Elevated blood lactate in COPD exacerbations associates with adverse clinical outcomes and signals excessive treatment with β2‐agonists

Raised blood lactate secondary to high dose β2‐agonist treatment has been reported in asthma exacerbations but has not been investigated during acute exacerbations of COPD (AECOPD). We explored associations of blood lactate measurements with disease outcomes and β2‐agonist treatments during AECOPD.


INTRODUCTION
Lactic acidosis can occur with tissue hypoxia (anaerobic tissue metabolism, Type A) or without tissue hypoxia (catecholamine-driven glycolysis, Type B). 1 High lactate predicts mortality in sepsis 2 and shock. 3 High lactate is also common during severe asthma exacerbations [4][5][6][7] attributed to exogenous catecholamine administration in the form of high-dose β 2 -agonist. In asthma, high lactate has not been associated with mortality, but can provoke dyspnoea and apparent clinical deterioration. 8 Surprisingly, there has been minimal investigation of lactate during acute exacerbations of COPD (AECOPD) despite inherent risks for this patient group to develop both anaerobic and catecholamine-driven lactic acidosis. High lactate may be relevant during AECOPD for several reasons. Tachypnoea driven by lactate may worsen dynamic hyperinflation. 9 As an anion, lactate will drive a metabolic acidosis that may delay liberation from non-invasive ventilation (NIV). 10 Patients hospitalized with AECOPD often manifest cardiac dysfunction 11,12 including tachyarrhythmias and elevated cardiac biomarkers, both associated with mortality. 13,14 Whilst standard doses of long acting β 2 -agonists (LABA) appear safe in stable COPD, 15 deleterious effects of high dose short acting β 2agonists (SABA) administered during severe AECOPD are plausible. To date, studies are limited [16][17][18][19][20] but lactate as a potential marker of excessive adrenergic stimulation has particular relevance in this population.
We postulated that elevated lactate may result from excessive β 2 -agonist dosages during AECOPD. We investigated lactate levels during AECOPD and associations with clinical outcomes and β 2 -agonist dosages.

METHODS
This study included independent retrospective and prospective cohorts, approved by Human Research Ethics Committee, Monash Health, Melbourne, Australia (HREC13291Q and HREC13134A).

Study populations
Retrospective cohort Patients over 40 years of age with a primary diagnosis of AECOPD admitted to two metropolitan hospitals in Melbourne, Australia over 12 months were identified via the Emergency Department (ED) electronic database. Case records were inspected by a pulmonologist (Martin I. MacDonald) to confirm AECOPD and exclude alternative diagnoses ( Figure 1). COPD diagnosis was accepted when confirmed by outpatient spirometry 21 (132/199 patients; 66.3%) or, if spirometry unavailable, a COPD diagnosis documented by a pulmonologist in a current/ex-smoker.

Prospective cohort
Consecutive admissions with AECOPD were recruited to a prospective observational study. 22 Lactate as a component of blood gas analysis was available in 142/156 patients (91%). COPD was confirmed by spirometry in all cases. Patients requiring immediate mechanical ventilation (MV) were excluded. Management of AECOPD in both cohorts was at the discretion of the treating clinicians and not guided by study investigators.
Demographics, comorbidities and pharmacotherapy were collected from electronic medical records. In the prospective cohort, patient history was obtained and physical examination performed.

Lactate measurements
In the retrospective cohort, lactate measurements were obtained from a combination of venous and arterial samples. In the prospective cohort, all lactate measurements were venous. For patients with multiple blood gas analyses, the highest lactate measurement within 24 h of admission was analysed. β 2 -agonist treatment β 2 -agonist treatment (salbutamol) was administered by nebulisation and/or metered dose inhaler (MDI) via spacer. To facilitate dose comparison of cumulative salbutamol exposure, 10 puffs of salbutamol was considered equivalent to 2.5 mg of nebulized salbutamol. Each puff of salbutamol via MDI was therefore counted as 0.25 mg 'nebulised equivalent'. 23

SUMMARY AT A GLANCE
β 2 -agonists can cause lactic acidosis but this has not been studied in AECOPD. Lactate was elevated in $50% of people hospitalized with AECOPD and associated with metabolic abnormalities, higher rates of NIV and cumulative β 2 -agonist dosages. Injudicious salbutamol may be driving apparent clinical deterioration via lactic acidosis.
Exacerbation severity was assessed by (elevated blood urea nitrogen, altered mental status, pulse >109 per minute and age >65 years, BAP-65) score. 25 AECOPDs were managed according to international treatment guidelines. 26 Mortality at 12 months was confirmed by review of electronic medical records or telephonically with next-of-kin.

