Jump to comment:

• Response to Laura W Lund, and Jeremy D Kimmel ALung Technologies January 16, 2020
  Vito Marco Ranieri, ALAIN COMBES and TOMMASO TONETTI
  Published on: 5 February 2020
• Response to "Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study" (Combes et al,, 2019 Dec;74(12))
  Laura W Lund, Jeremy D Kimmel
  Published on: 5 February 2020

• Published on: 5 February 2020

  Response to Laura W Lund, and Jeremy D Kimmel ALung Technologies January 16, 2020

  • Vito Marco Ranieri, Anesthesia and Intensive Care Medicine Alma Mater Studiorum University of Bologna Policlinico di Sant'Orsola
  • Other Contributors:
    • ALAIN COMBES, CARDIOLOGY AND CRITICAL CARE
    • TOMMASO TONETTI, ANESTHESIA AND CRITICAL CARE

  The SUPERNOVA trial was a prospective observational phase II study supported by an unrestricted grant from three companies (Alung, Maquet, and Novalung) and by the European Society of Intensive Care Medicine (ESICM). The three companies provided equipment and covered costs for data monitoring, site visits, and insurance fees. The grant (€171,000) was made available to ESICM that supported data collection and analysis, and all administrative costs. As owner of the data, ESICM appointed the two principal investigators (AC and VMR) and the independent Data and Safety Monitoring Board (Jukka Takala, Chair). The study included 95 patients. The proportion of patients who achieved ultra-protective settings by 24 hours was 82%. Number of patients that experienced severe and ECCO2R-related adverse events was 2 (2%) and 37 (39%)1. Retrospective analysis of these data showed that (a) efficacy of ECCO2R to facilitate further reduction of tidal volume was lower with smaller artificial lungs and running at lower blood flow than with larger artificial lungs and running at a higher blood flow2; (b) haemolysis and bleeding was higher with the former than with the latter2; (c) applying these data to a previously described theoretical model3 we predicted that incorporating higher CO2 removal rates as factors to design randomized clinical trial might substantially reduce screening and sample size requirements4.
  In her letter, Dr Lund, expressed several concerns about these findin...

  Show More

  The SUPERNOVA trial was a prospective observational phase II study supported by an unrestricted grant from three companies (Alung, Maquet, and Novalung) and by the European Society of Intensive Care Medicine (ESICM). The three companies provided equipment and covered costs for data monitoring, site visits, and insurance fees. The grant (€171,000) was made available to ESICM that supported data collection and analysis, and all administrative costs. As owner of the data, ESICM appointed the two principal investigators (AC and VMR) and the independent Data and Safety Monitoring Board (Jukka Takala, Chair). The study included 95 patients. The proportion of patients who achieved ultra-protective settings by 24 hours was 82%. Number of patients that experienced severe and ECCO2R-related adverse events was 2 (2%) and 37 (39%)1. Retrospective analysis of these data showed that (a) efficacy of ECCO2R to facilitate further reduction of tidal volume was lower with smaller artificial lungs and running at lower blood flow than with larger artificial lungs and running at a higher blood flow2; (b) haemolysis and bleeding was higher with the former than with the latter2; (c) applying these data to a previously described theoretical model3 we predicted that incorporating higher CO2 removal rates as factors to design randomized clinical trial might substantially reduce screening and sample size requirements4.
  In her letter, Dr Lund, expressed several concerns about these findings related to (a) the improper categorization of ECCO2R device; (b) the non-controlled collection and biased analysis of safety data.
  First, the scientific basis of distinction between lower and higher CO2 extraction relies on blood flow rate (350-550 ml/min vs 800-1000 L/min) and membrane lung surface (0.6 vs 1.3m2) that are the main determinant of CO2 extraction. A recent experimental study tested several lung membranes with varying surface areas (0.4, 0.8, 1.0 and 1.3 m2) at different stepwise increases of blood flow (from 250 to 1000 ml/min) at a constant sweep gas of 8 L/min. The combination of blood flow of 750-1000 ml/min with a membrane lung surface of at least 0.8 m2 was optimal in correcting respiratory acidosis highlighting the concept that a complex interplay exists between blood flow rate and membrane lung surface that may affect the efficiency of CO2 extraction5. Moreover, despite the engineering aspects of CO2 removal elegantly discussed by Drs Lund and Kimmel, the retrospective analysis of the SUPERNOVA database revealed that, with the due recognition of the limits related to the retrospective analysis and the lack of measurement of amount of CO2 extracted, the clinical performance of the different technological settings was substantially different2. Second, because of the retrospective nature of our data2,3, all analyses were purely descriptive and we only performed bivariate comparisons. The initial sample size decided in the protocol was pragmatic and not intended to reach a specific power or any Type 1 error calculation. If we take as an example the rate of patients reaching a tidal volume of 4mL/kg at 24h, the power was strong enough to detect a difference between groups (64% vs 92%) and p value was lower than 0.01 meaning that it is unlikely to have such a difference only by chance. It should also be noted that respiratory rate and minute ventilation needed to keep PaCO2 within 20% of its baseline value (by protocol) were significantly higher for the device running at the lower blood flow2. Analysis of 70 patients treated with ECCO2R within the great Paris area, showed that biological hemolysis and bleeding were significantly higher with the device we classified as low CO2 extraction than with the device we classified high CO2 extraction6.

