UCLA researchers in the Department of Medicine have developed a logistic regression model using DNA methylation biomarkers that enables early, noninvasive prediction of chronic lung allograft dysfunction up to two years before clinical disease onset.
BACKGROUND: Chronic lung allograft dysfunction (CLAD) is the leading cause of long-term graft failure and mortality following lung transplantation. Current surveillance methods rely heavily on pulmonary function testing and histopathology. However, these methods typically detect disease only after irreversible damage has occurred. Thus, there is a critical unmet need for minimally invasive, cost-effective biomarkers capable of identifying patients at high risk for CLAD well before clinical onset of disease, enabling earlier intervention and improved patient outcomes.
INNOVATION: UCLA researchers have leveraged targeted bisulfite sequencing (TBS-seq) to profile DNA methylation patterns from bronchoalveolar lavage (BAL) cell pellets collected from lung transplant recipients. TBS-seq combines the digital resolution of bisulfite sequencing with the affordability and scalability of methylation arrays, while allowing customizable targeting of clinically relevant CpG loci. Using data from a longitudinal lung transplant cohort, researchers performed genome-wide association testing to identify CpG sites associated with future CLAD risk while accounting for covariates including age, sex, infection status, acute injury, and cell type composition. These loci were incorporated into a penalized logistic regression model that accurately predicts CLAD development 1-2 years prior to clinical diagnosis, achieving excellent predictive performance (AUC = 0.97). After correcting the methylation levels for covariates such as age, sex and medical center, the prediction still reaches AUC of 0.84.
POTENTIAL APPLICATIONS:
- Prediction of CLAD following lung transplant
- Can be used with transplant immunomodulatory therapies
- Extension to other transplant types or chronic inflammatory lung diseases
ADVANTAGES:
- Predicts CLAD risk 1-2 years prior to clinical onset
- High predictive accuracy (AUC = 0.97)
- Lower cost than traditional sequencing-based methylation assays
- Customizable, targeted probe design
- Compatible with routinely collected BAL samples
- Scalable & can be used for longitudinal monitoring
DEVELOPMENT-TO-DATE: Researchers have used targeted bisulfite sequencing of a longitudinal cohort of lung transplant recipients to develop a penalized logistic regression model that predicts development of CLAD risk 1-2 years prior to clinical evidence of disease with an AUC = 0.97.
Related Papers (from the inventors only): Pickering H, Arakawa-Hoyt J, Llamas M, Ishiyama K, Sun Y, Parmar R, Sen S, Bunnapradist S, Hays SR, Singer JP, Schaenman JM, Lanier LL, Reed EF, Calabrese DR, Greenland JR; CMV Systems Immunobiology Group. Cytomegalovirus-associated CD57 + KLRG1 + CD8 + TEMRA T cells are associated with reduced risk of CMV viremia in kidney transplantation and chronic allograft dysfunction in lung transplantation. Hum Immunol. 2025 May;86(3):111285. doi: 10.1016/j.humimm.2025.111285. Epub 2025 Mar 21. PMID: 40120236.
TDG Keywords:
Diagnostic Markers > Targets and Assays
Diagnostic Markers > Immunology
Software & Algorithms > Artificial Intelligence & Machine Learning
Software & Algorithms > Bioinformatics
Software & Algorithms > Statistical Models
Therapeutics > Respiratory and Pulmonary
Keywords: targeted bisulfite sequencing, DNA methylation profiling, epigenetic biomarkers, chronic lung allograft dysfunction, lung transplant, bronchoalveolar lavage, machine learning, logistic regression, disease prediction, transplant rejection, clinical risk stratification, diagnostics