Page 49 - ECBIP 2021_Program

MODERATED e-POSTER SESSION 02:
Diseases of the pleura
PP14
Background: Pleuroscopy with pleural biopsy has high sensitivity for malignant pleural effusion (MPE). Because MPEs tend to recur, facilitating concurrent diagnosis and treatment of MPE during pleuroscopy is desired. However, proceeding directly to treatment at the time of diagnostic pleuroscopy requires confidence in the on-site diagnosis.The study’s primary objective was to create a predictive model to estimate the probability of MPE during diagnostic pleuroscopy.
Research Question: How well can MPE be diagnosed using a predictive model during diagnostic pleuroscopy?
Study Design and Methods: We performed a prospective observational multicenter cohort study of consecutive patients undergoing diagnostic pleuroscopy. We used a logistic regression model to evaluate the probability of final pathologic malignancy with relation to visual assessment, rapid on-site evaluation (ROSE) of touch preparation (low, high, and indeterminate malignancy probability), and presence of pleural nodules/masses on computed tomography (CT) (no or yes). To assess the model’s validity, a bootstrapped training/testing approach was utilized. The receiver operating characteristic area under the curve (AUC) was estimated.
Results: Of the 201 patients in the study, 103 had MPE. The logistic regression showed that compared to low probability of malignancy on visual assessment, high probability had higher odds of being associated with malignancy (OR=34.68, 95% CI=9.17-131.14, p<0.001).The logistic regression showed that compared to low probability of malignancy on ROSE of touch preparation, high probability had higher odds of being associated with malignancy (OR=11.63, 95% CI=3.85-35.16, p<0.001). Pleural nodules/masses on CT had higher odds of being associated with malignancy (OR=6.61, 95% CI=1.97- 22.1, p=0.002). A multivariable logistic regression model of final pathologic status with relation to visual assessment, ROSE of touch preparation, and presence of pleural nodules/masses on CT had an AUC of 0.94 (95% CI=0.91, 0.97).
Interpretation: A prediction model using visual assessment,ROSE of touch preparation,and CT scan findings demonstrated good predictive power for malignancy. Using this approach may facilitate concurrent diagnosis and treatment of MPE during pleuroscopy.
PREDICTING MALIGNANT PLEURAL EFFUSION DURING DIAGNOSTIC PLEUROSCOPY, A PROSPECTIVE MULTICENTER STUDY
Horiana Grosu1, Ryan Kern2, Fabien Maldonado3, Roberto Casal1, Clark R. Andersen4, Frangiskos Frangopoulos5, Liang Li4, Georgie Eapen1, David Ost1, Carlos Jimenez1, Bruce Sabath1, Erik Vakil1, Audra Schwalk1, Matheu Marcoux1, Ala Eddin Sagar1, Faria Nasim1, Julie Lin1, Moiz Salahudin1, Hasan Muhammad Arain1, Laila Noor1, Diana Montanez1, Michalis Michael6, John Stewart7, Ilias Porfyridis5
1 Department of Pulmonary Medicine,The University of Texas MD Anderson Cancer Center, Houston, USA 2 Department of Medicine, Division of Pulmonary and Critical Care Medicine, Mayo Clinic, Rochester, USA 3 Department of Thoracic Surgery, Vanderbilt University Medical Center, Nashville, USA
4 Department of Biostatistics,The University of Texas MD Anderson Cancer Center, Houston, USA
5 Pulmonary Department, Nicosia General Hopsital, Nicosia, Cyprus
6 Department of Cytopathology, Nicosia General Hospital, Nicosia, Cyprus
7 Department of Cytopathology,The University of Texas MD Anderson Cancer Center, Houston, USA
Figure 1: A. Logistic regression–adjusted probability of final pathologic malignancy among levels of visual assessment. B. Logistic regression–adjusted probability of final pathologic malignancy among levels of touch preparation assessment. C. Logistic regression–adjusted probability of final pathologic malignancy among levels of presence of pleural nodules/ masses on computed tomography. Catseye plots illustrate the normal distributions of
the model-adjusted means, with shaded +/- standard error intervals, transformed to the probability scale.The distributions of the model-adjusted means have been transformed from the logit scale to the probability scale; distributions near 0% or 100% have been distorted accordingly The horizontal lines in the cat’s eye plots indicate 50% probabilities, and standard errors are shaded.
Figure 2: Receiver operating characteristic area under the curve (AUC) from the logistic regression model of final pathologic status
with relation to visual assessment, touch preparation assessment, and presence of pleural nodules/masses on computed tomography
ECBIP 2021 49
6th European Congress
for Bronchology and Interventional Pulmonology
OCTOBER ECBIP 15 - 17
ATHENS - GREECE 2021
Book of Abstracts