Materials and Methods
We conducted a prospective observational study of singleton pregnancies in patients 18 years old or older. All patients were attending our center for routine first-trimester screening for aneuploidy between 11 weeks 0 days and 14 weeks 0 days of gestation. The dataset also includes 277 patients from our previous study on the association between 3DPD and preeclampsia.
The study was approved by our institutional review board and written informed consent was obtained from all participants. Patients with multiple gestation or fetal anomalies were excluded. Demographics were collected from prenatal electronic medical records or obtained via interview if electronic medical record data were incomplete. Data collected included maternal age, parity, ethnicity, medical history (including chronic hypertension, diabetes, antiphospholipid syndrome, sickle cell disease, and tobacco use), and obstetric history (including previous fetal demise, FGR, or preeclampsia). Maternal weight and height were obtained and the body mass index (BMI) was calculated.
Because there were no previous studies using our 3DPD method in FGR, an a priori power calculation was performed using the mean difference in placental Doppler indices in patients who developed preeclampsia compared with those who did not from our prior publication. Using that assumption, we determined we would need 59 patients with FGR for vascularization index (VI), 60 patients with FGR for flow index (FI), and 50 patients with FGR for vascularization flow index (VFI) to predict a difference of 6.7 for VI, 5.6 for FI, and 4.6 for VFI, if it exists, between groups with 80% power and type I error rate of α = 0.05.
Ultrasound and 3DPD evaluation were performed by 1 of 3 study authors (P.D., J.G., and M.R.) using the Voluson 730 Expert (GE, Milwaukee, WI). Crown-rump length and nuchal translucency measurements were obtained, and then attention was turned to the uteropacental interface while avoiding the fetus. Power Doppler settings were preadjusted on all machines to allow the capture of weak signals that are common in the intervillous space. Settings were maintained constant in all cases as follows: image quality high, 1; color gain, 1.6; pulse repetition frequency, 0.6 kHz; and wall motion filter, 50 Hz. All other Doppler and sonographic settings were at the manufacturer’s default settings.
After documenting the placental location, a 3-dimensional volume box was placed to contain the myometrium and placenta. A sweeping angle of 50 degrees was used while the 3-dimensional volume was acquired. The area with maximal vascularity was identified while scrolling through the sectional planes. Using the virtual organ computer-aided analysis program (VOCAL; GE), poles were placed at the upper edge of the UPCS and at the placental surface abutting the amniotic fluid to include the area of maximum vascularity. A spherical volume biopsy was then obtained and analyzed to determine the VI, FI, and VFI. The VI is the ratio between color voxels and total voxels expressed in percentages and is thought to approximate the number of vessels in the selected area. The FI is the sum of the color voxels’ signal intensity divided by the number of voxels, indicating the average intensity of blood flow and is expressed as a whole number between 0 and 100. The VFI is the sum of the color voxels’ signal intensity divided by the total tissue voxels, representing both vascularization and blood flow, and is also quantified as a whole number between 0 and 100.
Interobserver and intraobserver variability were evaluated for a subgroup of patients by 2 of the study authors (J.G. and M.R.). For interobserver variability, the offline evaluation of a single volume acquisition was independently evaluated by each author. For intraobserver variability, the same author (J.G.) then evaluated the same volume at a greater than 6 month interval.
Outcome data were collected for all patients who delivered at our institution or who were reachable by phone contact provided at the time of enrollment. The primary outcome was FGR, defined as a birthweight less than the 10th percentile, which was evaluated in 2 ways: first, using a traditional population birthweight centile table (<10 pop ), and second, using a birthweight less than the 10 th percentile of the optimal weight for gestational age based on a formula customized for each patient (<10 cust ). The customized formula considered the following individual data for each patient: maternal BMI, parity, neonatal sex, and gestational age at delivery.
Statistical analysis
The 3DPD indices were compared between patients who had neonates with FGR (using population and customized growth curves) and unaffected patients. Maternal and demographic characteristics for both the growth restriction groups and the non–growth-restriction groups were compared using a Student t test or Wilcoxon’s tests when comparison groups were not asymptotically normal. The χ 2 test was used for all categorical data that had sufficient sample size, and a Fisher exact test was used for sparse categorical data.
