Which Body Composition Parameters on Computed Tomography Are More Successful in Predicting the Prognosis of COVID-19 Patients? : Journal of Computer Assisted Tomography

Secondary Logo

Journal Logo

Chest Imaging

Which Body Composition Parameters on Computed Tomography Are More Successful in Predicting the Prognosis of COVID-19 Patients?

Ufuk, Furkan MD; Utebey, Ayse Ruksan MD; Yavas, Huseyin Gokhan MD∗,†; Baser Oncel, Sevin MD; Akbudak, Ismail Hakki MD§; Sari, Tugba MD

Author Information
Journal of Computer Assisted Tomography 47(1):p 58-66, 1/2 2023. | DOI: 10.1097/RCT.0000000000001387

Abstract

The coronavirus disease 2019 (COVID-19) poses a significant problem for global health, and COVID-19 patients may present with respiratory failure, pneumonia, thromboembolic events, and multiorgan failure.1,2 It has been reported that advanced age, immunosuppression, hypertension, coronary artery disease, heart failure, malignancy, diabetes mellitus, chronic lung disease, and obesity may play an essential role in the prognosis of COVID-19.2,3 Moreover, various body composition parameters (BCPs) on computed tomography (CT) have been shown to be associated with poor outcomes in COVID-19 patients.4–9 However, in these studies, different researchers investigated the effect of different BCPs in diverse populations. For example, Ufuk et al4 reported that decreased pectoralis muscle area (PMA) was significantly associated with prolonged hospital stay and intubation. Gupta et al5 reported that the increased coronary artery calcification (CAC) severity (CACS) on chest CT was significantly associated with intubation and mortality. Petersen et al6 reported that increased visceral adipose tissue area (VAA) was associated with intubation and intensive care unit (ICU) stay. Besutti et al7 showed that increased pectoralis muscle density (PMD) was a protective factor for hospitalization, intubation, and mortality. Kottlors et al8 showed that increased waist circumference (WC), low paraspinal muscle circumference (PMC), and low waist to paravertebral muscle circumference ratio (WPR) are poor prognostic indicators for the need of ICU stay. However, the most valuable prognostic factor(s) among these BCPs in COVID-19 patients is unknown. Thus, this study aimed to compare the predictive effect of different BCPs in COVID-19 patients and determine the most important BCP(s) in COVID-19 patients for prognosis.

MATERIALS AND METHODS

This study was approved by the institute review board (approval number: 601167787/71074) of our hospital and conducted in accordance with the Helsinki Declaration (2013 Edition).

Study Population

Symptomatic adult patients with positive reverse transcription–polymerase chain reaction test results for COVID-19 admitted to our center between December 1, 2020, and March 31, 2021, were consecutively reviewed. Patients who had an unenhanced chest CT within 24 hours of admission, which was obtained according to the statement of the Fleischner Society,10 were investigated. Those who had complete data on clinical outcomes (mortality and intubation) at a minimum 30-day follow-up were included among these patients. Patients who underwent contrast-enhanced chest CT, patients younger than 18 years, pregnant patients, patients with missing clinical data, vaccinated against COVID-19, and patients with inappropriate CT images for evaluation were excluded.

Image Acquisitions

Unenhanced chest CT examinations were performed while the patient was lying in the supine position and during deep inspiration using a multidetector CT scanner (Ingenuity 128, Philips Healthcare). Low-dose CT protocol covering the chest was used, and image acquisition parameters were as follows: 64 × 1-mm slice collimation, 0.5-second rotation time, 250 to 300-mm field of view, 100-kilovoltage tube voltage, and 50-milliampere seconds (mAs, effective tube current-time product). The raw data were reconstructed with an iterative model reconstruction (strength level of 2, Philips Healthcare) algorithm with a slice thickness of 1 mm.

Clinical Examination

The patients' comorbidities (such as malignancy, immunosuppression, diabetes mellitus, chronic renal failure, hypertension, heart failure, chronic liver failure, and chronic lung disease) were assessed from the hospital archive.

