Every year, thousands of women arrive at emergency departments with a positive pregnancy test but no answer to the most urgent question in early obstetrics: where is the pregnancy? When neither the uterus nor the tubes show clear signs of a gestational sac on the first scan, clinicians face a pregnancy of unknown location, a diagnostic limbo in which a normal early pregnancy, a failing pregnancy, and a life-threatening ectopic pregnancy can all look identical. A new study from Wuhu Fanchang District People’s Hospital in China, published in Frontiers in Medicine, reports that a simple blood measurement of the hormone progesterone, layered on top of transvaginal ultrasound and an established hCG-based risk model, may substantially sharpen the ability to separate ectopic pregnancies from the rest at the critical 48-hour reassessment visit.
Ectopic pregnancy, in which a fertilized egg implants outside the uterine cavity, most often in the fallopian tube, accounts for more than 95 percent of such cases and remains one of the leading causes of death in early pregnancy. Its incidence is rising, driven in part by the spread of assisted reproductive technology, where reported rates reach 2 to 5 percent per cycle, well above natural conception. The clinical difficulty is that early ectopic pregnancies are often silent or atypical: there may be no pain, no bleeding, and no visible mass on the initial ultrasound. The pregnancy is simply unlocated, and the clock starts ticking toward tubal rupture, hemorrhage, loss of fertility, or worse.
The research team, led by Leilei Liu, Yao Cai, and Tingting Wang, followed 148 women with pregnancies of unknown location between January 2022 and December 2025. Final outcomes, confirmed either by surgical pathology or by structured follow-up, divided the cohort into 49 ectopic pregnancies and 99 non-ectopic outcomes, the latter comprising 62 intrauterine pregnancies and 37 spontaneously resolving pregnancies. All patients underwent an initial transvaginal ultrasound to confirm that the pregnancy was unlocated, followed by a standardized 48-hour reassessment that included a repeat ultrasound, a second β-hCG measurement, and serum progesterone testing by electrochemiluminescence immunoassay on a Roche Cobas e601 analyzer with coefficients of variation below 5 percent.
The baseline differences between the two groups were striking. Women who ultimately proved to have ectopic pregnancies had a median serum progesterone of just 10.80 nmol/L, compared with 48.50 nmol/L in the non-ectopic group. Their initial β-hCG levels were lower (a median of 780 versus 1,850 IU/L), their 48-hour β-hCG rise was far shallower (a mean change rate of 15.20 percent versus 58.30 percent), and their endometrial thickness was markedly reduced (8.30 versus 16.12 mm). Abnormal adnexal findings on the follow-up ultrasound, such as heterogeneous masses, the tubal ring sign, or the doughnut sign, were detected in 79.59 percent of ectopic pregnancies but only 9.09 percent of the others. A history of prior ectopic pregnancy and pelvic inflammatory disease were also far more common among ectopic cases.
The mechanistic logic behind these differences is well understood. In an ectopic implantation, dysfunctional trophoblastic cells fail to stimulate the corpus luteum adequately, so progesterone secretion falls. Meanwhile, the gestational sac growing inside the fallopian tube provokes local inflammation and edema, generating the characteristic sonographic signs. The impaired trophoblastic function also explains the sluggish hCG dynamics: the mean 48-hour change rate of 15.20 percent in the ectopic group, with a large standard deviation of 22.50 percent, implies that many ectopic pregnancies actually showed declining hCG over the 48-hour window, a regressive pattern consistent with failing trophoblast in an unsupportive implantation site.
To quantify diagnostic performance, the team applied logistic regression and receiver operating characteristic analysis, anchored entirely at the 48-hour reassessment visit. Multivariable analysis identified four independent predictors: log-transformed progesterone (odds ratio 0.157, meaning each increase in progesterone was strongly protective against an ectopic classification), abnormal adnexal ultrasound (OR 26.00), prior ectopic pregnancy (OR 4.33), and high-risk classification by the M6NP model (OR 16.27). The M6NP model, an hCG-based risk calculator derived from the M6 framework of Van Calster and colleagues that uses only the log-transformed initial β-hCG and the 48-hour hCG ratio with its quadratic term, had already been externally validated in several European cohorts, making it a natural backbone for the combined strategy.
