Dataset Information

Individualized resuscitation strategy for septic shock formalized by finite mixture modeling and dynamic treatment regimen.

ABSTRACT:

Background

Septic shock comprises a heterogeneous population, and individualized resuscitation strategy is of vital importance. The study aimed to identify subclasses of septic shock with non-supervised learning algorithms, so as to tailor resuscitation strategy for each class.

Methods

Patients with septic shock in 25 tertiary care teaching hospitals in China from January 2016 to December 2017 were enrolled in the study. Clinical and laboratory variables were collected on days 0, 1, 2, 3 and 7 after ICU admission. Subclasses of septic shock were identified by both finite mixture modeling and K-means clustering. Individualized fluid volume and norepinephrine dose were estimated using dynamic treatment regime (DTR) model to optimize the final mortality outcome. DTR models were validated in the eICU Collaborative Research Database (eICU-CRD) dataset.

Results

A total of 1437 patients with a mortality rate of 29% were included for analysis. The finite mixture modeling and K-means clustering robustly identified five classes of septic shock. Class 1 (baseline class) accounted for the majority of patients over all days; class 2 (critical class) had the highest severity of illness; class 3 (renal dysfunction) was characterized by renal dysfunction; class 4 (respiratory failure class) was characterized by respiratory failure; and class 5 (mild class) was characterized by the lowest mortality rate (21%). The optimal fluid infusion followed the resuscitation/de-resuscitation phases with initial large volume infusion and late restricted volume infusion. While class 1 transitioned to de-resuscitation phase on day 3, class 3 transitioned on day 1. Classes 1 and 3 might benefit from early use of norepinephrine, and class 2 can benefit from delayed use of norepinephrine while waiting for adequate fluid infusion.

Conclusions

Septic shock comprises a heterogeneous population that can be robustly classified into five phenotypes. These classes can be easily identified with routine clinical variables and can help to tailor resuscitation strategy in the context of precise medicine.

SUBMITTER: Ma P

PROVIDER: S-EPMC8273991 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundThe decision of when to stop septic shock resuscitation is a critical but yet a relatively unexplored aspect of care. This is especially relevant since the risks of over-resuscitation with fluid overload or inotropes have been highlighted in recent years. A recent guideline has proposed normalization of central venous oxygen saturation and/or lactate as therapeutic end-points, assuming that these variables are equivalent or interchangeable. However, since the physiological determinants of both are totally different, it is legitimate to challenge the rationale of this proposal. We designed this study to gain more insights into the most appropriate resuscitation goal from a dynamic point of view. Our objective was to compare the normalization rates of these and other potential perfusion-related targets in a cohort of septic shock survivors.MethodsWe designed a prospective, observational clinical study. One hundred and four septic shock patients with hyperlactatemia were included and followed until hospital discharge. The 84 hospital-survivors were kept for final analysis. A multimodal perfusion assessment was performed at baseline, 2, 6, and 24 h of ICU treatment.ResultsSome variables such as central venous oxygen saturation, central venous-arterial pCO2 gradient, and capillary refill time were already normal in more than 70% of survivors at 6 h. Lactate presented a much slower normalization rate decreasing significantly at 6 h compared to that of baseline (4.0 [3.0 to 4.9] vs. 2.7 [2.2 to 3.9] mmol/L; p?<?0.01) but with only 52% of patients achieving normality at 24 h. Sublingual microcirculatory variables exhibited the slowest recovery rate with persistent derangements still present in almost 80% of patients at 24 h.ConclusionsPerfusion-related variables exhibit very different normalization rates in septic shock survivors, most of them exhibiting a biphasic response with an initial rapid improvement, followed by a much slower trend thereafter. This fact should be taken into account to determine the most appropriate criteria to stop resuscitation opportunely and avoid the risk of over-resuscitation.

