What this covers

Oncologist Azra Raza joins Russ Roberts on EconTalk to argue that cancer treatment has stalled. The core claim is blunt: the main tools available to oncologists—surgery, chemotherapy, and radiation, collectively termed slash, poison, burn—have changed little in fifty years, and the survival gains the field celebrates come almost entirely from earlier detection and smoking prevention, not from better drugs or techniques. Raza, drawing on decades treating acute myeloid leukemia, traces this stagnation to a fundamental mismatch between how cancer actually behaves and how the field tries to fight it. The conversation moves between clinical observation, economic incentive, and the philosophy of complexity, with Roberts offering pushback and alternative perspectives throughout.

The range spans several overlapping failures. Reductionism in drug development—testing molecules on mice, then scaling to humans—fails catastrophically because cancer is heterogeneous and mutates constantly; ninety-five percent of clinical trials for cancer fail in translation. The patent system and third-party payment structures create perverse incentives: Medicare pays for marginal efficacy gains (two months of life extension) at enormous cost, while thousands of redundant trials chase the same immune therapy across different cancers rather than pursuing novel targets. Raza also challenges screening—PSA tests, mammograms, colonoscopy—as often causing more harm than good through false positives and unnecessary intervention. Her proposed alternative abandons the hunt for a magic bullet in favor of detecting the first malignant cell by studying all five biological compartments—cells, DNA, RNA, proteins, metabolites—in real human tissue using modern genomic, proteomic, and metabolomic approaches. The discussion includes Raza's reflections on listening to dying patients, the role of emergence in complex systems, and why groupthink in science funding blocks breakthroughs—alongside skeptical questions about whether such a pivot is feasible or even wise.

Sharpest takeaway

Raza argues that 50 years of cancer research have produced little real progress in treatment—the field is still using 'slash, poison, burn' while gains come almost entirely from earlier detection—and that resources should be redirected from a failed reductionist, animal-model approach toward detecting the 'first cell' using multi-compartment human data and modern technology.

  • Survival gains come mostly from early detection and anti-smoking, not from better treatments, which remain unchanged for 50 years
  • The reductionist, mouse-model drug-development pipeline fails because cancer is heterogeneous and constantly mutating
  • Misaligned incentives (patents, third-party payment for marginal efficacy, redundant trials) waste resources that should fund early detection of the first cell

The argument · threads6 threads · 34 claims
0.85

Cancer is heterogeneous, multivariate, and patient-specific; reductionist approaches targeting single molecules or pathways cannot succeed.

6 pointscentrality 5/5
  • All research on cancer should have one and only one goal—to benefit the cancer patient; studying biology in mice, rabbits, fruit flies, and zebrafish is fine for understanding mechanisms, but if it is not therapy-driven and will not help the patient, it is of no use, and spending hundreds of billions chasing intracellular signaling pathways in artificially created mouse tumors is a waste.

    “My contention, Russ, is that all research on cancer should have one, and only one, goal, which is to benefit the cancer patient.”

  • Instead of the reductionist focus on one gene or one protein signature, we should study all five biological compartments simultaneously—cells, DNA, RNA, proteins, and metabolites—in serially collected actual human samples, using genomic, proteomic, and metabolomic technologies and machine learning, to detect cancer at its earliest stage.

    “now we have five compartments--cells, RNA, DNA, proteins, metabolites. Why should we look at any one of them... We should be looking at everything simultaneously.”

  • The intellectual infrastructure of drug development—testing on mice, then a larger animal, then humans—is fundamentally mistaken for cancer: results almost never translate from mouse to human, with about 95% of cancer clinical trials failing, because cancer is multivariate, heterogeneous, and different in every person and even within one tumor over time.

    “the idea that we can create a tumor in an animal model, treat it with a given agent and then bring those results to the human bedside. Well, this is met with unmitigated disaster.”

  • Treating cancer like one disease is like treating Africa like one country—a tumor is not even the same in two sites in the same patient or on two different days in the same patient, so a reductionist approach trying to find one molecule and develop one magic bullet cannot work amid the chaos of the tumor, its microenvironment, signaling, immune response, and angiogenesis.

    “I always say that treating cancer like one disease is like treating Africa like one country. It's not the same even in two sites in the same patient, or two days in the patient. The tumor has changed.”

  • Despite 60 years of searching, we have not found molecules expressed only by cancer cells and not by normal cells—we cannot find the unique 'zip code' of a cancer cell—which is why targeted and immune therapies destroy whole organs rather than just the cancer.

    “despite looking for 60 years, we have not been able to find molecules that are expressed only by cancer cells and not by normal cells. In other words, we can't find the address, the unique zip code, of a cancer cell so far.”

