Impact of Acute Exacerbation and Its Phenotypes on the Clinical Outcomes of Chronic Obstructive Pulmonary Disease in Hospitalized Patients: A Cross-Sectional Study

Acute exacerbations of COPD (AECOPD) are clinically significant events having therapeutic and prognostic consequences. However, there is a lot of variation in its clinical manifestations described by phenotypes. The phenotypes of AECOPD were categorized in this study based on pathology and exposure. In our cross-sectional study, conducted between 1 January 2016 to 31 December 2020, the patients were categorized into six groups based on pathology: non-bacterial and non-eosinophilic; bacterial; eosinophilic; bacterial infection with eosinophilia; pneumonia; and bronchiectasis. Further, four groups were classified based on exposure to tobacco smoke (TS), biomass smoke (BMS), both, or no exposure. Cox proportional-hazards regression analyses were performed to assess hazard ratios, and Kaplan–Meier analysis was performed to assess survival, which was then compared using the log-rank test. The odds ratio (OR) and independent predictors of ward admission type and length of hospital stay were assessed using binomial logistic regression analyses. Of the 2236 subjects, 2194 were selected. The median age of the cohort was 67.0 (60.0 to 74.0) and 75.2% were males. Mortality rates were higher in females than in males (6.2% vs. 2.3%). AECOPD-B (bacterial infection) subjects [HR 95% CI 6.42 (3.06–13.46)], followed by AECOPD-P (pneumonia) subjects [HR (95% CI: 4.33 (2.01–9.30)], were at higher mortality risk and had a more extended hospital stay (6.0 (4.0 to 9.5) days; 6.0 (4.0 to 10.0). Subjects with TS and BMS-AECOPD [HR 95% CI 7.24 (1.53–34.29)], followed by BMS-AECOPD [HR 95% CI 5.28 (2.46–11.35)], had higher mortality risk. Different phenotypes have different impacts on AECOPD clinical outcomes. A better understanding of AECOPD phenotypes could contribute to developing an algorithm for the precise management of different phenotypes.


Introduction
Chronic obstructive pulmonary disease (COPD) is a preventable and treatable condition characterized by non-reversible airflow restriction responsible for significant morbidity, mortality, and healthcare expenditure globally [1,2]. In 2017, there were 545 million chronic respiratory disorders, of which COPD accounted for roughly 50%, with a global prevalence between 9% to 13% [3][4][5]. Sixty-five million individuals worldwide have moderate-tosevere COPD, and it is the third most common cause of death [6]. More than 3.23 million the study, constituting the Mysuru COPD (MYCO) Cohort. During the study period, age, gender, comorbidities, length of hospital stay, admission to ward or ICU, outcome (survived/expired), and hematological investigations (neutrophils, lymphocytes, eosinophils, monocytes, basophils, RBC count, and platelet count) were documented and recorded from the hospital records. NLR (Absolute neutrophil count/Absolute lymphocyte count) and Charlson's comorbidity score were calculated. Any additional secondary diagnoses were defined according to the ICD-10 coding system. Subjects' demographic and clinical characteristics are presented in Table 1 based on pathology and Table 2 based on exposure.

Definitions
COPD is a common, preventable, treatable disease characterized by persistent respiratory symptoms and airflow limitation due to airway and/or alveolar abnormalities, usually caused by significant exposure to noxious particles or gases and influenced by host factors including abnormal lung development [2]. AECOPD is defined as acute worsening of respiratory symptoms (such as increase in coughing, quantity of sputum, breathlessness, or increase in purulence of sputum) that results in additional therapy [50]. Cor pulmonale, pulmonary hypertension, and other additional diagnoses are defined according to the ICD10 coding system. Charlson's comorbidity index was calculated as a composite risk of all associated comorbidities, where a higher score predicts a higher risk of mortality or higher resource-use [51].
