A Comprehensive Study on Prediction Reliability of The Severity of Computed Tomography Involvement in Patients with Covid-19

Purpose: Computed tomography (CT) emerges as a high–sensitivity tool in diagnosing SARS-CoV-2 virus on admission, even with the cases of negative reverse transcription polymerase chain reaction (RT-PCR). Moreover, CT plays a significant role in the evaluation of disease severity. In this study, we aimed to identify several parameters that could aid in evaluating the initial chest CT severity score (CT-SS). Methods: : A total of 348 RT-PCR positive patients were divided into three groups by evaluating the chest CT severity score (CT-SS) in detail. These three groups were defined as the CT-SS obtained 0-7, 8-15, 16-40 classified as mild, moderate and severe involvement, respectively. Patients with end-stage malignancy or immunodeficiency were excluded from the study. All CT images were evaluated by two chest radiologists, unaware of the clinical data. Results: The analysis of categorical variables show that the chest CT-SS tends to increase with higher CHA2DS2VASC risk score (RS) (p= 0.001), M-CHA2DS2VASC RS (p = 0.001), and CHADS2 RS (p = 0.003). Moreover, age, hypertension, cardiothoracic ratio, aortic diameter, white blood cell count, neutrophil counts, neutrophil-lymphocyte ratio, c-reactive protein, D-dimer, ferritin, fibrinogen, blood urea nitrogen, and lactate dehydrogenase were also found to be associated with higher chest CT-SS. A 5-variable multivariable linear regression model, consisting of more frequently used variables, suggests a higher CHA2DS2VASC RS as the only statistically significant variable predicting a higher severity score Conclusion: The CHA2DS2VASC RS may assist clinicians in predicting the CT-SS in COVID-19 patients, even if they are asymptomatic.


Introduction
The world is struggling with the contagious novel SARS-CoV-2 virus (COVID-19) since the rst case was identi ed in December 2019.
With the second wave, the COVID 19 pandemic continues to increase its impact across the world. Chest computed tomography (CT) is a prominent method in the diagnosis COVID-19, including the diagnosis in patients that are con rmed to have SARS-CoV-2 infection but with RT-reverse transcription polymerase chain reaction (RT-PCR) negative ndings (1) Besides, the widespread CT involvement is associated with the mortality and hospitalization in intensive care (2). The chest CT severity score (CT-SS) was described by Yang et al. (3), which is a relatively easily applicable, semi-quantitative method for assessing severity.
Recently, studies have shown that the CHA2DS2VASc RS can also be used as a mortality risk assessment scale in COVID-19 patients (5). Tertiles of CHA2DS2VASc RS, except stroke and female gender, have a mortality association with COVID-19 (6)(7)(8)(9). In contrast, male sex is associated with increased mortality (11). Therefore, several researchers have recently developed a modi ed (M) -CHA2DS2VASc RS that may predict mortality for use in COVID-19 patients (12). Moreover, mortality studies in COVID-19 have been associated with an increased risk of all-cause death during hospitalization of some demographic and laboratory parameters (such as chronic kidney disease, increased neutrophil-lymphocyte ratio, lactate dehydrogenase) (8)(9)(10)(11)(12)(13)(14)(15). Ground-glass opacity (GGO) and consolidation are early predictors of clinical deterioration in CT imaging. In asymptomatic cases may have pneumonia, and clinical deterioration may occur rapidly. (16) We think, even if the patient is clinically stable, CT plays a crucial role in decision-making for hospitalization. Thus, with the data detected in this study, we planned to use CT more effectively and accurately.

Study population
This retrospective study enrolled a total of 367 COVID-19 patients who admitted to our hospital between October 2020 and November 2020. Following inclusion criteria were used for the study: I) to have a con rmed diagnosis of COVID-19 by RT-PCR using both nasopharyngeal and throat swab samples, II) to be evaluated by CT at admission, III) age ≥ 18 years old. Malignancy, immunode ciency, pregnancy and absence of all medical records were determined as exclusion criteria for the study. A total 19 patients diagnosed with COVID-19 were excluded from the study for these reasons. All CT images were independently assessed by two chest radiologists with more than ve years of experience, blinded to the clinical data and laboratory indicator, and the demographic data of the patients were obtained from the National Health Database and the hospital database.
The study was conducted in accordance with the principles of the Helsinki Declaration and approved by the local Institutional Review Board.

Clinical and laboratory assessment
Demographics and medical records of the patients were reviewed and noted.
CHADS2 RS, CHA2DS2-VASc RS, and M-CHA2DS2-VASc RS were calculated in accordance with previous publications [17][18][19]. Also, the parameters that contribute to the scoring systems were analyzed individually. We additionally assessed levels of neutrophillymphocyte ratio (NLR), c-reactive protein (CRP), lactate dehydrogenase (LDH), cardiac troponin, and blood urea nitrogen (BUN) from the patients' records. Ischemic stroke was de ned as an episode of neurological dysfunction caused by focal cerebral, spinal, or retinal infarction [20]. These embolic events, the presence of clinical signs was con rmed by imaging modalities (magnetic resonance imaging with or without CT angiography). The clinical diagnosis of the stroke was made by a neurologist. Hypertension (HT) de ned as systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg measured in the supine position or use of antihypertensive medications [21], diabetes mellitus (DM) de ned as fasting blood glucose level ≥ 126 mg/dl or HbA1c level ≥ 6.5% or use of hypoglycemic medications, congestive heart failure de ned as recommended by the working group of the European Society of Cardiology [22].

