Peripheral blood and bone marrow from the first 20 consecutive patients who tested positive for COVID-19 and underwent postmortem examination were evaluated. There were 12 men and 8 women, with an age range of 32 to more than 89 years (median, 63 years). There were 11 White non-Hispanic patients, 8 Hispanic patients, and 1 Black non-Hispanic patient. The cohort showed the following comorbidities: 80% with hypertension, 50% with diabetes, 25% with asthma, and 15% with immunosuppression (renal transplant, methotrexate use for psoriasis, and chronic prednisone use for sarcoidosis). The body mass index (BMI) ranged from 19.8 to 69.3 (median, 32.7); 14 (70%) of 20 patients were obese (with BMI >30). Six of 20 patients were former smokers, whereas the other 14 were never smokers; there were no current smokers in the cohort. One patient had chronic lymphocytic leukemia (CLL) Table 2. All 20 patients were admitted with hypoxemic respiratory failure and/or fever, accompanied by altered mental status in 15 cases and clinical findings of meningo-encephalitis including seizures in 1 case. Three patients died at outside facilities, with details of their presentation not available. Duration of hospitalization before death ranged from less than 1 day to 15 days. The principal cause of death in all 17 in-house cases was hypoxemic respiratory failure. Three patients died at outside facilities and were transferred following death to our institution for autopsy; although cause of death was not provided, the autopsy findings in these 3 patients were all consistent with respiratory failure.
Complete blood counts (CBC) were obtained during hospitalization in 17 patients with autopsy Table 3. Their hematologic abnormalities are summarized in Figure 1 and included leukocytosis in 5 patients on admission (WBC count, 15.8–24.4 × 109/L; median, 20.1 × 109/L) and in 15 patients on their last CBC (WBC count, 11.6–57.4 × 109/L; median, 19.0 × 109/L), elevated neutrophils in 7 patients on admission (ANC, 8.66–18.03 × 109/L; median, 15.1 × 109/L) and in 12 patients on their last CBC (ANC, 10.11–42.5 × 109/L; median, 15.5 × 109/L), lymphopenia in 11 patients on admission (ALC, 0.09–0.97 × 109/L; median, 0.65 × 109/L) and in 5 patients on their last CBC (ALC, 0.58–0.79 × 109/L; median, 0.63 × 109/L), anemia in 6 patients on admission (hematocrit, 32.7–39.8 L/L; median, 38.0 L/L) and in 13 patients on their last CBC (hematocrit, 28.2–38.7 L/L; median, 34.9 L/L), and thrombocytopenia in 6 patients on admission (platelets, 71–149 × 109/L; median, 127 × 109/L) and in 4 patients on their last CBC (platelets, 67–147 × 109/L; median, 110 × 109/L). One patient had leukopenia and neutropenia on admission (WBC count, 3.3 × 109/L; ANC, 1.7 × 109/L); no patient had leukopenia or neutropenia on their last CBC. One patient had thrombocytosis (platelets, 472 × 109/L) on admission, and another had thrombocytosis (platelets, 520 × 109/L) in the last CBC. A two-tailed paired t-test comparison of the patients' first and last CBCs revealed that there was a statistically significant increase in WBC count (mean, 10.21 × 109/L; 95% confidence interval [CI], 3.85–16.57 × 109/L; P = .0036) and ANC (mean, 8.77 × 109/L; 95% CI, 3.26–14.28 × 109/L; P = .0042) and significant reduction in hematocrit (mean, 5.59 L/L; 95% CI, 2.75–8.43 L/L; P = .0007).
Hematologic data on admission and last CBC before death for the 17 patients with in-house autopsy. Gray bar demonstrates normal range for WBC count (P = .0036) (A), absolute neutrophil count (ANC; P = .0042) (B), absolute lymphocyte count (ALC) (C), absolute monocyte count (AMC) (D), hematocrit (P = .0007) (E), and platelets (F).
