Splenic kaposiform hemangioendothelioma presenting as insidious consumptive coagulopathy

Abstract

A healthy 2-year-old male with a history of mild persistent asthma visited the outpatient hematology clinic to evaluate intermittent epistaxis. There was no history of recent illness. His exam was notable for left upper extremity bruising secondary to recent venipuncture, but no lymphadenopathy or hepatosplenomegaly. Complete blood count (CBC) revealed hemoglobin 9.3 g/dL, leukocyte count 20.7 109/L, and platelet count 33 k 109/L. Peripheral blood smear exhibited polychromasia, numerous large platelets, and a few reactive lymphocytes. The differential diagnosis for both thrombocytopenia and anemia include processes leading to impaired hematopoiesis, increased cell destruction, or abnormal cell pooling. Impaired hematopoiesis could stem from infectious suppression, infiltrative processes (leukemia, metabolic disorders, etc.), or inherited and acquired bone marrow failure syndromes. Possible destructive or consumptive processes included microangiopathy, disseminated intravascular coagulation (DIC), or immune destruction. Sequestration within the liver or spleen was considered, but he was without hepatosplenomegaly to suggest a site of abnormal pooling. We approached his work-up with a stepwise approach. Screening tests were normal for common infectious diseases implicated in marrow suppression (Table 1). Since the patient appeared well and did not exhibit any findings that would prompt an immediate evaluation for malignancy or sepsis, we decided to repeat a short interval CBC to see if his cytopenias self-resolved. When they did not, further evaluation of the bone marrow was warranted. Bone marrow biopsy and aspirate did not demonstrate marrow infiltration or failure (Table 1). Lab work was sent to investigate consumptive processes. Lactate dehydrogenase levels and reticulocyte counts were elevated (Table 1), suggesting a hemolytic process. A hemoglobinopathy screen and glucose-6-phosphate dehydrogenase testing were normal, though active hemolysis can skew results. In an ill-appearing child, the presence of hemolysis and thrombocytopenia should prompt an investigation for DIC or microangiopathy, but in this well appearing-child, an immune process was thought likelier. We, therefore, pursued an immunologic work-up, including screening for autoimmune lymphoproliferative syndrome (ALPS) (Table 1). The results were unrevealing except for a low IgG level at 228 (505–1280 mg/dL) with normal lymphocyte subsets. Low IgG is often associated with systemic autoimmune disorders, vasculitides, atopy, and asthma. We attributed the low IgG and eosinophilia to the patient's known history of asthma. A consumptive autoimmune process was high on the differential, and the involvement of two cell lines would be classified as Evans syndrome. Evans syndrome, though never a primary diagnosis, is treated upfront with corticosteroids or intravenous immunoglobulin (IVIG). Therefore, he received oral prednisone 4 mg/kg/day for 4 days and later one dose of IVIG 0.8 g/kg for therapeutic and diagnostic purposes. There was no clinical improvement of the cytopenias. The patient exhibited worsening bruising, petechiae, and lip hematoma over 1 week. With no response of the thrombocytopenia and anemia to immunosuppressive agents, a trial platelet transfusion with a post-transfusion count check to assess response was performed. The 1-h post-transfusion platelet count was 111 k 109/L but fell to 13 k 109/L within 48 h. His hemoglobin nadired at 6.5 g/dL, prompting a packed red blood cell (PRBC) transfusion. Without an external source of hemorrhage and undetectable haptoglobin < 31 mg/dL (41–165 mg/dL), we concluded worsening hemolysis. Direct coombs testing was negative. IVIG is sometimes associated with development of hemolytic anemia, but with positive hemolysis markers prior to IVIG therapy, we did not believe IVIG was to blame. Moreover, his exam became remarkable for splenomegaly over 3 weeks, expanding from 2 to 4 cm below the costal margin. On ultrasound, the spleen was described as “inhomogeneous” and measured 11.1 cm (> 97% for age). We became concerned for either occult lymphoma with secondary autoimmune phenomena, or vascular malformation with cellular consumption. Computed tomography scan with contrast of the neck, chest, abdomen, and pelvis was notable for splenomegaly to 14.6 cm with diffuse inguinal and axillary lymphadenopathy without an identifiable mass (Figure 1A,B). Magnetic resonance imaging (MRI) brain and body with and without contrast and arteriography were obtained to assess for a vascular malformation or another