CT image evaluation
Chest CT imaging was performed using a 64-detector CT scanner (Somatom Sensation 64 Eco; Siemens). All patients were examined in the supine position. CT images were then acquired during inspiratory and expiratory phases with or without contrast using. Our CT implication indications were; RT-PCR testing is not available, delayed RT-PCR testing; RT-PCR test is negative, but the patient with high clinical of suspicion of COVID-19; patients with high comorbidity burden who have a higher risk of disease progression; older age; physical exam with severe symptoms and sings; patients may have presentations that indicate COVID-19 complications such as pulmonary embolism, pulmonary brosis; cases are not responding to the treatment; case with acute clinical deterioration and patients have pre-existing chronic lung disease as WHO recommendations (23). We classi ed the CT data using the previously-mentioned the CT severity score CT-SS (3). The chest CT-SS uses the ground-glass opacity, interstitial opacity, and air trapping criteria. Air trapping was assessed with the inspiratory and expiratory scans. Chest CT-SS determines the 20 regions of the lung separately as point 0-1-2. Zero points describe no involvement; one point describes with less than 50% involvement; two points describe more than 50% involvement. It is worth mentioning that the maximum score for a patient is limited to 40 (3). Following the assignment of CT-SSs, the patients were divided into three CT-SS groups of 0-7, 8-15, and 16-40, classi ed as mild, moderate, and severe involvement, respectively ( Fig. 1).
We also evaluated the patients according to the Radiological Society of North America Expert (RSNA) consensus statement category after CT-SS (24).
In addition to these radiological evaluations, the diameter of the ascending aorta and the cardiothoracic ratio (CR) were determined from CT's in all patients.

Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 20.0 (IBM Corp. Armonk, NY, USA). Descriptive statistics were reported as mean ± standard deviation for continuous variables with normal distribution or median (25th-75th percentiles) values for continuous variables without normal distribution and as frequency with percentages for the categorical variables. The Shapiro-Wilk and Kolmogorov-Smirnov tests were used to test the normality of the distribution of continuous variables. Categorical variables were compared with Chi-square or Fisher exact tests as appropriate. Kruskal-Wallis and One-Way-ANOVA tests were used for comparisons of continuous variables as appropriate. When the variances were homogeneous in the posthoc tests, Tukey test was performed and Tamhane's T2 test was performed in a homogeneous condition. After Kruskal Wallis test, Kruskal Wallis 1-way ANOVA (k samples) test was performed for post-hoc tests. Friedman test was used when Repeated Measures ANOVA test was not provided. In 2x2 comparisons between categorical variables, Pearson's Chi-square test was used if the expected value (> 5), chi-square yates test was used if the expected value was (3)(4)(5), and Fisher's Exact test was used if the expected value (< 3). In comparisons greater than 2x2 between categorical variables, the Pearson Chi-square test when the expected value is (> 5), and the Fisher-Freeman-Halton test when the expected value is (< 5). Post-hoc tests after chi-square were done by the Bonferroni method. The signi cance level was accepted as p < 0.05 in all statistical analyses.

Results
The study cohort consisted of 348 patients (mean age: 66 (22-97) years; male:174) showing intermediate and typical involvement in the RSNA classi cation resulted in 148 patients (42.6%) with a score of 0-7 (mild), 104 patients (29.8%) with a score of 8-15 (moderate), and 96 patients (27.6%) with a score of 16-40 (severe). Demographic and clinical features as well as the laboratory parameters of the study group according to the CT-SS tertiles are shown in Table 1and 2. In categorical variables, the chest CT-SS tends to increase with higher CHA2DS2VASC RS (p = 0.001), M-CHA2DS2VASC RS (p = 0.001), and CHADS2 RS (p = 0.003). The datadriven tertiles of CHA (2) HT and vascular disease, which are tertiles of CHADVASC RS and M-CHA2DS2VASC RS, are associated with higher chest CT-SS (p < 0.001, Table 1). Also, old age, CR, aortic diameter, white blood cell (WBC) count, neutrophil count, NLR, CRP, D-dimer, ferritin, brinogen, CRP, BUN, LDH (p < 0.001, for all), and AST levels (p = 0.011) tend to increase progressively from a lower CT-SS to a higher CT-SS score. On the other hand, lymphocyte count tends to decrease from a lower CT-SS to a higher chest CT-SS tertile (p < 0.001, Table 2).

