User login
Although genomic testing is useful when an interstitial lung disease diagnosis is uncertain, the testing results themselves aren’t sufficient to make the diagnosis, Daniel Dilling, MD, FCCP, said in a presentation at the annual meeting of the American College of Chest Physicians, which was held virtually.
The genomic classifier (Envisia, Veracyte) helps differentiate idiopathic pulmonary fibrosis (IPF) by detecting usual interstitial pneumonia (UIP), the hallmark pattern of this interstitial lung disease.
However, UIP is just one piece of the larger diagnostic puzzle, according to Dr. Dilling, professor of medicine in the interstitial lung disease program at Loyola University Medical Center in Maywood, Ill.
“Remember, it’s just a pattern, and not a diagnosis of IPF,” Dr. Dilling said in his presentation.
Genomic classifier results correlate well with both histologic and radiographic UIP pattern, studies show.
However, Dr. Dilling said the value of the genomic classifier is not in isolation.
“We don’t use this in a vacuum,” he said. “It increases our confidence and consensus, but it has to be incorporated into a multidisciplinary discussion group.”
Part of the diagnostic pathway
Dr. Dilling said the genomic classifier should be considered part of a diagnostic pathway in uncertain cases, particularly when the risk of surgical lung biopsy is high.
Current clinical practice guidelines recommend surgical lung biopsy for histopathologic diagnosis when clinical and radiologic findings are not definitive for IPF, the speaker said.
However, surgical lung biopsy carries some risk, and sometimes it can’t be done, he added.
In his presentation, Dr. Dilling cited a systematic review and meta-analysis of 23 studies looking at surgical lung biopsy for the diagnosis of interstitial lung diseases.
The postoperative mortality rate was 3.6% in that meta-analysis, published in 2015 in the Journal of Thoracic and Cardiovascular Surgery.
“The final decision regarding whether or not to perform a [surgical lung biopsy] must be based on the balance between benefits to establish a secure diagnosis and the potential risks,” authors wrote at the time.
Mortality risk is higher in immunocompromised and acutely ill patient populations, according to Dr. Dilling, who added that as many of 19% of patients will have complications from surgical lung biopsy.
Genomic classifier studies
In a proof-of-principle study, published in 2017 in the Annals of the American Thoracic Society, authors described how they used machine learning to train an algorithm to distinguish UIP from non-UIP pattern in tissue obtained by transbronchial biopsy (TBB).
The top-performing algorithm distinguished UIP from non-UIP conditions in single TBB samples with specificity of 86% and sensitivity of 63%, according to investigators, who said at the time that independent validation would be needed before the genomic classifier could be applied in clinical settings.
In a prospective validation study, published in 2019 in The Lancet Respiratory Medicine, the genomic classifier identified UIP in TBB samples from 49 patients with a specificity of 88% and sensitivity of 70%.
Excluding patients with definite or probable UIP as shown on high-resolution computed tomography, results show that the classifier had a sensitivity of 76%, specificity of 88%, and positive predictive value of 81%.
“The performance of the test is good, even in that scenario,” Dr. Dilling said.
Real-world results
Dr. Dilling also highlighted a “real-world” study, published earlier in 2021, demonstrating that UIP pattern recognized by a genomic classifier had encouraging sensitivity and specificity when combined with high-resolution CT and clinical factors.
That study included 96 patients who had both diagnostic lung pathology and a transbronchial lung biopsy for molecular testing with the classifier.
The classifier had a sensitivity of 60.3% and a specificity of 92.1% for histology-proven UIP pattern, investigators said in their report, which appears in the American Journal of Respiratory and Critical Care Medicine.
Local radiologists identified UIP with a sensitivity of 34.0% and specificity of 96.9%. But adding genomic classifier testing to local radiology testing increased the diagnostic yield, investigators said, with a sensitivity of 79.2% and specificity of 90.6%.
“This might suggest that the implementation of this into a local [multidisciplinary discussion] with your local radiology expertise might really improve your recognition of UIP,” Dr. Dilling said.
Dr. Dilling reported disclosures related to Bellerophon, Boehringer Ingelheim, Genentech, Nitto Denko, and Lung Bioengineering.
