Long-term RV dysfunction was noted in nearly half of ARDS survivors who developed RV dysfunction during the course of their ARDS. Logistic regression analysis revealed that a higher RVEDA/LVEDA ratio during ARDS was the only significant risk factor for the occurrence of long-term RV dysfunction in ARDS survivors, while other echocardiographic parameters such as LVEF or TAPSE, or RV fraction area change (RVFAC)/RV systolic pressure (RVSP) (RV-PA coupling) were not found to be statistically significant. In a univariate analysis, RV dysfunction based on the ASE definition, RVEDA/LVEDA ratio > 0.6 + septal dyskinesia and RVEDA/LVEDA ratio > 1 were also associated with a higher odds of developing long-term RV dysfunction. This article represents the first study investigating long-term RV function in patients with non-COVID-19 ARDS.
While the data for long-term RV dysfunction in ARDS survivors is sparse. In our study, we found that 50.7% of patients had long-term RV dysfunction. A meta-analysis of 26 studies (3671 patients) showed that long-term RV dysfunction at median follow-up of 18 months was 18.1% in survivors of pulmonary embolism [19]. In patients with sub-massive pulmonary embolism, the prevalence of RV dysfunction at 18 months was higher at 40% illustrating the significant burden of long-term RV dysfunction in critical illness survivors [20]. In a study of survivors of severe ARDS due to coronavirus disease 2019 (COVID-19), 58–78% were found to have cardiac involvement on cardiac MRI performed a median (IQR) 71 (64–92) days from COVID-19 diagnosis [21]. Huang et al. also showed COVID-19 survivors post-discharge had a significantly reduced RV function compared to controls [22].
Interestingly, Ou et al. found that sepsis survivors had a significantly higher risk of cardiovascular events 1 year after discharge [23]. Similarly, Beesley et al. reported significantly higher cardiovascular events in patients with high or low left ventricular global longitudinal systolic strain in sepsis survivors [24]. Unfortunately these studies did not quantify the impact of RV dysfunction in these patients.
The development of RV dysfunction in ARDS is multifaceted. The increase in RV afterload occurs from an imbalance between vasoconstrictors and vasodilators, endothelial injury, hypoxic and hypercapnic vasoconstriction. The increase in pulmonary vascular resistance is further augmented by positive pressure ventilation and positive fluid balance [8]. In addition, primary contractile impairment of RV has also been postulated. Dessap et al. reported that driving pressure ≥ 18 cmH2O, PaCO2 ≥ 48 cm2O, and P/F ratio < 150 mmHg are independent risk factors of developing RV dysfunction during ARDS [25]. In our study, there was no significant difference in median driving pressure, PaCO2, or P/F ratio on the day of echocardiogram between patients who had long-term RV dysfunction and those did not. [26]. This may also explain the absence of any significant difference between the baseline and disease-specific clinical parameters between patients with and without long-term RV dysfunction. Similar to our study, cardiac involvement after recovery from COVID-19 was reported to be independent of preexisting conditions, severity, and overall clinical course [21].
There is a possible physiologic rationale for why we observed an association of RVEDA/LVEDA with long-term RV dysfunction while not detecting similar associations among other markers such as LVEF, TAPSE, S’, or FAC, which are indicators of myocardial contractility. The inflammatory cytokines and catecholamine associated with critical illness often dysregulate myocardial function, resulting in decreased contractility[27]. These changes are often reversible, normalizing upon recovery from disease [27, 28]. RVEDA/LVEDA, on the other hand, indicates a severity of RV dysfunction out of proportion to LV dysfunction, including right ventricular failure. While isolated RV enlargement can occur from stress or critical illness, it is much more indicative of high RV afterload than TAPSE, S’, or FAC. We observed that the patients with long-term RVD trended towards having higher RV systolic pressure during ARDS (48 vs 28.5 mmHg, p = 0.07). We speculate that the RVEDA/LVEDA may be more indicative of severe cardio-pulmonary disease that might be less likely to recover as rapidly as the impaired contractility seen in critical illness.
“RV-Protective” management strategy in addition to the currently recommended “Lung protective” strategy is currently being suggested to improve outcome in ARDS [29]. It becomes even more important to study the impact of RV protective strategies as RV dysfunction during ARDS not only increases short-term mortality but may also associated with long-term RV dysfunction as shown in our study [29–31]. Future studies in ARDS and COVID-19 ARDS should not limit the impact of RV protective interventions to short-term outcomes but also consider evaluating the continued benefit of these interventions.
Our study has several important limitations. First, we only included patients who survived ARDS and had a repeat TTE after the discharge. There were likely to be reasons why TTE was performed and this suggests the possibility that included patients might already have a higher risk or prevalence of RV dysfunction than patients without any symptoms or signs. Due to the nature of the retrospective study, it was difficult to completely eliminate this potential selection bias. Second, the complex anatomy of RV makes an adequate image-acquirement challenging especially in patients with ARDS. Transesophageal echocardiogram (TEE) was reported to have higher sensitivity to identify RV dysfunction [32], however, the access to TEE is limited in most ICUs. In this study, we have utilized a previously described multi-modal definition using TAPSE, S’, FAC of RV, semi-qualitative RV size and function as recommended by both the American society of Echocardiography the Preferred Reporting Items for Critical care Echocardiography Studies (PRICES) project endorsed by the European Society of Intensive Care Medicine [33, 34] and we believe that we minimized the risk of missing RV dysfunction by using these standardized echocardiographic parameters. Third, this was a single-center small study and the generalizability of our findings needs to be clarified with larger studies. Last, our regression model’s limited sample size may be underpowered to detect some associations. Given the small sample size at a single-center and potential selection bias due to retrospective nature, a larger prospective study investigating long-term RV function in ARDS survivors to enable our knowledge of causation and association to long-term clinical outcome including mortality is warranted.