Radiology

Serial deep-learning CT response score predicts survival in immunotherapy-treated advanced NSCLC

In 1830 patients with advanced NSCLC on immune checkpoint inhibitors, a fully automated Serial CT Response Score (Serial CTRS) predicted overall survival from baseline and 12-week CTs. Higher Serial CTRS probabilities of 12-month survival were strongly associated with longer overall survival, independent of age, sex, PD-L1, histology, and tumor volume. Serial CTRS outperformed RECIST and simple tumor volume change for overall-survival risk discrimination in multi-institutional routine cohorts and an independent phase 1 trial. Predictive value persisted across PD-L1 strata and RECIST response categories, including stable disease. These data support Serial CTRS as an early imaging biomarker to stratify immunotherapy benefit and potentially guide treatment adaptation. 1

Updated ESGAR consensus refines MRI-based primary staging in rectal cancer

A 26-expert ESGAR panel used a RAND-UCLA approach to reach 96% consensus on MRI acquisition, interpretation, and reporting for primary rectal cancer staging. The guideline adopts the sigmoid take-off as the landmark separating rectal from sigmoid cancers, standardizing case inclusion and reporting. Mesorectal fascia involvement is now defined as tumor, irregular nodes, or EMVI within 1 mm, tightening circumferential resection margin risk assessment. Nodal staging moves to a patient-level cN assessment with updated lateral nodal criteria, including a 7 mm size threshold. Diffusion-weighted imaging is recommended only for limited roles in baseline staging, emphasizing optimized T2-weighted sequences and structured reports aligned with modern treatment pathways. 2

Multimodal CT reduces thrombolysis use without improving outcomes in early-window stroke

In the PRACTISE randomized trial, 271 thrombolysis-eligible patients within 4.5 hours were assigned to non-contrast CT alone or CT plus CTA and CTP. Among patients without non-contrast CT contraindications, multimodal CT led to fewer thrombolysis treatments than CT alone, with an adjusted odds ratio of 0.46. Door-to-decision and door-to-needle times, early neurological improvement, and 90-day functional outcomes were similar between imaging strategies. Symptomatic intracerebral hemorrhage occurred only in the CT-alone group, while mortality was numerically lower with multimodal CT but not statistically significant. These findings suggest multimodal CT can safely reduce thrombolysis in selected patients without delaying treatment or altering short-term outcomes. 3