key: cord-0868859-5i9b5q0d
authors: Tan, Li; Kang, Xia; Ji, Xinran; Li, Gaoming; Wang, Qi; Li, Yongsheng; Wang, Qiongshu; Miao, Hongming
title: Validation of predictors of disease severity and outcomes in COVID-19 patients: a descriptive and retrospective study
date: 2020-05-19
journal: Med
DOI: 10.1016/j.medj.2020.05.002
sha: 91d6ce2aebccc482d37db38558d66164ba1b5ba4
doc_id: 868859
cord_uid: 5i9b5q0d

Abstract Background The severity and outcome of COVID-19 cases has been associated with percentage of circulating lymphocytes (LYM%), levels of C-reactive protein (CRP), interleukin 6 (IL-6), procalcitonin (PCT), lactic acid (LA) and viral load (ORF1ab Ct). However, the predictive power of each of these indicators in disease classification and prognosis remains largely unclear. Methods we retrospectively collected information on the above parameters in 142 patients with COVID-19, stratifying them by survival or disease severity. Findings CRP, PCT, IL-6, LYM% and ORF1ab Ct were significantly altered between survivors and non-survivors. LYM%, CRP and IL-6 were the most sensitive and reliable factors in distinguishing between survivors and non-survivors. These indicators were significantly different between critically ill and severe/moderate patients. Only LYM% levels were significantly different between severe and moderate types. Among all the investigated indicators, LYM% was the most sensitive and reliable in discriminating between critically ill, severe and moderate types, and between survivors and non-survivors. Conclusions CRP, PCT, IL-6, LYM% and ORF1ab Ct, but not LA, could predict prognosis and guide classification of COVID-19 patients. LYM% was the most sensitive and reliable predictor for disease typing and prognosis. We recommend that LYM% be further investigated in the management of COVID-19.

Coronavirus disease 2019 (COVID- 19) is an acute respiratory infective disease caused by the severe acute respiratory syndrome coronavirus 2 virus. [1] [2] The COVID-19 pandemic has been spreading worldwide rapidly since March 2020. As of late of April 2020, it was reported that more than 1,800,000 individuals had been diagnosed and this disease caused over 200,000 deaths. The rapid increase in cases has led to heavy burdens on public healthcare resources and medical facilities. [3] [4] According to the New Coronavirus Pneumonia Diagnosis Program (5th edition), published by the National Health Commission of China, the disease severity of patients with COVID-19 can be divided into 4 categories, i.e. mild, moderate, severe and critical. 5 Current experience reveals that the majority of infected individuals (approx. over 80%) are not severely affected and can recover without medical intervention, whereas a small number of cases need to be carefully treated and hospitalized. [6] [7] [8] [9] The mortality rate for severe cases, particularly those who are critically ill, is quite high. It is therefore critical to identify reliable predictors for disease severity to improve outcomes and conserve medical resources.

Recent studies have shown that a variety of risk factors are associated with the prognosis of COVID-19. Patients with cardiovascular diseases, diabetes and other comorbidities are often subject to acute respiratory distress syndrome, shock, multi-organ failure, cytokine storm and other serious 6 complications in COVID-19. 8,10-12 These patients commonly have a poor prognosis. However, patients with moderate COVID-19 and no underlying diseases can also develop the complications described above and progress to the severe or critically ill types. 11, 13 According to the latest guidelines published by the National Health Commission of China, the diagnosis of severe and critically ill patients must rely on complex procedures such as imaging tests and blood gas analysis. In a pandemic situation, it is difficult to perform these diagnostic examinations on all patients. It is therefore particularly important to identify more easily detectable and earlier predictors to achieve extensive screening of patients and optimize the allocation of medical resources.

