EUVIMEDEUVIMED entdecken
Open Access

Predicting Primary Graft Dysfunction in Systemic Sclerosis Lung Transplantation Using Machine‐Learning and CT Features

Jatin Singh iD, Xin Meng, Joseph K Leader, John Ryan, Lucas Pu, Rachel Deitz iD, Ernest G Chan iD, Norihisa Shigemura, Chadi A Hage, Pablo G Sanchez, Jiantao Pu

Clinical Transplantation · 2025-07 · 39 (7)

DOI: 10.1111/ctr.70217 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Clinical Transplantation, online erschienen am 24.06.2025. Verfasst von Jatin Singh und 10 weiteren Autor:innen. Beteiligte Einrichtung: Department of Radiology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA. Hauptthemen laut MEDLINE: Lung Transplantation, Tomography, X-Ray Computed, Scleroderma, Systemic, Machine Learning. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 33 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: UPMC Hillman Developmental Pilot Program, NCI NIH HHS, (NIH).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Primary graft dysfunction (PGD) is a significant barrier to survival in lung transplant (LTx) recipients. PGD in patients with systemic sclerosis (SSc) remains especially underrepresented in research.

Methods

We investigated 92 SSc recipients (mean age 51 years ± 10) who underwent bilateral LTx between 2007 and 2020. PGD was defined as grade 3 PGD at 72 h post‐LTx. A comprehensive set of CT image features was automatically computed from recipient chest CT scans using deep learning algorithms. Volumetric analysis of recipients’ lungs and chest cavity was used to estimate lung‐size matching. Four machine learning (ML) algorithms were developed to predict PGD, including multivariate logistic regression, support vector machine (SVM), random forest classifier (RFC), and multilayer perceptron (MLP).

Results

PGD was significantly associated with BMI >30 kg/m 2 ( p = 0.009), African American race ( p = 0.011), lower Preop FEV1 ( p = 0.002) and FVC ( p = 0.004), longer waitlist time ( p = 0.014), higher lung allocation score (LAS) ( p = 0.028), and interstitial lung disease ( p = 0.050). From CT analysis, PGD was significantly associated with decreased lung volume ( p < 0.001), increased heart‐chest cavity volume ratio ( p < 0.001), epicardial ( p = 0.033) and total heart ( p = 0.049) adipose tissue, and five cardiopulmonary features ( p < 0.050). Oversized donor allografts estimated using CT analysis were significantly associated with PGD ( p < 0.050). The MLP model achieved a maximum AUROC of 0.85 (95% CI: 0.81–0.88) in predicting PGD with four features: Preop FEV1, heart‐chest cavity volume ratio, waitlist time, and donor to recipient chest cavity volume ratio .

Conclusion

CT‐derived features are significantly associated with PGD, and models incorporating these features can predict PGD in SSc recipients.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Lung TransplantationTomography, X-Ray ComputedScleroderma, SystemicMachine LearningPrimary Graft DysfunctionHumansFemaleMaleMiddle AgedPrognosisFollow-Up StudiesRisk FactorsAdultRetrospective StudiesGraft Survival

Stichwörter der Autor:innen

acute rejectionlung transplantationmachine learningprimary graft dysfunctionsclerodermasystemic sclerosis

Autor:innen und Institutionen

  • Department of Radiology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.
  • Division of Lung Transplant and Lung Failure, Department of Cardiothoracic Surgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA.
  • Department of Surgery, Section of Thoracic Surgery, University of Chicago, Chicago, Illinois, USA.
  • Division of Pulmonary Medicine and Critical Care, Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA.
  • Department of Bioengineering, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.
  • Department of Ophthalmology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.

ORCID-Profile: Jatin Singh, Rachel Deitz, Ernest G Chan

Förderung

  • UPMC Hillman Developmental Pilot Program
  • NCI NIH HHS
  • (NIH)
  • National Institutes of Health (NIH)
  • NIH HHS
  • NHLBI NIH HHS

Zitieren

APA

Singh, J., Meng, X., Leader, J. K., Ryan, J., Pu, L., Deitz, R., Chan, E. G., Shigemura, N., Hage, C. A., Sanchez, P. G., & Pu, J. (2025). Predicting Primary Graft Dysfunction in Systemic Sclerosis Lung Transplantation Using Machine‐Learning and CT Features. Clinical Transplantation, 39(7). https://doi.org/10.1111/ctr.70217

Vancouver

Singh J, Meng X, Leader JK, Ryan J, Pu L, Deitz R, et al. Predicting Primary Graft Dysfunction in Systemic Sclerosis Lung Transplantation Using Machine‐Learning and CT Features. Clinical Transplantation. 2025;39(7). doi:10.1111/ctr.70217

BibTeX
@article{Singh2025,
  author = {Singh, Jatin and Meng, Xin and Leader, Joseph K and Ryan, John and Pu, Lucas and Deitz, Rachel and Chan, Ernest G and Shigemura, Norihisa and Hage, Chadi A and Sanchez, Pablo G and Pu, Jiantao},
  title = {Predicting Primary Graft Dysfunction in Systemic Sclerosis Lung Transplantation Using Machine‐Learning and CT Features},
  journal = {Clinical Transplantation},
  year = {2025},
  volume = {39},
  number = {7},
  doi = {10.1111/ctr.70217}
}
Kostenlos starten →