Dr Xiu-Feng Huang, Department of Gynecology and Obstetrics, Women's Hospital, School of Medicine, Zhejiang University, 2 Xue Shi Road, Hangzhou, Zhejiang 310006, China. Email: firstname.lastname@example.org
Aim: A tubal ectopic pregnancy (EP) in the first trimester remains a major life-threatening complication for the mother. We aim to determine whether serum reproductive hormones may be clinically useful in the early identification of a tubal EP.
Methods: A total of 109 age-matched patients with a serum β-human chorionic gonadotropin (β-hCG) concentration <2000 IU/L were enrolled, including 68 patients with a tubal EP, 22 with a viable intrauterine pregnancy (vIUP) and 19 with a non-viable intrauterine pregnancy (nIUP). Serum was collected during the first trimester of pregnancy and assayed for β-hCG, follicle-stimulating hormone (FSH), luteinizing hormone (LH), testosterone (TE), progesterone (P), estradiol (E2) and prolactin (PRL) by using automated electrochemiluminescence immunoassays.
Results: Patients with an EP had significantly lower levels of daily β-hCG variation (Δβ-hCG/day), P and E2, and significantly higher levels of LH and FSH than IUP patients (P < 0.05). As an EP diagnostic marker, progesterone demonstrated a sensitivity of 100% at the cutoff of 86.01 nmol/L. The combination of E2 with Δβ-hCG/day reached a specificity of 100% for EP evaluation. To identify non-viable pregnancies (including EPs and nIUPs), progesterone demonstrated a sensitivity of 95.40% and a specificity of 90.91% at the cutoff of 63.2 nmol/L; the diagnostic power of the receiver operating curve was 0.9702.
Conclusions: A combination of Δβ-hCG/day, P and E2 may help distinguish EPs and nIUPs from vIUPs, facilitating earlier diagnosis and the timely implementation of medical treatment to prevent tubal rupture.
An ectopic pregnancy (EP) is a common complication among women of reproductive age, with >98% of occurrences happening in the Fallopian tubes. Approximately 1–2% of all pregnancies in Europe and the USA are ectopic, and the incidence of an EP is much higher in developing countries.1 Ectopic pregnancies remain a major cause of maternal morbidity and mortality worldwide. A combination of transvaginal ultrasound and serial β-human chorionic gonadotropin (β-hCG) measurement is currently the most common approach for the diagnosis of an EP; however, if the serum β-hCG concentration is <2000 IU/L, the intrauterine gestational sac may be nearly invisible, meaning that ultrasound cannot discriminate between intrauterine and ectopic pregnancies successfully.2 Therefore, most ultrasound laboratories use β-hCG levels between 1000 to 2000 IU/L as the threshold for which an intrauterine pregnancy should be visualized by vaginal ultrasound.3 A single measurement of serum β-hCG is not helpful in distinguishing the location of a pregnancy. In the emergency setting, it is valuable to be able to differentiate between an EP and an intrauterine pregnancy (IUP) without delay, in order to decrease the time to diagnosis and reduce the possibility of a fatal tubal rupture.
New serum markers for the earlier diagnosis of EP are emerging; for example, interleukin (IL)-8,4 IL-6, tumor necrosis factor-α,5‘triple marker’ vascular endothelial growth factor (VEGF)/(pregnancy associated plasma protein [PAPP]-A × progesterone [P]),6 VEGF,7 activin8 and activin β-B9 have all been studied for their potential utility as markers of ectopic pregnancy. Although these new serum markers may lead to earlier diagnosis, none has been proven to have sufficient sensitivity or specificity for the diagnosis of EP, and they are not currently used in the clinic.
