Available online at www.annclinlabsci.org 10 Annals of Clinical & Laboratory Science, vol. 44, no. 1, 2014

Erythroid Nuclear Irregularities: a Review of Fetal and Neonatal Autopsies and Correlation with Clinical Features Karen M. Chisholm and Amy Heerema-McKenney*

Department of Pathology, Stanford University School of Medicine, Stanford, CA, USA.

Abstract. In our autopsy experience, abnormal erythroblast nuclear contours are frequently observed in the stillborn fetus or neonate without marrow failure disorders. Bone marrow and liver slides from autopsies of fetuses and infants less than six months of age were analyzed for the percent erythroblasts with nuclear irregularities, and correlated with gestational age at birth, days of life, cause of death, postmortem interval, presence of hydrops, and at death. In total, 77 cases had sufficient marrow or liver erythroblasts for review, including 37 stillborns and 40 liveborns. Erythroid nuclear irregularities in >10% of erythroid precursors were present in either the liver or marrow in 54% of stillborns and 68% of liveborns, more com- monly seen in the liver. Cases with <1% abnormal erythroblasts were rare. Fetuses with >10% abnormal erythroblasts in the liver were more likely to have died in utero, whereas those with less were more com- monly terminations (p=0.008). No significant association between the extent of abnormal erythroblasts and the presence of or hydrops was observed. While the finding of erythroblasts with nuclear irregulari- ties is common in stillborns and liveborns and could be solely a postmortem artifact, we cannot exclude a potential fetal erythropoietic response to hypoxic stimuli. Dyspoietic-appearing erythroblasts alone should not be used as the basis for the diagnosis of a marrow failure disorder at autopsy.

Key words: Autopsy, dyserythropoiesis, erythroid precursors, fetus, neonate, and nuclear irregularities.

Introduction neonates may have a history of non-reassuring fetal status, abnormal umbilical artery Doppler studies, Nucleated red blood cells are a common finding in and decreased umbilical artery pH [9-11]. the peripheral blood of a neonate. Many reports Nucleated red blood cells are increased due to post- have detailed findings of increased circulating nu- natal stressors as well, including hypoxia due to cleated red blood cells in newborns experiencing pulmonary and heart disease [12]. Tissue hypoxia is acute, subacute, and chronic asphyxia. Neonates thought to stimulate erythropoietin expression, in with intrauterine growth restriction, pre-eclampsia, turn increasing erythropoiesis with the presence of maternal hypertension, maternal smoking, mater- precursor cells in the circulation [13-15]. nal diabetes, and congenital infections including Erythropoietin levels are elevated in neonates with toxoplasmosis, syphilis, rubella, cytomegalovirus, growth retardation and hypoxia [10,16,17]. The and parvovirus have been found to have increased erythropoietin stimulus also augments extramedul- red blood cell production and circulating nucleated lary hematopoiesis in the neonatal liver. red blood cells [1-4]. Acute and subacute hypoxia, which can be due to events during labor [5-7] and The regular, round erythroblast nucleus may appear [8], have also been found to cause irregular in certain conditions. In dyserythropoie- an increase in peripheral blood nucleated red blood sis, the nucleus can display irregularities including cells; the closer the event to birth, the lower the budding, notching, internuclear bridging, karyor- number of nucleated red blood cells [5]. These rhexis, multinuclearity, nuclear hyperlobation, or megaloblastoid changes. The cytoplasm can have *Current affiliation: Cleveland Clinic Pathology and Laboratory abnormal accumulation of iron in mitochondria Medicine Institute. Address correspondence to Amy Heerema- McKenney, MD, Pathology and Laboratory Medicine Institute, (ringed sideroblasts) or vacuolization. Cleveland Clinic, 9500 Euclid Avenue, L21, Cleveland, OH 44195; phone: 216 444 5777; e mail: [email protected] Dyserythropoiesis is a common morphologic

0091-7370/14/0100-010. © 2014 by the Association of Clinical Scientists, Inc. Erythroid nuclear irregularities 11

Table 1. Fetal (Stillborn) and Neonatal (Liveborn) Characteristics.

