Clinical Significance of CBC and WBC Morphology in the Diagnosis and Clinical Course of COVID-19 Infection

Olga Pozdnyakova, MD, PhD; Nathan T. Connell, MD, MPH; Elisabeth M. Battinelli, MD, PhD; Jean M. Connors, MD; Geoffrey Fell, MS; Annette S. Kim, MD, PhD


Am J Clin Pathol. 2021;155(3):364-375. 

In This Article


Comparison of Laboratory Values Between COVID-19–Positive ICU and Non-ICU Patients

COVID-19–positive patients demonstrated striking numerical peripheral blood WBC and inflammatory marker abnormalities (Table 1 and Supplemental Table 1; all supplemental materials can be found at American Journal of Clinical Pathology online). Although overall mean WBC counts were in the normal range, as seen in previous reports,[9,10] patients with COVID-19 had increased ferritin (mean, 1,905 μg/L) and CRP (mean, 124 mg/L) as well as lymphopenia (mean absolute lymphocyte count [ALC], 0.97 × 109/L) and monocytopenia (mean absolute monocyte count [AMoC], 0.44 × 109/L; Supplemental Table 1). The laboratory values, including both the markers of inflammation and CBC parameters, were significantly different between the ICU and non-ICU COVID-19–positive patients. Patients in the ICU settings had significantly higher ferritin (2,359.22 vs 971.85 μg/L, P = .013) and CRP (205.05 mg/L vs 103.03 mg/L, P < .0001) levels, a reflection of the hyperinflammatory state that has been described in a subset of COVID-19–positive patients with more severe disease presentation.[1] In addition, patients in the ICU setting had significantly higher WBC counts with associated neutrophilia and left-shifted granulopoiesis (WBC, 9.96 vs 6.00 × 109/L, P = .0009; absolute neutrophil count, 8.32 vs 4.21 × 109/L, P = .001; absolute immature granulocyte [IG] count 2.46 vs 0.64 × 109/L, P = .0212). Importantly, patients in the ICU were more likely to have lymphopenia (ALC, 0.75 vs 1.24 × 109/L, P = .0124) and increased nucleated RBCs (0.24 vs 0.03 × 109/L, P = .0462), while there was no statistical difference in monocyte count (AMoC, 0.47 vs 0.40 x 109/L, P = .5820).

Comparison of Cellular Morphology Between COVID-19–Positive ICU and Non-ICU Patients

The most striking findings in patients with COVID-19 were the WBC morphologic changes observed on peripheral blood smears (Table 2 and Image 1). Abnormal morphologic features were present in 100% of patients with COVID-19 (Supplemental Table 1). Although the constellation of morphologic findings was distinct, some of them resembled changes associated with other viral or bacterial infections. The most uniform morphologic finding was cytoplasmic vacuolization that was present in multiple cell types with varying frequency. Monocytes demonstrated the most impressive vacuolization, with numerous large coalescing vacuoles seen in 71% of patients. Smaller cytoplasmic vacuoles were also present in neutrophils (83%), lymphocytes (53%), and eosinophils (13%). Only two patients with COVID-19 did not demonstrate any cytoplasmic vacuoles. Neutrophil toxic granulation (97%), large granular lymphocytes (LGLs, 86%), and atypical lymphocytes (54%) were also frequent in patients with COVID-19. Importantly, there were different patterns of abnormal morphologic changes in lymphocytes and monocytes between ICU and non-ICU patients with COVID-19. Monocytes with large coalescent cytoplasmic vacuolization or atypical lymphocytes (grade >0) were more prevalent in non-ICU patients (odds ratios [ORs], 0.21, P = .0051 and 0.21, P = .0006, respectively), while myeloid left shift was associated with ICU status (OR, 3.10, P = .0109). The latter confirmed the CBC finding of increased immature granulocytes in the ICU COVID-19–positive patients (see above). When considering only grade 3 morphologic changes (present in >25% of cell lineage), there was a trend for an inverse correlation of LGLs and ICU status (OR, 0.43, P = .0528; data not shown).

