Unexplained elevated maternal serum a-fetoprotein is not

predictive of adverse perinatal outcome in an indigent urban population

Owen P. Phillips, MD: Joe Leigh Simpson, MD,. Cynthia D. Morgan, RNC, MS: Richard N. Andersen, PhD: Lee P. Shulman, MD: Carole M. Meyers, MD,. Baha Sibai, MD,. David C. Shaver, MD,. Elizabeth A. Tolley, PhD,b and Sherman Elias, MD·

Memphis, Tennessee

OBJECTIVE: The null hypothesis of this study is that in an urban, indigent obstetric population at high risk

for adverse perinatal outcome, unexplained elevations of maternal serum a-fetoprotein are not an additional predictor of adverse perinatal outcome.

STUDY DESIGN: Perinatal outcomes of 72 patients from a clinic for indigent patients with unexplained

elevated maternal serum a-fetoprotein levels were compared with those of matched controls from the same population with normal maternal serum a-fetoprotein levels. Subjects and controls were matched for

age, race, parity, and presence or absence of Hollister risk factors. The frequency of adverse perinatal

outcome in the two groups was subjected to matched-pair x2 analysis. RESULTS: Adverse perinatal outcome occurred in 38.9% (28 of 72) of subjects with unexplained elevated

maternal serum a-fetoprotein levels 2:2.5 multiples of the median, compared with 31.9% (23 of 72) of

controls with normal maternal serum a-fetoprotein levels (p = 0.5). No statistically significant difference in adverse perinatal outcomes was found. CONCLUSIONS: Elevated maternal serum a-fetoprotein levels offer little if any additional predictive value

for adverse perinatal outcome in populations already at high risk for such outcomes on the basis of

obstetric or socioeconomic criteria. (AM J OBSTET GYNECOL 1992;166:978-82.)

Key words: Maternal serum a-fetoprotein, adverse perinatal outcome

Several studies have shown an association between adverse perinatal outcome and unexplained elevated levels of maternal serum a-fetoprotein (AFP).'-5 Such adverse perinatal outcomes included fetal loss, preterm labor, preterm rupture of membranes, pregnancy-in­duced hypertension, stillbirth, intrauterine growth re­tardation (IUGR), and abruptio placentae. I 5 Investi­gators logically have recommended that patients with unexplained elevated maternal serum AFP should be considered high risk for adverse perinatal outcome and should have very close, but as yet undefined, prenatal follow-up.4-7

These studies, however, were all conducted in ob­stetric populations that were predominantly low risk. For example, the study population of Milunsky et al." was mostly white (94%) and received prenatal care from

From the Department of Obstetrics and Gynecology' and Biostatistics and Epidemiology: University of Tennessee, Memphis. Supported in part by National Institutes of Health grant No. A123479, Contract No. N01-HD-8-2904, the March of Dimes Birth Defects Foundation, the United States Agency for International De­velopment, and a University of Tennessee Chair of Excellence ULS.). Received for publication July 2,1991; revised September 3,1991; accepted September 13, 1991. Reprint requests: Owen P. Phillips, MD, Department of Obstetrics and Gynecology, University of Tennessee, Memphis, 873 Jefferson Ave., Room E-102, Memphis, TN 38163. 6/1/33786

978

private obstetricians. The study populations of Burton'

in North Carolina and Robinson et al.5 in California were derived from regional neural tube defect screen­ing programs that included private practice patients. Studies of Wald et al.,1 Brock et al.! and Smith' were from Great Britain, with demographic information not offered_ In addition, matched-pair analysis was not per­formed in any of these studies.

The purpose of this study was to determine whether unexplained elevated maternal serum AFP would show additional predictive value in patients already identified to be high risk for adverse perinatal outcome on the basis of other obstetric or socioeconomic criteria.

Material and methods

Subjects and controls, to be defined below, received obstetric care between Jan. 1, 1987, and Dec. 31,1989, at the Regional Medical Center, Memphis, Tennessee. In 1990 this population was 79.6% black, 89% indigent (78% on Medicaid, 11 % uninsured), and 61 % unwed, with an estimated 10% incidence of maternal substance abuse. All patients were offered maternal serum AFP screening between 15 and 20 weeks.

