Low PEEP Mechanical Ventilation and PaO2/FiO2 Ratio Evolution in COVID-19 Patients

Invasive mechanical ventilation (IMV) is the standard treatment in critically ill COVID-19 patients with acute severe respiratory distress syndrome (ARDS). When IMV setting is extremely aggressive, especially through the application of high positive-end-expiratory respiration (PEEP) values, lung damage can occur. Until today, in COVID-19 patients, two types of ARDS were identified (L- and H-type); for the L-type, a lower PEEP strategy was supposed to be preferred, but data are still missing. The aim of this study was to evaluate if a clinical management with lower PEEP values in critically ill L-type COVID-19 patients was safe and efficient in comparison to usual standard of care. A retrospective analysis was conducted on consecutive patients with COVID-19 ARDS admitted to the ICU and treated with IMV. Patients were treated with a lower PEEP strategy adapted to BMI: PEEP 10 cmH2O if BMI < 30 kg m−2, PEEP 12 cmH2O if BMI 30–50 kg m−2, PEEP 15 cmH2O if BMI > 50 kg m−2. Primary endpoint was the PaO2/FiO2 ratio evolution during the first 3 IMV days; secondary endpoints were to analyze ICU length of stay (LOS) and IMV length. From March 2 to January 15, 2021, 79 patients underwent IMV. Average applied PEEP was 11 ± 2.9 cmH2O for BMI < 30 kg m−2 and 16 ± 3.18 cmH2O for BMI > 30 kg m−2. During the first 24 h of IMV, patients’ PaO2/FiO2 ratio presented an improvement (p<0.001; CI 99%) that continued daily up to 72 h (p<0.001; CI 99%). Median ICU LOS was 15 days (10–28); median duration of IMV was 12 days (8–26). The ICU mortality rate was 31.6%. Lower PEEP strategy treatment in L-type COVID-19 ARDS resulted in a PaO2/FiO2 ratio persistent daily improvement during the first 72 h of IMV. A lower PEEP strategy could be beneficial in the first phase of ARDS in critically ill COVID-19 patients.

Relevant pathophysiological understanding of COVID-19 reported by Cronin [8], Nieman [9], Gattinoni [10], and Bendjelid [11] identified specific lung features in the early stages of the disease. According to these features, Habashi et al. stratified COVID-19 ARDS in two different groups, identifying a L-type and a H-type ARDS [12] based on different lung pathophysiology. The first population presents a higher lung compliance compared to "classic" ARDS patients, due to a probable alveolitis, with a shunt effect due to loss of local hypoxic vasoconstriction. The second population described presents a low lung compliance and a pattern of "baby-lung" compatible with classic ARDS [2,3,10,11,[13][14][15]. More recent evidences suggested that in patients with L-type ARDS a less aggressive approach implementing a lower PEEP strategy may be favorable [2,[16][17][18][19]. PEEP values between 8 and 10 cmH 2 O, intended as lower PEEP strategy, were suggested to be adequate in this setting [18].
During ARDS, also in the setting of critically ill COVID-19 patients, the PaO 2 /FiO 2 ratio is typically used as a prognostic stratification parameter [20]; moreover, it determines lung respiratory efficiency, acting as a primary clinical indicator of hypoxemia [20], allowing to properly evaluate changes in patients' respiratory status. The aim of this project was to verify if, in critically ill COVID-19 patients, a clinical management implementing a lower PEEP strategy during IMV was safe and efficient comparing to usual standard of care, analyzing the PaO 2 /FiO 2 ratio.

