key: cord-0026505-1m95556p authors: Wang, Jennifer M.; Han, MeiLan K.; Labaki, Wassim W. title: Chronic obstructive pulmonary disease risk assessment tools: is one better than the others? date: 2021-10-13 journal: Curr Opin Pulm Med DOI: 10.1097/mcp.0000000000000833 sha: 3bb08236bb4b4164be9af6aa449c37ffa1a37229 doc_id: 26505 cord_uid: 1m95556p Risk assessment tools are essential in COPD care to help clinicians identify patients at higher risk of accelerated lung function decline, respiratory exacerbations, hospitalizations, and death. RECENT FINDINGS: Conventional methods of assessing risk have focused on spirometry, patient-reported symptoms, functional status, and a combination of these tools in composite indices. More recently, qualitatively and quantitatively assessed chest imaging findings, such as emphysema, large and small airways disease, and pulmonary vascular abnormalities have been associated with poor long-term outcomes in COPD patients. Although several blood and sputum biomarkers have been investigated for risk assessment in COPD, most still warrant further validation. Finally, novel remote digital monitoring technologies may be valuable to predict exacerbations but their large-scale performance, ease of implementation, and cost effectiveness remain to be determined. SUMMARY: Given the complex heterogeneity of COPD, any single metric is unlikely to fully capture the risk of poor long-term outcomes. Therefore, clinicians should review all available clinical data, including spirometry, symptom severity, functional status, chest imaging, and bloodwork, to guide personalized preventive care of COPD patients. The potential of machine learning tools and remote monitoring technologies to refine COPD risk assessment is promising but remains largely untapped pending further investigation. Chronic obstructive pulmonary disease (COPD) is a heterogeneous condition characterized by different clinical phenotypes and highly variable disease courses [1 & ]. In the context of this complex heterogeneity, risk assessment tools are very valuable to clinicians as they seek to identify patients at higher risk of accelerated lung function decline, exacerbations, hospitalizations, and death. Initial studies mainly focused on prognostic data obtained from spirometry as the forced expiratory volume in 1 s (FEV 1 ) remains an important predictor of mortality [ ] have also been recognized as important for long-term prognosis. Furthermore, these metrics have been combined into a variety of composite indices [7 && ] that provide comprehensive risk assessment in COPD. More novel risk assessment tools include chest imaging-based metrics, such as the presence and severity of emphysema, large airways disease, and small airways abnormalities [8, 9] . Certain blood biomarkers, such as peripheral eosinophilia [10,11 && ,12], are associated with long-term outcomes in COPD and may also help predict response to therapy. Finally, new technologies, such as remote digital monitoring devices and mobile applications can provide personalized patient care and detect early clinical decline [13 && ,14 & ]. A summary of these risk assessment tools is provided in Table 1 . The severity of chronic respiratory symptoms, as assessed by validated tools, such as the Modified Medical Research Council (mMRC) dyspnea scale [19] and CAT [20 & ], is associated with long-term morbidity and mortality in COPD. Similarly, both the classic (presence of a productive cough for at least 3 months per year for 2 consecutive years) and SGRQ (productive cough that is near daily in frequency or occurring several days a week for 4 weeks) definitions of chronic bronchitis have been linked to an increased risk of COPD exacerbations [3] . However, only the SGRQ definition was associated with severe exacerbations requiring a visit to the emergency department or hospitalization. Even among ever-smokers without airflow obstruction, chronic bronchitis has been associated with accelerated lung function decline, higher hospitalization rates, and increased mortality [21 & ,22] . A history of COPD exacerbations is the best predictor of future exacerbations, regardless of Global Initiative for Chronic Obstructive Lung Disease (GOLD) grade [23] . Repeat exacerbations lead to excess FEV 1 decline over time, especially in patients with GOLD 1 COPD [24] . Furthermore, in a study of the UK Clinical Practice Research Datalink, the number of COPD exacerbations during the baseline year was associated with increased subsequent mortality over a mean of 4.9 years following a doseresponse relationship [25] . The GOLD 2017 ABCD classification based on symptom burden