A new single-center retrospective study published in the European Journal of Nuclear Medicine and Molecular Imaging aims to change that. Researchers at Beijing Tsinghua Changgung Hospital analyzed 290 consecutive patients with unresectable hepatocellular carcinoma who underwent yttrium-90 selective internal radiation therapy between October 2022 and December 2025. Among them, 91 were flagged on pre-treatment imaging as potentially receiving tumor blood flow from the inferior phrenic artery, and 78 ultimately had angiographically confirmed supply from that vessel. The team then compared the two main ways interventional radiologists handle the artery: threading microspheres directly through it into the tumor, or deliberately plugging it with coils so that tumor blood flow is redirected back into the liver’s own arterial network.

The technical foundation of the study lies in meticulous pre-procedural mapping. Every patient underwent hepatic digital subtraction angiography combined with cone-beam computed tomography, a pairing that allows physicians to see not just the two-dimensional course of the vessels but the full three-dimensional territory each artery perfuses. A technetium-99m macroaggregated albumin injection followed by SPECT/CT simulation calculated the lung shunt fraction, ensuring that too many microspheres would not escape to the lungs, and post-treatment PET/CT verified where the radioactive spheres actually landed. Using a 2.1 French goose-neck microcatheter, the team achieved superselective catheterization of the inferior phrenic artery in 88.6 percent of attempted cases, demonstrating that even this small, tortuous vessel can be reliably accessed when combined angiographic and tomographic guidance is used.

The decision-making algorithm the researchers propose hinges on two variables: the volume of tumor fed by the inferior phrenic artery and the anatomy of any non-target branches that could carry microspheres into dangerous territory. When the artery perfused at least 20 milliliters of tumor and risky branches could be managed, the team favored direct infusion of yttrium-90 microspheres through the vessel, reserving prophylactic coil embolization for high-risk offshoots such as suprarenal, esophageal, pericardial, or pulmonary shunt branches. When the perfused tumor volume fell below 20 milliliters or non-target vessels could not be safely catheterized, the team simply embolized the artery trunk with coils, wagering that intrahepatic arteries would compensate. Forty-five patients landed in the infusion group and 33 in the embolization group, with the infusion cohort predictably carrying heavier tumor burdens and more advanced disease.

The anatomical findings reinforce how distinctive the inferior phrenic artery can be. The right inferior phrenic artery dominated as the parasitic feeding vessel in 89.8 percent of cases, with the left artery accounting for only 7.7 percent and dual supply appearing in 2.5 percent. Most of these arteries originated from the celiac trunk or directly from the abdominal aorta, though a striking 21.8 percent arose from the renal artery. The median radioactivity delivered through the inferior phrenic artery was a modest 0.3 gigabecquerels, yet the dosimetric payoff was substantial: the median tumor-absorbed dose in the infusion group reached 92 Gy, with three-quarters of those patients receiving at least 61 Gy, approaching the 100 to 250 Gy window widely recommended for hepatocellular carcinoma.

Perhaps the most eye-catching result concerns the embolization strategy, which has always rested on an assumption that had never been carefully quantified. If you plug the parasitic artery, does tumor blood flow really reroute to intrahepatic vessels in time to matter? By comparing intra-procedural cone-beam CT with post-treatment PET/CT, the researchers found that approximately 88 percent of the territory originally supplied by the inferior phrenic artery received yttrium-90 microspheres through compensatory intrahepatic pathways, with more than half of lesions achieving complete hemodynamic compensation. Critically, the compensatory rate held steady regardless of tumor size, suggesting that flow redistribution after coil embolization is a robust phenomenon rather than a fortunate accident, and providing quantitative reassurance comparable to the 83.8 percent redistribution rate previously reported with glue embolization.

On efficacy, direct infusion through the inferior phrenic artery consistently outperformed embolization. Using the modified Response Evaluation Criteria in Solid Tumors, the objective response rate in the infusion group climbed from 60.0 percent at one month to 82.3 percent at three months, whereas the embolization group moved from 48.5 percent to 69.7 percent over the same intervals. The gap between the groups widened to 12.6 percentage points by three months, and the complete response rate in the infusion group surged from 17.8 percent to 46.7 percent, an absolute gain of 28.9 percentage points that far outpaced the 21.2 percent rise seen in the embolization cohort. Disease control rates exceeded 93 percent at one month in both groups, underscoring that neither strategy leaves patients behind.

