Idiopathic pulmonary fibrosis (IPF) is the most important of the restrictive lung diseases. Itis rare, irreversible, and fatal, with a median survival of 3-5 years from diagnosis. The progression of IPF is unpredictable and highly variable, with some patients experiencing rapid deterioration and others slow decline. In fact, the clinical natural history of IPF is often characterized by acute exacerbations (AEs) involving impaired respiratory function. This leads to a decline in quality of life and an increased social and health burden for patients.Given these implications, the ability to predict an exacerbation event early would help improve patients' quality of life and reduce health care costs. Therefore, the development of this doctoral project involved the identification of possible biomarker fingerprints that would predict exacerbation events at the time of diagnosis. To this end, 80 patients with IPF in the stable phase of the disease (a prerequisite for enrollment) were recruited and various parameters including age, sex, spirometric values, and blood concentrations of some markers of oxidative stress and inflammation were recorded. Patients were followed for one year and finally classified into two groups, exacerbated and nonexacerbated, according to the frequency of exacerbations. The collected data were analyzed with a univariate logistic regression model, and a ROC curve was generated for each variable. Of the 32 variables considered, only two showed a significant area under receiver operating characteristics (AUROC), even though showing weak diagnostic accuracy in relation to exacerbation events. To improve predictive ability, a multivariate logistic regression approach was used to develop a model by combining the variables. A composite model of six predictors was obtained, which included in addition to gender, parameters of disease progression/worsening (Forced Vital Capacity or FVC and Lung Diffusing Capacity for carbon monoxide or DLCO), biomarkers of inflammation (Systemic Immune-inflammation Index or SII) and oxidative stress (PSH and TBARS). AUROC indicated that the model had a good AUC value (0.841) with a sensitivity of 88% and specificity of 70%, respectively. The Hosmer-Lemeshow calibration test produced non-significant P-values (χ2=1). The P-values (χ2= 9.0308, P= 0.3397) indicate a good calibration. In conclusion, this newly developed combined index could be useful for predicting recurrent exacerbations in IPF patients at the time of diagnosis and implementing plans to limit disease progression.
A new logistic regression derived combined index for early predictionof recurrent exacerbation in IPFpatients / Mellino, Sabrina. - (2023 Jul 10).
A new logistic regression derived combined index for early predictionof recurrent exacerbation in IPFpatients
_PhD THESIS Sabrina Mellino.pdf
Descrizione: A new logistic regression derived combined index for early prediction of recurrent exacerbation in IPF patients
Tipologia: Tesi di dottorato
Dimensione 2.18 MB
Formato Adobe PDF
|Visualizza/Apri Richiedi una copia