BACKGROUND: Thrombocytopenia is common among small-for-gestational-age (SGA) neonates (birth weight <10th percentile reference range), but several aspects of this thrombocytopenia are unclear, including the incidence, typical nadir, duration, association with preeclampsia, mechanism, and risk of death.
METHODS: Using 9 years of multihospital records, we studied SGA neonates with ≥2 platelet counts <150 000/μL in their first week.
RESULTS: We found first-week thrombocytopenia in 31% (905 of 2891) of SGA neonates versus 10% of non-SGA matched controls (P < .0001). Of the 905, 102 had a recognized cause of thrombocytopenia (disseminated intravascular coagulation, early-onset sepsis, or extracorporeal membrane oxygenation). This group had a 65% mortality rate. The remaining 803 did not have an obvious cause for their thrombocytopenia, and we called this “thrombocytopenia of SGA.” They had a mortality rate of 2% (P < .0001) and a mean nadir count on day 4 of 93 000/μL (SD 51 580/μL, 10th percentile 50 000/μL, 90th percentile 175 000/μL). By day 14, platelet counts were ≥150 000/μL in more than half of the patients. Severely SGA neonates (<1st percentile) had lower counts and longer thrombocytopenia duration (P < .001). High nucleated red cell counts at birth correlated with low platelets (P < .0001). Platelet transfusions were given to 23%, and counts typically more than tripled. Thrombocytopenia was more associated with SGA status than with the diagnosis of maternal preeclampsia.
CONCLUSIONS: SGA neonates with clearly recognized varieties of thrombocytopenia have a high mortality rate. In contrast, thrombocytopenia of SGA is a hyporegenerative condition of moderate severity and 2 weeks’ duration and is associated with evidence of intrauterine hypoxia and a low mortality rate.
- CMV —
- DIC —
- disseminated intravascular coagulation
- ECMO —
- extracorporeal membrane oxygenation
- MPV —
- mean platelet volume
- NRBC —
- nucleated red blood cell count
- SGA —
- small for gestational age
What’s Known on This Subject:
Small-for-gestational-age neonates are at risk for thrombocytopenia during the first days and weeks after birth. However, the incidence, duration, severity, responsible mechanism, value of platelet transfusions, and risk of death from this variety of neonatal thrombocytopenia are unknown.
What This Study Adds:
Ten percent of thrombocytopenic small-for-gestational-age neonates have a recognized cause for low platelets (aneuploidy, extracorporeal membrane oxygenation, disseminated intravascular coagulation); they have a high mortality rate (65%). Ninety percent have a moderate, transient (2 weeks), hyporegenerative thrombocytopenia with a low mortality rate (2%).
Neonates who are born small for gestational age (SGA) (<10th percentile birth weight for a reference population) are at risk for having thrombocytopenia in the first weeks after birth.1–6 Some of these neonates have varieties of thrombocytopenia that are readily recognized, such as the consumptive thrombocytopenias accompanying disseminated intravascular coagulation (DIC), extracorporeal membrane oxygenation (ECMO), or early-onset sepsis or the hyporegenerative thrombocytopenias associated with marrow-failure syndromes. Other SGA neonates have a type of thrombocytopenia with no obvious explanation, a variety sometimes termed by exclusion “thrombocytopenia of SGA.”1–6 That variety has been the subject of previous reports, but many aspects remain unclear. As a step toward enhancing the knowledge base of the thrombocytopenia of SGA, we conducted an analysis of all SGA infants born in our health care system during a 9-year period. The aims were (1) to identify a group of thrombocytopenic SGA neonates in whom the thrombocytopenia was not a readily apparent variety; and (2) in that group (termed thrombocytopenia of SGA), to identify the incidence, nadir, severity, and duration of the thrombocytopenia and whether it was more closely associated with preeclampsia versus SGA status, assess the responsible mechanisms, and describe the outcomes.
