To examine the relationship between hematologic parameters at birth and prenatal progression of Doppler abnormalities in fetal growth restriction (FGR).
To examine the relationship between hematologic parameters at birth and prenatal progression of Doppler abnormalities in fetal growth restriction (FGR).
The study was a secondary analysis of FGR patients (abdominal circumference < 5th percentile and umbilical artery pulsatility index (UA-PI) elevation) with at least three examinations prior to delivery. Prenatal progression was classified as rapid, moderate or slow based on the interval between diagnosis and delivery and the extent of UA, middle cerebral artery and ductus venosus Doppler abnormalities. Associations between diagnosis-to-delivery interval, Doppler Z-scores, progression and hematologic parameters at birth were examined.
Of 130 patients, 54 (41.5%) had rapid, 51 (39.2%) moderate and 25 (19.2%) slow deterioration, delivering within 4, 6 and 9 weeks of diagnosis, respectively. The strongest association of moderate and rapid deterioration was with a low platelet count (r2 = 0.37 and 0.70, respectively; P < 0.0001). In patients with moderate deterioration, platelet count correlated inversely with UA-PI (ρ = −0.44, P = 0.001) and was lowest when end-diastolic velocity was absent. With rapid progression, platelet count correlated inversely with nucleated red blood cell count (ρ = −0.51, P < 0.001) but no longer with UA-PI.
Our observations suggest a relationship between prenatal clinical progression of FGR and hematologic abnormalities at birth. Accelerating cardiovascular deterioration is associated with decreased platelet count, which can be explained by placental consumption or dysfunctional erythropoiesis and thrombopoiesis. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.
In patients with fetal growth restriction (FGR), the severity and rate of cardiovascular deterioration determine the gestational age at which interventions need to be considered[1-6]. The rate of systemic fetal cardiovascular deterioration is related to increasing blood-flow resistance in the umbilical artery (UA), which reflects the loss of cross-sectional perfusion area in the terminal villous arterial tree[2-4, 7]. These villous vascular abnormalities may originate from faulty placentation early in pregnancy or from acquired loss of previously perfused villi, predisposing the fetus to intraplacental platelet activation, microthrombosis and lower platelet counts at birth and beyond[8-10]. Fetuses with markedly abnormal flow in the UA are at particularly increased risk for neonatal thrombocytopenia. We hypothesize that platelet count at birth is related not only to UA blood-flow resistance prior to delivery, but also to the rate of preceding cardiovascular deterioration. The existence of this association could suggest that the hematologic profile at birth not only is a consequence of placental dysfunction but also may be a contributor to the preceding clinical deterioration in FGR.
The aim of this study was to test the hypothesis that, in patients with FGR due to placental disease, the reduction in platelet count at birth is related to the rate of preceding cardiovascular deterioration.
FGR was studied in a multicenter collaboration in patients enrolled with the following inclusion criteria: (1) singleton pregnancy; (2) gestational age determined by certain last menstrual period and confirmed by sonography before 20 weeks' gestation; and (3) FGR diagnosed by abdominal circumference < 5th percentile and elevated UA pulsatility index (UA-PI) to ≥ 2 SD above the gestational mean[3, 12]. For this secondary analysis we evaluated a subset of prospectively enrolled patients who had had at least three Doppler examinations before delivery and in whom neonatal hematologic parameters had been obtained at birth. Fetuses with chromosomal or structural abnormalities were excluded. All centers utilized a prospectively defined standardized fetal surveillance protocol for enrolled patients. The choice of surveillance intervals and timing of delivery were at the discretion of the managing obstetrician and dictated by the local standard of care. The original study protocol was approved by the appropriate research regulatory authorities at each participating site.
Doppler measurements included UA, middle cerebral artery (MCA) and ductus venosus (DV) PI. We have previously identified three statistically distinct forms of cardiovascular progression in FGR. We applied the same system and categorized clinical progression as slow (defined by UA-PI elevation that did not exceed 3 SD), moderate (deterioration of arterial and venous Doppler parameters taking more than 4 weeks) and rapid (loss of UA end-diastolic velocity within 2 weeks and progression to abnormal DV Doppler (PI > 3 SD above gestational age mean) within 4 weeks).
Perinatal characteristics, delivery details, Apgar scores and umbilical cord arterial pH were ascertained. After delivery, hemoglobin (Hb) concentration, platelet count and nucleated red blood cell (NRBC) count per/100 white blood cells (WBCs) were determined in the hospital laboratory from a venous sample collected in a tube coated with ethylenediaminetetraacetic acid. Significant thrombocytopenia was defined as a manually confirmed platelet count of < 100 000/mL, and an NRBC count > 35 NRBC/100 WBC was considered elevated[9, 11, 14].
