http://informahealthcare.com/gye ISSN: 0951-3590 (print), 1473-0766 (electronic) Gynecol Endocrinol, Early Online: 1–5 ! 2015 Informa UK Ltd. DOI: 10.3109/09513590.2015.1031103
ORIGINAL ARTICLE
How can we better predict the risk of spontaneous miscarriage among women experiencing threatened miscarriage? Chee Wai Ku1,2, John C. Allen Jr1, Rahul Malhotra1, Han Chung Chong3, Nguan Soon Tan3,4, Truls Østbye1, Sze Min Lek1, Desiree Lie1, and Thiam Chye Tan1,2
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
1
Duke-National University of Singapore Graduate Medical School, Singapore, 2Department of Obstetrics and Gynecology, KK Women’s and Children’s Hospital, Singapore, 3School of Biological Sciences, Nanyang Technological University, Singapore, and 4Institute of Molecular and Cell Biology, A*STAR, Singapore Abstract
Keywords
This study seeks to establish progesterone and progesterone-induced blocking factor (PIBF) levels as predictors of subsequent completed miscarriage among women presenting with threatened miscarriage between 6 and 10 weeks of gestation. Our secondary objective was to assess the known maternal risk factors, toward development of a parsimonious and clinicianfriendly risk assessment model for predicting completed miscarriage. In this article, we present a prospective cohort study of 119 patients presenting with threatened miscarriage from gestation weeks 6 to 10 at a tertiary women’s hospital emergency unit in Singapore. Thirty (25.2%) women had a spontaneous miscarriage. Low progesterone and PIBF levels are similarly predictive of subsequent completed miscarriage. Study results (OR, 95% CI) showed that higher levels of progesterone (0.91, 95% CI 0.88–0.94) and PIBF (0.99, 95% CI 0.98–0.99) were associated with lower risk of miscarriage. Low progesterone level was a very strong predictor of miscarriage risk in our study despite previous concerns about its pulsatile secretion. Low serum progesterone and PIBF levels predicted spontaneous miscarriage among women presenting with threatened miscarriage between gestation weeks 6 to 10. Predictive models to calculate probability of spontaneous miscarriage based on serum progesterone, together with maternal BMI and fetal heart are proposed.
Miscarriage, predictive, progesterone, progesterone-induced blocking factor
Introduction Threatened miscarriage, defined as an ongoing pregnancy associated with vaginal bleeding, with or without abdominal pain [1], is the most common gynecological emergency, occurring in 15–20% of pregnancies [2]. Approximately 15% of women with threatened miscarriage progress to completed spontaneous miscarriage [3]. Currently, there are no standardized clinicianfriendly miscarriage risk assessment tools and no standard progesterone or Progesterone-Induced Blocking Factor (PIBF) cutoff levels accepted as ‘‘low risk’’ [4]. This study proposes progesterone and its downstream effector PIBF as biomarkers for risk assessment of spontaneous miscarriage in women who present in early pregnancy with threatened miscarriage. Progesterone is a critical hormone during implantation. It sustains decidualization, controls uterine contractility and promotes maternal immune tolerance to the fetal semi-allograft [5]. Risk of miscarriage is significantly higher among women with low serum progesterone [4], although cutoff levels for predicting completed miscarriage vary from 512 to 516 ng/mL among studies [6–8]. Progesterone-induced blocking factor (PIBF) is a pivotal mediator in progesterone-dependent immunomodulation [9,10]
Address for correspondence: Chee Wai Ku, KK Women’s and Children’s Hospital, Singapore. E-mail: [email protected]
History Received 16 December 2014 Revised 13 February 2015 Accepted 16 March 2015 Published online 2 June 2015
and has a regulatory role on anti-fetal immune responses during pregnancy [11]. In a recent study PIBF levels were significantly lower in women with threatened miscarriage who progressed to completed miscarriage compared with women who progressed to delivery [12]. Unlike progesterone, PIBF is not secreted in a pulsatile fashion and therefore not subject to gestational variability in serum concentration. Thus, it may be a better predictor of completed miscarriage than serum progesterone levels. Our primary aim was to establish progesterone and PIBF levels as predictors of subsequent completed miscarriage among women presenting with threatened miscarriage between 6 and 10 weeks of gestation. Our secondary objective was to assess known risk factors [4] towards development of a parsimonious and clinicianfriendly risk assessment model for predicting risk of completed miscarriage. The institutional review board at SingHealth (CIRB ref: 2010/ 620/D) approved the study.
Methods The study was conducted at the Kandang Kerbau Women’s and Children’s Hospital (KKH), the largest maternity hospital in Singapore. All obstetric emergencies are presented through the emergency unit of KKH. Study participants were women aged 19 years and above presenting at the KKH emergency unit between October 2011 and February 2012. Study inclusion criteria were a single intrauterine pregnancy and presenting with pregnancy-related vaginal
2
C. W. Ku et al.
Gynecol Endocrinol, Early Online: 1–5
bleeding between 6 and 10 weeks of gestation, calculated by ultrasonography using fetal pole size. For women with no fetal pole (but with the presence of yolk sac) on ultrasound (17.6%), the number of days since the last menstrual period (LMP) was used to calculate gestational age. Study exclusion criteria were previous episodes of vaginal bleeding during the current pregnancy, progesterone treatment in the current pregnancy, diagnosis of inevitable miscarriage, blighted ovum or planned pregnancy termination. Serum progesterone and PIBF levels were obtained from maternal blood samples taken at presentation. Blood was
collected into plain tubes and centrifuged for 10 min at 3000g within 2 h of collection. Serum samples were stored at 20 C until analysis. Serum progesterone was measured in the KKH clinical laboratory using a commercial ARCHITECT progesterone kit (Abbott, Ireland) and serum PIBF was measured in the research laboratory in Nanyang Technological University, Singapore, using a commercial ELISA kit, according to manufacturer’s protocol (Cusabio, China). Information on maternal demographic, health, obstetric and lifestyle factors (Table 1) was collected using a questionnaire administered verbally in either English or Chinese.
