Abstract
Objective:
Metabolic syndrome is known to predict outcomes in COVID-19 Acute Respiratory Distress Syndrome (ARDS), but has never been studied in non-COVID-19 ARDS. We therefore aimed to determine the association of metabolic syndrome with mortality among ARDS trial subjects.
Design:
Retrospective cohort study of ARDS trials’ data.
Setting:
An ancillary analysis was conducted using data from seven ARDS Network and Prevention and Early Treatment of Acute Lung Injury Network randomized trials within the Biologic Specimen and Data Repository Information Coordinating Center database.
Patients:
Hospitalized patients with ARDS and metabolic syndrome (defined by obesity, diabetes, and hypertension) were compared to similar patients without metabolic syndrome (those with less than three criteria).
Interventions:
None.
Measurements and Main Results:
The primary outcome was 28-day mortality. Among 4,288 ARDS trial participants, 454 (10.6%) with metabolic syndrome were compared to 3,834 (89.4%) controls. In adjusted analyses, the metabolic syndrome group was associated with lower 28-day and 90-day mortality when compared with control (adjusted odds ratio [aOR], 0.70 [95% CI, 0.55–0.89] and 0.75 [95% CI, 0.60–0.95], respectively). With each additional metabolic criterion from 0 to 3, adjusted 28-day mortality was reduced by 18%, 22%, and 40%, respectively. In subgroup analyses stratifying by ARDS etiology, mortality was lower for metabolic syndrome vs. control in ARDS caused by sepsis or pneumonia (at 28 days, aOR 0.64 [95%CI, 0.48-0.84] and 90 days, aOR 0.69 [95%CI, 0.53-0.89]), but not in ARDS from non-infectious causes (at 28 days, aOR 1.18 [95% CI, 0.70–1.99] and 90 days, aOR 1.26 [95% CI, 0.77–2.06]). Interaction p=0.04 and p=0.02 for 28- and 90-day comparisons, respectively.
Conclusions:
Metabolic syndrome in ARDS was associated with a lower risk of mortality in non-COVID-19 ARDS. The relationship between metabolic inflammation and ARDS may provide a novel biological pathway to be explored in precision medicine-based trials.
Keywords: Metabolic Syndrome, Acute Lung Injury, Pneumonia, Sepsis, Obesity
Introduction:
Acute Respiratory Distress Syndrome (ARDS) is a life-threatening, inflammatory lung injury characterized by the dysregulation of innate immune responses and cell-mediated damage to alveolar barriers.(1) This condition results in severe hypoxemia requiring invasive mechanical ventilation and death in over 40% of cases.(2) ARDS accounts for 10% of all intensive care unit (ICU) admissions, and has increased dramatically as a result of SARS-CoV-2.(3, 4) Despite advances in the understanding of ARDS, clinical heterogeneity remains a substantial challenge and is widely considered to be the primary reason that numerous randomized clinical trials (RCT) of novel therapeutics have failed to show benefit.(5) Recent studies suggest ARDS patients may be classified into a hyper- or hypo-inflammatory subphenotype which may be associated with differential treatment responses in secondary analyses of completed RCTs.(6, 7) However, biologic plausibility is yet to be fully demonstrated, and there remains a need for mechanistically-informed subphenotype delineation.
Metabolic syndrome is a clinical syndrome characterized by obesity, insulin resistance, hypertension, and dyslipidemia.(8) The innate immune dysfunction and chronic adipose tissue inflammation seen in poor metabolic health is described as “meta-inflammation” and shares many pathophysiologic traits with ARDS, including macrophage-driven proinflammatory cytokine production and endothelial dysfunction.(1, 9–11) COVID-19, in particular, has highlighted the relationship between poor metabolic health and critical illness.(3, 12) Although obesity(13–15) and diabetes(16–18) have been studied individually in non-COVID-19 ARDS, overall poor metabolic health, such as that seen with metabolic syndrome, has not been studied.
We hypothesized that by clustering metabolic syndrome risk factors to identify patients with likely meta-inflammation, we can predict prognosis for patients with ARDS. The objective of this study was to determine the association of metabolic syndrome, as a surrogate for poor metabolic health, with ARDS-related mortality and outcomes.
Materials and Methods:
Study design, setting, and participants
This was an ancillary analysis conducted using data from subjects enrolled in seven ARDS Network (ARDSNet) and Prevention and Early Treatment of Acute Lung Injury (PETAL) Network trials listed in eMethods. Details of each trial were previously published,(19–25) but included patients with ARDS identified by American-European Consensus Conference criteria or Berlin criteria.(26) Data were collected from the National Heart, Lung, and Blood Institute (NHLBI) Biologic Specimen and Data Repository Information Coordinating Center (BioLINCC) database. Raw patient-level data from each study were reviewed and collated, forming a new dataset that included patient demographics, outcomes, body mass index (BMI), comorbidity categories (cancer, hepatic/gastrointestinal, immunocompromised, pulmonary, renal), primary etiologies of ARDS (sepsis, pneumonia, trauma, aspiration, transfusion, other), and treatment group assignment. The protocol, “The relationship between metabolic syndrome and acute respiratory distress syndrome: an ancillary analysis of the ARDSNet and PETAL databases,” was reviewed by the NHLBI Data Repository Program Officer and Tulane University Biomedical Institutional Review Board (#2021-1240; January 03, 2023) who determined activities of this study were not human subjects research as defined by the Common Federal Rule. Procedures followed in this study were in accordance with the ethical standards of the Tulane Institutional Review Board and Helsinki Declaration of 1975. This study is reported following Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.