Methods
After approval by the Ethical Committee (Ethics Committees of Canton Ticino; Dec 2020, CE TI 3775) and in accordance with local federal rules, a retrospective analysis was conducted on consecutive patients with acute respiratory distress due to COVID-19 pneumonia admitted to the ICU during two pandemic waves (from March 2 to April 10, 2020, and from October 5, 2020, to January 15, 2021). All critically ill COVID-19 patients' relevant data like demographics, severity score (NEMSnine equivalents of nursing manpower use score, SAPSsimplified acute physiology score), clinical information, and laboratory/radiological results were obtained during patients' hospitalization from electronic health records. Standard laboratory tests included complete blood count, CRP, ferritin, ASAT, ALAT, blood ionogram, creatinine, urea, D-dimer, Prothrombin Time (PT), activated partial thromboplastin time (aPTT), fibrinogen, blood gas analysis, SvO 2 , NT-pro-BNP, blood, and urine cultures and urine analysis for Legionella pneumophila antigen. All patients underwent chest x-ray and transthoracic echocardiography, to assess the global cardiac function before every pronation cycle. Performing a chest CT scan was considered at ICU admission, if the examination had not been performed within the preceding 24 h.
At ICU admission, COVID-19 patients underwent noninvasive ventilation (NIV) through C-PAP or high-flow nasal cannula (HFNC); in case of worsening, defined as 2 consecutive ROX-index lower than 3.85 [21,22] or residual respiratory distress after 1 h of NIV, patients underwent IMV. Patients who did not develop neither dyspnea nor worsening in ROX-index were treated with NIV.
After endotracheal intubation, a lower PEEP strategy based on BMI was adopted: PEEP values of 10, 12, and 15 cmH 2 O were applied, respectively, for patients with BMI <30, 30-50, and >50 kg/m 2 . Once PEEP adjustment was performed, according to the ARDSnet PEEP table [13,23] and PV-tools ventilatory measurements, FiO 2 was adapted to maintain a SpO 2 greater than 92% and a PaO 2 > 60 mmHg (8 kPa). A protective ventilation strategy (TV 6-8 ml kg −1 , P plat < 30 cmH 2 O) with permissive hypercapnia (pH > 7.20) was adopted [24], with pronation cycles of 16 h beginning at the admission. A deep sedation was maintained to pursue a Richmond Agitation and Sedation Scale (RASS) of −4 during the first 36 h, combined with muscle relaxation in case of patient-ventilator asynchrony [25]. Given the thrombogenic diathesis of COVID-19 patients [24,26], all patients were treated with an intermediate-prophylaxis, switched to a therapeutic dose in case of high risk of venous thromboembolism [27]. All clinical, ventilatory, and biological data were reported.
Primary endpoint was to report the PaO 2 /FiO 2 ratio evolution during IMV after application of lower PEEP according to BMI during the first 3 days of IMV. Secondary endpoints were to report and analyze patients ICU LOS and IMV length, further describing patients' demographic characteristics, clinical complications rate, and critical care outcome.

Statistical Analysis
Descriptive statistics was performed to summarize the collected clinical data. Gaussian distribution was verified by Kolmogorov-Smirnov test. Differences between patient outcomes were studied with a t-test for independent groups or with a Mann-Whitney test if a non-parametric analysis was required. Similarly, comparison of clinical evolution over time was performed with a paired t-test or with a non-parametric Wilcoxon test, depending on data distribution. A study of differences between groups of categorical data was carried out with chi-square statistics. The significance level of p value was established to be <0.01, with a confidence interval (CI) of 99%. Statistical data analysis was performed using the SPSS.26 package (SPSS Inc., Armonk, NY; USA).

