Biological clustering supports both Dutch andBritish hypotheses of Asthma and COPD

Michael A Ghebre

University of Leicester

III Annual PGR ConferenceUniversity of Leicester, UK

31st March 2014

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Introduction and Objectives

Severe asthma and chronic obstructive pulmonary disease (COPD)are complex and heterogeneous diseases.

”Dutch hypothesis” => Asthma and COPD are the same diseasewith the same underlying mechanism.

”British hypothesis” => Asthma and COPD are two distinctdisease with di↵erent underlying mechanism.

We investigated whether they represent distinct or overlappingconditions in terms of their available measurements

We do have access to a large number of measurements on 86asthma and 75 COPD subjects; i.e.

Demographic characteristics

Clinical and lung function

Biological characteristics (i.e sputum cytokines)

This study was validated on an independent severe asthma (n=166)and COPD (n=58) using two approaches.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary statistics of some characteristics...

Table 1. Demographic, lung function and clinical characteristics across Asthma and COPDVariable Asthma=86 COPD=75 P-valueMale [n(%)] 43 (50) 53 (70.7) 0.008Current or Ex- smokers [n (%)] 32 (37.2) 72 (96.0) < 0.0001Pack -year history 4.6 (2.98 - 7.26) 40 (34.46 - 46.39) < 0.0001Age (years)+ 54 (1.3) 69 (1.1) < 0.0001Duration of Disease (years) 21 (16.4 - 26.5) 5 (4.12 - 6.55) < 0.0001BMI (kg/m2)+ 30.4 (0.8) 25.7 (0.5) < 0.0001Pre FEV1/FVC ratio (%)+ 67.6 (1.5) 49.8 (1.6) < 0.0001Pre FEV1 Predicted (%)+ 74.6 (2.4) 45.4 (2.1) < 0.0001Post FEV1 Predicted (%)+ 79.8 (2.4) 47.1 (2.1) < 0.0001Sputum Neutrophil count (%)+ 63.2 (2.5) 69.7 (2.5) 0.07Sputum Eosinophil count (%)+ 2.1 (1.38 - 3.1) 1.4(0.98 - 1.93) 0.14Sputum Macrophage count (%)+ 16.7 (13.41 - 20.78) 16.2 (13.3 - 19.8) 0.84Data presented as Geometric mean with corresponding confidence interval unless stated, + Mean (SEM)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Cytokines across Asthma and COPD

Figure : A=Higher in Asthma and B=Higher in COPD

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Where Asthma(A) and COPD (C) overlap?

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

Cluster analysis is unsupervised statistical techniques which identifieshomogeneous subgroups and creates labels for further supervisedtechniques (like linear regression or discriminant analysis)

The main aim this study is to segment Asthma and COPD patientsin to subgroups by their biological (i.e. sputum cytokines)characteristics

This might help to identify subjects phenotype rather than diseasespecific categorisation( which is already labelled as asthma orCOPD)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

Cluster analysis is unsupervised statistical techniques which identifieshomogeneous subgroups and creates labels for further supervisedtechniques (like linear regression or discriminant analysis)

The main aim this study is to segment Asthma and COPD patientsin to subgroups by their biological (i.e. sputum cytokines)characteristics

This might help to identify subjects phenotype rather than diseasespecific categorisation( which is already labelled as asthma orCOPD)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

Cluster analysis is unsupervised statistical techniques which identifieshomogeneous subgroups and creates labels for further supervisedtechniques (like linear regression or discriminant analysis)

The main aim this study is to segment Asthma and COPD patientsin to subgroups by their biological (i.e. sputum cytokines)characteristics

This might help to identify subjects phenotype rather than diseasespecific categorisation( which is already labelled as asthma orCOPD)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

Cluster analysis is unsupervised statistical techniques which identifieshomogeneous subgroups and creates labels for further supervisedtechniques (like linear regression or discriminant analysis)

The main aim this study is to segment Asthma and COPD patientsin to subgroups by their biological (i.e. sputum cytokines)characteristics

This might help to identify subjects phenotype rather than diseasespecific categorisation( which is already labelled as asthma orCOPD)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Clustering heterogeneous subjects

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Clustering heterogeneous subjects

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Clustering heterogeneous subjects

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

There are 18 sputum cytokines measurements

Factor analysis was performed on these cytokines, and fourindependent factor scores were extracted

