\|WN1HH\I$\W\\IUHWIUHWHUWMWHW‘HI NWERSITY LIBRARIES llllllllllllllllllllllll lllllll llll 3 1293 01405 178 ll This is to certify that the thesis entitled ABNORMAL TRIPLE TEST RESULTS AND THE RELATION TO ADVERSE PREGNANCY OUTCOMES presented by Robert D. Thomas has been accepted towards fulfillment of the requirements for M. S. degree in Epidemiology (IL/£5444: wane Major professor Date April 15, 1996 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution LIEBRARY Michigan State University PLACE fl RETURN BOX to man this ohockout from your rooord. TO AVOID FINES Mum on or odor. duo duo. DATE DUE DATE DUE DATE DUE Mona-on ABNORMAL TRIPLE TEST RESULTS AND THE RELATION TO ADVERSE PREGNANCY OUTCOMES By Robert D. Thomas A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Epidemiology 1996 ABSTRACT ABNORMAL TRIPLE TEST RESULTS AND THE RELATION TO ADVERSE PREGNANCY OUTCOMES By Robert D. Thomas The triple test is routinely used as a prenatal screening procedure to detect an increased risk for neural tube defects or Down syndrome. We hypothesized that women who had high test results for both the AFP and hCG components of the triple test would be most at risk for adverse pregnancy outcomes when compared to women with normal test results. We retrospectively selected women based on their triple test results from the screening databases of the Henry Ford Hospital (Detroit) and Michigan State University prenatal screening programs. Four groups were assembled: those with elevated alpha- fetoprotein (AFP) (22.0 MOM), those with elevated human chorionic gonadatropin (hCG) (23.0 MoM), those with both elevated AFP and hCG, and a normal comparison group. Our results indicated that women with elevated hCG levels were at increased risk for fetal defects. Women with elevated AFP levels were at significantly increased risk for stillbirth or spontaneous abortion (SAB), preterm delivery, and all outcomes examined combined. Women with elevated levels of both AFP and hCG were at an increased risk for stillbirth or SAB, fetal defects, preterm delivery, low birthweight, small for gestational age, and all adverse outcomes combined. To my family and Joy iii ACKNOWLEDGEMENTS I’d like to acknowledge the work of the two screening centers at Henry Ford Hospital and Michigan State University for providing the raw data used in this study. I’d also like to thank the members of the research committee: Rachel Fisher, Jackie Roberson, Suzanne Diment, and Claudia Holzman for their input on the interpretation and cleaning of the raw data. The final draft of this thesis would not have been possible without the feedback and constructive comments from my thesis advisor, Claudia Holzman, and the other members of my thesis committee: Rachel Fisher, Jack Lorenz, and Dorothy Pathak. Thank you for focusing me on areas in need of elaboration and clarification. Finally, I’d like to acknowledge the faculty and staff of the Program in Epidemiology. I have thoroughly enjoyed my graduate education in this superb program. Your dedication to educating students and supporting their needs is commendable. As the first graduate of the program, I’d like to say, “We’re on our way!”. iv TABLE OF CONTENTS LIST OF TABLES .................................................................................. vii LIST OF FIGURES ................................................................................ viii CHAPTER 1 BACKGROUND INFORMATION ................................................................ 1 Alpha-fctoprotein ............................................................................ 1 Human chorionic gonadatropin ............................................................ 3 Unconjugated estriol ........................................................................ 4 Evolution of the triple test .................................................................. 5 Current debate in triple test screening ................................................... 10 Current techniques in triple test screening .............................................. 13 CHAPTER 2 PREVIOUS LITERATURE ....................................................................... 16 Elevated AFP and adverse pregnancy outcomes ....................................... 16 Elevated hCG and adverse pregnancy outcomes ....................................... 24 Both elevated AFP and elevated hCG and adverse pregnancy outcomes ........... 31 CHAPTER 3 OUR STUDY ........................................................................................ 37 Introduction ................................................................................. 37 Methods ...................................................................................... 37 Statistical methods ......................................................................... 39 Results ....................................................................................... 41 Discussion ................................................................................... 48 Future directions ........................................................................... 54 APPENDDI A FOLLOW-UP INFORMATION FORM, MICHIGAN STATE UNIVERSITY . PRENATAL SCREENING PROGRAM ........................................................ 57 APPENDD( B FOLLOW-UP INFORMATION FORM, HENRY FORD HOSPITAL PRENATAL SCREENING PROGRAM ........................................................................ 58 V APPENDIX C INFANT ABNORMALITY CATEGORIES .................................................... 59 APPENDIX D MATERNAL COMPLICATIONS PRIOR TO PREGNANCY CATEGORIES ........... 60 LIST OF REFERENCES ........................................................................... 61 vi LIST OF TABLES TABLE 1 SUMMARY OF ARTICLES REPORTING ON ELEVATED AFP LEVELS AND ADVERSE PREGNANCY OUTCOMES ............................................... 17 TABLE 2 SUMMARY OF ARTICLES REPORTING ON ELEVATED hCG AND ADVERSE PREGNANCY OUTCOMES ...................................................... 25 TABLE 3 SUMMARY OF ARTICLES REPORTING ON ELEVATED AFP AND ELEVATED hCG WITH ADVERSE PREGNANCY OUTCOMES ....................... 32 TABLE 4 MATERNAL AND PREGNANCY CHARACTERISTICS BY SCREENING STATUS ........................................................................... 41 TABLE 5 SPONTANEOUS ABORTION, STILLBIRTH, AND INFANT DEFECTS BY SCREENING STATUS ........................................................................... 42 TABLE 6 FETAL GROWTH AND PRETERM DELIVERY FOR ALL LIVE BIRTHS BY SCREENING STATUS ...................................................................... 44 TABLE 7 FETAL GROWTH AND PRETERM DELIVERY FOR LIVE BIRTHS WITHOUT FETAL DEFECTS BY SCREENING STATUS ................................ 46 TABLE 8 ANY ADVERSE OUTCOME BY SCREENING STATUS ........................... . ...... 47 vii LIST OF FIGURES FIGURE 1 MEDIAN AFP VALUES FOR THE MICHIGAN STATE UNIVERSITY PRENATAL SCREENING PROGRAM ....................................................... 2 FIGURE 2 MEDIAN hCG VALUES FOR THE MICHIGAN STATE UNIVERSITY PRENATAL SCREENING PROGRAM ....................................................... 4 FIGURE 3 MEDIAN uE3 VALUES FOR THE MICHIGAN STATE UNIVERSITY PRENATAL SCREENING PROGRAM ....................................................... 5 viii Chapter 1 BACKGROUND INFORMATION Determining a method for detecting women with pregnancies at risk for chromosomal and structural abnormalities as well as other adverse outcomes has always been of primary importance to researchers in fetal development. By identifying women particularly at risk for certain abnormalities, medical care and interventions can be concentrated on those who need them. Today one of the most important of these screening procedures is the triple test which is used to screen for women with pregnancies at increased risk for trisomies 21 (Down syndrome) and 18 and neural tube defects (NTD’s) such as anencephaly, encephalocele, and open spina bifida. The triple test measures three separate analytes found in maternal serum: alpha-fetoprotein (AFP), human chorionic gonadotropin (hCG), and unconjugated estriol (uE3). Alpha-fetoprotein AFP was first identified independently by Bergstrand (2) and Gitlin (35) in 1956. This glycoprotein is primarily synthesized by the fetal liver and to a lesser extent the yolk sac and fetal gastrointestinal tract. AFP is a single polypeptide chain with several isoforrns. AFP is analogous to albumin in the adult, although its function is not fully understood. Early in fetal life, AFP is the predominant circulating fetal protein (2). It reaches the maternal circulation via the placenta or diffusion across the fetal membranes. AFP is measured by commercially