Stillborn Liveborn

Subjects 37 40 Gestational weeks of age at birth, average (range) 25.3 (15.0-39.0) 33.9 (23.0-40.7) Weeks lived, average (range) N/A 4.6 (0.1-19.6) Adjusted gestational age for weeks lived, average (range) N/A 37.1 (23.0-50.0) Abnormal karyotype, number (percent) 3 / 30 (10%) 4 / 22 (18%) Hydrops present 3 3 Hematocrit at death N/A 38.9 (9.7-49.6) due to termination 16 (43.3%) N/A Stillbirth due to intrauterine fetal demise 14 (37.8%) N/A Stillbirth due to preterm labor 7 (18.9) N/A Cause of death due to congenital anomalies 15 (40.5%) 29 (72.5%) Cause of death due to 17 (46.0%) 5 (12.5%) Cause of death due to hydrops 3 (8.1%) 0 Cause of death due to other reasons 2 (5.4%) 6 (15%) Postmortem interval in hours, average (range) 41.1 (3-315) 37.1 (1-139) Bone marrow evaluable for dyserythropoiesis 19 27 Liver evaluable for dyserythropoiesis 34 28 change seen in low grade and high grade myelodys- Materials and Methods plastic syndromes of adults and children [18]. In neonates and infants, dyserythropoiesis is found in With IRB exemption, the Stanford pathology database congenital dyserythropoietic (CDA). CDA was queried for autopsies of fetuses and infants less than type II typically has binucleate erythroblasts or six months of age between July 2008 and October 2011. erythroblasts with cloverleaf morphology in the Bone marrow and liver slides from all identified cases were analyzed for the percentage of erythroblasts with bone marrow. CDA type I generally demonstrates nuclear irregularities out of total erythroblasts on H&E internuclear chromatin bridges together with tri- stained slides. The percentage of this form of dyserythro- and tetra-nucleate cells. The rarer CDA type III has poiesis was graded as <1%, 1-10%, or >10% by both abundant giant erythroid precursors with frequent authors. Slides with poor cellular preservation due to au- multinucleation (reviewed in [19]). tolysis or insufficient representation of erythropoiesis were excluded from review. After slide review, autopsy Dyserythropoiesis is described in case reports of hy- reports were reviewed for gestational age at birth, days of dropic liveborns and stillborns with erythroblasto- life, cause of death, postmortem interval, presence of hy- sis [20-24]. In many of these cases, the findings do drops, and hematocrit at death. Concurrent cytogenetic not neatly fit into one of the three classic categories reports were also reviewed to identify abnormal karyo- types. Individual variables were analyzed using two- of CDA. Clearly, some cases of are tailed Fisher exact tests, Student’s t-Test, and chi-square caused by severe anemia due at least in part to tests. CDA, but how specific are abnormally shaped erythroblast nuclei for a marrow failure disorder at Glycophorin C immunohistochemical staining was per- autopsy? In our experience, erythroblasts with nu- formed on select liver and bone marrow cases to confirm clear irregularities, including blebbing, are a com- that these nuclear changes are indeed present in ery- mon finding in the fetus and neonate at autopsy. throid precursors and are not an apoptotic change of This observation is not well documented in the lit- myeloid cells. Serial sections of 4 µm were cut from cho- erature. We report the frequency of erythroblasts sen formalin-fixed paraffin embedded tissue blocks and with nuclear irregularities in autopsy liver and bone used for immunohistochemical analysis. The glycopho- rin C antibody used was the DAKO monoclonal mouse marrow sections and describe the clinical associa- anti-human clone Ret40f (Carpinteria, CA) at a 1:100 tions of this finding. dilution. The staining was performed on the Ventana 12 Annals of Clinical & Laboratory Science, vol. 44, no. 1, 2014