Table 1. Serum progesterone and PIBFa levels, and maternal characteristics at baseline, overall and by pregnancy status at 16 weeks gestation.
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
Pregnancy status at 16 weeks gestation Maternal characteristics Serum biological markers Progesterone, mean ± SD (nmol/L) PIBFa, mean ± SD (ng/ml) Demographics Age, mean ± SD (years) Age of spouse, mean ± SD (years) Marital status Married Single Highest educational level University degree (%) ITE or polytechnic (%) Primary and secondary (%) Number of children None (%) 1 or more (%) Health, obstetric and lifestyle factors Patients on progesterone after threatened miscarriage (%) Planned pregnancy Planned (%) Unplanned (%) Fetal Pole Present (%) Absent (%) Fetal Heart Present (%) Absent (%) Gestation age at presentation with threatened miscarriage By dates, mean (weeks) By CRLb, mean (weeks) Previous miscarriage Yes (%) No (%) BMIc, mean ± SD (kg/m2) Medical comorbidities Diabetes mellitus (%) Hypertension (%) Thyroid disease (%) Yes (%) No (%) Gynecological disease (%) Yes (%) No (%) Smoking Yes (%) No (%) Exposed to second hand smoke at home Yes (%) No (%) Experienced nausea during pregnancy (%) Yes (%) No (%) PSSd, mean ± SD (score) a
b
Total (N ¼ 119)
Ongoing pregnancy (N ¼ 89)
Spontaneous miscarriage at or before (N ¼ 30)
p Value
51.4 ± 26.4 255 ± 136
60.0 ± 23.5 293 ± 127
26.4 ± 17.1 142 ± 92
50.0001 50.0001
29.8 ± 4.9 33.2 ± 5.4
29.3 ± 4.8 32.9 ± 5.4
31.3 ± 4.7 33.9 ± 5.1
0.05 0.37
91.6 8.4
92.1 7.9
90.0 10.0
0.71
43.7 31.9 24.4
42.7 30.3 27.0
46.7 36.7 16.6
0.55
52.1 47.9
55.1 44.9
43.3 56.7
0.30
96.6
97.8
93.3
0.60
60.5 39.5
59.6 40.4
63.3 36.7
0.83
82.4 17.6
93.3 6.7
50.0 50.0
50.0001
73.1 26.9 8.0 ± 1.5
87.6 12.4 8.0 ± 1.5
30.0 70.0 8.0 ± 1.7
50.0001
7.3 ± 1.5
7.6 ± 1.5
6.3 ± 0.8
0.0003
20.2 79.8 23.3 ± 4.84
20.2 79.8 23.7 ± 5.1
20.0 80.0 22.1 ± 3.8
0.77
1.00 0.11
0 0
0 0
0 0
0.8 99.2
1.1 98.9
0 100.0
1.00
9.2 90.8
12.4 87.6
0 100.0
0.06
2.5 97.5
3.4 96.6
0 100.0
0.57
37.8 62.2
40.5 59.5
30.0 70.0
0.39
63.0 37.0 17.2 ± 5.8
70.8 29.2 16.9 ± 5.5
40.0 60.0 18.1 ± 6.7
0.004
c
d
PIBF ¼ progesterone-induced blocking factor, CRL ¼ crown-rump length, BMI ¼ body mass index, PSS ¼ perceived stress scale.
0.34
Progesterone miscarriage
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
DOI: 10.3109/09513590.2015.1031103
The primary outcome of interest was spontaneous miscarriage defined as evacuation of the uterus after inevitable or incomplete miscarriage, or complete miscarriage with an empty uterus, by the 16th week of gestation. Most miscarriages occur before week 16 of gestation [13]. Participants were contacted by telephone at the end of week 16 to determine the pregnancy outcome. A priori power analysis was performed using computer simulations on a logistic regression model with progesterone and PIBF as continuous predictors assuming a miscarriage rate of 20%. Multiple scenarios based on published literature [6–8] were simulated. A sample of n ¼ 100 patients was indicated for 80% power at ¼ 0.05 to detect clinically relevant differences in serum progesterone and PIBF levels in women with ongoing pregnancies compared to women experiencing spontaneous miscarriage. Statistical analysis was performed using SAS version 9.2 (SAS Institute Inc., Cary, NC). Baseline values of continuous variables were compared using two-sample t-tests between women with ongoing pregnancy beyond week 16 and women experiencing spontaneous miscarriage before 16 weeks; categorical variables were compared using Fisher’s exact test (2-sided, ¼ 0.05). Univariate and multivariate logistic regression analyses were used to assess progesterone and PIBF levels and other maternal factors on risk of spontaneous miscarriage. Stepwise logistic regression was used to develop a model predictive of miscarriage during the first 16 weeks of pregnancy. A tabular nomogram was constructed from the fitted model.