  Although we strongly agree that any comparisons between devices must be considered with caution due to the retrospective nature of our analysis, and understand the concerns expressed by ALung Technologies, we claim our commitment as independent scientists to make available to the clinical and scientific community all data for carrying out future studies and optimizing the clinical management of our patients
   

  1. Combes A, Fanelli V, Pham T, et al. Feasibility and safety of extracorporeal CO2 removal to enhance protective ventilation in acute respiratory distress syndrome: the SUPERNOVA study. Intensive Care Med 2019;45(5):592-600. doi: 10.1007/s00134-019-05567-4 [published Online First: 2019/02/23]
  2. Combes A, Tonetti T, Fanelli V, et al. Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study. Thorax 2019;74(12):1179-81. doi: 10.1136/thoraxjnl-2019-213591 [published Online First: 2019/08/15]
  3. Goligher EC, Amato MBP, Slutsky AS. Applying Precision Medicine to Trial Design Using Physiology. Extracorporeal CO2 Removal for Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2017;196(5):558-68. doi: 10.1164/rccm.201701-0248CP [published Online First: 2017/06/22]
  4. Goligher EC, Combes A, Brodie D, et al. Determinants of the effect of extracorporeal carbon dioxide removal in the SUPERNOVA trial: implications for trial design. Intensive Care Med 2019;45(9):1219-30. doi: 10.1007/s00134-019-05708-9 [published Online First: 2019/08/23]
  5. Karagiannidis C, Strassmann S, Brodie D, et al. Impact of membrane lung surface area and blood flow on extracorporeal CO2 removal during severe respiratory acidosis. Intensive Care Med Exp 2017;5(1):34. doi: 10.1186/s40635-017-0147-0 [published Online First: 2017/08/03]
  6. Augy JL, Aissaoui N, Richard C, et al. A 2-year multicenter, observational, prospective, cohort study on extracorporeal CO2 removal in a large metropolis area. J Intensive Care 2019;7:45. doi: 10.1186/s40560-019-0399-8 [published Online First: 2019/08/28]

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 5 February 2020

  Response to "Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study" (Combes et al,, 2019 Dec;74(12))

  • Laura W Lund, Vice President of Clinical Science ALung Technologies
  • Other Contributors:
    • Jeremy D Kimmel, Vice President of New Technology

  Dear Editor,

  We read with great interest the recently published article in Thorax by Combes and colleagues titled “Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study” [1]. In this article, the authors present brief, post-hoc analyses of safety and efficacy data derived from the SUPERNOVA trial, a single-arm, multi-center, pilot study assessing the feasibility and safety of extracorporeal carbon dioxide removal (ECCO2R) to facilitate ultra-protective ventilation in patients with moderate acute respiratory distress syndrome (ARDS) [2]. The study was conducted at 23 centers, each of which used one of three different ECCO2R devices.

  We wish to communicate significant concerns regarding improper categorization of ECCO2R device performance as well as important study limitations impacting interpretation and value of the presented data. The differentiation between devices based on the terms “higher CO2 extraction” and “lower CO2 extraction” is incorrect based on supporting evidence and engineering principles summarized in this letter. In addition, safety data was presented and statistically compared without including available associated data that would bring in to question the implications of the analyses. As the manufacturer of one of the ECCO2R devices used in the SUPERNOVA pilot study, we are strong believers in the life-saving potential of ECCO2R technology and its...

  Show More

  Dear Editor,

  We read with great interest the recently published article in Thorax by Combes and colleagues titled “Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study” [1]. In this article, the authors present brief, post-hoc analyses of safety and efficacy data derived from the SUPERNOVA trial, a single-arm, multi-center, pilot study assessing the feasibility and safety of extracorporeal carbon dioxide removal (ECCO2R) to facilitate ultra-protective ventilation in patients with moderate acute respiratory distress syndrome (ARDS) [2]. The study was conducted at 23 centers, each of which used one of three different ECCO2R devices.