Data were collected retrospectively as deidentified limited data sets from archived Intermountain Healthcare records. Intermountain Healthcare is a not-for-profit healthcare system that owns and operates 19 hospitals with labor and delivery units in Utah and Idaho. The information collected was limited to the information in this report. Patient records were accessed if the neonate had a date of birth from January 1, 2004, to December 31, 2013. The Intermountain Healthcare Institutional Review Board approved this as a deidentified data–only study as not requiring the consent of individual subjects.
Blood Cell Counts and Platelet Transfusions
Platelet counts and mean platelet volumes (MPVs) were determined in all hospitals with the Beckman Coulter LH750 Hematology Analyzer (Fullerton, CA) from 2004 to mid-2012. After mid-2012, platelet counts and MPVs were determined by using Sysmex counters (Sysmex America, Lincolnshire, IL). All blood tests were performed in accordance with Intermountain Healthcare Laboratory Services standard operating procedures. Nucleated red blood cells were quantified using either automated cell counts, performed in accordance with the hematology analyzer manufacturer’s instructions, or manual enumeration, performed by certified medical technologists on Wright-stained blood smears, counting a minimum of 100 nucleated cells per test. The reference intervals for complete blood count parameters are those we previously published from Intermountain Healthcare databases.7
Guidelines for administering platelet transfusions in the Intermountain Healthcare NICUs during this period have been published.8,9 All platelet transfusions were type specific or AB (Rh positive or negative) and were derived from apheresis. They were not pooled or volume-reduced, but all were subjected to leukofiltration and irradiation and administered in a volume of 10 to 15 mL/kg body weight. Gestational age was determined by obstetrical assignment unless changed by the neonatologist on the basis of gestational age assessment (physical examination and neurologic-neurodevelopmental findings).
Neonates were classified as SGA if their weight at birth was <10th percentile for gestational age, using normative values from our Intermountain Healthcare population.10 Severity of SGA was classified according to 3 categories: <1st percentile (severe), first to fifth percentile (moderate), and sixth to 10th percentile (mild). Preeclampsia was identified from case-mix records using the following definitions from the International Classification of Diseases, Ninth Revision, Clinical Modification (Ingenix Expert, Eden Prairie, MN): 6425, 6426, and 6427.
SGA neonates were matched 1:1 with neonates from the same hospitals born during the same period of time who were not SGA. Matching was performed on the basis of gestational age (within 1 week) and year/month of birth (within 1 month). Early thrombocytopenia was defined as ≥2 platelet counts <150 000/μL during the first week after birth. Using chart reviews and electronic reviews of laboratory data, patients whose data would otherwise qualify for inclusion in the SGA and thrombocytopenia group were excluded if they were recognized to have another variety of thrombocytopenia. These specifically included early-onset sepsis, congenital cytomegalovirus (CMV) or other congenital viral infection, ECMO treatment, immune-mediated thrombocytopenia, aneuploidy, varieties of severe congenital thrombocytopenia, or DIC. DIC was diagnosed by the combination of hypofibrinogenemia for age,11 schistocytosis,12 prolongation of prothrombin time and activated partial thromboplastin time for age,11 and elevated D-dimers.
Data Collection and Statistical Analysis
The program used for data collection was a modified subsystem of Clinical Workstation. The 3M Company (Minneapolis, MN) approved the structure and definitions of all data points for use within the program. Data were managed and accessed by authorized data analysts. Means and standard deviations were used to express values in groups that were normally distributed, and medians and ranges to express values in groups that were not. Differences in categorical variables were assessed by using the Fisher exact test or χ2. Student nonpaired t test was used to assess continuous variables. Statistical analysis used the R Foundation package (Statistical Computing, Vienna Austria). The mixed-effects model used NIME software, version 3.1-105, also from the R package. Statistical significance was set as P < .05.
Incidence, Severity, and Duration of Thrombocytopenia
During the 9-year period studied, 24 036 neonates were admitted to an Intermountain Healthcare NICU, and 3964 of these were SGA (Fig 1). Of the SGA infants, 2891 had ≥2 platelet counts measured in the first week, and 905 (31.5%) of these had ≥2 platelet counts <150 000/μL, thereby qualifying for the definition of early (first-week) thrombocytopenia. Thus the incidence of thrombocytopenia in our SGA neonates was 31.5%, which was higher than the 10.0% incidence of early thrombocytopenia among 2891 non-SGA control neonates matched for gestational age (P < .0001) (Fig 1). The 2891 non-SGA controls were well matched with the 2891 SGA neonates on the basis of gestational age (Table 1), but (as expected) the non-SGA control group had a higher birth weight; lower rates of cesarean delivery, maternal eclampsia or preeclampsia, and mortality; and proportionately more males compared with the SGA neonates.
