me to conclude that its benefits will be accessible to all who need it only if fewer people need it. Most presentations about cancer, including this one, fail to highlight sufficiently the importance of efforts to prevent cancers or detect them at early, readily curable stages. Yet, these are by far the most cost-efficient ways to avoid deaths from cancer. If far fewer people had advanced cancers, precision cancer medicine would be more sustainable.

The fact is that we possess powerful measures that could prevent far more deaths if they were more effectively deployed. Vaccines (such as the HPV and hepatitis vaccines), and the elimination of cancer-promoting diatheses (tobacco use, tanning parlors, radon exposure, obesity, etc.) could, by current estimates, avoid up to half of cancer deaths worldwide. Their impact can be maximized only by more enlightened and vigorous use public policies, influence marketing, and direct personal lifestyle interventions.

Sadly, there are no certain preventative measures available or on the immediate horizon for many of the most common and lethal forms of cancer. Next best is detection of such cancers when they can be cured by local means (surgery, local radiation therapy). We need robust predictive biomarkers and far more sensitive and precise imaging capabilities that lower the limits of reliable detection to a few thousand cancer cells from our present limit of millions of cells. At these levels, cancers could be eliminated with minimal image-guided surgery. These tiny tumors would also be far easier to defeat with limited exposures to our best targeted drugs because they would be less heterogeneous, and much less ensconced in a co-opted host microenvironment.

Prevention and early detection research must, therefore, be prioritized and far better funded. A relatively small tax on tobacco products, tanning parlors, snack foods, smoked, cured, and red meats, high sugar soft drinks, alcohol, and carcinogen-generating energy products would increase the funds available for these areas several-fold, and yield more improvements in longevity and quality of life than many of the things that are covered by current taxes on those items.

As academic physicians, we have been privileged to be part of momentous advances in life sciences research and health care. Cataract surgery, once requiring hours of surgery and many days of recovery in the hospital can now be done safely between two stops at the shopping mall. Statins delay and often prevent cardiovascular deaths, and prolonged survival is the norm for many of those undergoing solid organ and bone marrow transplantation. Gene therapy is emerging as a curative modality for sickle cell anemia, hemophilia, and immunodeficiency disorders. Gene editing is on the horizon. As these and many more stunning advances have changed the face of our profession, cancer care has plodded through the past half century with real but far more modest improvements in outcomes. The purpose of this discourse has been to point out that at last progress toward meaningful control and increasingly more frequent cures of human cancers is accelerating rapidly. Thousands of patients are alive because of therapies introduced only a few years ago. The evolving paradigm of cancer precision medicine promises to maximize the benefit of these agents. To be available to all who could benefit from precision medicine, we must accomplish major changes in policy, reimbursement incentives, the operating model for care delivery, and deployment of scarce and expensive expertise and technologies. These attainments alone, however, will not achieve sustainability unless fewer patients are in need of this inherently expensive mode of care. This leads to the inescapable conclusion that we must also reduce the incidence of advanced cancers by major investments in prevention and early detection of cancer.

DISCUSSION

Dale, Seattle: Beautiful. Perfect talk for this morning and realistic and optimistic. We just have our fingers crossed for good health, right? Anyone want to ask Ed questions before we move ahead? Yes?

Hochberg, Baltimore: Wonderful, wonderful presentation. And a real plus for getting up on a Sunday morning! I had two questions: the first was if you could comment on the use of chemo-prevention as part of the prevention strategy. And the second is if you would comment on the cost per quality of life adjusted years gained. Because, in your figure about the cost of drugs, which is now approaching $100,000 per year, if we could look at that in terms of cost per quality years of life, and then put into the spectrum of what society might be willing to pay, we might find, eventually, that we may not be able to afford the therapy even though patients want it.

Benz, Boston: Thank you very much, Marc. They are both good points. With regard to chemo-prevention, you know the Holy Grail would be “Could one take something every day that would prevent one from getting cancer?” Selenium was popular once as a preventive for prostate cancer. It turned out not to work, at least for the majority of people. Anti-oxidants were thought to be good preventative drugs, but they actually turned out, paradoxically, to be associated with slightly worse outcomes at least in terms of developing some cancers. The field of chemo-prevention is in its infancy. I would argue that we’re not going to know how to use chemo-prevention drugs for quite a while. Our fledgling efforts have revealed that individual people are highly variable to the degree to which they would be benefitted or potentially harmed by such agents. One good example is the best known and most effective chemo-preventative agent, aspirin. It is now clear that aspirin can be effective for preventing or at least slowing the progression of colon cancers, but only in patients who have polymorphisms for the appropriate genes associated with the multiple metabolic effects of aspirin.