Project description:BackgroundWe hypothesized that the cardiovascular responses to Surviving Sepsis Guidelines (SSG)-defined resuscitation are predictable based on the cardiovascular state.MethodsFifty-five septic patients treated by SSG were studied before and after volume expansion (VE), and if needed norepinephrine (NE) and dobutamine. We measured mean arterial pressure (MAP), cardiac index (CI), and right atrial pressure (Pra) and calculated pulse pressure and stroke volume variation (PPV and SVV), dynamic arterial elastance (Eadyn), arterial elastance (Ea) and left ventricular (LV) end-systolic elastance (Ees), Ees/Ea (VAC), LV ejection efficiency (LVeff), mean systemic pressure analogue (Pmsa), venous return pressure gradient (Pvr), and cardiac performance (Eh), using standard formulae.ResultsAll patients were hypotensive (MAP 56.8 ± 3.1 mmHg) and tachycardic (113.1 ± 7.5 beat min-1), with increased lactate levels (lactate = 5.0 ± 4.2 mmol L-1) with a worsened VAC. CI was variable but > 2 L min-1 M-2 in 74%. Twenty-eight-day mortality was 48% and associated with admission lactate, blood urea nitrogen (BUN), and creatinine levels but not cardiovascular state. In all patients, both MAP and CI improved following VE, as well as cardiac contractility (Ees). Fluid administration improved Pra, Pmsa, and Pvr in all patients, whereas both HR and Ea decreased after VE, thus normalizing VAC. CI increases were proportional to baseline PPV and SVV. CI increases proportionally decreased PPV and SVV. VE increased MAP > 65 mmHg in 35/55 patients. MAP responders had higher PPV, SVV, and Eadyn than non-responders. NE was given to 20/55 patients in septic shock, but increased MAP > 65 mmHg in only 12. NE increased Ea, Eadyn, Pra, Pmsa, and VAC while decreasing HR, PPV, SVV, and LVeff. MAP responders had higher pre-NE Ees and lower VAC. Dobutamine was given to 6/8 patients who remained hypotensive following NE. It increased Ees, MAP, CI, and LVeff, while decreasing HR, Pra, and VAC. At all times and all steps of the protocol, CI changes were proportional to Pvr changes independent of treatment.ConclusionsThe cardiovascular response to SSG-based resuscitation is highly heterogeneous but predictable from pre-treatment measures of cardiovascular state.

Project description:Septic shock is the leading cause of cardiovascular failure in the intensive care unit (ICU). Cardiac output is a primary component of global oxygen delivery to organs and a sensitive parameter of cardiovascular failure. Any mismatch between oxygen delivery and rapidly varying metabolic demand may result in tissue dysoxia, hence organ dysfunction. Since the intricate alterations of both vascular and cardiac function may rapidly and widely change over time, cardiac output should be measured repeatedly to characterize the type of shock, select the appropriate therapeutic intervention, and evaluate patient's response to therapy. Among the numerous techniques commercially available for measuring cardiac output, transpulmonary thermodilution (TPT) provides a continuous monitoring with external calibration capability, whereas critical care echocardiography (CCE) offers serial hemodynamic assessments. CCE allows early identification of potential sources of inaccuracy of TPT, including right ventricular failure, severe tricuspid or left-sided regurgitations, intracardiac shunt, very low flow states, or dynamic left ventricular outflow tract obstruction. In addition, CCE has the unique advantage of depicting the distinct components generating left ventricular stroke volume (large cavity size vs. preserved contractility), providing information on left ventricular diastolic properties and filling pressures, and assessing pulmonary artery pressure. Since inotropes may have deleterious effects if misused, their initiation should be based on the documentation of a cardiac dysfunction at the origin of the low flow state by CCE. Experts widely advocate using CCE as a first-line modality to initially evaluate the hemodynamic profile associated with shock, as opposed to more invasive techniques. Repeated assessments of both the efficacy (amplitude of the positive response) and tolerance (absence of side-effect) of therapeutic interventions are required to best guide patient management. Overall, TPT allowing continuous tracking of cardiac output variations and CCE appear complementary rather than mutually exclusive in patients with septic shock who require advanced hemodynamic monitoring.

