Limited role of MSAFP screening for prenatal omphalocele detection

Abstract

Objectives:

Although serum screening for aneuploidies has become less prevalent, maternal-serum alpha-fetoprotein (MSAFP) screening for body-wall defects remains widespread. We explored whether MSAFP screening is associated with earlier omphalocele detection than ultrasound alone.

Methods:

This is a retrospective cohort study of prenatally detected omphalocele cases at our center from 2007–2023. We explored the association between MSAFP screening, gestational age at omphalocele detection, and clinical outcomes.

Results:

Among 101 pregnancies with prenatally diagnosed omphalocele, 27 (26.7%) had MSAFP screening. The median gestational age at MSAFP screening was 17w4d. Of those who received MSAFP screening, 11 (41%) had values ≥2.5 multiples of the median (MoM) and 16 (59%) were not elevated. MSAFP results did not correlate with omphalocele size and were not associated with prenatal or postnatal outcomes. MSAFP screening did not result in earlier suspicion for or confirmation of omphalocele (p=0.97 and p=0.87, respectively). In contrast, first-trimester ultrasound screening was associated with earlier suspicion for and confirmation of omphalocele (p<0.01 and p=0.01, respectively). There were no clinical or demographic differences between those who received MSAFP screening and those who did not (including BMI or commute distance to an urban center).

Conclusion:

MSAFP screening is not associated with earlier omphalocele detection. Furthermore, in pregnancies with prenatally diagnosed omphalocele, the results of MSAFP screening are not predictive of clinical outcomes.

Introduction

Omphalocele is a birth defect characterized by abnormal development of the ventral abdominal wall that results in herniation of abdominal contents into the base of the umbilical cord. While omphaloceles are often amenable to surgical repair, severe cases result in prolonged NICU admissions, respiratory morbidity, and death. Furthermore, although omphaloceles can occur sporadically, a large proportion are associated with genetic conditions. For these reasons, prenatal detection of omphaloceles is critical, as it enables early genetic testing and delivery planning. Before the widespread availability of ultrasonography, screening for ventral abdominal wall defects relied on measuring alpha-fetoprotein in the maternal serum (MSAFP).(1–3) MSAFP was originally introduced in the 1970s and 1980s,(4,5) and has been incorporated into screening paradigms including the quadruple marker screen and the integrated screen. As prenatal ultrasonography has improved in resolution and sensitivity for body-wall defects, and as NIPT aneuploidy screening has supplanted serum screening, the use of MSAFP has declined leaving its utility and role uncertain.(6–8) It is unclear whether MSAFP screening still results in improved or earlier detection of body-wall defects compared to ultrasound alone.

Arguments can be made for and against second-trimester MSAFP screening. An isolated MSAFP level in a patient not otherwise undergoing multiple marker screening has the benefit of early risk stratification (particularly when first-trimester anatomy ultrasound is not available) because it can be obtained as early as 15 weeks, which is earlier than the typical second-trimester anatomy ultrasound.(4) Furthermore, in patients with suboptimal body-wall acoustics (such as obesity or prior abdominal surgery), MSAFP may enhance detection of body-wall defects. However, drawbacks of MSAFP screening include test anxiety, the discomfort and cost of an additional blood draw (more than 90% of MSAFP draws are solely to measure this analyte(9)), and frequent false-positive results.(10) The MSAFP test itself has relatively poor sensitivity and specificity compared to sonography because an elevated result can be due not only to structural defects but also to placental conditions, fetal congenital nephrosis or demise.(4,11) Moreover, at a time when obstetric risk calculators are eschewing self-reported or provider-presumed race as a risk modifier,(12,13) and despite new evidence that race adjustment is not useful in MSAFP interpretation,(14) most laboratories still consider maternal race in MSAFP reports.(3,15)