Severity of Pneumonia Assessment

Two radiologists with 3 and 2 years of experience in thoracic imaging (observers 1 and 2, respectively) evaluated the chest CT images in consensus to determine the severity of pneumonia (SoP). Observers were unaware of the outcomes of the patients. Chest CT evaluations were performed in a workstation (IntelliSpace Portal, Philips, the Netherlands) using lung window (window center [WC], −600 HU; window width, 1500 HU) settings. Pneumonia severity score (PSS. the percentage of lung involvement) was calculated in each patient using a simple method as previously described.4,11 The percentage of each lung lobe involvement was calculated as follows: no involvement = 0 points, 1%–25% volume of the lung lobe involvement = 1 point, 26%–50% involvement = 2 points, 51%–75% involvement = 3 points, and 76%–100% involvement = 4 points. The scores of each lung lobe were summed and multiplied by 5 to obtain the PSS (in percent).

Body Composition Analysis

All chest CT images were anonymized before assessing BCPs, and participants with COVID-19 were randomly numbered. Body composition parameters were assessed by a radiologist (observer 3) with 7 years of experience in cardiothoracic radiology using a commercially available medical image viewer (OsiriX MD; Pixmeo SARL, Bernex, Switzerland). Each BCP was evaluated in different sessions, and the observer was unaware of measurements in previous sessions and clinical data of the patients.

  1. The sum of all patients' bilateral pectoralis muscles area (PMA) was automatically measured from the transverse CT image just above the arch of the aorta. Pectoralis muscles were shaded manually using an attenuation range of −29 and 150 HU, as previously described by Ufuk et al.4
  2. The CACS was evaluated in the 4 main coronary arteries (right, left main, left anterior descending, and left circumflex coronary arteries) as described by Gupta and colleagues.5 Coronary artery calcification was scored as 0 to 3 points for each coronary artery (a score of 1: the involvement of less than one-third of the coronary artery length [CAL]; 2: the involvement of one-third to two-thirds of the CAL; 3: >two-thirds of the CAL) and summed to obtain a total CAC score of 0 to 12. Total scores were 0: no CAC, 1–3: mild CAC, 4–5: moderate CAC, and ≥6: severe CAC. Coronary artery calcification scoring was not performed in patients with coronary artery bypass graft or coronary stents.
  3. On axial CT images, VAA was measured on a single slice at the midlevel of the first lumbar vertebra (L1) between −190 and −30 HU, as described by Petersen et al.6
  4. On axial chest CT images, the right PMD was measured just above the aortic arch. Pectoralis major and minor muscles were shaded manually using a predefined attenuation range of −29 HU and 150 HU, as Besutti et al7 described.
  5. On axial CT images, the body (waist) circumference (WC) and PMC were measured using a freehand region of interest tool at the central level of the 12th thoracic vertebra. The mean of the right and left side PMC was calculated. Finally, waist to PMC ratio (WPR) was calculated according to the formula defined by Kottlors et al8 as follows: WPR = WC/[(right + left PMC) / 2].
  6. On axial chest CT images, the cross-sectional areas of the paravertebral muscles (the erector spinae, spinalis thoracis, longissimus thoracis, and iliocostalis lumborum muscles) were measured at the T5 vertebral level (T5MA), as described by Schiaffino et al.9

Examples of BCP assessments are depicted in Figure 1. To investigate the interobserver agreement, assessment of SoP and measurements of BCPs described above were repeated by observer 2 in the first 100 consecutive patients. Observer 2 was unaware of the patients' clinical data measurement results of observer 3.

F1
FIGURE 1:
Examples of evaluation of BCPs on chest CT images. A, Evaluation of the PMA between −50 and 90 HU in the transverse chest CT image just above the level of the aortic arch (pectoralis major and minor muscles are colored). B, An unenhanced chest CT image shows calcification in the left anterior descending coronary artery (red circle). C, Axial CT image at the midlevel of the first lumbar vertebra (L1) shows VAA measurement between −190 and −30 HU. D, Pectoralis muscle density assessment on an axial CT slice above the aortic arch level, using an attenuation range of −29 to 150 HU. E, The WC and PMC measurements at the T12 level. F, The assessment of paravertebral muscles (the erector spinae, spinalis thoracis, longissimus thoracis, and iliocostalis lumborum muscles) at the T5 vertebral level (T5MA).

