I. Introduction

Septic shock is associated with a high mortality risk related to progressive tissue hypoperfusion [1]. However, despite extensive research on the best monitoring and resuscitation strategies, many uncertainties remain. Over-resuscitation, particularly when inducing fluid overload, might contribute to a worse outcome [2]. Fluid overload more likely occurs when fluids are administered to fluid unresponsive patients, but also when inappropriate resuscitation goals are pursued, or a "one-size-fits-all" strategy is followed [3].

From a hemodynamic point of view, several pathogenic mechanisms determine a progressive circulatory dysfunction[4]. While loss of vascular tone and relative hypovolemia predominate in early phases, more complex mechanisms such as endothelial and microcirculatory dysfunction, progressive vasoplegia, and myocardial dysfunction may be involved later [4]. In effect, from a clinical point of view, many patients despite been fluid loaded in pre-intensive care unit (ICU) settings, are still evidently hypovolemic and benefit from further administration of fluid boluses[5]. Others, however, present very low diastolic arterial pressures (DAP) reflecting profound vasoplegia, and recent data suggest that these patients may benefit from early norepinephrine (NE) instead of fluids [6,7]; on the contrary, administering fluids may fail to correct vascular tone and increase the risk of fluid overload [2]. In addition, a recent echocardiography-based study confirms that a relevant myocardial dysfunction is present in a significant number of patients, and that several cardiovascular phenotypes with a potentially different therapeutic approach may be recognized [8]. Unfortunately, despite research efforts [9,10], no universally applicable clinical phenotyping method for septic shock patients has been translated to usual practice. This is particularly problematic since echocardiography is not immediately available in the majority of centers worldwide, and therefore initial decisions on fluid resuscitation are usually based on clinical grounds and tend to follow the "one-size-fits-all" principle, leading to the risk of fluid overload [3].

II. ANDROMEDA-SHOCK Study

The skin territory lacks auto-regulatory flow control, and therefore, sympathetic activation impairs skin perfusion during circulatory dysfunction [11], a phenomenon that can be evaluated by peripheral perfusion assessment. A robust body of evidence confirms that abnormal peripheral perfusion after initial [12,13] or advanced [14,15] resuscitation is associated with increased morbidity and mortality. A cold clammy skin, mottling or prolonged capillary refill time (CRT) have been suggested as triggers for fluid resuscitation in patients with septic shock [5]. Moreover, the excellent prognosis associated with CRT recovery, its rapid-response time to fluid loading, its relative simplicity, its availability in resource-limited settings, and its capacity to change in parallel with perfusion of physiologically relevant territories such as the hepatosplanchnic region [16], constitute strong reasons to consider CRT as the target for fluid resuscitation in septic shock patients.

ANDROMEDA-SHOCK was a multicenter, randomized controlled trial comparing CRT- versus lactate-targeted resuscitation in patients with early septic shock [5]. The main outcome was 28-day mortality, and secondary outcomes included organ dysfunction and treatment intensity. The hypothesis was that targeting CRT would lead to decreased mortality and organ dysfunction due to less fluid administration and treatment intensity. The intervention period was of 8h and included sequential stepwise interventions starting with fluid challenges, followed by vasoactive-related interventions if necessary, until the target was reached (normalization of CRT versus normalization or decrease >20% every 2h in lactate levels). CRT was assessed with a standardized method in the 28 participating centers in Latin America.

CRT-targeted resuscitation was associated with lower mortality (34.9% vs. 43.4%), beneficial effects on organ dysfunction, and less intensity of treatment. The superiority of this strategy was also supported by a subsequent Bayesian analysis [17]. This landmark study has received considerable attention by the medical community, and the associated JAMA editorial proposes CRT assessment as a legitimate and adequate target to guide fluid resuscitation in the ICU and potentially in pre-ICU or resource-limited settings [18]. The worldwide impact and immediate application of CRT-targeted resuscitation makes further research an urgent task. The key novelty of A2 is to combine a CRT-targeted strategy with a clinical hemodynamic phenotyping that may aid to personalize initial resuscitation with potential additional fluid-sparing effects.

III. Rationale for ANDROMEDA-SHOCK-2 intervention strategy.

There are four relevant actions that may increase the efficacy of A2 intervention strategy to decrease mortality and organ dysfunction due to potential fluid-sparing effects or at least by promoting a more rational septic shock resuscitation.

