|Home | About | Journals | Submit | Contact Us | Français|
To determine the sensitivity, specificity, and predictive values of prenatal ultrasound detection of fetal upper extremity anomalies at a single tertiary care center in a large patient cohort. Our secondary purpose was to assess factors affecting prenatal detection including the presence of associated anomalies.
We performed a retrospective review of prenatal ultrasound and postnatal clinical records from each pregnancy evaluated with a prenatal ultrasound at the *BLINDED* Washington University Department of Obstetrics and Gynecology over a 20-year period. We searched for upper extremity anomaly diagnosis codes pre- and postnatally and correlated with clinical postnatal follow up to determine prevalence, sensitivity, specificity, predictive values, and associated conditions.
A total of 100,856 pregnancies were evaluated by prenatal ultrasound, which included 843 fetuses diagnosed with a musculoskeletal anomaly (prevalence 1/120) and 642 with an upper extremity anomaly (prevalence 1/157). The postnatally confirmed sensitivity for prenatal ultrasound detection of an upper extremity anomaly was 42%. Sensitivity was lower in cases isolated to the upper extremity (25% vs 55%). Sensitivity was highest for conditions affecting the entire upper extremity (70–100%) and lowest for those affecting the digits alone (4–19%). Fetuses with limb reduction defects, radial longitudinal deficiency, phocomelia, arthrogryposis, abnormal hand positioning, and cleft hand had a higher likelihood of having an associated anomaly.
At our tertiary referral center, there was a notable prevalence of upper extremity anomalies, however the overall sensitivity for detecting them with prenatal ultrasound was low. This was disappointing given the value of prenatal identification of anomalies for parental counseling. Prenatal diagnosis of anomalies affecting the entire upper limb was more reliable than diagnosis of more distal anomalies.
III - Diagnostic
Prenatal detection and diagnosis of upper limb anomalies provides an opportunity for parental counseling and may prompt earlier diagnosis and treatment of associated conditions. Ultrasound remains the primary modality for fetal evaluation,  and detection of fetal anomalies has improved substantially over the last several decades as a result of technological advances, improved resolution, standardization of prenatal ultrasound protocols, and training of diagnosticians. [2–7] Recent studies have shown the sensitivity for detection of major fetal anomalies to be over seventy percent, although the detection of musculoskeletal anomalies remains lower, ranging from eighteen to forty percent. [6,8] Evaluation of the upper extremity, especially the small structures of the hand, can be difficult, with sensitivities between twenty and thirty percent. [2,7,9] Detection rates tend to be higher in tertiary care centers, for high-risk patients, and for fetuses with multiple anomalies. 
The aims of this study were to evaluate the performance of prenatal ultrasound detection of fetal upper limb defects at a single tertiary care center, to determine which upper limb defects were the most and least likely to be diagnosed prenatally, to understand relative frequency of upper extremity anomalies, and to determine if upper extremity anomalies with concomitant associated conditions had higher detection rates.
After approval by the human research protection office at *BLINDED*Washington University School of Medicine, we performed a retrospective review of all pregnancies evaluated with a prenatal ultrasound at the *BLINDED*Washington University Department of Obstetrics and Gynecology between January 1990 and January 2010. This is a tertiary referral center that provides care to both high and low risk pregnant women and performs both standard and specialized obstetric ultrasound examinations. Registered diagnostic medical sonographers with certification in obstetrics & gynecology performed the examinations, and maternal-fetal medicine specialists interpreted them. A dedicated nurse coordinator acquired postnatal clinical follow up data on 94% of births, allowing for validation of prenatal ultrasound findings and identification of anomalies not diagnosed by ultrasound (false negatives).
The Department of Obstetrics & Gynecology database includes all pregnancies with prenatal or postnatal diagnoses of any anomaly. This database contains indications for the ultrasound, prenatal ultrasound fetal diagnostic codes and clinical notes from each examination, number of ultrasounds per pregnancy, gestational age at each ultrasound, outcome of each pregnancy, postnatal diagnostic codes and clinical notes, and all associated diagnoses for each fetus.
The database was queried for pre- and postnatal institutional fetal anomaly diagnostic codes for upper extremity conditions including major and minor skeletal anomalies, limb reduction defects, abnormal hand positioning, arthrogryposis, polydactyly, syndactyly, and amniotic bands. Arthrogryposis was defined as fetal joint flexion or extension contractures with rigidity involving more than one large joint (hip, knee, ankle, elbow, or wrist). Abnormal hand positioning was defined as clenched hands or overlapping digits. Obstetricians use the diagnosis of abnormal hand positioning as a potential marker for the presence of non-orthopedic malformations rather than as an independent diagnosis potentially requiring treatment. Limb reduction defect was defined by shortening and/or absence of the bone(s) of one or more extremities. When possible, we supplemented the database with a review of the medical records at our institution to further refine and subcategorize diagnoses such as ulnar and radial longitudinal deficiency, cleft hand, symbrachydactyly, and phocomelia.