Maternal serum AFP assays were performed with the Abbott enzyme immunoassay. An elevated maternal se­rum AFP level was defined as 2:2.5 multiples of the

Volume 166 Number 3

Table I. Unequivocal risk factors for adverse perinatal outcome*

Cardiac disease (class I to IV)

Tuberculosis, active

Chronic pulmonary disease Thrombophlebitis Endocrinopathy Epilepsy (on medication)

Two abortions (sponta­neous)

Previous preterm or small­for-gestational-age in­fants

Isoimmunization (ABO, etc.)

Incompetent cervix Previous preeclampsia

Smoking (> 1 pack per day)

Prior fetal or neonatal death

Diabetes mellitus Hypertension Chronic renal disease Congenital-chromosomal

anomalies Hemoglobinopathies

Isoimmunization (Rh)

Alcohol or drug abuse

Habitual abortions

*Adapted from Preexisting Risk Guide, Hollister Inc., Lib­ertyville, III.

median, adjusted for weight, race, and diabetes melli­tus. If maternal serum AFP was 2.5 to 2.9 multiples of

the median and the patient's estimated gestational age was s 18 weeks, our policy was to repeat the assay 1 week later. If the repeat value was <2.5 multiples of the median, maternal serum AFP was considered nor­mal and the patient was not included in the study. If a repeat sample was 2:2.5 multiples of the median, if the initial value was 2:3.0 multiples of the median, or if gestational age was 2: 18 weeks, high-resolution ultra­sonographic examination was performed. The 18-week cutoff was chosen to ensure that patients potentially carrying an anomalous fetus later in the second trimes­ter would be evaluated in time that all pregnancy op­tions could be offered, including termination. If ges­tational age that was based on ultrasonographic mea­surements differed from gestational age that was based on onset of last menses by > 14 days, maternal serum AFP was reinterpreted with ultrasonographic dating; values reinterpreted as being within the normal range were considered explained. Amniocentesis was offered to patients having no ultrasonographic explanation for elevated maternal serum AFP levels. Patients who de­clined amniocentesis were included in our analysis; pa­tients who declined ultrasonography or were found to have multiple gestations were excluded.

An elevated maternal serum AFP value was consid­ered unexplained if, after correcting for erroneous ges­tational age, multiple gestation, and fetal death , no abnormality in fetal anatomy was detected on ultra­sonographic examination. Patients who had oligohy­dramnios «2 cm pocket of amniotic fluid on ultra­sonography) were excluded because of the known association between oligohydramnios, unexplained el­evated maternal serum AFP, and poor outcome.s

Standard Hollister forms (Hollister Inc., Libertyville,

Unexplained elevated AFP and adverse perinatal outcome 979

Table II. Risk factors that do not unequivocally place a patient at risk for adverse perinatal outcome*

Age <15 or >35 yr 2:7 Deliveries Hemorrhage during pre-

vious pregnancy Pregnancy without familial

support Age 2:40 yr

Significant social problems

<8th Grade education Infants 2:4000 gm Surgically scarred uterus

Second pregnancy in 12 mo

Prior neurologically dam­aged infant

Infertility

*Adapted from Preexisting Risk Guide, Hollister Inc., Lib­ertyville, Ill.

m.) were completed by perinatal nurses or house staff at the initial visit on all patients. This form permits information to be recorded concerning the patient'S past medical history, family history, previous pregnancy outcomes, and presence or absence of 33 factors that help identify pregnancies at risk for adverse outcome. In our opinion 21 of the 33 risk factors unequivocally place a patient at increased risk for adverse outcome (Table I); however, the other 12 factors are social fac­tors or other factors that do not, in our opinion, un­equivocally place a patient at increased risk (Table II). Subjects of this study were thus subgrouped according to presence or absence of any of the 21 unequivocal medical risk factors (Table I).