BenchMark XT instrument (Tucson, AZ), using a heat- was identified in 3 of the 37 infants: one case of induced epitope retrieval with Ventana solution CC1 Parvovirus B19 infection, another with a hyper- (cell conditioning 1) and EDTA retrieval solution (pH coiled , severe fetal thrombotic vas- 8.0). The immunohistochemical reactions were visual- culopathy and fetal-maternal hemorrhage, and the ized using Ventana’s iView DAB Detection Kit with di- third with placental mesenchymal dysplasia. Sixteen aminobenzidine (DAB) as the chromogen. of the 37 (43.3%) stillborn neonates were early terminations, 14 of 37 (37.8%) died in Results utero (IUFD), and the remaining seven were too premature to survive preterm labor. Most of the Evaluable Bone Marrow and Liver. Between July stillborns had congenital anomalies (15/37, 40.5%) 2008 and October 2011, 52 stillborn fetuses and or placental causes of death (17/37, 46.0%). 46 liveborn infants under the age of six months had Congenital anomalies included hydrocephalus, autopsies performed with microscopic examina- ventriculomegaly, agenesis of the corpus callosum, tions. Of the stillborn autopsies, 41 had bone mar- diaphragmatic hernia, and Tetralogy of Fallot. row slides and all 52 had liver slides available. Of Placental causes of death included severe amniotic the 41 stillborn bone marrow slides, 19 were suffi- fluid infection sequence (4), severe fetal thrombotic cient for morphologic evaluation of erythropoiesis vasculopathy (4), acute and/or chronic abruptions (22 were excluded for poor preservation or insuffi- (5), twin-twin transfusion syndrome (2), excessive cient amount of marrow). Of the 52 stillborn liver perivillous fibrin deposition (1), umbilical cord slides, 34 were evaluable while 18 were too auto- rupture or compromise (7), and combinations of lyzed for analysis. These 19 evaluable bone marrow the above. Most of the stillborns who died in utero and 34 evaluable liver slides represented 37 (IUFD) had placental causes of death (12 of 14). stillborns. The average postmortem interval for all stillborns was 41.1 hours, with a range of 3-315 hours, and a Of the 46 autopsies on liveborns, 36 had bone mar- median of 27 hours. Of these autopsies, 76% were row slides and 45 had liver slides available. Of the performed within 48 hours. 36 liveborn bone marrow slides, 27 had evaluable bone marrow for morphologic review of erythro- The 40 liveborns had an average gestational age at poiesis. Of the 45 stillborn liver slides, 28 were birth of 33.9 weeks. They lived for an average of 4.6 evaluable for morphologic review of erythropoiesis, weeks (range <1 day to 137 days), with an average whereas 17 had too little or no extramedullary he- adjusted gestational age of 37.1 weeks (range 23- matopoiesis (EMH). The earliest age of the live- 50). Cytogenetics was successfully performed in 22 borns with too little or no extramedullary hemato- of the infants, of which four infants had abnormal poiesis was 17 days of life, or 29.4 weeks gestational karyotypes: one with Trisomy 13, one with del(17p) age when adjusted for . All infants or Smith-Magenis syndrome, one with del22q11.2 more than 10 weeks of age, or 45.7 weeks gesta- or DiGeorge Syndrome, and one with del22q13. tional age when adjusted for preterm birth, had no Three of the liveborn infants were born with hy- EMH. The 27 evaluable bone marrow and 28 eval- drops: one due to dilated cardiomyopathy, the sec- uable liver slides represented 40 liveborns. ond due to severe extensive fetal thrombotic vascu- lopathy of the placenta, and the third due to Stillborn and Liveborn Infant Characteristics. unknown causes. The most common cause of death Characteristics of the stillborns and liveborns are was congenital anomalies (29/40 = 72.5%), includ- presented in Table 1. The 37 stillborn infants had ing 18 infants with complex congenital heart an average gestational age of 25.3 weeks at birth. anomalies (hypoplastic left heart, transposition of Cytogenetics in the form of karyotypes or compara- the great arteries, and Tetralogy of Fallot), and six tive genomic hybridization were successfully per- infants with severe lung disease (alveolar capillary formed on 30 of the stillborns, identifying three dysplasia, bronchopulmonary dysplasia, and acute with genomic abnormalities: one with Trisomy 18, respiratory distress syndrome). Placental diseases one with Trisomy 21, and one with del(4p) consis- including severe amniotic fluid infection sequence, tent with Wolf-Hirschhorn Syndrome. Hydrops Erythroid nuclear irregularities 13

Figure 1. Erythroid nuclear irregularities in bone marrow and liver. A) Liveborn with bone marrow dyserythropoiesis including nuclear contour irregularities and blebbing, estimated at >10% (60X magnification). Arrows highlight dyseryth- ropoietic erythroblasts. B) Glycophorin C stain of bone marrow in A demonstrating erythroid nuclear irregularities >10% (60X magnification). Arrows highlight dyserythropoietic erythroblasts. C) Liveborn with liver dyserythropoiesis predomi- nantly in the form of nuclear contour abnormalities, blebbing, and multinucleate cells at this focus measuring >90%; en- tire liver estimated at >10% (60X magnification). D) Glycophorin C stain of liver in C demonstrating erythroid nuclear irregularities >10% (60X magnification). extensive chronic villitis, fetal thrombotic vascu- Erythroid nuclear morphology. Erythroid nuclear lopathy, and acute and chronic abruptions, were as- irregularities in bone marrow and liver H&E sociated with death in 5 of the infants, all of whom stained sections were quantified as <1%, 1-10%, died within the first six days of life. The “other” and >10% of total erythroblasts. Figure 1 demon- causes of death included Harlequin-type ichthyosis, strates examples of erythroblasts with nuclear ir- mitochondrial disorder, junctional epidermolysis regularities in the bone marrow and the liver, shown bullosa, heterotaxy-asplenia syndrome with sepsis, in both H&E forms, and with a glycophorin C and necrotizing enterocolitis. In 33 of the liveborn stain highlighting erythroid cells. The predominant infants, were available at or near the erythroblast nuclear change was nuclear blebbing, time of death. In 17 of these 33 liveborn infants, with rarer instances of internuclear bridging and the hematocrit was within the normal range for age; multinuclearity. Glycophorin C staining was per- however, 15 infants had low hematocrits for age formed on selected cases confirming H&E percent- (the lowest being 9.7%), and one had a high hema- ages of erythroblasts with nuclear irregularities. tocrit for age. The average postmortem interval for Iron stains for ringed sideroblasts were not per- all liveborns was 37.1 hours, with a range of 1-139 formed. As listed in Table 2, erythroblasts with hours, and a median of 27.5 hours. A total of 73% nuclear irregularities were a common finding in of the autopsies were performed within the first 48 both liveborns and stillborns, in both the bone hours. marrow (BM) and the liver. In general, nuclear 14 Annals of Clinical & Laboratory Science, vol. 44, no. 1, 2014

