Congenital Anerythremic Erythroleukemia Presenting as Hepatic Failure Open Access

Congenital leukemias are rare, with an incidence of 4.7 per 1 million live births per year, and represent 0.8% of all childhood acute leukemias.1 Unlike leukemias in older children, leukemias with myeloid phenotypes are most frequent in newborns, especially the myeloblastic and myelomonocytic subtypes M4 and M5, respectively, in the French-American-British classification system.1–3 Erythroleukemias (M6) are exceedingly rare, with only 4 cases reported in the literature.3–6 By definition, congenital leukemias originate in utero, and symptoms are present at birth or in the first 4 weeks of life.2 Diagnosis requires the presence of leukemic blasts in the bone marrow and/or the involvement of nonhematopoietic tissues as well as the absence of any disease that can cause leukemoid reactions mimicking congenital leukemia.1,3 Additional criteria that some authors consider necessary include autopsy confirmation and the absence of trisomy 21, which may be associated with abnormal hematopoiesis.1,2 Pathologists are rarely confronted with leukemias, which are usually diagnosed on the bone marrow aspirate.2 We report a case of congenital erythroleukemia, without blasts in the peripheral blood, that was suspected on hepatic biopsy and confirmed by autopsy.

A male neonate, weighing 2.46 kg, was born to a healthy 34-year-old gravida 2, para 2 mother at 34.5 weeks of gestation by cesarean section because of fetal cardiac rhythm abnormalities. The pregnancy history was unremarkable. There was no intrauterine growth retardation or hydrops fetalis. At birth, a physical examination revealed hepatosplenomegaly. Neither lymphadenopathy nor cutaneous lesions were present. Laboratory studies at birth showed a platelet count of 69 × 10⁹/L, a hemoglobin level of 18.1 g/dL, and a white blood cell count of 30.4 × 10³/μL, with a normal neutrophilic count, 13% of activated lymphocytes, 18% of blast cells of erythroid origin (as shown by flow cytometry), and 64% of circulating mature erythroblasts. The newborn became anemic 4 days later (hemoglobin level, 9.0 g/dL). Aspartate aminotransferase, alanine aminotransferase, γ-glutamyl transferase, and alkaline phosphatase concentrations were 515 U/L (expected concentration, <35), 137 U/L (expected concentration, <50), 188 U/L (expected concentration, <45), and 245 U/L (expected concentration, <90), respectively. The infant blood group was O Rh-positive, and the Coombs test was negative. The concentrations of urinary catecholamines and their derivatives were normal. Viral and fetomaternal infections (adenovirus, parvovirus B19, toxoplasmosis, syphilis, herpes simplex I and II, cytomegalovirus, and rubella) were excluded by serologic tests and cultures. Peripheral blast cells completely disappeared after 7 days, and the patient progressively recovered from his thrombocytopenia, with platelet counts spontaneously normal after 3 weeks. Anemia persisted with initially low reticulocyte counts; thereafter, increased reticulocyte counts were greater than 300 × 10⁹/L after 3 weeks, with red blood cell fragmentation and less than 3% of mature circulating erythroblasts. Two bone marrow aspirates performed at 3 and 4 weeks of life showed less than 5% of dystrophic blast cells of unknown origin, which were not phenotyped (Figure 1). Erythroid cells with marked dyserythropoiesis and mild dysmegakaryopoiesis were present at 35% to 38%. Genes that coded for heavy and light chains of immunoglobulins and for T-cell receptors were not rearranged. Chromosomal analysis of bone marrow and peripheral blood showed a normal karyotype of 46,XY. Because the patient's liver function deteriorated, a metabolic hereditary disease was searched for but was not found (mitochondrial cytopathy, type I hereditary tyrosinemia, galactosemia, abnormalities of β-oxidization of long-chain fatty acids and bile acid disorders, and carbohydrate-deficient-glycoprotein syndrome). At 5 weeks, the clinical course was complicated by ascites and jaundice. Multiple intrahepatic confluent nodules with periportal and celiac lymphadenopathy were discovered by ultrasonography. A liver needle biopsy was performed. The patient died 5 days later of liver failure after supportive care only. An autopsy was performed. Two days before death, no circulating blasts were observed. A third bone marrow aspiration was attempted for flow cytometry. Bone marrow fibrosis was indicated, since the aspiration was very difficult; thus, flow cytometry could not be performed because of insufficient material. Three percent of the blasts were counted in this hypocellular sample.

