This was a single-center study conducted at Nanjing Drum Tower Hospital. The fetuses were recruited based on an ongoing prospective cohort study, which focused on the performance of a detailed first-trimester ultrasound (FTU) scan in detecting fetal structural and chromosomal anomalies (18, 19).

The inclusion criteria were as follows: (1) singletons with abnormal cardiac findings detected in the first trimester between January 2018 and June 2020; (2) pregnancy was terminated before 18 gestational weeks; (3) parents agreed to undergo pm-MRI and conventional autopsy (available in fetuses over 13⁺⁶ gestational weeks).

This study was approved by the Local Ethics Committee (2013058) and all the parents signed written informed consent.

The Statistical Product and Service Solution (version 26.0 for Macintosh, SPSS, IBM Incorporation, Armonk, New York, USA) was used for data analysis. Continuous variables are reported as mean ± SD or median (range) depending on the data distribution. Categorical variables were expressed in percentages and frequencies.

One fetus terminated for orofacial cleft without cardiac anomalies at 13+1 gestational weeks was recruited for MRI exploratory scan and its cardiac structure MRI was shown in Figure 2. It showed consecutive MRI of fetal cardiac structure (four-chamber view of atria and ventricles, crux and atrioventricular valves; three-vessel view of the pulmonary artery, aorta, and superior vena cava; other structures such as branches of pulmonary veins, left/right bronchus, and right subclavian artery). Video corresponding to these images is available as Movie I in the Data Supplement.

All the 19 cases underwent pm-MRI examinations successfully and 8 of 19 cases (42.1%) terminated over 13⁺⁶ weeks were subjected to autopsies.

In 73.7% (14/19) cases (case 3–11, 13–14, 16, 18, 19), the pm-MRI diagnosis was consistent with the prenatal ultrasound scan in the first trimester and a representative case of hypoplastic left heart syndrome was shown in Figure 3.

The pm-MRI findings were inconsistent with prenatal ultrasound scan in 15.8% (3/19) cases including 2 cases with autopsy assessment: (a) in case 15 (Table 2), the fetus was found severe tricuspid regurgitation but failed to yield a precise diagnosis prenatally. The fetus was terminated at 17 gestational weeks due to Trisomy 21 diagnosed by CVS. The pm-MRI revealed an AVSD clearly and the autopsy confirmed the diagnosis (Figure 4); (b) in Case 17 (Table 2), the fetus was detected as transposition of the great arteries in FTU scan and was terminated at 17⁺⁶ weeks of gestation for trisomy 18. The pm-MRI showed both the aortic artery and pulmonary artery completely arose from the right ventricle and the autopsy confirmed the diagnosis of double-outlet right ventricle (Figure 5). (c) in Case 1 (Table 2), pm-MRI examination of the heart revealed malalignment ventricular septal defect instead of AVSD detected by FTU (Figure 6).

In 10.5% (2/19) cases, pm-MRI provided additional information: (a) in Case 2 (Table 2), the fetus was diagnosed as AVSD and suspected of great vessels malformation prenatally. After TOP, AVSD, and Tetralogy of Fallot were observed using pm-MRI (Figure 7); (b) in Case 12 (Table 2), the hypoplastic left heart was observed and coarctation of aorta was suspected in a prenatal ultrasound scan. The fetus was terminated at 12 gestational weeks and was diagnosed with Turner syndrome by CMA postnatally. Pm-MRI showed that the aorta diameter was significantly smaller than the pulmonary artery (0.37 vs. 0.65 mm) on the three-vessel view and the left ventricle was smaller than the right ventricle on the four-chamber view (Figure 8). Thus, the diagnosis of aortic coarctation was raised by pm-MRI.

This study showed that it was clear to visualize fetal heart structure in detail using 9.4-T pm-MRI with an optimized scanning protocol. Pm-MRI had concordant findings in 73.7% (14/19) cases, discordant findings in 15.8% (3/19) cases, and additional findings in 10.5% (2/19) cases when compared with prenatal ultrasound. Pm-MRI findings were concordant with autopsy in all the 8 CHD cases terminated over 13⁺⁶ weeks.

