Effect of Adjustment of Maternal Seruma-Fetoprotein Levels in Insulin-DependentDiabetes Mellitus

R.L. Kramer,1* Y. Yaron,1 J.E. O’Brien,3 G. Critchfield,3 M. Ayoub,1 M.P. Johnson,1,2 C.R. Qualls,4and M.I. Evans1,2

1Center for Fetal Diagnosis and Therapy, Division of Reproductive Genetics, Departments of Obstetrics andGynecology, Hutzel Hospital/Wayne State University, Detroit, Michigan

2Department of Pathology, Molecular Medicine and Genetics, Hutzel Hospital/Wayne State University,Detroid, Michigan

3Quest Diagnostics, Teterboro, New Jersey4General Clinical Research Center, University of New Mexico Health Sciences Center, Albuquerque, New Mexico

Our objective was to determine the effect ofthe 20% upward adjustment of maternal se-rum alphafetoprotein (MSAFP) in patientswith insulin-dependent diabetes mellitus(IDDM) on the number of patients thatwould be classified at increased risk forpregnancy complicated by either Down syn-drome (DS) or neural tube defect (NTD). Weretrospectively evaluated a database con-taining 63,110 patients who underwent mul-tiple serum marker screening between 14and 22 weeks gestation; 620 patients withIDDM had measurements of MSAFP ofwhich 479 also had measurements of b-HCG,allowing calculation of DS risk. IncreasedNTD risk was defined as MSAFP >2.5 MOMwhile increased DS risk was defined as a cal-culated risk >1/270. One IDDM patient de-livered an infant with a NTD; it was not de-tected on serum screening. No infants wereborn with DS. Of the 620 patients withMSAFP determinations, 9 had values >2.5MOM before adjustment. After upward ad-justment, 7 additional patients were identi-fied. Sixteen patients were identified at in-creased risk for DS before and after adjust-ment. Our data suggest that the 20% upwardadjustment of MSAFP increases by 78%, thenumber of patients who would require fur-ther evaluation for NTD’s. Although wewere able to identify 620 women with IDDM

who underwent serum screening for NTD,the low prevalence of NTD’s did not allow usto demonstrate an increased detection rate.The effect of upward adjustment of MSAFPon the number of patients categorized at in-creased DS risk appears to be minimal.Am. J. Med. Genet. 75:176–178, 1998.© 1998 Wiley-Liss, Inc.

KEY WORDS: MSAFP; NTD; Down syn-drome; IDDM

INTRODUCTION

Maternal serum a-fetoprotein (MSAFP) has beenemployed routinely as a serum marker in prenatalscreening for both open neural tube defects (NTD’s) inwhich levels are elevated [Crandall and Matsumoto,1986; Cuckle and Wald, 1987] and Down syndrome(DS) in which levels are decreased [Merkatz et al.,1984, Cuckle et al., 1984]. Elevated serum human cho-rionic gonadotropin (HCG) levels and decreased levelsof serum unconjugated estriol (UE3) have also been ob-served in pregnancies complicated by DS [Bogart et al.,1987, Canick et al., 1988]. Currently, the ‘‘doublescreen’’ (MSAFP, HCG) or the ‘‘triple screen’’ (MSAFP,HCG, and UE3) are used in conjunction with maternalage to assess the risk of DS [Evans et al., 1994].

Low MSAFP in association with IDDM was initiallydescribed in 1979 [Wald et al., 1979] and subsequentlywas confirmed by several investigators [Milunsky etal., 1982, Martin et al., 1990, and Wald et al., 1992].The median level of free a-HCG has also been shown tobe reduced in the pregnancies of women with IDDM[Wald et al., 1994]. As a result, it has become routinepractice to adjust upward the MSAFP multiple of themedian (MOM) value, by 20%, to enhance test sensi-tivity. This results in an increase in the number of

Presented in part as a poster at the 4th Joint Clinical GeneticsMeeting, February 28–March 2, 1997.

*Correspondence to: Ralph L. Kramer, M.D., Department ofObstetrics and Gynecology, Division of Reproductive Genetics,Hutzel Hospital/Wayne State University, 4707 St. Antoine Bou-levard, Detroit, MI 48201.

Received 20 March 1997; Accepted 11 August 1997

American Journal of Medical Genetics 75:176–178 (1998)

© 1998 Wiley-Liss, Inc.

candidates for whom amniocentesis may be recom-mended. This recommendation is based on the assump-tion that the additional patients identified as being atincreased risk are, in fact, at increased risk, thoughthis has yet to be confirmed. Moreover, the upwardadjustment of MSAFP will decrease the number ofwomen classified at increased risk of carrying a fetuswith DS. However, it is apparent that the magnitude ofchange will be smaller, given that the calculated risk ofDS is also dependent on maternal age, HCG, and, ifutilized, UE3. The purpose of the current study was todetermine the effect of the 20% upward adjustment ofthe MSAFP measurement in patients with IDDM onthe number of patients that would be classified at in-creased risk for either DS or NTD.