Interobserver and intraobserver variability and reproducibility were calculated using the intraclass correlation coefficients. The impact of shell size/placental thickness and distance of the uteroplacental interface to the skin on variation in 3DPD parameters was assessed using mixed-effects analysis of variance.
The distribution of VI, FI, and VFI was analyzed using the normal probability plots and Kolmogorov-Smirnov test. Because color Doppler might be affected by physical variables, we used a linear multivariate regression to identify possible confounding parameters that may be associated with Doppler performance. We then adjusted for these confounders using linear regression models and used the adjusted indices for the correlation and comparison analyses. We first assessed for correlation between adjusted Doppler indices and birthweight percentiles, using both population-based and customized centile curves. We then compared the mean difference in VI, FI, and VFI between patients who had FGR and unaffected patients using a Student t test. A value of P < .05 was considered statistically significant.
SAS version 9.2 (SAS Institute, Cary, NC) was used for statistical analysis.
Results
Five hundred seventy-seven patients enrolled in the study. Eleven patients were excluded for fetal anomalies, 1 patient had a voluntary termination, and 10 patients had spontaneous abortions. Outcome data were obtained by phone from 7 patients who did not deliver at our institution within 2 months from the expected due date. Twenty-nine patients were lost to follow-up.
Of the remaining 526 patients, 68 (13%) delivered infants at less than 10 pop . There were 29 patients for whom customized birthweight centiles could not be calculated because of incomplete data. Of the remaining 497, 63 (13%) delivered infants at less than 10 cust . Thirty-five patients were reclassified; 18 patients (3.6%) had neonates that were less than 0 pop but were considered normal weight by the customized formula, and 17 patients (3.4%) with neonates born at less than 10 cust were considered greater than the 10th percentile for birthweight based on the population-based table.
The demographic characteristics and pregnancy-related data were compared in the patients whose neonates were born less than 10 pop and those who were less than 10 cust with those who had appropriate for gestational age weight neonates, greater than 10 pop and greater than 10 cust , respectively ( Table 1 ). Neonates in the less than 10 pop group were more likely to be born to nulliparous mothers, were more likely to be female, and had a statistically although not clinically significant earlier gestational age at delivery (38.8 vs 39.3 weeks).
| Characteristic | Population (n = 526) | Customized (n = 497) | ||||
|---|---|---|---|---|---|---|
| <10% (n = 68) (13%) | ≥10% (n = 458) (87%) | P value | <10% (n = 63) (13%) | ≥10% (n = 434) (87%) | P value | |
| Maternal age, y | 28.1 ± 5.5 | 28.4 ± 5.6 | .70 | 28.7 ± 5.2 | 28.3 ± 5.7 | .57 |
| Ethnicity | .09 | .27 | ||||
| White | 5 (7.5%) | 57 (12.6%) | 6 (9.5%) | 54 (12.4%) | ||
| African American | 28 (41.8%) | 135 (29.7%) | 25 (39.7%) | 128 (29.5%) | ||
| Hispanic | 21 (31.3%) | 194 (42.7%) | 21 (33.3%) | 183 (42.2%) | ||
| Asian | 12 (17.9%) | 66 (14.5%) | 10 (15.9%) | 67 (15.4%) | ||
| Other | 1 (1.5%) | 2 (0.4%) | 1 (1.6%) | 2 (0.5%) | ||
| Nulliparity | .02 | .93 | ||||
