Abstract
Objective
Delta-like homolog 1 (DLK1) is a growth factor that is reduced in maternal sera in pregnancies with small for gestational age neonates. We sought to determine if DLK1 is associated with stillbirth (SB), with and without placental insufficiency.
Study Design
A nested case-control study was performed using maternal sera from a multicenter case-control study of SB and live birth (LB). SB and LB were stratified as placental insufficiency cases (small for gestational age <5% or circulatory lesions on placental histopathology) or normal placenta controls (appropriate for gestational age and no circulatory lesions). Enzyme-linked immunosorbent assay (ELISA) was used to measure DLK1. The mean difference in DLK1 was compared on the log scale in an adjusted linear regression model with pairwise differences, stratified by term/preterm deliveries among DLK1 results in the quantifiable range. In exploratory analysis, geometric means were compared among all data and the proportion of “low DLK1” (less than the median value for gestational age) was compared between groups and modeled using linear and logistic regression, respectively.
Results
Overall, 234 SB and 234 LB were analyzed; 246 DLK1 values were quantifiable within the standard curve. Pairwise comparisons of case and control DLK1 geometric means showed no significant differences between groups. In exploratory analysis of all data, adjusted analysis revealed a significant difference for the LB comparison only (SB: 71.9 vs. 99.1 pg/mL, p = 0.097; LB: 37.6 vs. 98.1 pg/mL, p = 0.005). In exploratory analysis of “low DLK1,” there was a significant difference between the odds ratio of having “low DLK1” between preterm cases and controls for both SB and LB. There were no significant differences in geometric means nor “low DLK1” between SB and LB.
Conclusion
In exploratory analysis, more placental insufficiency cases in preterm SB and LB had “low DLK1.” However, low DLK1 levels were not associated with SB.
Keywords: stillbirth, placental insufficiency, biomarker, growth restriction
Stillbirth (SB), defined as fetal death after 20 weeks’ gestation, occurs in 5.96 per 1,000 births in the United States.1 The rate of SB in the United States exceeds that of many other high-income countries.2 To reduce the SB rate, better predictive measures are needed.3 Placental insufficiency is a recognized risk factor for SB and is a broad term describing disordered placental function.4,5 It has been used in reference to clinical phenotypes including fetal growth restriction (FGR), SB, and preeclampsia, as well as abnormal measurements of placental function including ultrasound Doppler, magnetic resonance imaging (MRI), and biomarkers.6 However, fetal growth is the most readily utilized measure of placental insufficiency in clinical practice.
Using fetal growth as a marker of fetal well-being presents several challenges. Currently, screening for fetal growth abnormalities is primarily guided by preexisting maternal risk factors (e.g., chronic hypertension), acquired pregnancy conditions (e.g., gestational hypertension), or an abnormal fundal height measurement.7 There is no consensus regarding the optimal definition for FGR. Hence, various fetal growth and birth weight curves are commonly used, with similarly varied correlation with neonatal outcomes.8,9 Additionally, there is not a widely accepted method to discern a pathologically small fetus from one that is constitutionally small. Methods currently in use indicate pathologic growth restriction and thus increased risk of adverse fetal and infant outcomes including measurement of umbilical and uterine artery Doppler velocimetry, middle cerebral artery Doppler velocimetry, ductus venosus flow, and the cerebroplacental ratio. Of these, the umbilical artery Doppler has the most consistent association with neonatal outcomes and is recommended for risk stratification of the growth-restricted fetus.7,10
Recently, Cleaton et al demonstrated that delta-like homolog 1 (DLK1), a growth factor involved in adipose homeostasis, is derived from the fetus and is expressed in decreased amounts in maternal serum in the setting of FGR with abnormal umbilical artery Doppler flow.11 DLK1 may be involved in the modulation of maternal metabolism to promptly switch fatty acid oxidation in the accelerated starvation process that occurs in pregnant individuals.12 This accelerated starvation response is thought to be a protective mechanism by which to ensure adequate fetal nutrition.13 When DLK1 is absent, there is impaired maternal adaptation to the metabolic needs of pregnancy as reflected by lower maternal ketone levels and a positive association with DLK and fetal size.11 Thus, we sought to investigate DLK1 as a biomarker of placental insufficiency among SB and LB, with and without placental insufficiency.
Materials and Methods
This was a nested case-control study using maternal sera from a multicenter case-control study of SB and LB. The original study was performed from 2006 to 2008 with enrollment of mothers of SB and LB at the time of delivery. Attempts were made to enroll all deliveries with SB and a representative sample of those with LB with oversampling of preterm births (PTBs) and non-Hispanic black women to facilitate stratified analyses. Study participants were recruited from five catchment areas in Rhode Island, Massachusetts, Georgia, Utah, and Texas. Study design and methodology have been previously described in detail.14 Consenting participants had blood drawn at the time of enrollment around the time of delivery. The blood was collected in serum separator tubes and maternal serum was stored at −80°C.
DLK1 was measured in undiluted, thawed maternal serum by enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s specifications. (R&D Systems, Minneapolis, MN). Samples were run in duplicate and a human serum standard was used. Data were expressed in pg/mL. The intraassay coefficient of variation was 7.4% and the interassay coefficient of variation was 5.9%. The standard curve included DLK1 values from 93.75 to 6,000 pg/mL. The standard curve was generated using a 4-parameter logistic curve fit.