The use of anti-oxidants to reduce risk of prostate cancer followed a similar experience; in some patients there appeared to be benefit, in others no benefit, and in others potential harm. The difference among these groups appeared to be related to polymorphisms in the superoxide dismutase gene. So, chemo-prevention will need to follow the same precision medicine paradigm as for treatment of established cancers. The regimen will need to be customized to patients based in part on their pharmacogenetics, and their risk profile for particular cancers. Finally, we appear to be a very long way away from having any single agent that would be a “pan-preventative,” that is, a drug that would prevent all forms of cancer. Preventing even the major killers could well require one to commit to life-long ingestion of multiple medications whose cumulative toxicities could limit their use.

And, to the cost per month or years of life saved, I can’t give you precise figures with regard to the newer therapies because we don’t yet have the denominator — that is, the years of life saved. But even for conventional cancer therapies that have existed during the past 20 years or so, the estimate is that there are multiple trillions of dollars saved relative to the cost of untreated cancers. These numbers in my head came from studies by University of Chicago–affiliated economists. They compared the cost of care to the cost of years of life lost in terms of economic productivity and other social factors. You can kind of figure out that, on a per year of life saved basis, a great deal depends on the age of the patient. Saving one child from a childhood cancer yields a very high net economic benefit; pre-menopausal breast cancer lives saved would be pretty high in economic benefit, but prostate cancer not so high because most people either don’t die from it or die at an advanced age. So, there clearly is tremendous variation by the individual type of cancer, its lethality untreated versus treated, and the age at which the cancer occurs. You must look at the cost of not taking care of cancer in terms of lost productivity and other contributions to the economy and society. This is my long way of saying that I can’t give you precise figures and even highly quantified numbers would be meaningful only for individual cancers in particular subgroups of patients. People are now watching closely with these newer drugs because of their high unit cost, and because, clearly, the pharmaceutical companies would like to expand the indications for their use. I do think that you will see the tolerance for pricing and reimbursement increasingly related to proof of quality years of life saved.

Merajver, Ann Arbor: Thank you so much for such a wonderful summary and the optimism for the future. I have a comment and a question. The comment I have is, and it’s something that may be worth emphasizing, is that the targeted therapies and immunotherapies also have a lot of side effects. And there is this superficial understanding that chemo is horrible but these immunotherapy drugs cause nothing. And that’s absolutely not true. They have a whole spectrum of very serious side effect and a lot of drug interactions with other comorbidities as we get older we get diseases other than cancer. And the comorbidity interactions are even more profound than with chemotherapy. I have another question for you: and that is in my brief talk I presented a model where actually academia and universities take more of a role in costly drug development; but with $100,000 a month in cost, there will be no payment plan that will fix that problem in Africa. So, how do we really think completely out of the box? I gave an example of what we are doing in Michigan. Other universities are maybe thinking about this. But we are very determined to taking tumors that have no targets and getting to first in human trials within the university setting, versus farming it out or having pharmaceutical companies basically take it all the way. And then perform trials that don’t necessarily address things like resistance about which you so eloquently spoke. So I’m very interested on your perspective.

Benz, Boston: Thank you. So, you’re quite right. Immunotherapies, in fact all targeted therapies, are not free of side effects. You know, if it’s a drug, it’s got side effects. And unfortunately for cancer I can’t follow Marshall Wolf’s advice. He was my residency program director and I learned to follow his advice on just about everything in clinical medicine and it was great. He said use a drug only until people know it has side effects. But I have not yet been able to find a way to use a cancer drug before the side effects show up! So, yes, you get all kinds of colitis, pneumonitis, some patients with encephalitis, dermatitis because you’re revving up the immune system and over activating it in an effort to get rid of the tumor. So, that’s the reason it limits its use. Fortunately, many of these are dose related so you can get away with a lower dose. The targeted therapies have a variety of their individual side effects as well. Not quite as devastating as traditional chemotherapy but like every other drug you pay a price when you take it. I will just briefly comment about what you said about academic institutions taking on more of the act of drug development. I think most cancer centers are doing this now. In fact, it’s become sort of an unwritten criteria for doing well on the renewal of your core grant for the comprehensive cancer centers. There are roughly 300 or 400 potential molecular targets out there, and pharmaceutical companies are working on a small fraction of them — the ones they think are going to have a reasonable market. So, it’s up to us in the academic community to develop the others because the others that don’t seem relevant today are going to be relevant tomorrow. Imatinib, that drug I mentioned is a front-line therapy now and having a very important effect for treating GIST sarcomas. You never would’ve guessed it when you were setting about to develop that drug.

