European journal of nuclear medicine and molecular imaging
June 24, 2026 · via Europe PMC
Diagnostic accuracy of <sup>18</sup>F-FDG PET-CT scores in distinguishing polymyalgia rheumatica from other inflammatory rheumatic diseases: a multicentre retrospective study (RHUMATEP).
Quéré C, Devauchelle-Pensec V, Thuillier P, +22 more, Querellou S.
AI summary — for doctors, generated from the abstract only
FDG PET-CT scores such as Leuven/Groningen and the new Brest score distinguished polymyalgia rheumatica from other inflammatory rheumatic diseases, though accuracy dropped in external validation.
What was studied
Retrospective multicenter study of 232 patients (150 PMR, 50 axial SpA, 32 other) evaluating published PET-CT scores and a new Brest score in training and external validation cohorts against rheumatologist diagnosis.
Key findings
- Best scores (Leuven, Leuven/Groningen, Besancon, Brest) AUC 0.83-0.89 training vs 0.67-0.69 external
- Leuven/Groningen sensitivity/specificity 94.7%/67.9% (training) and 72.2%/55.2% (validation)
- Brest 96.4%/66% (training) and 86.1%/51.7% (validation)
- Heidelberg and Saint-Etienne algorithms: high sensitivity, low specificity
Clinical relevance
This suggests PET-CT scores may support PMR diagnosis in uncertain cases, but modest external specificity means they should complement clinical assessment.
Who should care
Limitations
Retrospective with substantial performance drop on external validation; reference diagnosis not blinded to PET-CT itself.
PET-CT patterns help but do not reliably clinch a diagnosis of polymyalgia rheumatica.
This is an AI-generated summary of the published abstract, not a clinical recommendation. It may not reflect the full text or subsequent updates — verify against the original source before acting on it.
Abstract
Purpose
To evaluate the diagnostic accuracy of 18F-FDG PET-CT-based scores to differentiate polymyalgia rheumatica (PMR) from other inflammatory rheumatic diseases.
Methods
This retrospective multicentre study included 232 patients with inflammatory rheumatic diseases (150 PMR, 50 axial spondylarthritis AxSpA and 32 other rheumatic diseases) divided into training and external validation cohorts. The gold standard final diagnoses were those that were upheld by an experienced rheumatologist, at least at 6 months' follow up, taking into account all available information (clinical data and evolution, radiological, biological, PET-CT) but blinded for the scoring system. 18F-FDG uptake at 29 anatomical sites was assessed using visual analysis and maximum standardized uptake value (SUVmax). Several scoring systems previously described in the literature were evaluated for PMR and AxSpA: Leuven and Leuven/Groningen, Besançon and Pean de Ponfilly scores; Heidelberg and Saint-Étienne algorithms for PMR; sacroiliac-to-sacrum ratio (SIJ/S) for AxSpA. A new Brest score was also developed based on the best performing sites (AUC ≥ 0.8) identified in the training cohort. Diagnostic performance was assessed for each score and algorithm.
Results
The highest diagnostic accuracy for PMR was observed with the Leuven, Leuven/Groningen, Besançon and Brest scores (AUC 0.83-0.89 in training and 0.67-0.69 in external validation cohorts). The Leuven/Groningen sensitivity and specificity were respectively 94.7%/67.9% (training) and 72.2%/55.2% (validation). The sensitivity of Pean de Ponfilly score was lower (78.9% (training) and 55.6% (validation)). The diagnostic performance of Brest was quite similar to Leuven scores (96.4%/66% in training; 86.1%/51.7% in validation). Heidelberg and Saint-Étienne algorithms demonstrated high sensitivity with low specificity.
Conclusion
Leuven/Groningen and Brest scores may represent promising tools for distinguishing PMR from other rheumatic diseases, given their ease of use and diagnostic performance.
DOI
10.1007/s00259-026-08013-y
PMID
42337036
Research metadata (titles, abstracts, authors, publication dates) is sourced from Europe PMC, which aggregates content from PubMed/MEDLINE, PubMed Central, and other providers under the U.S. National Library of Medicine (NLM/NCBI). NLM does not claim copyright on abstracts; the original journal publishers or authors may hold rights to the full articles. AI-generated summaries are derived only from the abstract, never the full text, and are always labeled as AI-generated.
Want more like this?
Browse more from the feed. A free account unlocks the full feed, saving and communities.
Back to the research feed