Table 2. Erythroid nuclear irregularities present in Stillborns and Liveborns.

Bone Marrow Liver

N <1% 1-10% >10% N <1% 1-10% >10%

Stillborn 19 4 (21.1%) 10 (52.6%) 5 (26.3%) 34 3 (8.8%) 11 (32.4%) 20 (58.8%) Liveborn 27 3 (11.1%) 11 (40.8%) 13 (48.1%) 28 0 9 (32.1%) 19 (67.9%) irregularities in greater than 10% of erythroblasts Nuclear irregularities in greater than 10% of eryth- were more commonly seen in the liver (63% of liver roblasts were present in the liver (67.9%) or bone slides) than in the bone marrow (39% of BM marrow (48.1%) in a total of 27 of 40 (68%) live- slides). Nuclear irregularities in more than 10% of borns. Liveborns with >10% erythroblasts with erythroblasts were present in either the liver nuclear irregularities in the bone marrow almost (58.8%) or marrow (26.3%) in a total of 20 of 37 always had >10% erythroblasts with nuclear irregu- (54%) stillborns. All stillborns with >10% erythro- larities in the liver, except for seven cases in which blasts with nuclear irregularities in the marrow had there was no extramedullary hematopoiesis to eval- >10% erythroblasts with nuclear irregularities in uate for liver dyserythropoiesis and one case which the liver. The average postmortem intervals were had 1-10% erythroblasts with nuclear irregularities shorter in stillborns with >10% erythroblasts with in the liver. The average postmortem intervals were nuclear irregularities in bone marrow and liver sec- shorter in bone marrows of liveborns with >10% tions (26.4 hours and 28 hours, respectively), com- erythroblasts with nuclear irregularities compared pared to stillborns with <10% erythroblasts with to those with <10% erythroblasts with nuclear ir- nuclear irregularities in bone marrow and liver sec- regularities (30.8 hours versus 39.2 hours, p=0.455), tions (50.8 hours and 62.75 hours, respectively) while the average postmortem intervals were longer (p=0.167 and p=0.061, respectively). Stillborns in livers from liveborns with >10% erythroblasts who died in utero were more likely to have more with nuclear irregularities compared to those with extensive erythroid nuclear irregularities (>10% <10% erythroblasts with nuclear irregularities (45.6 compared to <10%, p=0.032). Although not statis- hours versus 30.2 hours, p=0.310). The number of tically significant, stillborns who were terminated days of life, gestational age, and adjusted gestation- tended to have less extensive erythroid nuclear ir- al age for days of life did not correlate with percent regularities (>10% compared to <10%, p=0.075). of erythroblasts with nuclear irregularities in either Comparing the two mechanisms of death with per- the bone marrow or liver. cent erythroblasts with nuclear irregularities, those fetuses who had >10% erythroblasts with nuclear Figure 2 charts the percent of erythroblasts with irregularities in the liver were more likely to have nuclear irregularities in stillborns and liveborns in died in utero (IUFD), whereas those with less eryth- the bone marrow (A) and liver (B). As demonstrat- roblasts with nuclear irregularities were more com- ed, the percent of erythroblasts with nuclear irregu- monly terminations (p=0.008). The average post- larities in the bone marrow was slightly higher in mortem intervals between the stillborns with IUFD liveborns than in stillborns, but to no significant and those who were terminated were not signifi- value. The majority of stillborns had 1-10% eryth- cantly different (29.6 hours versus 59.8 hours, p = roblasts with nuclear irregularities in the bone mar- 0.168). Though most cases of IUFD were due to row, whereas the majority of liveborns had >10% placental disease, overall, the cause of death (con- erythroblasts with nuclear irregularities in the bone genital anomalies versus placental disease) did not marrow. In the liver, the majority of both stillborns affect the percentage of erythroblasts with nuclear and liveborns had >10% erythroblasts with nuclear irregularities in the bone marrow or liver. irregularities. The percentage did not correlate with Erythroid nuclear irregularities 15