Each absolute Doppler index measurement was corrected for gestational age by conversion to a Z-score (SDs from the gestational age mean) using local reference ranges. Absolute Hb values were converted into Z-scores utilizing reference ranges that account for gestational age and timing of blood sampling. These Z-scores were used for all statistical comparisons. Time from enrollment to delivery was computed. Non-parametric or parametric tests were used to analyze continuous variables and chi-square or Fisher's exact test for categorical variables after evaluation for distribution or cell size. The relationship between enrollment-to-delivery interval, pre-delivery Doppler indices and hematologic parameters was tested using correlation analysis. Multiple logistic regression was used to identify determinants of the rate of cardiovascular deterioration and linear regression to find the primary determinants of hematologic parameters. SPSS 19.0 (SPSS Inc., Chicago, IL, USA) was used; P < 0.05 was considered statistically significant and was further adjusted for multiple comparisons.
The inclusion criteria were met by 130 pregnancies complicated by FGR resulting in live births, and recruitment was at a median of 26.6 weeks' gestation (range, 22.5–33.4 weeks). Progression was slow in 25 (19.2%), moderate in 51 (39.2%) and rapid in 54 (41.5%) cases. Comparison of demographic and pregnancy characteristics indicated that the three groups were similar in gestational age at initial diagnosis, degree of growth delay and acid–base status at birth (Table 1). The enrollment-to-delivery intervals and gestational age at delivery reflected the rate of deterioration. Birth weight decreased significantly and major morbidity increased following more rapid deterioration. Comparison of the hematologic profiles at birth showed that rapid progression was associated with the lowest Hb concentration and platelet count but higher NRBC count (Table 2, Figure 1).
|Parameter||All (n = 130)||Slow deterioration|
(n = 25)
|Moderate deterioration (n = 51)||P*||Rapid deterioration (n = 54)||P†||P‡|
|Maternal age (years)||28 (18–45)||27.5 (18–40)||29 (18–45)||—||28 (18–41)||—||—|
|Parity||0 (0–3)||0 (0–2)||0 (0–3)||—||0 (0–2)||—||—|
|Caucasian||89 (68.5)||19 (76.0)||34 (66.7)||—||36 (66.7)||—||—|
|Black||39 (30.0)||6 (24.0)||16 (31.4)||—||17 (31.5)||—||—|
|Other||2 (1.5)||0 (0.0)||1 (2.0)||—||1 (1.9)||—||—|
|GA at enrollment (weeks)||26.6 (22.5–33.4)||27.1 (23.3–34.4)||27.0 (24–33.2)||—||26.3 (22.5–31.1)||—||—|
|Enrollment-to-delivery (days)||30 (2–90)||39 (10–86)||31 (2–90)||—||23 (2–59)||< 0.0001||< 0.001|
|No. of exams||5 (3–26)||5 (3–20)||5 (3–14)||—||6 (3–26)||—||—|
|UA-PI Z-score before delivery||6.8 (0.3–33.5)||2.4 (0.39–6.82)||4.1 (1.6–21.1)||< 0.0001||7.5 (0.3–33.5)||< 0.0001||< 0.0001|
|GA at delivery (weeks)||31.2 (25.5–39.1)||35.3 (28.1–39.1)||32.5 (26.4–38.3)||< 0.0001||29.6 (25.5–36.2)||< 0.0001||< 0.0001|
|Fetal indication for delivery||97 (74.6)||18 (72.0)||36 (70.6)||—||43 (79.6)||—||—|
|Cesarean delivery||112 (86.2)||14 (56.0)||46 (90.2)||< 0.01||52 (96.3)||< 0.0001|
|Birth weight (g)||980 (385–2370)||1415 (650–2370)||1100 (420–2260)||< 0.05||790 (385–1800)||< 0.0001||< 0.0001|
|Birth-weight percentile||0.6 (0–9.3)||0.7 (0.03–3.99)||0.6 (0–3.73)||—||0.2 (0.01–9.3)||—||—|
|5-min Apgar score < 7||11 (8.5)||1 (4.0)||3 (5.9)||—||7 (13.0)||—||—|
|Cord artery pH < 7.20||39 (30.0)||4 (16.0)||19 (37.3)||—||16 (29.6)||—||—|
|IVH||21 (16.2)||0 (0.0)||9 (17.6)||—||12 (22.2)||< 0.05||—|
|§Major morbidity||26 (20.0)||1 (4.0)||9 (17.6)||—||16 (29.6)||< 0.05||—|
|Parameter||All (n = 130)||Slow deterioration (n = 25)||Moderate deterioration (n = 51)||P*||Rapid deterioration (n = 54)||P†||P‡|
|Hb conc. (g/dL)||15.5 (7.0 to 20.2)||17.3 (12.5 to 20.1)||15.6 (8.1 to 20.2)||< 0.05||14.3 (7.0 to 19.6)||< 0.0001||< 0.005|
|Hb Z-score||−0.49 (−4.72 to 1.83)||0.39 (−2.16 to 1.72)||−0.48 (−3.67 to 1.27)||0.105||−0.61 (−4.72 to 1.83)||0.791||0.007|
|NRBC count (/100 WBC)||57 (1–3081)||12 (1–71)||57 (1–621)||< 0.001||190 (2–3081)||< 0.0001||< 0.0001|
|Platelet count (/mL)||145 (32–437)||221 (93–437)||147 (32–310)||< 0.0001||115 (34–318)||< 0.0001||< 0.05|
