Disparities Between Prenatal Ultrasound and Autopsy Findings in Pregnancies Resulting in Fetal Loss

This retrospective study aimed to assess disparities between prenatal ultrasound and autopsy findings in pregnancies that resulted in fetal loss, and to evaluate the diagnostic performance of prenatal ultrasound using postmortem examinations as a gold standard.

specific congenital malformations.With a wide variety of congenital abnormalities of body structure or function, these specific rates are of great significance regarding pregnancy outcomes and clinical implications in the future, thus, understanding the strengths and limitations of ultrasound modality is imperative for both parents and physicians.
Our study aims to underscore disparities between prenatal ultrasound findings and postmortem observations.We employed postmortem examinations as the gold standard for confirming prenatal results to assess the agreement between the modalities and to evaluate the diagnostic performance of prenatal ultrasound across various organ systems and specific findings.

Materials and Methods
This retrospective study included all cases of fetal autopsy that were performed at our tertiary medical center between the years 2006 and 2013.All postmortem reports were retrospectively compared with their prenatal ultrasound findings, serving as a gold standard reference to examine the diagnostic value of prenatal ultrasound.
Patients attending the study were referred to our tertiary center following the detection of fetal abnormality.A diagnostic ultrasound examination was conducted in our prenatal diagnostic unit by boardcertified obstetricians and gynecologists who have received specialized training in prenatal imaging.Our healthcare professionals follow international guidelines and in accordance with the International Society of Ultrasound in Obstetrics and Gynecology guidelines, 9 ensuring the highest standard of care in this area.This included an anatomical scan and morphological assessment of the entire fetus and the suspected malformation, using Voluson E6 and E8 ultrasound machines (GE Medical Systems, Kretz Ultrasound, Zipf, Austria), equipped with a transabdominal multifrequency 4-8 MHz probe or a transvaginal multifrequency 6-12 MHz probe. 9Complementary tests such as fetal magnetic resonance imaging (MRI), genetic counseling, karyotyping, chromosomal microarray, and biochemical and viral analyses were also performed when indicated.
An autopsy was offered in case of a diagnosed severe fetal malformation leading to either an elective termination of pregnancy (TOP) or intrauterine fetal death.After obtaining parental informed consent for both the postmortem examination and the study, all fetuses underwent a postmortem examination at our pathological institute.Our specialized pathologists, each with a fetopathology subspecialty, collaboratively conduct examinations, thereby ensuring the quality and reliability of our autopsy findings.A conventional autopsy was performed in our center as it remains the gold standard method to confirm fetal abnormalities. 10Inclusion criteria were fully detailed autopsy reports of singleton and twin pregnancies following TOP or intrauterine fetal death after 16 weeks of gestation and a detailed prenatal ultrasound scan with pathological findings.Additional information, such as demographic and medical details, obstetric history, prenatal care, and gestational age at diagnosis and at a fetal loss were also recorded in our database.Only anomalies that were considered visible by both ultrasound and autopsy, such as anatomic features of the fetus, were recorded in our data (Table 1).The primary outcome was to evaluate the diagnostic performance of prenatal ultrasound at the level of both organ systems and specific findings.Thus, we used autopsy reports as a reference to calculate the sensitivity, specificity, FP and FN rates of prenatal ultrasound in detecting various anomalies.
A secondary outcome was to evaluate the agreement between prenatal ultrasound and autopsy examinations.All cases were classified into organ systems based on the most severe malformation detected in the last ultrasound examination as follows: central nervous system (CNS), cardiovascular, pulmonary, gastrointestinal (GI), genitourinary (GU), musculoskeletal (MS), facial anomalies, and multiple anomalies which refer to cases with several severe anomalies.In each case, a comparison between ultrasound and autopsy findings was made by correlating the degree of the agreement into 4 groups based on a modified method described by Isaksen et al 11 : 1) Complete agreement between autopsy and ultrasound findings; 2) agreement with minor differences; 3) agreement with major differences; and 4) complete disagreement.A major finding was considered as a malformation that is imperative for the subsequent diagnosis but could not have altered the management of a pregnancy.A minor finding was considered as one that would not pose a significant health problem and have no consequences for the management.
The study protocol has gained the approval of the institutional review board regional (1943-15-SMC), and all patients gave informed consent.
SPSS software (IBM SPSS Statistics for Windows, version 23, IBM Corp., Armonk, NY, USA, 2015) and NCSS software (NCSS 2021, Kaysville, UT, USA, 2021) were used for statistical analysis.Continuous variables were evaluated for normal distribution and reported as mean (standard deviation) or median (interquartile range).Sensitivity, specificity, FP and FN rates were calculated to assess the diagnostic efficacy of prenatal ultrasound examination.