Statistical Analysis

Continuous variables were assessed for normality using the Kolmogorov-Smirnov test. Normally distributed data are presented as mean ± SD and compared by Student t test between groups. Nonnormally distributed data are described as median with interquartile range (IQR) and tested by the Kruskal-Wallis test. Pearson correlation coefficient (r) was used to explore the relationship between the BCPs and the SoP. Unadjusted odds ratios (ORs with 95% confidence of interval [CI] levels) were calculated using univariate binary logistic regression to estimate the potential associations of variables with intubation and death. Statistically significant variables (P < 0.05) in the univariate analysis were used for multivariable logistic regression analysis. The receiver operating characteristic curve (ROC) analysis with the area under the curve (AUC) was used to assess the performance of the BCPs. Moreover, the AUC values were compared using the DeLong method. The intraclass correlation coefficient (ICC) test and weighted κ statistics were used to investigate the consistency in the measurements made by the observers. An ICC or weighted κ score less than 0.5 is accepted as poor agreement, 0.5 to 0.75 as moderate agreement, 0.75 to 0.9 as good agreement, and greater than 0.9 as excellent agreement. MedCalc (MedCalc Software, Ostend, Belgium) and SPSS v26.0. (IBM, Armonk, NY) were used for analyses. A P value less than 0.05 was considered to indicate statistical significance.

RESULTS

Five hundred eighty-eight COVID-19 patients underwent chest CT examination at our center between the examined dates. Among these patients, 238 patients (121 males, 50.84%) were included in the study (Fig. 2). The overall median age of the population was 48 years (IQR, 36–63 years). Height and weight data were available for 151 patients (63.45%). The overall median height was 167 cm (IQR, 160–172 cm), median weight 74 kg (IQR, 61–80 kg), and body mass index was 24.6 kg/m2 (IQR, 23.6–27.5 kg/m2; Table 1). The mean DLP (dose length product) and CTDIvol (CT dose index) values were 58.2 ± 10.1 mGy*cm and 1.8 ± 0.46 mGy, respectively.

F2
FIGURE 2:
The flow chart of the study population. Figure 2 can be viewed in color online at www.jcat.org.
TABLE 1 - The Descriptive Characteristics of the Study Population
Characteristics Total Population (n = 238) Intubation (n = 24) Death(n = 15)
Median (IQR) Median (IQR) Median (IQR)
Age, y 48 (36–63) 64 (48–79) 69 (64–84)
Males, % 121 (50.84) 17 (70.83) 8 (53.33)
Weight, kg 74 (61–80) 77 (63–85) 79 (65–91)
Height, cm 167 (160–172) 166 (157–171) 164 (152–169)
Body mass index, kg/m2 24.6 (23.6–27.5) 28.9 (27.1–30.6) 29.3 (28–31.5)
Comorbidities, % 112 (47.06) 16 (66.66) 15 (100)
Hypertension, % 63 (26.47) 10 (41.66) 9 (60)
Diabetes, % 52 (21.85) 11 (45.83) 8 (53.33)
COPD/Asthma, % 32 (13.45) 7 (29.17) 5 (33.33)
CAD/Heart Failure, % 17 (7.14) 8 (33.33) 7 (46.66)
Malignancy, % 9 (3.78) 5 (20.83) 3 (20)
Immunosuppression, % 7 (2.94) 1 (4.17)
Renal failure, % 2 (0.84)
Cirrhosis, % 1 (0.42)
SoP 4 (0–7) 12 (9–14) 13 (8–17)
Results are expressed as median and interquartile range (IQR).
CAD, coronary artery disease; COPD, chronic obstructive pulmonary disease; IQR, interquartile range.

One hundred twelve patients (47.06%) had at least one comorbidity, and hypertension was the most frequent (63 of 238, 26.47%), which was followed by diabetes mellitus (52 of 238, 21.85%). The median hospitalization length was 1 day (IQR, 0–8 days), 120 patients had at least 1-day hospitalization, 31 of 238 patients (13.03%) were admitted to the ICU, 24 patients (10.08%) were intubated, and 15 patients (6.3%) died during at least 1-month follow-up. Of the 120 patients hospitalized for at least 1 day, 17 were hospitalized at their second admission within a week. The median oxygen saturation (Spo2) by pulse oximetry value at initial admission was 96% (IQR, 93%–100%), and the Spo2 value was significantly correlated with PSS (P = 0.001, r = −0.748).