A combination of multiple reproductive hormones may be useful in discriminating an EP from an IUP. Progesterone, a hormone produced by the corpus luteum early in pregnancy, is elevated with a viable IUP (vIUP), even when the serum β-hCG value is <1000 IU/L.10 The presence of low serum progesterone concentrations in patients with EP has been known since the late 1970s.11 A number of studies have assessed the role of progesterone in the diagnosis of ectopic pregnancy; however, a meta-analysis concluded that while serum progesterone measurement may identify patients at risk for EP, its discriminative capacity is insufficient to diagnose an ectopic pregnancy with a high degree of certainty.12 Serum β-hCG and estradiol (E2) may also help in the diagnosis of ectopic pregnancies.13 We speculated that a multiple-biomarker strategy, including β-hCG, follicle-stimulating hormone (FSH), luteinizing hormone (LH), testosterone (TE), progesterone, E2 and prolactin (PRL), may be efficient for distinguishing an EP or non-viable IUP (nIUP) from a vIUP.
In this retrospective study, we recruited 109 age-matched pregnant women, including 68 with a tubal EP, 22 with a vIUP and 19 with a nIUP, to explore: (i) whether the serum reproductive hormone levels differ in EPs and IUPs; (ii) the diagnostic power or clinical utility of serum reproductive hormones for the identification of patients with EPs; and (iii) the utility of a multiple-biomarker strategy to help distinguish EPs from IUPs in early pregnancy.
Patients and Methods
Patients were recruited from a population of pregnant women presenting to the Women's Hospital at Zhejiang University from January 2006 to January 2008 with complaints of vaginal bleeding or abdominal pain who were suspected to have an EP. All women included in the study had a history of regular menses and a normal last menstrual period (LMP), had conceived spontaneously and were not taking exogenous progestogen. Patients older than 40 years of age or with a gestational age >90 days were excluded. The diagnosis of a vIUP was made on follow-up by the demonstration of an intrauterine gestational sac on the ultrasound scan. A nIUP was diagnosed histologically after surgical evacuation of the uterine contents. An EP was diagnosed by laparoscopy and histological examination of the excised specimens. Written, informed patient consent was obtained and the study was approved by the Ethics Committee of the School of Medicine of Zhejiang University.
Serum hormone measurements
All blood samples were obtained through venipuncture and clotted at room temperature. After centrifugation, the serum samples were stored at −20°C until analysis.
Serum levels of β-hCG, FSH, LH, TE, P, E2 and PRL were measured with electrochemiluminescence immunoassays (ECLIA) on the Roche Modular E170 immunoassay analyzer (Roche Diagnostics, Indianapolis, IN, USA). The total imprecision for the assays are 6.8% for β-hCG at a level of 1500 IU/L, 4.5% for FSH at a level of 28 IU/L, 3.9% for LH at a level of 23 IU/L, 6.6% for TE at a level of 16 nmol/L, 6.2% for progesterone at a level of 28 nmol/L, 6.3% for E2 at a level of 780 pmol/L and 5.8% for PRL at a level of 24 ng/mL. The measuring ranges were: β-HCG, 0.1–10 000 IU/L; FSH, 0.1–200 IU/L; LH, 0.1–200 IU/L; TE, 0.069–52 nmol/L; progesterone, 0.095–191 nmol/L; E2, 18.4–15 781 pmol/L; and PRL, 0.047–470 ng/mL.
All data were analyzed with Prism software (GraphPad Software, San Diego, CA, USA) and were expressed as mean ± SD. The Kolmogorov–Smirnov test confirmed that the distributions of Δβ-hCG/day (Δβ-hCG/day = log2(β-hCG2/β-hCG1)/day), FSH, LH, TE, P, E2 and PRL concentrations were not Gaussian. β-hCG1 indicates the β-hCG level detected the first time and β-hCG2 indicates the β-hCG level detected the second time. The d-value indicates the days of the interval between collection of β-hCG1 and β-hCG2 (range: 1–14, mean: 3.97); therefore, the Mann–Whitney U-test and Kruskal–Wallis test followed by Dunn's post hoc test were used to compare intergroup differences of these markers' levels. The SPSS (Chicago, IL, USA) logistic regression procedure with the forward stepwise selection method using the likelihood-ratio statistic was applied to determine variable inclusion. Values of P < 0.05 were considered significant.