According to a table of random numbers, controls were selected from patients who had singleton preg­nancies and normal maternal serum AFP levels «2.5 multiples of the median and with a Down syndrome risk less than that of a 35-year-old woman at mid­trimester, i.e., 1 in 277). Subjects were matched for the presence or absence of any of 21 medical risk factors (i .e., risk factors present or not present), race (white vs black), age (5-year intervals), and parity (0 vs 2: I).

Pregnancy outcome data were gathered from hos­pital records by a single physician who was unaware of maternal serum AFP status at the time of review. Out­come variables specifically assessed were congenital malformations, spontaneous pregnancy loss <20 weeks, preteI'm labor «37 weeks), preterm rupture of membranes «37 weeks), IUCR, stillbirths, pregnancy­induced hypertension, and abruptio placentae. Statis­tical analysis was done by McNemar's test (matched­pair X 2 test)," a test that minimizes the effects of con­founding variables.

Results

Among 6208 patients undergoing maternal serum AFP screening during the study period, 88 (1.4%) had unexplained elevated values; three of the 88 were ex­cluded because multiple gestation. four because of oli­gohydramnios. and nine because outcomes could not

980 Phillips et al. March 1992 Am J Obstet Gynecol

Table III. Perinatal outcome in subjects with unexplained elevated maternal serum AFP levels 2:2.5 multiples of the median compared with matched controls by matched-pair analysis (McNemar's test)

Subjects (n = 72)

Controls (n = 72)

All adverse outcomes

28 (38.9%) P = 0.5

23 (31.9%)

Major congenital

malformations

2 (2.8%) P = 0.99 I (1.4%)

Pre term labor

8(11.1%) P = 0.48

12 (16.7%)

Preterm rupture of membranes

6 (8.3%) P = 0.75 4 (5.6%)

/UGR

6 (8.3%) P = 0.125 1 (1.4%)

Abruptio placentae

2 (2.8%) P = 0.625 2 (2.8%)

Pregnancy-induced hypertension

9 (12.5%) P = 0.99

10 (13.41 %)

Spontaneous abortion

4 (5.6%) P = 0.125

o

Table IV. Perinatal outcome in subjects (maternal serum AFP levels 2:2.5 multiples of the median) with risk factors and without risk factors compared with matched controls by matched-pair analysis (McNemar's test)

Major Preterm All adverse congenital Preterm rupture of Abruptio Pregnancy-induced Spontaneous

outcomes malformations labor membranes /UGR placentae hypertension abortion

Subjects 14 (38.9%) 2 (5.6%) 3 (8.3%) 1 (2.8%) 6 (16.7%) 1 (2.8%) 6 (16.7%) 1 (2.83%) with risk p = 0.359 P = 0.5 P = 0.34 factors

P = 1.0 P = 0.125 P = 1.0 P = 0.38

(n = 36) Controls 11 (30.6%) 1 (2.8%) 7 (19.4%) 1 (2.8%) 1(2.8%) 1(2.8%) 3 (8.3%) 0

with risk factors (n = 36)

Subjects 14 (38.9%) 0 5 (13.9%) 5 (13.9%) 0 1 (2.8%) 3 (8.3%) 3 (8.3%) without p = 0.6 P = 0.73 P = 0.73 P = 1.0 P = 0.34 P = 0.25 risk factors (n = 36)

Controls 12 (33.3%) 0 5 (13.9%) 3 (8.3%) without risk factors (n = 36)

be obtained. The mean age of the remaining 72 patients was 24.9 years (range 15 to 37 years). Sixty-four (89%) of subjects were black; eight (11 %) were white. Forty­six (64%) were multiparous. Unequivocal risk factors (Table I) were documented in 36 (50%) of the 72 sub­jects with unexplained elevated maternal serum AFP levels. The mean of the multiples of the median values in the control group was 1.16 ± 0.48 (SEM), indicating that the control group's maternal serum AFP values were not skewed high or low.

