The COPD Biomarkers Qualification Consortium Database: Baseline Characteristics of the St George's Respiratory Questionnaire Dataset

Relationship of CT densitometry to lung physiological parameters and health status in alpha-1 antitrypsin deficiency: initial report of a centralised database of the NIHR rare diseases translational research collaborative

OBJECTIVES

To establish a database network for the study of alpha-1 antitrypsin deficiency (AATD) and compare the results to CT lung density as the most direct measure of emphysema.

DESIGN

A central electronic database was established to permit the upload of anonymised patient data from remote sites. Prospectively collected CT data were recorded onto disc, anonymised, analysed at the coordinating centre and compared with the clinical features of the disease.

SETTING

Tertiary referral centres with expertise in the management of AATD focused on academic Biomedical Research Units and Wellcome Clinical Research Facilities.

PARTICIPANTS

Data were collected from 187 patients over 1 year from eight UK academic sites. This included patient demographics, postbronchodilator physiology, health status and CT. Analysis was undertaken at the coordinating centre in Birmingham.

RESULTS

Patient recruitment in the 12 months reached 94% of target (set at 200) covering the whole spectrum of the disease from those with normal lung function to very severe chronic obstructive lung disease. CT scan suitable for analysis was available from 147 (79%) of the patients. CT density, analysed as the threshold for the lowest 15% of lung voxels, showed statistically significant relationships with the objective physiological parameters of lung function as determined by spirometric Global Initiative for Chronic Obstructive Lung Disease (GOLD) severity staging (p<0.001) and carbon monoxide gas transfer (p<0.01). Density also correlated with subjective measures of quality of life (p=0.02).

CONCLUSIONS

Establishment of the network for data collection and its transfer was highly successful facilitating future collaboration for the study of this rare disease and its management. CT densitometry correlated well with the objective clinical features of the disease supporting its role as the specific marker of the associated emphysema and its severity. Correlations with subjective measures of health, however, were generally weak indicating other factors play a role.

Statistical methods for biomarker data pooled from multiple nested case-control studies

Pooling biomarker data across multiple studies allows for examination of a wider exposure range than generally possible in individual studies, evaluation of population subgroups and disease subtypes with more statistical power, and more precise estimation of biomarker-disease associations. However, circulating biomarker measurements often require calibration to a single reference assay prior to pooling due to assay and laboratory variability across studies. We propose several methods for calibrating and combining biomarker data from nested case-control studies when reference assay data are obtained from a subset of controls in each contributing study. Specifically, we describe a two-stage calibration method and two aggregated calibration methods, named the internalized and full calibration methods, to evaluate the main effect of the biomarker exposure on disease risk and whether that association is modified by a potential covariate. The internalized method uses the reference laboratory measurement in the analysis when available and otherwise uses the estimated value derived from calibration models. The full calibration method uses calibrated biomarker measurements for all subjects, including those with reference laboratory measurements. Under the two-stage method, investigators complete study-specific analyses in the first stage followed by meta-analysis in the second stage. Our results demonstrate that the full calibration method is the preferred aggregated approach to minimize bias in point estimates. We also observe that the two-stage and full calibration methods provide similar effect and variance estimates but that their variance estimates are slightly larger than those from the internalized approach. As an illustrative example, we apply the three methods in a pooling project of nested case-control studies to evaluate (i) the association between circulating vitamin D levels and risk of stroke and (ii) how body mass index modifies the association between circulating vitamin D levels and risk of cardiovascular disease.

Socioeconomic Status as a Determinant of Health Status Treatment Response in COPD Trials

Background: Randomized controlled trials (RCTs) often recruit patients from low and high socioeconomic status (SES) countries, but little is known about the effect of SES on clinical outcomes, particularly patient-centered measures of symptomatic benefit. Methods: Combined individual chronic obstructive pulmonary disease (COPD) patient data from the placebo and long-acting bronchodilator arms of 17 RCTs (from the COPD Biomarkers Qualification Consortium database) were analyzed. Health status was measured using the St George's Respiratory Questionnaire (SGRQ) (minimum clinically important difference [MCID]: 4 units). Trials were grouped into short-term (≤12 months) and medium-term (>12 months to 48 months). A participant's country of residence was categorized into Low/Medium or High SES using World Health Organization criteria. Results: Data from 19765 individuals (6109 Low/Medium SES) were available. Patients in Low/Medium SES countries had more severe disease at baseline. Improvement in SGRQ score with placebo was ≈2 units greater in Low/Medium than in High SES countries; at its greatest, the improvement from baseline exceeded the MCID in Low/Medium countries. This difference was maintained for at least 1 year. Improvement with bronchodilator was also greater in Low/Medium versus High SES countries; overall there was no evidence that the treatment effect versus placebo was different between countries of different SES status. Conclusions: Participants in Low/Medium SES countries experienced significantly larger treatment effects, irrespective of treatment group (placebo and bronchodilator). Despite this, COPD patients in Low/Medium SES countries experienced a health status gain from long-acting bronchodilator treatment that is similar to that seen in High SES countries.