primary lesion. MRI demonstrated heterogeneous splenomegaly, now measuring 16 cm, without discrete lesion or abnormal vascularity (Figure 1C–G). However, retrospectively, venous phase enhancing and nonenhancing sub-centimeter T2− hyperintense lesions and hypointense lesions were present within the spleen. There was also a T2-hyperintense, uniformly enhancing soft tissue retroperitoneal lesion. The differential diagnosis of this retroperitoneal lesion included retroperitoneal fibrosis, Ig4 related autoimmune disease, or reactive lymphadenopathy. A possibility of venous malformation was also considered; however, no flow voids were present in the retroperitoneum or spleen. The imaging did not appear consistent with a solid tumor or lymphoma. Thus, we were still without a diagnosis. Meanwhile, the patient experienced persistent oozing from the previous bone marrow biopsy site despite platelet transfusions and tranexamic acid. His coagulation profile became notable for a prolonged prothrombin time (PT) of 27 (11.5–16.1 s) and for prolonged partial thromboplastin time (PTT) of 64 (27–39.4 s). Both corrected on mixing studies, indicating a lack of plasma factor rather than the presence of an inhibitor. Plasma factor levels were notable for factor V of 0.24 (0.7–1.5 U/mL), factor VII of 0.60 (0.7–1.6 U/mL), factor VIII:C of 0.29 (0.6–1.5 U/mL), and normal factor II, IX, X, XI levels (Table 2). In addition, his fibrinogen was 33 (208–481 mg/dL) and D-dimer > 20 (0.09–0.53 μg/mL). Protein C and Protein S levels were 37 (70–150 %N) and 52 (58–128 %N), respectively. He began requiring daily infusions of PRBC, platelets, fresh frozen plasma (FFP), aminocaproic acid, vitamin K, and cryoprecipitate (CPP) to prevent bleeding from IV sites and mucocutaneous surfaces (Figure 2A–C). This coagulation profile was consistent with DIC, though we had no primary etiology. With evidence of a markedly abnormal spleen and lymphadenopathy, an unspecified lymphoproliferative disorder causing consumptive coagulopathy became our working diagnosis. First-line treatment of lymphoproliferative disorders consists of high-dose corticosteroid or IVIG, while second-line treatments typically employ mycophenolate mofetil or sirolimus. With a demonstrated lack of response to IVIG, we trialed methylprednisolone 4 mg/kg/day, eventually increasing to 30 mg/kg/day for 3 days. He also began mycophenolate mofetil 600 mg/m2/dose twice daily and, after 3 days, transitioned to sirolimus 2 mg/m2/day for continued immunosuppression. After nearly 2 weeks of mucocutaneous bleeding, his coagulopathy remained unresponsive to continuous product replacement and immunosuppression (Figure 2C). Ultimately, the need for a tissue diagnosis led us to splenectomy. The patient was transferred to the pediatric intensive care unit to optimize hemostasis before surgery. Continuous low-dose heparin 5 cc/kg/h was infused for 24 h to offset thrombin activity contributing to coagulopathy. However, the patient's PT, international normalized ratio (INR), and PTT did not improve on heparin infusion (Table 2) and was discontinued in favor of a continuous FFP 2 mg/kg/h. The patient underwent splenectomy via ex-laparotomy after preparation with 24 h of continuous FFP, seven units of CPP, and six units of pooled platelets. He tolerated the procedure well without any excess bleeding. Five hours after splenectomy, the patient's platelet count improved from 65 to 223 k 109/L, and fibrinogen increased from 150 to 310 mg/dL. Forty-eight hours post-operatively he continued prophylactic continuous FFP infusion 1–3 mg/kg/h, and received an additional one unit PRBC, one unit CPP, and two platelet units. Splenic pathology revealed a diffuse KHE with three spleniculi thought to account for the retroperitoneal fibrosis seen on MRI (Figures 1 and 3). The lymphadenopathy seen on imaging was deemed reactive, as described in other KHE cases. Immunosuppression was discontinued. The patient was evaluated in the clinic 10 days after discharge from the hospital. He had no bleeding symptoms. The platelet count had rebounded to 824 k 109/L, and there was complete laboratory resolution of his consumptive coagulopathy (Figure 2A–C). He continued to have significant eosinophilia, 3940 109/L for which he underwent further evaluation. Allergy panel testing was normal; it was concluded that the eosinophilia was related to his known history of asthma. As of 2 years post-splenectomy, he remains symptom-free with normal laboratory parameters and normal body imaging. Kaposiform hemangioendothelioma (KHE) is an exceptionally rare, locally aggressive vascular tumor found in cutaneous and deep tissue structures.1 