Although genomic testing is useful when an interstitial lung disease diagnosis is uncertain, the testing results themselves aren’t sufficient to make the diagnosis, Daniel Dilling, MD, FCCP, said in a presentation at the annual meeting of the American College of Chest Physicians, which was held virtually.
The genomic classifier (Envisia, Veracyte) helps differentiate idiopathic pulmonary fibrosis (IPF) by detecting usual interstitial pneumonia (UIP), the hallmark pattern of this interstitial lung disease.
However, UIP is just one piece of the larger diagnostic puzzle, according to Dr. Dilling, professor of medicine in the interstitial lung disease program at Loyola University Medical Center in Maywood, Ill.
“Remember, it’s just a pattern, and not a diagnosis of IPF,” Dr. Dilling said in his presentation.
Genomic classifier results correlate well with both histologic and radiographic UIP pattern, studies show.
However, Dr. Dilling said the value of the genomic classifier is not in isolation.
“We don’t use this in a vacuum,” he said. “It increases our confidence and consensus, but it has to be incorporated into a multidisciplinary discussion group.”
Part of the diagnostic pathway
Dr. Dilling said the genomic classifier should be considered part of a diagnostic pathway in uncertain cases, particularly when the risk of surgical lung biopsy is high.
Current clinical practice guidelines recommend surgical lung biopsy for histopathologic diagnosis when clinical and radiologic findings are not definitive for IPF, the speaker said.
However, surgical lung biopsy carries some risk, and sometimes it can’t be done, he added.
In his presentation, Dr. Dilling cited a systematic review and meta-analysis of 23 studies looking at surgical lung biopsy for the diagnosis of interstitial lung diseases.
The postoperative mortality rate was 3.6% in that meta-analysis, published in 2015 in the Journal of Thoracic and Cardiovascular Surgery.
“The final decision regarding whether or not to perform a [surgical lung biopsy] must be based on the balance between benefits to establish a secure diagnosis and the potential risks,” authors wrote at the time.
Mortality risk is higher in immunocompromised and acutely ill patient populations, according to Dr. Dilling, who added that as many of 19% of patients will have complications from surgical lung biopsy.
Genomic classifier studies
In a proof-of-principle study, published in 2017 in the Annals of the American Thoracic Society, authors described how they used machine learning to train an algorithm to distinguish UIP from non-UIP pattern in tissue obtained by transbronchial biopsy (TBB).
The top-performing algorithm distinguished UIP from non-UIP conditions in single TBB samples with specificity of 86% and sensitivity of 63%, according to investigators, who said at the time that independent validation would be needed before the genomic classifier could be applied in clinical settings.
In a prospective validation study, published in 2019 in The Lancet Respiratory Medicine, the genomic classifier identified UIP in TBB samples from 49 patients with a specificity of 88% and sensitivity of 70%.
Excluding patients with definite or probable UIP as shown on high-resolution computed tomography, results show that the classifier had a sensitivity of 76%, specificity of 88%, and positive predictive value of 81%.
“The performance of the test is good, even in that scenario,” Dr. Dilling said.
Real-world results
Dr. Dilling also highlighted a “real-world” study, published earlier in 2021, demonstrating that UIP pattern recognized by a genomic classifier had encouraging sensitivity and specificity when combined with high-resolution CT and clinical factors.
That study included 96 patients who had both diagnostic lung pathology and a transbronchial lung biopsy for molecular testing with the classifier.
The classifier had a sensitivity of 60.3% and a specificity of 92.1% for histology-proven UIP pattern, investigators said in their report, which appears in the American Journal of Respiratory and Critical Care Medicine.
Local radiologists identified UIP with a sensitivity of 34.0% and specificity of 96.9%. But adding genomic classifier testing to local radiology testing increased the diagnostic yield, investigators said, with a sensitivity of 79.2% and specificity of 90.6%.
“This might suggest that the implementation of this into a local [multidisciplinary discussion] with your local radiology expertise might really improve your recognition of UIP,” Dr. Dilling said.
Dr. Dilling reported disclosures related to Bellerophon, Boehringer Ingelheim, Genentech, Nitto Denko, and Lung Bioengineering.