Routine hematoxylin-eosin sections were prepared from formalin-fixed, paraffin-embedded tissues. Immunohistochemical studies were performed by indirect staining methods using antibodies against CD20 (1:100; Dako, Trappes, France), CD79a (1:50; Dako), CD3 (1:100; Dako), CD45 (1:400; Dako), glycophorin C (1:100; Dako), glycophorin A (1:50; Dako), BNH9 (1:50; Dako), factor VIII (1:200; Dako), CD41 (1:100; Dako), TdT (1:50; Dako), CD34 (1:150; Dako), CD43 (1:100; Dako), myeloperoxidase (1:50; Dako), CD10 (1:10; Tebu, Le Perray-en-Yvelines, France), CD68 (1:800, clone KP1; Dako), neuron-specific enolase (1:200; Dako), and CD1a (undiluted; Immunotech, Marseille, France).

Trisomy 21 and 11q23 rearrangements were searched for by interphase fluorescence in situ hybridization from a frozen liver sample with a locus-specific chromosome 21 probe (Kit Aneuvysion LSI 13/21, Vysis, Voisins Le Bretonneux, France) and a myeloid lymphoid leukemia probe (LSI MLL Dual Color, Vysis).

The liver biopsy showed extensive portal and perisinusoidal fibrosis without atrophic hepatocytes. There was a weak ductular proliferation in portal spaces. Hepatocytes were often clarified and vacuolated, and some were in giant cell transformation. No steatosis was observed. Perls staining did not show iron storage. No viral inclusions were noted. Portal spaces and sinusoids were infiltrated by many blast cells without maturation and by rare nests of extramedullary hematopoiesis (Figure 2). Blast cells expressed CD45. They were negative for CD43, CD34, CD41, CD3, CD20, CD79a, CD10, CD68, CD15, neuron-specific enolase, and CD1a.

At autopsy, mediastinal, retroperitoneal, celiomesenteric, and portal lymphadenopathy was detected. The weights of the spleen and liver were 30 g (expected weight, 13 g) and 225 g (expected weight, 133 g), respectively. The cut surface of the spleen was homogeneous. The liver was greenish and contained numerous millimetric pink nodules (Figure 3).

A histologic examination showed a diffuse and nodular blastic infiltration in the lymph nodes (Figure 4), liver, and adrenal glands. Splenic sinusoids of the red pulp and vessels of the renal cortex also were obliterated by blast cells. The bone marrow was infiltrated by 80% blast cells. The remaining cells were an admixture of dystrophic megakaryocytes and maturing myeloid elements. Reticulin myelofibrosis was conspicuous. Blast cells had round or multilobated nuclei with finely dispersed chromatin and one or a few nucleoli. Cytoplasm was abundant and intensely basophilic with Giemsa stain. Frequent binucleate and giant blast cells were seen. Extramedullary hematopoiesis and lymphoid depletion were noted in the thymus. The other organs (gastrointestinal tract, pancreas, bladder, prostate, testis, heart, lungs, and thyroid) were normal. The brain was not examined. Placental specimens were again carefully studied and showed blast cells in some allantochorionic vessels.

Blast cells expressed glycophorin A (Figure 5), glycophorin C, and BNH9. They were negative for factor VIII, CD41, myeloperoxidase, and TdT. Trisomy 21 and 11q23 rearrangements were not found by fluorescence in situ hybridization in the blast cells.