Project description:IntroductionRecent evidence suggests that early, aggressive resuscitation in patients with septic shock reduces mortality. The objective of this survey was to characterize reported resuscitation practices of Canadian physicians caring for adult critically ill patients with early septic shock.MethodsA scenario-based self-administered national survey was sent out to Canadian critical care physicians. One hypothetical scenario was developed to obtain information on several aspects of resuscitation in early septic shock, including monitoring and resuscitation end-points, fluid administration, red blood cell transfusion triggers, and use of inotropes. The sampling frame was physician members of Canadian national and provincial critical care societies.ResultsThe survey response rate was 232 out of 355 (65.3%). Medicine was the most common primary specialty (60.0%), most respondents had practiced for 6 to 10 years (30.0%), and 82.0% were male. The following monitoring devices/parameters were reported as used/measured 'often' or 'always' by at least 89% of respondents: oxygen saturation (100%), Foley catheters (100%), arterial blood pressure lines (96.6%), telemetry (94.3%), and central venous pressure (89.2%). Continuous monitoring of central venous oxygen saturation was employed 'often' or 'always' by 9.8% of respondents. The two most commonly cited resuscitation end-points were urine output (96.5%) and blood pressure (91.8%). Over half of respondents used normal saline (84.5%), Ringers lactate (52.2%), and pentastarch (51.3%) 'often' or 'always' for early fluid resuscitation. In contrast, 5% and 25% albumin solutions were cited as used 'often' or 'always' by 3.9% and 1.3% of respondents, respectively. Compared with internists, surgeons and anesthesiologists (odds ratio (95% confidence interval): 9.8 (2.9 to 32.7) and 3.8 (1.7 to 8.7), respectively) reported greater use of Ringers lactate. In the setting of a low central venous oxygen saturation, 52.5% of respondents reported use of inotropic support 'often' or 'always'. Only 7.6% of physicians stated they would use a red blood cell transfusion trigger of 100 g/l to optimize oxygen delivery further.ConclusionOur survey results suggest that there is substantial practice variation in the resuscitation of adult patients with early septic shock. More randomized trials are needed to determine the optimal approach.

Project description:BACKGROUND:Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection. Numerous studies have explored the complex and dynamic transcriptome modulations observed in sepsis patients, but a large fraction of the transcriptome remains unexplored. This fraction could provide information to better understand sepsis pathophysiology. Multiple levels of interaction between human endogenous retroviruses (HERV) and the immune response have led us to hypothesize that sepsis is associated with HERV transcription and that HERVs may contribute to a signature among septic patients allowing stratification and personalized management. METHODS:We used a high-density microarray and RT-qPCR to evaluate the HERV and Mammalian Apparent Long Terminal Repeat retrotransposons (MaLR) transcriptome in a pilot study that included 20 selected septic shock patients, stratified on mHLA-DR expression, with samples collected on day 1 and day 3 after inclusion. We validated the results in an unselected, independent cohort that included 100 septic shock patients on day 3 after inclusion. We compared septic shock patients, according to their immune status, to describe the transcriptional HERV/MaLR and conventional gene expression. For differential expression analyses, moderated t tests were performed and Wilcoxon signed-rank tests were used to analyze RT-qPCR results. RESULTS:We showed that 6.9% of the HERV/MaLR repertoire was transcribed in the whole blood, and septic shock was associated with an early modulation of a few thousand of these loci, in comparison to healthy volunteers. We provided evidence that a subset of HERV/MaLR and conventional genes were differentially expressed in septic shock patients, according to their immune status, using monocyte HLA-DR (mHLA-DR) expression as a proxy. A group of 193 differentially expressed HERV/MaLR probesets, tested in an independent septic shock cohort, identified two groups of patients with different immune status and severity features. CONCLUSION:We demonstrated that a large, unexplored part of our genome, which codes for HERV/MaLR, may be linked to the host immune response. The identified set of HERV/MaLR probesets should be evaluated on a large scale to assess the relevance of these loci in the stratification of septic shock patients. This may help to address the heterogeneity of these patients.