Although genomic testing is useful when an interstitial lung disease diagnosis is uncertain, the testing results themselves aren’t sufficient to make the diagnosis, Daniel Dilling, MD, FCCP, said in a presentation at the annual meeting of the American College of Chest Physicians, which was held virtually.
The genomic classifier (Envisia, Veracyte) helps differentiate idiopathic pulmonary fibrosis (IPF) by detecting usual interstitial pneumonia (UIP), the hallmark pattern of this interstitial lung disease.
However, UIP is just one piece of the larger diagnostic puzzle, according to Dr. Dilling, professor of medicine in the interstitial lung disease program at Loyola University Medical Center in Maywood, Ill.
“Remember, it’s just a pattern, and not a diagnosis of IPF,” Dr. Dilling said in his presentation.
Genomic classifier results correlate well with both histologic and radiographic UIP pattern, studies show.
However, Dr. Dilling said the value of the genomic classifier is not in isolation.
“We don’t use this in a vacuum,” he said. “It increases our confidence and consensus, but it has to be incorporated into a multidisciplinary discussion group.”
Part of the diagnostic pathway
Dr. Dilling said the genomic classifier should be considered part of a diagnostic pathway in uncertain cases, particularly when the risk of surgical lung biopsy is high.
Current clinical practice guidelines recommend surgical lung biopsy for histopathologic diagnosis when clinical and radiologic findings are not definitive for IPF, the speaker said.
However, surgical lung biopsy carries some risk, and sometimes it can’t be done, he added.
In his presentation, Dr. Dilling cited a systematic review and meta-analysis of 23 studies looking at surgical lung biopsy for the diagnosis of interstitial lung diseases.
The postoperative mortality rate was 3.6% in that meta-analysis, published in 2015 in the Journal of Thoracic and Cardiovascular Surgery.
“The final decision regarding whether or not to perform a [surgical lung biopsy] must be based on the balance between benefits to establish a secure diagnosis and the potential risks,” authors wrote at the time.
Mortality risk is higher in immunocompromised and acutely ill patient populations, according to Dr. Dilling, who added that as many of 19% of patients will have complications from surgical lung biopsy.
Genomic classifier studies
In a proof-of-principle study, published in 2017 in the Annals of the American Thoracic Society, authors described how they used machine learning to train an algorithm to distinguish UIP from non-UIP pattern in tissue obtained by transbronchial biopsy (TBB).
The top-performing algorithm distinguished UIP from non-UIP conditions in single TBB samples with specificity of 86% and sensitivity of 63%, according to investigators, who said at the time that independent validation would be needed before the genomic classifier could be applied in clinical settings.
In a prospective validation study, published in 2019 in The Lancet Respiratory Medicine, the genomic classifier identified UIP in TBB samples from 49 patients with a specificity of 88% and sensitivity of 70%.
Excluding patients with definite or probable UIP as shown on high-resolution computed tomography, results show that the classifier had a sensitivity of 76%, specificity of 88%, and positive predictive value of 81%.
“The performance of the test is good, even in that scenario,” Dr. Dilling said.
Real-world results
Dr. Dilling also highlighted a “real-world” study, published earlier in 2021, demonstrating that UIP pattern recognized by a genomic classifier had encouraging sensitivity and specificity when combined with high-resolution CT and clinical factors.
That study included 96 patients who had both diagnostic lung pathology and a transbronchial lung biopsy for molecular testing with the classifier.
The classifier had a sensitivity of 60.3% and a specificity of 92.1% for histology-proven UIP pattern, investigators said in their report, which appears in the American Journal of Respiratory and Critical Care Medicine.
Local radiologists identified UIP with a sensitivity of 34.0% and specificity of 96.9%. But adding genomic classifier testing to local radiology testing increased the diagnostic yield, investigators said, with a sensitivity of 79.2% and specificity of 90.6%.
“This might suggest that the implementation of this into a local [multidisciplinary discussion] with your local radiology expertise might really improve your recognition of UIP,” Dr. Dilling said.
Dr. Dilling reported disclosures related to Bellerophon, Boehringer Ingelheim, Genentech, Nitto Denko, and Lung Bioengineering.
FROM CHEST 2021