Results
A total of 136 cases of autopsies with a detected fetal malformation were documented during the study period.Of them, 112 (82.4%) autopsies were performed after elective TOP and 24 (17.6%)following intrauterine fetal death.The median gestational age at the time of diagnosis via ultrasound was 24.1 weeks (range 22.8-28.7),while the median gestational age at the time of fetal loss stood at 26.6 weeks (range 24-31.4;Table 1).
A total number of 375 congenital findings were detected at autopsy, compared to 309 in prenatal ultrasound.These were classified based on organ systems and presumed pathogenesis (such as agenesis, hypoplasia, or dysplasia) as depicted in Table 2.The CNS (16.3%, n = 61), facial (16.0%, n = 60), and cardiovascular (15.7%, n = 59) were the three most common organ systems detected in autopsy.Table 3 shows the diagnostic efficacy of prenatal ultrasound in detecting congenital findings compared to postmortem examination.At the level of organ systems, sensitivity was highest among malformations of the CNS, MS and cardiovascular systems (93.2%, 87.0%, and 84.4%, respectively).This was specifically noted in detecting midline and posterior fossa defects (93.3% and 88.9%, respectively), septal and vascular defects of the cardiac outflow (88.9% and 84.6%, respectively), and skeletal anomalies (90.0%).On the other hand, lower sensitivity rates were observed in ventriculomegaly of the lateral ventricles (77.8%), valvular and pericardial effusion (70.0% and 0.0%, respectively), and digital deformities (50.0%).Of all organ systems, sensitivity was lowest among the pulmonary, facial, and GI systems (13.0%, 32.3%, and 38.2% sensitivity, respectively).This was specifically noticeable in detecting pulmonary hypoplasia, and splenic and adrenal defects, as ultrasound had missed almost all autopsy findings (n = 21, 10, and 7, respectively).Additionally, despite being the second most common organ system detected in autopsy (n = 60), ultrasound showed low sensitivity in almost all facial malformations, regardless of their subtype.To better evaluate the prenatal ultrasound results, the different malformations were divided into three levels: low (<60% sensitivity), moderate (60%-80% sensitivity), and high (>80% sensitivity) (Table 4).
To provide a deeper understanding of the complexities in prenatal ultrasound diagnoses, we have chosen to describe some of the intriguing FP and FN cases in our study.In one case of terminated pregnancies, prenatal ultrasound suspected left-sided diaphragmatic hernia with the stomach below the diaphragm, along with   ascites, pericardial effusion, congested placenta, and increased flow in the liver and brain.Autopsy analysis revealed significant iron load in hepatocytes, splenomegaly, ascites, and subcutaneous edema.Table 5 demonstrates the distribution of the agreement between prenatal ultrasound and postmortem examination.The autopsy confirmed the primary prenatal diagnosis (categories 1-3) in 91.9% (n = 125) of the cases.General agreement (agreement with only minor discrepancies, category 1-2) was noted in 80.9% (n = 110) of the cases, with the highest rates in the CNS, cardiovascular and MS systems (85.7%, n = 36, 18, and 12, respectively) and a slightly lower rates in GI (80.0%, n = 4) and GU (78.6%, n = 11) systems.Autopsy added additional findings in 50.7% (n = 69) of the cases, with 11% additional major findings (n = 15, category 3) that were imperative for the subsequent diagnosis and 39.7% additional minor findings (n = 54, category 2).Cases related to CNS had the highest rate of 50% (n = 21) of additional missed minor malformations.Complete disagreement (category 4) was noted in 8.1% (n = 11) of the cases.Online supplemental Table 1 depicts in detail all cases of major differences or complete disagreement (categories 3 and 4).
In terms of agreement levels and fluid abnormalities, a notable pattern emerges, associating the prevalence of these abnormalities with observed agreement levels.In category 1 (complete agreement), 23.2% (n = 13/56) of the cases exhibited fluid abnormalities, while category 2 (agreement with minor differences) displayed such in 31.5% (n = 17/54).In Regarding the potential impact of fetal outcomes on agreement levels, our study incorporated 112 cases of TOP and 24 cases of intrauterine fetal demise (IUFD).For IUFD cases, 37.5% were classified in category 1, 25.0% in category 2, 16.7% in category 3, and 20.8% in category 4. Similarly, within the TOP group, 42.0% fell into category 1, 42.9% into category 2, 9.8% into category 3, and 5.4% into category 4.These findings suggest a higher prevalence of discrepancies among IUFD cases, while TOP cases exhibited enhanced agreement.