Bacterial infection was defined as the presence of positive sputum culture for pathogenic bacteria. Eosinophilia was defined as the presence of 3% or more eosinophils in the differential leukocyte count [52]. The presence of an ICD-10 code identified pneumonia and bronchiectasis subjects.
Smokers are defined as adult smokers with a smoking history of >10 pack years. The burning of various fuels, including wood, animal dung, and crop wastes, to produce the energy required for heating and cooking in many houses worldwide results in the production of biomass smoke [53]. Biomass exposure is defined as chronic exposure to biomass smoke for at least ten years [54,55]. Non-tobacco smoke and non-biomass smoke COPD might be due to causes other than smoking and biomass exposure, such as prolonged exposure to dust due to dusty occupations such as agricultural farming, sericulture farming, working in the cotton industry, chronic asthma, and post-TB sequelae.
The common reasons for hospitalization of AECOPD subjects in the general ward were failed response to initial medical management, severe symptoms, serious comorbidities, the onset of new physical signs, and acute respiratory failure (without using accessory respiratory muscles). Common symptoms for admission to respiratory or medical ICU are very severe symptoms, acute respiratory failure (with accessory respiratory muscle), persistent or worsening hypoxemia and/or severe respiratory acidosis, need for noninvasive or invasive mechanical ventilation, and hemodynamic instability [50].
The subjects were categorized based on pathology and exposure as below. Pathology phenotypes (six groups): 1.
AECOPD-Non-bacterial and non-eosinophilic (NBE) group, consisting of AECOPD subjects who had no bacterial infection and no peripheral blood eosinophilia; 2.
AECOPD-Eosinophilia (E) group, consisting of AECOPD subjects who had peripheral blood eosinophilia and no bacterial infections; 4.
AECOPD-Bacterial with eosinophilia (BE) group, consisting of subjects who had bacterial AECOPD with peripheral blood eosinophilia; 5.
Exposure groups (four groups): 1. Non-tobacco smoke and non-biomass smoke (NTS and NBMS) AECOPD group, consisting of AECOPD subjects who were non-smokers with no biomass exposure; 2.
Biomass smoke (BS) AECOPD group, consisting of AECOPD subjects who were exposed to biomass smoke; 4.
Tobacco smoke and biomass smoke (TS and BMS) AECOPD group, consisting of AECOPD subjects who were smokers with biomass exposure.
The subjects eligible were identified by the ICD-10 codes J44.1 and J44.9. Male and female subjects ≥ 40 years of age diagnosed with acute exacerbation of COPD were included in the study. Subjects with relevant missing data were excluded from the study.

Statistical Analysis
Statistical analysis was performed employing Jamovi (v1.6, The Jamovi project, SYD, AUS). Data were checked for normality of distribution using Shapiro-Wilk's test. If they were normally distributed, continuous variables were presented as mean ± standard deviation. If not, they were presented as median with their interquartile range. Categorical variables were presented as percentages. Statistical significance was assessed by the Chisquare test for categorical variables and by the Kruskal-Wallis test for continuous variables and expressed with medians and interquartile ranges (IQRs). Cox proportional-hazards regression analyses were performed to assess univariate and multivariate hazard ratios. The Kaplan-Meier method was used to draw survival curves, while the survival rates were compared using the log-rank test. The odds ratio (OR) and independent predictors of type of ward admission and length of hospital stay were assessed by binomial logistic regression analyses. A two-tailed p-value of <0.05 was considered statistically significant.
Overall, mortality was higher in females when compared to males (6.2% vs. 2.3%, respectively; p < 0.001) (Table S1), and in all the different phenotypes, higher mortality in females was observed than in males. This was such for cases of AECOPD-NBE (2.7% vs.  Table 2). When the type of exposure was stratified based on gender, we observed a significant difference between male and female mortality in NTS and NBMS-AECOPD (1.0% vs. 3.6%, respectively; p = 0.030) and BMS-AECOPD (3.6% vs. 7.3%, respectively; p = 0.456). Still, in TS-AECOPD, higher mortality was observed in males (2.7% vs. 0%, respectively; p = 0.614) and the TS and BMS-AECOPD group had only male patients, with a 6.3% mortality rate (Table S3).