prepared radioimmunoassay (RIA) or . enzymeimmunoassay (EIA) kits using monoclonal or polyclonal antibodies to AFP. The typical non-pregnant woman might have a serum AFP concentration of 1 ug/L, with an 2 upper limit of 5 ug/L (48). However, during pregnancy AFP produced by the fetus passes into maternal circulation. The concentration of AFP in maternal serum is approximately 1/1,000 (48) of that found in the amniotic fluid in midtrimester. In normal pregnancies, about 2/3rds of the AFP in maternal circulation crosses transplacentally and the rest crosses transamnionally (44, 62). Measurable concentrations of AFP can be detected in maternal serum beginning at the end of the first trimester. Maternal serum levels rise steadily during pregnancy to a peak of approximately 500 ug/L (in normal pregnancies) at around 32 weeks of gestation (48). Thereafter levels decline steadily until term. During the typical screening period (15 to 20 weeks gestational age), normal median maternal AFP levels range from around 20 ug/L at 15 weeks to 45 ug/L at 20 weeks (106). The increase is about 15 percent per gestational week (48). Figure 1 presents cross-sectional Figure 1. Median AFP values for the Michigan State University Prenatal Screening Program'. -1y—-non4flack -—|p—bwck AFP (ML) 13 14 15 16 17 18 19 20 21 22 23 24 (imlnfionalflmnDONooki) 'Adjusted for maternal weight. 3 data from the Michigan State University Prenatal Screening Program showing weight- adjusted median maternal concentrations of AFP at each gestational week of testing. Human chorionic gonadatropin hCG is a sialoglycoprotein that is composed of two subunits of unequal size, a and [3. The Ot-subunit is nearly identical to that of the pituitary hormones hFSH and hTSH, and the B-subunit is similar to that of the pituitary hormone hLH (98). These subunits exist either in free form or are noncovalently bonded to form intact hCG. This placental steroid is produced specifically by placental syncytiotrophoblasts which are in direct contact with maternal blood. The physiological action of hCG in early pregnancy is to sustain the corpus luteum beyond its normal lifetime. In this way, production of estrogen and progesterone, which are needed for the maintenance of the endometrium, continues, thus allowing the pregnancy to proceed. A common method of measuring maternal serum hCG is to use an immunoradiometric kit which detects free-[3 as well as intact hCG (sometimes called total hCG). Detectable amounts of hCG may be found in maternal serum around the time of implantation. Levels in maternal serum rise rapidly to a peak during the third month of gestation and then decrease substantially until about 20 weeks of gestation, after which hCG levels remain relatively consistent until delivery (98). Levels in the second trimester range from approximately 5,000 to 50,000 mIU/mL (106). Figure 2 shows cross- sectional maternal weight-adjusted median serum hCG concentrations from the Michigan State University Prenatal Screening Program during the testing period. 4 Figure 2. Median hCG values for the Michigan State University Prenatal Screening Program‘. 8 M36 (IUImL) 8 15 16 17 18 19 20 21 22 Goatlonol Ago (Woolto) Adjusted for maternal weight. Unconjugated estriol Estriol is one of the three major naturally occurring estrogcns. It is produced almost exclusively during pregnancy, and it is the major estrogen produced in the human fetus (26). Unconjugated estriol is a steroid product of the fetoplacental unit and is synthesized in a series of steps beginning with the synthesis of dehydroepiandrosterone sulfate (DHEAS) from cholesterol in the fetal adrenal. Then, in the fetal liver, DHEAS is transformed into 160t-OH-DHEA and finally metabolized into uE3 in the placenta. During pregnancy, the production of uE3 depends on an intact matemal-placental-fetal unit, and therefore uE3 is a marker closely correlated with fetal and placental growth. uE3 can be measured using radioimmunoassay (RIA) testing kits, and detectable amounts are found in maternal serum starting in the first trimester. Fetal-placental 5 production of uE3 leads to a progressive rise in maternal circulating uE3 levels during pregnancy. uE3 levels reach a peak which is about two to three orders of magnitude greater than non pregnant levels in late gestation (15, 16). During the second trimester, concentrations range from around 0.2 to 0.5 ug/L and levels increase approximately twenty percent per gestational week (106). Cross-sectional maternal weight-adjusted median uE3 values from the Michigan State University Prenatal Screening Program for each gestational week of screening are presented in Figure 3. Evolution of the Triple Test AFP screening was the first element of the triple test to come into place. It was brought about by the recognition in the early 1970’s that significant AFP elevations in some women were associated with abnormal pregnancy outcomes such as fetal distress, fetal demise, anencephaly, and open spina bifida (10, 11, 96, 97, 109). A study by Brock Figure 3. Median uE3 values for the Michigan State University Prenatal Screening Program’. uEs (nymL) 15 16 17 18 19 20 21 22 Geuuonel Age (Weeks) rAdjusted for maternal weight. 6 et al. (11) particularly ushered in the present era of prenatal screening with the discovery that anencephaly and open spina bifida were associated with increased levels of maternal AFP. The association between elevated maternal serum AFP and NTD’s was particularly important because NTD’s are relatively common and serious malformations. NTD’s result from a failure in the closure of the developing neural tube during the first month of pregnancy. Depending on the cut-off used, 90 percent of anencephaly and 75-85 percent of spina bifida can be detected by elevated maternal serum AFP levels (48). NTD incidence varies by geographic location. The incidence in England is 2.95 per 1,000 (19), while in the United States the incidence is approximately 1 or 2 per 1,000 (92, 31, 78). The discovery of this association lead to a large collaborative study in the United Kingdom (116, 108) and a concurrent study by Macri et al. (69) in the United States. These studies unequivocally demonstrated the value of maternal serum AFP as a screening test for NTD’s. Since there is overlap in AFP levels in normal and NTD pregnancies, however, AFP cannot be used alone as a diagnostic test. By the mid 1970’s to early 80’s, large screening programs were in place for the detection of NTD’s, and the American College of Obstetrics and Gynecologists recommended in 1986 that the assessment of AFP levels be offered to all pregnant women (85). The quest for a method of detecting women at increased risk for Down syndrome led to the remaining elements of the present triple test. The association between maternal age and the risk for Down syndrome formed the basis of the first diagnostic testing for Down syndrome. Women thirty-five or older are at an increased risk of having a Down syndrome fetus. At thirty-five years the Down’s risk during midtrimester is l in 270, and 7 by 45 years the risk increases to 1 in 20 (48). From the late 1960’s, when techniques for culturing cells became available, to the early 1980’s, amniocentesis was routinely offered to women who would be thirty-five or older at delivery. This was an inefficient screening strategy, however, because women thirty-five or older only comprise approximately five percent of all pregnancies (80) and up to eighty percent of Down syndrome infants are born to women under thirty-five years of age (126). In addition, amniocentesis is an invasive procedure and in a small percentage can produce fetal loss. Using maternal age as a screening procedure for Down syndrome only allowed a detection rate of 20 to 45 percent (60, 80). In 1984 Merkatz et al. (73) discovered that low values of AFP, which was already being used to screen for N'I'D’s, were associated with an increased risk of Down syndrome in the fetus. They observed that 25 percent of pregnancies with autosomal trisomies had maternal serum AFP values less than or equal to 0.4 multiples of the median (MoM), whereas only 11.2 percent of the normal pregnancies had values below this level. This opened the door for the first screening program for all women to detect Down syndrome. Subsequent tests on this screening procedure confirmed these results and found that use of AFP in combination with maternal age allowed the detection of 20 to 33 percent of Down Syndrome cases in women less than 35 years at delivery (82, 25, 27). Since AFP is not related to maternal age (113), elevated levels would also detect 20- 30 percent in all women regardless of age. The recognition that maternal