MATERIALS AND METHODS

This retrospective study included 63,110 patientswho underwent multiple serum marker screening be-tween 14 and 22 weeks gestation, over a 5-year period(3/91–5/96). Data were entered prospectively into acomputerized database (Quest Laboratories, Teterboro,NJ and Auburn Hills, MI). Pregnancy outcome was ob-tained on all patients utilizing a physician-completedquestionnaire; 620 patients with IDDM who had mea-surements of MSAFP were identified. Four hundredseventy nine of these patients also had measurementsof b-HCG, allowing calculation of DS risk. Increasedrisk of NTD was defined as an MSAFP level ù2.5MOM. A patient was considered to be at increased riskof having a fetus affected with DS if her calculated riskwas ù1/270 (the reported risk of detection of DS onamniocentesis at 16 weeks GA in a woman age 35). Thenumber and percentage of patients classified at in-creased risk before and after adjustment was deter-mined. The 95% confidence intervals were calculatedwith Stat Xact Turbo (Cytel, Inc., Cambridge, MA).

RESULTS

One patient with a NTD was identified among the620 patients screened with MSAFP. The MSAFP levelin this patient was 1.18 MOM after adjustment, andthis defect was not, therefore, detected on serumscreening. Hence, our observed positive predictivevalue was 0% (exact one-sided 95% CI 4 0–17%). Ofthe 479 patients with IDDM who were screened for DSbased on MSAFP, b-HCG, and maternal age, none wassubsequently delivered of an infant with DS.

Of the 620 patients with MSAFP determinations, 9had values >2.5 MOM before adjustment. After upward

adjustment of MSAFP, 7 additional patients (Table I)were identified at increased risk for NTD, an incrementof 78% (95% CI 4 32–162%). The confidence intervalabout the 78% is a conditional confidence interval,given that the 9 subjects initially identified are subse-quently excluded from the 620 subjects, i.e. the 7 addi-tional subjects were derived from the remaining pool of611 subjects. The exact conditional confidence intervalis computed using a binomial distribution based on the7 of the 611 subjects divided by the estimated probabil-ity of selection of the subjects before adjustment ofMSAFP, i.e. 9 out of 620.

Of the 479 patients with both MSAFP and b-HCGdeterminations, 16 patients were classified at in-creased DS risk before upward adjustment. After up-ward adjustment, no change in the number of patientsat risk was observed (conditional one-sided 95% CI 40–19%). The exact conditional confidence interval wascomputed using a binomial distribution, based on the16 of the 463 subjects, divided by the estimated prob-ability of selecting the subjects without MSAFP adjust-ment, i.e. 16 out of 479.

DISCUSSION

Our data suggest that the 20% upward adjustment ofMSAFP measurements increases the number ofwomen who undergo further evaluation for NTD’s by78%. The adjustment would, therefore, be likely to in-crease the number of women undergoing further evalu-ation, in a population of women who are known to be atincreased risk of carrying a fetus with a NTD [Milun-sky et al., 1982]. However, NTD’s are relatively uncom-mon anomalies, even in a population of patients withIDDM. Although we were able to identify 620 womenwith IDDM who underwent serum screening for NTD,the low prevalence of NTD’s did not allow us to dem-onstrate an increased detection rate.

It would seem that upward adjustment of MSAFPwould be beneficial, resulting in an increase in the sizeof the candidate pool of women screened, and who,therefore, would receive further evaluation. The risk offetal NTD’s in the additional women identified by up-ward adjustment has not been quantified. The questionmust be raised as to whether the additional womenidentified do, in fact, have the same risk as women withelevated MSAFP before upward adjustment. In orderto address this question, we compared the positive pre-dictive value of elevated MSAFP between the twogroups, based on certain assumptions.

The prevalence of NTD’s varies significantly in dif-ferent geographic areas for normal women as well asthose with IDDM [Milunsky et al., 1982, Kucera, 1971].Assuming a positive predictive value of 5.6% as sitedfor the California screened population in 1990 [Cran-dall, 1993], and a doubling of this risk in patients withIDDM [Kucera, 1971] then the positive predictive valuewould be approximately 11% in women with IDDM.Both of these values reflect the upward adjustment.For the purpose of sample size determination and un-der the alternative hypothesis (there is a 50% reduc-

TABLE I. Effect of MSAFP Adjustment

Beforeadjustment

Afteradjustment

Percentchange

(95% CI)

Patients at increasedNTD risk (%) 1.45 2.58 78

(32–162)Patients at increased

DS risk (%) 3.34 3.34 0(0–19)

MSAFP and Diabetes Mellitus 177

tion in the positive predictive value of elevated MSAFPin additional women identified), we assumed a reduc-tion in the positive predictive value from 14% to 7%.The 14% and 7% positive predictive values were ob-tained so that their weighted average would be the 11%noted above. The weighted average is computed usingweights of 100% and 78% applied to the 14% and 7%,respectively. For 80% power and a significance level of0.05, a sample size of 535 women with IDDM andMSAFP levels >2.5 MOM’s would be needed. This inturn, would require that approximately 21,000 womenwith IDDM be screened with MSAFP in order to an-swer the question.

