Eosinopenia is associated with adverse outcomes after COVID-19 infection: A perspective from Japan

. A limitation of this assessment is that a relatively small sample size of patients were enrolled from a single centre. However, this is the first prospective report from Japan showing that Eos = 0 is a risk factor for ICU admission and exacerbation of COVID-19, including transfer from the general ward to the ICU. In conclusion

To the Editors: Age, comorbidities (chronic obstructive pulmonary disease [COPD], heart disease, diabetes mellitus) and laboratory examinations (C-reactive protein [CRP], lactate dehydrogenase, lymphocyte count) are recognized as consistent predictors of severe disease and poor prognosis in coronavirus disease 2019 (COVID-19). 1 There is also interest in markers of more severe disease in specific populations or settings.
Several recent studies reported that eosinopenia (eosinophil count [Eos] = 0) is associated with severe pneumonia and poor prognosis in COVID-19. 2,3 However, Glickman et al. reported that race is a modulating factor in the relationship between COVID-19 prognosis and eosinopenia, and that eosinopenia was not a risk factor for COVID-19 mortality in Asian patients. 4 However, the proportion of Japanese people with the HLA-A*11:01 allele, which has a protective effect against COVID-19, was reported to be lower than in other Asian (Chinese and Indian) populations. 5 Therefore, eosinopenia may affect severity and mortality in Japanese COVID-19 patients. Hence, we sought to clarify the association between eosinopenia and disease severity and mortality of COVID-19 in Japanese people.
Since October 2010, we have been conducting a prospective cohort study enrolling consecutive community-onset pneumonia patients hospitalized at a 1166-bed tertiary hospital in Okayama prefecture. We now share our observations of a subset of these patients who were hospitalized with COVID-19 pneumonia from the period March 2020 to August 2021. Pneumonia was clinically diagnosed based on American guidelines. 6 Exclusion criteria were age <15 years, hospital-acquired pneumonia and steroid therapy for COVID-19 before admission. All patients gave their informed consent for participation by the opt-out method.
Patients' clinical characteristics, including COVID-19 severity on admission and its maximum severity during hospitalization, treatment for COVID-19 and outcomes were investigated. COVID-19 was diagnosed based on positive results by either PCR or antigen testing for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). COVID-19 severity was assessed using the Japanese criteria 7 and the Ordinal Scale (OS). 8 Treatment for COVID-19 was at the attending physician's discretion based on the prevailing Japanese Guideline for the Treatment of Coronavirus Infections at that time. 7 The criteria for intensive care unit (ICU) admission were as previously reported: unstable oxygenation despite oxygen administration at a fraction of inspired oxygen of 70% with a high-flow nasal cannula, unstable circulatory state or at the discretion of the attending physician. 9 ICU admission during hospitalization was evaluated, along with 30-day mortality and exacerbation of COVID-19 severity scores. Data are expressed as medians or interquartile range for continuous variables and raw numbers or percentages for categorical variables. Categorical variables were analysed using Fisher's exact test, and continuous variables were analysed using the non-parametric Mann-Whitney Utest. To assess risk factors for ICU admission, 30-day mortality and exacerbation of severity scores in COVID-19 patients, univariate and multivariate logistic regression analyses were performed for 10 variables (age, COPD, diabetes mellitus, chronic heart disease, hypertension, CRP, albumin, lactate dehydrogenase, lymphocyte count and Eos = 0) based on previous studies. 1-3 All significance tests were twotailed, with p < 0.05 considered significant. Analyses were performed using the R statistical software package (version 3.0.3, Vienna, Austria).
Results are now presented from 125 patients (89 males (71.2%), median age 69 [58-77] years), 78 (62.4%) with Eos = 0. High-flow nasal cannulas were used in 27 (21.6%), and invasive positive-pressure ventilation in 29 (23.2%) patients; 15 (12.0%) patients died within 30-days after admission. Forty-two patients (33.6%) were transferred to the ICU after admission, 24 (57.1%) <24 h after admission and 18 (42.9%) ≥24 h after admission. There were no significant differences in age (69 vs. 69 years, p = 0.28), sex (male 66.3% vs. 81.0%, p = 0.10), comorbidities and vital signs between general ward and ICU admission groups. CRP (9.43 vs. 6.22 mg/dL, p < 0.001), lactate dehydrogenase (441 vs. 313 U/L, p < 0.001), blood urea nitrogen (23 vs. 15 mg/dL, p = 0.002), white blood cell count (7.1 vs. 5.0 Â 10 3 /μL, p < 0.001) and the rate of Eos = 0 (81% vs. 53%, p = 0.003) were significantly higher in the ICU admission group, whereas albumin (2.9 vs. 3.3 g/dL, p < 0.001) was significantly lower in the ICU admission group. Regarding admission COVID-19 severity scores, there were more severe patients in the