Recent studies from different cohorts of patients have identified several factors, including viral load, [14] [15] lymphocytes percentage, [16] [17] C-reactive protein (CRP), [18] [19] interleukin-6 (IL-6), 17,20 procalcitonin (PCT) [21] [22] and lactic acid (LA) 23 as warning indicators of prognosis in COVID-19 patients. However, it is still unclear which of these factors are the most sensitive and reliable indicators for predicting the prognosis of COVID-19 in the early stage. Besides, early indicators for disease classification are also urgently needed. In the present study, based on the clinical information of 142 patients with COVID-19, we compare and validate the predictive power of several reported risk factors for disease classification and prognosis in COVID-19 in a descriptive manner. 7 To investigate the mortality-associated risk factors, we initially collected clinical information from admission to discharge (or death) on a cohort of COVID-19 patients, dividing them into survivors (n=117) and non-survivors (n=15). Firstly, the baseline information of these patients was analyzed. We found that older patients, especially those over 90 years of age, had a higher risk of mortality (Table S1) . Female patients had a higher recovery rate and lower mortality rate than males (Table S1 ). The non-survivor group had a higher ratio of comorbidity, such as cardiovascular disease and diabetes, than the survivors (Table S2 ). In addition, relative to the cured patients, the non-survivors also had an increased incidence of complications like acute respiratory disease syndrome, septic shock, hemorrhagic shock, gastrointestinal bleeding, organ dysfunction and multiple organ failure (Table S3) .

Secondly, we retrospectively analyzed serum levels of CRP, PCT, IL-6 and LA, percentage of lymphocytes (LYM%) as well as Ct values of viral tests in the survivors and non-survivors. After disease onset, the levels of all the aforementioned indicators showed little change in the survivors, while they were notably altered in the non-survivors ( Fig. 1A-F) . The levels of CRP and IL-6 in peripheral blood of non-survivors increased quickly after symptom onset and further rose to a significantly higher level as compared to survivors ( Fig. 1A and 1C) . A similar trend was observed for the PCT level, although it only began to increase in the late phase of disease course (Fig. 1B) . In contrast, after symptom onset, the LYM% in non-survivors decreased quickly 8 and remained at a significantly lower level as compared to survivors (Fig. 1D) .

Unexpectedly, there was no obvious difference in blood LA levels between survivor and non-survivor groups (Fig. 1E ). In addition, non-survivors had obviously higher levels of viral load indicated by ORF1ab Ct values as compared to survivors (Fig. 1F ). Taken together, indicators CRP, PCT, IL-6, LYM% and viral load could predict the mortality risk to varying extents. The strong predictive power of LYM% and IL-6 in mortality was also confirmed in a time-dependent ROC analysis (Fig. S1 ).

To descriptively study the potential of these measurements as predictors, we defined two assessment indexes. Initial day with difference (IDD) after disease onset was used to indicate the sensitivity of indicators in discriminating between two groups. Duration with difference (DD) in the disease course was Interestingly, in line with their better prognosis (Table S1 ), female patients also had higher levels of LYM% than males in the disease course according to a mixed model analysis (Fig. S2) . These findings indicate that the immune status might determine the prognosis of COVID-19 patients.

To validate the roles of the aforementioned indicators in disease classification, we retrospectively reclassified our cohort of patients into critically ill (n=25), severe (n=21) and moderate (n=96). The critically ill patients were mainly male (76%) and older than the severe and moderate patients (76±16 vs 49±15 or 56±16) (Table S4 ). A mixed model analysis indicated that the critically ill and severe/moderate types of patients had significant differences in the levels of CRP while in hospital ( Fig. 2A) . According to analysis with IDD and DD, we observed that CRP could be used to discriminate between the critically ill and severe/moderate COVID-19 patients, with a better result in typing between critically ill and moderate patients (Fig. 2B ). However, CRP levels could not distinguish between severe and moderate types ( Fig. 2A, B) .

We next performed a mixed model analysis to determine whether PCT levels could discriminate between critically ill and severe/moderate patients (Fig. 3A) .