Current recommendations from professional societies are inconsistent. The American College of Medical Genetics (ACMG), the American College of Obstetrics and Gynecology (ACOG), and the International Society for Prenatal Diagnosis (ISPD) have each issued contrasting position statements regarding the need for second-trimester maternal serum alpha-fetoprotein (MSAFP) screening in patients already undergoing sonographic and NIPT screening. In 2016, the ACMG held that “[Non-invasive prenatal screening] does not screen for open neural tube defects; therefore, maternal serum α-fetoprotein testing to screen for open neural tube defects should still be offered at 15–20 weeks of gestation.”(16) In contrast, ISPD recommends either a “second-trimester MSAFP or a second-trimester ultrasound.”(17) More recent opinions from ACOG have downplayed the role of MSAFP in the era of near-universal prenatal ultrasound screening.(4,18)

Given inconsistent recommendations regarding MSAFP testing along with limited studies evaluating the utility of MSAFP testing in modern prenatal care, prenatal care providers’ use of MSAFP is highly inconsistent.(9) To better understand whether MSAFP can benefit modern prenatal care, we explored whether MSAFP screening resulted in earlier omphalocele detection among a cohort of prenatally detected omphalocele cases treated at our institution.

Methods

This retrospective study was approved by the Baylor College of Medicine Institutional Review Board (protocol H-52419). We reviewed cases of prenatally detected omphalocele, which received prenatal evaluation and/or postnatal care at our fetal referral center from January 2007 to July 2023. Cases were identified by review of the electronic medical record and imaging software at Texas Childrens’ Hospital and the Harris County Hospital System for patients with an ICD/Grouper code of “Congenital Omphalocele”. Comparison with clinical case lists developed by the Radiology and Surgery departments did not yield additional omphalocele cases for review. For inclusion in this study, cases were prenatally detected with postnatal confirmation. Cases of non-omphalocele ventral-wall defects were excluded, including gastroschisis and umbilical hernia. For included cases, clinical records were reviewed through December 2023.

For each included case, we abstracted maternal, fetal, and neonatal variables including the timing of MSAFP blood draw, if it occurred. Due to the overall completeness of our dataset, samples for which we could not definitively determine whether MSAFP had been drawn (n=12) were treated as no-draw cases. We recorded the gestational age (GA) at first suspicion of omphalocele (defined as the GA of the first outside ultrasound on which there was concern for a ventral wall defect, or the first ultrasound at our institution on which there was non-diagnostic concern for a ventral wall defect) and at diagnosis of the omphalocele at our institution. Physiologic midgut herniation (i.e. <1cm at ≤12 weeks and not including liver) was not considered suspicious for or diagnostic of a ventral wall defect and these cases were excluded.

We explored the association between MSAFP screening, gestational age at omphalocele detection, and association with demographic variables using bivariate analyses, the Wilcoxon rank-sum test for nonparametric comparisons, and correlation coefficients. Because elevated BMI and more rural locations may be impediments to effective sonographic evaluation, we assessed whether MSAFP screening was more prevalent in patients with elevated BMI or whose home census tract was further from an urban center, which would suggest that these patients’ providers may utilize MSAFP as an adjunct screen. The geospatial analysis was performed using the 2010 Rural-Urban Commuting Area Codes (RUCA) dataset.(19) All statistical analyses were performed in R, which is a freely available software suite for statistical analysis and data visualization.(20)

Results

Of 971 records screened, 101 met all inclusion criteria (Figure 1). Of these pregnancies, 27 (26.7%) underwent MSAFP screening, with a median gestational age of 17w4d. No demographic or clinical variables were significantly associated with MSAFP screening (Table 1). Notably, BMI and the proximity of the patient’s home census tract to an urban center did not differ between groups who received and who did not receive MSAFP screening.

Figure 1. Flow diagram of the sample selection process.

Records were identified primarily through database searches. After deduplication and exclusion of ineligible cases, there were 101 eligible cases remaining for analysis. BWS, Beckwith-Wiedemann Syndrome.

Table 1.