The patient with a history of CLL had an absolute lymphocytosis consistent with persistent CLL. One patient had erythrocytosis (admission hemoglobin/hematocrit, 20.3 g/dL/64.1 L/L) as well as a neutrophilic leukocytosis and died less than 24 hours after admission, with the last CBC showing hemoglobin/hematocrit of 21.2 g/dL/66.1 L/L; erythropoietin was elevated at 45.5. CBC had been normal 4 months before admission. During hospitalization, the patient's physical examination noted moist mucous membranes, and no concerns for dehydration were noted. Five patients had giant platelets on peripheral smear. Plasma cells were identified on the peripheral smears for 2 patients.
Six patients had bacterial or fungal coinfections during their hospitalization: Klebsiella pneumoniae pneumonia, Staphylococcus aureus pneumonia, Staphylococcus capitis bacteremia, Enterococcus faecalis bacteremia, Escherichia coli urinary tract infection, and cutaneous Candida albicans infection. In addition to positive bacterial or fungal cultures (respiratory and blood cultures for the bacterial infections and wound cultures for the fungal infection), all 6 patients were clinically diagnosed with either a bacterial or fungal coinfection. Among these 6 patients, hematologic abnormalities (summarized in Figure 2) included leukocytosis in 1 patient on admission (WBC count, 4.66–15.80 × 109/L; median, 7.67 × 109/L) and in all 6 patients on their last CBC (WBC count, 11.55–18.96 × 109/L; median, 16.31 × 109/L), elevated neutrophils in 1 patient on admission (ANC, 3.85–15.10 × 109/L; median, 6.32 × 109/L) and in 4 patients on their last CBC (ANC, 3.85–15.68 × 109/L; median, 14.47 × 109/L), lymphopenia in all 6 patients on admission (ALC, 0.48–0.93 × 109/L; median, 0.61 × 109/L) and in 2 patients on their last CBC (ALC, 0.58–3.34 × 109/L; median, 1.2 × 109/L), thrombocytopenia in 3 patients on admission (platelets, 71–472 × 109/L; median, 137 × 109/L) and in 4 patients on their last CBC (platelets, 33–240 × 109/L; median, 138 × 109/L). Of the 5 patients with leukocytosis on admission, 1 patient belonged to the cohort of coinfected patients, whereas the other 4 patients did not have a known coinfection. The 15 patients with leukocytosis on their last CBC included all 6 patients with coinfection and 9 patients without known coinfection. Of the 7 patients with elevated neutrophils on admission, 1 had a known coinfection and 6 did not. Of the 12 patients with elevated neutrophils on their last CBC, 4 had known coinfections and 8 did not. The patient with the highest ANC (34.1 × 109/L) did not have known coinfection. On admission, 11 patients had lymphopenia, which included all 6 patients with coinfection and 5 patients without. On their last CBC, the 5 patients with lymphopenia included 2 with coinfection and 3 without. Finally, the 6 patients with thrombocytopenia on admission included 3 with coinfection and 3 without, whereas the 5 patients with thrombocytopenia on their last CBC included 3 with coinfection and 2 without. Among the cohort with coinfections, between the first and last CBCs, there was a statistically significant increase in WBC count (7.04 × 109/L [95% CI, 2.48–11.61 × 109/L], P = .0107) and reduction in hematocrit (7.08 L/L [95% CI, 0.95–13.21 L/L], P = .0312).
Hematologic data on admission and last CBC before death for the 6 patients with in-house autopsy and coinfections. Gray bar demonstrates normal range for WBC count (P = .0107) (A), absolute neutrophil count (ANC) (B), absolute lymphocyte count (ALC) (C), absolute monocyte count (AMC) (D), hematocrit (P = .0312) (E), and platelets (F).