Placental insufficiency category was the “key” exposure in this analysis. Placental insufficiency was defined as either having a small for gestational age (SGA; <5th percentile) birth weight or circulatory placental lesions on histopathology.15 Circulatory placental lesions included any retroplacental hematoma, chorionic plate fetal vascular thrombus, parenchymal infarction, avascular villi, and placental edema. Normal placenta controls did not have any of these features and included pregnancies with an appropriate for gestational age birth weight infant and no circulatory placental lesions. Deliveries with placental insufficiency comprised our cases, while deliveries without placental insufficiency were normal placental controls.
We categorized eligible deliveries into eight mutually exclusive strata defined by combinations of three binary characteristics as follows: (1) SB or LB, (2) gestational age at delivery (term ≥37 or preterm <37 weeks’), and (3) placental insufficiency (case or control). These categories were developed in an attempt to refine our clinical phenotypes as much as possible. From all eight stratification categories, we excluded deliveries prior to 20 weeks’ gestational age, fetal anomalies, multiple gestations, and deliveries for which a detailed placental analysis was not available. Women were also excluded if no stored serum was available.
From this eligible subset of Stillbirth Collaborative Research Network (SCRN) deliveries, we sampled evenly without replacement from each category (Fig. 1). With an intention of including 468 deliveries, we aimed to include 59 deliveries from each of the eight categories. However, there were insufficient samples available to include 59 deliveries in each group. Where 59 deliveries were not available in a category that stratum was sampled exhaustively, and additional deliveries were included from other categories of the same SB or LB status.
Fig. 1.
Study design. Selection of placental insufficiency cases and normal placenta controls with stratification by SB, LB, and term/preterm gestational age. LB, live birth; SB, stillbirth.
With the total number of available deliveries known before samples were processed for DLK1 (n = 468), we conducted a feasibility analysis and determined that 34 SB term cases of placental insufficiency and 21 SB term deliveries without placental insufficiency would achieve 80% power to reject the null hypothesis of equal means. Population mean difference was assumed to be 10.3 with standard deviations of 15.7 (controls) and 11.1 (cases), with α = 0.05, using a two-sided two-sample unequal-variance t-test.11 For other pairwise comparisons or pooled-group comparisons, we anticipated adequate power for detecting smaller differences as group sizes are larger.
Demographic and clinical characteristics were summarized for deliveries in each of eight study strata. Within one linear regression model, the following two sets of pairwise comparisons of log DLK1 within the quantifiable range were tested: (1) between placental insufficiency groups stratified by the four PTB and SB/LB categories, and (2) between SB and LB groups stratified by the four PTB and placental insufficiency categories. In an exploratory analysis, the same two sets of four pairwise comparisons were conducted including all DLK1 values (both within and outside the standard curve). Results from comparisons on the log scale were transformed using exponentiation and reported as geometric mean and 95% confidence interval (CI). Results are presented from unadjusted and adjusted models, with adjusted models including maternal age, body mass index, tobacco use, fetal sex, hypertensive disorders, and race.
For the primary endpoint, we conducted two sensitivity analyses. First, we excluded large for gestational age (LGA) deliveries (birth weight ≥ 90th percentile). Second, among LB normal-placental preterm deliveries, we excluded iatrogenic labor (medically induced labor or delivery without labor).
In a second exploratory analyses utilizing all results, “low DLK1” was defined as values falling below rank-based threshold of DLK1 level stratified by dichotomous gestational age at delivery (50th percentile, lower cut-offs were not feasible given available samples) and below the lower limit of detection. The proportion of “low DLK1” was compared between groups by a Chi-square test and modeled using logistic regression. In both linear and logistic adjusted models, maternal age, body mass index, tobacco use, fetal sex, hypertensive disorders, and race–ethnicity were included as covariates.
All analyses were conducted in SAS 9.4, using survey design–specific procedures, and SCRN analytical weights created for the purposes of placental analysis, reflecting the study design and availability of placentas (SAS Institute, Cary NC). The placental examination data weights were constructed to account for differences between patients who did and did not have a placental histologic examination performed.14
Results
Of the 518 SB and 1,200 LB with placental pathology results available, 234 SB and 234 LB were analyzed. Demographic variables of SB and LB groups were compared between placental insufficiency cases and normal placenta controls and are shown in Tables 1 and 2, respectively. Characteristics significantly differed in smoking and income between preterm LBs (Table 2), and payment type among term SB deliveries (Table 1), but otherwise there were no significant differences between groups, other than the expected differences in birth weight and gestational age given the study design.
Table 1.