The metabolic syndrome cohort, identified by the combination of obesity, diabetes, and high blood pressure, was modified from World Health Organization criteria and our previously described metabolic syndrome criteria in COVID-19 ARDS.(3, 8) Metabolic criteria were identified by BMI ≥ 30 kg/m2, history of diabetes, and a history of high blood pressure. Control patients were classified as any participant not meeting all three criteria. Dyslipidemia and prediabetes were not included as these data were not available. Also, since medical history of hypertension was not reported for the Statins for Acutely Injured Lungs from Sepsis (SAILS) trial, we used alternative criteria, systolic >130 mmHg and diastolic >85 mmHg on two occasions, to identify high blood pressure for participants in this trial.(27) A sensitivity analysis excluding SAILS was performed to support this approach. Prespecified subgroups were created to study patients with metabolic criteria treated equally (i.e., one, two, or three of three criteria) as well as individual metabolic criterion in isolation (i.e., obesity, diabetes, or hypertension).
Outcomes
The objective of this study was to determine the association of metabolic syndrome with ARDS-related mortality and outcomes. The primary outcome was 28-day mortality. Secondary outcomes included 90-day mortality, hospital and ICU length of stay, ICU-free days to day 28, ventilator-free days (VFDs) to day 28, and organ-failure free days to day 28. Detailed outcome descriptions are provided in eMethods. All outcomes were compared between metabolic syndrome and control groups for the main analyses.
Additionally, stratified analyses were conducted to estimate effect modification of ARDS etiology on the association of metabolic syndrome with mortality. As diabetes was previously found to be more common among sepsis-related ARDS (18), we hypothesized that metabolic syndrome may impact ARDS outcomes differently in those with infection-related ARDS compared to those with non-infectious ARDS. To test this hypothesis, analyses stratifying for primary ARDS etiology (e.g. those due to sepsis/pneumonia as compared to all other etiologies) were performed comparing primary and secondary outcomes.
Lastly, the primary outcome and 90-day mortality were compared in prespecified subgroups to study the different criteria used to identify metabolic syndrome. First, participants with each metabolic criteria treated equally (i.e., one, two, or three of three criteria) were compared to non-metabolic participants possessing none of the criteria. Second, participants with each individual metabolic risk factor in isolation (i.e., obesity, diabetes, or hypertension) were compared to those without the respective risk factor (i.e., diabetes vs no diabetes, etc.) to investigate which conditions, if any, may be associated with greater risk.
Sample Size Calculation
The sample size was based on the ability to detect a significant difference between metabolic syndrome and control for the risk of death. Because this study was an ancillary analysis, the sample size assessment was used to determine an adequate convenience sample size using pilot results from studies on metabolic syndrome and COVID-19.(3, 12) Mortality estimates for metabolic and control subjects were 20% and 16%, respectively. Based on these estimates and a one-to-three ratio of metabolic subjects to controls, 3,788 observations were needed to detect a significant difference in mortality with 80% statistical power and a two-sided type I error threshold of 5%. Since BioLINCC maintained available data on approximately 4,763 subjects, the study exceeded these minimum thresholds.
Statistical Analysis
Baseline characteristics were expressed as means and standard deviations, medians and interquartile ranges, or frequencies and percentages. Summary statistics and statistical models were based on complete case analysis because missing data were minimal and assumed to be missing at random (eTable 1). Additionally, multiple imputation was used to create 10 imputed datasets to account for missing metabolic syndrome values. Imputed data sets were analyzed separately using similar models to the main analyses and were reported as a pooled outcome across imputations. For the main analyses and those in the individual metabolic criteria subgroups, multivariable logistic regression models were used adjusting for age, sex, race, ethnicity, comorbidities, primary ARDS etiology, and treatment group assignment for each trial. The same predictor set was used in multivariate linear regression models with hospital length of stay as the dependent variable. Covariates included were prespecified using background knowledge of the factors connecting exposure to outcome. To determine effect modification for patients stratified by ARDS etiology, similar models were created, each for infection-related and non-infection-related subgroups, adjusting for age, sex, race, ethnicity, comorbidities, and treatment group assignment. Heterogeneity testing to determine an interaction was assessed according to Cochran’s Q and I2 tests.(28) Statistical analyses were performed using SAS Enterprise Guide (version 6.1, SAS Institute, Inc., Cary, NC), and SAS (version 9.4, SAS Institute, Cary, NC). Forest plots were generated using the STATA workflow and Metaprop command (version 16.1, StatCorp LLC, College, Station, TX). P values were 2-sided, and type I error was set at 5%.
Results:
Study Population
Among 4,763 observations maintained in the BioLINCC database, 118 (2.5%) observations were duplicated for subjects enrolled in multiple studies, and 357 (7.7%) were missing a component necessary for determining the metabolic syndrome group (261 [5.6%], 145 [3.1%], and 80 [1.7%] for hypertension, obesity, and diabetes, respectively) as seen in eTable 1. The final data set included 4,288 observations for complete case analyses (Figure 1). Baseline characteristics are presented in Table 1. The mean (SD) age was 52.7 (16.2) years with 2,039 (47.5%) females. The full cohort included 3,123 (72.8%) participants self-identifying as White, 688 Black (16.0%), and 336 Hispanic (7.8%). Overall, the most common metabolic criterion was hypertension (1,874 participants, 43.7%), then obesity (1,818 participants, 42.4%) and diabetes (1,025 participants, 23.9%). Primary ARDS etiologies were most frequently identified as pneumonia (2,408 participants, 56.2%) and sepsis (797 participants, 18.6%). Results were similar when using multiple imputation (eTable 2).