ICU admission group than in the general ward group using both Japanese criteria and OS (moderate I 1 [ Cumulative ICU admission curves of the eosinophil count = 0 and eosinophil count >0 groups. The number of ICU admission patients with eosinophil count = 0 is significantly higher than those with eosinophil count >0 ( p = 0.004). Regarding ICU admission beyond 24 h after hospital admission, there were 18 patients with eosinophil count = 0 who were admitted to the ICU more than 24 h after hospitalization, whereas none of the patients with eosinophil count >0 required ICU admission more than 24 h after hospitalization. HR, hazard ratio. (B,C): Severity of COVID-19 on admission and at its worst during hospitalization in patients in the eosinophil count = 0 and eosinophil count >0 groups. (B) Japanese criteria (COVID-19 severity was defined according to the Japanese criteria as follows 6  ) was a significant predictor of increasing severity during hospitalization based on the Japanese criteria, but not by OS (p = 0.09). Figure 1A shows cumulative ICU admission curves in the Eos = 0 and Eos > 0 groups, and Figure 1B,C shows changes in severity from admission to maximum severity in each patient in the Eos = 0 and Eos > 0 groups. From our observations, Eos = 0 is a risk factor for ICU admission and COVID-19 exacerbation during hospitalization. Additionally, ICU admission ≥24 h after hospitalization was only seen in patients with Eos = 0 at admission. Older age, presence of diabetes mellitus, high lactate dehydrogenase and low lymphocyte count were predictors of 30-day mortality, but not Eos = 0.
Previous studies reported that eosinopenia is a risk factor for severe pneumonia and poor prognosis in COVID-19. 2, 3 Xie and colleagues indicated a significantly higher rate of Eos = 0 in the severe than the non-severe COVID-19 group (100% vs. 37.6%, p = 0.001). 2 Using multiple logistic regression analysis, Huang et al. also showed that eosinopenia (Eos < 20/μL on admission) was a significant risk factor for transfer to the ICU in elderly patients (≥60 years). Furthermore, the number of patients transferred to the ICU was higher in those with Eos = 0 than Eos > 0 (60.9% vs. 12.7%). 3 Our data support these findings not only in elderly patients, but in a wide range of age groups (≥15 years). Based on Huang's 3 and our results, we recommend closely monitoring COVID-19 patients with moderate level II disease according to Japanese criteria for determining whether patients with Eos = 0 require early transfer to the ICU.
Eosinophils reportedly have an antiviral effect, including rapid capture and inactivation of the virus and viral load reduction. 10 In mouse models, eosinophils showed defensive effects against influenza and respiratory syncytial (RS) viruses. 11 Hence, eosinopenia might be associated with more severe COVID-19 disease, as with influenza and RS viral infection.
Regarding 30-day mortality, Eos = 0 was not a significant predictor. There are several possible reasons for this. First, we did not analyse serial Eos values during hospitalization. Ferastraoaru et al. reported that patients with Eos ≥ 150/μL during hospitalization had a better prognosis than COVID-19 patients with Eos < 150/μL. 12 Chen et al. showed significantly lower mortality in patients with an increase in Eos after admission than in those with continuous eosinopenia, even with severe or critical disease. 13 Therefore, not only Eos on admission, but also serial Eos values might be significant predictors of a poor prognosis in COVID-19. Second, the prognosis of COVID-19 in cases of eosinopenia varies by race, and eosinopenia was not identified as a consistent risk factor for mortality in Asian individuals. 4 Finally, although multivariate analysis evaluated 10 variables previously identified as significant prognostic factors, other variables which may include unknown factors could affect COVID-19 prognosis.
A limitation of this assessment is that a relatively small sample size of patients were enrolled from a single centre. However, this is the first prospective report from Japan showing that Eos = 0 is a risk factor for ICU admission and exacerbation of COVID-19, including transfer from the general ward to the ICU.
In conclusion, Eos = 0 is a risk factor for ICU admission and disease exacerbation, including for transfer to the ICU, in Japanese COVID-19 patients. COVID-19 patients with Eos = 0 should be managed carefully, since they might require ICU admission.

HUMAN ETHICS APPROVAL DECLARATION
This study was approved by the institutional review board of Kurashiki Central Hospital (approval number 3991). The study was performed in accordance with the ethical standards established in the 1964 Declaration of Helsinki and its later amendments. Based on the Ethical Guidelines for Medical and Health Research Involving Human Subjects of the Ministry of Health, Labour and Welfare in Japan, information about the study was posted on a publicly available website to inform potential research subjects and the general public about the research. All patients gave their informed consent to participate in this study using the opt-out method. Clinical