Discussion
Acute respiratory distress induced by SARS-CoV-2 is a critical clinical condition associated with COVID-19 infection [28,29]. In a multisystem disease such as COVID-19, a multidisciplinary approach is recommendable [30]; to minimize the high mortality rate potentially associated with COVID-19 pandemics, and to correctly manage this critical condition, adequate hospital resources, structured triage, and appropriate clinical training are required [30]. Even if the classic ARDS criteria were identified in COVID-19 patients [31], clinical evidences led us to consider that atypical aspects were also Fig. 1 CLM COVID-19 patients. Management of COVID-19 patients evaluated at our COVID-19 center during two pandemic waves (from March 2nd to April 10th, 2020, and from October 5th, 2020, to January 15th, 2021). ICU admission was performed according to standard selection criteria (SpO 2 < 85% and/or dyspnea and/or signs of mental confusion). Patients not on invasive MV were excluded from the analysis. evident, especially the lack of a reduced lung compliance with consequent hypercapnia [32]. The lack of "baby-lung" pattern [33] and the peculiar physiological characteristics of this ARDS [15,17,19,31], suggested the implementation of a specific ventilation setting, in particular concerning PEEP [18]. Based on the abovementioned aspects, we specifically applied a lower PEEP ventilatory setting to all patients receiving IMV, carefully tailored to lungs physiology and, in agreement to ARDSnet PEEP table, also to BMI [17,21,28,32]. As expected, after endotracheal intubation, we found patients easy to ventilate, with an average lung compliance higher than classic ARDS, without any sign of pCO 2 retention [34]. This approach was consistent with growing evidences, as suggested by Gattinoni et al. [33] and Bendjelid et al. [11], who identified the presence of two different ICU patient populations in COVID-19 ARDS. Although our strategy was different compared to the available literature [35] and to recent NIH guidelines [4], there are an ever-increasing number of evidences suggesting a lower PEEP strategy in the management of L-type ARDS, based on specific lung physiology [15,16,18]. In these cases, higher PEEP could cause lung overdistension, resulting in an increased driving pressure and lung damage [36][37][38][39][40]; moreover, PEEP levels greater than 10 cmH 2 O can induce the reduction of venous return, with consequent worsening of the circulation status, as well as local biotrauma and a worsening of alveolar damage [17,37,38]. In fact, a lower PEEP approach resulted beneficial in our patients' cohort; ventilatory data confirmed a rapid improvement in the oxygenation lung function already in the first 3 days after endotracheal intubation, without any sign of hypoventilation, suggesting that L-type ARDS [3,10,14,15] appeared to respond appropriately to lower PEEP treatment tailored to patients' BMI. Moreover, this ventilatory setting avoided us to induce both a reduction in alveolar ventilation and a worsening in arterial oxygen saturation due to alveolar over/under-distension.
Other groups reported a 50% and a 61.5% of death rate [41]; in our critically ill COVID-19 patients cohort, we registered a mortality rate of 31.6%. We supposed that a relatively low-pressure ventilation could prevent the transition from an initial alveolitis to an iatrogenic H-type ARDS, in which the ongoing inflammation is worsened by high levels of PEEP, through a ventilation-induced-lung injury (VILI) mechanism. Moreover, our patients cohort median ICU LOS was reported to be equivalent to other groups, like Bhatraju et al. [41], even including patients who were deceased in the ICU. This data could suggest that a lower PEEP strategy with a protective IMV approach can improve COVID-19 patients' in-hospital management, morbidity and mortality, although further studies are necessary to confirm this interesting hypothesis.
Our project was burdened by several limitations. Firstly, this study compared lower PEEP in consecutive critically ill COVID-19 patients, and it was not possible to compare the data to a control group, even if the reproducible results obtained in the two distinct pandemic waves strongly support our analysis. Secondly, it was a monocentric observational retrospective study, with a relatively small series of patients. In this regard, a comparison with current literature was performed; even if patient populations differed, results can be assumed to be consistent, as the cohorts are comparable in terms of disease severity and biochemical markers.

Conclusion
A lower PEEP treatment in critically ill COVID-19 patients on IMV resulted in rapid and progressive improvement of PaO 2 /FiO 2 ratio during the first 72 h. A more physiologybased IMV setting could help to implement the understanding of the ARDS pathophysiological mechanisms in COVID-19 patients management; further studies need to be performed to confirm this approach. Fig. 3 P/F ratio variation during MV. P/F ratio variation at ICU admission compared to the first, second, and third day of MV. All daily median PF values resulted significantly different compared to admission and compared to the day after, even with the use of low PEEP setting on MV. All differences resulted statistically significative (CI 99%, p < 0.001)