Project description:BACKGROUND:Fluid boluses are administered to septic shock patients with the purpose of increasing cardiac output as a means to restore tissue perfusion. Unfortunately, fluid therapy has a narrow therapeutic index, and therefore, several approaches to increase safety have been proposed. Fluid responsiveness (FR) assessment might predict which patients will effectively increase cardiac output after a fluid bolus (FR+), thus preventing potentially harmful fluid administration in non-fluid responsive (FR-) patients. However, there are scarce data on the impact of assessing FR on major outcomes. The recent ANDROMEDA-SHOCK trial included systematic per-protocol assessment of FR. We performed a post hoc analysis of the study dataset with the aim of exploring the relationship between FR status at baseline, attainment of specific targets, and clinically relevant outcomes. METHODS:ANDROMEDA-SHOCK compared the effect of peripheral perfusion- vs. lactate-targeted resuscitation on 28-day mortality. FR was assessed before each fluid bolus and periodically thereafter. FR+ and FR- subgroups, independent of the original randomization, were compared for fluid administration, achievement of resuscitation targets, vasoactive agents use, and major outcomes such as organ dysfunction and support, length of stay, and 28-day mortality. RESULTS:FR could be determined in 348 patients at baseline. Two hundred and forty-two patients (70%) were categorized as fluid responders. Both groups achieved comparable successful resuscitation targets, although non-fluid responders received less resuscitation fluids (0 [0-500] vs. 1500 [1000-2500] mL; p 0.0001), exhibited less positive fluid balances, but received more vasopressor testing. No difference in clinically relevant outcomes between FR+ and FR- patients was found, including 24-h SOFA score (9 [5-12] vs. 8 [5-11], p = 0.4), need for MV (78% vs. 72%, p = 0.16), need for RRT (18% vs. 21%, p = 0.7), ICU-LOS (6 [3-11] vs. 6 [3-16] days, p = 0.2), and 28-day mortality (40% vs. 36%, p = 0.5). Only thirteen patients remained fluid responsive along the intervention period. CONCLUSIONS:Systematic assessment allowed determination of fluid responsiveness status in more than 80% of patients with early septic shock. Fluid boluses could be stopped in non-fluid responsive patients without any negative impact on clinical relevant outcomes. Our results suggest that fluid resuscitation might be safely guided by FR assessment in septic shock patients. TRIAL REGISTRATION:ClinicalTrials.gov identifier, NCT03078712. Registered retrospectively on March 13, 2017.

Project description:Background:Assessment of tissue hypoxia at the bedside has yet to be translated into daily clinical practice in septic shock patients. Perfusion markers are surrogates of deeper physiological phenomena. Lactate-to-pyruvate ratio (LPR) and the ratio between veno-arterial PCO2 difference and Ca-vO2 (?PCO2/Ca-vO2) have been proposed as markers of tissue hypoxia, but they have not been compared in the clinical scenario. We studied acute septic shock patients under resuscitation. We wanted to evaluate the relationship of these hypoxia markers with clinical and biochemical markers of hypoperfusion during septic shock resuscitation. Methods:Secondary analysis of a randomized controlled trial. Septic shock patients were randomized to fluid resuscitation directed to normalization of capillary refill time (CRT) versus normalization or significant lowering of lactate. Multimodal assessment of perfusion was performed at 0, 2, 6 and 24 hours, and included macrohemodynamic and metabolic perfusion variables, CRT, regional flow and hypoxia markers. Patients who attained their pre-specified endpoint at 2-hours were compared to those who did not. Results:Forty-two patients were recruited, median APACHE-II score was 23 [15-31] and 28-day mortality 23%. LPR and ?PCO2/Ca-vO2 ratio did not correlate during early resuscitation (0-2 h) and the whole study period (24-hours). ?PCO2/Ca-vO2 ratio derangements were more prevalent than LPR ones, either in the whole cohort (52% vs. 23%), and in association with other perfusion abnormalities. In patients who reached their resuscitation endpoints, the proportion of patients with altered ?PCO2/Ca-vO2 ratio decreased significantly (66% to 33%, P=0.045), while LPR did not (14% vs. 25%, P=0.34). Conclusions:Hypoxia markers did not exhibit correlation during resuscitation in septic shock patients. They probably interrogate different pathophysiological processes and mechanisms of dysoxia during early septic shock. Future studies should better elucidate the interaction and clinical role of hypoxia markers during septic shock resuscitation.