Chest CT showed pneumonia in 171 of 238 patients (71.85%), and the median SoP was 4 (IQR, 0–7; Table 1). Ground-glass opacity was the most common predominant pattern found in 129 patients (54.2%), and 32 patients (15.97%) had consolidation predominantly. Consolidation and ground-glass opacities were equal in 4 patients (1.68%).

Severity of pneumonia, CACS, and WPR values were significantly higher in patients who died than in survivors (P = 0.0001, P = 0.003, and P = 0.001, respectively). Pectoralis muscles area, PMD, and T5MA values were significantly lower in patients who died than in survivors (P = 0.001, P = 0.0001, and P = 0.01, respectively). Severity of pneumonia, CACS, and WPR values were significantly higher in intubated patients than in nonintubated (P = 0.0001, P = 0.001, and P = 0.0001, respectively). Pectoralis muscle area, PMD, and T5MA values were significantly lower in intubated patients than in nonintubated (P = 0.003, P = 0.0001, and P = 0.011, respectively; Table 2). The correlation between BCPs and SoP is shown in Table S1, https://links.lww.com/RCT/A148.

TABLE 2 - The Comparison of BCPs
Intubated (n = 24) Not Intubated (n = 214) P Death (n = 15) Survivor (n = 223) P
SoP 11.5 ± 3.9 4 ± 4.3 0.0001 12.27 ± 4.5 4.26 ± 4.4 0.0001
CACS 2.42 ± 2.87 0.96 ± 1.93 0.001 3.67 ± 2.99 0.94 ± 1.9 0.003
VAA 132.6 ± 50.3 105.5 ± 71.62 0.073 109 ± 44.2 108.1 ± 71.6 0.948
WC 101.7 ± 9.39 95.1 ± 10.6 0.004 99.4 ± 10.16 95.55 ± 10.67 0.177
PMC 34.76 ± 3.97 36.4 ± 4.34 0.067 33.7 ± 4.05 36.47 ± 4.3 0.022
WPR 2.95 ± 0.31 2.63 ± 0.35 0.0001 2.96 ± 0.24 2.64 ± 0.36 0.001
PMA 27.36 ± 7.4 35.1 ± 12.3 0.003 24.78 ± 7.13 34.97 ± 12.09 0.001
PMD 20.4 ± 9.5 32.1 ± 9.4 0.0001 16.47 ± 10.26 31.9 ± 9.34 0.0001
T5MA 11.8 ± 1.7 13.8 ± 3.8 0.011 11.25 ± 1.5 13.77 ± 3.7 0.01
The results are expressed as mean and SD (mean ± SD).

Unadjusted and adjusted ORs for outcomes (intubation and death) are presented in Table 3 and Table 4. In multivariable logistic regression analysis, older age and male sex were significant predictors for intubation, and older age and presence of hypertension were significant clinical predictors for death. The increased pneumonia severity (SoP) was a significant predictor of intubation and death.

TABLE 3 - Univariate and Multivariable Analysis of Variables for Intubation
Intubation (n = 24)
Univariate Analysis Multivariable Analysis
Variable Unadjusted OR P Adjusted OR P
Age median, y 1.05 (1.02–1.08) <0.001 1.05 (1.02–1.09) 0.001
Female sex 0.39 (0.16–0.98) 0.04 0.33 (0.13–0.86) 0.023
Comorbidities 2.44 (1–5.94) 0.043 1.19 (0.43–3.28) 0.74
 Hypertension 4.86 (0.75–31.2) 0.122
 Diabetes 1.82 (0.19–17.28) 0.621
 COPD/asthma 1.9 (0.66–14.8) 0.33
 Immunosuppression 25 (5.6–110.8) 0.055
 Malignancy 1.7 (0.35–8.2) 0.722
 CAD/hearth failure 1.6 (1.47–1.94) 0.08
SoP 1.34 (1.21–1.49) <0.001 1.34 (1.2–1.51) <0.001
CACS 1.27 (1.09–1.48) 0.004 1.07 (0.88–1.43) 0.561
Any of CAC 4.76 (1.94–11.7) <0.001 1.51 (0.41–5.55) 0.532
VAA 1.01 (1–1.01) 0.082
WC 1.06 (1.02–1.11) 0.003 1.05 (0.99–1.1) 0.115
WPR 8.62 (2.76–26.9) <0.001 8.48 (2.22–32.34) 0.002
PMA 0.92 (0.88–0.98) <0.001 0.83 (0.76–0.91) <0.001
PMD 0.88 (0.84–0.93) <0.001 0.84 (0.78–0.91) <0.001
T5MA 0.81 (0.69–0.96) 0.005 0.79 (0.65–0.98) 0.028

