For more than two decades, I have practiced emergency medicine. The job trains you to recognize patterns quickly because sometimes the pattern is the difference between sending someone home and recognizing a catastrophe that is still assembling itself.
A heart attack arrives after years of vascular disease and suddenly becomes an emergency. A stroke may represent decades of biology revealed in minutes. Sepsis can turn a functioning immune system into something profoundly dysregulated over the course of hours. We stabilize the patient, treat the immediate problem, and move on to the next emergency. Over time, that began to bother me.
I realized that much of medicine operates this way even outside the emergency department. The cardiologist treats plaque after it has accumulated. The oncologist treats cancer after surveillance has failed. The neurologist treats cognitive decline after years of biology have already unfolded. The specialty changes, but the posture often does not: We become involved after the downstream consequence is visible.
Emergency medicine had also taught me another principle: source control. If a septic patient has an infected gallbladder, an obstructed kidney, or an abscess, antibiotics and fluids matter enormously, but they may not be enough if the source remains. We do not congratulate ourselves because the fever came down while leaving the abscess undrained. We keep asking what is driving the process.
Real physician voices, twice a week
Free, and one click to unsubscribe.
That question followed me out of the emergency department. What if some of the diseases we classify separately are downstream expressions of a shared upstream state? What if aging, cardiovascular disease, metabolic disease, impaired recovery, chronic infection, and cancer vulnerability overlap because the biological environment beneath them overlaps?
That was pattern recognition. And pattern recognition is where a hypothesis begins, not where it ends.
The question eventually led me to dendritic cells, particularly the conventional type 1 dendritic cell, or cDC1. Dendritic cells occupy an unusual position in immunity. T cells kill. B cells make antibodies. Macrophages engulf debris. Dendritic cells tell those other cells what the biological situation means.
They are instructors.
The cDC1 is especially important for directing cellular immune surveillance. One of its defining functions is the ability to produce IL-12p70 at the appropriate time. That signal helps license cytotoxic T cells and natural killer cells to respond to abnormal, infected, or malignant cells.
The model I began assembling asks whether chronic inflammatory signaling can progressively impair that instructional state. In the proposed chain, persistent inflammatory pressure drives sustained STAT3 activity and epigenetic repression around IRF8, a transcription factor essential to cDC1 identity and function. As that system loses competence, inducible IL-12p70 output falls and downstream surveillance becomes less effective.
That could matter far beyond what we usually call immune aging. The same failure could plausibly affect clearance of damaged and senescent cells, control of latent infection, tumor surveillance, vascular inflammation, tissue repair, and recovery from physiological stress. In that view, aging, chronic disease, and cancer are not identical diseases. They are different clinical surfaces that may share part of the same upstream biology.
Something else from everyday medicine made me take the idea more seriously. Nearly every emergency physician looks at RDW and albumin without giving either much thought unless something is obviously abnormal. Divide red-cell distribution width by albumin and you get the RDW-to-albumin ratio, or RAR.
RAR is remarkably nonspecific. It has been associated with outcomes in severe infection, cardiovascular disease, cancer, chronic disease, and population mortality.
Normally we treat nonspecificity as a weakness. I began wondering whether, in this case, it might be the clue.
RDW changes slowly because red cells circulate for months. Albumin changes on a different but still relatively slow time scale. If a patient arrives tonight critically ill with an already elevated RDW, tonight’s illness did not create that red-cell population. The patient brought that biology through the door.
The pneumonia is new. The ground it landed on is not.
That is why I have come to think of RAR not as another aging clock, but as a possible pressure gauge. An aging clock tries to tell us how far the body has traveled. A pressure gauge asks how much biological pressure is being applied now.
The distinction matters. Distance already traveled cannot be undone. Pressure can potentially be reduced.
I am not suggesting that RAR diagnoses immune aging, cancer, or any particular chronic disease. It does not. The more interesting possibility is that it may be a cheap clinical readout of the systemic inflammatory and physiological environment shared by several of them. If one ratio tracks reserve in the emergency department, systemic burden in cancer, and chronic inflammatory pressure in metabolic and vascular disease, perhaps it is not measuring the individual disease at all. Perhaps it is partly measuring the patient in whom the disease is occurring.
That remains a hypothesis. The next step is not to admire how neatly the pieces fit together. It is to try to break the model.
The central experiment is straightforward. Isolate cDC1 cells from younger and older humans across a range of biological rather than merely chronological aging. Measure IRF8 regulatory state and stimulate those cells under controlled conditions. Then measure inducible IL-12p70 competence.
If the proposed mechanism is correct, inflammatory burden should track loss of the IRF8-high, IL-12-competent state better than birthdays do. If it does not, the model needs to change. If manipulating the predicted upstream pathway fails to restore function, that is even more important evidence against it.
RAR creates another falsifiable test. If it really is reporting pressure on this shared system rather than simply serving as a generic illness score, changes in RAR should track relevant immune and inflammatory biology prospectively, and RDW should change more slowly than faster inflammatory markers as the underlying field improves. Those questions can be tested in datasets that already exist.
This project has changed the way I think about physician scholarship. Clinical medicine rewards pattern recognition because we must act before every uncertainty is resolved. Science demands something more. Observation is not mechanism. Mechanism is not causation. And a coherent explanation is not automatically a correct one.
Frontline physicians do have one unusual advantage. We see the downstream consequences of many specialties in the same room. Heart attack, stroke, infection, cancer, metabolic disease, and frailty stop belonging to separate departments when they arrive through the same emergency-department doors. That vantage point can generate useful questions.
But the moment we move from recognizing a pattern to proposing why it exists, the standard changes. We have to state what would prove us wrong.
After twenty years of treating what broke, I find myself increasingly interested in what was happening before it broke. In sepsis, we call that looking for the source.
Perhaps chronic disease deserves the same question.
Andrew Caravello is an emergency medicine physician, board-certified in emergency medicine and internal medicine, and the author of The Biological Reboot: The Dendritic Cell Theory of Aging. He is affiliated with Rowan-Virtua School of Osteopathic Medicine and Thomas Jefferson University Hospital.
His writing begins with a clinical question: Why do some people retain the capacity to recover while others lose it? His work brings together immunology, aging biology, oncology, metabolism, and epigenetics to examine the role of immune instruction in health and disease.
He is coauthor, with Andrew Blidy, of a 2026 hypothesis article in Frontiers in Immunology proposing a molecular chain that links senescence, cancer, and chronic infection to a correctable failure of dendritic cell instruction. The fuller scientific manuscript is available as a preprint.
Through the complete book, the Reader’s Edition, and a companion app, he makes complex biology accessible while distinguishing established findings from the framework’s proposed connections and questions that remain open. He shares updates on X.
