This Week in Radiology — Aug 30, 2026

Welcome to This Week in Radiology. This week we're covering 10 notable papers spanning cardiac imaging as a risk-stratification tool, contrast safety and patient preparation, and the recurring problem of diagnostic uncertainty — from equivocal prostate MRI to emphysema that clinical CT simply cannot see. Let's dive in.

We start with cardiac imaging, where the biggest trial of the week tested whether a scar seen on cardiovascular magnetic resonance should drive device therapy. In JAMA, the CMR GUIDE trial, reported by Selvanayagam and colleagues, randomised 353 patients with ischaemic or non-ischaemic cardiomyopathy, an ejection fraction between 36 and 50 percent, and late gadolinium enhancement scar on magnetic resonance imaging, to either a primary prevention defibrillator or an implantable loop recorder, and followed them for a median of just over six years [1]. The primary composite of sudden cardiac death or haemodynamically significant ventricular arrhythmia was not reduced — about eight percent in the defibrillator arm versus nine percent with the loop recorder, a difference that was not statistically significant. So the headline is a negative trial: scar on magnetic resonance imaging in this ejection fraction band does not, on its own, justify extending defibrillator indications. The nuance is worth knowing, though. Sudden cardiac death alone was less frequent with a defibrillator — three patients versus ten — while haemodynamically significant arrhythmia events were numerically more common, and a prespecified subgroup interaction suggested benefit confined to patients under 70. All-cause and cardiovascular mortality and heart failure hospitalisation were similar. For the radiologist reporting these studies, scar quantification remains prognostically interesting but is not yet a device-selection biomarker.

A similar caution applies to the other cardiac imaging biomarker of the week. In the European Journal of Radiology, Isaka and colleagues pooled 21 studies covering more than 17,000 patients to assess pericoronary adipose tissue attenuation on coronary CT as a marker of perivascular inflammation [8]. Elevated attenuation was associated with roughly a 60 percent higher risk of major adverse cardiovascular events, a similar increase in all-cause mortality, and around a doubling of cardiac mortality. The signal was consistent in the left anterior descending and right coronary arteries but not statistically significant in the circumflex. Crucially, the strength of the association varied by measurement software, and the authors flag small-study effects and publication bias, concluding that standardisation and prospective validation are needed before this enters routine reporting. Two papers, two journals, one shared message: promising imaging biomarkers of cardiac risk are not yet ready to change management on their own.

Our second theme is contrast safety, where two European Journal of Radiology papers challenge long-standing departmental rituals. Choi and colleagues examined what happens when you shorten preparative fasting before contrast-enhanced CT, comparing a six-hour nil-by-mouth policy for solids and liquids in the first half of 2023 with a three-hour solids-only fast, no liquid restriction, in the same period of 2024 — a total of more than 242,000 examinations across nearly 200,000 patients [3]. Acute adverse reactions to iodinated contrast ran at just over one percent in both periods, and after weighting there was no increase in overall reactions, nor in hypersensitivity, physiologic, or emetic reactions specifically. Not a single case of aspiration pneumonia occurred. The authors are appropriately careful — this is an observational before-and-after design, so it supports relaxing fasting protocols rather than formally proving non-inferiority. But it is by far the largest real-world dataset on a practice that inconveniences enormous numbers of patients for no demonstrated benefit.

Alongside that, van der Molen and colleagues revisit corticosteroid and antihistamine premedication for patients with a prior hypersensitivity reaction [6]. Their argument is that the enthusiasm for premedication is inherited from the era of ionic high-osmolar agents. With modern non-ionic low-osmolar contrast, the evidence base is weak, efficacy against moderate-to-severe reactions is poor, breakthrough reactions occur and can themselves be severe, and costs rise. What works better is accurate documentation of the culprit agent and switching to a different contrast medium, ideally one with a different side chain, after allergological evaluation. They preserve one niche: emergency imaging of a patient with a prior severe reaction where the culprit agent cannot be identified. Taken together with the fasting data, the practical implication for department protocols is to stop doing the low-yield things and invest instead in documenting reaction history properly.

Our third theme is procedural and protocol innovation, and here the news is more affirming. In the American Journal of Neuroradiology, Grandhi and colleagues interrogated Medicare fee-for-service claims from 2018 through 2023 for chronic or subacute non-traumatic subdural haematoma, propensity matching just over 1,100 patients who had surgical evacuation alone against 577 who had evacuation plus concurrent middle meningeal artery embolization [2]. At 180 days, disease-related readmissions fell from about 16 percent to 10 percent, and reintervention fell from about 14 percent to just over seven percent — roughly a 40 to 50 percent relative reduction in each. Payments went the same direction: total all-cause payments were about four thousand dollars lower per patient with embolization, and for haematoma-related encounters the difference was closer to three and a half thousand dollars, driven by less acute-care hospital utilisation. This is registry data with all the attendant confounding caveats, but it is real-world corroboration of the randomised signal, and it makes the health-economic case as well as the clinical one.