Pregnancy outcome data are summarized in Table III. Adverse perinatal outcome occurred in 38.9% (28 of 72) of subjects with unexplained elevated levels 2::2.5 multiples of the median, compared with 31.9% (23 of 72) of controls with normal levels (p = 0.5). Further, no statistically significant difference was found when subjects and controls were compared for any specific outcome (Table III). No stillbirths occurred in either the subject or control group.

To determine whether identifying Hollister risk fac­tors helped predict perinatal outcome in patients with unexplained elevated maternal serum AFP levels, sub­jects with unexplained elevated levels and risk factors were compared with matched controls having normal levels and risk factors (Table IV). No statistically sig-

0 1 (2.8%) 7 (19.4%) 0

nificant difference in adverse outcomes was observed. Similarly, comparing patients with unexplained ele­vated levels but no risk factors with matched controls with both normal levels and no risk factors again yielded no statistically significant difference with re­spect to adverse perinatal outcome (Table IV).

Additional analysis was performed comparing those subjects with maternal serum AFP value 2::3.0 multiples of the median with matched controls (Table V). Thirty­three subjects (45.8%) had values 2::3.0 multiples of the median. Adverse perinatal outcome occurred in 16 (48.5%) of these subjects, compared with 12 (36.4%) controls (p = 0.42). Again, no statistically significant difference for any specific outcome was found when analyzed according to presence or absence of risk fac­tors (Table VI).

Finally, verifying the high-risk nature of our popu­lation is the observation that the frequency of adverse outcome in controls with risk factors was similar to that in controls without risk factors (30.6% vs 33.3%, Table IV). This seems to suggest that in our population preex­isting risk factors do not identify patients at high risk. However, six of these 36 controls without risk factors were young primigravid women with pregnancy-in­duced hypertension. In our population pregnancy-in-

Volume 166 Number 3

Unexplained elevated AFP and adverse perinatal outcome 981

Table V. Perinatal outcome in subjects with unexplained elevated maternal serum AFP levels :::::3.0 multiples of the median compared with matched controls by matched-pair analysis (McNemar'S test)

Subjects (n = 33)

Controls (n = 33)

All adverse outcomes

16 (48.5%) P = 0.42

12 (36.4%)

Major congenital

malformations

1 (3.0%) P = 0.5 1 (3.0%)

Preterm labor

6 (18.2%) P = 1.0

5 (15.2%)

Preterm rupture of membranes

3(9.1%) P = 1.0 2 (6.1%)

IUGR

4 (12.1%) P = 0.125 1 (3.0%)

Abruptio placentae

1 (3.0%) P = 0.5 1 (3.0%)

Pregnancy-induced hypertension

5 (15.2%) P = 0.75 5 (15.2%)

Spontaneous abortion

3 (9.1%) P = 0.25

o

Table VI. Perinatal outcome in subjects (maternal serum AFP :::::3.0 multiples of the median with risk factors and without risk factors compared with matched controls by matched-pair analysis (McNemar's test)

Major Preterm All adverse congenital Preterm rupture of Abruptio Pregnancy-induced Spontaneous

outcomes malformations labor membranes IUGR placentae hypertension abortion

Subjects 14 (38.9%) 1 (5.0%) 3 (15.0%) 1(5.0%) 4 (20.0%) 1 (5.0%) 3 (15.0%) 1 (5.0%) with risk p = 0.359 P = 1.0 P = 1.0 P = 1.0 P = 0.38 P = 1.0 P = 0.63 factors (n = 20)

Controls 11 (3Q.6%) 1 (5.0%) 3 (15.0%) 1(5.0%) 1 (5.0%) 1 (5.0%) 1 (5.0%) 0 with risk factors (n = 20)

Subjects 7 (53.9%) 0 3(23.1%) 2 (15.4%) 0 0 2 (15.4%) 2 (15.4%) without p = 0.5 P = 1.0 P = 1.0 risk factors (n = 13)

Controls 6 (46.2%) 0 2 (15.4%) 1 (7.7%) without risk factors (n = 13)

duced hypertension is very common, occurring in 19.4% of controls without risk factors in this study. If young primigravid women without risk factors and pregnancy-induced hypertension are excluded from analysis, controls with Hollister risk factors indeed show a trend toward more adverse outcomes (13.2%) than controls without risk factors (7.9%).

