Discussion
This study evaluated the prognostic value of pulmonary artery systolic pressure (PASP) in predicting outcomes of weaning patients with COVID-19 from mechanical ventilation. The adjusted HR for a 1-mmHg increase in PASP was 0.94 (95% CI 0.90‒0.97) for weaning (P < 0.001), and PASP was negatively correlated with the PaO2:FiO2 (r = -0.48, P < 0.01). The appropriate cutoff value of PASP for ventilator weaning was 39.50 mmHg, with a sensitivity of 100% and a specificity of 96.15%
Epidemiological studies of COVID-19 have shown that the mortality rate of critical patients is 49%, accounting for 5% of confirmed patients7. Recent studies have revealed that the severity of the disease is an independent predictor of a poor prognosis8. The effective and appropriate treatment for critically ill patients seems to be a key for decreasing mortality. Successful ventilator weaning often indicates a good prognosis. At present, the conditions for clinical discontinuation from mechanical ventilation are the minute ventilation volume, maximum inspiratory pressure, respiratory frequency, rapid shallow breathing index, 0.1-second airway closure pressure and so on9, 10. However, a previous study showed that attempts at extubation failed in approximately 20% of mechanically ventilated patients, and these patients spent more than 40% of the total mechanical ventilation time in the overall weaning process11. Improper extubation time leads to hypoxemia, and can even lead to a need for tracheal reintubation. On the other hand, prolonged mechanical ventilation time increases the risk of related complications (such as infection, gastrointestinal bleeding, deep vein thrombosis, etc.)12. Due to the recurrent nature of the disease symptoms, many patients had met the clinical indicators and were temporarily weaned off the ventilator, but then required mechanical ventilation again. Therefore, in order to ascertain the correct timing for weaning patients off the ventilator, more accurate and specific predictors appropriate to the characteristics of COVID-19 are needed.
In this study, we found statistically significant differences in biochemical indicators (IL-6 and D-dimer) between the weaning success and weaning failure groups. The former indicated that the immune response of patients who failed to be weaned off the ventilator was stronger, and the latter reflected the fibrinolytic dysfunction in these patients. The levels of PCT and CRP were higher than the normal values, without significant differences between the two groups, indicating that both groups showed enhanced inflammation. In terms of myocardial injury markers, the levels of TnI and myoglobin were significantly higher in group II than in group I, indicating that patients in whom ventilator weaning failed had more serious heart injury.
We also found that patients who failed to wean off the ventilator had a thicker interventricular septum and a higher PASP on echocardiography, which was consistent with a previous study13. The thickened interventricular septum may be related to the cytokine storm associated with COVID-19. In our study, the levels of IL-6 were significantly elevated in both groups, and was higher in patients who were not successfully weaned from the ventilator. The cytokine storm may increase the permeability of the vascular wall and could lead to myocardial edema14, 15. In addition, hypoxia can also lead to an early and substantial inflammatory response and cell damage16, and may also lead to some myocardial thickening and edema.
Among the echocardiographic measurements, PASP appeared to be most related to the severity of the patients’ condition. Our research showed that PASP had a moderate correlation with the OI, which indicated that PASP could partly reflect the lung condition of COVID-19 patients. The result of Cox analysis showed that PASP was an independent risk factor of ventilator evacuation in critically ill COVID-19 patients. For every 1 mmHg increase in PASP, the risk of weaning from the ventilator was 0.94 times higher, even after adjusting for other risk factors. The ROC curve showed that the best cutoff value for PASP for predicting the success of weaning off the ventilator was 39.50 mmHg, with a sensitivity of 100% and a specificity of 96.15% (P < 0.0001), and an AUC is 0.99, which showed the good power of the test. We suggest that PASP and other criteria for ventilator weaning should be considered together. In critically ill patients, if the PASP is higher than 39.5 mmHg, ventilator weaning should not be considered.