and exacerbation history [26] did not predict survival as well as the GOLD 2011 classification, which additionally incorporated FEV 1 [27] . The discriminative power of the 2011 and 2017 classifications for future exacerbations was nonetheless similar. It must be noted, however, that the main goal of these classifications is to guide clinical management strategies rather than predict long-term outcomes. COPD risk assessment tools include spirometry, patientreported symptoms, history of exacerbations, functional assessments, composite indices as well as novel blood, sputum, and chest imaging biomarkers. No single metric can fully capture the risk of poor longterm outcomes (including accelerated lung function decline, exacerbation, hospitalization, and death), given the clinical heterogeneity of COPD. Combinations of metrics through validated composite indices typically provide best risk assessment in COPD; ongoing work on machine learning tools may help refine risk assessment even further. Remote digital monitoring technologies to detect early clinical deterioration are promising risk assessment tools but require further research with regards to their performance and implementation. Chest computed tomography (CT) scans are frequently ordered in patients with COPD to screen for lung cancer, rule out acute pulmonary emboli, and assess candidacy for lung transplantation and lung volume reduction interventions. Although the primary indication for these scans is not long-term prognostication, they provide a wealth of information with regards to risk assessment in COPD through qualitative and quantitative evaluations of emphysema, large and small airways disease, pulmonary vascular abnormalities, and interstitial lung abnormalities (ILAs ]. In addition, visual emphysema scored on chest CT using the Fleischner Society classification system is associated with a higher risk of mortality following a dose-response relationship [34] . Similarly, quantitative emphysema defined as the percentage of lung volume with voxels less than -950 Hounsfield Units on noise-filtered low-dose CT scans has been independently associated with lung ]. The size distribution of low attenuation clusters on CT and the spatial heterogeneity of emphysema have also been shown to predict the risk of exacerbation, rate of lung function decline, and long-term mortality in COPD patients [36 & ]. Parametric Response Mapping (PRM) is a chest imaging analytic technique that pairs inspiratory and expiratory images to distinguish between emphysema and nonemphysematous air trapping referred to as functional small airways disease. Functional small airways disease has been associated with 5-year FEV 1 decline [9], 5-year emphysema progression [37] , and risk of consistent exacerbations, defined as at least one exacerbation every year for 3 years [8] . With regards to large airway disease, although airway wall thickness was not independently associated with mortality, bronchiectasis conferred a higher risk of hospitalization and mortality in patients with COPD [38 & ]. From a pulmonary vascular standpoint, increased pulmonary artery diameter (defined as a pulmonary artery : aorta ratio >1) has been associated with higher mortality and incidence of severe exacerbations [39,40 & ]. COPD patients with CT findings of ILAs, including reticular abnormalities, nodularity, ground glass opacities, traction bronchiectasis, honeycombing, nonemphysematous cysts or other evidence of architectural distortion, had an increased annual risk of moderate-to-severe COPD exacerba- ]. Further, progression of these radiologic fibrotic changes was associated with a higher rate of annual decline in FEV 1 and forced vital capacity (FVC). One of the most well studied biomarkers in COPD has been peripheral eosinophilia. A high blood eosinophil count (EOS) has been associated with a higher risk of moderate and severe COPD exacerbations [10] and a faster decline in FEV 1 [11 && ]. Importantly, change in blood EOS count after initiation of inhaled corticosteroid (ICS) therapy carries prognostic implications in patients with COPD. In a post hoc analysis of the Inhaled Steroids in Obstructive Lung Disease in Europe (ISOLDE) trial, patients with EOS suppression of at least 200 cells/ml following ICS initiation experienced slower FEV 1 decline and a lower incidence of COPD exacerbations over 3 years of follow-up, while those with EOS increase of at least 200 cells/ml following ICS initiation experienced opposite outcomes [42 && ]. When measured in the setting of an acute COPD exacerbation requiring hospitalization, low EOS (<50 cells/ml) has been associated with concurrent