Discussion
The major nding of the current study is that the CHA2DS2VASC RS has a distinctive ability to predict chest CT-SS in patients with COVID-19. Previous COVID-19 radiology studies do not provide a thorough investigation of the association between the chest CT severity and patient's comorbidity conditions and laboratory ndings. Statistically signi cant ndings in this study can be well used as a guide to screen the patients and effectively identify the ones that may have severe CT involvement on admission in daily practice.
The COVID 19 is a disease that can progress with a severe immune response, as well as have a thrombus burden [25][26][27]. The mechanism of action of thrombosis has not been fully elucidated. It is thought that the development of coagulopathy adversely affects the prognosis of the patient. D dimer, platelet count, prothrombin time, brinogen level should be used to detect the coagulopathy [28]. Increased D-dimer and massive brin formation are associated with poor prognosis, prolonged mechanical ventilation, and death [29,30]. D-dimer levels can also be used to administrate anticoagulants for COVID -19 patients [31]. The CHA2DS2VASC RS predicts the thromboembolic originated ischemic stroke in patients with or without atrial brillation (AF) [5]. It also predicts the risk strati cation and death in patients with COVID-19 [6]. Recently, Cetinkal et al reported that M-CHA2DS2VASC RS score, derived from CHA2DS2VASC RS, well predicts the in-hospital mortality [12]. Our study indicated that the CT-SS score increases gradually with increasing CHA2DS2VASC RS and M-CHA2DS2VASC RS. Besides, analyses revealed that CHA2DS2VASC RS was more speci c than other risk scores.
An irregular immune response is one of the leading causes of death in patients. This immune response progresses with leukocytosis, neutrophilia, monocytosis and lymphopenia. Recently, several investigators reported that NLR can facilitate in categorizing the disease severity and progression in patients, thereby enabling us to make appropriate and informed clinical decisions [32]. Hence, we compared leukocytosis and NLR with the chest CT-SS. Our results suggest that the NLR and leukocytosis correlate with high chest CT-SS. The dysregulated immune response mentioned above may cause macrophage activation syndrome (MAS) in patients. The pathophysiology of MAS has not been fully identi ed. However, MAS has many different parameters revealed in various studies [33]; some of which are correlated with the high CT-SS, such as ferritin, LDH, CRP, AST. Therefore, in some studies, these parameters have been identi ed as a predictor of morbidity and mortality [13,33].
The relationship between COVID-19 poor outcomes and CR or aortic diameter has not yet been investigated. However, in studies performed with non-covid patients, ascending aorta dilation and increased CR are associated with increased mortality [27,31]. The current data suggests a statistically signi cant correlation between high CT-SS and ascending aortic diameter and CR elevation.
If the CT was not applicable, CR was obtained from the posterior-anterior lung lm and may provided convenience in clinical practice.
Old age and HT are also associated with increased violence and mortality in COVID-19 patients [6,10]. Coronary artery disease, one of the vascular diseases, shows that mortality increases in patients with COVID-19 [10]. In our study, patients with vascular diseases, including aortic atheroma plaques, carotid artery stenosis, coronary artery disease, and peripheral artery disease plaques, showed more severe CT involvement. Advanced age, vascular disease, and hypertension, which are elements of both scoring systems, predict an increasing chest CT-SS score. Moreover, CKD was previously reported to be associated with increased mortality in patients with COVID-19 [9]. However, in this study, no statistical relationship was found between CKD and chest CT-SS. Also, recently published studies have shown that a higher initial BUN indicate a higher in-hospital mortality rate. Similarly, in our result demostrated that the chest CT-SS gradually increases as the BUN gets higher, unlike creatinine.
Yang et al. designed the chest CT-SS for rapid and objective assessment in COVID-19 patients [3]. Colombi et al have recently demonstrated that the level of CT involvement shows clinical outcome and mortality [2]. However, there is no study investigating the parameters that predict the degree of involvement in CT using such a large sample and variable size.. In asymptomatic cases may have pneumonia, and clinical deterioration may occur rapidly (16). We think, even if the patient is clinically stable, CT plays a crucial role in decision-making for hospitalization. On the other hand, previous COVID-19 radiology studies do not provide a thorough investigation of the association between the chest CT severity and patient's comorbidity conditions, clinical scoring systems, and laboratory ndings. Statistically signi cant ndings in this study can be well used as a guide to screen the patients and effectively and accurately identify the ones that may have severe CT involvement on admission in daily practice. To the best of our knowledge, this is the rst comprehensive analysis of comorbidity conditions or scales, radiological and biochemical characteristics that could aid in evaluating the CT involvement severity.

Limitations of the study
The present study has several limitations. First of all, this was a retrospective study and included a relatively small patient population. Second, the uncertainty of the time interval between onset of symptoms and hospitalization. Lastly, this analysis does not contain information about mortality and long-term follow-up of patients.

Conclusion
In conclusion, the CHA2DS2VASC RS can assist clinicians in predicting chest CT-SS in COVID-19 patients, even if they are asymptomatic. However, this nding should be con rmed by larger studies.