In this study of the first 20 consecutive patients who died of COVID-19 and had a postmortem examination performed at our institution, there was a high prevalence of comorbidities that have been associated with poor outcome, including hypertension, diabetes mellitus, and obesity.[5,13–15] Common hematologic abnormalities included neutrophilic leukocytosis, lymphopenia, anemia, and thrombocytopenia; the anemia and thrombocytopenia were mild in most cases. Lymphopenia is commonly associated with COVID-19,[16–19] with more pronounced lymphopenia associated with more severe COVID-19 disease.[14,15,19,20] The lymphopenia may be secondary to the cytokine storm induced by SARS-CoV-2. In addition, lymphocytes are reported to express ACE2 receptors on their surface and may become directly infected and killed by SARS-CoV-2.
Neutrophilic leukocytosis became more prevalent over time so that nearly all (15 of 17) evaluable patients demonstrated an elevated ANC on their last CBC. In this cohort, the increases in both WBC count and ANC between the first and last CBCs were statistically significant. Neutrophilic leukocytosis has been documented previously in association with COVID-19 infection, although it was more common and with more striking elevation of neutrophils in some patients in this series compared with other studies.[6,22] In some studies, leukopenia appeared to be more common than neutrophilia, and only a small minority of patients had leukocytosis.[4,19,23] Median WBC count and ANC were higher among our deceased patients than in unselected hospitalized patients with COVID-19 at our institution, suggesting that a more pronounced neutrophilic leukocytosis may correlate with worse outcome in patients with COVID-19. In several studies from Wuhan and other areas in China, patients with severe COVID-19 disease had a higher WBC count[14,16,18] and higher levels of neutrophils[14–16,18,20,24] than patients without severe disease, although the WBC counts and ANCs in these studies were often still within the normal range, in contrast to the striking neutrophilia noted in a subset of our patients. In a study from Singapore, no patient had an elevated WBC count on admission, although patients in the ICU tended to develop neutrophilia during hospitalization; however, their median ANC was lower than that seen in the last CBC of our patients. In 2 studies from Washington State in the United States,[17,25] WBC count was higher on average than in most studies from China but still not as pronounced as in some of our patients.
The cause of elevation of WBC count and ANC in patients with COVID-19 remains uncertain. Bacterial or fungal coinfection may play a role and has been described in a few patients from Wuhan. Additional studies have noted that patients with severe COVID-19 infection are more likely to have bacterial coinfections, likely secondary to lowered immune function. In our series, 6 patients had bacterial or fungal coinfection that may have contributed to their leukocytosis; nonetheless, the hematologic abnormalities were similar between the coinfected patients and patients without known coinfection in this cohort. Both the overall cohort and the patients with coinfections showed a statistically significant increase in WBC count and reduction in hematocrit, with a significant increase in ANC also noted in the overall cohort but not among the patients with coinfections. All patients without coinfection were monitored closely for signs of bacterial and/or fungal infection during their hospitalizations and received routine (often multiple) blood and urine cultures, all of which showed no growth. No clinical suspicion of coinfection was documented in their electronic medical records. Consequently, although a neutrophilic leukocytosis should prompt a search for superimposed bacterial and/or fungal infections, patients with COVID-19 and no known coinfections can nevertheless develop a pronounced neutrophilic leukocytosis.
As part of the cytokine storm induced by SARS-CoV-2, levels of granulocyte colony-stimulating factor (GCSF) and granulocyte-macrophage colony-stimulating factor are elevated, with more pronounced elevation of GCSF in severely ill patients compared with those with milder disease. Differences between our patients and those described in Wuhan, China, and other areas of Asia may be related to various factors, including differences in immune response related to ethnicities or other types of comorbidities. Additional larger and international studies are needed to investigate these factors.
Bone marrow taken at autopsy usually demonstrated trilineage hematopoiesis, sometimes with myeloid hyperplasia and/or a left shift. Similar findings have been reported previously in a few European series.[5,13] We identified the presence of hemophagocytosis in all 19 (100%) evaluable autopsy bone marrows, excluding the necrotic bone marrow that was unevaluable. Hemophagocytic histiocytes were scattered throughout hematopoietic elements and did not form aggregates. A similarly high incidence of hemophagocytosis was noted in recent series of 17 and 35 patients with COVID-19 in Spain and the United Kingdom, respectively,[5,27] although a series of 4 patients with COVID-19 in North America noted the presence of hemophagocytosis (specifically, lymphophagocytosis) in the lymph nodes and spleen but not in the bone marrow. Similar to the recent study from Spain, we identified primarily erythrophagocytosis in the bone marrow. Hemophagocytosis of leukocytes was also noted occasionally. The frequent presence of hemophagocytosis, as noted in several prior larger studies,[5,27] suggests that COVID-19 can induce an HLH phenotype. Mechanistically, COVID-19 may predispose to HLH through activation of the IL-1/IL-6 pathway, including overproduction of IL-1β by macrophages.[9,29] Hemophagocytosis, however, is not a specific finding for HLH. Erythrocytosis in particular appears to be more frequently noted in patients with underlying infections, many of whom do not ultimately receive a clinical diagnosis of HLH. In our cohort, erythrophagocytosis was the primary form of hemophagocytosis seen, regardless of the patient's HScore.