The diagnostic sensitivity, specificity, predictive value, likelihood ratios (LR) and the area under the curve (AUC) of the receiver operating characteristic (ROC) were assessed using GraphPad Prism (GraphPad Software, San Diego, CA, USA). Linear regression was used to analyze the correlation between Δβ-hCG/day and P/E2 levels.
Patients with serum β-hCG concentration >2000 IU/L were excluded and 109 cases in total were enrolled in the present study, including 68 tubal EP and 41 IUP cases, the latter being further divided into 22 vIUP and 19 nIUP cases. The clinical characteristics of all these groups are summarized in Table 1. There were no statistically significant differences in age or estimated gestational age between the groups (P > 0.05).
Table 1. Clinical characteristics and serum reproductive hormone concentrations of study participants
P < 0.05 versus EP;
P < 0.01 versus EP. Data are provided as mean ± SD. E2, estradiol; EP, ectopic pregnancy; IUP, intrauterine pregnancy; nIUP, non-viable IUP; P, progesterone; PRL, prolactin; TE, testosterone; vIUP, viable IUP; β-hCG, β-human chorionic gonadotropin; Δβ-hCG/day = log2(β-hCG2/β-hCG1)/day.
Serum levels of LH, FSH, E2, P and Δβ-hCG/day were all significantly different between the EP and IUP groups (Table 1 and Fig. 1). In detail, Δβ-hCG/day was significantly lower in EP patients than in either IUP patients (P < 0.05) or vIUP patients (P < 0.01), but was significantly higher than in nIUP patients (P < 0.05). Changes in progesterone levels showed a similar trend for Δβ-hCG/day, in that the progesterone levels in the EP group were approximately half of those in the IUP group (P < 0.01) and one-quarter of those with a vIUP (P < 0.01), but at about same level of those with a nIUP (P > 0.05). Estradiol levels in EP were lower than in patients with vIUP (P < 0.01), but the difference between EP and IUP was not statistically significant (P > 0.05). On the contrary, women with an EP had significantly higher LH and FSH concentrations than those with an IUP (P < 0.01). Patients with a vIUP had the lowest LH and FSH levels. No statistically significant difference in serum testosterone or prolactin concentrations were identified between the EP and IUP groups (P > 0.05).
Clinical utility of LH, FSH, E2, P and daily β-hCG variation measurement for EP diagnosis
The diagnostic sensitivity, specificity, positive and negative LR and AUC of LH, FSH, E2, P and Δβ-hCG/day were evaluated by ROC curve analysis. The ROC AUC of LH, FSH, E2, P and Δβ-hCG/day were, respectively: 0.6799 (95% confidence interval [CI], 0.5733–0.7864), 0.6478 (95% CI, 0.5352–0.7604), 0.6008 (95% CI, 0.4837–0.7179), 0.6765 (95% CI, 0.5520–0.8009) and 0.6273 (95% CI, 0.5027–0.7520) (EP vs IUP, P > 0.05) (Fig. 2a). LH, FSH, E2, P and Δβ-hCG/day were included in the logistic regression analysis to discriminate between EP and IUP. Eventually, only P and Δβ-hCG/day were retained in the model as FSH, LH and E2 had dropped out of the final model.
In detail, daily β-hCG variation at the cutoff of 0.1632 demonstrated a sensitivity of 60.29% (95% CI, 47.70–71.96%) and a specificity of 60.98% (95% CI, 44.50–75.80%) as a single marker for EP differentiation (ROC AUC: 0.6273 [95% CI, 0.5027–0.7520]), with a positive and negative LR of 1.55 and 0.65, respectively (Fig. 2b).