The factor scores were used as input variables in K-means clusteranalysis

Thus, three biological clusters was identified

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

There are 18 sputum cytokines measurements

Factor analysis was performed on these cytokines, and fourindependent factor scores were extracted

The factor scores were used as input variables in K-means clusteranalysis

Thus, three biological clusters was identified

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

There are 18 sputum cytokines measurements

Factor analysis was performed on these cytokines, and fourindependent factor scores were extracted

The factor scores were used as input variables in K-means clusteranalysis

Thus, three biological clusters was identified

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

There are 18 sputum cytokines measurements

Factor analysis was performed on these cytokines, and fourindependent factor scores were extracted

The factor scores were used as input variables in K-means clusteranalysis

Thus, three biological clusters was identified

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Asthma and COPD Biological Clustering

There are 18 sputum cytokines measurements

Factor analysis was performed on these cytokines, and fourindependent factor scores were extracted

The factor scores were used as input variables in K-means clusteranalysis

Thus, three biological clusters was identified

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Three biological Asthma and COPD clusters

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Pattern of cytokines across the biological clusters

Table 2. Summary statistics across the three identified biological clusters in the test study

Variable Cluster 1 Cluster 2 Cluster 3 C1 vs. C2 C1 vs. C3 C2 vs.C3

(Asthma=55, COPD=3) (Asthma=28, COPD=19) (Asthma=2, COPD=39) P-value P-value P-value

Male [n (%)] 32 (55.2) 26 (55.3) 28 (68.3) 0.99 0.19 0.21

Current or Ex-smokers [n (%)] 22 (37.9) 29 (61.7) 40 (97.6) 0.015 <0.0001 <0.000

Age (years) + 55 (1.5) 60 (2.1) 67 (1.8) 0.038 <0.0001 0.008

Pre FEV1/FVC ratio (%)+ 69.0 (1.9) 58.5 (2.3) 49.7 (2.4) <0.0001 <0.0001 0.011

Pre FEV1 Predicted (%)+ 77.0 (2.7) 59.9 (3.7) 47.0 (3) <0.0001 <0.0001 0.01

Post FEV1 Predicted (%)+ 81.7 (2.7) 63.9 (3.9) 49.1 (3) <0.0001 <0.0001 0.005

Sputum Neutrophil count (%)+ 58.8 (3.1) 77.2 (3) 59.1 (3.1) <0.0001 0.95 <0.0001

Sputum Eosinophil count (%) 3.9 (2.4 - 6.4) 0.7 (0.5 - 0.9) 2.0 (1.25 - 3.17) <0.0001 0.039 <0.0001

CFU >107/ml or positive culture (n[%]) 8 (13.8) 26(55.3) 9 (21.9) <0.0001 0.29 0.001

IL-1� (pg/ml) 39.5 (30.8 - 50.8) 379.5 (257.3 - 559.8) 23.5 (17.2 - 32.2) <0.0001 0.025 <0.0001

IL-5 (pg/ml) 2.6 (1.6 - 4.2) 2.2 (1.4 - 3.4) 1.4 (0.9 - 2.2) 0.56 0.083 0.22

IL-6 (pg/ml) 21.3 (15 - 30.4) 271.4 (192.2 - 383.3) 486.2 (327.7 - 721.4) <0.0001 <0.0001 0.031

IL-6R (pg/ml) 163.2 (126.0 - 211.6) 433.4 (344.2 - 545.6) 112.4 (88.6 - 142.6) <0.0001 0.04 <0.0001

IL-10 (pg/ml) 0.33 (0.25 - 0.45) 5.5 (3.5 - 8.7) 0.34 (0.2 - 0.6) <0.0001 0.89 <0.0001

IL-13 (pg/ml) 10.4 (7.7 - 14.0) 4.8 (3.8 - 6.2) 3.5 (2.4 - 5.2) 0.001 <0.0001 0.18

CCL-26 (pg/ml) 12.4 (7.8 - 19.9) 5.0 (3.4 - 7.5) 2.9 (1.9 - 4.6) 0.004 <0.0001 0.081

TNF↵ (pg/ml) 1.4 (1. 1 - 1.9) 29.9 (19.5 - 45.9) 1.7 (1.1 - 2.5) <0.0001 0.62 <0.0001

Data presented as Geometric mean with corresponding confidence interval unless stated, + Mean (SEM) ; CFU= colony forming units;C=cluster

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation Study

This study was validated on independent severe asthma and COPDstudies.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation Study

This study was validated on independent severe asthma and COPDstudies.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Asthma and COPD Validation Study

For validation, independent severe asthma (n=166) and COPD(n=58) patients were used

Two classification techniques were applied; i.e.