Of the 905 thrombocytopenic SGA neonates, 102 had a condition (besides SGA) that might have caused their thrombocytopenia (Fig 1). These conditions included treatment with ECMO (n = 28), aneuploidy (n = 30, 3 of whom were also on ECMO), early-onset culture-positive bacterial sepsis (n = 6), congenital CMV (n = 6), congenital marrow failure syndromes (n = 4), alloimmune thrombocytopenia (n = 2), DIC (n = 8), and various malformation syndromes (n = 18). These 102 neonates were excluded from further analysis, leaving 803 with a condition we termed thrombocytopenia of SGA (Table 2).
Platelet counts in these 803 are shown in Fig 2. The reference interval for platelet counts of neonates during their first 3 weeks (5th to 95th percentile limits), which we published previously,13 is shown by the shaded area for comparison. The lowest platelet counts were typically on day 4, with a mean nadir of 93 000/μL (SD 51 580/μL, 10th percentile 50 000/μL, 90th percentile 175 000/μL). By day 14 of life, the platelet count had increased to ≥150 000/μL in half of the infants and was ≥100 000/μL in 70%. By day 21, the count was ≥150 000/μL in about two-thirds (Fig 2). The most severely SGA neonates (birth weight <1st percentile reference range) had lower platelet counts than those with moderate (P = .013) or mild (P = .005) SGA (Fig 3). The severe SGA group also had a longer duration of thrombocytopenia, with 50% having platelet counts <150 000/μL by 28 to 30 days.
SGA and Preeclampsia
Associations of thrombocytopenia with SGA status versus maternal preeclampsia were sought by comparing the lowest platelet count during the first 4 days of life among 4 groups of NICU patients matched for gestational age (within 1 week) (Fig 4). Neonates who were SGA but no maternal preeclampsia was diagnosed (Group 1) had lower platelet counts (mean 153 000/μL) than did those born to women with preeclampsia but were not SGA (Group 2) (197 000μL, P < .0001, 95% confidence interval on the difference 22 to 64). Using linear regression to account for gestational and birth weight, the point estimate of the difference in the 2 groups was still significant (P = .0189, 95% confidence interval on the difference 10 to 61). Among SGA neonates, the presence or absence of preeclampsia made no difference in the platelet count. Thus, preeclampsia may not be associated with a risk of neonatal thrombocytopenia over and above the risk associated with SGA status.
Nucleated Red Blood Cell Count, MPV, and Response to Platelet Transfusion
Among the 803 neonates we labeled as having thrombocytopenia of SGA, an elevated nucleated red blood cell count (NRBC, per microliter) at birth correlated with a low platelet count in the first week after birth (Fig 5) (P < .01). MPV measurements accompanying the platelet counts of the 803 thrombocytopenic neonates over their first week after birth are displayed in Fig 6. Overlying the values in the shaded area is the MPV reference interval (5th to 95th percentile limits) that we published previously.13
Of the 803 neonates with thrombocytopenia of SGA, 182 (23%) received 1 to 33 apheresis platelet transfusions. Reviews indicated that 98% of the transfusions were given prophylactically, meaning the patient did not have active bleeding. Platelet counts before the transfusions ranged from 6000 to 110 000/μL; median count was 49 000/μL. The rise in platelet count after transfusion, calculated by subtracting the pretransfusion count from the count performed 1 to 24 hours after transfusion, was 118 990 ± 59 736/μL. Figure 7 shows the increase and subsequent decrease in platelet counts after platelet transfusion. For ease of comparison, all platelet counts were normalized to percentage of the pretransfusion platelet count.
