KEY POINTS
- The prospective PRIDE study enrolled patients across four tertiary centres in the Netherlands. The analysis included 141 patients treated with neoadjuvant chemoradiotherapy using 41.4 Gy in 23 fractions with weekly carboplatin and paclitaxel, followed by surgery.
- Pathological complete response was recorded in 33 patients, while 108 were classified as non-pCR. Imaging with 18F-FDG PET/CT, diffusion-weighted MRI, and dynamic contrast-enhanced MRI was performed before, during, and after treatment.
- The strongest univariable predictor was the change in total lesion glycolysis during treatment, with an AUC of 0.68. Changes in post-treatment SUVmax and intratreatment ADCmean achieved AUCs of 0.63 and 0.62, respectively.
- Among 101 patients with complete multimodal imaging, the combined model using intratreatment and post-treatment TLG achieved a cross-validated AUC of 0.65 (95% CI 0.48–0.79), with 83% sensitivity and 59% specificity in the full dataset.
- Dynamic contrast-enhanced MRI parameters did not significantly predict pCR, and combining PET and MRI features did not produce clinically useful discrimination.
CLINICAL TAKEAWAY
Quantitative FDG-PET, diffusion-weighted MRI, and dynamic contrast-enhanced MRI should not be used alone to select patients for organ preservation or treatment adaptation after neoadjuvant chemoradiotherapy. This is high-relevance negative evidence from a prospective multicentre cohort, although incomplete imaging, complete-case modelling, failure to reach the planned sample size, and the absence of an independent holdout validation cohort remain important limitations.