1. Linear discriminant analysis (LDA)

2. The diseases status (Asthma or COPD), and IL-1� cuto↵ for theoverlap

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Asthma and COPD Validation Study

For validation, independent severe asthma (n=166) and COPD(n=58) patients were used

Two classification techniques were applied; i.e.

1. Linear discriminant analysis (LDA)

2. The diseases status (Asthma or COPD), and IL-1� cuto↵ for theoverlap

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Asthma and COPD Validation Study

For validation, independent severe asthma (n=166) and COPD(n=58) patients were used

Two classification techniques were applied; i.e.

1. Linear discriminant analysis (LDA)

2. The diseases status (Asthma or COPD), and IL-1� cuto↵ for theoverlap

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Asthma and COPD Validation Study

For validation, independent severe asthma (n=166) and COPD(n=58) patients were used

Two classification techniques were applied; i.e.

1. Linear discriminant analysis (LDA)

2. The diseases status (Asthma or COPD), and IL-1� cuto↵ for theoverlap

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Asthma and COPD Validation Study

For validation, independent severe asthma (n=166) and COPD(n=58) patients were used

Two classification techniques were applied; i.e.

1. Linear discriminant analysis (LDA)

2. The diseases status (Asthma or COPD), and IL-1� cuto↵ for theoverlap

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Characteristics of the Validation study

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Classification model was developed from the Test study using LDA

Then subjects were assigned to subgroups in which he/she has thehighest discriminant score/posterior probability.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Classification model was developed from the Test study using LDA

Then subjects were assigned to subgroups in which he/she has thehighest discriminant score/posterior probability.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Classification model was developed from the Test study using LDA

Then subjects were assigned to subgroups in which he/she has thehighest discriminant score/posterior probability.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Classification model was developed from the Test study using LDA

Then subjects were assigned to subgroups in which he/she has thehighest discriminant score/posterior probability.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Table 3. Coe�cients (betas) and prior probability of each cluster in the teststudy which used to predict class membership in the validation studyVariables Cluster 1 (�1k) Cluster 2 (�2k) Cluster 3 (�3k)IL-1� 1.63 2.17 0.96IL-5 -6.31 -6.50 -6.79IL-6 -2.62 -1.17 0.14IL-6R 6.28 5.58 4.72IL-8 6.64 7.59 7.31CCL-2 5.56 5.66 6.98CCL-4 3.59 3.83 4.66CCL-5 -1.36 -0.80 -2.83CCL-13 -0.04 -0.49 -0.21CCL-17 2.13 1.09 1.09CXCL-11 0.48 0.51 0.29TNF↵ -5.77 -5.28 -6.57Constant -63.39 -75.75 -79.11Prior Probability 0.33 0.33 0.33

Unit of all cytokines is pg/ml

Sij = �j + �j1Y1i + �j2Y2i + · · ·+ �jkYki + log(Pj)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Patterns of characteristics across validation subgroups

Table 4. Summaries across the validation subgroups which were identified using linear discriminant analysis (LDA)

Variable Group 1 Group 2 Group 3 G1 vs. G2 G1 vs. G3 G2 vs.G3

(Asthma=94, COPD=12) (Asthma=55, COPD=18) (Asthma=7, COPD=28) P-value P-value P-value

Male [n (%)] 65 (61.3) 40 (54.8) 25 (71.4) 0.38 0.28 0.1

Current or Ex-smokers 37 (34.9) 42 (57.5) 33 (94.3) 0.003 <0.0001 <0.0001

Age (year)+ 53 (1.3) 56 (2.0) 66 (2.0) 0.15 <0.0001 0.002

Pre FEV1/FVC ratio (%)+ 68.2 (1.4) 61.8 (1.9) 57.3 (1.8) 0.004 <0.0001 0.14

Pre FEV1 Predicted (%)+ 67.6 (2.2) 64.7 (2.9) 65.5 (3.2) 0.4 0.65 0.85

Sputum Neutrophil count (%)+ 59.1 (2.8) 72.3 (2.5) 58.4 (4.0) 0.001 0.88 0.003

Sputum Eosinophil count (%) 6.3 (4.9 - 8.2) 3.0 (2.4 - 3.8) 3.2 (2.3 - 4.4) <0.0001 0.003 0.79

