New research demonstrates that a magnetic resonance imaging (MRI)-based pancreatic risk stratification model (M-PRiSM) offers enhanced prognostic insight into postoperative survival for patients being treated for pancreatic ductal adenocarcinoma (PDAC).
For the retrospective multicenter study, recently published in Insights into Imaging, researchers evaluated the use of the M-PRiSM model for predicting overall survival (OS) in 540 patients with PDAC. The overall cohort was comprised of 282 patients in the development cohort, 122 patients in the internal validation group and two external validation groups with 80 patients and 56 patients, according to the study. The study authors noted that the M-PRiSM model incorporated MRI assessments of tumor margin, tumor size and the venous enhancement ratio (VER) as well as the CA19-9 blood test.
In external validation testing, the researchers found that the M-PRiSM model offered an average mean C-index of 69 percent for predicting postoperative OS in contrast to 53.5 percent for the 8th American Joint Committee on Cancer (AJCC) staging system and 56.5 percent for CA19-9 testing alone.
In comparison to 8th AJCC staging and CA19-9 alone, the M-PRiSM model also provided a higher average mean one-year AUC for predicting postoperative OS for the external validation cohorts (78.5 percent vs. 55.5 percent and 60 percent respectively. The study authors also noted a higher average mean two-year AUC for the M-PRiSM model (69.5 percent) vs. 8th AJCC staging (58 percent) and CA19-9 (60 percent).
“ … The M-PRiSM model, incorporating CA19-9, tumor margin, tumor size, and VER, offers superior predictive performance for postoperative OS in PDAC compared to existing models and offers robust pathological interpretability,” wrote lead study author Haitao Sun, M.D., who is associated with the Department of Radiology at Zhongshan Hospital, Fudan University and the Shanghai Institute of Medical Imaging in Shanghai, China, and colleagues.
Three Key Takeaways
- Enhanced prognostic accuracy. The MRI-based pancreatic risk stratification model (M-PRiSM) outperformed AJCC staging and CA19-9 alone, providing higher predictive accuracy for postoperative overall survival in PDAC.
- Key role of VER in prognosis. The M-PRiSM model leverages the venous enhancement ratio (VER) alongside tumor size, margin definition, and CA19-9, with VER providing a simple, quantitative measure of tumor contrast uptake that enhances prognostic accuracy and remains practical for routine clinical use.
- Therapeutic implications. High-risk patients identified by M-PRiSM showed significant survival benefit with adjuvant therapy, suggesting its potential utility for guiding precision treatment decisions in PDAC.
While computed tomography (CT) has been the primary imaging focus with other clinical/radiological models for predicting postoperative PDAC recurrence, the study authors maintained that “MRI can capture a broader spectrum of tumor characteristics associated with PDAC prognosis.”
In addition to the independent prognostic factors of tumor size and ill-defined tumor margins on MRI, the researchers said VER provides enhanced accuracy in assessing tumor contrast uptake and a practical alternative to evaluation of venous phase signal intensity.
“ … This simple quantitative metric has shown generalizability and operability in our multicenter validation cohorts, potentially mitigating the impact of different imaging equipment, scanning parameters, and patient variations. Moreover, compared with radiomics and deep learning-based models, which often rely on high-dimensional features and complex algorithms, this score is easier to interpret, simpler to implement, and more adaptable to routine clinical workflows, enhancing its practical applicability,” pointed out Sun and colleagues.
For patients identified as high-risk with the M-PRiSM model, the researchers noted that administration of adjuvant therapies was associated with longer median OS in both external validation cohorts (849 days vs. 484 days in the one group and 827 days vs. 489 days in the other cohort).
“These findings suggest that M-PRiSM-based risk stratification could serve as a valuable tool for guiding adjuvant therapy decisions and precision care in PDAC,” added Sun and colleagues.
(Editor’s note: For related content, see “Can Diffusion-Weighted MRI Have an Impact in Detecting Recurrent Pancreatic Cancer?,” “CT-Based Risk Scoring Outperforms AJCC TNM Staging for Predicting Recurrence of Pancreatic Ductal Adenocarcinoma” and “Predicting Diabetes on CT Scans: What New Research Reveals with Pancreatic Imaging Biomarkers.”)
In regard to study limitations, the authors acknowledged variability in MRI protocols, limited follow-up and small cohorts for external validation.