We identified each prenatal diagnosis as either a true positive (TP), false positive (FP), or a false negative (FN) after reviewing postnatal clinical findings. The true negatives group included the 100,013 live births during the study period that had a negative ultrasound for upper extremity anomalies and a post-natal exaimination with no coded diagnosis of an upper extremity anomaly. We then calculated prevalence, sensitivity (with confidence intervals), specificity, and predictive values for each diagnosis.
We categorized the indications for prenatal ultrasound into two groups. When patients were referred for prenatal ultrasound evaluation of a known or suspected fetal musculoskeletal malformation, they were labeled high-risk. Examinations performed for other indications were classified as other. Age at diagnosis was identified and categorized into either before or after 22 weeks of gestation, as beyond this age, pregnancy termination options are limited. We divided pregnancy outcomes into live birth (LB), stillbirth (SB), termination of pregnancy (TOP), and neonatal death (NND). Findings were classified as isolated if no other anomaly was present on postnatal clinical follow up or as associated with other systemic anomalies on postnatal clinical exaimination. These other systemic anomalies were categorized as central nervous system, facial, nuchal, cardiothoracic, abdominal, renal, aneuploidy, or specific syndrome diagnoses.
There were 100,856 pregnancies evaluated with an obstetric ultrasound at a gestational age greater than or equal to thirteen weeks during the study period. Of these, 350 (.3%) of the pregnancies were considered a high-risk pregnancy, the situation in which a musculoskeletal anomaly was suspected upon referral to our center, while 100,506 (99.7%) were referred to our institution for an indication for assessment other than a suspected musculoskeletal anomaly (low-risk). The low risk 100,013 fetuses had none of our screening anomalies present and resulted in live births, while 843 fetuses had a pre- or postnatal diagnosis of a musculoskeletal anomaly (prevalence of 1/120). There were 642 fetuses with a postnatally confirmed upper extremity anomaly, providing a prevalence of 1/157 in this population. The overall prevalence of each upper extremity anomaly in this population, as well as the prevalence in the high-risk and low risk pregnancy groups, is shown in Table 1.
Overall sensitivity for detection of a fetal upper extremity anomaly was 42% (confidence interval 38–46%). Table 2 shows the sensitivity (with confidence intervals), specificity and predictive values of prenatal ultrasound detection of each upper extremity anomaly diagnosis. Sensitivity of anomaly detection was highest for conditions affecting the entire upper extremity and lowest for those affecting the digits alone. Limb reduction defects were divided into limb versus digital anomalies, and a lower sensitivity was found for digit (24%) compared to the limb (76%). The highest rate of false positive results was in abnormal hand positioning, leading to a poorer positive predictive value (84%). Overall sensitivity was higher in cases with associated compared to isolated cases, and this was also true within each diagnosis.
Of the pregnancies with a diagnosed fetal musculoskeletal anomaly, 350 (42%) were considered high-risk. For 3 types of limb anomalies, more than fifty percent of pregnancies were high- risk: phocomelia (80%), amniotic constriction bands (59%), and abnormal hand positioning (53%). In all other diagnoses, less than fifty percent of fetuses were part of a high-risk pregnancy (the lowest being ulnar longitudinal deficiently at 11%), indicating that no anomaly was suspected at the first ultrasound at our center. In general, more fetuses with an associated diagnosis were a product of a high-risk pregnancy than fetuses with an isolated upper extremity anomaly (see Appendix).
Approximately 40% of fetuses with abnormal hand positioning, arthrogryposis, and phocomelia resulted in stillbirth or neonatal death compared to less than 20% for all other diagnoses. Pregnancies in which the fetus had multiple anomalies were terminated 8.5 times more often than those with an isolated upper extremity anomaly.
Over fifty percent of fetuses with a diagnosis of phocomelia (100%), abnormal hand positioning (89%), radial longitudinal deficiency (81%), arthrogryposis (67%), cleft hand (60%), and limb reduction defect (58%) had an associated anomaly confirmed on prenatal ultrasound, while in other diagnoses, associated anomalies were uncommon, including symbrachydactyly (20%), polydactyly (24%), ulnar longitudinal deficiency (33%), syndactyly (42%), and amniotic band syndrome (45%). Patients with upper extremity anomalies with associated conditions did not consistently have more ultrasound examinations performed than patients with isolated anomalies, nor were they consistently detected at an earlier gestational age. Instead these relationships varied with each diagnosis (see Appendix).
For each upper extremity diagnosis, associated conditions involving other organ systems are listed in Table 3, and associated syndromes and chromosomal abnormalities are listed in Table 4. Fetuses with radial longitudinal deficiency more commonly had multiple organ system involvement, hypoplastic thumb, and VACTERL. Fetuses with polydactyly and syndactyly had a large variety of other conditions involving other organ systems and syndromes. Fetuses with abnormal hand positioning more commonly had aneuploidy, especially trisomy 18, and facial anomalies. 60% of fetuses with cleft hand also had cleft foot. Ulnar longitudinal deficiency and symbrachydactyly had the lowest involvement of other organ systems.