1. Use of CRT as a target

A key aspect of ANDROMEDA-SHOCK was a more rational approach to fluid resuscitation based on selecting a more appropriate target such as CRT. In fact, almost a 25% of patients assigned to the CRT arm had a normal CRT at baseline, and thus received no further fluid resuscitation [10]. This fact is per se fluid sparing. In addition, a significant number of patients corrected CRT at 2h with initial fluid boluses. This rapid response of CRT to flow-increasing maneuvers makes CRT particularly suitable as a target. In addition, a recent clinical-physiological study suggests that normalization of CRT is associated to resolution of profound tissue hypoperfusion as assessed by hypoxia surrogates, and regional/sublingual microcirculatory variables [19]. For these reasons and other considerations described elsewhere [20], a CRT- targeted resuscitation will be applied as the intervention arm of A2.

2. Clinical hemodynamic phenotyping

Hypotension assessed by mean arterial pressure (MAP) is the hallmark of sepsis-related acute circulatory dysfunction, and since duration of hypotension is related to morbidity and mortality [21,22], current guidelines recommend fluids and vasopressors administered in a stepwise fashion to increase MAP levels to >65mmHg to ensure minimal organ perfusion pressure [3]. This "MAP-driven strategy" has probably lead to an unwanted side-effect, a reductionist approach to clinical hemodynamic monitoring where among numerous variables provided by the blood pressure signal, only MAP is considered for decision-making at the bedside. Moreover, this standardized or “one size fit all” resuscitation strategy is debatable since septic shock patients are highly heterogeneous. For example, a recent report showed that 30% of septic shock patients admitted to ICU were already fluid unresponsive [23], and others have suggested that early start of NE instead of fluids could be associated with better outcomes in predominantly vasoplegic patients [24].

The search for tools to identify hemodynamic phenotypes and personalize resuscitation has been highlighted as a priority in septic shock research agenda [25]. Indeed, a few recent studies have already reported the heterogeneity of hemodynamic and perfusion profiles [8,10]. Vieillard-Baron et al, could identify 5 hemodynamic clusters in a cohort of septic shock patients with transesophageal echocardiography, ranging from hypovolemia to left ventricular dysfunction despite similar MAP levels [8]. However, even though critical care echocardiography (CCE) has become a standard of care for septic shock [26], bedside 24/7 CCE is not available everywhere, particularly in resource-limited settings [27,28], hindering its immediate applicability. As an example, a delay in performing CCE for one-hour, a likely scenario in many ICUs around the world, may lead physicians to start the common "one-size-fits-all" approach, meaning continue pushing fluid boluses in the meantime without any physiological rationale.

Which variables to integrate into a clinical hemodynamic phenotyping.

Pulse pressure (PP) as a surrogate of stroke volume : Since early 1900’s, different researchers have tried to study the correlation between arterial PP and stroke volume [29]. It is an interesting target since arterial pressure is an easily accessible monitoring window into the heart function and its’ interactions with the vascular system. According to a 3-compartment Windkessel model, characteristics of the arterial system -one of the main determinants of PP along with stroke volume- will be determined by peripheral resistance, total arterial compliance and aortic characteristic impedance [30]. Mathematical derivations of this model provided the basis for cardiac output monitors based on pulse-contour analysis [31]. Moreover, multiple studies have shown, both in simulated [32] and real patients [33], and in different clinical scenarios [34] that PP can adequately track stroke volume. PP normally increases with age due to changes in aortic impedance, but broadly speaking, a PP <40 mmHg is clearly low and reflects a decrease in stroke volume, which could be explained either by a decreased preload or a severe cardiac dysfunction.

Diastolic arterial pressure: In normal conditions, DAP is mainly determined by vascular tone and it remains nearly constant from the ascending aorta to the peripheral vessels. Thus, detection of low DAP at peripheral vessels should reflect systemic vasodilation as long as the aortic valve is competent. However, DAP is not considered for septic shock definition, and with few exceptions, its relationship with clinical outcomes has not been widely described [35]. Nevertheless, evaluation of the loss of vascular tone through the severity of diastolic hypotension could have profound implications on therapeutic decisions since there are not robust clues to rapidly predict when hypotension will be sustainably corrected with fluid loading. Thus, rapid assessment of severity of vasodilation could influence therapeutic decisions such as the early introduction of NE [36], which theoretically may avoid unnecessary fluid administration while promptly restoring tissue perfusion. A low DAP (<50 mmHg) may impair the myocardial perfusion of the left ventricle (LV), especially in the case of tachycardia where diastolic time is limited [37].