Figure 1: Cohort of patients in ARDSnet and PETAL Databases.
ARDS=Acute Respiratory Distress Syndrome; PETAL= Prevention & Early Treatment of Acute Lung Injury; ALTA=Aerosolized Albuterol versus placebo for the Treatment of Acute lung injury; ALVEOLI=Assessment of Low tidal Volume and elevated End-expiratory volume to Obviate Lung Injury; EDEN=Early versus Delayed Enteral Feeding, FACTT=Fluid And Catheter Treatment Trial; LaSRS Late Steroid Rescue Study; PETAL-ROSE=Prevention and Early Treatment of Acute Lung Injury-Reevaluation Of Systemic Early neuromuscular blockade; SAILS=Statins for Acutely Injured Lungs from Sepsis.
Table 1:
Baseline Characteristics, Metabolic Syndrome vs. Control
| Characteristics | Total (n=4,288) |
Metabolic (n=454) |
Control (n=3,834) |
|---|---|---|---|
| Age, mean years ± SD | 52.7 ± 16.2 | 58.6 ± 11.8 | 52.0 ± 16.5 |
| Female, No. (%) | 2,039 (47.5) | 253 (55.7) | 1,786 (46.6) |
| Race, No. (%) | |||
| Black | 688 (16.0) | 98 (21.6) | 590 (15.4) |
| White | 3,123 (72.8) | 302 (66.5) | 2,821 (73.6) |
| Other/Not Reported | 477 (11.1) | 54 (11.9) | 423 (11.0) |
| Ethnicity, No. (%) | |||
| Hispanic/Latinx | 336 (7.8) | 62 (13.7) | 274 (7.2) |
| Not Hispanic/Latinx | 2,479 (57.8) | 287 (63.2) | 2,192 (57.2) |
| Unknown/Not Reported | 1,473 (34.4) | 105 (23.1) | 1,368 (35.7) |
| BMI a , mean ± SD | 29.9 ± 8.4 | 38.8 ± 7.3 | 28.8 ± 7.9 |
| Metabolic criteria, No. (%) | |||
| Diabetesb | 1,026 (23.9) | 454 (100.0) | 572 (14.9) |
| Obesityc | 1,818 (42.4) | 454 (100.0) | 1,364 (35.6) |
| Hypertensiond | 1,874 (43.7) | 454 (100.0) | 1,420 (37.0) |
| Primary ARDS Etiology, No. (%) | |||
| Sepsis | 797 (18.6) | 110 (24.2) | 687 (17.9) |
| Pneumonia | 2,408 (56.2) | 254 (56.0) | 2,154 (56.2) |
| Aspiration | 547 (12.8) | 38 (8.4) | 509 (13.3) |
| Transfusion | 77 (1.8) | 6 (1.3) | 71 (1.9) |
| Trauma | 205 (4.8) | 11 (2.4) | 194 (5.1) |
| Othere | 254 (5.9) | 35 (7.7) | 219 (5.7) |
| Comorbidities, No. (%) | |||
| Cancer | 293 (6.8) | 21 (4.6) | 272 (7.1) |
| Hepatic/Gastrointestinal | 440 (10.3) | 46 (10.1) | 394 (10.3) |
| Immunocompromised | 651 (15.2) | 48 (10.6) | 603 (15.7) |
| Pulmonary | 502 (11.7) | 92 (20.3) | 410 (10.7) |
| Renal | 112 (2.6) | 17 (3.7) | 95 (2.5) |
Body Mass Index (BMI) is weight (kilograms) divided by the square of height (meters).
History of diabetes.
BMI ≥30kg/m2.
History of hypertension or high blood pressure.
Other category included primary etiologies of ARDS determined by investigators to not be primarily caused by sepsis, pneumonia, trauma, aspiration, or transfusion. SD=Standard deviation. No=Number.
Metabolic Syndrome versus Control Cohort Outcomes
A total of 454 (10.6%) participants met eligibility for metabolic syndrome, and 3,834 (89.4%) were included as control. Participants with metabolic syndrome, as compared to control, were more commonly older (mean [SD] age, 58.6 [11.8] years versus 52.0 [16.5] years), female (253 [55.7%] versus 1,786 [46.6%]), and Black (98 [21.6%], versus 590 [15.4%]). BMI was higher for metabolic syndrome versus control (mean [SD], 38.8 [7.3] versus 28.8 [7.9]), as was the presence of pulmonary disease as a comorbidity (92 [20.3%] versus 410 [10.7%]). Participants with metabolic syndrome versus control were more likely to have ARDS from sepsis (24.2% versus 17.9%, respectively), but less likely to have ARDS from aspiration (8.4% versus 13.3%, respectively) or trauma (2.4% versus 5.1%, respectively).