Results Overall, 306 women were considered for participation between 1 October 2011 and 28 February 2012. Of 119 women analyzed, 89 (74.8%) had ongoing pregnancies at week 16 of gestation and 30 (25.2%) experienced spontaneous miscarriage at or before week 16 of gestation (Supplementary Figure 1). On ultrasound, 21 women (17.6%) did not have fetal pole at presentation and 32 women (26.9%) did not have a fetal heart at presentation. The mean serum progesterone and PIBF were significantly higher in women with ongoing pregnancy (Progesterone: 60.0 ± 23.5 nmol/L; PIBF: 293 ± 127 ng/mL) than in women with spontaneous miscarriage (Progesterone: 26.4 ± 17.1 nmol/ L; PIBF: 142 ± 92 ng/mL) (p50.0001). The two groups did not differ significantly in maternal age, education, medical comorbidities, BMI, gestation age at presentation, or perceived stress as assessed by the PSS score; however, they did differ in gestational age by CRL at presentation, fetal pole and fetal heart (p50.001) (Table 1). Elevated levels of serum progesterone and PIBF were protective against completed miscarriage at 16 weeks (Progesterone: OR ¼ 0.907, 95% CI, 0.875–0.942; PIBF: 0.982, 0.974–0.990). Area (95% CI) under the ROC curve (AUC) was 0.888 (0.812, 0.965) for serum progesterone and 0.832 (0.744, 0.919) for PIBF. AUC for progesterone and PIBF did not differ
statistically at p50.05 based on overlapping confidence intervals. Four candidate predictors were significant at p50.15 in univariate analysis: presence of fetal heart (Y/N) (p50.0001), presence of fetal pole (Y/N) (p50.0001), maternal age (p ¼ 0.055) and BMI (p ¼ 0.116) (Table 2). ROC analysis yielded a cutpoint for serum progesterone of 35 nmol/L (sensitivity ¼ 77%, specificity ¼ 88%, PPV ¼ 68%, NPV ¼ 92%); the PIBF cutpoint was 180 ng/mL (sensitivity ¼ 73%, specificity ¼ 81%, PPV ¼ 58%, NPV ¼ 91%). Prevalence of spontaneous miscarriage in our cohort was 25%. Four factors were selected in the stepwise multivariate analysis (p50.15): BMI, progesterone, PIBF and fetal heart, with AUC ¼ 0.924 (Table 3). Using a best-subsets algorithm, a parsimonious 3-variable model evolved incorporating progesterone (p50.001), fetal heart (p ¼ 0.019) and BMI (p ¼ 0.003) with AUC ¼ 0.923 (Table 4). The Progesterone Risk of Miscarriage Index (Prog-RMI) giving 16-week predicted probability of miscarriage for fetal heart (present/absent), BMI (within the study range) and serum progesterone concentration is presented in
Table 3. Stepwise multivariate analysis for predicting spontaneous miscarriage.
Parameter
Odds ratio (95% CIa)
p-value
0.215 (0.055, 0.837) 0.925 (0.883, 0.969) 0.849 (0.730, 0.988) 0.992 (0.983, 1.002)
0.034
Fetal Heart (Y/N) Progesterone (nmol/L) BMIc (kg/m2) PIBFd (ng/ml)
Odds ratio (95% CIa)
Progesterone (nmol/L) PIBFc (ng/ml) Fetal Heart (Y/N) Fetal Pole (Y/N) Maternal age (years) BMId (kg/m2)
0.907 0.982 0.060 0.072 1.092 0.923
a
(0.875, (0.974, (0.022, (0.024, (0.998, (0.836,
0.942) 0.990) 0.165) 0.216) 1.195) 1.020)
Cumulative AUCb (95% CI) 0.788 (0.698, 878) 0.905 (0.830, 0.980) 0.923 (0.858, 0.988) 0.924 (0.858, 0.989)
50.0001 0.027 0.106
a
CI ¼ confidence interval, bAUC ¼ area under the ROC curve, cPIBF ¼ progesterone-induced blocking factor, dBMI ¼ body mass index.
Table 4. ‘‘Best subset’’ 3-factor model to predict spontaneous miscarriage: progesterone, fetal heart and BMI.
Parameter
Odds ratio (95% CIa)
p-value
0.910 (0.871, 0.952) 0.206 (0.055, 0.773) 0.850 (0.733, 0.986)
50.001
Progesterone (nmol/L) Fetal Heart (N/Y) BMIc (kg/m2) a
CI ¼ confidence interval, c BMI ¼ body mass index.
Cumulative AUCb (95% CI) 0.888 (0.812, 0.965) 0.905 (0.830, 0.980) 0.923 (0.858, 0.988)
0.019 0.031
b
AUC ¼ area
under
Table 2. Univariate-based analysis for predicting spontaneous miscarriage. Parameter
3
p-value
AUCb
50.0001 50.0001 50.0001 50.0001 0.055 0.116
0.888 (0.812, 0.965) 0.832 (0.744, 0.919) 0.788 (0.698, 0.878) 0.7163 (0.622, 0.811) 0.619 (0.501, 0.736) 0.581 (0.463, 0.698)
CI ¼ confidence interval, bAUC ¼ area under the ROC curve, cPIBF ¼ progesterone-induced blocking factor, dBMI ¼ body mass index.
the
ROC
curve,
4
C. W. Ku et al.
Gynecol Endocrinol, Early Online: 1–5
Table 5. Progesterone Risk of Miscarriage Index (Prog-RMI): Predicted probability (%) of spontaneous miscarriage by 16 weeks gestation (PPSM-16) in women presenting with threatened miscarriage from 6 to 10 weeks of gestation based on progesterone concentration, fetal heart and BMI.
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
Fetal Heart No Yes Progesterone BMI (kg/m2) BMI (kg/m2) (nmol/L) 15–20 >20–25 >25 15–20 >20–25 >25 ≤5 >5–10 >10–15 >15–20 >20–25 >25–30 >30–35 >35–40 >40–45 >45–50 >50–55 >55–60 >60
Legend:
99 98 97 95 92 88 83 75 65 54 42 31 22
97 96 93 90 84 77 68 57 45 34 24 17 11
94 91 86 79 70 60 48 37 27 19 12 8 5
94 91 86 80 71 61 49 38 28 19 13 8 5
88 82 74 64 52 41 30 21 14 10 6 4 3
76 67 56 44 33 23 16 11 7 4 3 2 1
0–19 20–49 50–74 75–100
Table 5 (See Supplement Table I for associated ROC summary table.)
