#485 — The New Science of Cancer

Measuring Prevention Effectiveness in Research Is Challenging and Lengthy

Prevention science is exceptionally challenging because it aims to stop cancer before it starts, making it necessary to prove that interventions can prevent an event that, ideally, never occurs. Prevention trials typically involve large, healthy populations followed over many years to observe if cancer incidence decreases—a slow process since cancer rates might be as low as 100 cases in 100,000 people. As a result, these studies often stretch out over five to ten years to reach sufficient statistical power. To circumvent the lengthy timeline, researchers sometimes focus studies on people at high genetic risk, but for most cancer types, long-term follow-up is still necessary.

A major hurdle is the lack of reliable surrogate biomarkers for cancer prevention. In contrast to cardiovascular disease, where biomarkers like high cholesterol and hypertension accurately predict future events and guide intervention, cancer lacks universally reliable predictors. With heart disease, lowering blood pressure or cholesterol can serve as measurable goals for risk reduction. For cancer, no such surrogate exists for most types, so trials must wait for actual cancer cases, making research labor-intensive and slow.

"Inflamigens" Offer New Path in Cancer Prevention, Unlike Traditional Mutagens

Recent discoveries highlight a new category of carcinogens called "inflamigens," which activate dormant cancer cells through persistent inflammation rather than by causing DNA mutations, unlike traditional mutagens such as x-rays or formaldehyde. Instead of directly damaging genetic material, inflamigens alter the surrounding tissue environment—the "soil" in the seed-and-soil analogy—triggering dormant cells to initiate tumor growth.

A notable example is particulate air pollution: minute airborne particles do not act as classic mutagens but induce chronic macrophage-mediated inflammation. This prolonged stimulation changes tissue conditions, allowing pre-existing, dormant cancerous cells to proliferate. Asbestos, long recognized as a carcinogen, is now also believed to function through this inflammation-driven mechanism, rather than through mutagenicity alone.

No New Major Preventable Chemical Carcinogens With Widespread Impact Have Been Found Since the 1960s, Though Viral Carcinogens Like Hpv Are Targeted With Vaccines

Since the 1960s, scientists have not found any new preventable chemical carcinogens with a major impact across the general population. While substances like asbestos or formaldehyde have been removed from specific industrial environments, such carcinogens typically affect only niche groups. Broad environmental causes of cancer remain elusive, either because few such agents exist, the current methods to detect them are inadequate, or cancer risk comes from a combination of many subtle exposures ("a death by a thousand cuts").

In contrast, significant progress has occurred with viral carcinogens. Human papillomavirus (HPV), for instance, is a major cause of cervical cancer and is now preventable with highly effective vaccines. Well-controlled studies, such as large randomized trials in Sweden, demonstrate that vaccination against high-risk HPV strains can reduce future cervical cancer risk to zero, especially when administered to adolescents. Despite this, cervical cancer, which is entirely preventable, still leads to many deaths worldwide, largely due to gaps in vaccine implementation.

New research into inflamigens and their unique inflammatory signatures could soon enable the development of biomarkers to identify people at risk for inflammation-driven cancers, much as cholesterol serves in cardiac care. However, not all inflammation causes cancer; only the specific macrophage-mediated type fostered by inflamigens is implicated in cancer development. This means general anti-inflammatory drugs are insufficient for cancer prevention—targeted approaches addressing the precise inflammatory pathway are needed.

Genetic Counseling for High-Risk Individuals: Brca1, Brca2, P53 Mutations, Polygenic Risk

Certain individuals face markedly high cancer risk because of inherited mutatio ...

Bayes' Theorem Explains High False Positive Rates in Cancer Screening Despite Strong Sensitivity and Specificity

Siddhartha Mukherjee explains that understanding the high rates of false positives in cancer screening, especially with liquid biopsy and cell-free DNA tests, requires a grasp of Bayes' Theorem. While many companies promoting these tests emphasize their sensitivity and specificity, they often neglect the central role of prior probability—the background rate of cancer in the population. The likelihood that a positive screening actually reflects true early-stage cancer is dominated by how common cancer is in the screened group. This means that even a test with excellent sensitivity and specificity will mostly provide false positives when used on populations with a low underlying cancer rate.

Mukherjee illustrates this with a "needle-in-a-haystack" analogy: if a detector is 90% sensitive and 90% specific but you’re searching for a rare event (a needle among much hay), the detector will alarm for hay more frequently than it will actually find the needle. This explains why rare-cancer populations produce more false positives than true positives—even with strong diagnostic tools.

Liquid Biopsy and Cell-Free Dna Tests Are Poorly Marketed To Asymptomatic Populations, Ignoring Bayesian Limitations That Make Unavoidable False Positives Distressing

Mukherjee observes that companies marketing cell-free DNA and liquid biopsy tests to the general, asymptomatic population often highlight their low false positive rates but overlook the modest positive predictive value when prior risk (the Bayes rate) is low. Sam Harris shares a personal experience where a company advertised a false positive rate of one in 200, but after a positive test result, he faced a much higher purported chance of having cancer, leading to significant anxiety and confusion.