Safety is where lingering fears about extrahepatic radiation have historically been most vocal, and the study directly confronts them. Mean whole-lung absorbed dose was 9.53 Gy, comfortably below the 20 Gy clinical threshold, with no significant differences between groups in lung volume, lung shunt fraction, or lung dose. The overall adverse event rate was 17.8 percent in the infusion group and 9.1 percent in the embolization group, a difference that was not statistically significant, and every reported event was a mild to moderate, reversible thoracic complaint. No severe treatment-related complications occurred in either arm, an outcome the authors attribute to the discipline of combined DSA-CBCT anatomical evaluation paired with prophylactic embolization of non-target branches whenever they were visualized.

The study’s authors are candid about its limits. As a single-center retrospective analysis, it carries inherent selection bias, and the infusion group’s higher tumor burden and greater proportion of Barcelona Clinic Liver Cancer stage C patients may confound the efficacy comparison. Only three-month tumor response was assessed, leaving progression-free survival and overall survival unmeasured, and the team calls for long-term follow-up and multicenter prospective validation before the algorithm becomes standard practice. Even so, the work represents the largest head-to-head comparison of inferior phrenic artery strategies in yttrium-90 radioembolization to date, and it delivers something the field has lacked: a concrete, volume-based decision rule. Direct infusion when the artery feeds at least 20 milliliters of tumor and the plumbing cooperates; embolization when it does not. For the roughly one in three hepatocellular carcinoma patients whose tumors recruit this diaphragmatic artery, that clarity could mean the difference between an incomplete treatment and a truly definitive one.

Subject of Research: Inferior phrenic artery management strategies during yttrium-90 selective internal radiation therapy for hepatocellular carcinoma.

Article Title: Interventional strategies for inferior phrenic artery management in yttrium‑90 selective internal radiation therapy for hepatocellular carcinoma: A single‑center retrospective study

Article References: Liao, Y., Huang, X., Liang, Z., Ma, J., Feng, X., Fu, W., Liang, B., Yang, S., Li, G., Bao, L., Zhang, T., Zheng, L., Liu, C., Tang, M., Zhang, L., & Dong, J. (2026). Interventional strategies for inferior phrenic artery management in yttrium‑90 selective internal radiation therapy for hepatocellular carcinoma: A single‑center retrospective study. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08170-0

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08170-0

Keywords: yttrium-90, selective internal radiation therapy, hepatocellular carcinoma, inferior phrenic artery, radioembolization, interventional radiology, cone-beam computed tomography, lung shunt fraction, mRECIST, extrahepatic collateral supply, tumor response, coil embolization

The research, conducted at a single German center and published in CVIR Oncology, followed 13 consecutive patients who underwent percutaneous CT-guided cryoablation of histologically confirmed desmoid tumors between January 2019 and December 2024. Most were young women with tumors embedded in the rectus abdominis muscle of the abdominal wall, the classic setting for this rare disease, which strikes roughly two to four people per million each year. Twelve of the thirteen had already failed at least one line of systemic therapy, ranging from nonsteroidal anti-inflammatory drugs and hormonal agents to chemotherapy and, in one case, a gamma-secretase inhibitor.

Cryoablation destroys tissue by cycling it between extreme cold and thaw. In this study, interventional radiologists placed between 2 and 16 thin cryoprobes directly into each tumor under CT fluoroscopic guidance while the patient was under general anesthesia. A standardized double freeze-thaw protocol followed: two ten-minute freeze cycles separated by passive and active thawing phases. During each freeze, the growing ice ball is visible on CT as a sharply defined dark region, a unique advantage of cryoablation over heat-based ablation techniques, allowing operators to watch the lethal zone expand in real time and confirm it covers the tumor with at least a five-millimeter margin.

