Keywords
ARDS - COVID-19 - SARS-CoV-2 - latent class
Introduction
Severe infection from the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)
virus can lead to pneumonia and acute respiratory distress syndrome (ARDS).[1] Initial data from China reported that ARDS was present in nearly 42% of patients
with coronavirus disease 2019 (COVID-19) pneumonia and carried mortality over 50%.[2] SARS-CoV-2 virus-associated pneumonia and ARDS were often associated with hyperinflammation
and elevation of several serum inflammatory markers but usually less than what is
observed in non-COVID ARDS.[3] Elevated inflammatory markers such as C-reactive protein (CRP), interleukin (IL)-6,
etc., were associated with severe infection.[4]
Despite an objective case definition,[5] ARDS in intensive care units (ICUs) are often underrecognized.[6] Significant clinical heterogeneity in ARDS is well defined as ARDS can be triggered
by a myriad of pulmonary and extrapulmonary causes. The severity of ARDS is also variable
and classically defined by the degree of hypoxemia, which ranges from mild to severe.[7] Previous studies have found the existence of two distinct subphenotypes in non-COVID
ARDS patients.[8] The hyperinflammatory subphenotype of ARDS was associated with elevated inflammatory
biomarkers, a higher requirement of vasopressors, and worse clinical outcomes.[8] Identification of subtypes of ARDS is clinically significant as the majority of
the successful experimental therapies were ultimately found to be ineffective in the
general ARDS population.[7] We hypothesize that severe COVID-19 infection-associated ARDS might have several
latent subphenotypes and carry a different prognosis.
Methods
Nucleic acid-based laboratory test-confirmed SARS-CoV-2 infected adult (age > 18 years)
patients with acute respiratory failure (defined by partial pressure of oxygen [PaO2]/fraction of inspired oxygen [FiO2] ratio < 200 mm Hg) and chest X-ray showing bilateral infiltrates, admitted in an
ICU of a tertiary care teaching hospital, New Delhi, India, were recruited in this
study. The data collection period was between May 2020 and October 2020. Strict adherence
to the Berlin definition of ARDS[5] was not technically feasible as several patients were managed by oxygen therapy
through a high-flow nasal cannula. No formal sample size estimation was performed
for this study, and approximately 200 to 250 patients' data were planned to be included
in this analysis.
Patients' demographic (age, sex, height, weight), clinical conditions, and preexisting
comorbid illnesses were noted. In addition, patients' baseline laboratory parameters
that included hematological and biochemical investigations were noted, and all investigations
available within the first 24 hours of ICU admission were considered as “baseline.”
The respiratory support requirement at the time of ICU admission and clinical outcome
(survival or death) at the time of hospital discharge was also noted for all patients.
Results
Data of n = 233 patients with laboratory-confirmed SARS-CoV-2 infection leading to acute hypoxemic
respiratory failure (PaO2/FiO2 ratio < 200 mm Hg) were included in this study. The mean (standard deviation) age
of the patients was 53.3 (15) years, and 165 of all patients (proportion [95% confidence
[CI]] 70.8 [64.7–76.3]) survived hospital discharge. Sixty-two percent of all patients
were male. The median (interquartile [IQR]) PaO2/FiO2 ratio was 150 (135–195) mm Hg. Median (IQR) duration of hospital stay and ICU stay
were 11 (9–14) and 8 (6–11) days, respectively.
Initially, a two subphenotypic model was constructed with the available inflammatory
markers (neutrophil-to-lymphocyte ratio [NLR], lactate dehydrogenase, IL-6, serum
ferritin, and CRP). However, the model constructed by IL-6 and NLR had the lowest
Akaike information criterion (AIC) (AIC 4,735, entropy 0.90, bootstrap likelihood
ratio test p = 0.01, [Fig. 1]). A comparison of baseline variables between subphenotype 1 and subphenotype 2 has
been depicted in [Table 1]. Patients belonging to the subphenotype 2 had a lower baseline PaO2/FiO2 ratio, higher CRP, ferritin, and creatinine, and lower platelet count. Male preponderance
was also seen in subphenotype 2. However, baseline hemoglobin, total leukocyte count,
and international normalized ratio were similar in both the subphenotypes. Unadjusted
analyses revealed that the requirement of mechanical ventilation (odds ratio [OR]
[95% CI] 6.4 [3.4–11.8], p < 0.001) and hospital mortality (OR [95% CI] 6.2 [3.3–11.6], p < 0.001) were higher in the subphenotype 2. In addition, duration of hospital stay
(median [IQR] 11 [9–13] vs. 14 [11–17] days, p < 0.001) and duration of ICU stay (median [IQR] 7 [6–9] vs. 11.5 [8–14] days, p < 0.001) were significantly higher in subphenotype 2.