From the pool of candidate predictors, we estimated a multivariate logistic regression model for ICU outcome in COVID-19 infection, regressing upon sex, monocyte vacuolization status, presence of atypical lymphocytes, and left-shifted myeloid cells (Table 2). While holding other predictors constant, male sex increased the odds for being in the ICU 3.9-fold, and the presence of any left-shifted myeloid cells increased the odds 3.7-fold. Conversely, the presence of monocyte vacuolization or atypical lymphocytes decreased the odds of being in the ICU for COVID-19–positive patients by 0.21-fold and 0.23-fold, respectively.

Comparison of Laboratory Values and Cellular Morphology Between COVID-19–Positive Deceased and Alive Patients

There was a high mortality rate among the COVID-19–positive ICU patients of 39.2%: 20 patients died of the disease, while 31 patients completely recovered. To better understand the disease course and identify potential predictors of inferior outcome, we evaluated laboratory and WBC morphologic parameters at the time of diagnosis between these two groups of patients Table 3 and Table 4. The deceased group constituted an older patient population (mean age, 71.35 vs 59.42 years, P = .0065), but there was no significant difference in sex distribution, with males comprising most patients in both groups (male to female ratio of 1.50 and 1.58, P = .9266). Both groups had markedly elevated ferritin and CRP levels. No differences were observed in the CBC values and overall WBC morphology at the time of diagnosis. As noted above, the COVID-19–positive ICU patients had significant lymphopenia compared with the non-ICU COVID-19–positive patients. However, there was a trend to even more significant lymphopenia at presentation in the deceased group compared with ICU survivors (0.54 vs 0.90 × 109/L , P = .0495). The morphologic WBC changes, such as monocyte, lymphocyte, and neutrophil vacuolization; the presence of atypical lymphocytes or LGLs; and toxic granulation in neutrophils were similar between these two groups. The only significant difference was the much higher number of left-shifted/immature granulocytes (promyelocytes, myelocytes, and metamyelocytes) on peripheral blood of the deceased patients (60% vs 6%, P = .00003) noted on Cellavision morphologic evaluation (Table 4). However, there was no difference in immature granulocytes, as evaluated by the automated IG parameter (absolute IG count of 0.20 vs 0.29 × 109/L, P = .4776). This discrepancy between the manual and automated evaluation possibly reflects the selection bias of Cellavision in capturing leukocytes for manual differential analysis and its lower sensitivity due to a much smaller number of cells for analysis (115 leukocytes for Cellavision vs ~30,000 leukocytes for CBC).

Serial CBC data were available for 18 (90%) deceased patients, allowing us to assess the CBC numerical and morphology parameter dynamics from the ICU admission to the date of death. Despite the lack of differences in CBC at the time of the ICU admission between the patients who subsequently died or recovered, the former patients demonstrated a significant increase in absolute neutrophil count (ANC) over the course of the disease in 14 (77.8%) patients with the overall percent increase from already elevated ANC values that ranged from 4.8% to 208.3%, with the majority demonstrating greater than a 30% ANC increase over the baseline values (10/15, 66.6%). Only four (22.2%) patients demonstrated a decrease in ANC, with only two of them with percent change over 20% Figure 1A. WBC showed similar dynamics (data not shown). Another fascinating finding was a continuing decline in ALC from the ICU admission to demise. Nine (50%) of 18 patients demonstrated a decrease in ALC by 28.3% to 100%, while among the nine patients who showed an ALC increase, only four demonstrated normalized ALC of greater than 1 × 109/L Figure 1B. Interestingly, although AMoC fluctuated, it remained within the normal reference range during the course of the disease (data not shown). In terms of WBC morphology, it appeared that striking vacuolization with coalescing vacuoles in monocytes was less prominent in peripheral blood smears closer to demise (data not shown).

Figure 1.

A, Percent absolute neutrophil count (ANC) change between diagnosis and demise in coronavirus disease 2019 (COVID-19)–positive intensive care unit (ICU) patients. B, Percent absolute lymphocyte count (ALC) change between diagnosis and demise in COVID-19–positive ICU patients.

Comparison of Laboratory Values Between COVID-19–Positive and COVID-19–Negative ICU Patients

To rule out the possibility that the CBC and morphologic changes seen in the ICU COVID-19–positive patients could be attributed to the ICU environment rather than SARS-CoV-2 infection, we compared the findings between COVID-19–positive ICU patients (51 patients) with COVID-19–negative ICU patients (30 patients). The data were collected at the time of diagnosis for COVID-19–positive patients and at the time of ICU admission for COVID-19–negative patients.