Discussion We conducted a prospective cohort study of women presenting with threatened miscarriage at 6–10 weeks gestation to assess progesterone and PIBF levels at the time of presentation as predictors of subsequent completed miscarriage in combination with other known risk factors. Low progesterone and PIBF levels were both found to be predictive of subsequent completed miscarriage, with better accuracy (AUC) from progesterone – although the difference was not statistically significant. Our findings suggest that either progesterone or PIBF can be used independently for prognosticating spontaneous miscarriage by 16 weeks. Low progesterone was a strong predictor of miscarriage in our study with no apparent adverse effect on prognostic accuracy due to pulsatile secretion. It is possible that pulsatile secretion is assuaged in women presenting with high risk of threatened miscarriage, but this remains to be investigated. PIBF is a relatively new biomarker not routinely available in clinical laboratories. Our findings suggest that PIBF should be considered for future use in the prognosis and management of threatened miscarriage. Our proposed progesterone cutpoint of 35 nmol/L (11.0 ng/mL) (NPV 92%) for diagnosis of subsequent completed miscarriage is consistent with other cutpoints reported in the literature. In a retrospective study of 175 women with threatened miscarriage, Duan et al. proposed a serum progesterone cutpoint of 16 ng/mL (50.7 nmol/L) [6] for diagnosis of inevitable miscarriage. More recently, Arck et al. reported that serum progesterone levels 12 ng/mL (38.3 nmol/L) were associated with increased risk of completed miscarriage [4]. If the 35 nmol/L (11.0 ng/mL) progesterone cutpoint is validated in subsequent studies, it might be used to reliably predict women at low risk of miscarriage and exclude them from further follow-up and unnecessary treatment.
In addition to progesterone and PIBF, four other significant independent predictors of completed spontaneous miscarriage were identified: absence of fetal heart (OR ¼ 16.7), absence of fetal pole (OR ¼ 13.9), increase in maternal age (OR ¼ 1.09 or 9% increase in risk per year older in age) and reduction in BMI (OR ¼ 1.08 or 8% increase in risk per unit reduction (kg/m2) in BMI). Interestingly, despite many studies alluding to the importance of fetal heart in predicting viability of the fetus, and hence lowering the risk of spontaneous miscarriage, it was not as powerful a predictor as compared to progesterone and PIBF (AUC ¼ 0.788 versus 0.888 and 0.832 respectively). In the best subset analysis, a 3-factor model comprising progesterone, fetal heart and BMI has a cumulative AUC of 0.923, which has great potential to predict risk of spontaneous miscarriage. The Progesterone Risk of Miscarriage Index (Prog-RMI) allows a clinician to arrive at the predicted probability of spontaneous miscarriage with the readily quantifiable values of progesterone, fetal heart and maternal BMI, thus allowing for ease of implementation into routine clinical practice in the management of women presenting with threatened miscarriage. This will guide possible therapeutic interventions in women at high risk of miscarriage and allow clinicians to reassure women at low risk. Our study is limited by a relatively small sample size. However, the accuracy and precision of the predictive model was impressive with an AUC of 0.923. The margin of error for the predicted probability of miscarriage was also relatively tight, especially at the lower and higher values of progesterone. Because pregnancy losses were followed only to the end of the 16th week our model may not apply to losses occurring after that period. However there is value in an early prognostic tool for preparing women at high risk of miscarriage in early pregnancy. Future studies will include a larger population with more diverse ethnicities, BMI and behavioral characteristics to validate the proposed predictive model and address the impact that other factors may have on subsequent complete miscarriage and other pregnancy complications. Our study findings lead us to propose a novel method for predicting the risk of spontaneous miscarriage. Clinicians can, in the future, have a useful clinical tool for risk stratification and prognosis of completed miscarriage following threatened miscarriage using three readily obtainable and clinically relevant risk factors: progesterone levels, presence/absence of fetal heartbeat and maternal BMI to more accurately predict the risk of miscarriage using our Progesterone Risk of Miscarriage Index.
Declaration of interest Each author represents and warrants that he or she has no financial affiliation (e.g. employment, direct payments, stock holdings, retainers, consultantships, patent-licensing arrangements, or honoraria) or involvement within the last 3 years with any commercial organization with a potential financial interest in the subject or material discussed in the manuscript. The authors declare that there are no conflicts of interest.
References 1. Cunningham F. Williams obstetrics. New York: McGraw-Hill; 2001. 2. Jouppila P. Clinical consequences after ultrasonic diagnosis of intrauterine hematoma in threatened abortion. J Clin Ultrasound 1985;13:107–11. 3. Basama FM, Crosfill F. The outcome of pregnancies in 182 women with threatened miscarriage. Arch Gynecol Obstet 2004;270:86–90. 4. Arck PC, Rucke M, Rose M, et al. Early risk factors for miscarriage: a prospective cohort study in pregnant women. Reprod Biomed Online 2008;17:101–13.
Progesterone miscarriage
DOI: 10.3109/09513590.2015.1031103
5. Arck P, Hansen PJ, Mulac Jericevic B, et al. Progesterone during pregnancy: endocrine-immune cross talk in mammalian species and the role of stress. Am J Reprod Immunol 2007;58:268–79. 6. Duan L, Yan D, Zeng W, et al. Predictive power progesterone combined with beta human chorionic gonadotropin measurements in the outcome of threatened miscarriage. Arch Gynecol Obstet 2011; 283:431–5. 7. Arck PC, Rucke M, Rose M, et al. Early risk factors for miscarriage: a prospective cohort study in pregnant women. Reprod Biomed Online 2008;17:101–13. 8. al-Sebai MA, Kingsland CR, Diver M, et al. The role of a single progesterone measurement in the diagnosis of early pregnancy failure and the prognosis of fetal viability. Br J Obstet Gynaecol 1995;102:364–9.
9. Szekeres-Bartho J, Barakonyi A, Polgar B, et al. The role of gamma/ delta T cells in progesterone-mediated immunomodulation during pregnancy: a review. Am J Reprod Immunol 1999;42:44–8. 10. Szekeres-Bartho J. Progesterone-mediated immunomodulation in pregnancy: its relevance to leukocyte immunotherapy of recurrent miscarriage. Immunotherapy 2009;1:873–82. 11. Kelemen K, Bognar I, Paal M, Szekeres-Bartho J. A progesteroneinduced protein increases the synthesis of asymmetric antibodies. Cell Immunol 1996;167:129–34. 12. Hudic I, Fatusic Z. Progesterone-induced blocking factor (PIBF) and Th(1)/Th(2) cytokine in women with threatened spontaneous abortion. J Perinat Med 2009;37:338–42. 13. Regan L, Rai R. Epidemiology and the medical causes of miscarriage. Baillieres Best Pract Res Clin Obstet Gynaecol 2000; 14:839–54.