Figure 2. Percent erythroid nuclear irregularities per gestational age in the bone marrow (A) and liver (B). The gestational age in weeks at delivery for stillborns (diamonds) and the adjusted gestational age for days of life for liveborns (squares) are plotted with the percent erythroblasts with nuclear irregularities listed as <1%, 1-10%, and >10%. As shown, gestational age does not correlate with percent erythroblasts with nuclear irregularities in the bone marrow or liver. the gestational age at birth (in stillborns) or adjust- irregularities in the bone marrow did not correlate ed gestational age for days of life (in liveborns). with a low hematocrit (p=0.214); in fact, 67% of infants with low hematocrits had <10% erythro- Cases with less than 1% erythroblasts with nuclear blasts with nuclear irregularities in the bone mar- irregularities were rare. A total of six stillborns row. Similarly, more than 10% erythroblasts with (16%) demonstrated <1% erythroblasts with nu- nuclear irregularities in the liver did not correlate clear irregularities in the bone marrow and/or liver. with a low hematocrit (p=0.178). The average post- Five of these cases were delivered at 21 weeks gesta- mortem intervals between the liveborn infants with tional age or earlier. Three of these six cases were low hematocrits and those with normal or high he- terminations due to severe hydrocephalus, Wolf- matocrits were not significantly different (33.3 Hirschhorn Syndrome (del 4p), and hydrops with hours versus 31.8 hours, p=0.852). placental mesenchymal dysplasia, while the other three stillborns had causes of death listed as acute No significant association between the extent of placental abruption, with erythroid nuclear irregularities and hydrops was ob- excessive perivillous fibrin deposition, and Trisomy served. A total of 3 stillborns and 3 liveborns dem- 18. Only three liveborns (8%) had <1% erythro- onstrated hydrops. Of the hydropic stillborns, only blasts with nuclear irregularities, and all three were two had bone marrows which were satisfactory for identified in the bone marrow. These three live- evaluation: one bone marrow demonstrated <1% borns ranged in days of life from 17 to 137, though erythroblasts with nuclear irregularities while the they were all born at less than 30 weeks gestational other had >10%. This latter stillborn had a diagno- age. Their listed causes of death were necrotizing sis of Parvovirus B19 infection with erythroblasto- enterocolitis, hydrops with congenital anomalies sis fetalis. In the liver, one stillborn had <1% eryth- associated with Smith-Magenis syndrome, and end roblasts with nuclear irregularities, two stillborns stage liver disease with respiratory failure and dif- had 1-10%, and one had >10% erythroblasts with fuse intra-alveolar hemorrhage. nuclear irregularities (the case of Parvovirus). In liveborns, the three hydropic infants had <1%, Of the 33 liveborn infants with available hemato- 1-10%, and >10% erythroblasts with nuclear ir- crits at or near the time of death, 17 were within the regularities in the bone marrow. In the liver, one normal range for age, 15 were low, and one was infant had 1-10% erythroblasts with nuclear irreg- high. More than 10% erythroblasts with nuclear ularities while the second had >10% erythroblasts 16 Annals of Clinical & Laboratory Science, vol. 44, no. 1, 2014 with nuclear irregularities; the third infant had too with congenital heart defects and 6 infants with little extramedullary hematopoiesis for evaluation. pulmonary defects. These heart and pulmonary dis- The average postmortem intervals between the hy- eases would cause severe postnatal hypoxia, causing dropic stillborns and liveborns and those who were an increase in nucleated red blood cells [12], and not hydropic were not significantly different (32.7 likely “stress” dyserythropoiesis. hours versus 39.6 hours, p=0.705). The presence of “stress” dyserythropoiesis has long In both stillborns and liveborns, the extent of bone been recognized by hematopathologists in the set- marrow and liver erythroid nuclear irregularity did ting of increased marrow turnover, such as a re- not correlate with each other. Of evaluable liver and sponse to severe hemolytic anemia or hemorrhage. bone marrow pairs in stillborns, 8/16 had equal lev- Why the increased production should be associated els and 7/16 had more erythroblasts with nuclear with structural nuclear changes is unclear. irregularities in the liver than in the bone marrow. Interestingly, increased hemoglobin F levels (fetal In liveborns, 9/15 had equal levels and 5/15 had hemoglobin) have been associated with dyserythro- more erythroblasts with nuclear irregularities in the poiesis in numerous settings, including after severe liver than in the bone marrow. Of all 31 pairs, only hemorrhage, in myelodysplastic syndrome, and in two had fewer erythroblasts with nuclear irregulari- hemophagocytic lymphohistiocytosis [26,27]. A ties in the liver than in the bone marrow: a stillborn “fetal” gene program of erythropoiesis appears to be terminated for Wolf-Hirschhorn Syndrome (del associated with more nuclear contour irregularities 4p), and a liveborn with hydrops and severe exten- than the more orderly state of erythropoiesis in the sive chronic villitis of the placenta. unstressed post-natal mature marrow [5,26-28]. The relationship of CDA types I and II, with their Discussion recognized gene mutations, to stress dyspoiesis is also unclear. Possibly, some of the dyspoietic mor- Abnormal erythroblast nuclear contours are preva- phologic features in CDA are the result of increased lent in stillborn and liveborn neonates. Only one erythropoiesis due to the anemia, and not necessar- prior report has described dyserythropoiesis in still- ily the result of the gene mutation. borns. Pajor et al. [25] reported dyserythropoiesis in the liver in terminated stillborns as 10-13.5% This study identified more erythroblasts with nu- and in IUFD stillborns as 22.2-30.8%, suggesting clear irregularities in the liver than the bone mar- that dyserythropoiesis was increased due to in utero row. One possible explanation for this finding is hypoxia. This group’s findings are generally consis- that this change is more readily recognized in the tent with the data we present in this study. In our fetal liver, because of the easy identification of study, stillborns who died in utero (IUFD) were erythroblasts segregated in the hepatic sinusoids. more likely to have >10% erythroblasts with nucle- The bone marrow contains an admixture of all three ar irregularities (p=0.032), whereas stillborns who hematopoietic lineages, making more subtle abnor- were terminated were more likely to have <10% malities difficult to discern on H&E section. (p=0.075). Twelve of the 14 IUFD stillborns had Morphologic evaluation of the bone marrow is best significant placental disease, all of which were likely done on aspirate smears, which were not routinely associated with fetal hypoxia. The mechanism of performed on our autopsy cases. this erythroid change could be related to erythro- poietin elevation and resultant stimulation of “stress The percent erythroblasts with nuclear irregularities dyspoiesis” as red blood cell production is increased did not correlate with low hematocrits. Therefore it [13-15]. Sixty-eight percent of liveborns in our is unlikely that we had many true cases of congeni- study had >10% erythroblasts with nuclear irregu- tal dyserythropoietic anemia in our series. Though larities in the bone marrow or liver. The most one may expect that a lower hematocrit would be prominent cause of death among liveborn neonates associated with increased erythroblasts with nuclear was congenital anomalies, including 18 infants irregularities, the opposite was actually observed. Erythroid nuclear irregularities 17