Ten SGA neonates had severe thrombocytopenia (<50 000/μL) that persisted for at least 4 weeks (Table 2). Nine of the 10 were severely SGA (<1st percentile at birth). Four of these had no apparent explanation, other than the SGA status, for persistent severe thrombocytopenia. The other 6 had developed late-onset thrombocytopenia accompanying necrotizing enterocolitis or sepsis. Using data from the first days after birth, focusing on birth weight, gestational age, or platelet counts in the first 4 days, we were not able to recognize these 6 as distinct from the other 2885 in the group of 2891. Mean platelet counts of these 10 during their first 3 weeks (grouped as the 4 with no explanation for prolonged thrombocytopenia and the 6 with NEC or late-onset sepsis) are shown in Fig 2 along with the entire group of 803 with thrombocytopenia of SGA (box and whiskers) and the normal controls (shaded area). This group of 10 with persistent severe thrombocytopenia received 4 to 33 platelet transfusions. All of these were prophylactic, for platelet counts <50 000 to 60 000/μL, with no signs of bleeding.
Of the 1986 SGA neonates in whom thrombocytopenia was not identified, 34 (2%) died. In contrast, of the 905 SGA neonates in whom thrombocytopenia was identified, 85 (9%) died (P < .0001) (data shown in Supplementary Table 3). Of the 102 SGA neonates with known varieties of thrombocytopenia (eg, ECMO or DIC), 66 (65%) died. However, of the 803 SGA neonates with thrombocytopenia of SGA, 19 (2%) died (P < .0001 vs. SGA neonates with known varieties of thrombocytopenia). All thrombocytopenic SGA neonates with DIC who died had bleeding problems at the time of death, predominantly pulmonary hemorrhage. None of those with trisomy 18 or 13 who died had bleeding problems (shown in Supplementary Table 3).
Logistic regression was performed to identify clinical variables predictive of severe (<50 000/μL) and prolonged (≥2 weeks) thrombocytopenia, which we judged could be relevant to future testing of thrombopoietin receptor agonists such as romiplostim.14 This medication takes 7 to 10 days to elevate the platelet count; therefore, neonates recognized within the first days as being very likely to remain severely thrombocytopenic for ≥2 weeks might be candidates for experimental treatment. Three variables had predictive correlations: (1) gestational age at birth (shorter gestation was associated with higher likelihood of severe, persistent thrombocytopenia); (2) gender (males were more likely to have severe, persistent thrombocytopenia); and (3) lowest platelet count in the first week (generally on day 4). However, the models were insufficiently predictive for individual patients, and no variable had an odds ratio >1.45 predicting severe and prolonged thrombocytopenia.
Thrombocytopenia is a common problem among patients in NICUs.1,2,5,6,14–18 The majority have acquired varieties of consumptive thrombocytopenia accompanying bacterial or fungal sepsis or necrotizing enterocolitis.1,2,5,6 Rare varieties of hyporegenerative thrombocytopenia accompany aneuploidy or syndromes involving thrombopoietic failure.1–3 A separate variety of early thrombocytopenia has been described among neonates who are SGA. The first report of this entity was by Meberg et al from Oslo in 1977.19 They studied 23 neonates weighing <10th percentile who, with no other explanation, had ≥1 platelet count <100 000/μL in the first days after birth. The platelet counts generally increased to >150 000/μL by day 15 of life, and none had pathologic bleeding. The authors speculated that this thrombocytopenia was the result of chronic hypoxia in utero induced by placental insufficiency. In 1983, Shuper et al from Israel reported 14 SGA infants with ≥1 platelet counts <100 000/μL and no other explanation for the thrombocytopenia.20 Elevated NRBCs at birth were common in the thrombocytopenic neonates; thus they postulated, as had Meberg et al, that the condition was due to reduced platelet production associated with chronic intrauterine hypoxia. Our present study used a large multihospital dataset to better define this condition, which has come to be termed thrombocytopenia of SGA1–6.