Supplementary material available online Supplementary Figure 1 and Supplementary Table I Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
5
ORIGINAL ARTICLE
How can we better predict the risk of spontaneous miscarriage among women experiencing threatened miscarriage? Chee Wai Ku1,2, John C. Allen Jr1, Rahul Malhotra1, Han Chung Chong3, Nguan Soon Tan3,4, Truls Østbye1, Sze Min Lek1, Desiree Lie1, and Thiam Chye Tan1,2
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
1
Duke-National University of Singapore Graduate Medical School, Singapore, 2Department of Obstetrics and Gynecology, KK Women’s and Children’s Hospital, Singapore, 3School of Biological Sciences, Nanyang Technological University, Singapore, and 4Institute of Molecular and Cell Biology, A*STAR, Singapore Abstract
Keywords
This study seeks to establish progesterone and progesterone-induced blocking factor (PIBF) levels as predictors of subsequent completed miscarriage among women presenting with threatened miscarriage between 6 and 10 weeks of gestation. Our secondary objective was to assess the known maternal risk factors, toward development of a parsimonious and clinicianfriendly risk assessment model for predicting completed miscarriage. In this article, we present a prospective cohort study of 119 patients presenting with threatened miscarriage from gestation weeks 6 to 10 at a tertiary women’s hospital emergency unit in Singapore. Thirty (25.2%) women had a spontaneous miscarriage. Low progesterone and PIBF levels are similarly predictive of subsequent completed miscarriage. Study results (OR, 95% CI) showed that higher levels of progesterone (0.91, 95% CI 0.88–0.94) and PIBF (0.99, 95% CI 0.98–0.99) were associated with lower risk of miscarriage. Low progesterone level was a very strong predictor of miscarriage risk in our study despite previous concerns about its pulsatile secretion. Low serum progesterone and PIBF levels predicted spontaneous miscarriage among women presenting with threatened miscarriage between gestation weeks 6 to 10. Predictive models to calculate probability of spontaneous miscarriage based on serum progesterone, together with maternal BMI and fetal heart are proposed.
Miscarriage, predictive, progesterone, progesterone-induced blocking factor
Introduction Threatened miscarriage, defined as an ongoing pregnancy associated with vaginal bleeding, with or without abdominal pain [1], is the most common gynecological emergency, occurring in 15–20% of pregnancies [2]. Approximately 15% of women with threatened miscarriage progress to completed spontaneous miscarriage [3]. Currently, there are no standardized clinicianfriendly miscarriage risk assessment tools and no standard progesterone or Progesterone-Induced Blocking Factor (PIBF) cutoff levels accepted as ‘‘low risk’’ [4]. This study proposes progesterone and its downstream effector PIBF as biomarkers for risk assessment of spontaneous miscarriage in women who present in early pregnancy with threatened miscarriage. Progesterone is a critical hormone during implantation. It sustains decidualization, controls uterine contractility and promotes maternal immune tolerance to the fetal semi-allograft [5]. Risk of miscarriage is significantly higher among women with low serum progesterone [4], although cutoff levels for predicting completed miscarriage vary from 512 to 516 ng/mL among studies [6–8]. Progesterone-induced blocking factor (PIBF) is a pivotal mediator in progesterone-dependent immunomodulation [9,10]
Address for correspondence: Chee Wai Ku, KK Women’s and Children’s Hospital, Singapore. E-mail: [email protected]
History Received 16 December 2014 Revised 13 February 2015 Accepted 16 March 2015 Published online 2 June 2015
and has a regulatory role on anti-fetal immune responses during pregnancy [11]. In a recent study PIBF levels were significantly lower in women with threatened miscarriage who progressed to completed miscarriage compared with women who progressed to delivery [12]. Unlike progesterone, PIBF is not secreted in a pulsatile fashion and therefore not subject to gestational variability in serum concentration. Thus, it may be a better predictor of completed miscarriage than serum progesterone levels. Our primary aim was to establish progesterone and PIBF levels as predictors of subsequent completed miscarriage among women presenting with threatened miscarriage between 6 and 10 weeks of gestation. Our secondary objective was to assess known risk factors [4] towards development of a parsimonious and clinicianfriendly risk assessment model for predicting risk of completed miscarriage. The institutional review board at SingHealth (CIRB ref: 2010/ 620/D) approved the study.
Methods The study was conducted at the Kandang Kerbau Women’s and Children’s Hospital (KKH), the largest maternity hospital in Singapore. All obstetric emergencies are presented through the emergency unit of KKH. Study participants were women aged 19 years and above presenting at the KKH emergency unit between October 2011 and February 2012. Study inclusion criteria were a single intrauterine pregnancy and presenting with pregnancy-related vaginal
2
C. W. Ku et al.
Gynecol Endocrinol, Early Online: 1–5
bleeding between 6 and 10 weeks of gestation, calculated by ultrasonography using fetal pole size. For women with no fetal pole (but with the presence of yolk sac) on ultrasound (17.6%), the number of days since the last menstrual period (LMP) was used to calculate gestational age. Study exclusion criteria were previous episodes of vaginal bleeding during the current pregnancy, progesterone treatment in the current pregnancy, diagnosis of inevitable miscarriage, blighted ovum or planned pregnancy termination. Serum progesterone and PIBF levels were obtained from maternal blood samples taken at presentation. Blood was
collected into plain tubes and centrifuged for 10 min at 3000g within 2 h of collection. Serum samples were stored at 20 C until analysis. Serum progesterone was measured in the KKH clinical laboratory using a commercial ARCHITECT progesterone kit (Abbott, Ireland) and serum PIBF was measured in the research laboratory in Nanyang Technological University, Singapore, using a commercial ELISA kit, according to manufacturer’s protocol (Cusabio, China). Information on maternal demographic, health, obstetric and lifestyle factors (Table 1) was collected using a questionnaire administered verbally in either English or Chinese.
Table 1. Serum progesterone and PIBFa levels, and maternal characteristics at baseline, overall and by pregnancy status at 16 weeks gestation.