Low hematocrits were present in 15 of the 33 live- intervals. We found no statistical difference be- born infants with available hematocrit data; 67% of tween average postmortem intervals of liveborns these low hematocrit infants had <10% erythro- versus stillborns, infants with or without anemia, blasts with nuclear irregularities in the bone mar- with or without hydrops, or IUFD versus termina- row and 42% had <10% erythroblasts with nuclear tions. The average IUFD postmortem interval was irregularities in the liver. Anemia of prematurity has 29.6 hours, while that of the terminations was 59.8 been associated with poor red blood cell produc- hours. tion, attributed to low erythropoietin production by the liver and increased clearance [29-31]. However, it is known that dyserythropoiesis in- Erythropoiesis is not sufficiently stimulated to re- creases when specimens are stored at increased tem- plenish the red blood cell in anemia of prematurity. peratures. Wang and Glasser [32] noted an increase Interestingly, the stillborns and liveborns with <1% in dyserythropoiesis, including abnormal nuclear erythroblasts with nuclear irregularities were all ex- shapes, abnormal chromatin patterns, and cytoplas- tremely premature, at less than 21 weeks gestational mic changes with vacuoles, with increased length of age and less than 30 weeks gestational age, respec- storage of bone marrow biopsy specimens at room tively. We were unable to explore if transfusion oc- temperature compared to colder storage at 1°C-6°C. curred in any of the infants. We suspect that dys- As such, IUFDs remaining at human body tem- poiesis might be less common in the very premature peratures may have an artifactual increase in dys- patients in our series because there is less erythro- erythropoiesis. Wang and Glasser [32] noted that poietin-driven stimulation of erythropoiesis. the cytoplasmic changes they identified were more extensive than nuclear alterations; our current study Finding a cause of hydrops fetalis at autopsy can be did not concern cytoplasmic changes, as these al- a major challenge for the pathologist. Some cases terations are less conspicuous on bone marrow sec- are associated with congenital dyserythropoiesis tions. In addition, Hoffman et al. [33] identified [20-24]. The histologic hallmark of congenital dys- postmortem changes of erythroid precursors in erythropoiesis, abnormal nuclear erythroblast con- bone marrow aspirations, including bizarre nuclear tours, was not correlated with hydrops in this study. distortions of cloverleaf-like and pear-like configu- Three stillborn fetuses and three liveborn neonates rations, starting at three hours after death. demonstrated hydrops. Of these six cases, five had Therefore, post-mortem artifact is likely a signfi- evaluable bone marrow, of which two had <1% cant cause of erythroid nuclear irregularities ob- erythroblasts with nuclear irregularities. In the liver, served in this study. three of the five evaluable cases had <10% erythro- blasts with nuclear irregularities. In conclusion, abnormal erythroblast nuclear con- tours are common in stillborns and liveborns at au- Abnormal nuclear contours of erythroblasts could topsy. In the setting of hydrops, the pediatric pa- be a degradation artifact due to cell degeneration thologist may be tempted to cite congenital and apoptosis with increased postmortem interval dyserythropoietic anemia (CDA) as the cause of times. In this study, the average postmortem inter- hydrops when erythroblast nuclear contour abnor- val for both stillborns and liveborns was 39.0 hours, malities are observed. We would discourage such a with a range of 1-315 hours, and median of 27 diagnosis on morphologic grounds alone. True cas- hours. Seventy-four percent of autopsies were per- es of CDA-associated hydrops tend to be associated formed within the first 48 hours. The in utero de- with erythroblastosis not attributed to immune- mise to delivery interval of the stillborns is un- mediated destruction, Parvovirus B19 infection, or known. However, because only cases with fetal-maternal hemorrhage. A family history of sufficiently preserved liver or bone marrow histolo- CDA may be present, and additional studies such gy were selected, it is unlikely that there are many as electron microscopy or genetic testing should be cases with prolonged in utero demise to delivery used to support the diagnosis. 18 Annals of Clinical & Laboratory Science, vol. 44, no. 1, 2014