In primary analyses shown in Table 2, 106 (23.4%) participants with metabolic syndrome died within 28 days, compared with 977 (25.5%) control (unadjusted odds ratio [OR], 0.89 [95% CI, 0.71–1.12]). In multivariable analyses, metabolic syndrome was associated with significantly lower 28- and 90-day mortality (adjusted OR [aOR] 0.70 [95% CI, 0.55–0.89] and aOR 0.75 [95% CI, 0.60–0.95], respectively). Tests for heterogeneity across individual trials were insignificant (eFigure 1). Other secondary outcome analyses comparing metabolic syndrome with control were not statistically significant. Sensitivity analyses using a dataset generated with multiple imputation and a dataset excluding SAILS were consistent with the primary analyses and are reported in eTable 3 and eTable 4, respectively.
Table 2:
Outcomes in Metabolic Syndrome vs. Control
| Outcomes | Metabolic | Control | Unadjusted OR (95% CI) |
Adjusted ORa (95% CI) |
P-valuec |
|---|---|---|---|---|---|
| Mortality, No. (%) | |||||
| 28-day | 106 (23.4) | 977 (25.5) | 0.89 (0.71–1.12) | 0.70 (0.55–0.89) | 0.004 |
| 90-day | 134 (29.5) | 1,156 (30.2) | 0.97 (0.78–1.20) | 0.75 (0.60–0.95) | 0.017 |
| Hospital Length of Stay, days | Metabolic | Control | Adjusted Difference (95% CI)b | ||
| Hospital, median (IQR) | 17 (10, 26) | 15 (10, 26) | 1.26 (−0.59–3.12) | 0.183 | |
| Hospital, mean ± SD | 20.9 ± 14.9 | 20.2 ± 15.5 | 0.57 (−1.18–2.32) | 0.524 | |
| ICU, median (IQR) | 9 (6, 16) | 9 (5, 15) | 0.23 (−0.74–1.19) | 0.648 | |
| ICU, mean ± SD | 12.3 ± 10.3 | 12.2 ± 10.6 | 0.12 (−1.04–1.27) | 0.843 | |
| ICU Free Days (at 28 days), mean ± SD | 5.1 ± 6.8 | 4.4 ± 6.3 | 0.04 (−0.55–0.63) | 0.900 | |
| Ventilator Free Days (at 28 days), mean ± SD | 13.7 ± 10.4 | 14.4 ± 10.6 | 0.57 (−0.46–1.61) | 0.279 | |
| Organ Failure Free Days (at 28 days), mean ± SD | 9.3 ± 9.8 | 9.3 ± 9.6 | −0.08 (−0.92–1.07) | 0.883 | |
Multivariable logistic regression model adjusted for age, sex, race, ethnicity, comorbidities, primary ARDS etiology, and treatment group assignment.
Multivariable linear regression model adjusted for age, sex, race, ethnicity, comorbidities, primary ARDS etiology, and treatment group assignment.
P-value for adjusted analyses. OR=Odds ratio; IQR=Inter-quartile range; ICU=Intensive care unit; SD=Standard deviation.
Stratified analyses comparing ARDS caused by sepsis and pneumonia to other non-infectious etiologies are shown in Table 3 and eTable 5. Metabolic syndrome participants with a primary ARDS etiology of sepsis or pneumonia experienced significantly lower adjusted 28-day mortality (aOR 0.64 [95% CI, 0.48–0.84]) and 90-day mortality (aOR 0.69 [95% CI, 0.53–0.89]) as compared to control. However, in non-infectious ARDS, there were no statistically significant differences between metabolic syndrome and control for 28-day (aOR 1.18 [95% CI, 0.70–1.99]) or 90-day (aOR 1.26 [95% CI, 0.77–2.06]) mortality. Tests for heterogeneity were significant for an interaction for both 28-day (unadjusted p= 0.02, adjusted p=0.04) and 90-day mortality (unadjusted p<0.01, adjusted p=0.02).
Table 3:
Sepsis/Pneumonia vs. other ARDS etiologies
| 28-Day Mortality | ||||||||
|---|---|---|---|---|---|---|---|---|
| ARDS Etiologies | Metabolic, No. (%) | Control, No. (%) | Unadjusted OR (95% CI) | Adjusted ORa(95% CI) | P-valuec | Unadjusted Interaction P-valued | Adjusted Interaction P-valued | |
| Sepsis/Pneumonia | 82 (22.5) | 783 (27.6) | 0.76 (0.59–0.99) | 0.64 (0.48–0.84) | 0.001 | 0.02 | 0.04 | |
| Other Etiologies | 24 (26.7) | 194 (19.5) | 1.50 (0.92–2.45) | 1.18 (0.70–1.99) | 0.546 | |||
| 90-Day Mortality | ||||||||
| ARDS Etiologies | Metabolic, No. (%) | Control, No. (%) | Unadjusted OR (95% CI) | Adjusted ORa (95% CI) | P-valuec | Unadjusted Interaction P-valued | Adjusted Interaction P-valued | |
| Sepsis/Pneumonia | 103 (28.3) | 919 (32.4) | 0.83 (0.65–1.05) | 0.69 (0.53–0.89) | 0.005 | <0.01 | 0.02 | |
| Other Etiologies | 31 (34.4) | 237 (23.9) | 1.68 (1.06–2.65) | 1.26 (0.77–2.06) | 0.366 | |||
Multivariable logistic regression model adjusted for age, sex, race, ethnicity, comorbidities, and treatment group assignment.