Indeed, a fundamental physiological characteristic of DAP is to represent the upstream pressure for the perfusion of the LV, since it is only perfused during the diastole and not during the whole cardiac cycle, as it is the case for the right ventricle. Therefore, when the DAP is low (<50mmHg), as it is frequently the case in early septic shock, there is an increased risk of myocardial ischemia. This risk should be higher in patients with prior coronary artery disease (CAD) as the perfusion pressure is far lower downstream to the coronary artery stenosis compared to upstream. Indeed, a great proportion of patients with septic shock have cardiovascular comorbidities including known or unknown CAD, and more than 50% patients with septic shock have reduced coronary flow blood reserve [37] Therefore, it seems important to maintain the DAP above a certain level in septic shock patients in order to avoid an abrupt fall in the LV perfusion and ultimately the occurrence of myocardial ischemia, which could lead to a decrease in stroke volume, and systemic flow, thus further decreasing tissue and microcirculatory perfusion. Indeed, based on the simplicity and ease of application in clinical practice, DAP guided resuscitation may be considered in the early phase of septic shock [38].

3. Systematic and repeated assessment of fluid responsiveness.

The basis for the fluid resuscitation builds on the concept that sepsis and septic shock are conditions inducing absolute or relative hypovolemia due to a combination of external fluid losses, increased capillary leakage and pathological vasodilation. This is why fluid administration is considered a cornerstone in the resuscitation process to revert tissue hypoperfusion from septic shock [39]. Nevertheless, fluid administration is not free of adverse effects since volume expansion influences extra and intracellular electrolyte composition, acid-base equilibrium, and volume of distribution. Moreover, when excessively administered, fluids can induce interstitial edema that may then limit oxygen diffusion to the tissues [2]. In addition, excess of fluids can also interfere tissue perfusion by increasing downstream pressures and raising pressures surrounding capillary vessels. Indeed, observational studies suggest that larger volumes of resuscitation fluids and higher net fluid balances are associated with increased mortality rates in sepsis [2,40].

Fluid administration is the first line therapy to reverse sepsis-induced tissue hypoperfusion [3]. For this purpose, fluids are administered either as fluid loading at the emergency department [3], or later as fluid challenges during advanced ICU-based resuscitation [40]. However, as any other drug, fluids have a narrow therapeutic index. Insufficient fluid resuscitation may lead to progressive tissue hypoperfusion and organ dysfunction [20], while excess fluids could induce detrimental fluid overload.

Fluid responsiveness (FR) is a physiological cardiovascular condition where an increase in preload induced by a fluid bolus leads to an increase in cardiac output (CO) by more than 10–15% [41]. In non-fluid responsive (FR−) patients, fluid administration does not significantly increase CO and may contribute to congestion and fluid overload. The rationale to assess FR is then to try to optimize fluid resuscitation in critically ill patients by focusing fluid boluses in FR+hypoperfused patients and by preventing harmful fluid administration in FR− patients.

Multiple tests have been described to assess FR at the bedside [41]. They allow to determine the position of the patient’s heart on its systolic function curve, and are based upon changes in cardiac output or stroke volume resulting from various changes in preload conditions, induced by heart-lung interactions, postural maneuvers or by the infusion of small amounts of fluids. By applying the appropriate tests, FR can be assessed in a wide variety of clinical settings [42]. However, despite their relative simplicity, lack of cost, and side effects, the use of FR tests has not completely permeated into routine clinical practice [43]. One of the contributing factors is that even under the optimal conditions of their use, their sensitivity and specificity is not perfect, and all have significant limitations. This may be linked to the unreliability of the test, but also to the lack of precision of the measurement method used to estimate its effects. Assessment of FR is particularly difficult in spontaneous breathing patients [44]. In addition, it should be kept in mind that the relationship between cardiac preload and cardiac output is curvilinear, and is not a biphasic relationship. There is a continuum between the state of fluid responsiveness or unresponsiveness, and a grey zone within where decisions to administer fluids may be taken with more confidence the farthest away the results are from the reported validated cut-off value [39].