serum AFP levels were altered in the presence of a fetus with Down syndrome led to the investigation of a number of other potential serum 8 markers including hCG and uE3. A test even more effective than AFP for Down syndrome screening was discovered by Bogart et al. (4). In their study 11 out of 17 women (65 percent) with Down syndrome fetuses had elevated levels of intact hCG while only 1 among 74 unaffected pregnancies were elevated. They found that hCG levels were 2 to 2.5 times higher in women with a Down syndrome fetus at 18 to 22 weeks of gestation. Further studies followed confirming these results, and of 18 studies published between 1989 and 1991, the geometric mean hCG level was 2.05 times higher in Down pregnancies than that in unaffected pregnancies of the same gestational age (115). Then, in 1988 the final element of the triple test came into use when Canick, et al. (17) noted that lower than average uE3 values in the second trimester were associated with Down syndrome. They found that of the 22 affected pregnancies, 16 (73 percent) had uE3 MoM values less than normal (1.0) with the median MoM being 0.79. Additionally, Haddow et al. (46) reported a 6 percent reduction in Down syndrome detection rates when uE3 was eliminated from the screening procedure. Of the three biochemical markers, it is widely accepted that the best predictor of Down syndrome pregnancies is hCG, followed by uE3 and then AFP (113, 103). Though the three markers had each been shown to be associated with Down syndrome in the fetus, there was no evidence that combining them into one overall ‘risk score’ would be beneficial. In 1988, however, Wald et al. (113) proposed a combined risk assessment using AFP, hCG, uE3, and maternal age. They published a multivariate statistical model for evaluating levels of all three analytes to calculate a single patient-specific risk. The results of the three tests were combined into one overall ‘risk score’ which assessed the 9 chance of that woman delivering a Down syndrome infant. This was not a diagnostic test for Down syndrome, but rather the results indicated the level of risk for a particular pregnancy. Using the model they developed and a risk cut-off for Down syndrome of 1 in 250, Wald et al. retrospectively tested maternal serum samples from 77 affected pregnancies and 385 unaffected pregnancies. They were able to detect 67 percent of Down syndrome fetuses with a false positive rate of 5 percent. This was a significant improvement from the detection rates possible with only maternal age. Furthermore, they found that the levels of uE3, hCG, and AFP were independent of maternal age and only weakly correlated with each other. Using this information they were able to determine that the three measurements provided largely independent information, meaning that the three analytes used together in combination would be a more sensitive Down syndrome screening test than any of the individual analytes. Additionally, they found that with the detection rate remaining constant, the false-positive rate decreased with the addition of each marker, hCG causing the most dramatic reduction. Therefore the combined triple test had higher specificity as well as sensitivity. The triple test methods developed by Wald et al. (113) have since been subjected to prospective trials. Wald et al. (119) performed their own prospective trial and found a detection rate of 48 percent with a false positive rate of 4.1 percent. The first trial in the United States by Haddow et al. (46) used a risk cut—off of 1 in 190 and detected 66 percent of Down syndrome births. When they considered the reported incidence of spontaneous abortion of Down syndrome infants between mid-trimester and term, the true detection rate proved to be 58 percent, with 6.6 percent of all women testing positive. 10 The second trial in the United States by Phillips et al. (84) restricted the study population to those women under 35 years and used a risk cut-off of 1 in 274. They had a screen positive rate of 7.2 percent and were able to identify 57 percent of Down syndrome cases. Additionally, a study by Burton et al. (12) used a cut-off risk of l in 270 (equal to that of a woman 35 years or older). Their study found a screen positive rate of 10.4 percent, and they were able to detect 83 percent of Down syndrome births. When accounting for spontaneous abortions, the detection rate they reported was 67 percent. These studies confirmed the benefit of the triple test in Down syndrome detection. The major advantage of the triple test is that women of all ages can be screened without invasive procedures, thus allowing for prenatal identification of a greater proportion of Down syndrome cases. Chitty et al. (20), in their review of current techniques in screening, concluded that age alone can detect 30 percent of Down syndrome cases, age and hCG, 49 percent; age, hCG, and AFP, 56 percent; age, hCG, and estriol, 57 percent; and age, hCG, estriol, and AFP can detect 61 percent of Down syndrome cases. It is important to realize that even with a 61 percent detection rate, the triple test is not a diagnostic test for Down syndrome, but rather it is a screening test that defines a population of women who may require additional testing. Current Debate in Triple Test Screening Since the publication of Wald et al. (113), a number of studies have attempted to evaluate the extent to which the addition of hCG and uE3 measurements improve screening sensitivity. These works have lead to two controversies: first whether or not 11 the addition of uE3 adds to the sensitivity, and second whether free-B hCG is a superior marker compared to total hCG. Since Canick et al. (17) and Wald et al. (113) reported that uE3 was useful in screening for Down syndrome, several studies have questioned those findings (66, 67, 64, 102, 24, 60). Some studies found no improvement in detection rates for Down syndrome with the addition of uE3 (66, 67), and one study (24) found a decrease in detection rate with the addition of uE3 to the screening procedure. These studies are in the minority, however, and the findings could be due to a study population that is not representative of the screening population as a whole. Most research has shown that uE3 does improve detection rates. The debate lies in whether this small improvement in detection rates is cost effective given the correlation of uE3 with AFP levels (60). However, since uE3 variation within gestational week is smaller than that of the other markers, including it in the triple test can provide an additional check on gestational age accuracy. Therefore uE3 can be a useful measure regardless of detection rates. The second triple test debate that has arisen is whether free-B hCG is a better predictor of risk than total hCG. Macri et al. (65) measured the free-B protein in maternal sera from 29 cases of Down syndrome and 450 unaffected pregnancies. They found a significant increase in the level of free-B hCG in Down syndrome pregnancies. 83 percent of the Down pregnancies exceeded the median of the controls, and 52 percent exceeded twice the median. When the free—B levels were combined with AFP levels and 12 maternal age in pregnancies under 17 weeks of gestation, a detection rate of 80 percent with a false positive rate of 5 percent was reported. Spencer et al. (103) directly compared free-B hCG with total hCG univariately and in a multivariate risk analysis with AFP, uE3, and maternal age in 29 Down pregnancies and 145 controls. They found the median free-B subunit from affected pregnancies to be 2.06 times that of unaffected pregnancies, and total hCG to be 1.88 times higher. The univariate analysis found a 34 percent detection rate with the free-B subunit and 29 percent with total hCG. In the multivariate analysis, at a false positive rate of 5.9 percent, they found a detection rate of 66 percent for the free-B subunit compared with only 52 percent for total hCG measurement. The differences, however, were not statistically significant. This led Spencer et al. to conduct the large multicenter study mentioned earlier (102). In addition to their findings that uE3 was not useful, this study confirmed their earlier findings regarding free-B hCG. Comparing the markers against one another for effectiveness, they found that at a false positive rate of 5 percent the relative detection rates were AFP, 7 percent; uE3, 18 percent; maternal age, 28 percent; total hCG, 38 percent; and free-B hCG, 45 percent. They determined that the best combination of the various markers was free-B hCG, AFP, and maternal age, with a 66 percent detection rate at a false positive rate of 6 percent. This finding was significant, and