  We wish to communicate significant concerns regarding improper categorization of ECCO2R device performance as well as important study limitations impacting interpretation and value of the presented data. The differentiation between devices based on the terms “higher CO2 extraction” and “lower CO2 extraction” is incorrect based on supporting evidence and engineering principles summarized in this letter. In addition, safety data was presented and statistically compared without including available associated data that would bring in to question the implications of the analyses. As the manufacturer of one of the ECCO2R devices used in the SUPERNOVA pilot study, we are strong believers in the life-saving potential of ECCO2R technology and its role as a critical care lung support option. However, we also recognize the risks of ECCO2R therapy and the importance of providing users with all available safety data to enable accurate characterization of the risk profile. While we support the authors in publishing available safety data from this study, the data presented lacked critical information and represented key limitations that preclude the basis of the analysis as well as the stated conclusions.

  The SUPERNOVA trial was not designed to investigate potential differences in device performance and safety, nor was such an analysis planned a priori [3]. However, in this post-hoc analysis, the data was reviewed retrospectively to compare differences in safety and efficacy between two groupings of the 3 devices, defined as either “higher CO2 extraction” or “lower CO2 extraction”. The question of “how much CO2 extraction is needed” to achieve certain therapeutic objectives is important, but the rationale used to categorize ECCO2R devices as “higher” versus “lower” extraction to address this question is fundamentally incorrect and, hence, undermines the entire analysis. In the SUPERNOVA study, CO2 extraction was not actually measured. It was assumed that CO2 extraction differences can be inferred based solely on differences in the operational blood flow rates and membrane surface area between devices.

  The CO2 extraction rate of a given blood gas exchange device depends strongly on its unique gas transport efficiency [4,5]. In the device characterized as “low CO2 extraction” (the Hemolung RAS), gas transport efficiency is significantly increased by secondary flow patterns achieved through integration of a centrifugal pump within the fiber bundle to achieve higher CO2 extraction at lower flows with less membrane surface area and using smaller catheters. This is the defining design feature of the Hemolung device.
  Data supporting similar CO2 extraction rates in ECCO2R devices operating at lower vs higher blood flows can be found in the literature. Hermann et al. published the results of a small study prospectively measuring CO2 extraction rates in 10 patients treated with ECCO2R using one of the devices also used in the SUPERNOVA trial that has been characterized as “higher CO2 extraction” [6]. With a membrane surface area of 1.3 m2, CO2 extraction rates of 70 mL/min (range of 42-85 mL/min), normalized to an inlet PCO2 of 45 mmHg, were measured at a blood flow of 1.0 L/min and a sweep gas flow of 14 L/min [7].

  Barrett and colleagues recently published an in-vitro study characterizing CO2 extraction using the Hemolung RAS, which has a membrane surface area of 0.59 m2. CO2 extraction rates of 89 - 98 mL/min were measured at a blood flow of 0.4 L/min, a sweep gas flow of 10 L/min, and an inlet PCO2 of 45 mmHg [8]. The study followed ISO 7199:2016 standards for blood gas exchanger conformance testing and utilized three different CO2 extraction measurement methodologies.

  This data from the literature indicates that use of the higher versus lower CO2 extraction nomenclature is not appropriate without validated measures of actual CO2 extraction. When comparing one ECCO2R device to another, higher blood flow and/or higher membrane surface area does not necessarily correlate with CO2 extraction. Thus, the conclusions drawn in the post-hoc analysis regarding differences in device CO2 extraction are incorrect.

  So then, if not CO2 extraction, how can the observed differences between the two groups of devices be explained? Unlike the original publication of the SUPERNOVA results, this publication did not clearly reiterate the important limitations in the study design that can’t be ignored when trying to draw meaningful and scientifically sound conclusions from the data. As a pilot study conducted at 23 different sites in Europe and Canada, SUPERNOVA was not designed to generate standard-of-care or inter-device control data, thus post hoc analyses of data subsets are speculative. Many potential site-dependent differences or biases could exist in patient selection, cannulation techniques, anticoagulation management, ECCO2R management, and ventilation management. While many of these factors had constraints imposed by the protocol procedures, these procedures provided for enough variation in practice to open the door to bias impact. Thirty-three of the 95 patients enrolled were treated with the Hemolung device at five different centers with an enrollment variation of 1, 3, 7, 8, and 14 subjects. This variation alone is large enough to enable site-dependent differences just within subjects treated with the Hemolung device. One of the more notable limitations in the study design with respect to conducting between-device comparisons is that each site only used one of the three possible devices.