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
Pregnancy status at 16 weeks gestation Maternal characteristics Serum biological markers Progesterone, mean ± SD (nmol/L) PIBFa, mean ± SD (ng/ml) Demographics Age, mean ± SD (years) Age of spouse, mean ± SD (years) Marital status Married Single Highest educational level University degree (%) ITE or polytechnic (%) Primary and secondary (%) Number of children None (%) 1 or more (%) Health, obstetric and lifestyle factors Patients on progesterone after threatened miscarriage (%) Planned pregnancy Planned (%) Unplanned (%) Fetal Pole Present (%) Absent (%) Fetal Heart Present (%) Absent (%) Gestation age at presentation with threatened miscarriage By dates, mean (weeks) By CRLb, mean (weeks) Previous miscarriage Yes (%) No (%) BMIc, mean ± SD (kg/m2) Medical comorbidities Diabetes mellitus (%) Hypertension (%) Thyroid disease (%) Yes (%) No (%) Gynecological disease (%) Yes (%) No (%) Smoking Yes (%) No (%) Exposed to second hand smoke at home Yes (%) No (%) Experienced nausea during pregnancy (%) Yes (%) No (%) PSSd, mean ± SD (score) a
b
Total (N ¼ 119)
Ongoing pregnancy (N ¼ 89)
Spontaneous miscarriage at or before (N ¼ 30)
p Value
51.4 ± 26.4 255 ± 136
60.0 ± 23.5 293 ± 127
26.4 ± 17.1 142 ± 92
50.0001 50.0001
29.8 ± 4.9 33.2 ± 5.4
29.3 ± 4.8 32.9 ± 5.4
31.3 ± 4.7 33.9 ± 5.1
0.05 0.37
91.6 8.4
92.1 7.9
90.0 10.0
0.71
43.7 31.9 24.4
42.7 30.3 27.0
46.7 36.7 16.6
0.55
52.1 47.9
55.1 44.9
43.3 56.7
0.30
96.6
97.8
93.3
0.60
60.5 39.5
59.6 40.4
63.3 36.7
0.83
82.4 17.6
93.3 6.7
50.0 50.0
50.0001
73.1 26.9 8.0 ± 1.5
87.6 12.4 8.0 ± 1.5
30.0 70.0 8.0 ± 1.7
50.0001
7.3 ± 1.5
7.6 ± 1.5
6.3 ± 0.8
0.0003
20.2 79.8 23.3 ± 4.84
20.2 79.8 23.7 ± 5.1
20.0 80.0 22.1 ± 3.8
0.77
1.00 0.11
0 0
0 0
0 0
0.8 99.2
1.1 98.9
0 100.0
1.00
9.2 90.8
12.4 87.6
0 100.0
0.06
2.5 97.5
3.4 96.6
0 100.0
0.57
37.8 62.2
40.5 59.5
30.0 70.0
0.39
63.0 37.0 17.2 ± 5.8
70.8 29.2 16.9 ± 5.5
40.0 60.0 18.1 ± 6.7
0.004
c
d
PIBF ¼ progesterone-induced blocking factor, CRL ¼ crown-rump length, BMI ¼ body mass index, PSS ¼ perceived stress scale.
0.34
Progesterone miscarriage
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
DOI: 10.3109/09513590.2015.1031103
The primary outcome of interest was spontaneous miscarriage defined as evacuation of the uterus after inevitable or incomplete miscarriage, or complete miscarriage with an empty uterus, by the 16th week of gestation. Most miscarriages occur before week 16 of gestation [13]. Participants were contacted by telephone at the end of week 16 to determine the pregnancy outcome. A priori power analysis was performed using computer simulations on a logistic regression model with progesterone and PIBF as continuous predictors assuming a miscarriage rate of 20%. Multiple scenarios based on published literature [6–8] were simulated. A sample of n ¼ 100 patients was indicated for 80% power at ¼ 0.05 to detect clinically relevant differences in serum progesterone and PIBF levels in women with ongoing pregnancies compared to women experiencing spontaneous miscarriage. Statistical analysis was performed using SAS version 9.2 (SAS Institute Inc., Cary, NC). Baseline values of continuous variables were compared using two-sample t-tests between women with ongoing pregnancy beyond week 16 and women experiencing spontaneous miscarriage before 16 weeks; categorical variables were compared using Fisher’s exact test (2-sided, ¼ 0.05). Univariate and multivariate logistic regression analyses were used to assess progesterone and PIBF levels and other maternal factors on risk of spontaneous miscarriage. Stepwise logistic regression was used to develop a model predictive of miscarriage during the first 16 weeks of pregnancy. A tabular nomogram was constructed from the fitted model.
Results Overall, 306 women were considered for participation between 1 October 2011 and 28 February 2012. Of 119 women analyzed, 89 (74.8%) had ongoing pregnancies at week 16 of gestation and 30 (25.2%) experienced spontaneous miscarriage at or before week 16 of gestation (Supplementary Figure 1). On ultrasound, 21 women (17.6%) did not have fetal pole at presentation and 32 women (26.9%) did not have a fetal heart at presentation. The mean serum progesterone and PIBF were significantly higher in women with ongoing pregnancy (Progesterone: 60.0 ± 23.5 nmol/L; PIBF: 293 ± 127 ng/mL) than in women with spontaneous miscarriage (Progesterone: 26.4 ± 17.1 nmol/ L; PIBF: 142 ± 92 ng/mL) (p50.0001). The two groups did not differ significantly in maternal age, education, medical comorbidities, BMI, gestation age at presentation, or perceived stress as assessed by the PSS score; however, they did differ in gestational age by CRL at presentation, fetal pole and fetal heart (p50.001) (Table 1). Elevated levels of serum progesterone and PIBF were protective against completed miscarriage at 16 weeks (Progesterone: OR ¼ 0.907, 95% CI, 0.875–0.942; PIBF: 0.982, 0.974–0.990). Area (95% CI) under the ROC curve (AUC) was 0.888 (0.812, 0.965) for serum progesterone and 0.832 (0.744, 0.919) for PIBF. AUC for progesterone and PIBF did not differ
statistically at p50.05 based on overlapping confidence intervals. Four candidate predictors were significant at p50.15 in univariate analysis: presence of fetal heart (Y/N) (p50.0001), presence of fetal pole (Y/N) (p50.0001), maternal age (p ¼ 0.055) and BMI (p ¼ 0.116) (Table 2). ROC analysis yielded a cutpoint for serum progesterone of 35 nmol/L (sensitivity ¼ 77%, specificity ¼ 88%, PPV ¼ 68%, NPV ¼ 92%); the PIBF cutpoint was 180 ng/mL (sensitivity ¼ 73%, specificity ¼ 81%, PPV ¼ 58%, NPV ¼ 91%). Prevalence of spontaneous miscarriage in our cohort was 25%. Four factors were selected in the stepwise multivariate analysis (p50.15): BMI, progesterone, PIBF and fetal heart, with AUC ¼ 0.924 (Table 3). Using a best-subsets algorithm, a parsimonious 3-variable model evolved incorporating progesterone (p50.001), fetal heart (p ¼ 0.019) and BMI (p ¼ 0.003) with AUC ¼ 0.923 (Table 4). The Progesterone Risk of Miscarriage Index (Prog-RMI) giving 16-week predicted probability of miscarriage for fetal heart (present/absent), BMI (within the study range) and serum progesterone concentration is presented in
Table 3. Stepwise multivariate analysis for predicting spontaneous miscarriage.