References 17. Lemery DJ, Santolaya J, Serre AF,Denoix S, Besse GH, Vanlieferinghen PC, Bezou MJ, Gaillard G, Jacquetin B. 1. Soothill PW, Nicolaides KH, Campbell S. Prenatal asphyxia, Serum erythropoietin in small for gestational age fetuses. Biol hyperlacticemia, hypoglycemia, and erythroblastosis in growth Neonate 1994; 65:89-93. retarded fetuses. Br Med J (Clin Res Ed) 1987; 18. Brunning RD, Orazi A, Germing U, Le Beau MM, Portwit A, 294:1051-1053. Baumann I, Vardiman JW, Hellstrom-Lindberg E. 2. Yeruchimovich M, Dollberg S, Green DW, Mimouni FB. Myelodysplastic syndromes/neoplasms, overview. In: Swerdlow Nucleated red blood cells in infants of smoking mothers. SH, et al, (eds). WHO Classification of tumours of haemato- Obstet Gynecol 1999; 93:403-406. poietic and lymphoid tissues. Lyon, France: International 3. Green DW, Mimouni F. Nucleated erythrocytes in health in- Agency for Research on Cancer; 2008. p. 88-93. fants and in infants of diabetic mothers. J Pediatr 1990; 19. Kamiya T, Manabe A. Congenital dyserythropoietic anemia. 116:129-131. Int J Hematol 2010; 92:432-438. 4. Hermansen MC. Nucleated red blood cells in the fetus and 20. Carter C, Darbyshire PJ, Wickramasinghe SN. A congenital newborn. Arch Dis Child Fetal Neonatal Ed 2001; dyserythropoietic anaemia variant presenting as hydrops feta- 84:F211-215. lis. Br J Haematol 1989; 72:289-290. 5. Phelan JP, Ahn MO, Korst LM, Martin GI. Nucleated red 21. Williams G, Lorimer S, Merry CC, Greenberg CR, Bishop AJ. blood cells: a marker for fetal asphyxia? Am J Obstet Gynecol A variant congenital dyserythropoietic anaemia presenting as a 1995; 173:1380-1384. fatal hydrops foetalis. Br J Haematol 1990; 76:438-439. 6. Thilaganathan B, Athanasiou S, Ozmen S, Creighton S, 22. Roberts DJ, Nadel A, Lage J, Rutherford CJ. An unusual vari- Watson NR, Nicolaides KH. Umbilical cord blood erythro- ant of congenital dyserythropoietic anaemia with mild mater- blast count as an index of . Arch Dis Child nal and lethal fetal disease. Br J Haematol 1993; 84:549-551. Fetal Neonatal Ed 1994; 70:F192-194. 23. Tekinalp G, Sarici SU, Erdinc AS, Gogus S, Balci S, Gurgey A. 7. Korst LM, Phelan JP, Ahn MO, Martin GI. Nucleated red Lethal hydrops fetalis due to congenital dyserythropoietic ane- blood cells: an update on the marker for fetal asphyxia. Am J mia in a newborn: association of a new skeletal abnormality. Obstet Gynecol 1996; 175:843-846. Pediatr Hematol Oncol 2001; 18:537-542. 8. Salafia CM, Ghidini A, Pezzullo JC, Rosenkrantz TS. Early 24. Remacha AF, Badell I, Pujol-Moix N, Parra J, Muniz-Diaz E, neonatal nucleated erythrocyte counts in preterm deliveries: Ginovart G, Sarda MP, Hernandez A, Moliner E, Torrent M. clinical and pathologic correlations. J Soc Gynecol Investig Hydrops fetalis-associated congenital dyserythropoietic ane- 1997; 4:138-143. mia treated with intrauterine transfusions and bone marrow 9. Minior VK, Bernstein PS, Divon MY. Nucleated