PCT was equally sensitive and reliable classifying between critically ill and severe/moderate cases (Fig. 3B ). However, PCT levels could not distinguish between severe and moderate types.

Likewise, we analyzed whether fluctuations in circulating IL-6 were significantly altered between critically ill and severe/moderate patients. Similar to CRP, we observed that IL-6 could be used to discriminate between critically ill and severe or moderate patients but not between patients with moderate and severe COVID-19 (Fig. 4A ). This indicated that IL-6 had a similar effect in typing between the critically ill and severe patients, and between the critically ill and moderate ones (Fig. 4B ).

patients LYM% levels were notably different between critically ill, severe and moderate groups according to a mixed model analysis (Fig. 5A ). Severe and critical patients had a rapid decrease of LYM% after disease onset. Severe patients had a subsequent restoration of LYM% 4 weeks after disease onset, whereas critical patients did not and in some cases had already succumbed to the disease by that time (Fig. 5A) . LYM% showed the earliest sensitivity and longest reliability in discriminating between groups (Fig. 5B) . These results indicated that LYM% could be a sensitive and reliable indicator to predict 11 COVID-19 severity.

LA levels were measured in blood gas assay, an analysis which is not commonly performed in moderately ill patients. We retrospectively investigated blood LA levels in the severe (n=21) and critically ill patients (n=25). However, there was no difference in the LA levels between these two groups of patients ( Fig. 6A, B) 

Unlike the indicators CRP, PCT, IL-6 or LYM%, ORF1ab Ct values were significantly altered only between critically ill and moderate groups (Fig. 7A ).

Our analysis also showed that ORF1ab Ct values could not distinguish between severe and moderate or critically ill patients (Fig. 7B ). We validated that the LYM% could be a sensitive and reliable predictor to distinguish between critically ill, severe and moderate COVID-19 patients.

Previous studies also supported the conclusion that lymphocyte count and function are closely related to the disease status of COVID-19. 16, 24 Patients with lowered immunity, including elderly or immunocompromised individuals, commonly presented a lower level of lymphocytes and a worse prognosis after infection with SARS-CoV-2. 25 According to the present descriptive study, we showed that LYM% had the earliest IDD and longest DD among all the included indicators in disease classification and prognosis prediction. Notably, we found that LYM% was the only indicator that could distinguish between the moderate and severe patients. These results indicated that LYM% was the most sensitive and reliable indicator for COVID-19.

Notably, as we described above, the serum levels of IL-6 and CRP presented with a zigzag curve during the entire course of disease in critically ill patients. We speculate that this obvious change may be caused by medical intervention. Interestingly, the changes of serum levels of these two markers in moderate and severe patients were much more stable and lower than those in critical patients. These results indicate that refractory inflammatory reaction can predict a poor outcome. Patients with this feature should be prioritized for treatment in the early stage. In clinical practice, the curve of inflammatory indicators should be drawn dynamically, similar to body temperature.

The higher levels of inflammatory indicators and the lower level of LYM% leads to another interesting question: whether the inflammatory cytokine storm 13 is due to impaired lymphocytes, such as T cells. As we know, some T cell subsets, i.e. Treg, are responsible for inflammatory regulation. [26] [27] The impaired function of Treg cells may contribute to the uncontrolled inflammation in critical patients. It should be noted that in the late stage of non-survivors, all the levels of proinflammatory indicators, LYM% and lactic acid displayed an acute increase (Fig. 1A-E) . This phenomenon indicated a significant dysregulation of inflammation, immunity and metabolism in these dying patients.

high levels of viral load in the critically ill patients. 15 However, it should be noted that there were no obvious differences in the duration of viral shedding between the severe and non-severe patients, which was consistent with a previous study. 8 Recently, we reported a moderate patient with long duration of viral shedding for 49 days. 28 This phenomenon was in line with evidence of asymptomatic infection reported in some populations. 29 