the authors concluded they had clear evidence to justify the claim that free-B hCG was a superior marker to total hCG. 13 Wald et al. (118) went even further by testing separately for both free-[3 hCG and free-or hCG. In their study of 75 singleton Down pregnancies and 367 unaffected singleton pregnancies using a four-marker serum screening test (AFP, uE3, free-B hCG, free-0t hCG, and maternal age), they were able to detect 65 percent of Down syndrome pregnancies with a 5 percent false-positive rate. This was in comparison to a detection rate of 59 percent with the conventional triple test. The authors concluded that this four- marker test is the most effective method of prenatal screening for Down syndrome suitable for routine use. More research will be needed examining the four marker test before any conclusions about its effectiveness can be drawn, however. Current techniques in triple test screening Triple test screening is usually performed between 15 and 22 weeks gestation, with the optimal time being from 16 to 18 weeks gestation. Results of the test are commonly expressed as multiple of the median (MOM). This allows for comparison across different testing sites even though they may use different testing kits with varying levels of detection. It also allows for comparison of women tested at different gestational ages given that the normal values of the analytes change during pregnancy. The distribution of AFP, hCG, and uE3 values for women with uncomplicated pregnancies are analyzed at each testing center for each gestational week (or day) during the normal testing period. The median value from these distributions is identified, and it is assigned a value of 1.0. Every other value for that gestational week (or day) from that center is then expressed as a multiple of its median. 14 As previously mentioned, maternal age and accurate gestational age estimates are important covariates in determining the risk for Down syndrome and NTD’s in a fetus, and they are therefore taken into account when performing the triple test. Other variables can also affect the triple test for singleton pregnancies and need to be taken into account when interpreting the results. The most important of these variables are maternal weight, maternal race, and maternal insulin-dependent diabetes. All three markers are inversely associated with maternal weight (106). Heavier women have, on average, lower concentrations of AFP, hCG, and uE3 than lighter women. In other words, if no adjustment is made, too many lighter women and too few heavier women will have values above any given MoM risk cut-off value. This can be explained by the larger blood volume in heavier women (42, 110). It has been determined that this adjustment will reduce the false-positive rate by about 15 percent while not adversely affecting the detection rate (53). Maternal race also affects the levels of the three analytes in maternal serum. Serum samples from black women contain AFP concentrations that are approximately 10 to 15 percent higher than measurements from other racial groups (52). In other words, when screening for NTD’s and Down syndrome, too many blacks and too few whites would have results above the median if a single MoM scale was used. HCG levels appear to be higher in blacks, but this finding is not consistent among researchers (106). A study by Bogart et al. (3) found no significant differences in hCG levels between blacks, whites, and Hispanics; however the weight adjusted MoM’s of blacks were 9.8 percent higher and Orientals 16 percent higher than those for whites. uE3, on the other hand, may be 15 reduced by 5 percent in blacks (106). Because of this, most testing centers calculate MoM’s based on separate medians for blacks and non-blacks for all three analytes. Insulin-dependent diabetes has also recently been found to affect test results, although gestational diabetes does not appear to be associated with variations in concentrations (48). Women with insulin-dependent diabetes have been found to have AFP values that average 20 percent lower at any given time in the second trimester than women who do not have diabetes (111, 39). uE3 and hCG values are also lower, by about 10 percent, in women with insulin-dependent diabetes (106). This means that when assessing results using MoM’s, too few women with insulin-dependent diabetes will have values considered to be elevated if no adjustment is made. Because of this, it has now been recommended that corrections for insulin-dependent diabetes be instituted (106, 59). Other factors have also been found to affect analyte levels. AFP levels in multiple pregnancies are higher than singleton pregnancies by approximately the number of fetuses (117). Additionally, some drugs can affect AFP levels, so it is advisable to have women stop taking medications at least 12 hours in advance of having their blood drawn, if possible (106). Chapter 2 PREVIOUS LITERATURE Since the establishment of the triple test as a screening procedure for NTD’s and Down syndrome, many researchers have investigated the relationship between these analyte levels and other adverse pregnancy outcomes. To date, most research has focused on elevated AFP levels, although some research has been done on elevated hCG levels and to a lesser extent on pregnancies with both AFP and hCG elevated. Elevated AFP and adverse pregnancy outcomes The first studies associating unexplained AFP elevations with adverse pregnancy outcomes focused on low birthweight (LBW), which is a composite of both preterm delivery and fetal growth. Brock et al. (7, 9) first reported that women with elevated AFP levels not explained by multiple gestations, open neural tube defects, or demise were at increased risk of delivering a low birthweight infant. He found a LBW rate of 10.7 percent among infants born to women with AFP greater than 2.3 MoM compared to a background rate of 4.2 percent. Another early study by Wald et al. (1 14) used AFP greater than 3 MoM as a definition of elevated, and they too found an association with LBW (relative risk (RR) 4.7). In addition, Wald et al. found an increased risk of preterm birth (RR 5.8) and perinatal death (RR 3.5) among infants of women with elevated AFP. These reports initiated further investigation on the association of LBW and other adverse pregnancy outcomes with elevated AFP levels. Studies from 1977 to 1991 which assessed these associations are summarized in table 1. l6 17 Table 1. Summary of articles reporting on elevated AFP levels and adverse pregnancy outcomes. Study Year AFP LBW IUFD Peri D I’l'D Cut-off (LB) Brock et al. (7) 1977 2.3 + + + NR Wald ct al. (1 14) 1977 3.0 4» NR + + Macri et al. (68) 1978 2.0 + NR NR + Gordon et al. (37) 1978 2.0 - NR + + Brock ct al. (6) 1979 2.0 + + NR NR Brock et al. (8) 1980 2.0 + NR NR + Wald et al. (112) 1980 2.0 + NR NR NR Smith ct al. (100) 1980 2.0 + NR NR + Read or al. (87) 1980 2.0 + + NR + Macri et al. (70) 1982 2.0 NR + + NR Persson ct al. (83) 1983 2.3 + NR + + Haddow ct al. (41) 1983 2.0 + + NR NR Purdic er al. (86) 1983 2.5 + NR NR - Crandall et al. (22) 1983 2.0 + + + NR Evans or al. (29) 1984 3.0 + NR + NR Schnittger or al. (94) 1984 2.0 + NR NR NR Milunsky et al. (74) 1984 2.0 - «1» NR NR Hamilton ct al. (49) 1985 2.5 + + + + Furhrnan et a1. (33) 1985 2.5 + + NR NR Sccher ct al. (95) 1985 2.0 - NR NR NR Ghosh ct al. (34) 1986 2.0 + NR + NR Haddow ct a1. (47) 1987 2.0 + NR NR NR Doran et al. (28) 1987 2.5 + + + NR Nelson ct al. (81) 1987 2.5 NR + NR NR Burton et al. (13) 1988 2.5 + + + NR Robinson ct al. (89) 1989 2.5 + + NR NR Milunsky ct al. (75) 1989 2.0 + + + NR Crandall et al. (23) 1991 2.5 + + NR + Waller ct al. (120) 1991 2.0 NR + NR NR LB = live births only. + = risk elevated. - = no elevation in risk. NR = not reported. 