Predictive value of combined DCE-MRI perfusion parameters and clinical features nomogram for microsatellite instability in colorectal cancer

Abstract

Objectives: To develop a nomogram that combines dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) perfusion parameters, ADC values and clinical features to preoperatively identify microsatellite instability (MSI) in patients with colorectal cancer (CRC).

Methods: This retrospective study included 63 CRC patients who underwent preoperative DCE-MRI and had immunohistochemistry results available. Two radiologists, in a double-blind manner, placed two circular regions of interests in the area with the highest perfusion intensity on the DCE-MRI perfusion map and the corresponding area on the ADC map. Perfusion parameters and ADC values were measured, and the average values from both radiologists were used for subsequent analysis. Univariate analysis was performed to identify independent risk factors for MSI. A nomogram was then constructed by combining the most significant clinical risk factors with DCE-MRI perfusion parameters. The model's performance was evaluated using receiver operating characteristic (ROC) curves. Calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC) were used to assess the nomogram's clinical utility and net benefit.

Results: The nomogram prediction model, which combined PLT, LNM, K^trans, K_ep, iAUC, and ADC, demonstrated good predictive performance. The combined model had an AUC of 0.951 (95% CI: 0.903-0.998), an accuracy of 0.873, a sensitivity of 1.000, and a specificity of 0.818. Both the DCA and CIC demonstrated good clinical applicability and net benefit.

Conclusion: The nomogram method demonstrated good potential in the preoperative individualized identification of MSI status in CRC patients. This tool can assist clinicians in adopting appropriate treatment strategies and optimizing personalized stratification for CRC patients.

Keywords: Colorectal cancer; DCE; Magnetic resonance imaging; Microsatellite instability; Nomogram.

Fig. 1

Case Analysis of MSI and MSS Patients in Clinical Applications. A 73-year-old man with rectal cancer, immunohistochemical staining exhibited MSI (a–g), and a 53-year-old man with rectal cancer, immunohistochemical staining exhibited MSS (h–n); a and i ROI₁ in DCE-MRI perfusion maps. b and j ROI₂ in ADC maps; c–f immunohistochemical test results in the MSI patient [c, MLH1(−); d, MSH2(+); e, MSH6(+); f, PMS2(−)]and k–n immunohistochemical test results in the MSS patient [k, MLH1(+); l, MSH2(+); m, MSH6(+); n, PMS2(+)]; the total nomogram scores for MSI and MSS patients in g and h were 148 and 120, respectively, predicting MSI risk greater than 0.9 and less than 0.1, respectively

Fig. 2

The overall framework, imaging parameters, and the process of pathological assessment of this study. ADC apparent diffusion coefficient, DCE dynamic contrast-enhanced, IHC immunohistochemistry

Fig. 3

The violin plots illustrate the comparison of DCE-MRI parameters and ADC values between the MSI and MSS groups. MSS microsatellite stability, MSI microsatellite instability, ADC apparent diffusion coefficient

Fig. 4

Correlation analysis. A Correlation analysis between K_ep and ADC values; B correlation analysis between K^trans and iAUC; and C Heatmap of the correlation between the optimal clinical features, DCE-MRI parameters, and ADC values

Fig. 5

Nomogram based on DCE-MRI parameters, ADC values and clinical features

Fig. 6

Performance comparison of different models. A ROC curve of 10 predictive model performance comparison and B combination model calibration curve analysis

Fig. 7

Results of the decision analysis curve and clinical impact curve demonstrating excellent clinical application and net benefits of the combined model. A Comparison of decision analysis curves of three prediction models and B Clinical decision curve analysis of the combined model