The critical technical challenge lies in what surrounds these tumors. Because most desmoids in this cohort sat in the abdominal wall, the bowel loops immediately behind the target were the structures most at risk of catastrophic freeze injury. The team deployed protective dissection techniques in 76.9 percent of procedures: hydrodissection with 5 percent dextrose solution in just over half the cases, CO₂ pneumodissection in two patients, and air dissection in one. In laparoscopically assisted cases, surgeons created capnoperitoneum and directly observed the ice ball from inside the peritoneal cavity, retracting probes by pre-planned distances between freeze cycles to sculpt an enlarged zone of destruction. Every patient’s skin was protected with injected fluid and a warm saline-filled glove.

What distinguishes this study from earlier cryoablation series is its measurement strategy. Rather than tracking overall tumor diameter with standard RECIST criteria, the investigators systematically quantified the enhancing tumor volume on contrast-enhanced MRI, using the ellipsoid formula to calculate viable, cellular tissue separately from the non-enhancing fibrotic residue that persists long after successful ablation. This distinction proved dramatic. At a median follow-up of 14.5 months, median viable tumor volume collapsed from 70.5 cubic centimeters to just 1.97 cubic centimeters, a 98.3 percent reduction with a large statistical effect size. Yet the median entire lesion volume remained 6.7 cubic centimeters, and in the most striking individual case a patient retained a 251-cubic-centimeter residual mass of which less than 2 percent was living tissue.

That discrepancy carries a direct clinical warning. Conventional size-based response criteria, which measure the whole lesion including dead fibrotic tissue, would classify many of these successfully treated patients as having only stable or partial disease, badly underestimating how thoroughly the tumor had been devitalized. Applying modified RECIST criteria, which evaluate only the enhancing component, the authors found six of twelve evaluable patients achieved complete response and six achieved partial response, for an objective response rate of 100 percent. No patient showed progression during follow-up.

Objective MRI signal analysis corroborated the volumetric findings. The T2 contrast-to-noise ratio, a quantitative measure of how brightly tumor tissue stands out from background noise on T2-weighted images, fell from a median of 146.9 at baseline to 19.5 at last follow-up, reflecting replacement of highly cellular tumor by avascular fibrosis and coagulative necrosis. T2 signal intensity relative to muscle also dropped significantly. These quantitative markers align with prior work showing that intermediate T2 signal combined with nodularity and enhancement predicts post-ablation growth, offering clinicians measurable indicators of durable local control.

Patients themselves reported substantial benefit: twelve of thirteen described improved symptoms. The safety record, however, was not flawless. One patient, notably the only one treated without any prior therapy, suffered a severe grade 3b complication after the protective dissection was discontinued at cryoprobe retrieval before the ice ball had fully thawed. The residual ice contacted adjacent bowel and skin, causing bowel wall necrosis, cutaneous frostbite, and an enterocutaneous fistula requiring surgical repair; the patient was lost to follow-up. The remaining twelve patients experienced no complications at all, and the overall 7.7 percent major complication rate sits within the 0 to 17 percent range reported across published desmoid cryoablation series.

The results compare favorably with landmark studies. The prospective CRYODESMO-01 trial reported an 86 percent nonprogression rate at twelve months and a 79 percent objective response rate, while a ten-year pediatric and young adult series documented symptom improvement in 90 percent of patients but tumor volume reductions exceeding 75 percent in only 43 percent. The authors caution that their exceptional figures likely reflect selection bias and the small sample size, and that a median follow-up of 14.5 months cannot capture recurrences appearing years later.

Nevertheless, the study delivers a clear message for the multidisciplinary management of desmoid tumors. Cryoablation, positioned as a second- or third-line option after systemic therapy fails, achieves near-complete elimination of viable tumor with meaningful symptom relief and objective, quantitative MRI evidence of tumor death. The practical implication for radiologists and oncologists alike is to differentiate viable from total lesion volume when interpreting post-ablation imaging: large areas of residual non-enhancing tissue should never be mistaken for treatment failure. Larger prospective trials with patient-reported outcomes and extended follow-up are now needed to cement enhancing volume-based assessment as the standard endpoint for this increasingly credible, minimally invasive treatment.