Progesterone at a cutoff of 47.89 nmol/L showed a sensitivity of 92.65% (95% CI, 83.67–97.57%) with a specificity of 58.54% (95% CI, 42.11–73.68%) for EP, with a ROC AUC of 0.6765 (95% CI, 0.5520–0.8009), and positive and negative LR of 6.63 and 0.45, respectively. Progesterone, at a cutoff of 86.01 nmol/L, achieved a sensitivity of 100% (95% CI, 94.72–100.00%) with a specificity of 34.15% (95% CI, 20.08–50.59%) for EP (Fig. 2c).
To distinguish EP and nIUP from vIUP, progesterone at a cutoff of 63.2 nmol/L demonstrated a sensitivity of 95.40% (95% CI, 88.64–98.73%) and a specificity of 90.91% (95% CI, 70.84–98.88%), with a ROC AUC of 0.9702 (95% CI, 0.9414–0.9990), and positive and negative LR of 10.49 and 0.05, respectively (Fig. 2d).
For patients with progesterone <86 nmol/L (100% sensitivity for EP), their serum E2 and daily β-hCG variation data were plotted in Figure 3, with EP patients clustered in the top left and bottom right quadrants. To cluster the EP and IUP patients completely for the patients with P < 86 nmol/L, we determined the threshold according to the ROC curve results and graphed them together. According to the graph, it is hard to distinguish between EP and IUP with Δβ-hCG/day at the cutoff of 0.1632 IU/L; therefore, we chose the Δβ-hCG/day at a cutoff of 0.2134 IU/L. When E2 <520.6 pmol/L combined with Δβ-hCG/day >0.2134 (top left quadrant) or E2 >520.6 pmol/L combined with Δβ-hCG/day <0.2134 (bottom right quadrant), specificity and positive predictive value for EP diagnosis were 100%, although sensitivity was 51.47% with a negative predictive value of 55%.
Progesterone and E2 concentrations correlated well with the daily β-HCG variation in the IUP group (P < 0.01) with r2 of 0.4684 and 0.3821, respectively (Fig. 4). However, there was no significant correlation between P, E2 and daily Δβ-hCG/day in the EP group (P > 0.05).
Transvaginal ultrasonography in combination with serum β-hCG is commonly used for the diagnosis of an EP. Transvaginal ultrasound is an important tool that can reliably predict the presence of ectopic pregnancy.14 With transvaginal ultrasonography, viable intrauterine pregnancies should be visible with β-hCG levels above 2000 IU/L. Unfortunately, up to 15% of symptomatic patients have an indeterminate ultrasound at their initial presentation;15 that is, both diagnostic findings of an IUP as well as extrauterine findings suggestive of an EP are absent. In such scenarios, serial β-hCG levels are crucial for the diagnosis. If β-hCG levels continue to rise appropriately, a repeat vaginal ultrasound should be obtained when the β-hCG level reaches the threshold level of 2000 IU/L in order to attain unequivocal findings by transvaginal ultrasound. However, if a measurement could differentiate between ectopic and intrauterine pregnancy without delay, it would reduce unnecessary medical treatment and the possibility of tubal rupture, and would therefore improve prognosis. In this study, we enrolled patients with serum β-hCG levels <2000 IU/L and investigated the usefulness of a combination of serum markers for the earlier diagnosis of EP.
First, our study demonstrated significant differences in serum reproductive hormone levels between women with an ectopic pregnancy and those with an intrauterine pregnancy. Given that serum β-hCG levels vary substantially for both intrauterine and ectopic pregnancies, a single measurement of the β-hCG level is not helpful in practice. In most situations, serial β-hCG measurements are required. The doubling time associated with serum β-hCG is an important parameter for differentiating normal and abnormal pregnancies. In early gestations, the serum β-hCG doubles approximately every 1.98 days.16 We selected the daily β-hCG variation (Δβ-hCG/day = log2(β-hCG2/β-hCG1)/day) instead of the β-hCG level as a parameter in the present study. We found that the levels of Δβ-hCG/day, E2 and P were significantly decreased in women with an EP, especially when compared with those in women with a vIUP, which is consistent with previous research.17 In contrast, the levels of FSH and LH were significantly increased in women with an EP, especially when compared with women with a vIUP. It is possible that these hormones may represent efficient markers for distinguishing an EP and a nIUP from a vIUP.