|Platelet count < 100 000/mL||37 (28)||1 (4)||13 (25)||< 0.05||23 (43)||< 0.001||—|
A shorter enrollment-to-delivery interval correlated with a lower absolute Hb concentration and higher NRBC count (Spearman ρ = 0.33 and −0.38, respectively; P < 0.0001). Pre-delivery UA-PI, MCA-PI and DV-PI correlated with absolute Hb concentration, Hb Z-score and platelet and NRBC count. In addition the duration of DV-PI elevation also correlated with the NRBC count (Spearman ρ = 0.43; P < 0.005). The direction of the correlation coefficients indicates that faster rate and worse Doppler deterioration correlated with lower Hb concentration and platelet count and higher NRBC count. In addition, Hb concentration and platelet count were related directly to each other and inversely to NRBC count (Table 3).
|Parameter||Hemoglobin concentration||Hemoglobin Z-score||Platelet count||NRBC count|
|Last umbilical artery PI Z-score||−0.43||−0.20*||−0.41||0.51|
|Last middle cerebral artery PI Z-score||0.32||0.23†||0.33||−0.35|
|Last ductus venosus PI Z-score||−0.33||−0.21*||−0.29||0.46|
By logistic regression analysis, the strongest hematologic parameter associated with the difference between slow and moderate deterioration was the platelet count (r2 = 0.37, B = 9.147; P < 0.0001). The difference between slow and rapid deterioration was related to low platelet count (r2 = 0.70, B = 6.901; P < 0.0001). Hematologic parameters did not contribute to the difference between moderate and rapid deterioration. The absolute Hb concentration, but not the Hb Z-score, was independently related to the enrollment-to-delivery interval and UA-PI (r2 = 0.14, B = 3.375 and r2 = 0.24, B = −0.100; P = 0.001, respectively), while platelet count was related to the MCA-PI (r2 = 0.23, B = 19.687; P < 0.0001) and the NRBC count to DV-PI and cord artery pH (r2 = 0.16, B = 33.442 and B = 959.845, respectively; P = 0.003). The regression coefficients indicate that faster deterioration was associated with the lowest absolute Hb concentration, while greater degree of MCA brain sparing and DV index elevation was associated with low platelet and high NRBC counts, respectively.
To evaluate hematologic and cardiovascular interrelationships further, we correlated platelet count and UA-PI according to the three rates of deterioration. Here, a significant inverse correlation with platelet count was only observed after moderate progression (Spearman ρ = −0.439, P = 0.001, Figure 2a). Similarly, it was only in patients with moderate progression that loss of UA end-diastolic velocity was associated with lower platelet count (Figure 2b). In contrast platelet and NRBC count correlation coefficients were higher with faster progression (Spearman ρ slow progression, –0.219; moderate progression, –0.43; rapid progression, –0.51; all P < 0.05) and NRBC count elevation > 35 cells/100 WBC was associated with lower platelet count in patients with moderate and rapid progression (Figure 3). Automated mean platelet volume, available for 69 patients, was highest after moderate progression, while the range of distribution was greatest following rapid progression (Figure 4).
According to conventional theory, progressive microvascular platelet aggregation and terminal villous vascular occlusion lead to rising UA blood-flow resistance in FGR[1-4]. The interaction of these factors could also determine cardiovascular deterioration preceding delivery. By incorporating longitudinal observations, our findings add new information to cross-sectional studies[16-18]. We demonstrate that Hb concentration and platelet count are lower while NRBC count is higher with abnormal pH and more rapid and extensive cardiovascular deterioration. The reduced platelet count is the main hematologic factor associated with accelerated cardiovascular progression to abnormal venous Doppler. When this progression is moderate, taking more than 1 month, low platelet count is primarily associated with escalating UA blood-flow resistance. However, with rapid progression the low platelet count is no longer explained by placental blood-flow resistance but corresponds to the onset of accelerated NRBC release. Concurrently there is a low Hb level and failure to produce new large platelets.