Statistical analysis
Demographics were summarized using descriptive statistics for parametric or nonparametric data. Data are presented as number (percentage), mean ± standard deviation, or median [interquartile range]. Trends related to clinical outcomes were analysed by one-way analysis of variance (normally distributed data) or Kruskal-Wallis statistic (non-parametric data). Chi-square analyses were used for categorical data and relationships between different biomarkers were analysed by Pearson correlation.
Lactate levels were skewed and were log transformed prior to linear regression analyses. Univariate associations between lactate and subject characteristics were explored using students t-testing and Chi-square analyses where appropriate. Multiple linear regression was used to assess the relationship between lactate and salbutamol dose while controlling for potential confounding variables. All factors in which we found p < 0.2 on univariate analysis were entered into the final multivariate model. In addition, age, sex and acid-base measurements were retained in the linear regressions model irrespective of significance as potential important determinants and/or confounding factors related to associations between lactate and salbutamol dosage. Stepwise backward elimination was used beginning with the variable with the highest p-value. Inspection of the change in the adjusted R 2 and a likelihood-ratio test were both used to confirm that deleted factors did not contribute to the model. Assumptions underlying multiple regression analyses were examined and confirmed. The final model parameters were represented by unstandardized beta coefficients. Logistic regression models were applied in the prospective cohort to estimate odds ratios with 95% confidence intervals of possible risk factors for raised lactate. Age, sex and baseline FEV 1 were included in multivariable adjusted analyses. Statistical significance was accepted if p < 0.05 (twosided). All analyses were conducted on Stata MP 14.1 (Statacorp, College Station, TX).

Clinical characteristics of patient cohorts
Demographic and clinical characteristics of patients with normal (≤2.0 mmol/L) and elevated (>2 mmol/L) lactate measurements in each cohort are shown in Table 1.

Lactate and biochemical measurements
There was no correlation between lactate and CRP (À0.12, p = 0.14) with similar CRP levels (12.2  vs. 19.6 , p = 0.47) and less frequent fever (9% vs. 21%, p = 0.04) in the high lactate group. Patients with high lactate had lower pH and lower (actual) bicarbonate in both the retrospective and prospective cohorts ( Table 2). In  contrast, there was no difference in PvCO 2 between high and normal lactate levels in either cohort.

Evaluation of putative causes of elevated blood lactate measurements
Since lactic acidosis can occur during tissue hypoxia but also as a result of catecholamine excess (including β 2 -agonists), we evaluated elements reflecting these processes. Severe hypoxaemia (SpO 2 ≤ 88%) at initial presentation was no more common amongst those with versus without high lactate (48% vs. 43.6%, p = 0.7 and 41.6 vs. 42.9, p = 0.54). Systemic hypotension (SBP < 80 mm Hg), a surrogate index of tissue hypoperfusion, was uncommon and similar between lactate groups (6.7% vs. 4.2%, p = 0.28 and 1.4% vs. 1.4%, p = 0.98). Notably, lactate generally increased during the course of treatment. Amongst those with serial lactate T A B L E 2 Associations of blood lactate with characteristics of AECOPD events and clinical outcomes. Documented pre-hospital β 2 -agonist (salbutamol) dosage administration by paramedics was available in 99/142 (69.7%, prospective cohort). Amongst those with high initial lactate levels, prehospital salbutamol doses were higher (10.6 ± 9.1 vs. 6.8 ± 6.8, p = 0.02). In the retrospective cohort, quantification of salbutamol dose administered prior to venepuncture for lactate measurement could not be definitively confirmed. Total salbutamol dosages over the first 24 h of hospital presentation however, were greater in the high lactate group (40 mg  vs. 25 mg [10.2-35], p = 0.0001) and high treatment doses correlated with lactate (r = 0.4, p < 0.0001, Figure S2 in the Supporting Information). In the prospective cohort, cumulative salbutamol dosage administered prior to the first lactate measurement could be reliably calculated and was higher in the high lactate group (15 mg  vs. 10 mg [1.2-20], p = 0.017, Figure 2) with correlation between lactate and β 2 -agonist dose (r = 0.22, p = 0.008, Figure S2 in the Supporting Information).
In the prospective cohort (where accurate salbutamol dosage prior to lactate measurement could be confirmed), logistic regression analyses were used to analyse variables associated with elevated lactate (Table 3). In univariate analysis, neither prehospital hypoxaemia (SpO 2 < 88%), or FEV 1 were linked to elevated lactate. Exacerbation severity as measured by BAP-65 class was associated with elevated lactate values on univariate ( p = 0.02) analysis only. In multivariate analysis, only salbutamol dosage ( p = 0.01) and glucose ( p = 0.04) remained significant predictors of elevated lactate.