Demographic and clinical associations with MSAFP screening

Variable	MSAFP (n=27)	No MSAFP (n=74)	P value
Age, years			0.06a
Mean (SD)	26.7 (6.83)	29.42 (6.45)
Median (Min-Max)	28 (17–45)	30 (14–44)
IQR (Q1-Q3)	10.5 (20–30.5)	8.75 (25–33.75)
BMI			0.66a
Mean (SD)	27.67 (6.85)	27.93 (5.81)
Median (Min-Max)	25 (17.7–47.5)	28 (18.7–46)
IQR (Q1-Q3)	7.3 (23.5–30.8)	7.4 (23.4–30.8)
Parity, n (%)			0.18b
Nulliparous	18 (32.7)	37 (67.3)
Parous	9 (19.6)	37 (80.4)
Race, n (%)			0.79b
White	21 (25.3)	62 (74.7)
Black	5 (35.7)	9 (64.3)
Other	1 (25)	3 (75)
Ethnicity, n (%)			0.37b
Hispanic	14 (31.8)	30 (68.2)
Not Hispanic	13 (22.8)	44 (77.2)
Insurance, n (%)			0.21b
Medicaid or CHIP	17 (34.7)	32 (65.3)
Private insurance	9 (19.15)	38 (80.85)
Uninsured or not known	1 (20)	4 (80)
GA at entry to care, weeks			0.89a
Mean (SD)	10.38 (4.32)	11.79 (6.74)
Median (Min-Max)	10 (1–20)	10 (1–35)
IQR (Q1-Q3)	5.25 (8–13.25)	5 (8–13)
RUCA Commute Code*			0.69
Mean (SD)	1.52 (1.6)	1.27 (0.93)
Median (Min-Max)	1 (1–7)	1 (1–7)
IQR (Q1-Q3)	0 (1–1)	0 (1–1)
First-Trimester Scan, n (%)			0.28b
Yes	13 (28.9)	32 (71.1)
No	13 (30.2)	30 (69.8)
Unknown	1 (7.7)	12 (92.3)
GA at MSAFP draw, weeks		NA	NA
Mean (SD)	17.57 (1.67)
Median (Min-Max)	17.43 (15.14–21.57)
IQR (Q1-Q3)	2.43 (16.36–18.79)
GA at sonographic suspicion for abdominal wall defect, weeks			0.97a
Mean (SD)	18.25 (4.81)	18.7 (6.32)
Median (Min-Max)	18.29 (11–31.71)	18.29 (10–37)
IQR (Q1-Q3)	5.57 (14.64–20.21)	8.21 (13.11–21.32)
GA at sonographic confirmation of abdominal wall defect, weeks			0.87a
Mean (SD)	24.92 (6.54)	24.61 (6.51)
Median (Min-Max)	24.43 (12.29–37)	23.71 (11.57–38.86)
IQR (Q1-Q3)	10.5 (20.21–30.71)	7.96 (21.04–29)

Categorical data are presented as number (percentage). P values are not adjusted for multiple comparisons.

^aThe Wilcoxon rank-sum test was used to calculate the P value;

^bThe Fisher exact test was used to calculate the P value;

^*Rural-Urban Commuting Area Codes (RUCA), higher values reflect rural locales with longer commutes to an urban center.

BMI, body-mass index; GA, gestational age; Min, minimum; Max, maximum; MSAFP, maternal serum alpha-fetoprotein.

We next investigated whether MSAFP screening resulted in earlier detection of omphalocele. There was no difference between the MSAFP and no-MSAFP groups with regard to gestational age at sonographic suspicion or confirmation of omphalocele (Table 1). We further stratified these groups by first-trimester screening ultrasound. For cases with and without a first-trimester screening ultrasound, the addition of MSAFP screening was not associated with earlier detection. In contrast, first-trimester ultrasound screening was by itself significantly associated with earlier detection (Figure 2).

Figure 2. Gestational age at first suspicion of omphalocele.

MSAFP blood draw was not associated with earlier suspicion for omphalocele in subgroups which had or did not have a first-trimester ultrasound (Panel A). In contrast, first-trimester ultrasound demonstrated a strong association with earlier suspicion for omphalocele (Panel B).