The implementation of a detailed FTU scan has fundamentally changed prenatal care by moving the detection of major structural abnormalities, including cardiac abnormalities, to the early stage of gestation. A study by Syngelaki A showed that FTU scan in 11–13⁺⁶ weeks could diagnose all cases of tricuspid or pulmonary atresia, more than 90% cases of hypoplastic left heart syndrome, and AVSD (4). Our previous prospective study also suggested the vast majority of major CHDs (83.3%) can be detected during detailed FTU performed by experienced sonographers (18, 19). However, there is a longstanding desire for a postmortem reassessment of ultrasound diagnosis, especially in certain forms of CHDs terminated in the early stage of gestation. This study focused on postmortem assessment of fetuses with CHD diagnosed in the first trimester. In this study, we established a cohort based on a detailed FTU scan, which enables us to investigate a variety of CHD cases detected in the first trimester. Also, we developed the optimized protocol for pm-MRI examination of fetal hearts with TOP ranging from 12⁺⁶ to 17⁺⁶ gestational weeks. Studies have reported different methods of pre-scanning preparation before pm-MRI. A study by Votino C showed that postmortem fetal tissues were frozen at −20°C before an MRI scan could be beneficial for cardiac imaging (12). A study by Staicu A indicated that using 10% formalin solution before MRI scanning is beneficial for MRI imaging (6). In order to further improve the resolution of pm-MRI, we took out the visceral organs including intact fetal heart and lungs and then fixed them within 10% formalin for more than 2 weeks before the MRI scan. Though this could compromise the integrity of the fetus, it seems to be beneficial for improving the resolution of pm-MRI with presenting precise cardiac anatomy because of the complete immersion with formalin (Figure 2). Furthermore, we conducted exploratory scans using three fetal hearts in different gestational weeks terminated for extracardiac anomalies only. To obtain an optimal resolution, we found setting different parameters according to different fetal sizes before 18 weeks was essential. Thus, we conducted the MRI scan for CHD fetuses in different parameters according to their gestational weeks within practicable scanning time (<20 min each case) (shown in Table 1).

The diagnosis of CHD may be missed or wrongly interpreted in the antenatal scan, especially in the early gestational stage (20, 21). Zidere and colleagues demonstrated that their expertise in first-trimester diagnosis in continuing pregnancies scanned at least 6 weeks later had a false-positive diagnosis of CHD in 7/81 (9%) cases (22). Therefore, it was crucial to perform a postmortem evaluation of those terminated for CHD in the early stage. This study reported a false-positive case with AVSD diagnosis prenatally and pm-MRI revealed features of malalignment VSD. On the other hand, around 10% of fetuses terminated for aneuploidies below 14 weeks were non-diagnostic structural anomalies on the prenatal ultrasound (23). This study showed that 9.4-T pm-MRI could further reveal precise CHD diagnosis (AVSD) with clear images in the fetus terminated for Trisomy 21 with only tricuspid regurgitation on FTU (Case 15 in Table 2). The comparatively ideal performance of high-field pm-MRI for small fetal hearts in this study is encouraging, especially for those who were unavailable for conventional autopsy due to size restrictions.

In addition, it is impossible to always give a precise diagnosis of CHD in the early stage of pregnancy by prenatal ultrasound (24). For those suspicious of severe CHD without a precise diagnosis prenatally, the results of this study indicated that 9.4-T pm-MRI could provide the sound morphological basis for assessment of cardiac malformation terminated in the early stage. In Cases 2 and 12 (Table 2), more diagnostic information was provided by pm-MRI, and, therefore, a supplementation to the diagnosis was made.

Pathologic results of 11 cases in this study were not available owing to the technical limitations of the postmortem examination though pm-MRI provided us clear images of the cardiac structure of these cases. Considering the study is confined to a single center and the limited number of CHD cases, it requires more validation studies in more centers.