Other studies have also confirmed that PASP is related to prognosis in patients with cardiovascular disease. One study showed that when PASP > 45 mmHg, the prognosis of acute decompensated heart failure (ADHF) was poor; early right ventricular dilatation or dysfunction suggested a poor prognosis, while LVEF or left ventricular diastolic function could not predict prognosis17.
There are many reasons for the increased pulmonary artery pressure in critically ill COVID-19 patients. Anatomic reasons include inflammation of pulmonary arterioles involved in bronchitis and the surrounding inflammation, accompanied by thickening, spasm, or fibrosis of the blood vessel walls. Functional reasons include pulmonary vasoconstriction and spasm caused by a neurohumoral mechanism, such as hypoxia and respiratory acidosis, which increase pulmonary circulatory resistance. In addition, secondary injury (emphysema) is caused by long-term mechanical ventilation, which indirectly causes the rise of PASP. Computed tomography angiography of one case in our study confirmed that the spasticity and thrombus of pulmonary arterioles caused the PASP elevation and that PASP fluctuates during anticoagulation and vasodilation treatment. In addition to the pulmonary vasospasm caused by the pathological changes in the lung itself, and the decrease in pulmonary vascular compliance caused by pulmonary fibrosis, the accompanying cardiac diseases also have a significant impact. Acute virus replication in the myocardium and secondary inflammatory responses can lead to acute myocardial injury, some of which can lead to left ventricular systolic dysfunction and myocardial swelling18, that is, ADHF, which can aggravate the pulmonary congestion. With the progression in the disease course, the pulmonary artery pressure of these patients increased gradually, which increased the burden of the right heart. In the present study, two of the three patients who died showed a sudden decrease in right heart function and pulmonary artery pressure several days before death. Therefore, PASP elevation is a comprehensive indicator of the interaction between systemic inflammation and pathogen-mediated cardiovascular and respiratory system damage, which may be the mechanism underlying the worse outcomes.
PASP measurements in this study could have been influenced by various factors, such as nitric oxide (NO) treatment and increased blood volume due to fluid retention. NO gas is soluble in water and easily passes through a lipid membrane. NO can relax vascular smooth muscle, inhibit platelet aggregation, affect leukocyte chemotaxis, control inflammation, ease pain, regulate immunity, eliminate oxygen free radicals, and protect the gastric mucosa. NO can combine with hemoglobin in red blood cells and acts as a potent vasodilator19. Three of our patients were treated with NO, which temporarily reduced but did not reverse PASP.
Furthermore, hypoxia and hypercapnia can cause sympathetic nerve excitability, renal vasoconstriction, and decreased renal blood flow, resulting in water and sodium retention. The increased blood volume can lead to increased pulmonary circulation resistance. However, the patients in this study underwent strict volume management to avoid fluid retention as far as possible, and the left atrial diameters were rarely enlarged, as measured by echocardiography. A previous study also showed no significant association between echocardiographic indices of filling pressures and weaning failures20. Therefore, PASP in this study was less affected by pre-pulmonary circulation load.
PASP did not significantly associated with success of ventilator weaning in the logistic regression analysis; however, it showed an association with the outcome in Cox regression analysis. This may be explained by the addition of the time variable, for correction, in the Cox regression analysis; this result shows that continuous pulmonary hypertension, rather than temporary changes, may be more strongly related to the prognosis.
In this study, estimation of pulmonary artery pressure was based on TRPG plus RAP. The literature on mechanical ventilation indicates that CVP is equivalent to RAP; hence, we used CVP instead of RAP. For spontaneous respiration, we used the American Society of Echocardiography guidelines for estimation; sonographic measurement of the IVC diameter and collapsibility was used to estimate RAP21. In terms of the effect of mechanical ventilation on right atrial pressure, high positive end-expiratory pressure (PEEP; 8‒10 cmH2O) will affect the flow from the right heart to the left heart. However, the hemodynamics of the heart mainly depend on changes in the internal pressure of the heart, and our patients’ PEEPs were generally below 8‒10 cmH2O during mechanical ventilation; thus, the increase in the external pressure likely had little effect22.