Tumors May “Frack” Under Pressure, Reshaping How Cancer Drugs Spread

When doctors inject a drug directly into a tumor, the goal seems straightforward: place a high concentration of medicine exactly where it is needed while limiting exposure to healthy tissue. But tumors are not passive targets. Their dense, irregular structure can redirect fluid, concentrate treatment in some regions, leave other regions untouched and create escape routes into surrounding tissue. A new study suggests that, at clinically relevant injection rates, drugs may move through tumors in a process strikingly similar to hydraulic fracturing—the industrial technique commonly known as fracking. Rather than spreading smoothly through the tumor’s existing microscopic pores, the injected fluid appears to build pressure until it opens fracture-like pathways through the tissue.

The finding could change how physicians interpret the pressure inside an injection needle during intratumoral therapy. In experiments using nine freshly resected canine lung tumors, researchers observed a characteristic pressure pattern as they pushed a radiopaque liquid into the tissue. Pressure initially rose in an approximately linear fashion, then reached a sharply defined peak before entering a transition phase and eventually settling into a plateau. That sequence resembles the pressure curve recorded during an extended leak-off test in petroleum engineering, when fluid pumped into rock first stretches the surrounding formation, then triggers cracks and finally spreads through a growing network of fractures. In the tumors, the pressure peak appears to signal a mechanical failure in the tissue and the sudden creation of a low-resistance route for the injected agent.

Intratumoral delivery is increasingly attractive because it can expose cancer cells to high local drug concentrations without requiring the entire body to tolerate an equivalent systemic dose. The approach is already used or being investigated for chemotherapy, immunotherapy and oncolytic viruses such as talimogene laherparepvec, or T-VEC, which is approved for certain cases of advanced melanoma. Yet direct injection has produced inconsistent results. Previous studies have documented leakage into nearby tissues, rapid loss of drug through blood vessels, and highly uneven concentrations within tumors. A single injection can leave some areas effectively untreated while creating potentially toxic pockets elsewhere. Understanding how fluid actually travels after the needle enters a tumor is therefore essential to making local therapy safer and more effective.

The conventional explanation assumes that injected drugs move by convection through the pre-existing porous matrix of tumor tissue, followed by diffusion after the injection stops. In porous-media physics, pressure-driven flow through tissue can be described using Darcy’s law: the flow depends on pressure, hydraulic conductivity, fluid viscosity and the geometry of the pores. But the researchers argue that this mechanism cannot explain what happens at ordinary clinical flow rates. Tumor tissue has low hydraulic conductivity, meaning that it resists fluid movement. At the tiny flow rates of microliters per minute used in some laboratory models, existing pores may indeed carry fluid. Clinical injections, however, are typically performed at milliliters per minute. Delivering a standard T-VEC volume of 0.1 to 4 milliliters through the slower porous-flow regime could take from roughly two hours to as long as three days, making it impractical for treatment.

To test the alternative, the team used controlled syringe pumps to inject radiopaque contrast agents into six resected tumors at rates of 0.5 or 5 milliliters per minute. The injected volumes represented 12.5, 25 or 50 percent of the estimated tumor volume. A miniature pressure sensor positioned inside the 18-gauge needle recorded pressure continuously during and briefly after each injection. The remaining three tumors were injected at 0.1 or 1 milliliter per minute while fluoroscopy provided a real-time, two-dimensional view of the contrast agent’s movement. After injection, six tumors underwent micro-computed tomography, allowing the researchers to reconstruct the three-dimensional distribution of the agent and estimate how much remained in the tumor rather than escaping into surrounding tissue.

The images revealed patterns that would be difficult to reconcile with smooth, radially symmetric flow through pores. Instead of forming roughly spherical clouds around the needle tip, the contrast agent often traveled along thin, elongated tracks and accumulated near tumor edges, sometimes far from the injection site. In one fluoroscopy sequence, a narrow path appeared beside the needle tip just after the pressure maximum, connecting to a region that began filling only after the peak had been reached. The timing suggested that the pressure drop did not result from gradual seepage through the tissue. It marked the sudden opening of a new pathway. Across the six tumors examined by micro-CT, the fraction of injected agent retained within the tumor varied dramatically, from 3.45 to 58.33 percent.