A significant difference between the six groups was observed following Kaplan-Meier analysis using the log-rank test (p < 0.0001) (Figure 2), which showed that the AECOPD-B group had a significantly higher risk of mortality compared to the other groups. A pairwise comparison was also performed for each group (Table 3). We observed that intergroup comparisons for survival among conditions associated with bacterial infections such as AECOPD-B, AECOPD-BE, AECOPD-P, and AECOPD-BC were not statistically significant. Intergroup comparisons for survival between conditions related to bacterial infections (AECOPD-B, AECOPD-BE, AECOPD-P, and AECOPD-BC) and those without bacterial infections (AECOPD-E and AECOPD-NBE) were statistically significant, except for AECOPD-BE with AECOPD-NBE.  A significant difference between the four groups was observed following Kaplan-Meier analysis using the log-rank test (p < 0.0001), which showed that the BMS-AECOPD group had a significantly higher risk of mortality compared to the other groups ( Figure 3). A pairwise comparison was also performed for each group (Table 4). Three intergroup comparisons for survival were significant: BMS-AECOPD vs. NTS and NBMS-AECOPD, BMS-AECOPD vs. TS-AECOPD, and TS and BMS-AECOPD vs. NTS and NBMS-AECOPD.     Two models were considered to evaluate factors with independent association to mortality. Model 1 assessed exposures and model 2 considered only phenotypes based on pathology (Table 5). Univariate Cox proportional-hazard regression analysis for survival in model 1 showed that females, those older than 60, those with type 2 respiratory failure, patients with sepsis, and the AECOPD-B, AECOPD-P, and AECOPD-BC groups were associated with poor survival.  Table 5).  Univariate Cox proportional-hazard regression analysis for survival in model 2 showed that age higher than 60, type 2 respiratory failure, patients with sepsis, TS-AECOPD, BMS-AECOPD, and TS and BMS-AECOPD were associated with poor survival. On multivariate Cox proportional-hazard regression analysis, type 2 respiratory failure [HR (95% CI: 1. 75 (Table S5).

Discussion
We observed six phenotypes of acute exacerbation of COPD which impact clinical outcomes. To the best of our knowledge, this is the first attempt to compare the various AECOPD phenotypes based on pathology and exposure, their characteristics, and their clinical outcome in an LMIC country. We observed an overall mortality rate of 3.3%. We observed the highest mortality in AECOPD-B (HR: 6.42), followed by AECOPD-P (HR:4.33). The highest ICU admissions were for AECOPD-B (51%) and AECOPD-P (35.6%). Besides AECOPD-E and AECOPD-BC, which had lower Charlson's comorbidity indices, comorbidities were similar across various phenotypes. Among subjects with AECOPD, combined TS and BMS-AECOPD (HR 7.24) was observed to have a greater mortality risk than BMS-AECOPD (HR 5.28), followed by TS-AECOPD (HR 2.17), and an additive effect was observed. Female patients with AECOPD had a greater risk of death and poor outcome than male patients.