DISCUSSION
We observed elevated lactate that was not linked to hypoxia or infection in approximately half of people hospitalized with AECOPD. Instead, raised lactate was associated with metabolic acidosis, tachypnoea, tachycardia, hyperglycaemia and higher β 2 -agonist dosages. Strikingly, clinicians applied NIV almost four times more frequently in patients with high lactate. Although we report only an association, excessive doses of β 2 -agonists can cause lactic acidosis and combined manifestations of high lactate and high dose β 2 -agonists may provoke an impression of clinical deterioration. Persistent metabolic acidosis, tachycardia, tachypnoea and anxiety may be interpreted by clinicians as evidence of AECOPD unresponsive to therapy, rather than β 2 -agonist toxicity. Raised blood lactate could help to alert clinicians to β 2 -agonist overtreatment and merits further study.
Lactate elevation during AECOPD may be multifactorial. Anaerobic metabolism appeared unlikely to be a predominant factor as circulatory shock was rare and hypoxaemia was similar between groups. Importantly, lactate tended to rise during the course of treatment, a time when hypoxaemia was generally resolved but F I G U R E 2 Salbutamol dosage prior to lactate measurement.
T A B L E 3 Associations of clinical, biochemical and β 2 -agonist dosage with raised lactate measurements in the prospective cohort. cumulative β 2 -agonist dosages were increasing. β 2 -agonist dosage correlated with lactate in both cohorts and was a significant predictor of elevated lactate on univariate and multivariate analyses. Whilst in asthma, blood salbutamol levels are strongly correlated with lactate levels, 6 this laboratory investigation was not feasible in our setting. Beyond β 2 -agonists, alternative potential sources of lactate include increased respiratory muscle workload, 27 although high lactate levels have been observed in asthmatic patients with muscle paralysis. 28 The lungs themselves may be a potential source of lactate production but this has only been described in the context of acute lung injury. 29,30 Doses of β 2 -agonists administered were often high. International guidelines recommend conservative doses of bronchodilators in AECOPD although there is a limited evidence base from which to draw recommendations. 26,31 Whilst no clear threshold to define excessive β 2 -agonist dose in AECOPD has been established, a previous RCT compared dose regimens of 2.5 versus 5 mg nebulised every 4 h. 32 A dose >20 mg nebulised salbutamol had already been administered prior to lactate measurements (blood gas draw) in 31.4% of our prospective cohort. Importantly, escalating β 2 -agonist dosages in AECOPD have consistently been associated with limited additional bronchodilatation but increased side effects. 20 For example, a comparison of moderate versus high dose salbutamol for moderately severe hospitalized AECOPD suggested a trend towards longer hospitalization in the high dose group (9 vs. 6 days, p = 0.084). 32 Whilst very high dose β 2 -agonists has been recommended for severe asthma exacerbations 33 a similar approach may be deleterious in COPD. We found lactate to correlate with peak respiratory rate, potentially worsening dynamic hyperinflation during AECOPD. 9 In addition, lactate stimulates anxiety and lactate infusions have been used in psychiatry studies to induce panic attacks. 34,35 Even minor elevations in blood lactate can induce panic attacks in susceptible individuals. 36 Potential exists for a 'vicious circle' to be created whereby metabolic (lactic) acidosis, tachycardia, tachypnoea and anxiety due to overtreatment with β 2 -agonists may provoke an impression of clinical deterioration. This in turn may trigger increased treatment with β 2 -agonists and other modalities such as NIV.
The studies presented have a number of important limitations and do not validate lactate as a biomarker of β 2 -agonist overtreatment. Blood gas analyses were predominantly venous but pH, bicarbonate (actual) and lactate measurements obtained during venous or arterial blood gas sampling are relatively well matched in AECOPD. [37][38][39] Clinical management during AECOPD was at the discretion of treating hospital clinicians and treatment decisions such as increasing β 2 -agonist doses or instituting NIV were not standardized. Perhaps the key limitation in assessing the relationship of β 2 -agonists, lactate and clinical outcomes is potential confounding by exacerbation severity. High lactate measurements may simply reflect higher acuity presentations requiring higher β 2 -agonist doses and also greater need for non-invasive ventilation. Mitigating against this possibility are similar baseline severity in the high and low lactate groups in terms of hypoxia, PvCO 2 , baseline lung function, comorbidities and CAT scores. Exacerbation severity was assessed using BAP-65 scores and scores were slightly higher in the high lactate group. This was solely due to faster heart rates, potentially as a consequence of receiving higher β 2 -agonist doses. Our observations were performed during severe AECOPD, and whether similar relationships between lactate and β 2 -agonist dose exist in other settings (e.g., stable COPD, community AECOPD) was not studied.
In summary, elevated lactate measurements were prevalent during hospitalized AECOPD and associated with adverse metabolic and clinical outcomes. Raised lactate should prompt review of β 2 -agonist treatment to aid judicious AECOPD therapy. Apposite prospective studies are needed to validate lactate as a biomarker of β 2 -agonist excess in AECOPD.

CONFLICTS OF INTEREST STATEMENT
Martin I. MacDonald, Christian R. Osadnik and Paul T. King are Editorial Board members of Respirology and co-authors of this article. They were excluded from all editorial decision-making related to the acceptance of this article for publication. Philip G. Bardin is Co-Editor in Chief of Respirology and a co-author of this article. He was excluded from the peer-review process and all editorial decisions related to the acceptance and publication of this article. Peer-review was handled independently by Co-Editor in Chief Paul Reynolds to minimize bias. DATA AVAILABILITY STATEMENT Data available on request from the corresponding author due to privacy/ethical restrictions.

HUMAN ETHICS APPROVAL STATEMENT
This study included independent retrospective and prospective cohorts, approved by the Human Research Ethics Committee, Monash Health, Melbourne, Australia (HREC13291Q and HREC13134A). The prospective study obtained informed consent from all included study subjects.