Stillbirth demographics
| Characteristic | Value | Term | Preterm | ||||
|---|---|---|---|---|---|---|---|
| Placental category | Insufficient | Normal | p-Value | Insufficient | Normal | p-Value | |
| Unweighted (n) | 34 | 21 | 126 | 53 | |||
| Weighted (n) | 36 | 19 | 126 | 49 | |||
| Maternal age (at delivery) | <20 | 4 (12.5) | 2 (9.6) | 0.889 | 19 (15.4) | 3 (5.7) | 0.172 |
| 20–34 | 24 (66.8) | 14 (73.3) | 84 (66.3) | 40 (80.7) | |||
| 35+ | 7 (20.7) | 3 (17.1) | 23 (18.3) | 7 (13.6) | |||
| Maternal BMI | Mean (SE) | 27.2 (1.08) | 29.6 (1.99) | 0.29 | 27.5 (0.56) | 27.0 (0.96) | 0.651 |
| Maternal BMI (categorized) | <25 (under/normal weight) | 18 (51.0) | 8 (40.9) | 0.249 | 53 (43.1) | 20 (43.2) | 0.96 |
| 25–29.9 (overweight) | 6 (16.9) | 5 (25.3) | 37 (30.6) | 13 (27.0) | |||
| 30–34 (obese) | 7 (18.3) | 1 (4.3) | 14 (11.1) | 6 (12.6) | |||
| 35+ (morbidly obese) | 5 (13.8) | 5 (29.5) | 19 (15.3) | 8 (17.2) | |||
| Maternal race/ethnicity | White non-Hispanic | 13 (36.8) | 6 (32.8) | 0.394 | 42 (33.4) | 19 (38.0) | 0.509 |
| Black, non-Hispanic | 11 (30.4) | 2 (12.2) | 22 (17.3) | 11 (23.0) | |||
| Hispanic | 10 (28.6) | 9 (45.5) | 50 (39.7) | 17 (34.4) | |||
| Other | 1 (4.2) | 2 (9.4) | 12 (9.7) | 2 (4.6) | |||
| Marital status | Not married or cohabitating | 4 (11.7) | 2 (10.9) | 0.368 | 36 (29.1) | 14 (28.2) | 0.36 |
| Cohabitating | 6 (18.3) | 6 (36.0) | 30 (24.6) | 7 (15.5) | |||
| Married | 24 (69.9) | 9 (53.1) | 57 (46.2) | 27 (56.2) | |||
| Maternal education | 0–11 (secondary or less) | 6 (19.3) | 4 (25.4) | 0.857 | 23 (18.9) | 11 (22.0) | 0.595 |
| 12 (completed secondary) | 10 (30.0) | 5 (30.5) | 38 (31.1) | 11 (23.2) | |||
| 13+ (college) | 17 (50.8) | 7 (44.1) | 62 (50.0) | 26 (54.8) | |||
| Tobacco use (3 months prior to pregnancy) | Did not smoke | 27 (78.2) | 12 (72.5) | 0.564 | 102 (82.7) | 39 (81.9) | 0.936 |
| <10 cigarettes/day | 4 (11.5) | 4 (21.8) | 11 (9.2) | 4 (8.5) | |||
| 10+ cigarettes/day | 4 (10.4) | 1 (5.7) | 10 (8.1) | 5 (9.7) | |||
| Insurance/method of payment | Any public/private assistance | 25 (70.0) | 8 (39.2) | 0.022 | 79 (63.4) | 26 (53.4) | 0.226 |
| Commercial/HMO/military | 11 (30.0) | 12 (60.8) | 46 (36.6) | 23 (46.6) | |||
| Family income (last 12 months) | Any assistance | 18 (51.6) | 6 (36.9) | 0.316 | 57 (46.5) | 21 (43.6) | 0.735 |
| Only personal income | 17 (48.4) | 10 (63.1) | 65 (53.5) | 27 (56.4) | |||
| Prenatal care | 35 (96.3) | 19 (100) | NA | 115 (91.2) | 44 (88.7) | 0.627 | |
| Preeclampsia/gestational HTN | 5 (12.7) | 2 (9.2) | 0.692 | 15 (12.3) | 2 (4.0) | 0.108 | |
| Chronic hypertension | 6 (16.6) | 3 (13.9) | 0.794 | 13 (10.0) | 5 (9.6) | 0.934 | |
| Type 1 or 2 diabetes | 3 (9.3) | 1 (3.5) | 0.37 | 7 (5.4) | 3 (6.8) | 0.746 | |
| Gestational diabetes | 5 (12.9) | 3 (15.8) | 0.781 | 3 (2.0) | 2 (3.6) | 0.539 | |
| Birth weight (g) | Mean (SE) | 3,263 (102.9) | 3,416 (141.6) | 0.384 | 1,091 (81.76) | 1,446 (145.8) | 0.035 |
| Infant sex | Male | 19 (54.3) | 10 (51.2) | 0.825 | 62 (49.5) | 31 (62.8) | 0.117 |
Abbreviations: BMI, body mass index; HTN, hypertension; NA, not available; SE, standard error.
Note: Values presented as n (column%) unless otherwise noted.
Table 2.