We began by identifying all patients in the last 9 years admitted to an Intermountain Healthcare NICU with a birth weight <10th percentile.10 We then determined how many of these SGA neonates had ≥2 platelet counts drawn during their first week, and how many of those had ≥2 platelet counts <150 000/μL. A low platelet count in a neonate can be real or artifactual; the latter typically involves platelet clumping, either on the wound when blood is drawn slowly from a capillary puncture or within the specimen tube.21 We aimed to reduce the contamination of our dataset with artifactual thrombocytopenia by requiring 2 low counts.
After identifying all SGA neonates who had early (first-week) thrombocytopenia, we sought to cull those associated with a recognized cause of thrombocytopenia. This was done so that we could tentatively classify the others as having the exclusionary diagnosis of thrombocytopenia of SGA. We found that ∼10% of thrombocytopenic SGA neonates had a readily identifiable variety of thrombocytopenia, and that group had a high mortality rate (65%). In contrast, 90% had a low-mortality variety with the following characteristics. (1) The thrombocytopenia appears to be primarily hyporegenerative as opposed to accelerated platelet consumption or destruction. We made this judgment on the basis of normal MPVs and good responses to platelet transfusions. (2) The low platelet counts are associated with intrauterine hypoxemia (elevated NRBC) and with the pathology that creates fetal growth restriction, not the pathology that produces preeclampsia without fetal growth restriction. (3) The thrombocytopenia is typically only moderately severe (mean nadir count of 93 000/μL) with a typical nadir on about day 4 and a typical duration (days to a count >150 000/μL) of 2 weeks. However, those with severe SGA tend to have lower counts and a longer duration.
Cremer et al also suggested that thrombocytopenia of SGA is the kinetic result of reduced platelet production, by measuring the immature platelet fraction.22 They found a trend toward lower immature fractions (suggesting reduced platelet production) in thrombocytopenic SGA neonates compared with thrombocytopenic neonates who had infection, in whom the mechanism is likely accelerated platelet utilization.
Reduced platelet production was also the mechanism suggested by the work of Murray et al, who observed that preterm infants with thrombocytopenia had fewer circulating megakaryocyte progenitors than nonthrombocytopenic controls.23 Likewise, we reported few megakaryocytes in the marrow of 3 thrombocytopenic SGA neonates.24 Both Murray et al23 and Sola et al24 reported low plasma thrombopoietin concentrations, suggesting inadequate upregulation of thrombopoietin production. The hypothesis that the thrombocytopenia of SGA is primarily due to thrombopoietin deficiency in utero is consistent with these studies but remains to be confirmed.
The molecular mechanisms resulting in thrombocytopenia of SGA may be similar to that causing thrombocytopenia of perinatal asphyxia.25 We suspect that fetal hypoxia is causally involved in both varieties, which is consistent with the study of McDonald et al,26 in adult mice subjected to hypoxia, and the marrow culture studies of Saxonhouse et al.27,28
If the thrombocytopenia of SGA is determined to be the result of thrombopoietin deficiency, it will remain unclear whether thrombopoietin receptor agonists such as romiplostim or eltrombopag will be of value in treating this condition.14,29 Most neonates with the thrombocytopenia of SGA are not severely thrombocytopenic; thus perhaps no treatment is warranted for most. Moreover, romiplostim and eltrombopag require 7 to 10 days between commencement of dosing and a significant increase in platelet count, and half of the neonates with this variety of thrombocytopenia have platelet counts >150 000/μL by day 14. Because a subset of patients remain severely thrombocytopenic for >14 days, and some for >28 days, the efficacy, risks, and benefits of administering thrombopoietin agonists to that group might warrant future study, if they could be identified in the first week. We found that these are typically males born at very early gestational age with very low counts initially. However, we were unable to accurately predict during the first week which infants would still have severe, persistent thrombocytopenia after 2 weeks.
