Researchers Find HRCT Risk Prediction Model Helpful in Predicting RA-ILD Prognosis

August 8, 2020
Jaime Rosenberg
Jaime Rosenberg

A risk prediction model based on high-resolution computed tomography (HRCT) variables outperformed other models when predicting mortality among patients with rheumatoid arthritis-associated interstitial lung disease (RA-ILD).

New study findings indicate that a risk prediction model based on high-resolution computed tomography (HRCT) variables may be helpful in predicting the prognosis for patients with rheumatoid arthritis-associated interstitial lung disease (RA-ILD).1

The study of over 100 patients with RA-ILD demonstrated that 3 imaging parameters—fibrosis score (>20%), a usual interstitial pneumonia (UIP) pattern, and emphysema—were more effective than physiological parameters at predicting the prognosis of these patients. In multivariate analysis, these 3 characteristics, as well as old age (>60 years), were independent predictors of 5-year mortality.

According to the researchers, this risk prediction model outperformed other prediction models, including the Gender-Age-Physiology (GAP) model, a validated risk model used to predict the prognosis of idiopathic pulmonary fibrosis, which some studies have suggested could be appropriate for predicting the prognosis of patients with RA-ILD.

The study results showed that the GAP model may not be appropriate for predicting the prognosis of RA-ILD, likely due to the limited role of lung function and different demographic features in the group, according to the researchers. Notably, the GAP model did not account for principal prognostic factors for RA-ILD, such as a UIP pattern and emphysema on HRCT.

“Can we conclude that risk stratification for RA-ILD patients is possible with only age and high-resolution computed tomography (HRCT) data?” asked the authors of an accompanying editorial of the study.2 “The data from Kim et al’s study would suggest this is possible, which does have potential advantages in circumstances where pulmonary function tests are not available. However, it is important to consider whether the detailed HRCT scoring used here would be practicable in a general clinical setting.”

The study results do have positive implications, because a reliable risk prediction model for RA-ILD has eluded researchers as a result of the variable prognosis that’s difficult to predict.

The prediction model was developed using retrospective data of the 153 patients with RA-ILD whose disease was diagnosed between 1995 and 2015 (derivation cohort), as well as 149 patients whose disease was diagnosed between 2000 and 2017 (validation cohort).

For each of the 4 variables identified as predictors of prognosis, the researchers assigned points ranging from 0-2 and found that as total points increased, the survival rate decreased for the patients in the derivation cohort. The researchers then divided patients into 3 groups based on their total points: stage I (0 point), stage II (1-3 points), and stage III (4-5 points). The survival rates were significantly different at each stage.

Compared with the derivation cohort, patients in the validation cohort were older and had higher lung function, lower prevalence of emphysema, and higher fibrosis scores. When the researchers used the same risk prediction model on these patients, they found that the risk prediction was still useful for predicting mortality.

“To improve the applicability of the risk prediction model in patients with indeterminate fibrosis score (ranging from 10% to 30%), we investigated the substitution parameter for fibrosis score,” wrote the researchers. “In ROC analysis, the optimal cut-off level of [forced vital capacity (FVC)] for predicting 20% of the fibrosis score was 72% predicted (C-index = 0.775; P < .001), and as shown by quadratic function, 70% predicted of the FVC threshold corresponded to 20% of the fibrosis score threshold.”

The researchers found that when FVC (<70%) was substituted for fibrosis score (<20%) among patients in the derivation cohort who had an indeterminant fibrosis score, the survival rates were still significantly different at each stage.

Reference

1. Kim H, Lee J, Lee E, et al. Risk prediction model in rheumatoid arthritis-associated interstitial lung disease. Respirology. Published online May 22, 2020. doi:10.1111/resp.13848.

2. Parker M, Corte T. First risk, next reward? a new clinic-radiological risk model predicts mortalirt in rheumatoid arthritis-associated interstitial lung disease. Respirology. Published online July 30, 2020. doi: 10.1111/resp.13924.