Of those who underwent MSAFP screening, 11 (41%) had MSAFP values elevated ≥2.5 multiples of the median (MoM) and 16 (59%) were not elevated. MSAFP values did not correlate with omphalocele size (Spearman=0.09) or the presence of additional anatomic defects (phi=0.368). Although not powered for these comparisons, elevated MSAFP values were not associated with postnatal outcomes including fetal or neonatal death, respiratory outcomes, and type of surgical repair (Table 2). Only one case that underwent MSAFP screening experienced a prenatal or perinatal omphalocele rupture. In this case, the MSAFP MoM was 14.6 (the highest recorded in this study), though rupture was also recognized on prenatal ultrasound.

Table 2.

Associations of MSAFP results with postnatal outcomes

Variable	MSAFP < 2.5 MoM (n=14)	MSAFP ≥ 2.5 MoM (n=11)	P value
Fetal or neonatal death, n (%)			1.00b
Yes	3 (60)	2 (40)
No	11 (55)	9 (45)
Termination of pregnancy, n (%)			1.00b
Yes	0 (NaN)	0 (NaN)
No	14 (56)	11 (44)
Mode of delivery, n (%)			1.00b
Cesarean	12 (57.1)	9 (42.9)
Vaginal	2 (50)	2 (50)
GA at delivery			0.06a
Mean (SD)	37.07 (2.67)	35.22 (2.18)
Median (Min-Max)	38.29 (31.6–39.6)	34.43 (31.9–39.1)
IQR (Q1-Q3)	2.75 (36.2–38.9)	3 (33.9–36.9)
Need for intubation, n (%)			0.67b
Yes	4 (50)	4 (50)
No	10 (62.5)	6 (37.5)
Need for high-frequency oscillatory ventilation, n (%)			1.00b
Yes	1 (50)	1 (50)
No	13 (65)	7 (35)
Any respiratory support on DOL 30, n (%)			0.70b
Yes	5 (50)	5 (50)
No	9 (60)	6 (40)
Pulmonary hypertension, n (%)			1.00b
Yes	3 (60)	2 (40)
No	7 (50)	7 (50)
Surgical repair, n (%)			0.25b
Paint and wait	8 (72.3)	3 (27.7)
Primary	4 (44.4)	5 (55.6)
Staged	0 (0)	1 (100)
Length of NICU stay			0.74a
Mean (SD)	66.62 (84.59)	46.89 (35.55)
Median (Min-Max)	37 (3–317)	38 (6–106)
IQR (Q1-Q3)	57 (15–72)	51 (13–64)

Categorical data are presented as number (percentage). P values are not adjusted for multiple comparisons.

^aThe Wilcoxon rank-sum test was used to calculate the P value;

^bthe Fisher exact test was used to calculate the P value.

DOL, day of life; GA, gestational age; Min, minimum; Max, maximum; MSAFP, maternal serum alpha-fetoprotein; NICU, neonatal intensive-care unit.

Discussion

In this cohort of patients with prenatally diagnosed omphalocele, we found that MSAFP screening was not associated with earlier omphalocele detection. In contrast, first-trimester screening ultrasound was significantly associated with early omphalocele detection. It is well established that omphalocele is readily detected by first-trimester ultrasound.(21) We considered whether a possible benefit of MSAFP for early omphalocele detection was outweighed by first-trimester screening ultrasound, and thus whether MSAFP could be helpful in settings with limited ultrasound availability. However, MSAFP demonstrated no benefit even among cases without first-trimester ultrasound.

