Objective: To determine the diagnostic value of virtual autopsy using post mortem-MRI (pm-MRI) at 3Tesla (T) compared to classic autopsy for the confirmation of fetal structural anomalies and secondly to establish which cases of termination of pregnancy would benefit mostly from a virtual autopsy.
Methods: In each of 32 fetuses included in the study, 32 anatomical structures were assessed, after termination of pregnancy in the second trimester. All cases were evaluated by prenatal ultrasonography, virtual autopsy and classic autopsy, and then divided into four groups: Cerebral Group, Cardiac Group, Renal Group and Other Group (miscellaneous group). The concordance of virtual autopsy with classic autopsy was calculated overall and for each group and each structural item. Also, the concordance between the two methods was assessed using a diagnostic error score (DgE_score), calculated as the absolute value of the difference between the number of malformations detected by classic autopsy per case (CA score) and the number of malformations detected at virtual autopsy per case (VA score).
Results: Overall virtual autopsy demonstrated a diagnostic sensitivity (Se) compared to classic autopsy of 67.33% [95% CI 57.28-76.33], with a specificity (Sp) of 98.37% [95% CI 97.33-99.09], a positive predictive value (PPV) of 81.93% [95% CI 71.95-89.52], a negative predictive value (NPV) of 96.49% [95% CI 95.11-97.57] achieving a diagnostic accuracy of 95.31% [95% CI 93.83-96.52]. Overall, no statistic significant correlation was demonstrated between DgE_score and the gestational age of the fetuses or between DgE_score and the weight of the fetuses, but a significant correlation was revealed between the virtual autopsy and classic autopsy score. The diagnostic utility of virtual autopsy using pm-MRI at 3 T as compared to classic autopsy for each category of termination of pregnancy revealed in the Cerebral Group a Se of 80.00% [95% CI 28.36-99.49], with a 96.30% [95% CI 81.03-99.91], a PPV of 80.00% [95% CI 35.75-96.64] a NPV of 96.30% [95% CI 81.81-99.34], with a diagnostic accuracy of 93.75% [95% CI 79.19% to 99.23] and a Cohen's Kappa coefficient of 0.76 [95% CI 0.4494-1.0765]; in the Renal Group a Se and Sp of 100%, but in the Cardiac Group the Se was only 60.00% [95% CI 26.24-87.84], Sp 75% [95% CI 34.91-96.81], the PPV 75.00% [95% CI 44.92-91.69], NPV 60% [95% CI 38.87-77.96], with a diagnostic accuracy of 66.67% [95% CI 40.99-86.66] and a Cohen's Kappa coefficient of 0.32 [95% CI -0.07-0.76].
Conclusions: The results support virtual autopsy using pm-MRI at 3T as a reliable alternative to classic autopsy for the non-forensic analysis of second trimester fetuses. Analyzing the diagnostic utility of virtual autopsy using pm-MRI at 3 T for the confirmation of prenatal ultrasound findings in second trimester fetuses as compared to classic autopsy, the best results were obtained in the Cerebral and Renal Group. Reserved results were found in the Cardiac Group. Therefore, for the pregnancies with termination of pregnancy for cerebral or renal abnormalities, virtual autopsy by pm-MRI at 3T can be taken into consideration as a first-line investigation to confirm the prenatal findings.