Table 1
Comparison of baseline variables between two subphenotypes
|
Variable
|
All Patients
|
Subphenotype 1
|
Subphenotype 2
|
Significance
|
|
Age
|
55 (45–65)
|
55 (46–64)
|
56.5 (44.3–66.5)
|
p = 0.44
|
|
Sex (male/female)
|
145/88
|
100/67
|
45/21
|
p = 0.24
|
|
Hemoglobin
|
11.6 (9.1–13.5)
|
11.8 (9.3–13.5)
|
11.4 (7.9–13.1)
|
p = 0.14
|
|
TLC
|
10300 (6700–15500)
|
9900 (6710–15300)
|
10590 (6675–17775)
|
p = 0.44
|
|
Platelet
|
193 (130–270)
|
209 (135–289)
|
172 (101–224)
|
p = 0.005
|
|
INR
|
1.18 (1.08–1.3)
|
1.17 (1.1–1.23)
|
1.2 (1–1.5)
|
p = 0.42
|
|
Creatinine
|
1.2 (0.8–2.2)
|
1.1 (0.8–1.9)
|
1.6 (1.04–3.1)
|
p = 0.004
|
|
LDH
|
585 (406–808)
|
542 (404–739)
|
690 (445–937)
|
p = 0.009
|
|
Ferritin
|
716 (360–1385)
|
585 (329–1130)
|
975 (574–2000)
|
p < 0.001
|
|
Bilirubin
|
0.6 (0.4–0.95)
|
0.6 (0.4–0.9)
|
0.63 (0.5–1)
|
p = 0.24
|
|
PaO2/FiO2 ratio
|
142 (110–174)
|
156 (129–177)
|
110 (87–155)
|
p < 0.001
|
Abbreviations: FiO2, fraction of inspired oxygen; INR, international normalized ratio; LDH, lactate dehydrogenase;
PaO2, partial pressure of oxygen; TLC, total leukocyte count.
Fig. 1 Profile plot showing distribution of neutrophil-to-lymphocyte ratio and interleukin
(IL)-6 in both subphenotypes.
After adjustment of age and baseline PaO2/FiO2 ratio, subphenotype 2 had significantly higher mortality (OR [95% CI] 3.1 (1.4–6.9],
p = 0.007) and requirement of mechanical ventilation (OR [95% CI] 3 [1.1–8.1], p = 0.029). In addition, duration of hospital stay (t = 3.51, p > 0.001) and duration of ICU stay (t = 4.12, p < 0.001) were also significantly higher in subphenotype 2, even after adjustment
of age and baseline PaO2/FiO2.
Discussion
This study identified two subphenotypes of COVID-19 infection-associated ARDS and
subphenotype 2, which were associated with “hyperinflammation” and poor clinical outcomes.
Classically, ARDS is a heterogeneous disorder, and there are significant pathophysiological
variations between pulmonary and extrapulmonary ARDS.[9] Subphenotypes of ARDS were identified from secondary analyses of various randomized
controlled trials. Calfee et al identified two subphenotypes of ARDS; the hyperinflammatory
subphenotype was associated with the elevation of inflammatory biomarkers, acidosis,
and sepsis. Clinical outcome was also worse in that subphenotype.[8] Our findings are also quite similar to the previous data as we were also able to
identify two subphenotypes, and poor clinical outcome was evident with hyperinflammation.
Similar subphenotypes were identified in COVID-19 ARDS patients also.[10]
Identification of subphenotypes in COVID-19 pneumonia/ARDS patients has several clinical
implications. As patients with elevated inflammatory biomarkers have a worse clinical
outcome and higher requirements for mechanical ventilation, they should be managed
in a well-equipped center. During the time of the pandemic, the categorization of
the patients is important for proper resource utilization.
Identification of subphenotypes of COVID-19 ARDS has several therapeutic implications.
First, future randomized controlled trials can be designed for a particular subphenotype
to avoid heterogeneity in the clinical status of the participants. Apart from this,
anti-inflammatory drugs are the mainstay of treatment in COVID-19 pneumonia patients,
and they may be more useful in patients with a hyperinflammatory state.
Limitations
Our study has several limitations. Most importantly, the sample size of our study
was limited and strict adherence to the Berlin definition was not practically feasible.
Furthermore, we conducted this study during the so-called “first wave” of the COVID-19
pandemic in New Delhi. As the COVID-19 pandemic is evolving, variants of the SARS-CoV-2
virus also evolve, and disease severity also changes. Hence, there is some probability
that the study findings may not apply to other geographical locations or when other
virus variants are causative agents.
Conclusion
Two distinct but overlapping subphenotypes were identified in SARS-CoV-2-associated
respiratory failure. In addition, the hyperinflammatory subphenotype was associated
with a significantly poor short-term outcome even after adjustment of baseline disease
severity.