Parameter
Odds ratio (95% CIa)
p-value
0.215 (0.055, 0.837) 0.925 (0.883, 0.969) 0.849 (0.730, 0.988) 0.992 (0.983, 1.002)
0.034
Fetal Heart (Y/N) Progesterone (nmol/L) BMIc (kg/m2) PIBFd (ng/ml)
Odds ratio (95% CIa)
Progesterone (nmol/L) PIBFc (ng/ml) Fetal Heart (Y/N) Fetal Pole (Y/N) Maternal age (years) BMId (kg/m2)
0.907 0.982 0.060 0.072 1.092 0.923
a
(0.875, (0.974, (0.022, (0.024, (0.998, (0.836,
0.942) 0.990) 0.165) 0.216) 1.195) 1.020)
Cumulative AUCb (95% CI) 0.788 (0.698, 878) 0.905 (0.830, 0.980) 0.923 (0.858, 0.988) 0.924 (0.858, 0.989)
50.0001 0.027 0.106
a
CI ¼ confidence interval, bAUC ¼ area under the ROC curve, cPIBF ¼ progesterone-induced blocking factor, dBMI ¼ body mass index.
Table 4. ‘‘Best subset’’ 3-factor model to predict spontaneous miscarriage: progesterone, fetal heart and BMI.
Parameter
Odds ratio (95% CIa)
p-value
0.910 (0.871, 0.952) 0.206 (0.055, 0.773) 0.850 (0.733, 0.986)
50.001
Progesterone (nmol/L) Fetal Heart (N/Y) BMIc (kg/m2) a
CI ¼ confidence interval, c BMI ¼ body mass index.
Cumulative AUCb (95% CI) 0.888 (0.812, 0.965) 0.905 (0.830, 0.980) 0.923 (0.858, 0.988)
0.019 0.031
b
AUC ¼ area
under
Table 2. Univariate-based analysis for predicting spontaneous miscarriage. Parameter
3
p-value
AUCb
50.0001 50.0001 50.0001 50.0001 0.055 0.116
0.888 (0.812, 0.965) 0.832 (0.744, 0.919) 0.788 (0.698, 0.878) 0.7163 (0.622, 0.811) 0.619 (0.501, 0.736) 0.581 (0.463, 0.698)
CI ¼ confidence interval, bAUC ¼ area under the ROC curve, cPIBF ¼ progesterone-induced blocking factor, dBMI ¼ body mass index.
the
ROC
curve,
4
C. W. Ku et al.
Gynecol Endocrinol, Early Online: 1–5
Table 5. Progesterone Risk of Miscarriage Index (Prog-RMI): Predicted probability (%) of spontaneous miscarriage by 16 weeks gestation (PPSM-16) in women presenting with threatened miscarriage from 6 to 10 weeks of gestation based on progesterone concentration, fetal heart and BMI.
Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
Fetal Heart No Yes Progesterone BMI (kg/m2) BMI (kg/m2) (nmol/L) 15–20 >20–25 >25 15–20 >20–25 >25 ≤5 >5–10 >10–15 >15–20 >20–25 >25–30 >30–35 >35–40 >40–45 >45–50 >50–55 >55–60 >60
Legend:
99 98 97 95 92 88 83 75 65 54 42 31 22
97 96 93 90 84 77 68 57 45 34 24 17 11
94 91 86 79 70 60 48 37 27 19 12 8 5
94 91 86 80 71 61 49 38 28 19 13 8 5
88 82 74 64 52 41 30 21 14 10 6 4 3
76 67 56 44 33 23 16 11 7 4 3 2 1
0–19 20–49 50–74 75–100
Table 5 (See Supplement Table I for associated ROC summary table.)
Discussion We conducted a prospective cohort study of women presenting with threatened miscarriage at 6–10 weeks gestation to assess progesterone and PIBF levels at the time of presentation as predictors of subsequent completed miscarriage in combination with other known risk factors. Low progesterone and PIBF levels were both found to be predictive of subsequent completed miscarriage, with better accuracy (AUC) from progesterone – although the difference was not statistically significant. Our findings suggest that either progesterone or PIBF can be used independently for prognosticating spontaneous miscarriage by 16 weeks. Low progesterone was a strong predictor of miscarriage in our study with no apparent adverse effect on prognostic accuracy due to pulsatile secretion. It is possible that pulsatile secretion is assuaged in women presenting with high risk of threatened miscarriage, but this remains to be investigated. PIBF is a relatively new biomarker not routinely available in clinical laboratories. Our findings suggest that PIBF should be considered for future use in the prognosis and management of threatened miscarriage. Our proposed progesterone cutpoint of 35 nmol/L (11.0 ng/mL) (NPV 92%) for diagnosis of subsequent completed miscarriage is consistent with other cutpoints reported in the literature. In a retrospective study of 175 women with threatened miscarriage, Duan et al. proposed a serum progesterone cutpoint of 16 ng/mL (50.7 nmol/L) [6] for diagnosis of inevitable miscarriage. More recently, Arck et al. reported that serum progesterone levels 12 ng/mL (38.3 nmol/L) were associated with increased risk of completed miscarriage [4]. If the 35 nmol/L (11.0 ng/mL) progesterone cutpoint is validated in subsequent studies, it might be used to reliably predict women at low risk of miscarriage and exclude them from further follow-up and unnecessary treatment.