The researchers interpret the sharp pressure rise as elastic deformation of tissue surrounding the needle. As fluid enters faster than it can move away, the local tissue stretches and pressure climbs. Once a critical pressure is reached, the tissue becomes mechanically unstable. A crack may open, or a cavity may expand and tear into the surrounding material, allowing fluid to move rapidly through a newly created channel. The distinction between fracture and cavitation—a rapidly expanding void in a soft solid—is difficult to establish from the final shape alone. But the observed fluid distribution was thin, directional and anisotropic, rather than smoothly spherical. Its sudden appearance at the moment of pressure instability favored fracture-mediated transport, the authors say. The relatively large radius of an 18-gauge needle also places the injection conditions in a regime where fracture is expected to dominate across much of the relevant range of tissue stiffness.

After the pressure peak, the signals differed among tumors. In some, pressure fell before gradually approaching a plateau; in others, it briefly dropped and then rose again, suggesting that the first pathway may have connected to another compliant region that expanded until it reached its own failure threshold. A plateau can have several physical explanations, including stable fracture growth, continuous leakage from the tumor or extensive yielding of the tissue. Whatever the precise mechanism, the practical consequence is concerning: once a fracture reaches a blood vessel, lymphatic channel, needle tract or the tumor boundary, additional drug may leave the intended target. In the experiments, visible leakage occurred through several routes, including a large vessel, the needle insertion channel and a region of pre-existing necrosis. One tumor displayed almost immediate leakage because a necrotic area already extended to the tumor’s edge.

This pressure behavior could provide physicians with a simple real-time warning system. The investigators propose that injections should be paused after the first sharp pressure peak and before the signal reaches its eventual steady state, particularly when preventing leakage is more important than delivering the entire dose at one location. The remaining drug could then be administered at one or more different sites, using the same pressure-guided strategy. Such an approach might improve coverage while reducing the chance that a single fracture carries medicine out of the tumor. Earlier computational work by members of the team has suggested that dividing a cytotoxic dose among multiple injection sites can improve treatment effects compared with delivering the same amount centrally. It could also reduce the formation of high-concentration pockets that damage immune cells needed for an effective antitumor response.

Pressure monitoring has potential advantages over imaging-based guidance. Endobronchial ultrasound can help place needles and visualize injections, but it produces only two-dimensional information. Cone-beam computed tomography can locate radiopaque material but does not continuously track the injection, while fluoroscopy requires contrast agents and exposes patients to ionizing radiation. A pressure sensor inside the needle could operate continuously without adding imaging radiation and could be integrated into procedures already used to deliver drugs directly into tumors. The pressure curve might also help physicians tailor flow rate, needle size, fluid viscosity and injection location to the desired outcome, whether that is maximum retention, broad distribution or avoidance of a particular escape route.

The study remains an early mechanistic investigation rather than evidence that pressure-guided treatment is ready for routine cancer care. The experiments involved only nine resected canine lung tumors, which lacked active blood circulation, lymphatic drainage and the living mechanical environment of an intact tumor. Removing the tumors may have lowered interstitial pressure and changed their resistance to fracture, while differences in storage time could have altered tissue properties. The researchers also used two contrast agents with different viscosities, and they did not test the very low flow rates at which pore-based convection is expected to dominate. In a living tumor, active circulation could buffer pressure in some regions but also accelerate drug clearance after a fracture reaches the vasculature. Human studies will be needed to determine whether the same pressure signatures appear in vivo and whether acting on them improves drug retention, distribution and clinical outcomes. Even so, the work offers a vivid new picture of intratumoral therapy: at the speeds required in medicine, a tumor may behave less like a sponge quietly absorbing fluid and more like a stressed geological formation that suddenly splits open, turning a carefully placed injection into a rapidly expanding network of hidden channels.