The demographic, laboratory, CBC, and morphology parameters are presented in Table 5 and Table 6 and Supplementary Table 1. There was no significant difference in age and sex between these two groups. Ferritin and CRP levels were significantly higher in COVID-19–positive ICU patients than in COVID-19–negative ICU patients (2,359.22 vs 549.70, P = .0040; and 205.05 vs 100.40, P = .0002, respectively), supporting the finding of a robust inflammatory response in COVID-19 infection. While WBCs were elevated in both groups, COVID-19–positive patients had significantly lower WBC values than COVID-19–negative patients (9.96 vs 13.39 × 109/L , P = .0130). Hemoglobin values were significantly higher in COVID-19–positive ICU patients than in COVID-19–negative ICU patients (106.5 vs 90.0 g/L, P = .0075), suggesting more underlying comorbidities and prolonged disease state in the latter group. Univariate analysis of the WBC morphologic features demonstrated that most of the parameters were similar in the ICU patients with and without COVID-19. However, the ICU COVID-19–positive patients continued to show significantly less monocytes with abnormal vacuolization compared with the ICU COVID-19–negative patients by both univariate analysis (OR, 0.29, P = .0270) and multivariate analysis (Table 7; OR of 0.10 for grade 1 vacuolization, P = .0035, and OR of 0.17 for grade 3 vacuolization, P = .0146), a finding similar to the non-ICU comparison. In the multivariate analysis, the presence of atypical lymphocytes, especially plasmacytoid forms, was more predictive of the COVID-19 infection (OR, 5.52, P = .0392).

Exploratory Analysis of Morphology-associated Research Parameters in COVID-19–Positive and COVID-19–Negative Patients

Striking WBC morphology findings and differences between COVID-19–positive and COVID-19–negative patients, as well as COVID-19–positive ICU and non-ICU patients, prompted us to explore WBC research morphology-associated parameters that are measured (but not reported) on Sysmex XN-9000 hematology analyzers with all routine CBC on the WBC differential channel. These parameters are reported as a mean number of coordinates of ~30,000 cells distributed across x, y, and z axes and, to the best of our understanding, are based on the cell complexity (NE-SSC, LY-X, MO-X), size (NE-FSC, LY-Y, MO-Y), and level of maturity (NE-SFL, LY-Z, MO-Z). Table 8 summarizes the data for research morphologic parameters in neutrophils, lymphocytes, and monocytes between three groups of patients: COVID-19–positive ICU and non-ICU and COVID-19–negative ICU patients. Interestingly, despite striking differences in neutrophil, monocyte, and lymphocyte morphology between COVID-19–positive ICU and non-ICU patients observed on peripheral blood smears, all research morphology parameters showed similar values between these two groups, indicating the importance of morphologic review by an experienced professional. On the other hand, all neutrophil morphology parameters (NE-SSC, NE-FSC, and NE-SFL), as well as LY-Y and MO-Y, were significantly different between COVID-19–positive ICU and COVID-19–negative ICU patients, identifying a potential role of these research parameters beyond the manual morphologic review. Due to the lack of experience with these morphology research parameters, it is somewhat difficult to know the significance of these observed differences. However, our data suggest that neutrophils in COVID-19–positive ICU patients are less granular and smaller than neutrophils in COVID-19–negative ICU patients (NE-SSC, 150.99 vs 154.41, P = .0142; NE-FSC, 83.79 vs 86.82, P = .0051). Lymphocytes appear to contain lower RNA content, as suggested by the LY-Y parameter in COVID-19–positive ICU patients (LY-Y, 71.11 vs 74.22, P = .0367), while monocytes demonstrated significantly higher fluorescence light intensity consistent with higher RNA content (124.09 vs 115.15, P = .0227), suggesting that the monocytes might be more transcriptionally active. These exploratory data need to be confirmed by more extensive multi-institutional analyses, but they clearly indicate the added value of the research parameters to the conventional CBC parameters and morphologic analysis of peripheral blood smear.