One treatment option for neonates with thrombocytopenia of SGA is platelet transfusion, but as highlighted by Josephson et al, platelet transfusion strategies for neonates lack a strong evidence base.30 There is substantial disagreement on neonatal platelet transfusion thresholds, but it is widely accepted that platelet transfusions should be given to neonates if their platelet count falls below 20 000/μL.1,31,32 Counts that low should be rare in neonates with thrombocytopenia of SGA. From our present data, 90% of platelet count among neonates with this diagnosis will be >50 000/μL; thus perhaps 10% will be <50 000/μL. In our present analysis, 182 of the 803 neonates with thrombocytopenia of SGA received platelet transfusions. In retrospect, we question how many of those were beneficial. Very few had a platelet count <20 000/μL, most had a pretransfusion platelet count >50 000/μL, and almost all were transfused prophylactically with no bleeding problems prompting the transfusion.
We recognize several significant problems in our retrospective analysis. For instance, surely some relevant data were sometimes missing from the charts or electronic databases, and bias may have been unintentionally introduced. Also, platelet counts after platelet transfusions were not obtained on a standard schedule, and platelet transfusions were given to some and not to others without any clear explanation. Also, changes in obstetrical and neonatology practices occurred during the 9 years of study. The practice changes we recognize are the overall reduction in NICU transfusions during this period8 and the specific reduction in platelet transfusions on the basis of a change from platelet count–based guidelines to platelet mass–based guidelines.9 In addition, in our studies of associations with preeclampsia, we captured only those women coded as having HELLP (syndrome of hemolysis, elevated liver enzymes, and low platelets), eclampsia, or severe or moderately severe preeclampsia, not those with mild or unspecified preeclampsia; thus we might have ignored associations between thrombocytopenia and mild preeclampsia. Last, we did not identify the molecular mechanisms involved in the association between SGA status and thrombocytopenia. Clarifying this mechanism should be the topic of future investigations, as it might suggest novel preventive or treatment approaches.
Considering the weaknesses and strengths of our data, we maintain it is reasonable to conclude that about one-third of SGA infants have early thrombocytopenia. About 10% of these have an obvious cause of thrombocytopenia such as sepsis, aneuploidy, or ECMO treatment, and this group has a high mortality rate (65% in our dataset). However, 90% of thrombocytopenic SGA neonates have a different variety that could be termed thrombocytopenia of SGA. Those neonates should have a low mortality rate, yet some will probably receive multiple platelet transfusions.
- Address correspondence to Robert D. Christensen, MD, University of Utah Department of Pediatrics, 295 Chipeta Way, Salt Lake City, UT 84108. E-mail:
Drs Christensen and Sola-Visner and Ms Baer conceptualized and designed the study; Ms Baer carried out the initial analyses; Dr Christensen drafted the initial manuscript; Dr Snow and Ms Butler carried out the statistical analysis; Mr Henry carried out the statistical display; Ms Baer, Mr Henry, Drs Snow and Sola-Visner, and Ms Butler reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
Accepted for publication May 7, 2015
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This work was partially supported by US National Institutes of Health grant PO1HL046925 (Dr Sola-Visner).
POTENTIAL CONFLICT OF INTEREST: Dr Sola-Visner presented a webinar on evaluation of neonatal thrombocytopenia and was invited speaker at the Sysmex New England Users Group Meeting, discussing challenges in the evaluation and management of neonatal anemia and thrombocytopenia. The other authors have indicated they have no potential conflicts of interest to disclose.
- ↵Sola-Visner MC, Saxonhouse MA. Acquired thrombocytopenia. In: deAlarcon PA, Werner EJ, Christensen RD, eds. Neonatal Hematology, 2nd ed. Cambridge, UK: Cambridge University Press; 2013:162–164
- Baer VL,
- Henry E,
- Lambert DK,
- et al
- Muthukumar P,
- Venkatesh V,
- Curley A,
- et al.,
- Platelets Neonatal Transfusion Study Group
- McDonald TP,
- Cottrell MB,
- Clift RE,
- Jackson CW
- ↵Sola-Visner M. Platelets in the neonatal period: developmental differences in platelet production, function, and hemostasis and the potential impact of therapies. Hematol Am Soc Hematol Educ Program 2012;2012:506–511
- Josephson CD,
- Glynn SA,
- Kleinman SH,
- Blajchman MA,
- State-of-the Science Symposium Transfusion Medicine Committee
- Copyright © 2015 by the American Academy of Pediatrics