TABLE 4 - Univariate and Multivariable Analysis of Variables for Death
Death (n = 15)
Univariate Analysis Multivariable Analysis
Variable Unadjusted OR P Adjusted OR P
Age median, y 1.11 (1.06–1.16) <0.001 1.09 (1.03–1.15) 0.014
Female sex 0.9 (0.32–2.56) 0.002 0.77 (0.24–2.5) 0.665
Comorbidities
 Hypertension 14 (1.27–172.3) 0.029 15.3 (1.16–204.5) 0.039
 Diabetes 3.7 (0.32–43.1) 0.337
 COPD/asthma 1.83 (1.59–2.17) 0.452
 Immunosuppression 1.96 (1.68–2.35) 0.723
 Malignancy 1.87 (1.62–2.22) 0.512
 CAD/hearth failure 1.99 (1.7–2.38) 0.041 1.47 (1.32–1.68) 0.564
SoP 1.34 (1.19–1.5) <0.001 1.39 (1.19–1.62) <0.001
CACS 1.45 (1.22–1.73) <0.001 1.05 (0.8–1.38) 0.711
Any of CAC 15.4 (3.37–69.98) <0.001 1.99 (0.27–14.37) 0.496
VAA 1 (0.99–1.01) 0.965
WC 1.03 (0.98–1.09) 0.174 1.02 (0.99–1.04) 0.144
WPR 7.53 (2.07–27.4) 0.002 3.49 (0.53–22.84) 0.192
PMA 0.89 (0.82–0.96) <0.001 0.86 (0.76–0.97) 0.013
PMD 0.84 (0.78–0.90) <0.001 0.85 (0.78–0.93) 0.002
T5MA 0.74 (0.59–0.94) 0.003 0.84 (0.65–1.09) 0.184
Bold text indicates a statistically significant.

Multivariable logistic regression analysis revealed WPR, PMA, PMD, and T5MA values significantly associated with intubation (Tables 3, 4). The ROC analysis showed PMD (adjusted OR, 0.84; 95% CI, 0.78–0.91; P < 0.001) has the highest AUC value for prediction of intubation (AUC, 0.814; 95% CI, 0.759–0.862; P < 0.001; Fig. 3). According to the DeLong, the AUC value of PMD was significantly higher than PMA (P = 0.005) and T5MA (P = 0.001).

F3
FIGURE 3:
The ROC curve shows the BCPs in the prediction of intubation. Figure 3 can be viewed in color online at www.jcat.org.

The PMA and PMD values were significantly associated with death in multivariable logistic regression analysis, and the ROC analysis revealed PMD (adjusted OR, 0.85; 95% CI, 0.78–0.93; P = 0.002) has the highest AUC value for prediction of death (AUC, 0.871; 95% CI, 0.821–0.911; P < 0.001) and the cutoff value of PMD for death was less than 23.1 HU (Tables 3, 4; Fig. 4). The AUC values of PMD and PMA was not statistically significant (P = 0.077).

F4
FIGURE 4:
The ROC curve shows the BCPs in the prediction of death. Figure 4 can be viewed in color online at www.jcat.org.

There was a good-excellent interobserver agreement in the measurements of BCPs and SoP assessment (ICC and weighted κ scores range, 0.899–0.998; Table 5; Fig. 5).