First described in 1989 by Niedt et al., it possesses features reminiscent of infantile hemangioma and Kaposi sarcoma.2-4 Up to half of KHE cases may present as a life-threatening consumptive coagulopathy called Kasabach-Merritt Phenomenon (KMP).5 KHE arises from various anatomical origins and histologically is characterized by sheets of spindle cells in an infiltrative pattern with surrounding epitheloid cells and fibrin sheets. It has several distinct histological features which distinguish it from similar vascular tumors. KHE is lobular with slit-like vessels, mimicking Kaposi sarcoma; however, KHE can be differentiated by its lack of plasma cells and no central tumor focus.6 It shares features with tufted angioma, as both tumors derive from vascular endothelium with detectable CD-34, CD-31, and vascular endothelial growth factor receptor (VEGFR) markers,3 but KHE has a disorganized appearance and cannonball pattern.7 KHE also contains a lymphatic vascular component positive for D2-40 and PROX1, endothelial markers of lymphatic channels.8 The latter of these is a more recent finding explaining KHE's association with lymphatic abnormalities and has been apparent since the earliest reports.2 A useful immunohistochemical platelet marker is CD61, highlighting microthrombi.3 KHE most commonly arises in infancy, manifesting after birth, though both cutaneous and deep tissue KHE have been identified in adults. Cutaneous lesions present with a single firm, painful violaceus lesion with irregular borders. They can rapidly expand through the subcutaneous fat into muscle and local lymph node involvement. KHE originating in the deep tissues, such as the one in our case, are most commonly in retroperitoneal, mediastinal, muscle, and boney tissues.5 Their depth and poorly-defined nature can make a clinical diagnosis difficult. KHE's infiltrative and progressive growth differentiates it from classic infantile hemangioma, which has a proliferative phase followed by involution. As such, KHE has the potential to cause substantial deformation and dysfunction. Croteau et al. reported significant musculoskeletal complaints, including pain and decreased range of motion in a patient with KHE diagnosed at a later age.5 Other clinical complications include severe thrombocytopenia, lymphedema, and nearby vascular and organ structure compromise secondary to tumor growth.5 KMP is recognized in half of the cases of KHE, and may increase KHE-associated mortality by up to 30%.5, 9 KMP is a unique form of DIC stemming from generalized microthrombi formation from increased fibrin deposition and breakdown. The thrombocytopenia in KMP may be secondary to tumor vessel congestion triggering turbulent flow and secondary activation of procoagulant properties of the vascular endothelium.1, 3 Microangiopathic hemolysis can occur, with schistocytes and worsening anemia present,10 as in our case. Risk factors for the development of KMP include large size of the KHE tumor (> 5–8 cm), deep tissue location, and congenital KHE.2, 5 Our case highlights a rare presentation of KHE. Deep tissue KHE are less common than cutaneous lesions, and reports of isolated splenic KHE are exceedingly rare, with only two pediatric cases and two adult cases reported.11-14 Multifocal KHE has been reported as a simultaneously cutaneous and deep tissue structure, further illustrating the wide range of possible presentations.6, 15, 16 Out of these four reports of isolated splenic KHE, only Shabtaie et al.'s description of a prenatally diagnosed splenic mass was associated with KMP.13 In all but one of these reports, the splenic lesions were diagnosed on imaging. Mota et al. reported on splenic KHE in an adult who underwent splenectomy and was found to have additional KHE lesions in the liver and vertebrae not identified on original imaging.15 Unlike prior pediatric cases where there were focal splenic masses, our patient had diffuse spleen involvement, which displayed a T2-hyperintense and hypointense sub-centimeter focal lesion. Moreover, our patient did not display enhancement on arterial phase imaging, and there was no flow void within the spleen as there was in Shabtaie et al.13 Similar to our case, Abdulrahman et al. describe a KHE identifiable only by pathologic examination.11 Splenectomy was a reliable form of cure for all reported cases of splenic KHE except one (Mota et al.), implying possible improved outcomes for isolated splenic lesions with complete resection.15 No mode of imaging has been identified as superior for the diagnosis of KHE. It can be visualized on ultrasound, a reasonable approach for superficial cutaneous KHE, where the lesion appears