infection, longer ]. These results indicate that long-term mortality prediction in COPD is most accurate when respiratory parameters, such as lung function and dyspnea severity are combined with systemic factors, such as age, BMI, and exertional capacity. Beyond mortality prognostication, several composite tools have been used to predict other important outcomes, such as exacerbations, hospitalizations, and lung function decline. The Acute COPD Exacerbation Prediction Tool (ACCEPT), which was derived from pooled data of three randomized controlled trials, included several clinical and demographic variables, such as number of prior exacerbations, age, sex, BMI, smoking status, SGRQ score, postbronchodilator FEV 1 , and use of inhalers and oxygen therapy . The Summit Lab Score, which integrates age, BMI, smoking history, FEV 1 , heart rate, blood pressure, prior hospitalizations for COPD exacerbations, comorbidities (including myocardial infection, heart failure, and diabetes), and use of certain antithrombotic and antiarrhythmic medications, was associated with risk of exacerbations and length of hospital stay in COPD patients with ]. This model included quantitative CT imaging metrics and outperformed the BODE and ADO indices at predicting mortality at 7 years of follow-up. In addition to the aforementioned tools used for prediction of long-term outcomes in patients with stable COPD, other composite indices have been developed for short-term prognostication in the setting of acute COPD exacerbations. For example, the Integrated Pulmonary Index (combining endtidal carbon dioxide, respiratory rate, pulse rate, and oxygen saturation) and the Ottawa COPD Risk Score (using patient history, vital signs, imaging, lab work, and EKG) have helped emergency medicine physicians decide on disposition based on risk of severe short-term events within 30 days [72 & ]. These are just some of many available COPD risk assessment composite indices and are summarized in Table 2 . The widespread use of these indices ultimately depends on their practicality, cost, and performance in real-world settings. The newest frontier of COPD management involves the use of remote digital monitoring, electronic inhaler sensors, at-home spirometry devices, and mobile applications targeting early markers of clinical deterioration prior to a patient developing an exacerbation or presenting to the hospital. Remote monitoring devices can track parameters, such as heart rate, respiratory rate and pattern, sleep quality, physical activity, body temperature, oxygen saturation, and cough when connected to patients on clothing, via arm or wristbands, or directly to the torso or ear [13 && ]. In patients with stable COPD, parameters must be consistently obtained over a week but the data collected have been shown to correlate with symptom burden and inhaler usage [ ] but this technology remains limited by patient adherence. Although practical, at-home spirometry still faces challenges around its accuracy and lack of infrastructure within health systems to support its broad rollout [13 && ]. Further, several studies have examined whether phone or web-based applications can serve as risk assessment tools. For example, COPD-Predict TM , a novel application that uses a decision tree model to provide early warning signs of an exacerbation based on changes in symptoms, FEV 1 and C-reactive protein levels, has been shown to predict exacerbations, including severe exacerbations requiring hospitalization, in a small cohort of patients [14 & ]. Many health systems have not yet developed robust workflows to import these types of data into the electronic medical record nor have the associated reimbursement processes been fully established. Therefore, while remote digital monitoring technologies are promising tools to personalize preventive care in COPD, their performance, ease of implementation, and cost effectiveness still need to be further evaluated. Risk assessment tools have been extensively studied in COPD and have traditionally included spirometry, patient-reported symptoms, history of exacerbations, functional status, and combinations of these metrics in composite indices. More recently, specific blood, sputum, and chest imaging biomarkers have emerged as independent predictors of longterm outcomes in patients with COPD. In our practice, we administer CAT and record exacerbation history for all COPD patients to guide inhaler therapy and assess future exacerbation risk. We calculate the BODE score to estimate mortality risk and help guide decisions regarding lung transplantation. We also use the extent of emphysema and small airways