Discussion
Our primary objective was to shed light on the existing disparities between prenatal ultrasound findings and postmortem observations across various organ systems and specific anomalies.We aimed to distinctively enhance clinical decision-making by providing a comprehensive overview of ultrasound's strengths and limitations, emphasizing its role as the primary imaging modality in obstetric care.
Using autopsy reports, which serve as the gold standard for validating prenatal diagnoses, 12,13 our study examined the diagnostic value of prenatal ultrasound at the level of both organ systems and specific malformations.Our results indicate that the sensitivity of ultrasound differed dramatically between the organ systems, with the highest rates in detecting malformations of the CNS, cardiovascular and MS systems and the lowest in facial and pulmonary systems.Regarding CNS malformations, the high sensitivity and specificity rates together with the relatively low FP and FN rates indicate that prenatal ultrasound is a good rule-in and rule-out test for evaluating major CNS anomalies.At the level of specific malformations, sensitivity was higher in detecting midline defects of the brain, cardiac septal defects, skeletal malformations, and renal agenesis and lower in detecting ventriculomegaly of the lateral ventricles, valvular defects, pericardial and pleural effusion, pulmonary hypoplasia, splenic and adrenal defects, and renal hypoplasia or dysplasia.Facial malformations exhibited lower overall diagnostic rates.Some, like jaw and lip (33.3%), showed relatively higher sensitivities, while others, including nose (16.7%) and tongue and palate (12.5%), faced detection challenges.The relatively high specificities in these categories can be attributed to the complex nature of identifying facial malformations, or, in part, to the limited attention these structures receive during routine prenatal ultrasound scans, resulting in a higher number of negative results and a lower number of positive results.This, in turn, may contribute to a reduction in FP diagnoses and an increase in true negatives.Facial anomalies hold significant clinical importance within our study, as they represent one of the most prevalent malformation groups encountered.A total of 60 facial anomalies were identified, affecting 31 cases in our study.5][16] Remarkably, out of the 60 facial anomalies, only 10 were accurately identified in prenatal ultrasounds.The overall agreement within these 31 cases was moderate (categories 1-2), with 7 cases exhibiting substantial discrepancies (category 3).Although the underdetection of facial malformations did not lead to significant alterations in clinical management, the emphasis placed on diligently diagnosing these anomalies remains of paramount importance.This emphasis is rooted in the remarkable prevalence of facial anomalies within our high-risk population, where they often accompany major malformations as high-prevalence soft markers.Therefore, enhancing sensitivity in the detection of facial malformations serves as a crucial clinical goal with far-reaching implications for patient care and informed decision-making.
][6][7][8] These studies reported the highest sensitivity in detecting malformations of the CNS (sensitivity of 76%-100%) and cardiovascular system (sensitivity of 43%-100%), with similar lower rates in pulmonary and facial malformations (sensitivity of 9%-80% and 40%-60%, respectively).Even so, the corresponding rates vary widely in the literature due to the lack of a more unified classification system.
The alignment between primary sonographic diagnoses and autopsy findings was evident in 91.9% of cases (n = 125, categories 1-3), reflecting clinically accurate management, particularly in pregnancy termination scenarios at our medical center.In evaluating the disparities between ultrasound and postmortem outcomes, 11% (n = 15) cases displayed additional significant findings (category 3), and 8.1% (n = 11) experienced complete disagreements (category 4).Previous studies have demonstrated similar agreement rates of 83%-98.7%,8][19][20][21] We assume that these subtle distinctions can be largely attributed to our restrictive definition of complete agreement.
The increasing prevalence of fluid abnormalities from category 1 to 4, as detailed in the results, suggests a potential association between these abnormalities and the complexity of co-occurring malformations.3][24] An illustrative example of this influence involves a case with severe IUGR and an identified solitary left kidney during a prenatal ultrasound examination.Initially, the presence of oligohydramnios led to the hypothesis of uncompensated unilateral right kidney agenesis.Notably, postmortem analysis revealed the presence of bilateral small kidneys in this instance, shedding light on the potential for amniotic fluid discrepancies to confound ultrasound interpretations.This emphasizes the need for cautious interpretation, particularly in cases where amniotic fluid abnormalities might impact ultrasound accuracy.It underscores the importance of a comprehensive approach, including the use of MRI, for evaluating complex cases with such anomalies.
Given our experience with relatively high FPs in detecting dynamic anomalies like fluid accumulation, ventriculomegaly, and dynamic valvular malformation, it is essential to note that fluid accumulations detected via prenatal ultrasound (such as pleural and pericardial effusion and ascites) have the potential to shift during postmortem examinations, possibly resulting in FPs.This fluid redistribution is influenced by factors like the cessation of fetal circulation and cellular breakdown after death, resulting in changes in pressure gradients.Furthermore, handling, manipulation, and gravity during postmortem procedures contribute to these shifts.It is worth noting that disparities between ultrasound and postmortem findings do not always indicate genuine FPs; some reveal ultrasound's capacity to capture evolving post-fetal death conditions. 25This phenomenon is exemplified by the apparent postmortem resolution of prenatally detected ventriculomegaly, which is also subject to cerebrospinal fluid shifts after death. 26dditionally, prenatal ultrasound offers a dynamic modality that visualizes functional insights not always observable through postmortem examination, such as valvular insufficiencies, effusions, and ascites. 27This underscores the dynamic nature of fetal fluid distribution, influencing both prenatal and postmortem interpretations.Thus, it is crucial to emphasize that disparities between the modalities do not necessarily signify FPs but rather underscore an additional capability of ultrasound to capture anomalies at specific time points.This highlights the significance of considering the antemortem history or antenatal ultrasound examinations to discern these differences accurately.It is essential to emphasize that our intention is not to critique the ultrasound modality, which serves as the primary imaging modality in obstetric care.Instead, our goal is to stimulate a more profound discussion within the medical community, delving into the underlying factors contributing to the disparities.This future dialog holds can offer valuable clinical insights.We posit that certain disparities may be attributed to various factors, including the lack of inclusion of certain anomalies in the guidelines, variations in sonographer focus, the influence of normal variations and familial factors, as well as the natural course of organogenesis and fetal development.
The main limitation of the study is that our study includes relativity small cases in several rare malformations (ie, tracheoesophageal fistula), which were insufficient for accurate results.Nevertheless, our study examines the effectiveness of prenatal ultrasound to identify its strengths and limitations across various types of specific malformations and organ systems.By doing so, our study provides comprehensive insights regarding its applicability and improves clinical decision-making.
Most of the known data regarding the diagnostic values of prenatal ultrasound in specific malformations come from pediatric reports and clinical outcomes of unselected populations.Many studies have evaluated concordance between prenatal ultrasound and autopsy at the level of organ systems, including a recent systematic review of 3534 fetuses conducted to review correlation in 7 different organ systems. 21The main strength of our study is that, to the best of our knowledge, this is the first study to draw conclusions about the diagnostic values of prenatal ultrasound, not only across broader organ systems but also by delving deeply into specific malformations, thus setting our study apart from previous works.We believe that this approach distinctively enhances clinical decisionmaking by providing a comprehensive overview of ultrasound's strengths and limitations across different anomalies and organ systems.This data will allow both physicians and parents to achieve a reasonable degree of certainty regarding clinical decision-making when evaluating pregnancy outcomes.The second strength of our study is represented by the wide variety of congenital malformations that were referred to our hospital from regional clinics or hospitals, as our hospital serves as a large tertiary hospital and is a center of excellence in radiology, offering obstetrical and fetopathological services to the entire state.