Other ARDS etiologies include all primary etiologies other than sepsis or pneumonia.
P-value for adjusted analyses comparing metabolic to control.
P-values for comparing the interaction between metabolic and control across the two ARDS etiology strata of sepsis/pneumonia and other ARDS etiology. OR=Odds Ratio.
Individual Metabolic Syndrome Criteria Subgroup Analyses
Baseline characteristics for metabolic criteria subgroups are portrayed in eTable 6. When each criterion was considered similarly, there was a high prevalence of metabolic disease with 2,831 (66.0%) participants carrying at least one metric of poor metabolic health (1,457 [34.0%] met 0/3 criteria, 1,398 [32.6%] met 1/3 criteria, 979 [22.8%] met 2/3 criteria, 454 [10.2%] met 3/3 criteria). As the number of criteria increased from zero to one, two, or three, so did the rates of sepsis as the cause of ARDS (17.0%, 17.9%, 19.4%, and 24.2%, respectively).
As shown in Figure 2 and eTable 7, 28-day and 90-day mortality were compared between patients with escalating metabolic criteria (one, two, or three of three criteria), and the non-metabolic zero of three criteria subgroup. In adjusted analyses, the odds of dying were progressively lower with additional metabolic criterion as compared to a non-metabolic subgroup at both 28 days (aOR: 0.82 [95% CI 0.68-0.99], 0.78 [95% CI 0.64-0.96], and 0.60 [95% CI 0.46-0.79], respectively) and 90 days (aOR: 0.86 [95% CI 0.72-1.03], 0.82 [95% CI 0.68-1.00], 0.67 [95% CI 0.52-0.86], respectively).
Figure 2: Individual Metabolic subgroup analyses on 28-day and 90-day mortality.
Multivariable models include age, sex, race, ethnicity, comorbidities, primary ARDS etiology, and treatment group. Panel A: Subgroups were prepared to determine risks associated with each additional metabolic syndrome criteria treated equally (i.e., 1, 2, or 3 of 3 criteria) compared to a non-metabolic subgroup possessing 0 of the 3 metabolic criteria. Panel B: Metabolic syndrome criteria considered as individual risk factors (i.e., diabetes, obesity, hypertension) and comparisons made to participants without the respective risk factor (i.e., diabetes compared to no diabetes). None of the outcome-specific tests for heterogeneity yielded significant results.
Second, each metabolic criterion was also considered a unique risk factor (i.e., diabetes, obesity, hypertension) to determine if any specific criterion contributed to a greater extent. Comparisons were made between participants with each risk factor and participants who did not carry the respective risk factor (i.e., diabetes vs no diabetes, etc.). Multivariable analyses show that hypertension was independently associated with lower 28-day mortality (aOR 0.72 [95% CI, 0.62-0.85]) and 90-day mortality (aOR 0.78 [95% CI, 0.67–0.91]), obesity was independently associated with lower 90-day mortality (aOR 0.85 [95% CI, 0.73–0.98]), and diabetes was independently associated with lower 28-day mortality (aOR 0.83 [95% CI, 0.70–0.99]).
Discussion:
In this ancillary analysis of 4,288 ARDS patients randomized in ARDSNet or PETAL Network trials, inclusion in a metabolic syndrome group was significantly associated with lower 28-day and 90-day mortality after adjustment for age, sex, race, ethnicity, comorbidities, primary ARDS etiology, and treatment group assignment. Subgroup analyses and homogeneity testing were concordant with the primary results, indicating a potentially cumulative relationship with mortality where the associated risk declined with each additional metabolic syndrome criteria present. These results show that hypertension and diabetes were individually associated with lower 28-day mortality, while hypertension and obesity were associated with lower 90-day mortality. Interestingly, hypertension had the strongest association for both 28- and 90-day mortality, a finding previously not described. Lastly, stratified analyses indicate that these associations were seen only in infection-related ARDS, a hypothesis-generating result supported by heterogeneity testing that may help to delineate the relationship of ARDS and poor metabolic health.
Characterization of phenotypic heterogeneity among critically ill adults has, until recently, proven elusive. Recent studies using an unsupervised clustering approach, latent-class analysis, yielded a novel, biomarker-driven phenotypic characterization that can be used to predict outcomes in ARDS,(6, 7) however, the biological plausibility and bedside applicability of these phenotypes remain uncertain. In this study, commonly available risk factors were used to identify metabolic syndrome as a surrogate to estimate the burden of metabolic illness at the individual patient level. The results demonstrate an independent association between metabolic syndrome and reduced mortality among patients with ARDS. This finding is not entirely surprising, as obese patients often suffer from more severe ARDS, yet have similar or even lower mortality rates,(29) a result that could be explained by collider stratification bias,(30) but also may be related to an altered inflammatory response.(15) Additionally, diabetes has been shown to be more common in sepsis-associated ARDS compared with non-sepsis-associated ARDS, but the mechanisms underlying this relationship have not been explored in depth. As this is the first report of metabolic syndrome (as well as hypertension) being associated with reduced mortality in ARDS, molecular justification to explain these results are needed and may help to uncover biologic targets for which to improve personalized treatment options in ARDS.