TABLE 5 - Interobserver Agreement in the Evaluation of Quantitative and Semiquantitative Parameters
Observer 1 Observer 2 ICC or Weighted κ
SoP 6 (0–10) 6 (1–9) 0.929 (95% CI, 0.913–0.945)*
CACS 0 (0–2) 0 (0–2) 0.948 (95% CI, 0.925–0.971)*
VAA 111.5 (64–159.7) 103 (59.7–146.8) 0.907 (95% CI, 0.836–0.994)
WC 96.8 ± 10.9 95.1 ± 10 0.988 (95% CI, 0.977–0.999)
WPR 5.42 (4.9–5.88) 5.52 (5.02–5.96) 0.948 (95% CI, 0.897–0.999)
PMA 30 (23.5–38.2) 28.8 (22.7–36.4) 0.940 (95% CI, 0.676–0.998)
PMD 28 ± 11.2 29.7 ± 10.2 0.914 (95% CI, 0.610–0.996)
T5MA 12 (10.5–14.35) 13.3 (11.9–15.5) 0.899 (95% CI, 0.844–0.978)
According to the distribution of data, results are expressed as mean with SD (mean ± SD) or median with interquartile range (IQR).
*Weighted κ analyses were performed.

F5
FIGURE 5:
Bland-Altman plot for comparison between observers for assessment of (A) VAA, (B) WC, (C) WPR, (D) PMA, (E) PMD, and (F) the area of the T5MA. Figure 5 can be viewed in color online at www.jcat.org.

DISCUSSION

The present study results showed that older age, male sex, and presence of hypertension were significant clinical and demographic predictors in COVID-19 patients. The SoP, waist to paravertebral muscle circumference ratio (WPR), PMD and PMA, and paravertebral muscle area at the T5 vertebral level (T5MA) values were significant radiological predictors in COVID-19 patients. However, CACS, VAA, and WC did not show a statistically significant prognostic effect in multivariable regression analysis. Pectoralis muscle density had the highest AUC value to predict intubation and death.

Ufuk et al4 examined chest CT images of 130 patients with COVID-19. They reported that patients with higher PSSs and lower PMA values were independent predictors for poor clinical outcomes. Moreover, they reported an excellent interobserver agreement for PMA measurement in 60 consecutive patients.4 In line with these results, excellent interobserver agreement was found for PMA assessment in the present study (ICC, 0.990). In another study examining pectoral muscles in COVID-19 patients, Hocaoglu et al12 reported that pectoralis muscle volume and PMD were negatively correlated with bad outcomes in COVID-19 patients. However, multivariable analysis was not performed in the study by Hocaoglu et al,12 and the effects of PMA and PMD, adjusted for other variables, were not investigated.12 Recently, Menozzi et al13 assessed skeletal muscle area at the T12 vertebra level in 272 COVID-19 patients, and Menozzi et al13 reported that low skeletal muscle area is independently associated with poor clinical outcomes.

Gupta and colleagues5 investigated 180 COVID-19 patients, and Gupta et al5 reported that any CAC was independently associated with mechanical ventilation and death. Moreover, Gupta and colleagues5 showed that a high CAC score (CACS) was independently associated with poor outcomes. In the present study, although the high CAC score or the presence of any CAC was associated with poor outcomes in the univariate analysis, they lost significance when cofactors were included in multivariable analysis. We suggest that the difference between the study by Gupta et al5 and the present study may be due to the demographic difference in the patient populations or the number of patients. The median age of the patient population in the study by Gupta et al5 was 68 years, while the median age was 48 years in the present study. Moreover, in a meta-analysis conducted by Lee et al14 and including 8 studies and a total of 4542 individuals, an increase in mortality was found in COVID-19 patients with high CAC scores. However, they showed that the CAC score had no effect on intubation or ICU stay. However, there was no consensus on CAC scoring methods and threshold values in the studies included in this meta-analysis.14 Gupta et al5 showed an excellent interobserver agreement for CAC score assessment on chest CT examinations. Similarly, there was an excellent interobserver agreement for CAC score assessment in the present study (weighted κ = 0.948).