ANDROMEDA-SHOCK was the first major study that incorporated systematic per-protocol assessment of FR [45]. A post-hoc analysis of FR assessment in the trial found that FR status could be determined in 82% of early septic shock patients by using diverse tests depending on the clinical context; that 30% of patients were already non-fluid responsive before starting ICU-based resuscitation; and that despite receiving less fluids, non-fluid responders at baseline resolved hypoperfusion in a similar proportion than FR+ patients by following other steps of the protocol with no difference in clinically relevant outcomes. On the other hand, and despite current recommendations, passive leg raising (PLR) with assessment of PP (PLR-PP) was frequently used. This could be criticized since changes in PP during PLR have a low sensitivity although good specificity to assess FR [42]. Indeed, a positive test is reliable for detecting a FR+state, but a negative test is not, but some centers prefer to start with PLR-PP which is much faster and easier to be applied on a 24/7 basis especially in resource-limited settings [46]. In addition, the impact of assessing FR on congestion or fluid overload was not determined as it will be in A2.

In A2, FR assessment will be performed systematically in every patient with abnormal CRT randomized to the intervention arm and presenting a low PP. A chart with recommended techniques and limitations for each context will be provided (supplementary addendum). FR assessment will be repeated after every fluid bolus (500 ml of crystalloids in 30 minutes) to decide on further fluid administration if CRT target was not achieved. A FR-status or a safety issue may command to move to other steps of the intervention algorithm.

4. Selective Critical Care Echocardiography

The pathophysiology, prognostic value and potential management of myocardial dysfunction in septic shock has been extensively explored in recent years [47]. The widespread use of CCE has broadened our understanding of its incidence and severity, but particularly the landmark study from Vieillard-Baron et al., demonstrated the unequivocal coexistence of different cardiac phenotypes with potential relevant management implications [8]. Two published prospective databases from 12 different ICUs including echocardiographic monitoring performed by a transesophageal route at the initial phase of septic shock were merged for post hoc analysis. Hierarchical clustering in a principal components approach was used to define cardiovascular phenotypes using clinical and echocardiographic parameters. This clustering approach could characterize five different cardiovascular phenotypes but two are relevant for A2: patients with left ventricular (LV) systolic dysfunction (cluster 2, n = 64, 17.7%), and patients with right ventricular (RV) failure (cluster 4, n = 81, 22.5%). A left ventricular ejection fraction (LVEF) <40 and an aortic Velocity Time Integral (VTI) 0.6, and a central venous pressure (CVP)>8 mmHg [8] and was present in 22% of the patients. In this case, it is a signal to stop further fluid administration, and considering reducing PEEP levels and plateau pressure on the ventilator as recommended by current guidelines [49]. In A2 intervention algorithm, CCE will be obligatorily performed after failure of first tier interventions to correct CRT (see algorithm) to rule out the presence of severe LV systolic dysfunction, and RV failure (and command their specific management if detected). This approach is considered among basic skills by a recent CCE narrative review published by recognized experts in the field [49]. In addition, if already available CCE can aid to determine FR status at the top of the algorithm in patients with low PP. If at that moment, a significant cardiac dysfunction is diagnosed, the patient will proceed directly to the specific part of the algorithm (see above).

IV. Usual Care group

Recruited centers for A2 should exhibit a historical mortality for septic shock of 65 mmHg, HR94%, Hb > 7 gr/dl, and the use of NE as the first vasopressor and crystalloids as the fluid of choice. All data regarding insertion of invasive monitoring devices, intravenous-fluid resuscitation, vasoactive support, mechanical ventilation, and other supportive therapies will be collected by the study coordinator or monitors. Lead investigators at a site will not serve as the bedside treating physician for patients in the usual-care group.

V. Physiological substudies

Several substudies will be performed in all or part of the participating centers (see addendums) including venous congestion substudy (VExUS), sublingual microcirculation substudy, and renal doppler resistive index substudies. These substudies are highly relevant since they will compare the impact of A2 intervention strategy on effective organ perfusion microcirculatory flow, and venous congestion, thus determining if this resuscitative strategy is associated with a more harmonic resuscitation as expressed by a "perfusion without congestion" result.