18 Of the twenty-nine studies examined, twenty-three found that elevated AFP was associated with an increased risk of LBW, while only three studies failed to find an increased risk. Twenty-one studies reported relative risks or presented data from which relative risks could be calculated. The relative risks found in these studies ranged from 2 to 10. The nine studies using a cut-off of 2.0 MoM had relative risks ranging from 2 to 5. When a cut-off of 2.5 MOM was used, relative risks for eight studies ranged from 2 to 10. In the four studies which examined a cut-off of 3.0 MoM, the relative risks ranged from 3 to 5.2. Sixteen studies reported an association between elevated AFP and intrauterine fetal death (IUFD). In the eleven studies where relative risks were reported or could be calculated, risks for IUFD ranged from 3.2 to 21.0. Among the four studies which examined a cut-off of 2.0 MoM the relative risks ranged from 2 to 8. When a cut-off of 2.5 MoM was used, relative risks found in five studies ranged from 4.4 to 21. The two studies which examined a cut-off of 3.0 MoM found relative risks of 8.5 and 10.4. One of the problems, however, with evaluating IUFD is the timing of the death in relation to the triple test measurement, since unrecognized fetal demise could be the cause of the elevated analyte levels. Therefore in order to determine the true association of elevated levels with IUFD, studies must take this into account. There were also twelve studies which reported on the association between elevated AFP levels and neonatal or perinatal death. Nine of these studies reported relative risks or provided data from which they could be calculated. In these nine studies, relative risk estimates ranged from 3 to 10. When a cut-off of 2.0 MoM was used, the 19 relative risks from three studies ranged from 4.5 to 5.8. In the four studies that examined a cut-off of 2.5 MoM, relative risks ranged from 6 to 10. The two studies which examined a cut-off of 3.0 MoM found relative risks of 3.5 and 8. Nine studies reported on an association between elevated AFP levels and preterm delivery (PTD), while only one study failed to find an association. Four studies provided data from which relative risks were calculated. Gordon et al. (37) used a cut-off for elevated AFP of 2.0 MoM, and they reported a relative risk of 3.5. Persson et al. (83), who used a cut-off of 2.3 MoM found a relative risk of 2, and Hamilton et al. (49) reported a relative risk greater than 10 when a cut-off of 2.5 was used. Wald et al. (114) examined preterm delivery using an AFP cut—off of 3.0 MoM, and they found a relative risk of 5.8. The main concern in studies reporting on preterm delivery is the accuracy of gestational age reporting. If gestational age estimates are inaccurate at the time of testing, they can affect both the analyte test results and the estimated gestational age at birth. Katz et al. (54) reviewed eighteen studies that examined adverse pregnancy outcomes in women with elevated maternal serum AFP levels. These studies combined reported on over 225,000 women with AFP screening. They determined that the risk of low birthweight was increased 2 to 4 times in women with unexplained AFP levels greater than 2 MoM. Additionally, they reported that the risk of stillbirth was increased 2 to 8 times. Approaching the data from another perspective, Katz et al. determined that women with an unexplained AFP elevation have between a 10 and 33 percent chance of delivering an infant weighing less than 2,500 grams and a 20 to 38 percent chance of having an adverse pregnancy outcome (LBW, IUFD, IUGR, or preterm delivery). 20 Another review of the literature regarding elevated AFP and adverse pregnancy outcomes was done by Barbara Burton (14). She found that most studies examining low birthweight and high AFP report relative risks of around 2—5. She also found that approximately 15 percent of pregnancies that will ultimately deliver low birthweight infants will have elevated midtrimester AFP levels. However, if the pregnancy is complicated by IUGR, this detection rate could fall as low as 7 percent. Associations with fetal loss can often be complicated by unrecognized fetal demise at the time of testing, however, since this could be the reason for the elevated analyte levels. Burton concluded, however, that the evidence indicates that even with a viable fetus at the time of testing, there is an increased risk for later fetal loss or stillbirth when AFP levels are elevated. The relative risk ranged from 4.7-12.6. Burton also concludes that the positive predictive value for low birthweight appears to be relatively low (8-30% depending on the cut-off used). Waller et al. (121) published another review of the studies examining elevated AFP and adverse pregnancy outcomes. They focused their review on the question of whether pregnant women with high levels of AFP who have singleton fetuses without birth defects have an increased risk of low birthweight and perinatal death. Twenty-three studies were reviewed, and eighteen which had comparable measurements were reported on. In order to make these studies comparable on a consistent basis, they recalculated risk estimates in some studies which had failed to exclude conditions which could have explained the elevation in AFP, such as relevant birth defects, multiple gestations, missed abortions, and cases of underestimated gestational age. Waller et al. found eleven studies 21 that concurred in reporting an increased risk of fetal death in women with high levels of AFP. If a cut-off of 2.0 MoM was used, estimates of risk ranged from 3.2 to 4.4. For cut-offs of 2.5 and 3.0, relative risks ranged from 4.4 to 21 and 8.5 to 10.4, respectively. Fourteen studies reported an increased risk of low birthweight (<2,500 grams) in women with elevated AFP. For the cut-offs of 2.0, 2.5, and 3.0 the range of relative risks found were 2.8-5.0, 1.9-6.4, and 4.1-5.2, respectively. Estimates of risk in seven studies reporting an association between elevated AFP and neonatal death were 4.9 and 5.8 for a 2.0 MoM cut-off and 4.0 and 6.0 for a 2.5 MoM cut-off. In addition, all studies that examined more than one level of high AFP found a dose-response relation between AFP level and LBW incidence. The authors concluded that elevated AFP is predictive of women at risk for adverse pregnancy outcomes. Since these reviews several studies have been published examining the outcome of having a small for gestational age (SGA) infant, and its association with elevated AFP levels. In 1992, Capeless et al. (18) prospectively followed 381 screened women. They identified 23 (6 percent) women with AFP levels greater than 2.0 MoM. Fifteen percent of the elevated women delivered a SGA infant compared to only 6 percent of the comparison p0pulation. This translated into a relative risk for delivering a SGA infant of 2.54 in women with elevated AFP levels. In addition to these findings, Capeless et al. also reported an increased risk for preterm delivery (RR 6.35) and IUFD (RR 5.38) in women with AFP elevated above 2.0 MoM. Their study also highlighted the fact that AFP is not an effective screening test specifically for adverse pregnancy outcomes by pointing out that the positive predictive value was 49 percent and the sensitivity was 16 22 percent. In multivariate regression, elevated AFP remained significant meaning that it is an independent variable in adverse pregnancy risk assessment. The authors concluded, however, that it was unclear whether elevated AFP identified a new population at risk for these outcomes because of the high prevalence of historical risk factors such as previous SGA infant, PTD, IUFD, and pregnancy complications in their population. Morssink et al. (77) published a study in 1995 that examined SGA infants and preterm delivery in women who had AFP levels 2 2.5 MoM. They examined adverse pregnancy outcomes which included extremely SGA infant (< 2.3 percentile), SGA infant (<10th percentile), preterm delivery (< 37 weeks), and either SGA infant or preterm delivery. They excluded pregnancies with unknown outcome, a congenital anomaly, delivery before 25 weeks of amenorrhea, or known insulin-dependent diabetes. Separate analysis was done in those who had elevated hCG levels as well as elevated AFP levels. They found that 9.4 percent of women with elevated AFP levels had extremely SGA infants compared to 2.1 percent of the comparison population. The relative risk was 4.5 (p<.01). For SGA infant, 27.1 percent of the elevated population and 9.9 percent of the comparison population met the criteria, for a relative risk of 2.7 (p<.01). Fourteen percent of the elevated population versus 5.9 percent of comparisons had preterm delivery. The relative risk was 2.4 (p<.01). For either SGA infant or preterm delivery, 37.5 percent of the elevated population and 15.2 percent of the comparison population fell into this category. The relative risk of either having an SGA infant or a preterm delivery for a woman with elevated AFP was 2.5 (p<.01). 