Petersen et al6 retrospectively investigated a small population (30 patients) of COVID-19 patients, and they assessed VAA on axial CT images at the L1 vertebra level. Petersen et al6 showed that an increase in VAA area is significantly associated with ICU stay and intubation. Koehler and colleagues15 reported that the VAA at the third lumbar vertebrae (L3) level of the COVID-19 patients was significantly related to ICU admission. Although VAA values in patients who died or were intubated were higher than in those who remained, there was no statistical difference in the present study. Moreover, no significant relationship was found between VAA and prognosis in the present study. We suggest that this may be due to the number of cases between studies and the lack of multivariable analysis in previous studies.6,13 Moreover, the present study results showed an excellent agreement between observers for VAA assessment.

Besutti et al7 were investigated 318 COVID-19 patients, and they reported that incrementally increased PMD value had a protective effect on intubation or death (OR, 0.964; 95% CI, 0.934–0.996). Damanti et al16 were investigated 81 COVID-19 patients, and they reported that low muscle density on CT was associated with longer hospitalization, unsuccessful extubation, and death in COVID-19 patients. Similarly, the increase in PMD value was significantly protective for intubation (OR, 0.84; 95% CI, 0.78–0.91) and death (OR, 0.85; 95% CI, 0.78–0.93) according to the results of the present study. Moreover, the ROC analysis revealed that PMD showed the highest AUC value among BCPs for prediction of intubation (AUC, 0.814; 95% CI, 0.759–0.862) and death (AUC, 0.871; 95% CI, 0.821–0.911).

Kottlors et al8 defined the measurement of WC to paravertebral muscle circumference (PMC) ratio (WPR; WC/mean of the left and right PMC), and they evaluated the prognostic effect of WPR in 58 COVID-19 patients. Kottlors et al8 reported that WPR was an independent predictor for ICU stay, and there was an excellent agreement between observers for WPR measurement. Similarly, the present study results with a larger patient population demonstrated an excellent interobserver agreement for WPR measurement, and WPR was an independent predictor for intubation. Although WPR was a significant predictor of death in the univariate analysis, it lost significance by including cofactors in multivariable analysis.

Schiaffino et al9 retrospectively investigated 552 hospitalized COVID-19 patients, and they reported that the paravertebral muscle area at the T5 level (T5MA) is a significant predictor of ICU admission and death. They also found that the AUC of T5MA for ICU admission estimation was 0.83 (P = 0.38) and 0.87 for death (P = 0.001). Similarly, Koehler et al14 reported that skeletal muscle area measured at the fourth thoracic vertebrae (T4) level was significantly associated with bad outcomes (OR, 0.98, P = 0.026). Recently, Osuna-Padilla et al17 showed that low skeletal muscle mass in COVID-19 patients was associated with prolonged ICU and hospital stays. In line with these studies, the T5MA value was a significant predictor for intubation in the present study (OR, 0.79; 95% CI, 0.65–0.98; P = 0.028). Although T5MA was a significant predictor of death in univariate analysis, it lost its importance in multivariable analysis in the present study. This insignificance may be due to the number of deceased patients in the current study being lower than in the study by Schiaffino et al.9 In the present study, the AUC of T5MA for intubation estimation was 0.654 (95 CI, 0.590–0.714; P = 0.001) and 0.717 for death (95 CI, 0.656–0.774; P < 0.001).

The present study had several limitations. First, the present study was conducted retrospectively at a single center, and the sample size was relatively small. Nevertheless, this is the first study to compare the prognostic significance of CT-induced BCPs, which were significant in adult COVID-19 patients in previous studies. Second, while most previous studies investigated BCPs only in hospitalized patients, the present study investigated all COVID-19 patients who underwent chest CT examinations. Consistent with the ordinary course of the disease, we prefer to investigate all consecutive CT images of COVID-19 patients, as most patients with COVID-19 had a mild degree of illness at admission that did not require hospitalization. Third, not all COVID-19 patients with a chest CT examination were included because of exclusion criteria (lack of follow-up or clinical information). Therefore, it may have caused selection bias. Finally, height information and weight information were not available for all patients. Therefore, we preferred to use muscle areas, which have been shown to be successful in previous studies, rather than the skeletal muscle index.