red blood transplantation. Blood 2002; 100:356-358. cells in growth-restricted fetuses: associations with short-term 25. Pajor A, Kelemen E, Janosa M. Dyserythropoiesis in the fetal neonatal outcome. Fetal Diagn Ther 2000; 15:165-169. liver as an indicator of the intensity of stress caused by abor- 10. Bernstein PS, Minior VK, Divon MY. Neonatal nucleated red tion. Orv Hetil 1983; 124:619-622. blood cell counts in small for gestational age fetuses with ab- 26. Craig JE, Sampietro M, Oscier DG, Contreras M, Thein S. normal umbilical Doppler studies. Am J Obstet Gynecol 1997; Myelodysplastic syndrome with karyotype abnormality is as- 177:1079-1084. sociated with elevated F-cell production. Br J Haematol 1996; 11. Axt-Fliedner R, J. HH, Schmidt W. Nucleated red blood cell 93:601-605. counts in growth-restricted neonates with absent or reversed- 27. Kudo S, Harigae H, Watanabe N, Takasawa N, Kimura J, end-diastolic umbilical artery velocity. Clin Exp Obstet Kameoka J, Meguro K, Imaizumi M, Kaku M, Sasaki T. Gynecol 2002; 29:242-246. Increased HbF levels in dyserythropoiesis. Clin Chim Acta 12. Naeye RL, Localio AR. Determining the time before birth 2000; 291:83-87. when ischemia and hypoxemia initiated cerebral palsy. Obstet 28. Bard H, Gagnon C, Peri KG. HbF synthesis during stress Gynecol 1995; 86:713-719. erythropoiesis as determined by gamma-mRNA/non-alpha- 13. Clemons GK, Fitzsimmons SL, DeManincor D. mRNA quantification. Pediatr Res 1999; 45:684-686. Immunoreactive erythropoietin concentrations in fetal and 29. Stockman JAIII, Graeber JE, Clark DA, McClellan K, Garcia neonatal rats and the effects of hypoxia. Blood 1986; JF, Kavey RE. Anemia of prematurity: determinants of the 68:892-899. erythropoietin response. J Pediatr 1984; 105:786-792. 14. Widness JA, Teramo KA, Clemons GK, Garcia JF, Cavalieri 30. Widness JA, Veng-Pedersen P, Peters C, Pereira LM, Schmidt RL, Piasecki GJ, Jackson BT, Susa JB, Schwartz. Temporal re- RL, Lowe LS. Erythropoietin pharmacokinetics in premature sponse of immunoreactive erythropoietin to acute hypoxemia infants: developmental, nonlinearity, and treatment effects. J in fetal sheep. Pediatr Res 1986; 20:15-19. Appl Physiol 1996; 80:140-148. 15. Vatansever U, Acunas B, Demir M, Karasalihoglu S, Ekuklu 31. Dame C, Fahnenstich H, Freitag P, et al. Erythropoietin G, Ener S, Palo O. Nucleated red blood cell counts and eryth- mRNA expression in human fetal and neonatal tissue. Blood ropoietin levels in high-risk neonates. Pediatr Int 2002; 1998; 92:3218-3225. 44:590-595. 32. Wang LJ, Glasser L. Spurious dyserythropoiesis. Am J Clin 16. Snijders RJ, Abbas A, Melby O, Ireland RM, Nicolaides KH. Pathol 2002; 117:57-59. Fetal plasma erythropoietin concentration in severe growth re- 33. Hoffman SB, Morrow GWJ, Pease GL, Stroebel CF. Rate of tardation. Am J Obstet Gynecol 1993; 168:615-619. cellular autolysis in postmortem bone marrow. Am J Clin Pathol 1964; 41:281-286.