| Yes | 34 (50%) | 159 (34.9%) | 23 (36.5%) | 156 (35.9%) | ||
| No | 34 (50%) | 297 (65.1%) | 40 (63.5%) | 278 (64.1%) | ||
| BMI, kg/m 2 | 27.1 ± 6.4 | 27.9 ± 5.9 | .33 | 28.1 ± 5.8 | 27.7 ± 5.9 | .64 |
| Tobacco use | .09 | .70 | ||||
| None | 55 (90.2%) | 355 (92.2%) | 52 (91.2%) | 340 (91.9%) | ||
| Before pregnancy | 6 (9.8%) | 17 (4.4%) | 4 (7.0%) | 18 (4.9%) | ||
| During pregnancy | 0 (0%) | 13 (3.4%) | 1 (1.8%) | 11 (3.0%) | ||
| Diabetes | 1.00 | .33 | ||||
| Yes | 3 (4.4%) | 21 (4.6%) | 4 (6.4%) | 17 (4.0%) | ||
| No | 65 (95.6%) | 436 (95.4%) | 59 (93.7%) | 416 (96.0%) | ||
| Hypertension | .06 | .15 | ||||
| Yes | 9 (13.2%) | 31 (6.8%) | 8 (12.7%) | 30 (6.9%) | ||
| No | 59 (86.8%) | 426 (93.2%) | 55 (87.3%) | 403 (93.1%) | ||
| Prior preeclampsia a | .54 | .56 | ||||
| Yes | 4 (11.8%) | 27 (9.1%) | 5 (12.5%) | 25 (9.0%) | ||
| No | 30 (88.2%) | 269 (90.9%) | 35 (87.5%) | 252 (91%) | ||
| Prior FGR a | .64 | .19 | ||||
| Yes | 2 (7.1%) | 8 (3.2%) | 3 (8.8%) | 9 (3.9%) | ||
| No | 26 (92.9%) | 245 (96.8%) | 31 (91.2%) | 221 (96.1%) | ||
| Prior fetal demise a | 1.00 | .6 | ||||
| Yes | 0 (0.0%) | 8 (3.2%) | 0 (0.0%) | 8 (3.4%) | ||
| No | 29 (100%) | 245 (96.8%) | 33 (100%) | 229 (96.6%) | ||
| Sickle cell disease | .41 | .72 | ||||
| Yes | 0 (0.0%) | 2 (0.6%) | 0 (0.0%) | 2 (0.6%) | ||
| No | 51 (94.4%) | 339 (96.9%) | 49 (96.1%) | 320 (97.0%) | ||
| SCT | 3 (5.6%) | 9 (2.6%) | 2 (3.9%) | 8 (2.4%) | ||
| APLS/autoimmune disease | .02 | .02 | ||||
| Yes | 3 (4.4%) | 2 (0.4%) | 3 (4.8%) | 2 (0.5%) | ||
| No | 65 (95.6%) | 454 (99.6%) | 60 (95.2%) | 431 (99.5%) | ||
| CRL, mm | 63.7 ± 7.9 | 63.3 ± 8.4 | .73 | 62.8 ± 8.3 | 63.5 ± 8.4 | .50 |
| Placental location | .87 | .43 | ||||
| Anterior/fundal | 38 (55.9%) | 251 (54.8%) | 38 (60.3%) | 239 (55.1%) | ||
| Posterior | 30 (44.1%) | 207 (45.2%) | 25 (39.7%) | 195 (44.9%) | ||
| Preeclampsia | .11 | .02 | ||||
| Yes | 10 (14.7%) | 38 (8.3%) | 10 (15.9%) | 33 (7.6%) | ||
| No | 52 (76.5%) | 392 (85.8%) | 46 (73.0%) | 374 (86.4%) | ||
| GHTN | 6 (8.8%) | 27 (5.9%) | 7 (11.1%) | 26 (6.0%) | ||
| Delivery GA, wks | 38.8 (24.1–41.3) | 39.3 (26–41.4) | .0004 | 39.1 (24.1–41.3) | 39.3 (26–41.6) | .99 |
| Infant weight, g | 2425.6 ± 539.8 | 3353.2 ± 495.8 | < .001 | 2544.4 ± 608.3 | 3335.2 ± 515.6 | < .001 |
| Infant sex | .01 | .72 | ||||
| Female | 43 (63.2%) | 212 (47.2%) | 32 (50.8%) | 210 (48.4%) | ||
| Male | 25 (36.8%) | 237 (52.8%) | 31 (49.2%) | 224 (51.6%) | ||
As expected, when customized centiles that correct for fetal gender and parity were used, there was no difference in these characteristics between the less than 10 cust and the greater than 10 cust groups. However, less than 10 cust neonates were more likely to be born to women with preeclampsia. There was no difference in gestational age at delivery between the less than 10 cust and the greater than10 cust groups.
There was no difference between ethnicity, BMI, smoking, diabetes, chronic hypertension, prior preeclampsia or FGR, crown rump length, or placental location between any of the groups. Neonates classified as less than the 10th percentile by either method were more likely to have mothers with antiphospholipid syndrome or an autoimmune disorder.
Inter- and intraobserver correlation was calculated for 119 patients. The interobserver variability, intraobserver variability, and reproducibility were good for all vascular parameters ( Table 2 ).