disease on chest CT to determine candidacy for surgical or bronchoscopic lung volume reduction. In the setting of the complex heterogeneity of COPD with regards to both disease manifestation and progression, all available clinical information should be integrated to provide the best risk assessment. However, this strategy could be challenging in real-world settings depending on data accessibility, time constraints, and type of practice. Therefore, machine learning risk assessment tools may be very valuable in this context and warrant further investigation. Remote digital monitoring technologies may also prove to be another important risk assessment asset at the disposal of both patients and clinicians but questions regarding their accuracy, ease of use, and cost effectiveness still need to be addressed. In this multicenter longitudinal study of a primary care population of COPD patients, the CAT score was found to be linked to symptom severity, exacerbations, and overall physical and mental health, further confirming its role in COPD management. Fermont JM, Bolton CE, Fisk M, et al. Risk assessment for hospital admission in patients with COPD; a multicentre UK prospective observational study. PLoS One 2020; 15:e0228940. The Short Physical Performance Battery, which evaluates gait speed, balance, and sit-to-stand, was shown in this study to predict risk of hospitalization for COPD exacerbations as well as length of hospital stay. This study shows that reduced serum free light chains, which reflects impaired antibody production, was associated with an increased risk of COPD exacerbations; given that the trigger of these exacerbations is often a respiratory tract infection, adaptive immunity and antibody production are critical. Validation of COPDPredict TM : Unique combination of remote monitoring and exacerbation prediction to support preventive management of COPD exacerbations COPDPredict TM is one of the first digital applications that can provide early signs of an impending COPD exacerbation using decision tree models based on changes in symptoms, FEV 1 , and CRP levels Association of nonobstructive chronic bronchitis with respiratory health outcomes in adults This study, which pooled data from nine US general population-based cohorts, underscores the detrimental clinical consequences associated with chronic bronchitis symptoms in individuals without airflow obstruction Prognostic significance of chronic respiratory symptoms in individuals with normal spirometry Susceptibility to exacerbation in chronic obstructive pulmonary disease Acute exacerbations and lung function loss in smokers with and without chronic obstructive pulmonary disease Natural history of chronic obstructive pulmonary disease exacerbations in a general practice-based population with chronic obstructive pulmonary disease Global strategy for the diagnosis, management, and prevention of chronic obstructive lung disease 2017 report: GOLD executive summary Mortality and exacerbations by Global Initiative for Chronic Obstructive Lung Disease groups ABCD: 2011 versus 2017 in the COPDGene1 cohort This study shows the potential of automated emphysema quantification on lowdose chest CTs obtained for lung cancer screening to inform risk of lung cancer incidence Per cent low attenuation volume and fractal dimension of low attenuation clusters on CT predict different long-term outcomes in COPD Beyond severity of emphysema, this study shows how spatial clustering and distribution of emphysema are associated with long-term outcomes in COPD Voxel-wise longitudinal parametric response mapping analysis of chest computed tomography in smokers Impact of radiographic bronchiectasis in COPD This meta-analysis, which included data from 18 observational studies, confirms the long-term adverse implications of coexistent COPD and bronchiectasis Pulmonary arterial enlargement and acute exacerbations of COPD None. This study describes the first machine learning-derived mortality prediction model in COPD that includes quantitative CT imaging metrics, with a performance superior to those of the BODE and ADO indices. Kocak AO, Cakir Z, Akbas I, et al. Comparison of two scores of short term serious outcome in COPD patients. Am J Emerg Med 2020; 38:1086-1091. This study looked at a distinct population of emergency room COPD patients and found that the Integrated Pulmonary Index and the Ottawa COPD Risk Score could help physicians decide on disposition based on predicted risk of severe short-term events. This study shows the potential of electronic inhaler monitoring as a risk assessment tool in COPD patients with high healthcare utilization.