Bacterial infection associated with COPD exacerbation, pneumonia associated with COPD exacerbation, and sepsis are the risk factors most closely associated with a greater risk of death. Lower respiratory tract infections, both acute and chronic, are frequent in patients with AECOPD. Infections contribute considerably to these patients' poor clinical course and overall morbidity and mortality [56,57]. AECOPD patients have an impaired response to microbial colonization and infection. These impaired responses, such as impaired alveolar macrophage and toll-like receptor activity, increase infection vulnerability and promote an oxidative stress response that injures lung tissue [58,59]. Bacterial infections are responsible for more than 50% of AECOPD exacerbations [60]. Bacterial load in the lungs of AECOPD patients is an essential determinant of airway inflammation. Increased concentration of pathogens is directly correlated with greater intensity of neutrophilic airway inflammation [60,61]. Another model of exacerbation pathogenesis suggests that acquiring novel bacterial strains is a major driver of acute exacerbations [56,62]. Among different AECOPD phenotypes, the highest mortality rate was observed in the AECOPD-B phenotype (11.2%), with the highest hazard ratio of 6.42 compared to other AECOPD phenotypes. Other studies evaluating mortality rates in AECOPD with bacterial infections range from 2.4% to 8.2% [27,28]. The AECOPD-B phenotype was also associated with a longer duration of hospitalization, and had the highest risk of ICU admissions compared to other AECOPD phenotypes. Other studies evaluating the length of hospital stay ranged in their evaluation from 10 to 13 days [27][28][29], but had fewer ICU admissions than our study, with 51% of AECOPD-B patients admitted to ICU.
Pneumonia is a significant cause of hospitalization and mortality, especially among AECOPD patients, and is linked to abnormal host-defense mechanisms [63][64][65]. A study by Gutierrez et al. analyzed sputum samples from pneumonia patients, and compared them with AECOPD patients and patients with both pneumonia and AECOPD. They observed that the microenvironment present in the lung modulates the activation of macrophages, which may lead to disparities in cytokine generation and specific macrophage activation when AECOPD and pneumonia are present simultaneously; this can cause different inflammatory responses leading to different outcomes [65]. Furthermore, a study by Huerta et al. found that, compared to other cohorts, patients with AECOPD-P had a higher level of Interleukin 6, tumor necrosis factor-α, C-reactive protein, and procalcitonin [30]. Our cohort's prevalence of AECOPD-P was 14.1%, and they had the second highest mortality rate at 7.8%. Various studies have shown that AECOPD-P has mortality rates ranging from 3.4% to 13.2% [30][31][32][33][34][35][36]. In our study, AECOPD subjects with pneumonia had a significantly higher mortality risk than AECOPD-NBE and AECOPD-E as assessed by Kaplan-Meier analysis (30-day mortality). Similar results were also seen in other studies investigating AECOPD-P [37,38]. AECOPD-P had a longer duration of hospitalization, between 4 to 10 days, which was twice that of AECOPD-NBE. The length of hospitalization in other studies varied from 7 days to 16 days [30][31][32][33][34]. The percentage of ICU admission was significantly lower in previous studies, ranging from 2.6% to 26%, compared to our study (35.6%), which could be due to a higher percentage of sepsis patients (17.2%) among the AECOPD-P phenotype in our study. According to several studies, sepsis is more likely to occur in AECOPD patients due to corticosteroids, underlying comorbidities, and possibly compromised barrier function, leading to an increased risk of developing pneumonia [66][67][68]. A study performed by Chen et al. studied the impact of sepsis on COPD patients' outcomes, and they identified that COPD patients with sepsis had a higher risk of pneumonia, severe exacerbations, and mortality compared to COPD patients without sepsis [69].