Postmortem histologic examination of the bone marrow revealed normocellular marrow in 8 patients and hypercellular marrow in 12 patients Figure 3. There was complete maturation of myeloid elements, often with a left shift in 13 patients, left-shifted myelopoiesis with a paucity of fully mature elements in 6 patients, and myeloid elements that appeared too sparse to evaluate for maturation in 1 patient. The myeloid-to-erythroid ratio was increased in 11 patients, decreased in 1 patient, and within normal limits in 8 patients. The marrow of 1 patient appeared necrotic, precluding further evaluation. Megakaryocytes were increased in 3 patients, decreased in 3 patients, and appeared to be present in adequate numbers in the remaining cases. The patient with erythrocytosis had a hypercellular marrow (60%) with an apparently normal myeloid-to-erythroid ratio. One patient had lymphoid aggregates with the immunophenotype of CLL (CD20+, CD5+, LEF1+, cyclin D1−) but no peripheral blood lymphocytosis, likely representing monoclonal B lymphocytosis. The patient with CLL had scattered clusters of lymphocytes composed of CD20+ B cells and fewer CD3+ T cells. In general, patients with higher WBC count and ANC in their last CBCs had higher marrow cellularity. Hemophagocytosis, ranging from rare to many scattered cells, was identified histologically on H&E and/or CD68 immunohistochemical stains for all 19 evaluable patients.
A, Low power shows bone marrow that is hypercellular for age. B, Higher power shows trilineage hematopoiesis with maturing myeloid and erythroid elements, with a leftward shift in myeloid maturation. A large histiocyte with phagocytosis of multiple RBCs is seen at the center of the image (C, oil immersion). An immunostain for CD68 (D) shows scattered hemophagocytic histiocytes (arrows); the inset shows 1 hemophagocytic histiocyte with engulfed leukocytes (oil immersion).
In all cases, the hemophagocytic histiocytes were scattered in a background of hematopoietic elements. The hemophagocytic histiocytes did not form aggregates or replace zones of the marrow. The scattered histiocytes primarily demonstrated engulfed erythrocytes and occasional engulfed leukocytes (Figure 3).
Among the 20 patients, 17 had data for at least 4 of the 9 established criteria for HScore calculation. HScore ranged from 0 to 269 (median, 118). For patients with data for at least 7 of the 9 criteria, the median HScore was 125 (Table 3). Five patients had an HScore higher than 169.
Bone marrow of the 10 deceased patients who tested negative for COVID-19 was examined. The cohort consisted of 6 men and 4 women, with an age range of 48 to 84 years (median, 66 years). There were 7 White non-Hispanic patients and 3 Hispanic patients. The cohort showed the following comorbidities: 60% with hypertension, 40% with diabetes, 10% with asthma, and 10% with immunosuppression (renal transplant). The BMIs ranged from 19.0 to 31.6 (median, 26.3); 1 of the 10 patients was obese. The causes of death included infection in 40%, malignancy in 20%, cerebrovascular accident in 20%, cardiac arrythmia in 10%, and cardiogenic shock in 10%. Examination of the bone marrow revealed 3 instances of hemophagocytosis. Hemophagocytosis was identified in the bone marrow of a patient who died of cardiogenic shock following a protracted course in the intensive care unit (ICU) after open-heart surgery. In addition to hemophagocytosis, this patient's bone marrow showed extensive hemorrhage. Rare hemophagocytosis was identified in a patient who died of multifocal cerebral infarction secondary to emboli following a recent coronary artery bypass grafting surgery. The third patient with identified hemophagocytosis died of a hemorrhagic cerebrovascular accident; however, review of the patient's outpatient records revealed that there was a clinical suspicion of HLH, and the patient was undergoing further workup at time of death. This patient's HScore was 195. There was a statistically significant difference in the finding of hemophagocytosis between the COVID-19–positive and COVID-19–negative cohorts (P < .0001).
Bone marrow of 2 living patients who tested positive for COVID-19 was examined. The first patient was a 48-year-old man with diabetes mellitus type 2 and hypertension who underwent below-the-knee amputation for osteomyelitis. CBC showed anemia. Marrow from the amputation specimen showed trilineage hematopoiesis with slightly left-shifted myelopoiesis but no apparent hemophagocytosis. The HScore was 19. The second patient was a 63-year-old man with a pericardial effusion who underwent excision of a portion of pericardium, evacuation of a clot from the pericardial cavity, and removal of a portion of the xiphoid process. His CBC was remarkable for a neutrophilic leukocytosis that resolved and for anemia. The marrow showed maturing trilineage hematopoiesis with a few hemophagocytic histiocytes highlighted by an immunohistochemical stain for CD68.
Am J Clin Pathol. 2021;155(5):627-637. © 2021 American Society for Clinical Pathology