  • Cancer begins in one cell, but because that cell divides extremely rapidly—covering in hours what should take months—it makes DNA copying errors (mutations) with each division, so that every time a cancer cell divides there are potentially two entirely new cancers, producing within days a tumor of heterogeneous, biologically distinct cell populations.

    “every time a cancer cell divides into two, it picks up new mutations, which means now there are potentially two entirely new cancers.”

0.80

Current cancer treatments have not meaningfully improved in 50 years despite enormous research spending and technological advances elsewhere.

5 pointscentrality 5/5
  • Although we cure about 68% of cancers seen today, that success comes mostly from earlier detection, while the treatments themselves—surgery, chemotherapy, radiation ('slash, poison, burn')—are by and large the same ones used for 50 years.

    “while it is very true that we are killing 68% of cancers we see today, the reason we are curing them is mostly because of early detection. And, the treatments we are giving them are, by and large, the same treatments that we have been giving for 50 years--which is the slash, poison, burn.”

  • We have known since 1907 that it is not cancer itself that kills but the delay in treatment, so the earlier the disease is detected the better—yet we still rely on 50-year-old detection methods (PSA, mammograms, scopes) despite supreme technological advances elsewhere.

    “we know clearly since 1907 that it's not cancer that can kill but it's the delaying the treatment that kills; and the earlier we detect the disease the better”

  • The roughly 30% of patients we don't cure—those diagnosed at an advanced stage—have outcomes today that are roughly the same as they were 50 years ago, because slash, poison, burn has made no difference to them.

    “while it is very true that we are killing 68% of cancers we see today, the reason we are curing them is mostly because of early detection. And, the treatments we are giving them are, by and large, the same treatments that we have been giving for 50 years--which is the slash, poison, burn.”

  • Cancer mortality has fallen about 1% a year, translating to a 27% decline over three decades, but most of that came from two causes—the anti-smoking campaign finally showing results and earlier and earlier detection—rather than from better treatments.

    “we have brought down cancer mortality by 1% a year... that has translated in the last three decades into a 27% decline in mortality. But, most of that has happened for two reasons. One, because the anti-smoking campaign... and... we are detecting cancers earlier and earlier”

  • Raza was giving the same two drugs to treat acute myeloid leukemia in 1977 that she still uses in 2020, with the same dreadful results—and to this day we still don't know the precise mechanism of action of those two drugs.

    “I was giving the same two drugs in 1977 to treat a patient like her that I'm using today in 2020. The same two drugs with the same dreadful results.”

0.70

Early detection, not better drugs, drives cancer survival; delaying treatment fundamentally worsens outcomes regardless of therapeutic advances.

3 pointscentrality 5/5
  • Rather than rely on any single 50-year-old annual test, we should use the latest scanning, imaging, biomarking, genomic, proteomic, and metabolomic technologies to develop perhaps 500 different tests that can be amassed in a bar-code fashion and run rapidly to identify individuals in the earliest stages of cancer; half of resources should go to treating current patients and half to early detection and prevention.

    “we need to develop using the latest scanning, imaging, biomarking, genomic, proteomic, metabolomic technologies to find maybe 500 different tests that can be amassed in a bar code fashion... to identify individuals in the earliest stages of cancer.”

  • One in five new cancers occurs in a cancer survivor, because the underlying reasons that caused the first cancer remain and chemotherapy for the first cancer adds further damage; with 16-17 million survivors in 2020 rising to 20 million by 2025, this population should be most carefully monitored for the appearance of the first cell.

    “one in five new cancers appear in a cancer survivor. And, the worst case was my husband. He had the first cancer at 34 years of age, and a completely different one at 57, of which he died.”

  • Just as paradigms shifted overnight from typewriter to word processor and from phone booths to smartphones carried even by homeless beggars in impoverished countries, modern technology can transform cancer detection within a decade—as shown by genome sequencing dropping from 15 years and over a billion dollars to a few weeks for a few hundred dollars—so the problem is far more tractable than it appears.

    “what 20 years ago took 15 years to do to sequence the first genome and cost over a billion dollars, today can be done in a few weeks for the cost of a few hundred dollars. This is where technology can transform everything within a decade”

0.66

Serial sampling of actual human tissue enables identification of cancer's earliest precursors and risk factors through integrated molecular analysis.

5 pointscentrality 4/5
  • Myelodysplastic syndromes (MDS) are a group of pre-leukemic conditions in which blood counts fall and anemia develops; MDS is a disease in itself that can kill, but about a third of these patients go on to develop acute myeloid leukemia and die.

    “these syndromes were eventually gathered under the umbrella of myelodysplastic syndromes, or MDS, which is really a disease by itself that can kill, but a third of these patients can develop acute myeloid leukemia and die.”