23 Additionally, Silver et al. (99) in 1994 studied elevated AFP specifically in women with antiphospholipid antibodies. In their cohort of sixty pregnancies in women with median to high levels of IgG anticardiolipin antibodies, lupus anticoagulant, or both they found 13 (22 percent) had elevated AFP. None of the elevated AFP levels were explained by fetal anomalies, current fetal death, multiple gestation, incorrect dates, or vaginal bleeding. They found that pregnancies with elevated AFP values had a significantly higher incidence of fetal death (62 percent) than those without elevated AFP values (6 percent). Also, 77 percent of elevated AFP pregnancies versus 15 percent of non-elevated pregnancies resulted in perinatal loss. Both of these findings were statistically significant (p<.001). In addition to the adverse pregnancy outcomes discussed above, countless studies have also found associations between various pregnancy complications and elevated AFP levels. The complications that have been found in at least one study to be associated with elevated AFP levels include: abruptio placenta (18, 86, 75, 49, 83), maternal infection (48, 75), ologiohydramnios (63, 32, 55, 105), preeclarnpsia (122, 76), and pregnancy- induced hypertension (76, 49). The mechanism by which unexplained AFP levels are elevated in abnormal pregnancies has not been conclusively proven. Maternal levels of AFP may result from transport across either chorioamniotic membranes or the placenta. Since problems associated with elevated AFP levels can be related to uteroplacental disease, most investigators believe that pathology within the placenta leads to an increase in transfer of AFP across the matemal-fetal interface in women with unexplained AFP elevations. 24 There are several lines of support for the hypothesis that elevated AFP is a result of leakage through a defective feto-matemal placental barrier. First of all, most women with unexplained maternal serum elevations of AFP have normal amniotic fluid AFP levels (108). Secondly, a strong association between elevated AFP levels and fetal- matemal hemorrhage has been reported (50, 56, 61). Another line of support for this theory is that a strong relationship between vaginal bleeding and elevated AFP has been demonstrated (43, 83, 57). The fourth piece of supporting evidence is that several placental lesions (chorioangiomas, hematomas, hemangianomas) as well as chronic villitis have been associated with elevated maternal serum AFP levels (71, 91, 104). Finally, elevated levels of fetal red blood cells have been demonstrated in maternal serum when AFP levels are elevated (48). Therefore, because of this evidence, it appears that in most cases of unexplained AFP elevations the integrity of the uteroplacental interface has been breached allowing more fetal AFP to enter the maternal serum. Elevated hCG and adverse pregnancy outcomes Although the research regarding AFP and adverse pregnancy outcomes still far outweighs that for elevated hCG, recently several studies have reported on hCG levels and adverse pregnancy outcomes. These studies have become more focused and reliable as knowledge of hCG and testing strategies have improved. In all, eight articles have examined various adverse fetal outcomes. Table 2 summarizes the most common outcomes examined in these studies. In one of the first prospective studies of hCG as a marker for detecting Down syndrome, Bogart et al. (3) found that hCG was also a marker for other adverse pregnancy ‘7‘. 25 Table 2. Summary Of articles reporting on elevated hCG and adverse pregnancy OUICOIIICS. Study # of Cases Case Def. IUFD LBW (LB) P'I'D (LB) SGA (LB) Bogart et al. 255 >2.0 MOM l 1.6% 28.8% NR NR 1991 (3) Gravett et al. 7 >50 MOM NR 33% 50% NR 1992 (38) Goncn et al. 284 >2.5 MOM RR 4.1 NR RR 1.1 NR 1992 (36) (0.5-34.8) (0.6-2.2) >4.0 MOM NR NR RR 3.3 NR (1.3-8.2) Muller et al. 657 >95m RR 1.61 NR NR NR 1993 (79) percentile (0.64.4) Liepprnan et al. 225 22.0 NR RR 4.7 RR 2.9 RR 1.8 1993 (58) (1.8-12.2) (1.4-6.1) (LO-3.2) Tanaka et al. 42 >20 RR 28 NR NR RR 4.9 1993 (107) (2.6-306) (2.4-10.4) Wenstrom et al. 22 22.0 13.6% NR 18.2% NR 1994 (124) Morissink et al. 366 22.5 NR NR RR 1.4 RR 1.5 1995 (77) (0.8-2.0) (1.2-2.0) LB = live births only. NR = not reported. RR = relative risk. outcomes. In their study, 6 of 11 pregnancies with fetal aneuploidy (54.5 percent) had abnormal hCG levels (>2.5 MOM). Additionally, they found an 11.6 percent incidence of fetal death and 28.8 percent incidence Of low birthweight in women with elevated hCG levels. Inclusion of aneuploidy pregnancies could have affected these results. Also, corresponding AFP values for these women were not provided, and therefore some of these women could have had elevated AFP as well. Gravett et al.(38) published an article in which they identified seven women with hCG levels elevated above 5 MOM (one also had elevated AFP equal to 2.87 MOM). Of the six women with elevated hCG without elevated AFP, 3 (50 percent) had preterm 26 deliveries and 2 (33 percent) had low birthweight infants. This was in contrast to a preterm birth rate of 16.4 percent in a high risk comparison population. The authors concluded that since AFP was not elevated in these cases, hCG could represent an independent risk factor for adverse pregnancy outcome. The small number of cases reported on, however, limits the conclusions that can be drawn from this study. Gonen et al. (36) published an article in which they examined unexplained elevations of hCG greater than 2.5 MOM. They excluded pregnancies with fetal abnormalities, abnormal karyotypes, and a maternal serum AFP level greater than 2.5 MOM. 284 (4.7 percent) of the women screened had elevated hCG levels. They found that 10.3 percent of women with elevated hCG had pregnancies with fetal growth restriction compared to 5 percent of the normal hCG population (RR 2.8 95% CI l.0-7.0). They found no difference, however, between fetal loss rates (1.8 percent verses 0.5 percent), preterm delivery (7.0 percent verses 6.3 percent), Apgar score < 7 at five minutes (0.7 percent verses 1.8 percent), or neonatal death (0.7 percent verses 0.0 percent). However when the hCG cut-Off was raised to >40 MOM, preterm delivery became significant (RR 3.3 95% CI 1.3-8.2). The authors concluded that women with unexplained elevations of hCG are at an increased risk for adverse pregnancy complications and therefore may require careful Obstetric management to optimize pregnancy outcome. Muller et al. (79) published the results Of a cohort study examining the association of elevated maternal serum hCG levels with fetal chromosome anomalies and subsequent fetal death. The study was limited to singleton pregnancies screened between 14 and 20 27 weeks of gestation. hCG was evaluated as percentiles based on maternal age and gestational week at the time of testing. They found that women who had hCG elevations 295th percentile were at high risk of fetal and perinatal deaths. The relative risk for miscarriage was 3.0 and for in utero fetal death was 1.61. These results, however, could be partially explained by the inclusion Of Down syndrome and other aneuploidy cases in their analysis. Also, the previously mentioned problem of whether the fetal death led to the elevated level of the analyte was not considered. Liepprnan et al. (58) conducted a cohort study of the association of hCG and adverse pregnancy outcomes. Study participants were screened between 15 and 18 weeks of gestation, and they excluded women who had pregnancies with multiple gestations, insulin-dependent maternal diabetes, known fetal chromosomal abnormalities, and fetal anomalies detected by ultrasonography before screening. They examined LBW ((2500 grams), preterm delivery (<37 weeks), and SGA infant (<10‘h percentile) in 225 pregnancies with hCG 2 2.0 MOM and 235 pregnancies that screened positive for Down syndrome but had hCG < 2.0, AFP < 2.0, and uE3 > 0.5 MOM. Thirteen women with elevated hCG also had AFP 2 2.0 MOM, however excluding these women did not alter any of the analysis results, nor did differences exist when analysis was done separately for those with uE3 S 0.5 MOM and with uE3 > 0.5 MOM. Approximately 10 percent of women with elevated hCG and 2.1 percent of those in the comparison group had a LBW outcome. The relative risk was 4.7 (95% CI 1.8-12.2). After adjusting for maternal age, race, payment status, and gravidity, the relative risk was 4.0 (95% CI 1.699). Eleven percent of the elevated population and 3.8 percent of the comparison population delivered 28 prematurely for a relative risk of 2.9 (95 % CI 1.4-6.1). After adjustment, the relative risk was 2.8 (95% CI 1.4-5.8). SGA infant occurred in 13.3 percent of the elevated population and 7.2 percent of the comparison population. The relative risk was 1.8 (95% CI 1.0-3.2), and the adjusted relative risk was also 1.8 (95% CI 1.0-3.2). Trend analysis was significant for all three adverse outcomes, indicating that risk increased with increasing levels Of hCG. If hCG levels were 2.0 to 3.9 MOM, 4.0 to 5.9 MOM, or 26.0 MOM the corresponding LBW risks were 2.9 (95% CI 1.0-8.4), 8.7 (95% CI 3.2-25.1), and 20.3 (95% CI 4.0-104.0). For preterm delivery in the same categories the risks were 2.1 (95% CI 0.9-4.8), 4.6 (95% CI 2.0-10.6), and 11.8 (95% CI 2.6-54.5) respectively. For SGA births the risks corresponding to the categories were 1.3 (95% C10.7-2.5), 3.2 (95% CI 1.8-6.4), and 4.6 (95% C10.9-24.3). The authors conclude that elevated hCG in the midtrimester appear to be associated with adverse pregnancy outcome, that the magnitude of the risk correlates with hCG level, and that this risk is independent of any risks that might be associated with unexplained elevations Of AFP. Tanaka et al. (107) published a study in which 638 consecutive pregnant women were screened. Forty-two women (6.6 percent) had hCG elevations 2 2.0 MOM. SGA infant was determined using fetal growth curves constructed from Japanese data, with SGA infants defined as those below the mean minus 1.5 standard deviations. This can approximately be considered equal to the 7th percentile. They found that 19 percent of the elevated population delivered SGA infants compared to 3.9 percent of the population with hCG < 2.0 MOM. The relative risk was 4.9 (p<.001). Fetal death was also significantly associated with hCG 2 2.0 MOM. 4.8 percent Of the elevated population and 29 0.34 percent of the comparison population experienced a fetal death for a relative risk Of 28 (p<.05). Although this study failed to take into account AFP levels of the elevated hCG women, the authors concluded that an elevated hCG level may be used for selecting women at high risk for pregnancy complications. A study by Wenstrom et al. (124) examined 126 women with poor pregnancy outcomes compared to 126 women with normal outcomes. All women in their study underwent amniocentesis because of elevated analyte levels, historical risk factors, or advanced maternal age. They found that 14 percent Of complicated pregnancies versus 3 percent of normal pregnancies had elevated hCG levels (p=.01). The odds ratio associated with a poor obstetric outcome if hCG was elevated was 3.9 (95% CI 1.2-17.8). Of the 22 women with elevated hCG levels, 13.6 percent had a fetal death, and 18.2 percent delivered prematurely. Morrsink et al. (77) published a cross-sectional study examining elevated hCG and its association with extreme SGA infant (<23rd percentile), SGA infant (< 10th percentile), preterm delivery (<37 weeks, with the exclusion of infants with a birth weight below the 10th percentile), and either SGA infant or preterm delivery. Elevated hCG was defined as 2 2.5 MOM, and women with unknown pregnancy outcome, a congenital anomaly, delivery before 25 weeks of amenorrhoea, or known insulin-dependent diabetes were excluded from analysis. Women with both hCG and AFP elevated were analyzed separately. They found that 4.4 percent of those with elevated hCG versus 2.1 percent of those < 2.5 MOM delivered extremely SGA infants. The relative risk was 2.1 (p<.01), and the detection rate was 2.7 percent. For SGA infant, they reported that 15.5 percent of 30 the elevated population and 10.1 percent of the comparison population had the outcome, yielding a relative risk of 1.5 (p<.01) and a detection rate of 7.1 percent. Preterm delivery was not significantly associated with elevated hCG, with 8.6 percent of the elevated population and 5.9 percent Of the comparison population reported as delivering prematurely. They also reported that 22.7 percent of the elevated population and 15.4 percent of the comparison population had either SGA infants or preterm delivery. The relative risk was 1.5 (p<.01), and the detection rate was 6.8 percent. The authors concluded that women with elevated hCG are at increased risk for an adverse pregnancy outcome and should have some kind Of increased surveillance. Many of these above mentioned articles, as well as a few additional ones, also reported on the relationship between elevated hCG levels and various pregnancy complications. Elevated hCG was found to be associated with preeclampsia in both the study by Gravett et al. (38) and the study by Muller et al (79). Abruptio placenta and elevated hCG were associated in the Tanaka et al. (107) article and the Gravett et al. (38) article, but the Gonen et al. (36) article found no such relation. Gravett, et al.(38), Muller et al. (79), and Tanaka, et al. (107) all reported that elevated hCG was associated with preterm labor/premature rupture of membranes (PROM), but Gonen et al. (36) failed to find this association. Pregnancy induced hypertension (PIH) was associated with elevated hCG levels in the Gonen et al. (36) article as well as an article by Sorenson et al. (101), but not in the Tanaka et al. (107) article. Gonen et al. (36) failed to find an association between hCG elevation and either maternal diabetes or maternal bleeding. Additionally, an article by Los, et al. (63) found an association between elevated hCG and 31 oligohydramnios, and an article by Clark et al. (21) found that elevated hCG levels were associated with maternal lupus anticoagulant. The specific mechanism for hCG elevation in abnormal pregnancies is unknown. One theory is that it results from the breakdown of the maternal placental barrier, much like with AFP (40). Another possible mechanism might involve decreased oxygen tension in the placenta. This could cause cytotrophoblastic hyperplasia (90) and increased trophoblast (which are responsible for hCG production) volume leading to increased production of hCG (36). This later thesis has also been supported by studies in animal models (30). Both elevated AFP and elevated hCG and adverse pregnancy outcomes Recently researchers have begun investigating the risks associated with adverse pregnancy outcomes in the presence of concurrent elevations of both AFP and hCG. Because this is a new topic of interest, the research is only in its early stages and much more research will need to be done in the future to confirm the findings of the early studies. Five studies have reported on a relationship between adverse pregnancy outcomes and elevated levels of both AFP and hCG. These studies are discussed below and summarized in Table 3. The first study to report on this relationship was by Beekhuis et al (1). They noted that intrauterine fetal death (IUFD) occurred in four women from their screening program who were screen-positive for both NTD’s and Down syndrome (i.e. elevated AFP causing a screen positive result for NTD’s and elevated hCG driving a screen positive result for Down syndrome). This caused them to retrospectively evaluate all women who had been 32 Table 3. Summary of articles reporting on elevated AFP and elevated hCG with adverse pregnancy outcomes. Study # of Cases Case Def. IUFD SGA (LB) [DOB [’11) (LB) Beekhuis et al. 1 l hCG22.0 36% NR NR 50% 1992 (l) AFP22.0 Walters et a1. 4 AFP>2.0 50% NR NR 50% 1993 (123) DS>1:270 Gross et al. 5 AFP22.5 25% 100% 100% 50% 1994 (40) DS21:274 Wenstrom ct a1. 8 AFP22.5 37.5% NR 12.5% 50% 1994 (124) hCG22.0 Morssink et al. 21 AFP22.5 NR 38.1% NR 15.4% 1995 (77) hCG22.5 LB = live births only. DS = Down syndrome risk. NR = not reported. screened by their program in whom both AFP and hCG had been found to be 2 2.0 MOM. Eleven women fulfilled the criteria. Of these eleven women, only one had no pregnancy complications and only three delivered a healthy livebom child, even though all the fetal karyotypes were normal. The pregnancy complications noted in the ten women included hemolysis, elevated liver enzymes and low platelet count (HELLP), eclampsia, oligohydramnios, hemorrhage, and PROM. Four women had an intrauterine death with subsequent histopathical examination showing no fetal congenital anomalies. Three women terminated their pregnancies because of severe fetal congenital anomalies (omphalocoele, cystic urachus and prune belly, and cystic kidneys). Of the four pregnancies that continued, one was uncomplicated and ended in normal delivery of a healthy full-term (40 week) infant, one was delivered by cesarean section at 38 weeks because of severe pre-eclampsia, and two were delivered prematurely at 36 and 26 weeks of gestation (the later baby died four days 33 later). Ten placentas were examined, and showed infarction in four cases, ischaemic changes in two, severe chorionitis in one, and a normal histology in three. These results led the authors to conclude that when levels of both AFP and hCG are elevated they no longer correspond to the risk of having a fetus with a NTD or Down syndrome, respectively, but rather indicate a population at increased risk for other adverse pregnancy outcomes. Walters et al. (123) arrived at a similar conclusion relating adverse pregnancy outcome to elevated AFP and hCG, but only in the context of a positive triple analyte screen test for Down syndrome, whereby a decreased maternal