IL-1� (pg/ml) 54.1 (42.7 - 68.5) 526.6 (375.5 - 738.4) 42.4 (26 - 69.2) <0.0001 0.36 <0.0001

IL-5 (pg/ml) 1.4 (1.1 - 1.8) 1.2 (0.8 - 1.7) 0.9 (0.5 - 1.5) 0.43 0.085 0.34

IL-6 (pg/ml) 26.1 (19.9 - 34.3) 273.3 (210.6 - 354.6) 344.2 (237.2 - 499.3) <0.0001 <0.0001 0.32

IL-6R (pg/ml) 186.0 (135.5 - 255.3) 589.8 (476.6 - 729.8) 90.5 (50.3 - 162.8) <0.0001 0.013 <0.0001

IL-10 (pg/ml) 1.9 (1.5 - 2.4) 3.9 (2.8 - 5.4) 0.5 (0.3 - 0.9) 0.001 <0.0001 <0.0001

TNF↵ (pg/ml) 1.4 (1.2 - 1.7) 23.3(16.1 - 33.7) 1.8 (1.1 - 3.0) ¡0.0001 0.32 <0.0001

Data presented as geometric mean with corresponding confidence interval unless stated, + Mean (SEM); G=Group

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Cytokines profiles in the test and validation subgroups

Figure : Test Figure : Validation

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation Results: Using LDA

Validation Results

The sputum cellular and cytokines patterns were very similar between thetest and validation subgroups.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validating Asthma and COPD using LDA

Table 3. Coe�cients (betas) and prior probability of each cluster in the teststudy which used to predict class membership in the validation studyVariables Cluster 1 (�1k) Cluster 2 (�2k) Cluster 3 (�3k)IL-1� 1.63 2.17 0.96IL-5 -6.31 -6.50 -6.79IL-6 -2.62 -1.17 0.14IL-6R 6.28 5.58 4.72IL-8 6.64 7.59 7.31CCL-2 5.56 5.66 6.98CCL-4 3.59 3.83 4.66CCL-5 -1.36 -0.80 -2.83CCL-13 -0.04 -0.49 -0.21CCL-17 2.13 1.09 1.09CXCL-11 0.48 0.51 0.29TNF↵ -5.77 -5.28 -6.57Constant -63.39 -75.75 -79.11Prior Probability 0.33 0.33 0.33

Unit of all cytokines is pg/ml

Sij = �j + �j1Y1i + �j2Y2i + · · ·+ �jkYki + log(Pj)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

IL-1� across the three clusters

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

IL-1� across the three clusters at 130 pg/ml

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation using IL-1� cuto↵ for the overlap and diseases

Since IL-1� performed well in discriminating the overlap group in thetest study at 130 pg/ml cuto↵

This cuto↵ point was applied to the validation studies

With assumption that asthma and COPD are clinicallydistinguishable diseases

Then the validation dataset classified in to three subgroups

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation using IL-1� cuto↵ for the overlap and diseases

Since IL-1� performed well in discriminating the overlap group in thetest study at 130 pg/ml cuto↵

This cuto↵ point was applied to the validation studies

With assumption that asthma and COPD are clinicallydistinguishable diseases

Then the validation dataset classified in to three subgroups

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation using IL-1� cuto↵ for the overlap and diseases

Since IL-1� performed well in discriminating the overlap group in thetest study at 130 pg/ml cuto↵

This cuto↵ point was applied to the validation studies

With assumption that asthma and COPD are clinicallydistinguishable diseases

Then the validation dataset classified in to three subgroups

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation using IL-1� cuto↵ for the overlap and diseases

Since IL-1� performed well in discriminating the overlap group in thetest study at 130 pg/ml cuto↵

This cuto↵ point was applied to the validation studies

With assumption that asthma and COPD are clinicallydistinguishable diseases

Then the validation dataset classified in to three subgroups

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation using IL-1� cuto↵ for the overlap and diseases

Since IL-1� performed well in discriminating the overlap group in thetest study at 130 pg/ml cuto↵

This cuto↵ point was applied to the validation studies

With assumption that asthma and COPD are clinicallydistinguishable diseases

Then the validation dataset classified in to three subgroups

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Patterns of characteristics across the groups in validation