  The post-hoc analysis also presented comparisons of safety data to support statistical differences in rates of hemolysis and bleeding between the two groupings of devices. These findings were of significant concern because they suggested higher rates than previously observed. Following publication of this article, we requested further information associated with the incidences of hemolysis presented in the analysis. Review of the additional information revealed several relevant observations with important implications for assessment of severity and device-relatedness. For example, of the seven hemolysis events reported with the Hemolung device, six were reported at a single site that treated only eight subjects which suggests possible biased reporting of hemolysis (or perhaps other unknown site-specific anomalies). Five of the seven incidents of hemolysis were reported to have occurred only on Day 0 of treatment, but there were no accompanying baseline measures taken prior to initiation of ECCO2R. The actual levels of plasma-free hemoglobin on which these five incidents of hemolysis were based was between 100 – 330 mg/L, which is an uncommonly low threshold for reporting of device-related hemolysis. The Extracorporeal Life Support Organization defines moderate hemolysis as peak plasma hemoglobin of 500-1000 mg/L occurring at least once during extracorporeal therapy and sustained for at least two consecutive days [8]. Given the severity of illness in the underlying patient population, it cannot be ruled out that the low levels of plasma-free hemoglobin reported to have occurred on Day 0 weren’t present at baseline, making it difficult to determine device-relatedness. Furthermore, the use of simple bivariate comparisons in a post-hoc analysis cannot ignore Type 1 error inflation where the likelihood of observing a statistically significant p-value increases with the number of comparisons made. In short, the safety data presented is not controlled, is without baseline comparisons, reflects a reasonable likelihood of bias, and does not reflect a priori Type 1 error control for analysis.

  Additionally, the safety data was presented without any relevant details that would enable readers to effectively weigh the risks of ECCO2R against the risks of ventilator-associated lung injury. What impact did these events have on the patient? What medical interventions were required? How did the events impact ECCO2R therapy or ventilation status? Was ICU length of stay impacted? Without this perspective and without control data, the information presented could be misinterpreted or misused. The communication concludes with the statement, “Future randomised clinical trials assessing the overall benefit and harm of ultraprotective ventilation should be carried out with ECCO2R devices equipped with a larger membrane lung and blood flows between 800 and 1000 mL/min.” Given the many limitations of this post-hoc analysis, the incorrect differentiation in CO2 extraction between devices, and the paucity of information provided with the safety data, this conclusion is not supported.

  Respectfully,

  Laura W. Lund, Ph.D. Jeremy D. Kimmel, Ph.D.
  Vice President of Clinical Science Vice President of New Technology
  ALung Technologies, Inc. ALung Technologies, Inc.

  Section Bibliography

  1. Combes A, Tonetti T, Fanelli V et al. Efficacy and safety of lower versus higher CO2 extraction devices to allow ultraprotective ventilation: secondary analysis of the SUPERNOVA study. Thorax 2019; 74: 1179-1181
  2. Combes A, Fanelli V, Pham T, Ranieri VM. Feasibility and safety of extracorporeal CO2 removal to enhance protective ventilation in acute respiratory distress syndrome: the SUPERNOVA study. Intensive Care Medicine 2019; 45: 592-600
  3. Fanelli V, Ranieri MV, Mancebo J et al. Feasibility and safety of low-flow extracorporeal carbon dioxide removal to facilitate ultra-protective ventilation in patients with moderate acute respiratory distress sindrome. Critical Care 2016; 20: 1-7
  4. AAMI Standards and Recommended Practices. Association for the Advancement of Medical Instrumentation. Cardiovascular implants and artificial organs — Blood-gas exchangers (oxygenators). ANSI/AAMI/ISO 7199:2009/(R)2014 and ANSI/AAMI/ISO 7199:2009/A1:2012(R)2014.
  5. Hattler B, and Federspiel W: Chapter 6: Gas exchange in the venous system: Support for the failing lung. In: The artificial lung. Edited by Vaslef S, Anderson R. Georgetown, TX: Landes Bioscience; 2002: 133–174.

  Show Less

  Conflict of Interest:
  Both authors are employed by ALung Technologies, the manufacturer of one of the medical devices used in the Supernova study, specifically, the Hemolung Respiratory Assist System.

  • Back to top