Live birth demographics
| Characteristic | Value | Term | Preterm | ||||
|---|---|---|---|---|---|---|---|
| Placental category | Insufficient | Normal | p-Value | Insufficient | Normal | p-Value | |
| Unweighted (n) | 59 | 59 | 57 | 59 | |||
| Weighted (n) | 57 | 55 | 16 | 26 | |||
| Maternal age (at delivery) | <20 | 5 (9.0) | 4 (7.4) | 0.948 | 1 (7.5) | 3 (11.9) | 0.75 |
| 20–34 | 45 (78.8) | 44 (79.2) | 13 (77.2) | 20 (77.9) | |||
| 35+ | 7 (12.2) | 7 (13.4) | 2 (15.2) | 3 (10.2) | |||
| Maternal BMI | Mean (SE) | 24.7 (0.70) | 25.2 (0.74) | 0.577 | 27.3 (1.36) | 27.9 (1.24) | 0.718 |
| Maternal BMI (categorized) | <25 (under/normal weight) | 34 (61.4) | 32 (58.9) | 0.289 | 7 (45.5) | 11 (43.2) | 0.957 |
| 25–29.9 (overweight) | 14 (25.8) | 10 (17.5) | 4 (24.1) | 5 (20.9) | |||
| 30–34 (obese) | 5 (8.5) | 11 (20.3) | 3 (21.0) | 6 (21.2) | |||
| 35+ (morbidly obese) | 2 (4.3) | 2 (3.3) | 1 (9.4) | 4 (14.8) | |||
| Maternal race/ethnicity | White non-Hispanic | 31 (55.4) | 23 (41.2) | 0.243 | 4 (24.8) | 8 (30.1) | 0.415 |
| Black, non-Hispanic | 3 (4.5) | 8 (14.1) | 2 (14.7) | 4 (14.6) | |||
| Hispanic | 18 (32.2) | 22 (39.1) | 10 (59.5) | 12 (46.5) | |||
| Other | 4 (7.9) | 3 (5.7) | 0 (1.0) | 2 (8.8) | |||
| Marital status | Not married or cohabitating | 10 (17.1) | 5 (8.6) | 0.235 | 3 (21.2) | 6 (25.8) | 0.911 |
| Cohabitating | 15 (27.7) | 12 (22.0) | 5 (29.7) | 8 (31.0) | |||
| Married | 31 (55.2) | 38 (69.4) | 8 (49.1) | 11 (43.2) | |||
| Maternal education | 0–11 (secondary or less) | 12 (21.4) | 14 (25.4) | 0.848 | 4 (23.9) | 5 (21.0) | 0.274 |
| 12 (completed secondary) | 14 (25.1) | 12 (21.3) | 4 (24.1) | 11 (45.1) | |||
| 13+ (college) | 30 (53.5) | 29 (53.3) | 8 (52.0) | 8 (33.9) | |||
| Tobacco use (3 months prior to pregnancy) | Did not smoke | 45 (81.4) | 50 (92.0) | 0.104 | 15 (98.3) | 20 (80.4) | <.001 |
| <10 cigarettes/day | 6 (11.5) | 1 (2.3) | 0 (1.2) | 1 (2.1) | |||
| 10+ cigarettes/day | 4 (7.1) | 3 (5.7) | 0 (0.5) | 4 (17.4) | |||
| Insurance/method of payment | Any public/private assistance | 28 (49.7) | 29 (53.0) | 0.741 | 13 (82.7) | 21 (79.4) | 0.756 |
| Commercial/HMO/military | 28 (50.3) | 26 (47.0) | 3 (17.3) | 5 (20.6) | |||
| Family income (last 12 months) | Any assistance | 25 (44.7) | 27 (49.2) | 0.649 | 6 (40.2) | 17 (69.7) | 0.026 |
| Only personal income | 31 (55.3) | 28 (50.8) | 9 (59.8) | 7 (30.3) | |||
| Prenatal care | 57 (100) | 55 (100) | NA | 14 (88.1) | 24 (91.4) | 0.669 | |
| Preeclampsia/gestational HTN | 3 (5.4) | 4 (7.5) | 0.631 | 4 (23.7) | 5 (18.7) | 0.673 | |
| Chronic hypertension | 2 (3.1) | 1 (1.6) | 0.559 | 3 (16.9) | 2 (8.1) | 0.316 | |
| Type 1 or 2 diabetes | 3 (4.6) | 1 (2.6) | 0.527 | 0 (0) | 3 (11.5) | NA | |
| Gestational diabetes | 4 (7.2) | 10 (18.0) | 0.104 | 3 (16.6) | 3 (11.3) | 0.633 | |
| Birth weight (g) | Mean (SE) | 3,226 (82.12) | 3,528 (61.98) | 0.004 | 2,280 (141.4) | 2,632 (110.3) | 0.052 |
| Infant sex | Male | 28 (48.9) | 22 (39.1) | 0.313 | 9 (54.7) | 16 (59.9) | 0.715 |
Abbreviations: BMI, body mass index; HTN, hypertension; NA, not available; SE, standard error.
Note: Values presented as n (column%) unless otherwise noted.
Overall, 246 (52.6%) DLK1 values were quantifiable within the standard curve as depicted in Fig. 2 between dotted horizontal reference lines. DLK1 among preterm deliveries was distributed as follows: 43.8 as the 20th percentile, 64.9 as the 33rd percentile, and 96.4 as the 50th percentile. DLK1 among term deliveries was distributed as follows: 92.8 as the 20th percentile, 110.5 as the 33rd percentile, and 153.7 as the 50th percentile.
Fig. 2.
DLK1 values. The dotted lines denote the bounds of the standard curve. DLK1 is measured in pg/mL and plotted on a log scale. DLK1, delta-like homolog 1.
In the primary comparison, we analyzed data within the standard curve (Table 3). First, four pairwise comparisons across placental insufficiency category were assessed within each combination of SB and PTB; these showed no significant differences. Second, in the same model, DLK1 values were compared between SB and LB within each of four categories defined by placental insufficiency and PTB; there were no significant differences in DLK1 level according to SB or LB status. This interpretation was consistent across unadjusted and adjusted comparisons.
Table 3.