Table 5. Summaries across the validation subgroups which were identified using IL-1� cuto↵ and diseases status (asthma or COPD)

Variable Group 1 Group 2 Group 3 G1 vs. G2 G1 vs. G3 G2 vs.G3

(Asthma=103) (Asthma=63, COPD=26) (COPD=32) P-value P-value P-value

Male [n (%)] 66 (64.1) 51 (57.3) 22 (68.7) 0.34 0.63 0.26

Current or Ex-smokers 28 (27.2) 54 (60.7) 32 (100) <0.0001 <0.0001 <0.0001

Age (year)+ 49 (1.2) 57 (1.8) 68 (1.6) <0.0001 <0.0001 0.001

Pre FEV1/FVC ratio (%)+ 70.1 (1.2) 61.0 (1.7) 56.6 (2.4) <0.0001 <0.0001 0.16

Pre FEV1 Predicted (%)+ 71.2 (2.3) 64.1 (2.5) 60.2 (3.6) 0.033 0.019 0.4

Sputum Neutrophil count (%)+ 59.0 (2.8) 70.4 (2.5) 59.2 (4.2) 0.003 0.97 0.023

Sputum Eosinophil count (%) 5.6 (4.3 - 7.3) 3.5 (2.8 - 4.4) 3.9 (2.5 - 5.9) 0.009 0.12 0.68

IL-1� (pg/ml) 37.0 (29.1 - 47) 527.1(407.1 - 682.5) 40.0 (28.2 - 56.6) <0.0001 0.75 <0.0001

IL-5 (pg/ml) 1.2 (1.0 - 1.5) 1.3 (1.0 - 1.9) 1.0 (0.6 - 1.7) 0.69 0.45 0.37

IL-6 (pg/ml) 34.7 (25.6 - 47) 190 (138.4 - 261) 157.7 (88.8 - 280.2) <0.0001 <0.0001 0.56

IL-6R (pg/ml) 153(102.7 - 228.6) 549.4 (454.9 - 663.6) 101.7(74.1-139.5) <0.0001 0.17 <0.0001

IL-10 (pg/ml) 2.2 (1.8 - 2.6) 3.1 (2.2 - 4.3) 0.4 (0.2 - 0.6) 0.063 < 0.0001 <0.0001

TNF- (pg/ml) 1.6 (1.3 - 2.0) 14.4 (9.8 - 21.2) 1.3 (0.7 - 2.1) <0.0001 0.39 <0.0001

Data presented as geometric mean with corresponding confidence interval unless stated, + Mean (SEM) ; CFU= colony forming units; G=Group

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Cytokines profiles in the test and validation subgroups

Figure : Test Figure : Validation ( using IL-1�)

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Validation Results: Using diseases and IL-1� cuto↵

Validation Results

The sputum cellular and cytokines patterns were also very similarbetween the test and validation subgroups.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

Cytokine profiles in the test and validation by clusters

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

TNF↵ as alternative for IL-1�

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

TNF↵ as alternative for IL-1�

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Validating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

TNF↵ as alternative for IL-1�

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Table of contents

1 Introduction and Objectives

2 Asthma and COPD Biological Clustering

3 Asthma and COPD Validation StudyValidating Asthma and COPD using LDAValidating Asthma and COPD using IL-1� cuto↵

4 Summary and Conclusion

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

Three biological clusters were identified which are...

Cluster 1

Eosinophilic asthma predominant (95% asthma) high Th-2 cytokines

Cluster 2

Neutrophilic asthma and COPD overlap representing about one-third ofeach disease group, high in Th-1 and proinflmmatory(PI) cytokines

Cluster 3

Non-neutrophilic COPD predominant (95% COPD) and high in PIcytokines

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

Three biological clusters were identified which are...

Cluster 1

Eosinophilic asthma predominant (95% asthma) high Th-2 cytokines

Cluster 2

Neutrophilic asthma and COPD overlap representing about one-third ofeach disease group, high in Th-1 and proinflmmatory(PI) cytokines

Cluster 3

Non-neutrophilic COPD predominant (95% COPD) and high in PIcytokines

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

Three biological clusters were identified which are...

Cluster 1

Eosinophilic asthma predominant (95% asthma) high Th-2 cytokines

Cluster 2

Neutrophilic asthma and COPD overlap representing about one-third ofeach disease group, high in Th-1 and proinflmmatory(PI) cytokines

Cluster 3

Non-neutrophilic COPD predominant (95% COPD) and high in PIcytokines

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

Three biological clusters were identified which are...

