Combined Indexes of Serum NLR, CRP, LDH, IL-6 and D-Dimer Levels are a Biomarker of Disease Severity and Prognostic Predictor in Patients with Coronavirus Disease 2019

Background: Coronavirus disease-2019 (COVID-19) has become a worldwide emergency and has had a severe impact on human health. Inammatory factors have the potential to either enhance the eciency of host immune responses or damage the host organs with immune overreaction in COVID-19. Therefore, there is an urgent need to investigate the functions of inammatory factors and serum markers that participate in disease progression. Methods: In total, 54 COVID-19 patients were enrolled in this study. Disease severity was evaluated by clinical evaluation, laboratory tests, and computed tomography (CT) scans. Data were collected at: admission, 3–5 days after admission, when severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA detection became negative, and composite endpoint. Results: We found that the positive rate in sputum was three times higher than that in throat swabs. Higher levels of C-reactive protein (CRP), lactate dehydrogenase (LDH), D-dimer (D-D), interleukin-6 (IL-6) and neutrophil-to-lymphocyte (NLR) or lower lymphocyte counts suggested more severe disease, and the levels of cytokines and serum markers were intrinsically correlated with disease progression. When SARS-CoV-2 RNA detection became negative, the receiver operating characteristic (ROC) curve demonstrated that LDH had the highest sensitivity independently, and four indicators (NLR, CRP, LDH, and D-D) when combined had the highest sensitivity in distinguishing critically ill patients from mild ones. Conclusions: Monitoring dynamic changes in NLR, CRP, LDH, IL-6, and D-D levels, combined with CT imaging and viral RNA detection in sputum, could aid in severity evaluation and prognosis prediction and facilitate COVID-19 treatment. Linear correlations between the serum markers and blood cells Linear correlations between CRP and lymphocyte count, (B) LDH and lymphocyte counts, and (C) NLR and lymphocyte count were negatively correlated, while (D) LDH and CRP, (E) D-D and CRP, (F) CRP and IL-6, (G) NLR and CRP, neutrophil count and CRP, (I) NLR and IL-6, (J) procalcitonin and LDH, (K) NLR and LDH, (L) neutrophil count and NLR, (M) neutrophil count and D-D, (N) NLR and procalcitonin were positively correlated. CRP: C-reactive LDH: lactate dehydrogenase;

recovered from COVID-19 pneumonia, the host immune interactions undergo many phases such as incubation, syndromic, and recovery periods, in which the virus initiates replication, reaches a peak at 5-6 days after symptom onset, and then gradually decreases, respectively [6]. Correspondingly, the host immune system initiates a cytokine storm or the release of a speci c serum protein that is accompanied by disease progression, and a tide of cytokine storms (e.g., interleukin [IL]-6) may result in MODS and fatality [7]. The cytokine storm is considered the top reason for death among COVID-19 MODS patients [8,9]. Therefore, accurate monitoring of in ammatory factors plays an important role in the judgment of disease progression and the selection of treatment strategies. In ammatory factors can either enhance the e ciency of host immune responses or damage host organs with immune overreaction in COVID-19 [10]. Because the functions of in ammatory factors and serum markers that participate in disease progression are controversial, therefore warranting urgent exploration, this study aimed to determine the following aspects: (1) pro ling the trends of in ammatory factors and serum markers between mild and severe cases and (2) assessing the speci city and sensitivity of COVID-19-associated in ammatory markers and their joint roles in severity evaluation that may further guide clinical treatment or prognosis prediction.

Study design and participants
We retrospectively reviewed patient database and focused on the changing trends in cytokines and serum markers and their associations with the severity and prognosis of COVID-19. A total of 54 patients with a COVID-19 diagnosis were hospitalized in the First A liated Hospital of Bengbu Medical College from January 2020 to March 2020. Hospitalization duration was longer than two weeks for all patients, and each patient underwent severity assessment during disease progression, including clinical evaluation, laboratory tests, and computed tomography (CT) scans. Data were collected at: admission, 3-5 days after admission, when viral RNA detection became negative, and composite endpoint. The study was approved by the Ethics Committee of the First A liated Hospital of Bengbu Medical College (approval no. 2020KY067).

