In the article “Intraindividual Comparison of 18F-PSMA-1007 PET/CT, Multiparametric MRI, and Radical Prostatectomy Specimens in Patients with Primary Prostate Cancer: A Retrospective, Proof-of-Concept Study” by Kesch et al. (J Nucl Med. 2017; 58:1805–1810), the reported sensitivities, specificities, and positive and negative predictive values for mpMRI and the 18F-PSMA-1007 PET/CT combination are not correct due to a systematic calculation error. The correct values appear in italics in the paragraphs as well as the table below. Despite these errors, the main conclusions have not changed. The authors regret the error.
In the abstract:
18F-PSMA-1007 PET/CT had an NPV of 68% and accuracy of 75%, and mpMRI had an NPV of 63% and accuracy of 73% for total agreement. Near-total agreement analysis resulted in an NPV of 83% and an accuracy of 93% for 18F-PSMA-1007 PET/CT and 91% and 87% for mpMRI, respectively.
In the “Results” section:
MpMRI had 60% sensitivity, 88% specificity, 86% PPV, 64% NPV, and accuracy of 73% for total agreement and 85% sensitivity, 91% specificity, 92% PPV, 83% NPV, and accuracy of 87% for near-total agreement. The combination of 18F-PSMA-1007-PET/CT and mpMRI had 85% sensitivity, 75% specificity, 81% PPV, 81% NPV, and accuracy of 81% for total agreement and 99% sensitivity, 85% specificity, 89% PPV, 99% NPV, and accuracy of 93% for near-total agreement (Table 2).
Subsequently, we stated in the “Discussion” section on page 1809, “Compared with mpMRI, 18F-PSMA-1007 … had a worse sensitivity and NPV for total agreement.” However, 18F-PSMA-1007 had a worse specificity for total agreement.
- © 2019 by the Society of Nuclear Medicine and Molecular Imaging.