Cluster 1

Eosinophilic asthma predominant (95% asthma) high Th-2 cytokines

Cluster 2

Neutrophilic asthma and COPD overlap representing about one-third ofeach disease group, high in Th-1 and proinflmmatory(PI) cytokines

Cluster 3

Non-neutrophilic COPD predominant (95% COPD) and high in PIcytokines

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

Three biological clusters were identified which are...

Cluster 1

Eosinophilic asthma predominant (95% asthma) high Th-2 cytokines

Cluster 2

Neutrophilic asthma and COPD overlap representing about one-third ofeach disease group, high in Th-1 and proinflmmatory(PI) cytokines

Cluster 3

Non-neutrophilic COPD predominant (95% COPD) and high in PIcytokines

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

This study was validated on independent study using two methods,and reveal very similar pattern in cytokines and sputum cellular

IL-1� (at 130 pg/ml) and TNF↵ (at 5 pg/ml) preformed reasonablywell in discriminating the overlap subgroup, and could be potentialfuture markers

Conclusion

In conclusion, sputum cytokines can determine distinct and overlappinggroups of patients with severe asthma and COPD, and may contribute toimproved patients classification to enable stratified medicine!

However,for generalisation this findings need to be replicate inmulti-studies

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

This study was validated on independent study using two methods,and reveal very similar pattern in cytokines and sputum cellular

IL-1� (at 130 pg/ml) and TNF↵ (at 5 pg/ml) preformed reasonablywell in discriminating the overlap subgroup, and could be potentialfuture markers

Conclusion

In conclusion, sputum cytokines can determine distinct and overlappinggroups of patients with severe asthma and COPD, and may contribute toimproved patients classification to enable stratified medicine!

However,for generalisation this findings need to be replicate inmulti-studies

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

This study was validated on independent study using two methods,and reveal very similar pattern in cytokines and sputum cellular

IL-1� (at 130 pg/ml) and TNF↵ (at 5 pg/ml) preformed reasonablywell in discriminating the overlap subgroup, and could be potentialfuture markers

Conclusion

In conclusion, sputum cytokines can determine distinct and overlappinggroups of patients with severe asthma and COPD, and may contribute toimproved patients classification to enable stratified medicine!

However,for generalisation this findings need to be replicate inmulti-studies

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

This study was validated on independent study using two methods,and reveal very similar pattern in cytokines and sputum cellular

IL-1� (at 130 pg/ml) and TNF↵ (at 5 pg/ml) preformed reasonablywell in discriminating the overlap subgroup, and could be potentialfuture markers

Conclusion

In conclusion, sputum cytokines can determine distinct and overlappinggroups of patients with severe asthma and COPD, and may contribute toimproved patients classification to enable stratified medicine!

However,for generalisation this findings need to be replicate inmulti-studies

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Summary and Conclusion

This study was validated on independent study using two methods,and reveal very similar pattern in cytokines and sputum cellular

IL-1� (at 130 pg/ml) and TNF↵ (at 5 pg/ml) preformed reasonablywell in discriminating the overlap subgroup, and could be potentialfuture markers

Conclusion

In conclusion, sputum cytokines can determine distinct and overlappinggroups of patients with severe asthma and COPD, and may contribute toimproved patients classification to enable stratified medicine!

However,for generalisation this findings need to be replicate inmulti-studies

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

References

Carolan BJ and Sutherland ER (2013); Clinical phenotypes of chronic obstructivepulmonary disease and asthma: Recent advances.

Tabachnick BG and Fidell LS (2007); Using Multivariate Statistics. FIFTH ed.

Everitt B (2011); An Introduction to Applied Multivariate Analysis with R.

Gregory R and Samuelsen KM (2008); Advances in Latent Variable MixtureModels.

Barnes PJ (2008); The cytokine network in asthma and chronic COPD disease.

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD

Introduction and ObjectivesAsthma and COPD Biological Clustering

Asthma and COPD Validation StudySummary and Conclusion

References

Acknowledgement

Prof Chris Brightling, Prof Paul Burton & Dr Chris Newby

Dr Mona Bafadhel & Dr Dhan Desai

Michael A Ghebre University of Leicester Biological clustering supports both Dutch and British hypotheses of Asthma and COPD