serum uE3 level was also present. In their data from the Vermont Prenatal Screening Program, they identified four women (age 25-34) who met the double screen-positive criteria of AFP > 2.0 MOM and an increased second trimester risk for fetal Down syndrome of greater than 1 in 270 (hCG range: 1.93-4.62 MOM), and thirteen women (age 18-35) who had screening results of both AFP and hCG > 2.0 with a Down syndrome risk of less than 1 in 270. None of these pregnancies were affected by Down syndrome or NTD’s. A comparison of mean biomarker levels in the two groups showed that the only difference between the two groups was that uE3 was significantly lower in those pregnancies with a Down syndrome risk greater than 1 in 270. Their results showed that three of the four pregnancies in the double screen- positive group ended in an adverse outcome: two in IUFD and one in a preterm delivery at 29 weeks. In the thirteen women not screen-positive for both NTD and Down . syndrome, however, all four completed pregnancies produced normal males. These 34 results led the authors to conclude that only those pregnancies in which the pattern of the three biochemical markers along with maternal age yielded an increased Down syndrome risk of greater than 1 in 270, while AFP was elevated, were associated with an adverse pregnancy outcome. A study published by Gross et al. (40) reported on five women who were screen positive for both NTD, defined as AFP22.5, and Down syndrome, defined as risk 21 in 274 (hCG range: 2.62-3.65) from an original sample of 14,857 screened women. Four women elected to undergo amniocentesis, and all exhibited normal karyotype infants and normal amniotic fluid AFP levels. The Gross et al. (40) findings were consistent with those found previously by Walters et al (123). In their study, all five women had adverse pregnancy outcomes. IUGR occurred in all five pregnancies, and two were terminated because of this. One additional pregnancy resulted in intrauterine fetal demise, and the remaining two pregnancies resulted in live births. Of the two live births, one was delivered at 38 weeks by emergency Cesarean section for fetal distress, and the other was delivered by Cesarean section at 32 weeks for severe pre-eclampsia and fetal distress. Both live births had weights below the tenth percentile for their respective gestational ages. The authors concluded that although predictive value estimates cannot be offered regarding the incidence of adverse pregnancy outcomes in the context of both an elevated AFP and a positive serum triple screen test for Down syndrome because Of the rarity of these cases, their experience suggests that there is an increased incidence of fetal demise, growth retardation, or infection in these cases. 35 Wenstrom et al. (124) examined analyte levels in 126 women with adverse pregnancy outcomes such as preterm delivery, stillbirth, and IUGR while excluding aneuploidy and structural abnormalities and compared them to 126 women with normal outcomes but who had an amniocentesis for other reasons (maternal age, family history of defects). They found that 37.5 percent of the women with both high had a fetal death, 50 percent had a preterm delivery, and 12.5 percent had IUGR. The results indicated both elevated hCG and AFP levels were significantly associated with preterm delivery and fetal death, and the odds ratio associated with any poor obstetric outcome was 7.4 (95% CI 1.1-168.2). Morssink et al. (77) conducted a relatively large study in comparison to those previously discussed. Their study identified twenty-one women with both AFP and hCG level 2 2.5 MOM after excluding those with unknown pregnancy outcome, a congenital anomaly, delivery before 25 weeks of amenorrhoea, or known insulin-dependent diabetes. These women comprised 0.2 percent of the screened population of 8,892 singleton pregnancies. They found that 23.8 percent of the population with both markers elevated had an extremely (< 2.3 percentile) small for gestational age infant compared to only 2.2 percent Of those with normal test results. This produced a relative risk of 10.9 (p<.01) and a detection rate of 2.7%. When SGA infant (<10th percentile) was examined, 38.1 percent of the pregnancies resulted in SGA infants compared to 10.3 percent of the normal population for a relative risk of 3.7 (p<.01) and a detection rate of 0.9 percent. For preterm delivery (<37 weeks), 15.4 percent of the population with both AFP and hCG elevated and 6.0 percent of the normal population resulted in a preterm birth. This result 36 was not found to be statistically significant. When SGA infant or preterm delivery were . examined, 47.6 percent of the population with both AFP and hCG elevated and 15.7 percent of the normal population had these pregnancy outcomes. The relative risk was 3.0 (p<.01), and the detection rate was 0.8 percent. The authors concluded that a combined elevation of both AFP and hCG produced a higher positive predictive value and relative risk than AFP or hCG alone. The mechanism for both markers being elevated is unknown. One possible explanation derives from a morphometric study by Boyd and Keeling (5). Their study showed an increased mean volume of placental parenchyma and a greater villous surface area in patients with raised AFP levels. This could lead to the elevated hCG since it is placental in origin (51). Reduced perfusion could explain the relation of high levels of both AFP and hCG with adverse pregnancy outcome. Reduced perfusion is a stimulus for the formation of trophoblastic tissue which produces a larger amount of hCG (88). On the other hand, reduced perfusion causes infarction and ischaemic changes in the placenta (1) which could be a cause in the adverse pregnancy outcomes seen in these studies. Chapter 3 OUR STUDY Introduction The purpose of this study is to examine the relationship between abnormal AFP and/or hCG levels and adverse pregnancy outcomes. We built on earlier studies by assessing elevations of AFP and hCG both separately and together in order to consider the effect of each marker on the various adverse pregnancy outcomes. In the process, we have constructed the largest study population of women with both elevated AFP and hCG levels to date to improve on the limited statistical power of previous small studies. We hypothesized that women who had high test results for both AFP and hCG would be most at risk for adverse pregnancy outcomes when compared to women with normal test results. Methods For this retrospective cohort study, women were selected from the screening databases of the Henry Ford Hospital (Detroit) and Michigan State University prenatal screening programs. These are two of the leading screening programs in Michigan, and together they provide the triple test for approximately 36,000 women annually in Michigan. Both programs use similar methods and protocols for their testing. The three analytes were assayed using FDA approved kits. The MSU program used the Kallestad AFP/OB for AFP detection, the Diagnostic Systems Laboratories kit for uE3, and the Serono kit for hCG detection. Henry Ford Hospital also used the Kallsted kit for AFP detection, however for uE3 they used the Amersharn 3rd trimester kit until September 37 38 1991, the Amersham 2nd trimester kit until January 1992, and then the Diagnostic Systems Laboratories kit. For hCG they used the Serono kit until September 1991, the Amersham 2Ind trimester kit until August 1992, and then the Coming Magic Lite kit. All of the hCG kits measured both free-B and intact hCG by using antibodies specific for the B-subunit. The MSU screening facility expresses results based on the median for the week of pregnancy the women were tested, and Henry Ford expresses results based on the median by day. Results were expressed as multiples of the median (MOM) to allow - comparisons of women at different gestational ages and between the two sites. All results for AFP were adjusted for gestational age and maternal weight at the time of testing, and separate medians were used for black and non-black women. MOM’s for the other two analytes were adjusted for gestational age at the time of testing, but were not adjusted for maternal weight nor calculated separately by race as the protocol for these markers did not require adjustment during the entire time period these women were tested. Women were selected for participation in this study based on their triple test results. Eligible women were screened between January 1, 1991 and July 31, 1994, and all were screened between 15 and 22 weeks of pregnancy. Women were divided into four groups: Group 1) Both high AFP and hCG: AFP22.0, hCG23.0 Group 2) High AFP only: AFP22.0, 0.70