Subjects with AECOPD and both eosinophilia and bacterial infection (AECOPD-BE) formed the smallest group (2.4%) in our study, with a mortality rate of 5.7%. It had the third highest ICU admission (30.2%) with odds of 4.33 (95% CI: 2. 3-8.13) and the length of hospital stay ranged from 4 days to 9 days with odds of 2.29 (95% CI: 1.29-4.06) compared to AECOPD-NBE. We could not find any studies showing a direct relationship between AECOPD and bacterial infection and eosinophilia. Conversely, we found several studies showing an inverse relationship between eosinophil count and bacterial load, suggesting that the presence of less than 2% eosinophils in AECOPD events may indicate bacterial infection [81][82][83]. It is unclear what causes the drop in eosinophil levels during bacterial infection. Still, possible reasons might be the adrenal glucocorticoid activation in reaction to the stress of bacterial infection, or the rapid accumulation of eosinophils at the inflammatory site possibly causing a decrease in the number of eosinophils in circulation [84,85] Mitochondrial DNA which plays a vital role in regulating cellular metabolism, signaling pathways, and cell death [86] can be released by eosinophils as an innate immune response to pathogens to form extracellular mitochondrial DNA traps and express specific pattern-recognition receptors, such as Toll-like receptor 4, which can recognize pathogens and trigger an immune response [87,88]. Eosinophil granule proteins also possess bactericidal activity [89]. Furthermore, high levels of cell-free mitochondrial DNA have been found in the plasma of former smokers affected by COPD [90], in the serum of mice that developed emphysema induced by chronic exposure to cigarette smoke for 6 months [90], and also in the bronchoalveolar lavage of mice acutely exposed to cigarette smoke [91]. In addition, in a small single-center study, high levels of total cell-free DNA were detected in the plasma of patients with COPD exacerbations admitted to the hospital, and were associated with an increased risk of 5-year mortality [92]. It has been also shown that cells other than eosinophils could release mitochondrial DNA, including neutrophils [93]. In the context of COPD and its main risk factor (cigarette smoke), it has been found that human bronchial alveolar cells exposed to cigarette smoke release mitochondrial DNA by extracellular vesicles and cells debris, and this, in turn, may trigger the upregulation of typical proinflammatory cytokines observed in COPD [90].
Bronchiectasis is highly prevalent in AECOPD patients, ranging from 4% to 72% [94]. It is associated with severe bronchial inflammation, higher functional impairment, higher frequency and severity of exacerbations, and increased hospitalization in AECOPD patients [95][96][97]. The prevalence of AECOPD-BC in our study was 7.7%, with a mortality of 4.2%. We observed a similar length of hospital stay (4 to 8 days) in our AECOPD-BC cohort compared to other studies, which ranged from 5.7 days to 9.6 days [39][40][41]. Sánchez-Muñoz et al. observed 5.5% ICU admissions in the AECOPD-BC cohort [41] in contrast, in our study, we observed 23.2% ICU admissions in the AECOPD-BC cohort. Our study found that the AECOPD-BC phenotype was higher in female than in male patients, similar to observations made in other similar studies [98][99][100][101].
One of the important risk factors for non-smoking COPD is exposure to biomass fuels [102], which are critical for activities of daily living such as cooking and heating, especially in developing nations [103]; almost 3 billion people worldwide use biomass fuels as their primary source of energy. Biomass fuels account for 50-90% of household energy in many developing countries. Biomass combustion fumes contain several pollutants with potential to cause long-term damage to the lungs. This includes particulate matter, volatile organic compounds such as and formaldehyde and benzene, and other organic matter, including polycyclic aromatic hydrocarbons such as benzopyrene [54]. The US Environmental Protection Agency has established National Ambient Air Quality Standards, and the recommended daily average of PM 10 is below 150 µg m −3 . The recommended annual average is below 50 µg m −3 . In homes that use biomass fuels, the average PM 10 concentrations vary between 200 to 5000 µg m −3 , depending on the presence of a chimney, ventilation (windows, cooking with doors open), and the type of fuel and stove. Cigarette smoke contains more than 4000 chemicals, including many toxic compounds and wellknown risk factors for developing COPD [44,[104][105][106][107].