Comment

Our matched-pair analysis failed to confirm the ob­servations of others that patients with unexplained el­evated maternal serum AFP have an increased risk of adverse perinatal outcome regardless of whether a cut­off of :::::2.5 or :::::3.0 multiples of the median was used. We realize that the number of subjects in this study is relatively small, limiting the power. To achieve a power of 0.80 and a significance level of 0.05, a sample size of 350 subjects would be required to detect a difference of 10.5% in adverse outcomes. However, our findings suggest that the clinical significance of unexplained el­evated maternal serum AFP levels in an urban, indigent population having an a priori increased risk for adverse perinatal outcome needs to be reconsidered.

When subjects were subgrouped according to pres-

P = 0.69 P = 0.5

0 0 4 (30.8%) 0

ence or absence of risk factors and compared with con­trols (matched for the presence or absence of risk fac­tors), no predictive value for elevated maternal serum AFP was found. IUGR occurred in six (16.7%) subjects with unexplained elevated levels and one (1.4%) con­trol; this difference is not statistically significant (p =

0.125). A larger sample size may have demonstrated significance. However, all subjects with IUGR had preexisting factors that placed them at high risk for such an outcome (e.g., significant drug and alcohol use, n = 4; prior neonatal death, n = 2; hypertension, n = 1).

In our subjects with preexisting risk factors, an unex­plained elevated maternal serum AFP level did not fur­ther define a subpopulation at even higher risk. Fur­thermore, no difference in adverse perinatal outcome was observed when subjects with unexplained elevated levels and no Hollister risk factors were compared with controls with normal levels and no Hollister risk factors. Irrespective of whether women actually had preexisting Hollister risk factors, maternal serum AFP did not ap­pear to have additional predictive value for adverse perinatal outcome.

In summary, the association between unexplained

982 Phillips et al.

maternal serum AFP levels and adverse perinatal out­come may not apply to all obstetric populations. Our study differs from others'-6 in that it involved an in­digent urban population that is already known to be at increased risk for adverse perinatal outcome. Unex­plained elevated maternal serum AFP does not appear to have additional predictive value in this population. We therefore suggest further studies from institutions whose populations are similar to ours.

REFERENCES I. Wald N, Cuckle H, Stirrat GM, et al. Maternal serum alpha­

fetoprotein and low birth-weight. Lancet 1977;2:268-70. 2. Brock DJ, Barron L, Duncan P. Significance of elevated

mid-trimester maternal plasma alpha-fetoprotein values. Lancet 1979;2:1281-2.

3. Smith ML. Raised maternal serum alpha-fetoprotein levels and low birth weight babies. Br J Obstet Gynaecol 1980;87: 1099-102.

March 1992 Am J Obstet Gynecol

4. Burton BK. Outcome of pregnancy in patients with unex­plained elevated or low levels of maternal serum alpha­fetoprotein. Obstet Gynecol 1988;72:709-13.

5. Robinson L, Grau P, Crandall BF. Pregnancy outcomes after increasing maternal serum alpha-fetoprotein levels. Obstet GynecoI1989;74:17-9.

6. Milunsky A, Jick SS, Bruell CL, et al. Predictive values, relative risks, and overall benefits of high and low maternal serum Il-fetoprotein screening in singleton pregnancies: new epidemioligic data. AM J OBSTET GYNECOL 1989;161:291-7.

7. Garver KL. Update on MSAFP policy statement from The American Society of Human Genetics. Am J Hum Genet 1989;45:332-4.

8. Dyer SN, Burton BK, Nelson LH. Elevated maternal serum Il-fetoprotein levels and oligohydramnios: poor prognosis for pregnancy levels. AM J OBSTET GYNECOL 1987;157: 336-9.

9. Rosner B. Hypothesis testing: categorical data. In: Fun­damentals of biostatistics. Boston: Duxbury Press, 1986;335.