Similar to the current findings of hemophagocytosis in patients with COVID-19, previous studies on severe acute respiratory syndrome (SARS) demonstrated variable bone marrow histologic changes. Trilineage hematopoiesis without evidence of hypoplastic marrow or HLH was reported in one study. Another study documented hemophagocytosis in the lungs of patients infected with SARS but did not examine the bone marrow of these patients, whereas hemophagocytosis was identified in the bone marrow of SARS-infected patients in 2 additional studies.[33,34]
Aberrant inflammatory response has been implicated in the pathogenesis of COVID-19 in several recent studies.[9,29] The hyperinflammatory response may take the form of macrophage-activation syndrome, also known as secondary HLH. In one study, patients with clinical findings consistent with HLH were noted to show elevated HScores, which estimate the likelihood of HLH.[9,12] Secondary HLH, in contrast to primary/hereditary hemophagocytic syndrome, is a reactive condition that may occur in response to infection, autoimmune diseases, or malignancies. Despite the potential importance of HLH as a driver of the proinflammatory cytokine response in some patients with COVID-19, to date there have been few reported analyses of the bone marrow in affected patients. A recent study documented the HScores of 8 patients who tested positive for COVID-19 but did not examine their bone marrow. Identification of underlying bone marrow alterations in patients with COVID-19 provides important data to support HLH as one of the manifestations of COVID-19 and furthers our understanding of its pathophysiology.
Secondary polycythemia, characterized by erythrocytosis with an elevated erythropoietin level, was noted in one autopsy patient who had a normal CBC documented 4 months previously. Secondary polycythemia can develop in people residing at high altitudes, those with pulmonary disease, and rarely as a paraneoplastic phenomenon. The striking elevation of the hemoglobin and the hematocrit in this patient was unusual and out of proportion to those in hypoxemic patients with lung disease and may represent an unusual manifestation of SARS-CoV-2 infection. Of note, rare patients positive for HIV with polycythemia that eventually resolves, sometimes in response to antiretroviral therapy, have been reported.[36,37] To date, there has been only one prior report of (presumed secondary) polycythemia in patients with COVID-19.
We identified hemophagocytosis in 3 of the 10 bone marrows of patients who tested negative for COVID-19. Two of the patients had had recent cardiac surgeries with protracted ICU courses. In one case, the bone marrow showed extensive hemorrhage. The hemophagocytosis seen may represent a normal response to the hemorrhage. In the third case, there was a clinical suspicion of HLH. The finding of hemophagocytosis in the patient's bone marrow and an HScore of 195 raise a strong possibility of the diagnosis of HLH. Although the cohort was small, this finding further highlights the striking prevalence of hemophagocytosis in the bone marrow of patients with COVID-19 relative to the typical prevalence of hemophagocytosis in COVID-19–negative autopsies (P < .0001).
This study has several limitations. First, given the nature of a single-institution study with a limited sample size, our estimates of the incidence of peripheral blood abnormalities and hemophagocytosis may not be precise. Second, because this study primarily focused on patients with fatal COVID-19 disease, the incidence of hemophagocytosis in all patients with COVID-19, including those with asymptomatic or mild disease, remains unclear. Nevertheless, our findings are consistent with other reports of worse outcomes among patients with a clinical profile resembling HLH. Third, our autopsy cohort examined to date was predominantly White, limiting generalization of these findings to the overall racial and ethnic demographics of the United States or to the national infection and mortality trends. Finally, our analysis was based primarily on autopsy material with variable autolysis and degenerative change, which may render optimal evaluation of some histologic features difficult.
In conclusion, we identified a high incidence of neutrophilic leukocytosis, lymphopenia, and hemophagocytosis in patients with fatal COVID-19. We noted the frequent presence of hemophagocytosis in the bone marrow of deceased patients, present in 100% of evaluable bone marrow. Our findings support the burgeoning theory that COVID-19 induces HLH in a subset of patients. Finally, the absence or the presence of only mild hemophagocytosis in 2 bone marrows from living patients with COVID-19 may support the hypothesis that patients with an HLH phenotype experience worse clinical outcomes.
Am J Clin Pathol. 2021;155(5):627-637. © 2021 American Society for Clinical Pathology