The sensitivity of prenatal ultrasound for detection of musculoskeletal and limb anomalies has been low historically. In 1991, Levi published a series of 16,072 pregnant women with prenatal ultrasound and found a 40% sensitivity for detection of any type of anomaly, with a 17% sensitivity for detection of limb and skeletal anomalies.  In 1992, Stoll found a 15% sensitivity for isolated anomalies and 48% sensitivity for multiple anomalies for second trimester prenatal ultrasounds.  Detection of major anomalies has improved over the last two decades as a result of improvements in technology and technique, although detection of limb anomalies remains low. The Eurofetus study, in 1999, showed an overall sensitivity for detection of any anomaly to be 61%, with identification of musculoskeletal anomalies much lower and similar to the findings of Levi, et al at 18%. 
More recent studies have shown improved detection of upper extremity anomalies, with more accurate diagnosis in the presence of associated anomalies and conditions involving the entire limb. The 2005 EUROCAT study of 4,366 fetuses with anomalies found a prenatal detection rate for both upper and lower limb reduction defects of 34%.  Stoll et al found a higher prenatal detection rate for limb reduction defects with associated malformations (49%) compared to isolated limb reduction defects (25%), and Pakjrt et al found a high detection rate for fetuses with short or absent radii and/or ulnae with associated aneuploidy or genetic syndromes (70%). [10,11] Stoll et al also found previously that detection of more proximal limb reduction defects was better (23–50%) than detection of hand or finger limb reduction defects (0–8%).  Gray et al found that 31% of upper extremity anomalies were detected prenatally, however only 18% were correctly diagnosed.  Those that were missed tended to involve the hand and fingers more than the forearm or entire upper extremity. Our overall detection sensitivity of 42%, with higher sensitivity in fetuses with associated conditions and more proximal involvement, is better than these previously published reports. Our sensitivity for more proximal anomalies was markedly improved compared to prior studies, ranging from 75–100%.
Suboptimal detection of upper extremity anomalies is related to several factors; the current fetal imaging guidelines mandate only a cursory examination of the upper and lower limbs during the standard second trimester examination, which may contribute to a high false negative rate.  The ideal window for visualizing the fetal hands is at the late first and early second trimester, when the fingers tend to be extended and abducted; ultrasound examinations at a later gestational age are limited by fetal size, position, and flexed digits.  The use of three-dimensional ultrasound has also been shown to improve detection of hand anomalies by as much as 50%, [7,9,14] however is not currently recommended for routine use by the American College of Obstetrics and Gynecology. [1,13] We did not routinely use three- dimensional imaging in our patients.
Several benefits are derived from the prenatal diagnosis of upper extremity anomalies.  The primary advantage is the opportunity for more refined and prognostic prenatal counseling and preparation. Parents are given the chance to discuss their child’s diagnosis with a variety of specialists and receive genetic counseling. For treatable anomalies, a team may be assembled to prepare for postnatal care. Some families will consider pregnancy termination for major untreatable anomalies, and several studies have shown higher rates of pregnancy termination after early prenatal diagnosis of major untreatable anomalies. [2,5,6,12] However, conditions with a high rate of false positives should be considered with caution since they can mislead parents and clinicians in their decisions and recommendations.
The identification of an upper extremity anomaly should instigate a more detailed evaluation of the entire fetus, given the high rate of associated conditions. [11,15] Over 50% of the fetuses with limb reduction defects, phocomelia, radial longitudinal deficiency, arthrogryposis, abnormal hand positioning, and cleft hand in our study population had an associated condition involving a major organ system. Several diagnoses were also associated with a high-risk of aneuploidy (Table 4); these fetuses should be considered for diagnostic testing for aneuploidy or chromosomal microarray analysis.
The primary strength of our study was the large study population and the availability of coupled prenatal sonographic and postnatal clinical information for each pregnancy, allowing an accurate assessment of the performance of prenatal ultrasound diagnosis. We were also able to subdivide diagnoses to determine more specific sensitivity data than are available in the literature. Our data are derived from a single center with a consistent sonographic protocol, and, therefore, avoided the variability seen in multicenter studies.
This study may overestimate the sensitivity of obstetric ultrasound for prenatal diagnosis of upper extremity anomalies, as our institution is a tertiary referral center, prompting more thorough examinations. Additionally, registered diagnostic medical sonographers with certification in obstetrics and gynecology perform the exams; these are interpreted by maternal-fetal medicine specialists. These factors provide an optimal setting for prenatal identification of upper extremity anomalies. While technological advances occurred over the course of the data collection, our equipment was serially replaced and, therefore, we cannot identify which studies were performed on which machines. The small number of fetuses diagnosed with certain anomalies is a limitation of our study. Additionally, heterogeneity in gestational age at first ultrasound at our institution may have introduced bias into our sensitivity data since not all patients were evaluated during the ideal window between the late first and early second trimesters.
Funding: This publication was supported by the Washington University Institute of Clinical and Translational Sciences grant UL1 TR00488, from the National Center of Advancing Translational Science.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.