  • The sequencing of the human genome required simultaneous collaboration and competition among hundreds of thousands of investigators, showing that incentivizing a new goal—the first cell—and letting the race begin could similarly accelerate progress in cancer detection.

    “the human genome required collaboration and competition simultaneously between hundreds of thousands of investigators”

  • Acute myeloid leukemia is a liquid cancer that is easier to study than a solid tumor like lung or ovarian cancer, which is why Raza focused on it when she began her career in 1977.

    “acute myeloid leukemia, a liquid cancer which is easier to study than a solid tumor like lung cancer, ovarian cancer, etc.”

  • Since 1984 Raza has banked bone marrow and blood samples from her own MDS and leukemia patients, accumulating over 60,000 samples from thousands of patients—none from another investigator—allowing serial study of cells, RNA, DNA, proteins, and metabolites as patients progressed from pre-leukemia to acute leukemia, enabling researchers to work backward and identify risk factors.

    “I have now collected over 60,000 samples from thousands of patients. Not one cell in my tissue bank has come from another investigator... I can go in, look at the cells, look at RNA, DNA, proteins, and metabolites in a serial fashion--as the patient progressed from pre-leukemia to acute leukemia.”

  • When an immune therapy is approved for one cancer, rather than running one trial to test it elsewhere, thousands of redundant trials (e.g., 2200) ask the same question—each costing millions, with Phase One around $25 million and Phase Three from $500 million to $2.5 billion—so a 'me too' rush to make money stifles creativity and the development of novel targets.

    “instead one trial there are not 500 trials, there are not 1000, there are 2200 trials going on to ask the same question. Now, each clinical trial costs millions of dollars. Phase One costs $25 million, and Phase Three costs anywhere from $500 million to $2.5 billion dollars.”

0.56

Current therapies cannot distinguish cancer from normal cells, causing severe organ and systemic damage regardless of efficacy.

3 pointscentrality 4/5
  • Chemotherapy is called 'poison' because it cannot distinguish a normal cell from a cancer cell; it kills rapidly-dividing cells, which is why its worst side effects (hair loss, nausea) hit rapidly-dividing normal tissues like hair follicles and the GI tract, and it works only because cancer cells divide faster than normal cells.

    “Chemotherapy cannot distinguish between a normal cell and a cancer cell. But, it kills rapidly-dividing cells. That's why side effects of chemotherapy affect rapidly-dividing normal cells the most.”

  • With current immune therapies, the only patients who respond are those who experience the most severe side effects—the cytokine storm that puts their life at stake; patients who don't experience this horrendous reaction don't eventually respond to the therapy at all.

    “the only patients who respond are the ones who experience the most severe side effects, called the cytokine storm, which basically puts the patient's life at stake. That, if they survive it, they will enter remission. Those patients who don't experience this horrendous cytokine storm, they don't even eventually respond”

  • CAR-T cellular therapy, while a dramatic scientific achievement at killing every last cancer cell, also cannot differentiate normal from cancer cells—the engineered T-cells will seek out and kill any cell expressing the target receptor (e.g., CD-19), effectively destroying the whole organ plus off-target cells, a limitation investigators rarely point out.

    “nowhere do investigators point out that these T-Cells also cannot differentiate between a normal cell and a cancer cell. So, what they actually do is kill the whole organ.”

0.43

Treating the whole person rather than just the disease pathology always offers hope for healing and improved outcomes.

3 pointscentrality 3/5
  • An average of two minutes of uninterrupted conversation must occur before a patient starts to express their real concerns, yet doctors are known to interrupt a patient every 18 seconds, and about 80% of physicians' time is now spent in front of computers versus 20% in face-to-face interaction—so doctors must learn to 'hear more hearingly.'

    “An average of two minutes of conversation has to occur uninterrupted before the patient will start to express... what is keeping them up at night... And, yet doctors are known to interrupt a patient every 18 seconds.”

  • Reading fiction is important for a caregiver because it lets one poach on the experience of others, stand in their shoes, and watch whole lives and the consequences of decisions unfold; declaring feelings for one character and not another gives insight into oneself, which helps in dealing with real human tragedies.

    “reading fiction is so important because it allows one to poach on experience of others. It allows one to stand in the shoes of others... this gives you insights into yourself.”

  • When we treat a disease we have a 50% chance of winning and 50% of losing, but if we treat the person we have a 100% chance of winning—because healing the person, not just curing the cancer, is always possible.

    “if we treat a disease, any time we treat a disease, we have a 50% chance of winning and 50% chance of losing. But, the way I have tried to look at it in the book is if we treat the person, then we have 100% chance of winning.”