In the present study, based on the information obtained from 142 patients with COVID-19, we analyzed the predictive power of several reported indicators for disease severity and prognosis. We found that CRP, PCT, IL-6, LYM% and viral load could predict prognosis and guide classification of COVID-19

patients to different extents. The percentage of circulating lymphocytes was the most sensitive and reliable predictor for disease outcome and classification, 14 especially for the typing of severe and moderate cases. During the COVID-19 pandemic, precise classification and prognosis prediction are critical for saving insufficient medical resources, stratifying treatment and improving the survival rate of critically ill patients. We recommend that LYM% be used independently or in combination with other indicators in the management of COVID-19.

Firstly, this study is a retrospective study in a single center, and the conclusions obtained need to be further verified by other centers. Secondly, the sample size of the non-survivor group in this study is not large enough to exclude bias in the results of the analysis. Thirdly, this study only suggests the potential of a series of indicators in disease typing and prognosis, but their validation and implementation in the clinic need to be further explored. 

The authors declare no competing interests. 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Hongming Miao (hongmingmiao@sina.com).

This study did not generate new unique reagents.

This study did not generate or analyze datasets.

All cases were taken from the General Hospital of Central Theater Command 20 blood oxygen saturation ≤ 93% (at rest), PaO 2 / FiO 2 ≤ 300 mmHg, or pulmonary inflammation that progresses significantly within 24 to 48 hours> 50%, it was classified as severe; respiratory failure, shock, and organ failures that require intensive care were critically ill. Among them, mild patients were not admitted in this designated hospital.

The patient can be discharged if he/she simultaneously meets the following conditions recommended in the guidelines: 1) The body temperature returns to normal for more than 3 days. 2) Respiratory symptoms improve significantly. 3)

Pulmonary imaging shows significant improvement in acute exudative lesions.

swabs and other respiratory tract specimens (at least 24 hours apart).

Serum CRP levels were determined by the turbidimetry (Lifotronic PA900, Shenzhen, China). Serum PCT and IL-6 concentrations were measured by the electrochemiluminescence immunoassay (Roche COBAS e411, Mannheim, Germany). Routine blood tests including LYM% were performed by a hematology analyzer (Sysmex XN-9000, Kobe, Japan). LA levels were determined by a blood gas analyser (Radiometer ABL800, Copenhagen, Denmark).

The pharyngeal swab of patients were collected and the total RNA was extracted using the nucleic acid extraction kit (Tianlong, Xi'an, China). The 21 quantitative RT-PCR assay was performed using a SARS-CoV-2 nucleic acid detection kit (Chinese Center for Disease Control and Prevention recommended) according to the manufacturer's protocol (DAAN Gene, Guangzhou, China). Open reading frame 1ab (ORF1ab) was simultaneously amplified and tested during the quantitative RT-PCR assay. A cycle threshold value (Ct-value) less than 40 was defined as a positive test result, and a Ct-value of 40 or more was defined as a negative test.

In this study, the basic information, complete blood count, serum biochemical test, inflammatory indicators, viral load and disease outcome of all included patients were collected from admission to discharge or death. This study was approved by the Ethics Committee of the hospital. All subjects signed informed consent forms at admission to hospital.

In this study, GraphPad 8.01.244, SPSS 18.0 and R 3.6.3 were used for data mapping and statistics. Mann-whitney U test was used for comparison between two groups of continuous data, and kruskal-wallis H was used for comparison between multiple groups. Group comparisons for categorical variables were performed by using the Chi-square or Fisher exact probability methods. The laboratory examination indexes were expressed as the means ± s.e.ms. and were analyzed using a mixed model with repeated measure. If the difference between two groups was significant, a descriptive study would be 22 performed. Initial day with difference and duration with difference between two groups were used to respectively indicate the sensitivity and reliability of predictors for discrimination roughly. To calculate the hazard ratios (HR) of IL-6 and LYM%, the univariable Cox proportional hazard ratio regression model was performed. The sensitivity and specificity of prognostic indicators were evaluated by the time-dependent receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) was quantified by the timeROC package. All tests were bilateral, and P< 0.05 was considered statistically significant. For each parameter of all data presented, * indicates P<0.05, ** indicates P<0.01 and *** indicates P<0.001.