We observed the highest mortality rates in BMS-AECOPD (7.1%), followed by TS and BMS-AECOPD (6.3%). Various studies have shown mortality rates for TS-AECOPD ranging from 6.8% to 29% [47,[108][109][110][111][112][113]; for BMS-AECOPD, ranging from 8.4% to 18% [47,112]; and for NTS and NBMS-AECOPD, ranging from 3.8% to 24.67% [108][109][110]113]. Biomass exposure was predominantly seen in females, and included 93.4% of females and 6.6% of males. Among females exposed to biomass, mortality rates were 7.3% versus 3.6% among males exposed to biomass. A higher number of subjects with BMS-AECOPD had diabetes mellitus, hypertension, obesity, obstructive sleep apnea, and renal failure than subjects with TS and BMS-AECOPD, who were all males. Female gender was an independent risk factor for AECOPD-related mortality in our study, and few reasons have been identified from previous studies. Females have smaller airways and lungs [114]; there are gender differences in lung physiology and microbiota composition, which may influence the severity and progression of chronic respiratory disease states [100,115]; the respiratory microbiome is vulnerable to various host immunological and inflammatory effects, and seems to have sex-specific characteristics [116,117]; and chronic persistent inflammation is thought to be more harmful in females, increasing tissue damage and worsening illness severity [118]. The gender divide in human health and disease could be primarily explained by variations in genetics and sex steroid hormones, both in kind and concentration [119].
There is a steady decline in lung function with advancing age, which increases the risk of dyspnea and the prevalence of chronic pulmonary disorders in older people [120,121]. The prevalence of AECOPD is two to three times higher in patients older than 60. With increasing age, comorbidities also increase, but these comorbidities manifest earlier in AECOPD patients than in non-COPD patients [122]. The aging process in the lungs and AECOPD shares many similarities. Many signs of aging are present in AECOPD, indicating that accelerated aging may be a pathogenic factor in AECOPD [123].
To our knowledge, this is one of the most extensive studies in India on the survival of different phenotypes of AECOPD based on both pathologies and exposure. The study's main drawback is that it has a monocentric design, making it harder for our findings to generalize. Only routine bacterial cultures were performed for bacterial identification. Other causes of AECOPD, such as viral AECOPD, were not investigated. Sputum induction was not performed; therefore, further characterization of AECOPD could not be achieved. In the future, more advanced techniques-such as the Biofire test-to identify viruses, bacteria, parasites, yeast, and antimicrobial resistance genes, in addition to sputum induction, can help to identify AECOPD phenotypes with even greater resolution.

Conclusions
In conclusion, we observed significant gender differences, with females having a much higher mortality risk than males. AECOPD-B patients, followed by AECOPD-P patients, were at higher mortality risk and had a more-extended hospital stay. Subjects with biomass exposure and who smoked, followed by subjects with biomass exposure, were at higher mortality risk. Different phenotypes have different impacts on AECOPD clinical outcomes. A better understanding of AECOPD phenotypes could contribute to developing an algorithm for the precise management of different phenotypes.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/toxics10110667/s1, Table S1: Clinical characteristics of AECOPD subjects based on gender; Table S2: Clinical Characteristics of AECOPD subjects based on pathology with gender stratified; Table S3: Clinical Characteristics of AECOPD subjects based on exposure type with gender stratified; Table S4: Binomial logistic regression for the type of ward as an outcome; Table S5: Binomial logistic regression for the length of hospital stays as an outcome.
Author Contributions: M.K.U., S.U., K.G. and P.A.M. were involved in conceptualizing the research idea and creating the research design. M.K.U., A.P., J.B.S. and P.V. were involved in the acquisition of data and analysis. M.K.U., A.P., J.B.S., P.V., S.U., K.G. and P.A.M. were involved in the interpretation of data and verified the underlying data. M.K.U., A.P., J.B.S., P.V. and P.A.M. were involved in drafting the manuscript's intellectual content. P.A.M., S.U. and K.G. were involved in the critical revision of the manuscript's intellectual content. M.K.U., A.P., P.A.M., S.U. and K.G. managed the submission process. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: Patient consent was waived because this was a retrospective study.
Data Availability Statement: All data generated or analyzed during this study are included in this published article and are available from the corresponding author upon reasonable request.

Conflicts of Interest:
The authors declare that this article's content has no conflict of interest.