as a heterogeneous, hyperechoic mass with irregular borders, and high vascularity.17, 18 Our patient's ultrasound revealed heterogeneous echotexture splenomegaly. However, MRI with gadolinium enhancement is currently the preferred method of imaging for the diagnosis of KHE. Lesions classically appear as poorly defined, hypointense thickened soft tissue masses, often involving multiple tissue planes on T1-weighted images.5 T2–weighted MRI imaging commonly demonstrates a hyperintense mass with reticular stranding in the subcutaneous fat.5 MRI with arteriography may provide benefit in defining venous and arterial patterns within the lesion but cannot confirm a diagnosis.17 KHE lesions are heterogeneously hyperintense or mildly hyperintense and demonstrate speckled hypointensity on T2-weighted images. The hypointense areas represent hemosiderin deposition and hyaline stromal response.19 KHE lesions can be hypervascular, containing arterial feeders and draining veins that appear as flow voids.19 KHE can mildly restrict diffusion, and this finding can differentiate it from infantile hemangiomas. For complicated deep tissue lesions as in this report, histological confirmation is necessary for diagnosis and is considered the gold standard. More recent studies have focused on identifying unique genomic and vascular serum markers, such as guanine nucleotide-binding protein subunit alpha-14 (GNA14) and angiopoietin-2 (ANG-2), to aid in rapid diagnosis.20 Despite multiple imaging modalities and current research, KHE is still misdiagnosed or has delayed discovery, stressing the importance of histological examination and research. There is no standard approach to treatment. There are limited data on long-term therapy, and management depends on location, size, and secondary complications. Steroids are considered the first-line treatment with variable success. More recent studies have identified improved tumor burden and lab abnormalities when combined with vincristine or sirolimus.21 As either single or dual therapy, vincristine inhibits endothelial proliferation, leading to KHE regression and reduced complications from KMP; it is also effective in steroid-resistant KHE.22, 23 Sirolimus is indicated for previously-treated and relapsed KHE.21, 24, 25 Surgical removal or embolization is more invasive approaches for extensive, infiltrative deep tissue KHE. The bleeding risk from surgical excision is of great concern in cases exhibiting KMP, and the tumor itself may be technically challenging to remove completely, further contributing to the bleeding risk. Without treatment, KHE can either proliferate or remain stable without secondary complications. KHE has little potential to metastasize despite involving the lymphatic system. Mortality typically stems from hemorrhage, infection, or compromise of nearby organ structures, and is estimated to be 12%.9 There have been no reports of actual spontaneous regression. Presence of KMP, deep tissue location, and tumor invasion of nearby structures are poor prognostic factors.3, 19 In summary, our patient presented with a rare form of deep-tissue splenic KHE and life-threatening consumptive coagulopathy. A true diagnosis was not achieved by imaging alone. He ultimately required tissue identification by splenectomy, which rapidly improved his coagulopathy and symptomatic bleeding. This case is unusual in the sense that at initial presentation, his only symptom of KHE was mucocutaneous bleeding. The bleeding manifested weeks before the splenic tumor was detectable on examination or imaging. Given this rare deep-tissue presentation in a pediatric patient who exhibited KMP, the case serves as valuable indication for clinicians to have a high index of suspicion for occult vascular tumors in the setting of unexplained aggressive consumptive coagulopathy. This work was supported by funding HD071834 (to DND). Additional support was provided by the UPMC Children's Trust Young Investigator Foundation Fund 020284 (to DND). Dr. Reyes-Múgica is in part supported by the Marjory K. Harmer endowment for research in Pediatric Pathology. The authors have no competing financial interests to disclose. M Shiel, R Kalpatthi, and D Nolfi-Donegan treated the patient. M Goldenberg, M Shiel, M Reyes-Múgica, R Kalpatthi, S Subramanian, and D Nolfi-Donegan wrote the manuscript. M Goldenberg and M Shiel contributed equally to the project and manuscript. Verbal/Written informed consent was obtained from the patient's guardian for publication of this case report. A copy of the written consent is available for review by the Editor-in-Chief of this journal. Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.