The issue is that advertising a low type I error rate (such as 0.5%) is misleading if the consumer's risk is low—most positive results will still be false alarms. Many people, as Harris notes, do not find assurance in these rates because the probability that a positive test reflects real, actionable early-stage cancer remains modest in general screening populations. Consequently, these tests produce distress and unnecessary follow-up, while failing to deliver real benefits to low-risk, asymptomatic individuals.

Cell-Free Dna and Other Screening Tests Should Be for Populations With High Cancer Risk, Including Those With Family Histories, Genetic Risk Factors, or Prior Cancer Diagnoses

Mukherjee argues that these screening tools are best deployed in populations with elevated baseline risk, which increases the probability that a positive test truly indicates cancer. This includes people with strong family or genetic histories or those who have had cancer previously and are in remission. Minimal residual disease monitoring—using cell-free DNA to detect the earliest return of cancer—is especially justified for those in remission who face substantial recurrence risk.

For instance, in patients with myeloma in remission, monitoring for cell-free DNA can help detect minimal residual disease, allowing for early intervention while the disease burden is low and more treatable. Likewise, individuals with mutations or polygenic risk factors, or with strong familial cancer history, are more likely to benefit from this kind of screening because their higher prior probability of cancer means that positive results are more likely to be true positives.

Mri Challenges Similar to Blood Screening; Sequential Imaging May Lower False Positives

MRI-based screening shares analogous limitations with blood tests. A first-time, full-body MRI may reveal findings that prompt invasive follow-up, even when those findings have existed harmlessly for years. Most people are uncomfortable with a "wait-and-see" approach for ambiguous resu ...

Cancer Mortality Rates Decrease Due to Advances in Prevention, Detection, and Treatment

Siddhartha Mukherjee highlights the substantial decline in U.S. cancer mortality rates over the past two decades. Twenty years ago, cancer claimed 200 lives per 100,000 people, a figure that has dropped to about 140 deaths per 100,000. This absolute progress is due to a combination of factors: prevention (with the most significant impact from reduced smoking rates), improvements in early detection, and advances in treatment.

Immunotherapy Meds Removing Cancer's Defenses Yield Cures, With 20% of Advanced Lung Cancer Patients Surviving Five Years

Immunotherapy represents a watershed in cancer treatment, mobilizing the patient’s own immune system against cancer. These drugs disrupt cancer’s ability to hide from immune cells or redirect the immune system to attack the cancer directly. Their impact on certain cancer types has been transformative. In non-small cell lung cancer, previously deemed incurable at advanced stages, about 20% of patients now survive five years post-immunotherapy, far beyond previous expectations. In some cases, the word "cure" now applies, a dramatic shift in outlook for previously fatal cancers.

Immunotherapy has also yielded real progress in cancers such as bladder and advanced breast cancer. With advanced therapies, including immunologic and antibody treatments, some people with breast cancer are living 5, 10, even 15 years after diagnosis, with many remaining active and functional. For several cancers, what was once a deadly diagnosis has now become a chronic, and sometimes even curable, condition, thanks to improvements in therapy that differ in their mechanism from one tumor type to another.

Iterative Advances Improve Survival in Newly Diagnosed Multiple Myeloma Patients Over Decades

Multiple myeloma serves as another example of steady, incremental progress. If you track survival rates by year of diagnosis, there is a consistent pattern: from 1990 onward, people diagnosed with multiple myeloma at the same disease stage have lived longer with each passing decade. This improvement results from continual innovation in therapeutic approaches.

Car-t Cell Therapies Achieve 50-60% Cure Rates in Resistant Leukemias and Blood Cancers

CAR T-cell therapies have been revolutionary for certain blood cancers. By engineering a patient's T cells to target cancer cells, this approach has achieved striking results in diseases like acute lymphoid leukemia (ALL) in children. Historically, 80–90% of children with ALL could be cured using aggressive chemotherapy, but this left a vulnerable group—10–15% with relapsed or chemotherapy-resistant disease. CAR T-cell therapy now brings five-year survival rates of 50–60% for these children, representing a breakthrough for a population where prior options ran out.

However, CAR T therapy is much less effective against solid tumors. The microenvironment around solid tumors—composed of blood vessels, immune, and supportive cells—prevents T cells from infiltrating and killing the cancer. Despite ongoing research combining CAR T with other treatments to change this microenvironment, a true solution for most solid tumors remains elusive.

Targeted Therapies: Initial Success in Previously Intractable Pancreatic Cancer

Targeted therapies are also broadening the reach of effective cancer treatment. A RAS inhibitor, recently tested in clinical trials for pancreatic cancer, extended median survival from six months (on previous standard therapy) to 13 months. While not a cure, this is the "first crampon" or foothold in two decades against a cancer whose mortality rates had not budged. The success raises new questions—such as why relapses still occur after 13 months—leading to further innovation in an iterative, ste ...

Artificial intelligence (AI) is transforming multiple facets of cancer medicine. Across detection, drug discovery, clinical design, and prevention, AI enhances accuracy, efficiency, and understanding. However, these advancements also underscore the need for human expertise and oversight.