Staying in neuroradiology, Peret and colleagues describe a feed-and-swaddle abbreviated magnetic resonance protocol, which they call Focused Abbreviated Survey Techniques, for infants with suspected Sturge-Weber syndrome [10]. The clinical problem is a real gap: early-onset seizures affect roughly three quarters to nine tenths of affected children and predict poor neurological outcome, yet consensus guidelines recommend screening magnetic resonance imaging only at one year of age, and conventional protocols are long enough to require sedation in infants. Their institutional experience is that a rapid disease-targeted protocol can detect leptomeningeal involvement without sedation, enabling earlier neurology referral, parental education about seizure recognition, low-dose aspirin, and identification of severe hemispheric disease for surgical planning. This is a clinical report rather than a comparative trial, so treat it as a protocol template rather than evidence of outcome benefit — but it is a model for how abbreviated, targeted imaging can move screening earlier.

Our final theme is diagnostic uncertainty, and four papers approach it from different organ systems. Two RadioGraphics reviews tackle interpretive ambiguity head-on. Nicola and colleagues address the PI-RADS 3 prostate lesion — equivocal by definition, and a driver of both overdiagnosis and missed significant cancer [4]. They identify what inflates the PI-RADS 3 rate: suboptimal image quality, post-treatment change, non-contrast protocols, and less-experienced readers. The remedies are unglamorous but actionable — structured reporting, strict application of the decision rules, image quality assessment programmes, patient preparation, and folding in ancillary biomarkers such as prostate-specific antigen density to make the biopsy decision risk-based rather than reflexive. Tan and colleagues, also in RadioGraphics, examine how magnetic resonance imaging for hepatocellular carcinoma diverges between East and West [5]. In East Asia, where hepatitis B predominates and ablation is favoured, guidelines prioritise sensitivity and broader magnetic resonance screening; in Western systems, where transplant allocation raises the stakes for specificity, criteria for definite hepatocellular carcinoma are stricter. Interestingly, treatment response assessment has converged globally, shifting from gross size reduction toward tumour viability — a change that matters particularly for immunotherapy, where size criteria underestimate benefit. If you read or contribute to multicentre international studies, these regional priors explain a lot of apparent disagreement.

Two further papers quantify what our current imaging misses or can predict. In European Radiology, Van der Rauwelaert and colleagues compared clinical CT low-attenuation area against automated micro-CT measurement of mean linear intercept in 89 patients spanning never-smokers to severe chronic obstructive pulmonary disease, with histological validation in a subset [7]. Micro-CT correlated strongly with histology, but among patients whose clinical CT showed no emphysema by the five percent low-attenuation threshold, close to a third OF PATIENTS had elevated mean linear intercept on micro-CT — underdiagnosed emphysema concentrated among ever-smokers, pre-chronic obstructive pulmonary disease, and mild disease. A reminder that a negative quantitative CT does not exclude early airspace destruction. And in the European Journal of Radiology, Sebastian and colleagues reviewed 101 arthroscopically confirmed bucket-handle meniscal tears; repair was performed in about two thirds OF PATIENTS [9]. Altered signal in the central flipped fragment raised the odds of a non-repair decision roughly fivefold, and extension to the posterior root roughly sevenfold, with each one-grade increase in central degeneration pushing the surgeon further toward meniscectomy and then debridement. Interobserver agreement was good, so this is a reproducible four-grade score you can add to a preoperative knee report.

If you only have time for one paper this week, make it the fasting study in the European Journal of Radiology [3]. Nearly a quarter of a million examinations is the strongest real-world evidence yet for a change every department can implement immediately, at zero cost, with a direct benefit to patient comfort and scheduling.

Here are the key takeaways from this week in Radiology. First, magnetic resonance scar in patients with an ejection fraction of 36 to 50 percent did not justify defibrillator implantation in a randomised trial, and pericoronary fat attenuation, while prognostic, remains too method-dependent for routine reporting. Second, shortening pre-contrast fasting to three hours for solids with free liquids was not associated with more acute contrast reactions, and premedication for prior hypersensitivity is best replaced by accurate documentation and switching the agent. Third, middle meningeal artery embolization added to evacuation for chronic subdural haematoma cut six-month readmissions and reinterventions by roughly 40 to 50 percent in Medicare data, and lowered costs. Fourth, abbreviated feed-and-swaddle magnetic resonance protocols can close the screening gap in suspected Sturge-Weber syndrome without sedation. And fifth, uncertainty is reducible: structured reporting and quality programmes shrink the PI-RADS 3 problem, a simple four-grade central degeneration score predicts meniscal repairability, and micro-CT reminds us that a normal quantitative CT does not rule out early emphysema.

That's your roundup for This Week in Radiology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And think of one colleague — in any specialty — who never has time to keep up with the literature. Tell them about AudioScholar: a free ten-minute weekly for every specialty, to listen to in any podcast app, or to read at audioscholar dot C C.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.