Baseline Severity as Predictor of Change in St George's Respiratory Questionnaire Scores in Trials of Long-acting Bronchodilators with COPD Patients

Background: In trials oflong-acting bronchodilators, health status is an important trial outcome, however the influence of baseline severity on response measured by St George's Respiratory Questionnaire (SGRQ) is not known. We have compared SGRQ changes between patients with chronic obstructive pulmonary disease (COPD) of mild-moderate severity or dyspnea (Global initiative for chronic Obstructive Lung disease [GOLD] grades 1 and 2; modified Medical Research Council [mMRC] grades 1 and 2) to those with severe-very severe severity or dyspnea (GOLD grades 3 and 4; mMRC grades 3 and 4). Methods: Combined individual patient data from the COPD Biomarkers Qualification Consortium database (trials of long-acting bronchodilators) were used comprising of patients from short-term (≤1-year duration; n=10802) and medium-term (2-4 years' duration; n=8963) studies. A repeated measures analysis of variance (ANOVA) was used to determine the effects of baseline severity (GOLD/mMRC) on SGRQ response to treatment. All treatment arms were combined. Results: In short-term studies, milder patients showed a greater response than those with more severe disease in terms of GOLD grade (partial Eta² = 0.03, p < 0.0001) and mMRC grade (partial Eta² = 0.05, p < 0.0001). Similar results were seen in the medium-term studies (partial Eta² = 0.02, p < 0.0001; mMRC: partial Eta² = 0.05, p < 0.0001,). Conclusions: Patients with less severe airflow limitation and less severe dyspnea showed larger improvements in SGRQ score than more severely obstructed or dyspneic patients. Although these severity influences are small (2%-5% of the variance in SGRQ score), they do suggest that pre-specified separate analyses are warranted to test for differences in response, based on baseline severity.

Responder Analyses for Treatment Effects in COPD Using the St George's Respiratory Questionnaire

Background: Patient-reported outcomes data in clinical trials are usually reported as mean values, interpreted in comparison to a minimum clinically important difference (MCID) and ignoring the possibility of a sizable proportion of patients experiencing a worthwhile benefit when the majority did not. This analysis tested the reliability of calculated responder rates (from chronic obstructive pulmonary disease [COPD] patients) with the St George's Respiratory Questionnaire (SGRQ) using a range of responder cut-points above and below the MCID (4 units). Methods: Individual patient data (i.e., data from long-acting bronchodilator [LAB] and inhaled corticosteroids [ICS]/long-acting beta2-agonist [LABA] randomized clinical studies) in the COPD Biomarker Qualification Consortium database were used: short-term (≤1-year duration; 14,814 patients,) and medium-term (2-4 years; 12,043 patients). Responder rates versus placebo across SGRQ score change thresholds ranging from -1.5 to -8.0 were tested; differences were expressed as the odds ratio (OR) of a patient exceeding the threshold versus no change or deterioration. Results: The ORs measuring benefit of active treatment were similar across thresholds in short-term studies (LAB, ORs 1.40-1.42; LABA/ICS, 1.50-1.56) and medium-term LAB studies (ORs 1.34-1.43), whereas ORs in medium-term studies with LABA/ICS intervention showed a trend for higher response rates at higher values of threshold cut-points (1.64-1.79). In short-term studies, different thresholds had little effect on the OR between active drugs versus a trend for lower ORs with lower thresholds in medium-term studies. Conclusions: The OR for a treatment effect compared with placebo appears consistent across a range of responder cut-points. In medium-term trials, the treatment difference between active drugs suggests that use of a lower threshold would not increase the odds of observing a measured treatment difference.