The association of poor metabolic health and systemic inflammation is actually quite well-characterized and thought to be driven by proinflammatory cytokines released through a phenotypic switch in adipose tissue macrophage polarization.(10, 11, 31) The prevailing model posits that this phenotypic shift from an anti-inflammatory to pro-inflammatory state, termed ‘meta-inflammation’, leads to the development of diabetes and metabolic syndrome. The pathophysiology of ARDS, similar to metabolic syndrome, is largely driven by activation of the innate immune system, chemoattraction of immune cells via cytokine release, and endothelial dysfunction.(1, 9) How these overlapping pathophysiologic responses relate to outcomes in ARDS and critical illness remain controversial. Could it be that metabolic syndrome imposes an altered host resilience that is adapted to respond to ARDS-triggering events such as sepsis or pneumonia? This would not be entirely novel given what is known about adipose tissue macrophage polarity among critically ill adults (32) as well as sepsis-associated ARDS and diabetes.(13, 18) In this study, instead of focusing on a single metabolic illness such as diabetes or obesity alone, we used a global assessment of impaired metabolic health, as a surrogate for meta-inflammation, to identify a class of patients which might respond differently in the context of acute inflammation. To explore this relationship, we stratified participants according to their primary cause of ARDS being sepsis and pneumonia and compared them to non-infectious causes of ARDS (i.e., trauma, transfusion, aspiration). The results indicate that the association of metabolic syndrome with reduced mortality was seen largely in the those suffering from infection-related ARDS. Although only hypothesis-generating, these results may help inform our understanding of ARDS in the setting of poor metabolic health.
One surprising result in this study was that premorbid hypertension appeared to have the strongest impact on the association with reduced 28-day and 90-day mortality. However, hypertension has not been previously explored in ARDS. This is in contrast to diabetes and obesity. For example, a history of diabetes may be associated with a lower risk of ARDS development,(16, 17, 33) but these results have been debated(14, 34), and data are lacking on its impact with mortality in patients with ARDS. Similarly, the relationship of obesity with mortality among patients with ARDS remains unresolved, with some studies showing no association,(35) while others suggest higher severity but no change or even improved mortality.(29, 36) This finding, the “obesity paradox”,(37) may have several explanations, including that it could be a spurious association. However, biologic differences may exist. For instance, cytokine markers of innate immunity (i.e. IL-6, IL-8) were decreased in patients with obesity and ARDS raising the possibility that inflammation is attenuated when exposed to an ARDS-triggering event, while at the same time, a marker of endothelial activation and injury, von-Willebrand Factor, was increased and associated with improved survival.(15) This association also seems paradoxical, but as von-Willebrand Factor is associated with hypertension(38) and metabolic syndrome(39), this biomarker deserves further study in this population.
There are several limitations in this study. Although BioLINCC provides excellent data for patients meticulously diagnosed with ARDS, confounding variables such as socioeconomic status and health system characteristics were not available. Selection bias due to missing data is also a limitation, but because data were compiled from well-executed RCTs, missing data were few and found to produce similar results with multiple imputation (eTable 2 and eTable 3). Collider stratification bias is also a potential explanation for these findings, as patients in this study were conditioned on having ARDS and obesity could be considered an ARDS risk factor. Additionally, misclassification bias was carefully considered a priori, but since it was expected to be non-differential (or random), its impact is likely minimal. Most definitions of metabolic syndrome include obesity, high blood pressure, diabetes or prediabetes, and dyslipidemia. However, neither hemoglobin A1c nor cholesterol were available to characterize prediabetes or dyslipidemia. We anticipated this limitation, expecting that any misclassification bias in this setting would again be non-differential, and as such, would dilute effect estimates and bias our results to the null if present. To this point, our sample size was estimated assuming a one to three ratio of metabolic syndrome to control, yet we only found 10.6% of patients met criteria for metabolic syndrome. Therefore, it is possible that these results were underpowered leading to the failure to detect a significant difference in unadjusted analyses. Lastly, we used datasets from different trials each with a different study focus conducted over three decades. Data collection may not have been uniform, as exemplified in SAILS where hypertension was not explicitly collected. As a result, we utilized alternative criteria for high blood pressure in that study to identify participants with probable hypertension understanding that hospital measurements for this purpose are a limitation. Supporting this approach, a sensitivity analysis excluding SAILS showed similar results.
Conclusions:
Metabolic syndrome, defined as the combination of obesity, diabetes, and high blood pressure, is significantly associated with decreased mortality in ARDS and may be most specific in ARDS secondary to sepsis or pneumonia.
Supplementary Material
Key Points.
Question:
Can a history of poor metabolic health predict outcomes in acute respiratory distress syndrome (ARDS) trial participants?
Findings:
In this secondary analysis of 7 randomized clinical trials including 4,288 patients with ARDS, the presence of metabolic syndrome significantly reduced the odds of 28- and 90-day mortality (by 30% and 25%, respectively). This finding was only seen in ARDS caused by sepsis or pneumonia and stronger with each criterion added from 0 to 3 (mortality reduced by 18%, 22%, and 40%, respectively).
Meaning:
In ARDS trial participants, poor metabolic health was associated with lower mortality, particularly when caused by sepsis or pneumonia.