Pairwise comparison of DLK1 (within the standard curve)
| Study group | Unadjusted Placental category Insufficient (cases) |
Normal (controls) | p-Value(1) | Adjusteda Placental category Insufficient (cases) |
Normal (controls) | p-Value(1) |
|---|---|---|---|---|---|---|
| SB term | 240.4 (191.2–302.3) | 233.2 (169.8–320.2) | 0.8777 | 216.6 (161.7–290.2) | 230.7 (162.5–327.6) | 0.7501 |
| LB term | 267.0 (202.5–352.0) | 205.4 (163.8–257.6) | 0.1493 | 220.2 (166.3–291.5) | 188.2 (136.0–260.4) | 0.3647 |
| p-Value(2) | 0.5656 | 0.5217 | 0.9269 | 0.276 | ||
| SB preterm | 217.8 (184.7–256.8) | 217.4 (182.8–258.6) | 0.9882 | 201.6 (157.2–258.6) | 183.9 (138.4–244.6) | 0.5009 |
| LB preterm | 197.4 (133.1–292.8) | 177.7 (147.9–213.4) | 0.6332 | 164.6 (99.5–272.2) | 158.2 (118.7–210.8) | 0.8738 |
| p-Value(2) | 0.6511 | 0.1163 | 0.41 | 0.2901 |
Abbreviations: DLK1, delta-like homolog 1; LB, live birth; SB, stillbirth.
Notes: Analysis of DLK-1 values within the standard curve. Values reported as geometric mean and 95% confidence interval.
Two sets of four pairwise comparisons are reported from each model,
between placental insufficiency cases and normal placenta controls, and
between SB and LB deliveries.
Adjusted for maternal age, body mass index, tobacco use, fetal sex, hypertensive disorders, and race–ethnicity
In exploratory analyses, we analyzed all data within and beyond the standard curve, (Table 4). First, four pairwise comparisons across placental insufficiency cases and normal placenta controls were compared within each combination of SB and PTB. In unadjusted analyses, there were significantly lower levels of DLK1 in the preterm SB and LB cases compared with controls (SB: 93.1 vs. 139.6 pg/mL, p = 0.031; LB: 52.0 vs. 126.4 pg/mL, p = 0.009). In adjusted analysis, this significant difference remained for the LB comparison only (SB: 71.9 vs. 99.1 pg/mL, p = 0.097; LB: 37.6 vs. 98.1 pg/mL, p = 0.005). Second, in the same model, DLK1 values were compared between SB and LB within each of four categories defined by placental insufficiency and PTB; no significant differences in DLK1 levels were present.
Table 4.
Exploratory analysis of all DLK1 values (exploratory analysis of all DLK1 values)
| Unadjusted Placental category Insufficient (cases) |
Normal (controls) | p-Value(1) | Adjusteda Placental category Insufficient (cases) |
Normal (controls) | p-Value(1) | |
|---|---|---|---|---|---|---|
| SB term | 219.5 (171.8–280.4) | 129.5 (77.3–216.8) | 0.0697 | 172.3 (121.0–245.5) | 119.8 (67.2–213.7) | 0.2067 |
| LB term | 167.5 (122.8–228.4) | 127.8 (101.5–161.0) | 0.1703 | 124.5 (81.5–190.2) | 95.4 (63.3–143.9) | 0.1810 |
| p-Value(2) | 0.1792 | 0.9642 | 0.1104 | 0.4118 | ||
| SB preterm | 93.1 (75.8–114.3) | 139.6 (102.9–189.4) | 0.0309 | 71.9 (50.0–103.5) | 99.1 (65.9–149.1) | 0.0973 |
| LB preterm | 52.0 (27.8–97.4) | 126.4 (103.1–155.0) | 0.0085 | 37.6 (19.4–72.8) | 98.1 (68.7–140.0) | 0.0049 |
| p-Value(2) | 0.0838 | 0.5956 | 0.0526 | 0.9582 |
Abbreviations: DLK1, delta-like homolog 1; LB, live birth; SB, stillbirth.
Notes: Exploratory analysis of all DLK-1 values. Values reported as geometric mean and 95% confidence interval.
Two sets of four pairwise comparisons are reported from each model,
between placental insufficiency cases and normal placenta controls, and
between SB (SB) and LB deliveries.
Adjusted for maternal age, body mass index, tobacco use, fetal sex, hypertensive disorders, and race-ethnicity
In sensitivity analyses excluding LGA deliveries and iatrogenic preterm LB deliveries, results were robust. First, excluding n = 21 LGA deliveries across strata for analyses of in-range DLK1, unadjusted and adjusted findings were similar to that reported in Table 3; our eight key comparisons remained nonsignificant, and geometric mean estimates were generally within 10% of the estimate including LGA. Comparisons including all data within and beyond the standard curve were similar to those in Table 4; significant differences in DLK1 across placental insufficiency among SB preterm, LB preterm deliveries persisted in unadjusted analysis, and among LB preterm deliveries in adjusted analyses when LGA deliveries were excluded. Second, excluding n = 8 of27 LB preterm deliveries with iatrogenic labor, for in-range DLK1, unadjusted and adjusted findings were similar to that reported in Table 3; comparisons with LB preterm deliveries remained nonsignificant, the geometric mean estimate was lower when limited to spontaneous preterm birth (SPTB). Comparisons including all data within and beyond the standard curve were similar to that in Table 4 when n = 19 iatrogenic deliveries were excluded from 56 preterm LB normal-placental deliveries. In the pairwise comparison of DLK1 across placental insufficiency category among LB preterm deliveries, the significant difference persisted in unadjusted and adjusted analyses when limited to SPTB livebirth normal-placental deliveries.