Ai Enhances Cancer Detection as a "Companion Diagnostic" in Radiology and Pathology, Supporting Human Experts

AI has emerged as an indispensable partner in radiology and pathology. Rather than replacing experts, it acts as a “companion,” supporting professionals through diagnostic collaboration. AI systems are trained on huge datasets—for example, hundreds of thousands or even millions of cancer images, such as melanoma—compared to the relatively limited exposure of human pathologists. In practice, this means AI can act in triage or as a second opinion, identifying subtle abnormalities or confirming findings that a human might overlook.

In mammography, AI helps reduce the miss rate in early breast cancer detection by alerting radiologists to overlooked patterns, functioning almost like a colleague whispering advisories while reviewing scans. This approach extends to lung cancer screening in high-risk patients and skin lesion assessment for melanoma. In all these cases, AI’s broad exposure and pattern recognition power improve accuracy, bolstering confidence in diagnoses.

In pathology, studies show that AI can distinguish between malignant and benign lesions more accurately than many pathologists, thanks to the vast array of images it processes. However, research adoption varies, and AI is generally viewed as a collaborative tool rather than a replacement for human judgment.

Ai In Drug Discovery: Mastering Complex Medicinal Chemistry Beyond Pattern Learning

In drug discovery, AI’s role extends beyond simply detecting patterns in images. The field requires mastery of complex rules governing chemical interactions with proteins—a level of reasoning that far surpasses simple pattern recognition. Traditional medicinal chemists spend decades learning how to create molecules that interact with specific biological pockets.

Siddhartha Mukherjee describes how his company, Manus AI, tackles this challenge. Since there are not enough real-world examples for AI to simply learn drug discovery by rote, AI must be taught the principles of medicinal chemistry—how molecules should interact to inhibit or activate cancer proteins effectively. This represents a significant leap, as AI needs explicit training in the fundamentals of chemistry to engineer drugs rather than just recognize them. While examples of full AI-driven drug success remain limited, this approach holds promise for reducing the time and cost involved in bringing targeted cancer therapies to patients.

Ai-driven Target Identification in Cancer Therapy

Another domain where AI excels is target identification. Every cancer drug works by binding to a protein “target,” and finding the right protein is paramount. AI assists by processing high-dimensional cellular data to identify which proteins (“locks”) should be targeted to halt cancer growth. This multidimensional analysis finds optimal targets more efficiently than traditional methods, driving the discovery of new treatment strategies.

Ai Enhances Clinical Trial Design and Recruitment Efficiency

AI powers major improvements in clinical trials, both in design and recruitment. For recruitment, AI searches electronic medical records—under strict privacy rules such as HIPAA protections—to identify candidates who best fit specific trials or drugs. Natural language models can efficiently sift through mountains of data to find suitable participants, accelerating trial enrollment and potentially improving outcomes.

Beyond recruitment, AI optimizes trial structure through adaptive trials. Unlike stati ...

Funding Cuts and Policy Changes Create Chaos in American Research and Drug Development

Severe funding cuts to organizations like the CDC, threatened reductions to many other agencies without any substantial budget increases, and misaligned scrutiny toward research have created a chaotic environment for American medical research. Essential research often faces misplaced oversight, while some areas lacking proper scrutiny get ignored. Such administrative decisions can dismantle organizations with the stroke of a pen—disbanding teams, undermining institutional knowledge, and causing skilled individuals to lose training, jobs, and interest in the field. Once disrupted, rebuilding these scientific ecosystems can take years or decades due to the loss of expertise and the slow process of retraining and restoring lost progress.

U.S. Losing Position in Pharma Innovation; China's Drug Imports to Rise From $5b (2020) To $60-70b (2025)

The U.S. has historically been a leader in pharmaceutical innovation, exporting groundbreaking medicines and holding a competitive technological edge. However, underfunding and policy disruption threaten this standing. In 2020, the U.S. imported $5 billion worth of drugs from China, a number projected to skyrocket to $60-70 billion by 2025. This trend signals a dramatic shift of economic value and technological leadership toward Chinese biotech companies, reflecting not a lack of American innovation but the direct consequences of disrupted funding and unstable policy. This loss is difficult to reverse and erodes a key source of national economic and scientific strength.

Fragile Supply Chains for Medical Products Like Sterile Saline Threaten Hospital Function and Surgical Capacity

The COVID-19 pandemic exposed critical weaknesses in supply chains for essential medical products. Despite recognizing the need for domestic production, policy has continued to favor offshoring, increasing U.S. dependence on foreign manufacturing for vital resources like sterile saline. Hospitals, regardless of their advanced technologies, rely on basic supplies such as intravenous saline, without which fundamental operations and procedures become impossible. Failure to secure robust, onshore manufacturing for these products puts the entire healthcare system at risk. Building domestic manufacturing capacity could simultaneously enhance supply chain resilience and create jobs, yet this remains mostly unfulfilled in federal policy.

Trust in Science Erodes due to Policy Actions and Poor Communication on Research, Drug Pricing, and Medical Advances

Public trust in science has suffered due to both top-down policy decisions and communication failures. Scientists' isolation in "ivory towers" has contributed to a d ...