Sources of funding and support:
The study was funded by the 2022 SCCM-Weil Research Grant (Denson) on behalf of Discovery, the Critical Care Research Network of the Society of Critical Care Medicine, the United States Critical Illness and Injury Trials Group, and the Critical Care Pharmacotherapy Network as well as by a Roadmap Scholar Award (Denson) from the Louisiana Clinical and Translational Science Center funded by U54 GM104940 from the National Institute of General Medical Sciences of the National Institutes of Health. As the data for this study were collected originally in the NHLBI ARDS Network and PETAL Network trials, we would like to acknowledge all participants and research teams who participated in these trials. Furthermore, we thank the Biological Specimen and Data Repository Information Coordinating Center of the NHLBI that made the data available to perform this study.
Role of Funder/Sponsor Statement:
The National Institutes of Health and the Society of Critical Care Medicine provided grant funding to the institutions for study investigators. The content of this manuscript is solely the responsibility of the authors and does not represent the official views of the NHLBI or National Institutes of Health. However, excluding the above examples of institutional support for investigators or research staff, the funding organizations of this study had no direct role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Copyright Form Disclosure:
Dr. Pagliaro disclosed that she is employed by the Bedford VA and Mass General Hospital. Dr. Denson’s institution received funding from the Weil Research Grant from the Discovery Research Network of SCCM 2022 and the Roadmap Scholar Award from the Louisiana Clinical and Translational Science Center funded by the National Institute of General Medical Sciences (U54 GM104940); received support for article research from the National Institutes of Health. The remaining authors have disclosed that they do not have any potential conflicts of interest.
Footnotes
Declaration of interests
All authors declare no competing interests relevant to this work.
Meeting Presentation: Preliminary results from this study were presented in the form of an abstract at the 2022 American Thoracic Society International Conference and the 2023 Society of Critical Care Medicine Congress both in San Francisco, CA.
Data sharing
Data from these studies can be provided to others upon reasonable request on approval of a written request to the corresponding author and the Biologic Specimen and Data Repository Information Coordinating Center. Data from the US National Heart, Lung, and Blood Institute was accessed through the Biologic Specimen and Data Repository Information Coordinating Center public repository.
References:
- 1.Thompson BT, Chambers RC, Liu KD: Acute Respiratory Distress Syndrome. New England Journal of Medicine 2017; 377(6):562–572 [DOI] [PubMed] [Google Scholar]
- 2.Rubenfeld GD, Caldwell E, Peabody E, et al. : Incidence and outcomes of acute lung injury. The New England journal of medicine 2005; 353(16):1685–1693 [DOI] [PubMed] [Google Scholar]
- 3.Denson JL, Gillet AS, Zu Y, et al. : Metabolic Syndrome and Acute Respiratory Distress Syndrome in Hospitalized Patients With COVID-19. JAMA Netw Open 2021; 4(12):e2140568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bellani G, Laffey JG, Pham T, et al. : Epidemiology, Patterns of Care, and Mortality for Patients With Acute Respiratory Distress Syndrome in Intensive Care Units in 50 Countries. Jama 2016; 315(8):788–800 [DOI] [PubMed] [Google Scholar]
- 5.Khan YA, Fan E, Ferguson ND: Precision Medicine and Heterogeneity of Treatment Effect in Therapies for ARDS. Chest 2021; 160(5):1729–1738 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Calfee CS, Delucchi K, Parsons PE, et al. : Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials. Lancet Respir Med 2014; 2(8):611–620 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sinha P, Delucchi KL, McAuley DF, et al. : Development and validation of parsimonious algorithms to classify acute respiratory distress syndrome phenotypes: a secondary analysis of randomised controlled trials. Lancet Respir Med 2020; 8(3):247–257 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Grundy SM, Cleeman JI, Daniels SR, et al. : Diagnosis and management of the metabolic syndrome: an American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation 2005; 112(17):2735–2752 [DOI] [PubMed] [Google Scholar]
- 9.Aggarwal NR, King LS, D’Alessio FR: Diverse macrophage populations mediate acute lung inflammation and resolution. Am J Physiol Lung Cell Mol Physiol 2014; 306(8):L709–725 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sutherland JP, McKinley B, Eckel RH: The metabolic syndrome and inflammation. Metab Syndr Relat Disord 2004; 2(2):82–104 [DOI] [PubMed] [Google Scholar]
- 11.Esser N, Legrand-Poels S, Piette J, et al. : Inflammation as a link between obesity, metabolic syndrome and type 2 diabetes. Diabetes Res Clin Pract 2014; 105(2):141–150 [DOI] [PubMed] [Google Scholar]
- 12.Xie J, Zu Y, Alkhatib A, et al. : Metabolic Syndrome and COVID-19 Mortality Among Adult Black Patients in New Orleans. Diabetes Care 2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gong MN, Bajwa EK, Thompson BT, et al. : Body mass index is associated with the development of acute respiratory distress syndrome. Thorax 2010; 65(1):44–50 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gajic O, Dabbagh O, Park PK, et al. : Early identification of patients at risk of acute lung injury: evaluation of lung injury prediction score in a multicenter cohort study. Am J Respir Crit Care Med 2011; 183(4):462–470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Stapleton RD, Dixon AE, Parsons PE, et al. : The association between BMI and plasma cytokine levels in patients with acute lung injury. Chest 2010; 138(3):568–577 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Moss M, Guidot DM, Steinberg KP, et al. : Diabetic patients have a decreased incidence of acute respiratory distress syndrome. Critical care medicine 2000; 28(7):2187–2192 [DOI] [PubMed] [Google Scholar]