Congenital leukemias generally occur during normal pregnancies, with rare cases of in utero death.1 The symptomatology is varied, but a tumoral syndrome including hepatosplenomegaly and skin lesions is especially frequent.1,2 The other localizations (eg, lymph nodes, testis, kidney) are more unusual.1 

The blood cell count often shows hyperleukocytosis with a variable percentage of circulating blasts.1 Some cases of leukemia without circulating blast cells have been reported, especially in erythroleukemia.3 The bone marrow aspirate that usually confirms the diagnosis also elicits a precise immunophenotype of leukemia. The French-American-British classification of leukemias applies imperfectly to the congenital leukemia group because, unlike leukemia in older children, the phenotype of blast cells in newborns often shows a coexpression of different lineage markers.2 Cytogenetic analysis may show a constitutional chromosomal abnormality (eg, trisomies 13 and 9, monosomy 7) or changes restricted to the leukemic cells, particularly the myeloid lymphoid leukemia/11q23 rearrangement.7 

Differential diagnoses are numerous. In this case report, fetomaternal blood incompatibility and hemolytic diseases causing anemia with erythroblastosis were easily excluded. Indeed, the initial transient erythroblastosis was considered a nonspecific leukemoid reaction of the premature baby. The clinical and biologic presentation was also compatible with a fetal or fetomaternal infection, a metabolic disease, a tumoral lesion (congenital leukemia, Pepper syndrome, and Langerhans histiocytosis), and a transient myeloproliferative disorder of Down syndrome. All of these can be shown by hematologic abnormalities or hepatomegaly with liver test perturbation.1 Infectious diseases and metabolic disorders were excluded by biologic tests. Moreover, a leukemoid reaction associated with sepsis has never, to our knowledge, been described as a tumor syndrome.4 The catecholamine concentration and blast cell negativity for neuron-specific enolase on liver biopsy did not indicate a disseminated neuroblastoma. Letterer-Siwe disease was very unlikely without cutaneous lesions. Moreover, cell morphology and negativity for CD1a ruled out this hypothesis. Blast cells expressing CD45 could correspond to a leukemia or to a transient myeloproliferative disorder of Down syndrome. The liver involvement that occurs in 10% of transient myeloproliferative disorders of Down syndrome is often fatal.8,9 It is characterized histologically by panlobular fibrosis, cholestasis, and extramedullary hematopoiesis, as described in this case report, but also by iron deposits and a megakaryoblastic infiltrate.8,10 These megakaryoblasts express CD34, platelet markers (CD41 and factor VIII), and sometimes glycophorin A.8 Myelofibrosis is not unusual.10 Analysis of the karyotype is determinant and shows trisomy 21 that may be constitutional or restricted to the blastic population.8 

The autopsy of the baby of this study confirmed the diagnosis of congenital erythroleukemia and showed a multivisceral infiltration of blast cells expressing the erythroid markers glycophorin A and C. Although such findings are in agreement with the definition of congenital leukemia, the presentation is nevertheless atypical. Indeed, the successive bone marrow aspirates showed only a low number of unclassifiable blasts with dysmyelopoiesis and less than 50% erythroid cells. The absence of cytogenetic abnormalities was atypical, although some neonatal leukemias are reported with a normal medullary karyotype.3 Finally, the myelofibrosis and liver fibrosis observed in this case are more commonly seen in megakaryocytic leukemia and transient myeloproliferative disorder of Down syndrome than in erythroleukemia.10 

The treatment of congenital acute myeloid leukemias is not well codified.1 Chemotherapy, which is very toxic to newborns because of hepatic and renal immaturity, must be determined case by case. The mortality rate due to infectious complications is high during induction chemotherapy.1,2 The prognosis of congenital leukemia is poor, whatever the type, with only about 30% of children alive 6 months after the diagnosis.3,7 

We thank Prof T. Molina, Department of Pathology, Hôtel Dieu, Paris, and Dr J. Selves, Department of Pathology, Hôpital Purpan, Toulouse, France, for reviewing the slides.