In addition to progesterone and PIBF, four other significant independent predictors of completed spontaneous miscarriage were identified: absence of fetal heart (OR ¼ 16.7), absence of fetal pole (OR ¼ 13.9), increase in maternal age (OR ¼ 1.09 or 9% increase in risk per year older in age) and reduction in BMI (OR ¼ 1.08 or 8% increase in risk per unit reduction (kg/m2) in BMI). Interestingly, despite many studies alluding to the importance of fetal heart in predicting viability of the fetus, and hence lowering the risk of spontaneous miscarriage, it was not as powerful a predictor as compared to progesterone and PIBF (AUC ¼ 0.788 versus 0.888 and 0.832 respectively). In the best subset analysis, a 3-factor model comprising progesterone, fetal heart and BMI has a cumulative AUC of 0.923, which has great potential to predict risk of spontaneous miscarriage. The Progesterone Risk of Miscarriage Index (Prog-RMI) allows a clinician to arrive at the predicted probability of spontaneous miscarriage with the readily quantifiable values of progesterone, fetal heart and maternal BMI, thus allowing for ease of implementation into routine clinical practice in the management of women presenting with threatened miscarriage. This will guide possible therapeutic interventions in women at high risk of miscarriage and allow clinicians to reassure women at low risk. Our study is limited by a relatively small sample size. However, the accuracy and precision of the predictive model was impressive with an AUC of 0.923. The margin of error for the predicted probability of miscarriage was also relatively tight, especially at the lower and higher values of progesterone. Because pregnancy losses were followed only to the end of the 16th week our model may not apply to losses occurring after that period. However there is value in an early prognostic tool for preparing women at high risk of miscarriage in early pregnancy. Future studies will include a larger population with more diverse ethnicities, BMI and behavioral characteristics to validate the proposed predictive model and address the impact that other factors may have on subsequent complete miscarriage and other pregnancy complications. Our study findings lead us to propose a novel method for predicting the risk of spontaneous miscarriage. Clinicians can, in the future, have a useful clinical tool for risk stratification and prognosis of completed miscarriage following threatened miscarriage using three readily obtainable and clinically relevant risk factors: progesterone levels, presence/absence of fetal heartbeat and maternal BMI to more accurately predict the risk of miscarriage using our Progesterone Risk of Miscarriage Index.
Declaration of interest Each author represents and warrants that he or she has no financial affiliation (e.g. employment, direct payments, stock holdings, retainers, consultantships, patent-licensing arrangements, or honoraria) or involvement within the last 3 years with any commercial organization with a potential financial interest in the subject or material discussed in the manuscript. The authors declare that there are no conflicts of interest.
References 1. Cunningham F. Williams obstetrics. New York: McGraw-Hill; 2001. 2. Jouppila P. Clinical consequences after ultrasonic diagnosis of intrauterine hematoma in threatened abortion. J Clin Ultrasound 1985;13:107–11. 3. Basama FM, Crosfill F. The outcome of pregnancies in 182 women with threatened miscarriage. Arch Gynecol Obstet 2004;270:86–90. 4. Arck PC, Rucke M, Rose M, et al. Early risk factors for miscarriage: a prospective cohort study in pregnant women. Reprod Biomed Online 2008;17:101–13.
Progesterone miscarriage
DOI: 10.3109/09513590.2015.1031103
5. Arck P, Hansen PJ, Mulac Jericevic B, et al. Progesterone during pregnancy: endocrine-immune cross talk in mammalian species and the role of stress. Am J Reprod Immunol 2007;58:268–79. 6. Duan L, Yan D, Zeng W, et al. Predictive power progesterone combined with beta human chorionic gonadotropin measurements in the outcome of threatened miscarriage. Arch Gynecol Obstet 2011; 283:431–5. 7. Arck PC, Rucke M, Rose M, et al. Early risk factors for miscarriage: a prospective cohort study in pregnant women. Reprod Biomed Online 2008;17:101–13. 8. al-Sebai MA, Kingsland CR, Diver M, et al. The role of a single progesterone measurement in the diagnosis of early pregnancy failure and the prognosis of fetal viability. Br J Obstet Gynaecol 1995;102:364–9.
9. Szekeres-Bartho J, Barakonyi A, Polgar B, et al. The role of gamma/ delta T cells in progesterone-mediated immunomodulation during pregnancy: a review. Am J Reprod Immunol 1999;42:44–8. 10. Szekeres-Bartho J. Progesterone-mediated immunomodulation in pregnancy: its relevance to leukocyte immunotherapy of recurrent miscarriage. Immunotherapy 2009;1:873–82. 11. Kelemen K, Bognar I, Paal M, Szekeres-Bartho J. A progesteroneinduced protein increases the synthesis of asymmetric antibodies. Cell Immunol 1996;167:129–34. 12. Hudic I, Fatusic Z. Progesterone-induced blocking factor (PIBF) and Th(1)/Th(2) cytokine in women with threatened spontaneous abortion. J Perinat Med 2009;37:338–42. 13. Regan L, Rai R. Epidemiology and the medical causes of miscarriage. Baillieres Best Pract Res Clin Obstet Gynaecol 2000; 14:839–54.
Supplementary material available online Supplementary Figure 1 and Supplementary Table I Gynecol Endocrinol Downloaded from informahealthcare.com by Nyu Medical Center on 06/04/15 For personal use only.
5
