The tumor regrew from the inside out. Not a drug-sensitivity problem. A transport problem.
A nanoparticle chemotherapy shrinks a mouse tumor. Imaging looks promising. But section the tumor and stain for drug — and you find a band only along the outer rim. The core is untouched. The particles circulated. They leaked from tumor vessels. They accumulated. But they stalled at the vessel wall, unable to push through the dense, pressurized interior. Within weeks, the tumor regrows from the inside out. This is not a drug-sensitivity problem. It is a transport problem — built into the tumor's own architecture.
The same tyranny of distance that forced tumors to build blood vessels governs how far a drug can travel through tissue.
Start with the number that governs everything. A cluster of cells can survive on passive diffusion alone — oxygen drifting from a nearby vessel — only up to about one to two millimeters in diameter. Beyond that, the interior starves. Oxygen itself, molecular weight 32, can only diffuse about 100 to 200 micrometers from a vessel before it is fully consumed. A drug molecule is ten to one hundred times larger. A 100 nanometer nanoparticle is 100,000 times larger. The same physics that forced the tumor to recruit blood vessels governs how far a drug can travel through tumor tissue.
HIF-1α drives VEGF; VEGF sprouts capillaries. What grows instead of orderly vasculature is a mess — and that mess is now the drug delivery problem.
When a tumor recruits vessels, it hijacks the angiogenesis machinery. Hypoxia activates HIF-1-alpha, which drives expression of VEGF, which sprouts new capillaries. But the normal regulation — the balanced interplay of signals producing orderly, functional vasculature — is lost. Tumor vessels have chaotic branching, abrupt diameter changes, blind ends, and arteriovenous shunts. Blood flow is intermittent. Endothelial cells have gaps rather than tight junctions. Pericytes are sparse or absent. Lymphatic drainage is impaired. The tumor grew precisely by solving its transport problem inadequately — and that inadequacy is now the drug delivery problem.
Leakiness admits the particle. Then the pressure it created pushes it back toward the rim.
Here is the pivot on which the whole chapter turns. The leaky vessel walls are the basis of the EPR effect — enhanced permeability and retention — first described by Matsumura and Maeda in 1986. Particles in the 10 to 200 nanometer range leak out through endothelial gaps, and then stay, because the impaired lymphatics cannot drain them out. EPR is the physical basis for the entire field of nanoparticle drug delivery. But the same features that produce EPR also produce the delivery problem. Leaky vessels pour fluid into the tumor interstitium faster than it can drain. Pressure builds — elevated interstitial fluid pressure — and that elevated pressure means net fluid flow is outward, from the pressurized core toward the lower-pressure periphery. Leakiness admits the particle and then the pressure it created pushes it back toward the rim.
Rim-dominant delivery is not a measurement artifact — it is physics. Total accumulation can look good while spatial distribution is disastrous.
Interstitial fluid pressure in tumors is typically five to ten times higher than in normal tissue. The IFP at the tumor center can reach 30 to 40 millimeters of mercury; the periphery, where the tumor meets normal tissue with functional lymphatics, is much lower. Fluid moves down pressure gradients — so net interstitial fluid flow is outward, centrifugal, from the high-pressure core to the lower-pressure periphery. A particle trying to diffuse inward is fighting this current. For large particles, which diffuse slowly through collagen matrix, the outward convective push dominates and they barely move. This is the physical explanation for rim-dominant delivery. The whole-organ measurement can look good while the spatial distribution is disastrous. Accumulation at the rim is real accumulation — it simply does not reach the cells that matter.
The rim cells killed by the drug were not the cells most likely to generate recurrence. The core cells were — and hypoxic selection had spent months making them harder to kill.
The poorly perfused, drug-inaccessible tumor core is also hypoxic — and this compounds the transport problem in two ways. Radiation therapy requires oxygen for most of its cell-killing effect. Hypoxic cells are roughly three times more radioresistant than well-oxygenated cells. So the region hardest to reach by drug delivery is simultaneously the region most resistant to radiation. More important for understanding recurrence: hypoxia selects for aggressive cells. HIF-1-alpha drives expression of genes for glycolytic metabolism, invasion, and survival under stress. A tumor growing under hypoxic pressure selects, generation by generation, for cells that tolerate oxygen deprivation and evade apoptosis. The cells in the unreached core are biologically the most dangerous cells in the mass — and exactly the cells the drug never reached.
EPR is real and was the rationale for decades of development. Whether it is large enough in human tumors to be clinically meaningful is a more complicated question.
The EPR effect was named in 1986, and for three decades it was the central rationale for nanoparticle drug delivery. The revisionist view, which hardened around 2016 when Wilhelm and colleagues compiled delivery efficiency across hundreds of nanoparticle studies, is sobering: the median delivery efficiency to tumors — the fraction of the administered dose that actually reaches the tumor — is approximately 0.7 percent. Ninety-nine percent of administered nanoparticles go elsewhere, primarily to liver and spleen. This does not mean nanoparticles do not work. It means EPR is a real but modest effect, highly variable across tumor types and patients, and that the barriers described in this chapter account for most of the gap between the clean in-vitro killing curve and the disappointing in-vivo result.
Still open: noninvasive IFP measurement, vascular normalization efficacy, and what fraction of human tumors have co-opted vasculature where EPR barely operates.
What is not puzzling is the physics. Pressure pushes fluid outward. Dense matrix slows diffusion. Hypoxic cells selected for survival cluster in the unreachable core. The tumor built itself in a way that protects its most dangerous inhabitants from the drugs sent to kill them. Three unsolved problems remain honest ones: we cannot yet measure tumor IFP noninvasively in patients. We do not know what fraction of human tumors grow by co-opting existing vasculature rather than building new leaky vessels — for those tumors, EPR barely operates and passive nanoparticle accumulation may be the wrong strategy entirely. And vascular normalization — improving tumor vessel function to lower IFP and improve delivery — remains compelling in principle and inconsistently demonstrated in practice.
Cancer Nanomedicine · Chapter 2 · Tumor Transport Barriers
That is the frame for everything ahead in this series. A nanoparticle can be perfectly engineered — right size, right surface, right payload — and still fail because of physics it cannot control. Outward pressure. Dense matrix. A hypoxic core that selected for the most dangerous cells over months. Learn the transport barriers first. Every strategy you encounter in nanomedicine — EPR exploitation, active targeting, matrix degradation, vascular normalization — is an answer to one specific barrier. Knowing which barrier you are addressing is the whole game.