In conclusion, SoP and BCPs on chest CT enabled reliable prediction of intubation and death in COVID-19 patients. While WPR, PMA, PMD, and T5MA values enabled reliable prediction for intubation, only PMA and PMD values enabled reliable prediction for death. Moreover, among the BCPs, PMD showed the highest success in predicting intubation and death. We recommend the use of PMD in addition to pneumonia severity and comorbidities in the risk assessment of COVID-19 patients.

REFERENCES

1. Zhu N, Zhang D, Wang W, et al. A novel coronavirus from patients with pneumonia in China, 2019. N Engl J Med. 2020;382:727–733.
2. Ufuk F, Savaş R. Chest CT features of the novel coronavirus disease (COVID-19) Turk. J Med Sci. 2020;50:664–678.
3. Idilman IS, Telli Dizman G, Ardali Duzgun S, et al. Lung and kidney perfusion deficits diagnosed by dual-energy computed tomography in patients with COVID-19–related systemic microangiopathy. Eur Radiol. 2021;31:1090–1099.
4. Ufuk F, Demirci M, Sagtas E, et al. The prognostic value of pneumonia severity score and pectoralis muscle area on chest CT in adult COVID-19 patients. Eur J Radiol. 2020;131:109271.
5. Gupta YS, Finkelstein M, Manna S, et al. Coronary artery calcification in COVID-19 patients: an imaging biomarker for adverse clinical outcomes. Clin Imaging. 2021;77:1–8.
6. Petersen A, Bressem K, Albrecht J, et al. The role of visceral adiposity in the severity of COVID-19: highlights from a unicenter cross-sectional pilot study in Germany. Metabolism. 2020;110:154317.
7. Besutti G, Pellegrini M, Ottone M, et al. The impact of chest CT body composition parameters on clinical outcomes in COVID-19 patients. PLoS One. 2021;16:e0251768.
8. Kottlors J, Zopfs D, Fervers P, et al. Body composition on low dose chest CT is a significant predictor of poor clinical outcome in COVID-19 disease—a multicenter feasibility study. Eur J Radiol. 2020;132:109274.
9. Schiaffino S, Albano D, Cozzi A, et al. CT-derived chest muscle metrics for outcome prediction in patients with COVID-19. Radiology. 2021;300:E328–E336.
10. Rubin GD, Ryerson CJ, Haramati LB, et al. The role of chest imaging in patient management during the COVID-19 pandemic: a multinational consensus statement from the Fleischner society. Radiology. 2020;296:172–180.
11. Li K, Fang Y, Li W, et al. CT image visual quantitative evaluation and clinical classification of coronavirus disease (COVID-19). Eur Radiol. 2020;30:4407–4416.
12. Hocaoglu E, Ors S, Yildiz O, et al. Correlation of pectoralis muscle volume and density with severity of COVID-19 pneumonia in adults. Acad Radiol. 2021;28:166–172.
13. Menozzi R, Valoriani F, Prampolini F, et al. Impact of sarcopenia in SARS-CoV-2 patients during two different epidemic waves. Clin Nutr ESPEN. 2022;47:252–259.
14. Lee KK, Rahimi O, Lee CK, et al. A meta-analysis: coronary artery calcium score and COVID-19 prognosis. Med Sci (Basel). 2022;10:5.
15. Koehler J, Boirie Y, Bensid L, et al. Thoracic sarcopenia as a predictive factor of SARS-COV2 evolution. Clin Nutr. 2022;S0261–5614:00032–00032.
16. Damanti S, Cristel G, Ramirez GA, et al. Influence of reduced muscle mass and quality on ventilator weaning and complications during intensive care unit stay in COVID-19 patients. Clin Nutr. 2021;S0261–5614:00375–00377.
17. Osuna-Padilla IA, Rodríguez-Moguel NC, Rodríguez-Llamazares S, et al. Low muscle mass in COVID-19 critically-ill patients: prognostic significance and surrogate markers for assessment. Clin Nutr. 2022;S0261–5614(22)00070–X. doi:10.1016/j.clnu.2022.02.019.
Keywords:

computed tomography; COVID-19; sarcopenia; prognosis; skeletal muscle

Supplemental Digital Content

Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.