Laboratory con rmation and treatment
We collected sputum and throat swab specimens from all patients at admission and used RT-PCR for the detection of RNA of SARS-CoV-2 (tested on an ABI 7500 system, USA). Viral RNAs were extracted using a commercial kit speci c for SARS-CoV-2 (Da An Gene Co., Ltd., Guangzhou, China). The specimens were considered positive if the cycle of threshold (Ct) value of the ORF-1ab and the N gene was not higher than 40 and negative if the Ct value was undetermined. Specimens with a Ct value between 40 and 42 of double genes or single gene were repeated and considered positive if the repeat results were the same as the initial result. If the repeat Ct values were undetermined, they were considered negative. These detections were started at the admission time point and were repeated every 24 h. Speci cally, laboratory tests included routine blood tests (SYSMEX, XE-5000), which revealed the whole content of blood cells (e.g., red and white blood cell quantity and ratio, platelet [PLT] quantity, and neutrophil-to-lymphocyte ratio [NLR]) and serum biochemistry tests (measured using cobas 8000, Roche, Switzerland) (e.g., C-reactive protein [CRP] and lactate dehydrogenase [LDH]). The coagulation function (e.g., D-dimer [D-D]) was measured using CS5100 SYSMEX, and procalcitonin (PCT) was measured with a uorescence immunochromatographic system (Wondfo,  and tests for other respiratory pathogens were performed. All the patients were treated according to the Guidelines of the COVID-19 Diagnosis and Treatment (GCDT), issued by the National Health Committee of China.

Criteria of clinical assessment
According to GCDT, we classi ed COVID-19 into three clinical subtypes: mild, moderate, and severe. Patients that just had slight clinical symptoms without radiological changes are classi ed as mild.
Patients that had fever, respiratory distress, and a signs of pneumonia after CT image are classi ed as moderate. Patients that had any of the following are classi ed as severe: (1) respiratory rate > 30 times/min; (2) Sp O2 ≤ 93%; (3) Pa O2 / Fi O2 ≤300 mmHg; or (4) CT scan showing pulmonary lesions developed quickly within 1-2 days. Patients that required mechanical ventilation because of respiratory failure, with signs of septic shock or even multiple organ failure are critically severe cases, which are also include in the severe group.
CT imaging ndings, as indicators of disease severity, were classi ed into the following four types: (1) healthy type, which did not exhibit alterations on pulmonary imaging; (2) mild type, which manifested ground-glass opacities and consolidation as well as thin and small subpleural patches in either single or bilateral lobes; (3) progressive type, which showed large lesions and multiple lung lobes that were involved in the bilateral lungs, accompanied by bronchial retraction, bronchiectasis, and interlobular pleural thickening; and (4) severe type, in which the bilateral lungs exhibited diffuse lesions with uneven distribution of density and large areas of ground-glass opacities. Large lung lesions resulted in a "white lung," with or without thickened interlobular pleura, bilateral pleura, and pleural effusion. Speci cally, CT imaging was critical dependence for clinical severity assessment (Supplemental Fig. 1). In this study, healthy, mild, progressive, and severe types of CT imaging were scored as 0, 1, 2, and 3, respectively.
Based on the clinical progression, the outcomes of COVID-19 were classi ed as 4 types: fully recovered, improved, exacerbation, and death.

Statistical analysis
We used SPSS (IBM SPSS software, Chicago, IL) and GraphPad Prism 5 (GraphPad Software, San Diego, CA) for statistical analysis. Since the data presented signi cant variance, we also used the two-tailed unpaired Student's t test to evaluate the differences between two groups, and the chi-square test for multiple group comparisons.
We used the receiver operating characteristic (ROC) curve to assess the sensitivity and speci city of disease-associated cytokine factors and serum markers, in which a more substantial area under the ROC curve (AUC) indicated a higher accuracy. Since these markers can re ect disease severity, we used the ROC curve to evaluate the independent or joint sensitivity of the markers for disease progression. The judgment of disease progression was based on clinical assessment combined with CT imaging evaluation, regarding the mild/moderate type as negative (score: 0), whereas the severe/critically severe type was positive (score: 1). Hospitalization duration was more than two weeks, and the ROC curve calculations were repeated at three time points each week. In addition, we also adopted a linear correlation model to analyze the correlations between these serum markers. P < 0.05 was considered statistically signi cant. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, no signi cant difference.