With values for the 20th percentile, and 33 percentile falling below the lower limit of detection for one or both groups, results are presented from low DLK1 defined as below the median (i.e., 50th percentile) and below the lower limit of detection. The proportion of cases having “low DLK1” values and corresponding p-values are shown in the left panel of Fig. 3. In this analysis, there was a significant difference between the odds ratio (OR) of having “low DLK1” between preterm cases and controls for both SB and LB (SB: adjusted OR [aOR] = 2.33, 95% CI: 1.12–4.84; LB: aOR = 5.89 95% CI: 1.68–20.62). There were no significant differences in “low DLK1” between SB and LB (right panel, Fig. 3).
Fig. 3.
Exploratory analysis of “low DLK1” in all data. The proportions and associated p-values are shown in the left panel. The odds ratio of having “low DLK1” was compared between groups, shown in the right panel. “Low DLK1” defined as values below the gestational age specific median, <96.36 pg/mL for preterm deliveries, <153.68 pg/mL for term deliveries. Adjustment includes maternal age, body mass index, tobacco use, fetal sex, hypertensive disorders, and race–ethnicity. CI, confidence interval; DLK1, delta-like homolog 1.
The level of DLK1 among a more rigorously defined placental insufficiency group was assessed by comparing cases meeting both SGA <5th percentile and circulatory placental lesion criteria to cases meeting only one criterion, as well as to controls. Mean levels of DLK1 among cases meeting both criteria (190.2 pg/mL [95% CI: 112.5, 321.7]) were not significantly different from cases meeting only one criterion (237.7 pg/mL [95% CI: 207.3, 272.6] p = 0.4), or controls (207.1 pg/mL [95% CI: 184.9, 232.1] p = 0.8), adjusted for gestational age and SB/LB. Similarly, mean levels of DLK1 among cases with SGA <5th percentile (220.2 pg/mL [95% CI: 172.2, 281.6]) were not significantly different from either cases defined without SGA (238.2 pg/mL [95% CI: 204.6, 277.5] p = 0.6), or controls (207.1 pg/mL [95% CI: 184.9, 232.1] p = 0.7), adjusted for gestational age and SB/LB.
Discussion
Circulating maternal DLK1 has previously been shown to be lower in pregnancies affected by pathologic FGR as compared with those with normal weight fetuses and those with constitutionally small fetuses.11 We investigated this association in pregnancies with placental insufficiency and in those affected by SB. We found no difference in DLK1 levels between placental insufficiency cases and controls in both LB and SB pregnancies. We also found no difference in DLK1 level between LB and SB when matched for gestational age and placental insufficiency case or normal placenta control status. However, in exploratory analysis of all results, preterm LB placental insufficiency cases had significantly lower values of DLK1 compared with normal placenta controls.
Our analysis is limited by the fact that 47.4% of results fell outside of the standard curve, thus limiting our power to detect significant differences between groups. Our DLK1 levels were also substantially lower than those observed in prior work.11 This may be due to stored sample degradation since our samples were collected between 2006 and 2008. However, other proteins have been stable after storage at ≥80°C from the same sample set.16–18 Additionally, prior work has been successful in assessing DLK1 levels in stored maternal plasma.11 Another explanation for low DLK1 levels may be that we included both preterm and term deliveries. Term and preterm groups were chosen to account for differential DLK1 expression through gestation, although we cannot fully correct for differences in placental development and soluble protein expression. Circulating maternal DLK1 increases as pregnancy progresses and inclusion of preterm cases and controls may have resulted in a larger percentage of low levels which were not quantifiable within the standard curve.11 Further examination of those samples that fell outside of the standard curve demonstrated that indeed many of these arose from preterm and lower birth weight samples. This limited threshold which we could choose in defining “low DLK1,” precluding us from selecting lower rank-based definitions. The characteristics of DLK1 expression through gestation in humans is not yet well characterized.19,20 Another factor that may have impacted DLK1 levels is that our samples were collected around the time of delivery while prior work examined maternal DLK1 level in ongoing pregnancies.11 The impact of delivery on maternally circulating DLK1 levels is not known but given that DLK1 is of fetal origin, it may decrease rapidly following delivery of the fetus.
Additionally, our ability to detect significant differences in DLK1 between cases and controls was likely impacted by the availability of placental insufficiency cases. We defined placental insufficiency by neonates with SGA <5% or circulatory placental lesions. This definition is less robust than cases with both clinical and histologic evidence of placental insufficiency and likely includes some constitutionally small infants with normal placentas. We were not able to meet both criteria for all cases owing to fixed availability of samples and did not have antenatal data regarding other measures of FGR such as umbilical artery Doppler measurements. There was also significant overlap in the DLK1 values between cases and controls indicating that this is not an effective discriminatory marker in this population.