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

COVID-19-New Insights on a Rapidly Changing Epidemic

Italy in a Time of Emergency and Scarce Resources: The Need for Embedding Ethical Reflection in Social and Clinical Settings

The First 75 Days of Novel Coronavirus

CoV-2) Outbreak: Recent Advances, Prevention, and Treatment

Coronavirus Disease 2019 (COVID-19): A Perspective from China

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

COVID-19 patients' clinical characteristics, discharge rate, and fatality rate of meta-analysis

Korean Society of Infectious Diseases and Korea Centers for Disease Control and Prevention

Disease 2019 in the Republic of Korea from

Clinical features of COVID-19 in elderly patients: A comparison with young and middle-aged patients

Predictors of Mortality for Patients with COVID-19 Pneumonia Caused by SARS-CoV-2: A Prospective Cohort Study

Severe Outcomes Among Patients with Coronavirus Disease 2019 (COVID-19) -United States

SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients

Quantitative Detection and Viral Load Analysis of SARS-CoV-2 in Infected Patients

Complex Immune Dysregulation

Patients with Severe Respiratory Failure

C-reactive protein levels in the early stage of COVID-19

C-reactive protein correlates with computed tomographic findings and predicts severe COVID-19 early

Induction of pro-inflammatory cytokines (IL-1 and IL-6) and lung inflammation by Coronavirus-19 (COVI-19 or SARS-CoV-2): anti-inflammatory strategies

Procalcitonin in patients with severe coronavirus disease 2019 (COVID-19): A meta-analysis

Risk factors of critical & mortal COVID-19 cases: A systematic literature review and meta-analysis

Epidemiological and clinical features of COVID-19 patients with and without pneumonia in Beijing, China. medRxiv

Dysregulation of immune response in patients with COVID-19 in

COVID-19 Presents High Risk to Older Persons

The Pivotal Role of Regulatory T Cells in the Regulation of Innate Immune Cells

Control of inflammation by integration of environmental cues by regulatory T cells

A special case of COVID-19 with long duration of viral shedding for 49 days. medRxiv

CoV-2 Infections in Residents of a Long-Term Care Skilled Nursing Facility

We thank all the medical and scientific workers for the efforts in fighting against 

 27 2020. MMWR Morb Mortal Wkly Rep 69, 377-381.

(1) Inflammatory cytokines, lymphocytes and viral load predict prognosis in COVID-19 patients.(2) Inflammatory cytokines, lymphocytes and viral load indicate disease severity.(3) Lymphocytes and C-reactive protein distinguish between severe and moderate types.(4) Lymphocyte count is the most effective indicator of disease severity and prognosis.

In this descriptive and retrospective study, the authors demonstrate that viral load, lymphocyte percentage, C-reactive protein, interleukin-6 and procalcitonin can predict prognosis and guide classification of COVID-19 patients. Among those indicators, lymphocyte percentage is the most sensitive and reliable predictor for disease typing and prognosis.

There is an urgent need to identify factors that can predict outcome and disease severity in COVID-19 patients. In this study, researchers from the Third Military Medical University in Chongqing, China, retrospectively enrolled 142 COVID-19 patients and measured their percentage of circulating lymphocytes, as well as C-reactive protein, interleukin 6, procalcitonin, lactic acid and viral load, factors that have been associated to the severity of the disease. The authors find that the percentage of circulating lymphocytes is the most sensitive and reliable predictor of disease severity and outcome, and that levels of C-reactive protein and interleukin 6 can predict outcome. These findings may be useful in the treatment of COVID-19 patients and management of medical resources during the pandemic.