Conclusions
This study provides important insight regarding the disparities between prenatal ultrasound and postmortem examinations.Our findings support the overall high diagnostic performance of prenatal ultrasound in detecting congenital malformations, while contributing to our comprehensive understanding of its strengths and limitations across various types of organ systems and specific malformations.

Table 1 .
Clinical Characteristics of the Study Population (n = 136) PM, postmortem; GA, gestational age; TOP, termination of pregnancy; IUFD, intra-uterine fetal death.a Mean AE standard deviation.b Median (interquartile range).Sade et al-Disparities Between Prenatal Ultrasound to Autopsy

Table 2 .
Classification of the Detected Findings in Ultrasound and Postmortem Examinations CCAM, congenital cystic adenomatoid malformation; IUGR, intrauterine growth restriction.

Table 3 .
Diagnostic Performance of Prenatal Ultrasound in Detecting Congenital Findings Compared to Postmortem Examination tem examinations confirmed CMV infection and identified splenomegaly, which was initially missed.In another case, prenatal ultrasound suggested potential lymphatic, endocrine, or metabolic disorder due to

Table 4 .
Sensitivity of Prenatal Ultrasound in Detecting Specific Malformations

Table 5 .
Categorization of Agreement Between Prenatal Ultrasound and Postmortem Examination Findings