MRI‐based nomogram estimates the risk of recurrence of primary nonmetastatic pancreatic neuroendocrine tumors after curative resection

Background

Accurate estimation of the recurrence of pancreatic neuroendocrine tumors help with prognosis, guide follow‐up, and avoid futile treatments.

Purpose

To investigate whether MRI features could preoperatively estimate the recurrence of pancreatic neuroendocrine tumors (PNETs) and to refine a novel prognostic model through developing a nomogram incorporating various MRI features.

Study Type

Retrospective.

Population

In all, 81 patients with clinicopathologically confirmed nonmetastatic PNETs.

Field Strength/Sequences

1.5 T MR, including T₁‐weighted, T₂‐weighted, and diffusion‐weighted imaging sequences.

Assessment

Qualitative and quantitative MRI features of PNET were assessed by three experienced radiologists.

Statistical Tests

Uni‐ and multivariable analyses for recurrence‐free survival (RFS) were evaluated using a Cox proportional hazards model. The MRI‐based nomogram was then designed based on multivariable logistic analysis in our study and the performance of the nomogram was validated according to C‐index, calibration, and decision curve analyses.

Results

MRI features, including tumor size (hazard ratio [HR]: 14.131; P = 0.034), enhancement pattern (HR: 21.821, P = 0.032), and the apparent diffusion coefficient (ADC) values (HR: 0.055, P = 0.038) were significant independent predictors of RFS at multivariable analysis. The performance of the nomogram incorporating various MRI features (with a C‐index of 0.910) was improved compared with that based on tumor size, enhancement pattern, and ADC alone (with C‐index values of 0.672, 0.851, and 0.809, respectively). The calibration curve of the nomogram exhibited perfect consistency between estimation and observation at 0.5, 1, and 2 years after surgery. The decision curve showed that a nomogram incorporating three features had more favorable clinical predictive usefulness than any single feature.

Data Conclusion

MRI features can be considered effective recurrence predictors for PNETs after surgery. The preliminary nomogram incorporating various MRI features could assess the risk of recurrence in PNETs and may be used to optimize individual treatment strategies.

Level of Evidence 4

Technical Efficacy Stage 2

J. Magn. Reson. Imaging 2018.

from #Head and Neck by Sfakianakis via simeraentaxei on Inoreader http://bit.ly/2VdICcT