Viability of intrauterine pregnancy in women with pregnancy of unknown location: prediction using human chorionic gonadotropin ratio vs. progesterone

Abstract

Objectives: To find the best predictor of the viability of intrauterine pregnancies of uncertain viability (IPUVs). An IPUV is defined as an empty intrauterine gestational sac measuring less than 20 mm in diameter or an intrauterine gestational sac containing a fetal pole with a crown-rump length (CRL) of less than 6 mm with no fetal cardiac activity.

Methods: This was a prospective observational study comparing the following two markers for the prediction of viability in IPUVs: human chorionic gonadotropin (hCG) ratio (hCG 48 h : hCG 0 h); and serum progesterone at first presentation. All women classified with a pregnancy of unknown location (PUL) were followed up until final diagnosis: failing PUL, viable or non-viable intrauterine pregnancy (IUP), or ectopic pregnancy. Those PULs found to have an IPUV at follow-up transvaginal ultrasound scan (TVS) were included in the final analysis. Receiver-operating characteristics (ROC) curves were used to compare the performance of the hCG ratio and progesterone. Using logistic regression, the added value of combining both variables over single-variable prediction was also evaluated. Missing values were imputed using multiple imputation.

Results: In total, 12 572 consecutive first-trimester women were scanned. Of these, 1003 (8%) women were classified as having PULs at first scan; 379 (37.8%) PULs were confirmed as IPUVs at follow-up scan, with 82.6% (276/334) found to be viable IUPs and 17.4% (58/334) non-viable IUPs on a second follow-up scan. Forty-five PULs were lost to follow-up. For prediction of viability in IPUVs, the area under the ROC curve (AUC) for the hCG ratio was 0.756 (95% CI, 0.686-0.826) whilst that for progesterone was 0.678 (95% CI, 0.603-0.753). The difference in AUC was 0.078 (95% CI, 0.014-0.169, P = 0.098), suggesting that the hCG ratio was at least as good as initial progesterone. Furthermore, there was no clear threshold for initial progesterone, whilst for hCG a ratio of 2 was suggested (sensitivity 78%, specificity 67%). Logistic regression modeling demonstrated that progesterone did not improve the discriminatory power of the hCG ratio (AUC = 0.758).

Conclusions: The hCG ratio may be preferred to single measurements of progesterone to predict the viability of IPUVs in the PUL population.