The incremental results are the heart of the study. Serum progesterone alone achieved an AUC of 0.851 with an optimal cutoff of 12.36 nmol/L, closely matching the 10 to 15 nmol/L range reported in prior literature. Abnormal adnexal ultrasound alone reached an AUC of 0.852 with 90.91 percent specificity, and the continuous M6NP predicted probability reached 0.858. But when ultrasound findings were combined with the M6NP model, the apparent AUC rose to 0.913. Adding progesterone pushed it to 0.947, with sensitivity of 91.84 percent and specificity of 90.91 percent, and the highest Youden index of the study at 0.828. The DeLong test confirmed the improvement was statistically significant (P = 0.035), and the integrated discrimination improvement of 0.059 (P = 0.005) indicated that progesterone contributed genuinely new information beyond hCG dynamics and sonographic anatomy, likely reflecting early placental endocrine activity rather than luteal function alone.
Perhaps the most clinically compelling finding emerged within the M6NP risk stratification. Among women the model flagged as high-risk, progesterone combined with ultrasound achieved an apparent AUC of 0.945, with sensitivity of 95.12 percent and specificity of 91.67 percent. In the low-risk group, progesterone’s diagnostic role weakened, but its high specificity of 91.82 percent still helped exclude ectopic pregnancy. Subgroup analyses by age, duration of amenorrhea, and prior ectopic history showed no significant interaction, suggesting progesterone behaves as a relatively stable biomarker across patient characteristics, although the small subgroup sizes, including only 20 women with prior ectopic pregnancy, make these comparisons exploratory.
The authors are careful to frame these numbers as preliminary. The model was developed and evaluated in the same dataset without bootstrap or internal validation, so all AUC values are apparent, in-sample estimates. The single-center retrospective design, the limited number of ectopic events, the possibility of selection bias from excluding women who needed immediate treatment or lacked complete 48-hour data, and the potential misclassification of spontaneously resolving pregnancies, which may occasionally represent resolving ectopics, all temper the conclusions. The precise timing of the follow-up ultrasound relative to the 48-hour blood draw was not uniformly recorded, leaving open the possibility of temporal information leakage that only prospective studies with standardized assessment timing can resolve.
Even with those caveats, the study offers a concrete, testable pathway for a problem that obstetricians confront daily. The intended application is not to replace surgical or histopathological confirmation, nor to guide immediate treatment at the first visit, but to support risk re-stratification and follow-up planning at the 48-hour reassessment, after the repeat hCG result and follow-up ultrasound are available. If external validation in larger, multicenter cohorts confirms these findings, the combined strategy of progesterone, transvaginal ultrasound, and hCG-based risk modeling could reduce unnecessary diagnostic surgeries, prevent excessive follow-up, and, most importantly, catch the ectopic pregnancies that currently slip through risk stratification as low-risk before they rupture. For a condition in which hours matter, a hormone measurement that costs little and adds measurable discrimination may prove to be one of the most consequential incremental tools in early pregnancy medicine.
Subject of Research: Ectopic pregnancy diagnostics
Article Title: Incremental diagnostic value of serum progesterone combined with transvaginal ultrasound in differentiating early ectopic pregnancy: a single-center real-world study
Article References: Liu, L., Cai, Y., & Wang, T. (2026). Incremental diagnostic value of serum progesterone combined with transvaginal ultrasound in differentiating early ectopic pregnancy: a single-center real-world study. Frontiers in Medicine, 13, Article 1949190. https://doi.org/10.3389/fmed.2026.1949190
Image Credits: AI Generated
DOI: 10.3389/fmed.2026.1949190
Keywords: ectopic pregnancy, progesterone, transvaginal ultrasound, M6NP model, pregnancy of unknown location, β-hCG, diagnostic accuracy, obstetrics, Frontiers in Medicine, Incremental, diagnostic, value
News Source: Harold Sullivan. (October 17, 2026). Progesterone Blood Test Plus Ultrasound Sharply Improves Early Ectopic Pregnancy Detection. Scienmag.