As noted above, no infants with DS were born to the479 patients in our study. It is readily apparent that wewould require a significantly larger number of patientsin order to assess the ultimate effect of the upwardadjustment of MSAFP levels on the identification ofwomen at increased risk of carrying a fetus with DS.Intuitively, upward adjustment of MSAFP would seemto be detrimental, resulting in a decrease in the size ofthe candidate pool for screening. Whether the ultimateeffect is beneficial or detrimental and by what magni-tude, is not known. Reasoning by analogy to NTD de-tection, it is clear that an extremely large number ofsubjects would be required to answer this question.

Thus, our data suggest that the 20% upward adjust-ment of MSAFP increases the number of patients whowould require further evaluation for NTD’s. However,the low prevalence of NTD’s did not allow us to dem-onstrate an increased detection rate. The effect of up-ward adjustment of MSAFP on the number of patientscategorized at increased DS risk appears to be mini-mal. The determination of whether women identifiedafter MSAFP adjustment have the same risk for DSand NTD as those identified before adjustment, wouldrequire an extremely large sample.

REFERENCESBogard MH, Pandian MR, Jones OW (1987): Abnormal maternal serum

chorionic gonadotropin levels in pregnancies with fetal chromosomeabnormalities. Prenat Diag 7:623–630.

Canick JA, Knight GJ, Palomaki GE, Haddow JE, Cuckle (1988): Lowsecond trimester levels of maternal serum unconjugated oestriol inpregnancies with Down’s syndrome. Brit J Obstet Gynecol 95:330–333.

Crandall BF (1993): Prenatal screening for the detection of neural tubedefects. In: Simpson JL and Elias S (eds): ‘‘Essentials of Prenatal Di-agnosis.’’ New York: Churchill Livingstone Inc., pp. 253–274.

Crandall BF, Matsumoto M (1986): Routine amniotic fluid alphafetopro-tein assay: Experience with 40,000 pregnancies. Am J Med Genet 24:143–149.

Cuckle H, Wald N (1987): The impact of screening for open neural tubedefects in England and Wales. Prenat Diag 7:91–99.

Cuckle HS, Wald NJ, Lindenbaum RH (1984): Maternal serum alpha-fetoprotein measurement: A screening test for Down syndrome. Lancet1(8383):926–9.

Evans MI, O’Brien JE, Dvorin MP (1994): Biochemical screening. CurrentOpin Obstet Gynecol 6:453–458.

Kucera J (1971): Rate and type of congenital anomalies among offspring ofdiabetic women. J Reprod Med 7:73–82.

Martin AO, Dempsey LM, Minogue J, Liu K, Keller J, Tamura R, FreinkelN (1990): Maternal serum a-fetoprotein levels in pregnancies compli-cated by diabetes: Implications for screening programs. Am J ObstetGynecol 163:1209–1216.

Merkatz IR, Notowsky HM, Macri JN, Johnson WE (1984): An associationbetween low maternal serum alpha-fetoprotein and fetal chromosomalabnormalities. Am J Obstet Gynecol 148:886–894.

Milunsky A, Alpert E, Kitzmiller JL, Younger MD, Neff RK (1982): Pre-natal diagnosis of neural tube defects. VIII. The importance of serumalpha-fetoprotein screening in diabetic women. Am J Obstet Gynecol142:1030–1032.

Wald NJ, Cuckle H, Boreham J, Stirrat GM, Turnbull AC (1979): Maternalserum alpha-fetoprotein and diabetes mellitus. Brit J Obstet Gynaecol86:101–105.

Wald NJ, Cuckle HS, Densem JW, Stone RB (1992): Maternal serum un-conjugated oestriol and human chorionic gonadotrophin in pregnancieswith insulin-dependent diabetes: Implications for screening for Down’ssyndrome. Brit J Obstet Gynaecol 99:51–53.

Wald NJ, Densem JW, Cheng R, Collishaw S (1994): Maternal serum freea- and free b-human chorionic gonadotrophin in pregnancies with in-sulin-dependent diabetes mellitus: Implications for screening forDown’s syndrome. Prenat Diag 14:835–837.

178 Kramer et al.