As a result of the past decade’s dramatic shift away from serum screening in favor of NIPT, there is significant heterogeneity in prenatal care providers’ use of MSAFP screening.(9) While existing evidence suggests that MSAFP no longer offers a benefit for prenatal detection of body-wall defects,(10,22) large studies are lacking and available clinical guidance is inconsistent. In clinical practice, MSAFP results are most valued when sonographic visualization of the body-wall is limited. However, in the present study and in a past study of a different population undergoing MSAFP screening,(9) there was no association between MSAFP use and body-mass index, a factor known to impair prenatal sonographic detection.(23) Furthermore, we found no association between use of MSAFP screening and proximity to urban centers, which are likely to have improved care access compared to rural locales. Therefore, this study provided no evidence that mid-trimester MSAFP screening improves prenatal omphalocele detection, and further found no evidence that providers preferentially use MSAFP as a adjunct in populations that may be more difficult to screen: those who reside further from an urban center or with a higher BMI.

We next explored whether MSAFP screening has prognostic utility in cases of prenatally diagnosed omphalocele. Although the only prenatally ruptured omphalocele that underwent MSAFP screening had the highest MSAFP value, this clinical outcome was also detected by ultrasound. We also found no association between elevated MSAFP results and a variety of postnatal outcomes. Thus, we found no role for mid-trimester MSAFP screening in the prenatal diagnosis or evaluation of omphalocele. These results should prove useful as professional societies re-evaluate the role of prenatal serum screening for aneuploidy and body-wall defects.

This study has several strengths. These data are derived from a large modern cohort of prenatally detected omphaloceles. There are few available data investigating the benefits of MSAFP screening for the detection of body-wall defects in modern cohorts(24) and recommendations from professional societies regarding MSAFP use are inconsistent, making studies of this topic timely. Detailed information regarding demographics, ultrasound timing, and MSAFP timing have enabled granular investigation into the potential temporal benefit of MSAFP for omphalocele detection.

However, this study is not without limitations. With this study’s retrospective design we are unable to answer whether AFP would help detect the very rare cases that are undetected by ultrasound and are postnatally ascertained.(25,26) Although our sample size was not sufficient to detect very small improvements in omphalocele detection attributable to MSAFP screening, there was neither an association nor a trend toward improved detection in MSAFP cases. Also, in this study we did not explore the role of MSAFP in detection of gastroschisis. MSAFP levels are on average twice as high in gastroschisis than in omphalocele, and as such MSAFP has better sensitivity for detecting gastroschisis.(27) Finally, the majority of the patients in this cohort identify as White and are insured which may contribute to the early average gestational age at entry to care and illustrate differences in access to care. Because cases were identified by methods that should be agnostic to race and ethnicity, this skew is not a result of ascertainment bias but accurately reflects the omphalocele patients that were cared for at our institution. While care should be taken when extending these results to diverse populations, the early gestational age at entry to care improved the power to test the effectiveness of first-trimester ultrasound and MSAFP for omphalocele detection.

In this clinically and demographically diverse cohort we found that MSAFP screening was not associated with earlier omphalocele detection or key outcomes including omphalocele size, the presence of additional anomalies, or postnatal morbidity. Because the incidence of omphalocele is higher among some racial minority groups, it is important to assess whether the use of or deprioritization of MSAFP screening for body-wall defects would disproportionately affect these groups.(28) Additional work should be done in modern cohorts to determine whether MSAFP screening improves prenatal detection of gastroschisis or neural tube defects, which can be more challenging to diagnose than ventral wall defects. Until additional data are available, it will remain unclear whether mid-trimester MSAFP screening represents a misuse of resources or a missed screening opportunity.

What is already known about this topic?

Mid-trimester maternal-serum alpha-fetoprotein (MSAFP) measurement is a widely but inconsistently used screening tool to assess for fetal body-wall defects, including omphalocele. In the era of non-invasive prenatal testing and high-resolution first-trimester ultrasonography, it is unclear whether MSAFP screening improves prenatal detection or management of omphalocele.

What does this study add?

We found that MSAFP screening does not lead to earlier omphalocele detection, either in patients that are or are not undergoing first-trimester ultrasonography. Furthermore, MSAFP results were not associated with prenatal or postnatal outcomes. Our findings suggest that, in a modern setting, MSAFP screening does not improve omphalocele detection or management.

Data availability:

Due to privacy restrictions, individual-level raw data are not available for sharing outside our institution. Please direct data requests to the corresponding author.