Strengths and Limitations
A strength of our work is access to a large cohort of SB cases to study DLK1 levels. Despite this, we did not observe significant differences in DLK1 between SB and LB of matched case or control and gestational age status. This suggests that DLK1 is associated with placental insufficiency but not SB. Our results may have been affected by a limited number of term placental insufficiency SB cases with a resultant oversampling of preterm SB cases (Fig. 1). With a larger number of term patients in which a more rigorous definition of placental insufficiency could be applied, we might have observed a more pronounced difference in DLK1 levels as seen in prior work.11 Additionally, it is possible that maternal DLK1 levels may have fallen following an antepartum SB with varying retention time prior to diagnosis and delivery. Cleaton et al demonstrated that maternally circulating DLK1 is derived from the fetus but it is not clear how an antepartum fetal death may affect its expression.11 Despite these unknown factors, many of the DLK1 values in the SB deliveries were within the standard curve which suggests that some amount of DLK1 persists after fetal death.
Improved detection of pathologic FGR and pregnancies at risk for SB is needed. Data informing routine ultrasound screening for FGR in the third trimester are limited.21,22 While routine serial ultrasonography improves detection of FGR, the challenge of differentiating pathologic from constitutional FGR remains.23 As discussed, third trimester ultrasound use is based on risk stratification incorporating prepregnancy and pregnancy characteristics with use of Doppler velocimetry to differentiate constitutional versus pathologic FGR.24–27 Given the imperfect prediction of pathologic FGR using ultrasound measures, other biomarkers have been explored. However, currently available genetic screening, angiogenic, hormonal, endothelial stress, and cytokine biomarkers have low positive predictive value for pathologic FGR.28
Conclusion
Given these knowledge gaps, novel approaches for identifying those pregnancies at the highest risk for placental insufficiency and resultant adverse outcomes are needed. Our data, in addition to existing literature, suggest that maternally circulating DLK1 is a promising biomarker of placental insufficiency but not SB. DLK1 deserves further study to better define its positive predictive value and utility for adverse clinical outcomes in pregnancies at risk for placental insufficiency and SB.
Key Points.
Maternally circulating DLK1 is correlated with placental insufficiency.
Maternally circulating DLK1 is not correlated with SB.
DLK1 is a promising marker for placental insufficiency.
Funding
This analysis was supported by Oregon Health and Science University through grant no.: U01-HD087182 for Ms. Gaffney’s effort. This investigation was supported by the University of Utah Study Design and Biostatistics Center, with funding in part from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through grant no: 8UL1TR000105 (formerly grant no.: UL1RR025764). This work, including the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review and approval of the manuscript, was supported by grant funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, grant no.: U10-HD045953, Brown University, RI; grant no.: U10-HD045925 Emory University, GA; grant no.: U10-HD045952 University of Texas Medical Branch at Galveston, TX; grant no.: U10-HDO45955 University of Texas Health Sciences Center at San Antonio, TX; rant no.: U10-HD045944 University of Utah Health Sciences Center, UT; and grant no.: U01-HD045954 RTI International, RTP.
Footnotes
This study was conducted at 59 hospitals in portions of five states: Rhode Island, Massachusetts, Georgia, Texas, and Utah.
Conflict of Interest
None declared.
References
- 1.MacDorman MF, Gregory ECW. Fetal and perinatal mortality: United States, 2013. Natl Vital Stat Rep 2015;64(08):1–24 [PubMed] [Google Scholar]
- 2.Flenady V, Wojcieszek AM, Middleton P, et al. ; Lancet Ending Preventable Stillbirths study group Lancet Stillbirths In High-Income Countries Investigator Group. Stillbirths: recall to action in high-income countries. Lancet 2016;387(10019):691–702 [DOI] [PubMed] [Google Scholar]
- 3.Page JM, Silver RM. Interventions to prevent stillbirth. Semin Fetal Neonatal Med 2017;22(03):135–145 [DOI] [PubMed] [Google Scholar]
- 4.Stillbirth Collaborative Research Network Writing Group. Causes of death among stillbirths. JAMA 2011;306(22):2459–2468 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Getahun D, Ananth CV, Kinzler WL. Risk factors for antepartum and intrapartum stillbirth: a population-based study. Am J Obstet Gynecol 2007;196(06):499–507 [DOI] [PubMed] [Google Scholar]
- 6.Hunt K, Kennedy SH, Vatish M. Definitions and reporting of placental insufficiency in biomedical journals: a review of the literature. Eur J Obstet Gynecol Reprod Biol 2016;205:146–149 [DOI] [PubMed] [Google Scholar]
- 7.Fetal Growth Restriction. Fetal growth restriction: ACOG practice bulletin, number 227. Obstet Gynecol 2021;137(02):e16–e28 [DOI] [PubMed] [Google Scholar]
- 8.Blue NR, Beddow ME, Savabi M, Katukuri VR, Mozurkewich EL, Chao CR. A comparison of methods for the diagnosis of fetal growth restriction between the Royal College of Obstetricians and Gynaecologists and the American College of Obstetricians and Gynecologists. Obstet Gynecol 2018;131(05):835–841 [DOI] [PubMed] [Google Scholar]