Demographic characteristics
From January 2020 to March 2020, a total of 54 patients were hospitalized in the First A liated Hospital of Bengbu Medical College. In this patient cohort, there were 23 mild cases (42.6%), 22 moderate cases (40.7%), and 9 severe/critically severe cases (16.7%) according to the initial evaluation of clinical severity. In addition, age and sex distributions were as follows: 31 younger patients (57.3%, y < 60) and 23 older patients (42.7%, y ≥ 60); 22 female patients (40.7%) and 32 male patients (59.3%). Their demographic characteristics are shown in Table 1. Clinical classi cations and their associated laboratory test and CT imaging results In the total patient cohort, we tested the positivity of SARS-CoV-2 RNA at the rst, second, and third weeks after hospitalization and strati ed the positivity ratio into the mild, moderate, and severe groups. In this strati cation, the mild group had fewer positivity days than the moderate and severe groups, but the comparison did not reach signi cance (Table 2). In addition, sputum had three times the positivity ratio than throat swabs, which were sampled and detected at the same time 68 times. This result suggests that the sputum test was more accurate and reliable ( Fig. 1A-B). Curiously, the Ct value of the COVID-19 ORF1ab gene (second and third weeks) and the N gene (second week) in the sputum between the two groups indicated that the moderate and severe groups had lower virus replication than the mild group (Table 2). In addition, the laboratory ndings of blood cells and serum markers indicated that disease severity was negatively correlated with the counts of lymphocytes and monocytes and albumin levels but positively correlated with the levels of D-dimer, alanine aminotransferase (ALT), aspartate aminotransferase (AST), blood urea nitrogen, creatine kinase (CK), LDH, total bilirubin, IL-6, glucose and CRP (p < 0.05; Table 3).
Notably, the reduction of lymphocyte or monocyte counts indicated the potential immune cell exhaustion that represented disease severity. Dynamic pro ling of blood cells and serum markers COVID-19 progression is a dynamic process wherein lymphocytes and the levels of serum cytokines and markers change dynamically. Therefore, we evaluated the typical indexes involved in this progression as follows: lymphocyte count, neutrophil count, white blood cell count, CRP, LDH, D-D, and NLR. In those indexes, we found that lymphocyte count reduction and the severe group had signi cantly lower lymphocyte counts than the moderate and mild groups, and the lymphocyte quantity gradually recovered in the following two weeks and reached a healthy level in the third week ( Fig. 2A). Serum levels of CRP, LDH, D-D, and NLR were increased but gradually decreased to the normal level with disease recovery ( Fig. 2B-E). The CRP level in the severe group was high in the initial ve days, sharply reduced on days 6-9, then interstitially rebounded to a high level on days 10-13, and nally reached the normal level on days 15-21 (Fig. 2B). The LDH level showed a steady declining trend in the severe and moderate groups, while it was consistently low in the mild group (Fig. 2C). Furthermore, the D-D level and NLR were high in the severe group; they gradually declined on days 1-10 and reached a steady level in the following 11-23 days; these indexes remained steady and relatively low in the mild and moderate groups (Fig. 2D, E). The neutrophil count and white blood cell count were not steady and did not represent signi cant changing trends among the three subgroups (Fig. 2F, G).

Linear correlations between serum markers and blood cells
The linear correlation model showed that the serum markers and lymphocyte count had intrinsic correlations with disease progression. In correlation analysis, CRP and lymphocyte count (L), LDH and L, and NLR and L were negatively correlated (Fig. 3A-C), while the remaining markers (LDH and CRP, D-D and CRP, CRP and IL-6, NLR and CRP, neutrophil count and CRP, NLR and IL-6, procalcitonin and LDH, NLR and LDH, neutrophil count and NLR, neutrophil count and D-D, and NLR and procalcitonin) were positively correlated ( Fig. 3D-N). These signi cant correlations indicated that the combined indexes might be better indicators of disease severity.