- 17.Gong MN, Thompson BT, Williams P, et al. : Clinical predictors of and mortality in acute respiratory distress syndrome: potential role of red cell transfusion. Critical care medicine 2005; 33(6):1191–1198 [DOI] [PubMed] [Google Scholar]
- 18.Sheu CC, Gong MN, Zhai R, et al. : Clinical characteristics and outcomes of sepsis-related vs non-sepsis-related ARDS. Chest 2010; 138(3):559–567 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Matthay MA, Brower RG, Carson S, et al. : Randomized, placebo-controlled clinical trial of an aerosolized β2-agonist for treatment of acute lung injury. Am J Respir Crit Care Med 2011; 184(5):561–568 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Brower RG, Lanken PN, MacIntyre N, et al. : Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome. N Engl J Med 2004; 351(4):327–336 [DOI] [PubMed] [Google Scholar]
- 21.Rice TW, Wheeler AP, Thompson BT, et al. : Initial trophic vs full enteral feeding in patients with acute lung injury: the EDEN randomized trial. Jama 2012; 307(8):795–803 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wiedemann HP, Wheeler AP, Bernard GR, et al. : Comparison of two fluid-management strategies in acute lung injury. The New England journal of medicine 2006; 354(24):2564–2575 [DOI] [PubMed] [Google Scholar]
- 23.Steinberg KP, Hudson LD, Goodman RB, et al. : Efficacy and safety of corticosteroids for persistent acute respiratory distress syndrome. The New England journal of medicine 2006; 354(16):1671–1684 [DOI] [PubMed] [Google Scholar]
- 24.Truwit JD, Bernard GR, Steingrub J, et al. : Rosuvastatin for sepsis-associated acute respiratory distress syndrome. The New England journal of medicine 2014; 370(23):2191–2200 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Moss M, Huang DT, Brower RG, et al. : Early Neuromuscular Blockade in the Acute Respiratory Distress Syndrome. The New England journal of medicine 2019; 380(21):1997–2008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ranieri VM, Rubenfeld GD, Thompson BT, et al. : Acute respiratory distress syndrome: the Berlin Definition. Jama 2012; 307(23):2526–2533 [DOI] [PubMed] [Google Scholar]
- 27.Muntner P, Shimbo D, Carey RM, et al. : Measurement of Blood Pressure in Humans: A Scientific Statement From the American Heart Association. Hypertension 2019; 73(5):e35–e66 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Cumpston M, Li T, Page MJ, et al. : Updated guidance for trusted systematic reviews: a new edition of the Cochrane Handbook for Systematic Reviews of Interventions. Cochrane Database Syst Rev 2019; 10:Ed000142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.O’Brien JM Jr., Phillips GS, Ali NA, et al. : Body mass index is independently associated with hospital mortality in mechanically ventilated adults with acute lung injury. Critical care medicine 2006; 34(3):738–744 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Banack HR, Kaufman JS: Does selection bias explain the obesity paradox among individuals with cardiovascular disease? Annals of Epidemiology 2015; 25(5):342–349 [DOI] [PubMed] [Google Scholar]
- 31.Lumeng CN, Bodzin JL, Saltiel AR: Obesity induces a phenotypic switch in adipose tissue macrophage polarization. J Clin Invest 2007; 117(1):175–184 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Langouche L, Marques MB, Ingels C, et al. : Critical illness induces alternative activation of M2 macrophages in adipose tissue. Crit Care 2011; 15(5):R245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yu S, Christiani DC, Thompson BT, et al. : Role of diabetes in the development of acute respiratory distress syndrome. Critical care medicine 2013; 41(12):2720–2732 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ji M, Chen M, Hong X, et al. : The effect of diabetes on the risk and mortality of acute lung injury/acute respiratory distress syndrome: A meta-analysis. Medicine (Baltimore) 2019; 98(13):e15095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.O’Brien JM Jr., Welsh CH, Fish RH, et al. : Excess body weight is not independently associated with outcome in mechanically ventilated patients with acute lung injury. Ann Intern Med 2004; 140(5):338–345 [DOI] [PubMed] [Google Scholar]
- 36.Morris AE, Stapleton RD, Rubenfeld GD, et al. : The association between body mass index and clinical outcomes in acute lung injury. Chest 2007; 131(2):342–348 [DOI] [PubMed] [Google Scholar]
- 37.Zhi G, Xin W, Ying W, et al. : “Obesity Paradox” in Acute Respiratory Distress Syndrome: Asystematic Review and Meta-Analysis. PLoS One 2016; 11(9):e0163677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Blann AD, Naqvi T, Waite M, et al. : von Willebrand factor and endothelial damage in essential hypertension. J Hum Hypertens 1993; 7(2):107–111 [PubMed] [Google Scholar]
- 39.Lim HS, Lip GYH, Blann AD: Plasma von Willebrand Factor and the Development of the Metabolic Syndrome in Patients with Hypertension. The Journal of Clinical Endocrinology & Metabolism 2004; 89(11):5377–5381 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data from these studies can be provided to others upon reasonable request on approval of a written request to the corresponding author and the Biologic Specimen and Data Repository Information Coordinating Center. Data from the US National Heart, Lung, and Blood Institute was accessed through the Biologic Specimen and Data Repository Information Coordinating Center public repository.