- 9.Blue NR, Beddow ME, Savabi M, Katukuri VR, Chao CR. Comparing the Hadlock fetal growth standard to the Eunice Kennedy Shriver National Institute of Child Health and Human Development racial/ethnic standard for the prediction of neonatal morbidity and small for gestational age. Am J Obstet Gynecol 2018;219(05):474.e1–474.e12 [DOI] [PubMed] [Google Scholar]
- 10.Martins JG, Biggio JR, Abuhamad A Society for Maternal-Fetal Medicine (SMFM). Electronic address: pubs@smfm.org. Society for Maternal-Fetal Medicine Consult Series #52: diagnosis and management of fetal growth restriction: (replaces Clinical Guideline Number 3, April 2012). Am J Obstet Gynecol 2020;223(04):B2–B17 [DOI] [PubMed] [Google Scholar]
- 11.Cleaton MAM, Dent CL, Howard M, et al. Fetus-derived DLK1 is required for maternal metabolic adaptations to pregnancy and is associated with fetal growth restriction. Nat Genet 2016;48(12):1473–1480 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Charalambous M, Da Rocha ST, Radford EJ, et al. DLK1/PREF1 regulates nutrient metabolism and protects from steatosis. Proc Natl Acad Sci U S A 2014;111(45):16088–16093 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Metzger BE, Ravnikar V, Vileisis RA, Freinkel N. “Accelerated starvation” and the skipped breakfast in late normal pregnancy. Lancet 1982;1(8272):588–592 [DOI] [PubMed] [Google Scholar]
- 14.Parker CB, Hogue CJR, Koch MA, et al. ; Stillbirth Collaborative Research Network. Stillbirth Collaborative Research Network: design, methods and recruitment experience. Paediatr Perinat Epidemiol 2011;25(05):425–435 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Bukowski R, Hansen NI, Willinger M, et al. ; Eunice Kennedy Shriver National Institute of Child Health and Human Development Stillbirth Collaborative Research Network. Fetal growth and risk of stillbirth: a population-based case-control study. PLoS Med 2014;11(04):e1001633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Silver RM, Parker CB, Reddy UM, et al. Antiphospholipid antibodies in stillbirth. Obstet Gynecol 2013;122(03):641–657 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Silver RM, Parker CB, Goldenberg R, et al. Bile acids in a multicenter, population-based case-control study of stillbirth. AmJ Obstet Gynecol 2014;210(05):460.e1–460.e9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Varner MW, Silver RM, Rowland Hogue CJ, et al. ; Eunice Kennedy Shriver National Institute of Child Health and Human Development Stillbirth Collaborative Research Network. Association between stillbirth and illicit drug use and smoking during pregnancy. Obstet Gynecol 2014;123(01):113–125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Floridon C, Jensen CH, Thorsen P, et al. Does fetal antigen 1 (FA1) identify cells with regenerative, endocrine and neuroendocrine potentials? A study of FA1 in embryonic, fetal, and placental tissue and in maternal circulation. Differentiation 2000;66(01):49–59 [DOI] [PubMed] [Google Scholar]
- 20.Carlsson C, Tornehave D, Lindberg K, et al. Growth hormone and prolactin stimulate the expression of rat preadipocyte factor-1/delta-like protein in pancreatic islets: molecular cloning and expression pattern during development and growth of the endocrine pancreas. Endocrinology 1997;138(09):3940–3948 [DOI] [PubMed] [Google Scholar]
- 21.Bricker L, Medley N, Pratt JJ. Routine ultrasound in late pregnancy (after 24 weeks’ gestation). Cochrane Database Syst Rev 2015;(06):CD001451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Monier I, Blondel B, Ego A, Kaminiski M, Goffinet F, Zeitlin J. Poor effectiveness of antenatal detection of fetal growth restriction and consequences for obstetric management and neonatal outcomes: a French national study. BJOG 2015;122(04):518–527 [DOI] [PubMed] [Google Scholar]
- 23.Sovio U, White IR, Dacey A, Pasupathy D, Smith GCS. Screening for fetal growth restriction with universal third trimester ultrasonography in nulliparous women in the Pregnancy Outcome Prediction (POP) study: a prospective cohort study. Lancet 2015;386(10008):2089–2097 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Brar HS, Platt LD. Reverse end-diastolic flow velocity on umbilical artery velocimetry in high-risk pregnancies: an ominous finding with adverse pregnancy outcome. Am J Obstet Gynecol 1988;159(03):559–561 [DOI] [PubMed] [Google Scholar]
- 25.Kalache KD, Dückelmann AM. Doppler in obstetrics: beyond the umbilical artery. Clin Obstet Gynecol 2012;55(01):288–295 [DOI] [PubMed] [Google Scholar]
- 26.Smith GCS, Yu CKH, Papageorghiou AT, Cacho AM, Nicolaides KHF et al. Medicine Foundation Second Trimester Screening Group. Maternal uterine artery Doppler flow velocimetry and the risk of stillbirth. Obstet Gynecol 2007;109(01):144–151 [DOI] [PubMed] [Google Scholar]
- 27.Yu CKH, Smith GCS, Papageorghiou AT, Cacho AM, Nicolaides KHF et al. Medicine Foundation Second Trimester Screening Group. An integrated model for the prediction of pre-eclampsia using maternal factors and uterine artery Doppler velocimetry in unselected low-risk women. Am J Obstet Gynecol 2006;195(01):330. [DOI] [PubMed] [Google Scholar]
- 28.Gaccioli F, Aye ILMH, Sovio U, Charnock-Jones DS, Smith GCS. Screening for fetal growth restriction using fetal biometry combined with maternal biomarkers. Am J Obstet Gynecol 2018;218(2S):S725–S737 [DOI] [PubMed] [Google Scholar]