Independent and joint sensitivities of COVID-19-associated markers
In this study, we found that either higher levels of CRP, LDH, D-D, and NLR or lower lymphocyte counts suggested more severe disease. Therefore, we used the ROC curve to calculate their sensitivity in detecting COVID-19 progression, regarding the mild/moderate type as negative and the severe/critical severe type as positive. The ROC curve showed that CRP had the highest independent sensitivity in  (Fig. 4D).
Although most patients had no syndromes or mild syndromes, fatal MODS can develop rapidly within a few days in severe cases [13]. Therefore, COVID-19 treatments essentially require practical evaluation of the disease condition and expected judgment of disease progression, and both the evaluation and judgment urgently require laboratory evidence for clinical guidance [14].
According to our paired detection results of sputum and throat swab samples of the same patients assessed 68 times, the accuracy of sputum detection is signi cantly higher than that of throat swabs. Upper respiratory tract samples are now widely used to detect viral RNA for the diagnosis of COVID-19, we must also remind that throat swabs are more suitable for broad-spectrum screening. For suspected cases, sputum and throat swab samples must be combined to improve the detection rate.
COVID-19 disease progression (incubation, syndromic, and recovery periods) involves virus-host interactions through which the host immune system recognizes and presents the virus-speci c antigen to effective T and B cells and thereby clears the virus. In this process, pyroptosis of infected epithelial cells can release many DAMPs and PAMPs thereby attracting lymphocyte in ltration [6]. Furthermore, extensive and severe infection sites could rapidly attract excessive lymphocyte in ltration within a short time, thereby reducing the quantity of blood lymphocytes [15]. Speci cally, this reduction was mainly attributed to the lymphocyte decrease and NLR increase in blood cell counts, and recovered patients usually had lymphocyte restoration. Moreover, the cell count of neutrophils increased 5-9 days after viral infection and then gradually decreased. This rise and fall of neutrophil counts may be associated with bacterial infection that stimulates the bone marrow to produce neutrophils instantly, and bacterial infections usually occur one week after the onset of viral infections. Therefore, a continuous reduction in lymphocyte counts and increased NLR indicate a worsening trend of disease progression, and an increased number of neutrophils suggests potential bacterial infection in COVID-19.
In this study, we also found that in ammatory cytokines and serum markers were correlated with COVID-19 tissue damage and lymphocyte counts. LDH is a cytoplasmic glycolytic enzyme expressed in almost every tissue and could represent the extent of tissue damage in COVID- 19 [16, 17], in which severe pneumonia has a high level of LDH [18] and associated DAMPs and PAMPs. Alveolar macrophages can recognize the DAMPs and PAMPs released by the pyroptosis of endothelial cells, thereby initiating cytokine secretion (e.g., IL-1β, IL-18, and TNF-α). The IL-1β, TNF-α, and Toll-like receptor signaling pathways can activate innate immune cells and effective T cells to produce IL-6 [19], which circulates to the liver and induces an extensive range of acute-phase proteins, such as CRP, serum amyloid A (SAA), haptoglobin, brinogen, and α1-antichymotrypsin [20]. In addition, IL-6 can promote the nal maturation of B cells into antigen-speci c antibody-producing plasma cells [21]. Therefore, excessive in ammation [22], which occurs as a high level of LDH, may result in macrophage pyroptosis and lymphocyte exhaustion [23], and a large amount of IL-6 is produced in this process. Increased IL-6 expression [24] was correlated with high levels of CRP, SAA, and D-D ( brinogen degradation product) and lymphocyte reduction (decreased lymphocyte count and increased NLR). It is di cult to de ne which one is the earliest and decisive factor, but our data demonstrate that CRP, LDH, and D-dimer levels are signi cantly higher in severe patients than in mild patients in the rst two weeks. These indicators may lead to the formation of the cytokine storm in severe patients and COVID-19 exacerbation.
The in ammatory cytokines and serum markers analyzed in this study had individual speci city and may be used to evaluate speci c disease progression time points, which may help better understand disease progression [25]. However, this study contained only a limited sample size, which is a limitation. Therefore, studies with more samples are warranted for further validation.

Conclusions
In summary, our study validated the changing trends of lymphocytes that correlated with in ammatory cytokines and serum markers, in which decreased lymphocytes were correlated with increased CRP, LDH, and NLR. Therefore, when SARS-CoV-2 RNA detection became negative, LDH independently and