Tweardy, Houston: Ed that was a great presentation. Very comprehensive. One question I had relates to the challenge that arises as we further dissect and atomize, if you will, the different diagnosis of cancers. How do we actually get the trials done in any one center to actually demonstrate that the therapeutic molecule designed to go to the specific target is having a beneficial effect?

Benz, Boston: Thank you David. We must look at the impact of the new way of approaching cancer research and treatment as a mandate to reformulate the paradigm for performing clinical trials. In fact, I think cancer is the canary in the coal mine for what’s going to go on in the rest of medicine as precision medicine takes hold in the approach to other major disease types. In cancer centers, we are increasingly moving to what we call bucket or basket trials where we look at putting the patient into the trial, or not, based on molecular profiling of her or his tumor rather than on anatomic site. I actually prefer the term basket trials because bucket trials for cancer has an unfortunately ominous indication! Increasingly, we’re going to be doing n-of-one trials where you, based on molecular profiling, decide to try a drug in an individual patient. The only way you’re going to be able to decide if that drug works for that particular molecular tumor profile is to put that patient on the drug, and submit the results of the patient’s outcome into an international database and compare the outcomes of that patient with others who are quite similar and who have received the same drug on the basis of their profiling. Then one can aggregate all the data, and hopefully, pull back out some insight as to the range of situations in which the drug to likely to be most effective. Traditional randomized blinded trials will be increasingly difficult to perform as we start to look at smaller and smaller slices of the population having cancers that are truly similar based on molecular profile.

Oates, Nashville: I’d be interested in your thoughts about in the future the use of genomic profiling as a guide to chemo-prevention. The elegant meta-analysis by Roswell and colleagues of the controlled clinical trials on aspirin indicates that in those trials if you develop the colon cancer during the trial your risk of further metastasis was reduced by 83%. Eighty-three percent in reduction in metastatic disease which is the cause of death in those patients. So, do you envision that may be the way of the future to decide how one might select patients who would go on aspirin treatment as a chemo-prevention?

Benz, Boston: This has to be the one and only moment I could be like JFK at a press conference and simply say…Yes! In this case I can say “yes” because one of the best chemo-preventatives you can take is actually aspirin. Aspirin is good for preventing cancer in addition to its value in heart disease. It clearly seems to have a beneficial effect on reducing mortality from colon cancer even if you start taking it after it’s been diagnosed provided it’s not at a very advanced stage already. You know, people argue about NSAIDs and things like that in terms of their chemo-preventative value. I do think, and I’m telling you my own thinking on this, that during the next 5 to 10 years the cost of doing a genomic sequence at the time of birth, or at the time after conception and before birth, is going to drop to the point where everybody will have access to that kind of information if they can get reasonable access to a reasonably sophisticated site of care. At that point you might know enough about each patient as a “host” of potential cancers that you could design a personalized chemo-prevention cocktail that might actually work. Something like that will be necessary if we’re going to know who’s going to be at risk to get what kinds of cancer, and therefore would have the best benefit from a particular set of chemo-prevention agents. There is another form of chemo-prevention called secondary chemo-prevention. It sounds like an oxymoron because in this case you already have the cancer. The goal of secondary chemo-prevention is to prevent recurrence and metastasis. The use of tamoxifen to prevent breast cancer recurrence is, possibly, the best known example; in addition, a number of the men in this room have taken androgen suppressing agents to slow the progression or prevent the recurrence of prostate cancer. This is secondary prevention. The study you mention adds aspirin to that list. This is another reason why everyone who comes into the door of our care centers should, in the not too distant future, have their genomes and their tumors sequenced, or analyzed by whatever other high throughput technologies we can bring to bear, to develop a highly detailed personalized picture of the individual and his or her cancer. I also think that a lot of the chemo-prevention progress that we’re going to have to explore is going to depend more on what we learn to understand from epigenomics than what we are going to get simply from mutational analysis now. This is because the epigenome is in many ways more sensitive to the environment as well as to our intrinsic genomic makeup; epigenomic phenomena also open up broad regions along chromosomes that can have a far more profound effect than single mutations.