LAUSR

To the Editor: In a recent issue of Critical Care Medicine, we read with interest the article by Dangers et al (1) who prospectively studied 105 patients with acute decompensated heart failure implanted with venoarterial extracorporeal membrane oxygenation (VA-ECMO). Of those patients, 61 (58%) were nonsurvivors at 1 year after hospital admission. The significant predictive factors for 1-year mortality were reportedly preextracorporeal membrane oxygenation (ECMO) Sequential Organ Failure Assessment (SOFA) score greater than 11, idiopathic cardiomyopathy, duration of pre-ECMO cardiac disease greater than 2 years, and pre-ECMO blood lactate greater than 4 mmol/L. As these results have important implications for the indication of ECMO implantation in patients with cardiogenic shock complicating acute decompensated heart failure, we appreciate the clinically useful information provided by this research. Several factors that potentially affected the current results need to be taken into account. First, the lactate levels and SOFA score before ICU admission should be considered. The authors reported differences of 4–5 days between shock onset-toECMO interval and ICU admission-to-ECMO interval; however, there was no information provided on the treatment for shock and/or acute decompensated heart failure before ICU admission. These intervals and therapies before ICU admission could have affected the lactate levels, SOFA score, and subsequent clinical outcomes. Jansen et al (2) demonstrated that the lactate levels changed according to the length of treatment and that reduced lactate levels were associated with favorable outcome. It has been previously demonstrated that successful treatment reduces the SOFA scores and is associated with favorable outcome (3). Adding the information on trends of improvement or deterioration of the pre-ECMO SOFA score may enable a more accurate evaluation. Second, the Kaplan-Meier curves (Fig. 2 in [1]) are nearly parallel between the periods of 3 months and 1 year after enrollment, suggesting that the pre-ECMO SOFA score was more appropriate for predicting short-term mortality rather than 1-year mortality, which was the predefined outcome. Hryniewicz et al (4) also showed that short-term mortality does not largely differ from long-term mortality in patients with refractory cardiogenic shock managed with VA-ECMO. As shown in Table 2 (1), long-term outcome could have been affected by several therapeutic factors, including the application of heart transplantation. Evaluating pre-ECMO predictive factors for short-term mortality using a Cox regression analysis adjusted by treatment factors would therefore be clinically useful. Third, the rationale for calculating the cut-off values for the SOFA score is unclear. It was previously demonstrated that a SOFA score greater than 14 was a significant predictive factor The authors reply: We thank Kyo et al (1) for their interest in our study (2), recently published in Critical Care Medicine, and their comments. First, we agree with them that having data on the course of patients between onset of shock and extracorporeal membrane oxygenation (ECMO) would have been interesting. Unfortunately, we do not have these data: indeed, 47% of them were referred from other hospital (cardiology department or ICU) to our ICU to discuss heart transplantation because of refractory cardiogenic shock and were implanted at admission. Data on clinical course before ECMO are not retrievable for these patients. Furthermore, although we agree that a decrease in Sequential Organ Failure Assessment (SOFA) score and/or for mortality in patients with refractory cardiogenic shock post cardiac arrest (3). Czobor et al (5) showed that in patients who had undergone VA-ECMO, the SOFA score of survivors was significantly lesser than that of nonsurvivors (9 and 16, respectively). We consider that the authors should have entered the variables as a continuous variable in the analysis or defined the optimal cut-off value using the receiver operating characteristic curve in the current cohort. In conclusion, we suggest that the authors provide additional data to enable the precise evaluation of the factors associated with mortality in this study (1). Supported, in part, by a Japan Society for the Promotion of Science KAKENHI Grant (numbers JP 16K09541, 17K11573), the Strategic Information and Communications R&D Promotion Programme, and the Japan Agency for Medical Research and Development. The authors have disclosed that they do not have any potential conflicts of interest.