The biology underlying desmoid tumors helps explain why local destruction can be so effective despite the tumors’ infiltrative habits. Most sporadic cases harbor activating mutations in the CTNNB1 gene, which stabilizes beta-catenin and drives the Wnt signaling pathway that fuels fibroblast proliferation. A minority arise in patients with familial adenomatous polyposis, where germline APC mutations set the stage. Because these tumors rarely metastasize, controlling the local disease is essentially controlling the whole disease, which is why a technique that devitalizes the entire enhancing component can translate into durable clinical benefit.

The demographic pattern of the disease also deserves emphasis. Desmoid tumors show a striking predilection for young women, and pregnancy and the postpartum period are recognized risk factors, with estrogen thought to modulate tumor behavior in at least a subset of patients. This means many patients face decades of life ahead of them at diagnosis, making the avoidance of radical surgery with its high recurrence rates and functional morbidity particularly consequential. An abdominal wall tumor in a woman of childbearing age, for example, can compromise core strength, posture, and future pregnancies if treated with wide resection, whereas a percutaneous needle-based approach preserves the muscle envelope even if some fibrotic residue remains.

The shift toward active surveillance as a first step deserves further context. Longitudinal observational cohorts have shown that a substantial fraction of desmoid tumors, in some series approaching half, either regress spontaneously or remain stable without any intervention. This observation, formalized in consensus guidelines from the Desmoid Tumor Working Group, fundamentally changed the treatment algorithm: intervention, whether systemic therapy, radiation, surgery, or ablation, is now reserved for tumors that progress on surveillance or cause significant symptoms. The patients in the cryoablation study fit squarely into this modern paradigm, having been selected precisely because their disease demanded treatment after systemic options were exhausted or declined.

On the mechanistic side, cryoablation’s lethality depends on achieving sufficiently low temperatures throughout the target volume. Vascularized tissue dies reliably when cooled to roughly minus twenty to minus forty degrees Celsius, with ice crystal formation disrupting cell membranes, vascular stasis producing ischemia, and the slow thaw phase amplifying the injury. The double freeze-thaw protocol used here exploits this last phenomenon: the second cycle kills cells already sensitized by the first, and the freeze-thaw alternation damages the microvasculature more thoroughly than a single prolonged freeze. This is why the authors insisted on an ablation margin of at least five millimeters beyond the enhancing tumor boundary, since the lethal isotherm sits inside the visible edge of the ice ball.

The choice of 5 percent dextrose in water for hydrodissection is also deliberate rather than arbitrary. Isotonic saline conducts electrical current and would be hazardous with radiofrequency ablation, but for cryoablation the more relevant property is thermal: fluid placed between tumor and bowel acts as a physical and thermal buffer, pushing heat-rich structures out of the freeze zone. Dextrose solution also avoids the theoretical concern of saline’s higher thermal conductivity accelerating ice growth toward protected structures.

Finally, the emergence of gamma-secretase inhibitors, exemplified by nirogacestat’s regulatory approval based on randomized trial evidence, has added a systemic option that did not exist for most of the study period. How ablation and these targeted agents will be sequenced remains an open question, and head-to-head comparisons incorporating volumetric endpoints, patient-reported symptom scores, and cost-effectiveness analysis will be needed to define each therapy’s proper place in a disease whose natural history is as variable as its treatment options are expanding.

Subject of Research: Percutaneous CT-guided cryoablation of desmoid tumors with volumetric MRI outcome assessment

Article Title: Percutaneous CT-guided cryoablation of desmoid tumors: single-center experience with volumetric midterm outcome

Article References: Schaaf, M., Sattler, T., Spatz, J., Surwald, S., Schraut, J., & Jakobs, T. F. (2026). Percutaneous CT-guided cryoablation of desmoid tumors: single-center experience with volumetric midterm outcome. CVIR Oncology, 2(1), Article 22. https://doi.org/10.1007/s44343-026-00060-4

Image Credits: AI Generated

DOI: 10.1007/s44343-026-00060-4

Keywords: desmoid tumor, cryoablation, aggressive fibromatosis, interventional radiology, CT-guided ablation, volumetric assessment, mRECIST, MRI, hydrodissection, soft tissue tumor, minimally invasive, local recurrence