For the clinical utility of single biomarkers, to our surprise, the diagnostic power of daily β-hCG variation, P, LH, FSH and E2 are all insufficient to diagnose EP with certainty (Fig. 2a); however, several important insights may be gleaned from this finding. Progesterone at a cutoff of 86 nmol/L achieved a sensitivity of nearly 100% for EP diagnosis (Fig. 2c), which is consistent with former studies.18 Therefore, progesterone levels >86 nmol/L may be used to rule out EP. In clinical practice, the time delay necessary for distinguishing a viable IUP from a nIUP or EP is often distressing for patients and practitioners. Since EP and nIUP are both non-viable pregnancies, we tried to distinguish EP and nIUP from vIUP by using progesterone. A cutoff of 63.2 nmol/L for progesterone may identify patients at risk of a non-viable pregnancy with a high sensitivity and specificity (Fig. 2d). Distinguishing a vIUP by using progesterone will help to avoid unnecessary diagnostic laparoscopic procedures and hospitalization.
The women with P > 86 nmol/L were identified as having a vIUP in this study. For the remaining patients, the combination of E2 and daily β-hCG variation reached a diagnostic specificity of nearly 100% (Fig. 3). At this time in the pregnancy, the β-hCG level may be still <2000 IU/L, making it difficult to judge the viability by transvaginal ultrasonography. Women with E2 <520.6 pmol/L and daily β-hCG variation >0.2134, or E2 >520.6 pmol/L and daily β-hCG variation <0.2134, can be ruled in as having an EP and can be treated with medical therapy in a timely manner. Economic evaluations showed that timely medical treatment was associated with reduced direct and indirect costs for EP. For example, when the β-hCG level is >1500 IU/L, the cost-effective benefits may lessen due to the need for prolonged follow-up and surgical intervention.19 The success rate of single-dose methotrexate was reported to be 97% at a β-hCG level below 2000 IU/L, but dropped to 74% for β-hCG levels >2000 IU/L.20–22 It is obvious that a combination of estradiol and daily β-hCG variation is helpful in the early diagnosis of an EP, which is valuable, not only for improving the treatment effect, but also in reducing the associated medical costs.
There are fair correlations among daily β-hCG variation and the serum concentrations of P and E2 in women with an IUP, but not in women with an EP. Logically, a combination of E2 and daily β-hCG variation is helpful in the diagnosis of an EP. The associations among daily β-hCG variation, P and E2 support the idea that, in the case of an ectopic pregnancy, progesterone and estradiol are derived predominantly from the trophoblast and not from the corpus luteum.17
Our study investigated the utility of reproductive hormones for the early detection of an EP. The levels of Δβ-hCG/day, E2, P, FSH and LH differed between ectopic pregnancies and intrauterine pregnancies. If the progesterone level was ≥86 nmol/L, the diagnosis of ectopic pregnancy could virtually be excluded with nearly 100% certainty. For the patients with progesterone <86 nmol/L, the combination of estradiol and daily β-hCG variation may rule in ectopic pregnancy with nearly 100% certainty. A cutoff of 63.2 nmol/L may identify patients at risk of a non-viable pregnancy with a high sensitivity and specificity. Early EP diagnosis permits conservative treatment, allowing for subsequent preservation of fertility. It can also contribute to timely implementation of medical treatment and prevention of tubal rupture. Due to the small sample size of the present study, the clinical applicability of these findings needs to be evaluated in multi-center, large prospective studies.
This work was supported by grants from the National Natural Science Foundation of China (No. 81170310, 81070541, 81170620, 81170587 and 30901604), the Specialized Research Fund for the Doctoral Program of Higher Education of China (No. 20100101120155), the Science Foundation of Zhejang Province (No. Y2090455), and the National Basic Research Program of China (973 Program) (No. 2012CB944901).