Body mass index is associated with pulmonary gas and blood distribution mismatch in COVID-19 acute respiratory failure. A physiological study

Abstract

Background: The effects of obesity on pulmonary gas and blood distribution in patients with acute respiratory failure remain unknown. Dual-energy computed tomography (DECT) is a X-ray-based method used to study regional distribution of gas and blood within the lung. We hypothesized that 1) regional gas/blood mismatch can be quantified by DECT; 2) obesity influences the global and regional distribution of pulmonary gas and blood; 3) regardless of ventilation modality (invasive vs. non-invasive ventilation), patients' body mass index (BMI) has an impact on pulmonary gas/blood mismatch.

Methods: This single-centre prospective observational study enrolled 118 hypoxic COVID-19 patients (92 male) in need of respiratory support and intensive care who underwent DECT. The cohort was divided into three groups according to BMI: 1. BMI<25 kg/m² (non-obese), 2. BMI = 25-40 kg/m² (overweight to obese), and 3. BMI>40 kg/m² (morbidly obese). Gravitational analysis of Hounsfield unit distribution of gas and blood was derived from DECT and used to calculate regional gas/blood mismatch. A sensitivity analysis was performed to investigate the influence of the chosen ventilatory modality and BMI on gas/blood mismatch and adjust for other possible confounders (i.e., age and sex).

Results: 1) Regional pulmonary distribution of gas and blood and their mismatch were quantified using DECT imaging. 2) The BMI>40 kg/m² group had less hyperinflation in the non-dependent regions and more lung collapse in the dependent regions compared to the other BMI groups. In morbidly obese patients, gas and blood were more evenly distributed; therefore, the mismatch was lower than in other patients (30% vs. 36%, p < 0.05). 3) An increase in BMI of 5 kg/m² was associated with a decrease in mismatch of 3.3% (CI: 3.67% to -2.93%, p < 0.05). Neither the ventilatory modality nor age and sex affected the gas/blood mismatch (p > 0.05).

Conclusion: 1) In a hypoxic COVID-19 population needing intensive care, pulmonary gas/blood mismatch can be quantified at a global and regional level using DECT. 2) Obesity influences the global and regional distribution of gas and blood within the lung, and BMI>40 kg/m² improves pulmonary gas/blood mismatch. 3) This is true regardless of the ventilatory mode and other possible confounders, i.e., age and sex.

Trial registration: Clinicaltrials.gov, identifier NCT04316884, NCT04474249.

Keywords: COVID-19; acute respiratory failure; dual-energy computed tomography; mechanical ventilation; obesity; ventilation/perfusion mismatch.

FIGURE 1

CONSORT flowchart diagram showing the flow of participants through each stage of the study.

FIGURE 2

Methods. Representative example of the regions of interest applied to pairs of images. All selected images underwent a semiautomatic delineation of the regions of interest corresponding to the lung parenchyma. The resulting gas and blood distribution maps were subsequently divided into ten gravitational levels. Abbreviations: ND: non-dependent; D: dependent.

FIGURE 3

HU distribution for gas (above) and blood (below) maps for the three BMI groups. Histogram bin width equal to 5 HU; for each bin, the values are reported as a percentage of the total voxels (mean ± SD). * To mark differences between the labeled value and the corresponding value for BMI>40. Analysis of variance (ANOVA) followed by multiple comparisons with Bonferroni correction (α < 0.05). Abbreviations: BMI: body mass index; HU: Hounsfield units.

FIGURE 4

Gravitational HU distribution for gas (left) and blood (right) maps in the BMI<25 group. Data illustrating the ten gravitational levels. Histogram bin width equal to 5 HU; y-axis: percentage of total voxels [mean ± SD]. Abbreviations: ND: non-dependent; D: dependent; HU: Hounsfield unit.

FIGURE 5

Gravitational HU distribution for gas (left) and blood (right) maps in the BMI>40 group. Data illustrating the ten gravitational levels. Histogram bin width equal to 5 HU; y-axis: percentage of total voxels [mean ± SD]. Abbreviations: ND: non-dependent; D: dependent; HU: Hounsfield unit.

FIGURE 6

Gas/blood mismatch analysis divided into three BMI groups. Regional 28 distribution is divided into ten gravitational levels for gas/blood match (violet), and 29 mismatch, where blue represents gas exceeding blood and pink represents blood exceeding gas. Compared to the 30 other groups, the BMI>40 group showed a lower, more homogeneously distributed 31 gas/blood match along the gravitational axis. ANOVA followed by multiple comparisons 32 with Bonferroni correction (α < 0.05). * To mark differences between the labelled value 33 and the corresponding value for the BMI>40 group. # To mark differences between 34 contiguous gravitational levels (p > 0.05).

FIGURE 7

Sensitivity analysis testing the influence of BMI and other potential confounding factors on gas/blood mismatch. The four independent tested variables tested were 1) age (1 year), 2) sex (M), 3) BMI (5 kg/m²), and 4) exposure to invasive mechanical ventilation (yes). The multiple linear regression model tested their effect on gas/blood mismatch (dependent variable) based on DECT. Abbreviations: M: male; BMI: body mass index.

FIGURE 8

Second sensitivity analysis for two separate subgroups of patients based on the ventilation mode. Sensitivity analysis testing the influence of BMI and other potential confounding factors on gas/blood mismatch based on DECT. The four independent tested variables tested were 1) age (1 year), 2) sex (male), 3) BMI (5 kg/m²), and 4) exposure to invasive mechanical ventilation (yes). The multiple linear regression model tested their effect on gas/blood mismatch (dependent variable). In this case the model was applied separately in two subgroups of patients (invasive ventilation vs. spontaneous breathing). Abbreviations: M: male; BMI: body mass index.