It has been demonstrated that consumption of activated platelets in the placenta increases the risk for fetal thrombocytopenia, especially when there is temporary stasis or reversed umbilical blood flow during end-diastole[11, 19-21]. However, similarly to the findings of Martinelli et al., consideration of additional vascular beds indicates that platelet count is more strongly related to MCA brain sparing, a circulatory finding that correlates with hypoxemia independently of UA end-diastolic velocity[22, 23]. When the rate of clinical evolution is also taken into account, it is apparent that the relationship between platelet count and UA blood-flow resistance is most pronounced with moderate deterioration. With more rapid deterioration, platelet count no longer relates to placental blood-flow resistance, as the increasing NRBC count emerges as the primary determinant. This finding requires examination of the relationship between red blood cell and platelet production in the fetus.
Both red blood cells and platelets originate from early myeloid CD 34+ progenitors after differentiation into erythroid and megakaryocyte cell lines under the influence of erythropoietin (EPO), and thrombopoietin (TPO)[24, 25]. The primary stimulus for EPO production is hypoxia, while TPO activity increases when circulating platelet numbers fall[24-26]. During normal fetal development, hematopoiesis moves from extramedullary sites such as the liver and spleen to the bone marrow by the beginning of the third trimester, while production of platelets continues in the bone marrow and blood[24, 25, 27]. Enhanced stimulation of hematopoiesis due to fetal hypoxemia first leads to compensatory increase in red blood cells. However, persistence of hypoxemia reactivates extramedullary hematopoiesis and it is from these sites with sufficiently sized capillary fenestrations that large NRBCs can readily enter the circulation in response to worsening acid–base status[24, 25, 29, 30]. In contrast to NRBCs, large new platelets can enter the circulation freely, and accordingly increased platelet production in response to TPO results in elevation of the mean circulating platelet volume (as seen in the group with moderately progressive disease). Because of their shared progenitors, red blood cell and platelet numbers are inversely related, which is most notable in the presence of significant extramedullary hematopoiesis and elevated NRBC count[14, 24, 25, 32, 33]. This competition for progenitors can become so exaggerated that growth-restricted neonates with thrombocytopenia may temporarily lose their ability to promote megakaryocyte precursors despite increased TPO levels[25, 29, 33]. Accordingly, placental dysfunction produces two distinct mechanisms for thrombocytopenia – increased platelet consumption and decreased platelet production. While the risk for consumption is related to placental blood-flow resistance, extramedullary hematopoiesis in the context of chronic hypoxemia appears to be the risk factor for decreased production.
It is therefore plausible that placental platelet activation sets the stage for progressive villous occlusion and accelerated loss of UA end-diastolic velocity. At this time increased platelet demand can be met through the production of large immature platelets. However, with persistent chronic hypoxemia, the ability to generate new platelets becomes limited, as extramedullary production of red cell precursors escalates. Accordingly, the mean platelet volume, a reflection of the number of large immature platelets, falls. In patients with rapid clinical deterioration, dysfunctional thrombopoiesis and erythropoiesis can explain the complex hematologic profile at birth[24, 25, 33]. The secondary role of placental platelet activation as a contributor to clinical deterioration also explains why fetal antiplatelet therapy is ineffective in delaying progression if initiated after UA end-diastolic velocity is lost[34-40].
This is a secondary analysis of a study that was not designed to evaluate hematologic impacts of placental dysfunction. Our sample size was defined by the availability of Doppler and hematologic information, and the population was skewed towards severe placental dysfunction. We have no information on the balance of hematopoietic regulatory factors (EPO/TPO). The progression categories used here had not been described at the time of the original study. However, an analysis of progression rates using continuous variables yielded the same results and accordingly is not presented here. Finally, we are relating longitudinal cardiovascular data to a single hematologic profile at birth, as we did not have blood samples during evolving compromise, forcing us to hypothesize about the preceding evolution of hematologic abnormalities. In spite of these apparent limitations our longitudinal analysis of Doppler examinations raises questions about the fetal vascular and hematologic responses that have implications for basic and clinical research. Our findings stress the importance of concurrent evaluation of circulation, metabolic state, hematologic profile and expression of regulatory substances to clarify the mechanisms that operate as placental function worsens. A reappraisal of the role of fetal antiplatelet therapy may be warranted, recognizing the potential importance of the timing of initiation of therapy.
In summary, we suggest that placental platelet consumption and progressive villous vascular occlusion are associated with accelerated cardiovascular deterioration in FGR. A subsequent increase in hypoxemia-stimulated extramedullary erythropoiesis of red blood cells increases competition for platelet precursors, limiting the generation of new platelets. Rapid cardiovascular deterioration culminates in dysfunctional erythropoiesis with absence of compensatory polycythemia and widespread release of NRBCs into the circulation.