WHEN ABNORMAL HCG LEVELS RESULT IN A NORMAL PREGNANCY

Abstract

Thresholds of the rise in human chorionic gonadotropin (hCG) that distinguish early normal and abnormal pregnancies have evolved with time and experience and are still not 100% accurate. Using a data set of women from a state where law mandates conservative management of pregnancies of unknown viability, and therefore no intervention for abnormal rise of hCG in the clinically stable patient, we assessed how often hCG thresholds misclassify a normal pregnancy as abnormal. All women seen at the University of Missouri Health Care System who had at least two consecutive beta hCG quantitative serum levels (initial hCG <5000 mIU/ML) between 1/2015 - 3/2020 were evaluated. Collected data included date and time of each serum hCG result, patient presenting symptom, and final pregnancy outcome (early pregnancy loss (EPL), ectopic pregnancy (EP), intrauterine pregnancy (IUP), live birth). We assessed 1) A Linear fit model for hCG rise for all initial values1 (53% rise in 2 days); 2) Quadratic model for hCG rise based on 3 ranges of initial values2 (49%, 40%, and 33% rise based on initial hCG of <1500, 1500-3000, and >3000 mIU/ML, respectively); and 3) Pre-determined thresholds in hCG based on a conservative threshold of at least 15% rise per day3 to determine which model most accurately predicts possible viability. A total of 1295 patients with a pregnancy of unknown location had at least two consecutive serum hCG values and 689 patients met criteria of initial hCG < 5000 mIU/ML with second hCG value between 1 -7 days thereafter. Final outcome was 462 (67%) women with EPL, 58 (8%) with EP, and 169 (25%) with a viable IUP. Depending on the model, up to 9 IUP (9%) were misclassified by hCG values. Of these, 6 (0.9% of total) resulted in a live birth, 1 was an elective abortion, 1 was a fetal demise at 37 weeks gestation, and 1 was lost to follow up. With more conservative thresholds, the number of misdiagnosed IUP declined, but the majority of EPL and EP were identified. Patterns of hCG rise as determined by ROC analyses were able to predict live birth and miscarriage with up to a 95% accuracy. Current models of hCG rise will identify women with EPL and EP but will also classify some intrauterine pregnancies as nonviable. Levels as low as a 15% rise may be needed to minimize this error. In the clinically stable patient with a desired pregnancy, continued watchful observation should be the norm.