Development and validation of the FIP-Score for the screening of FIP1L1::PDGFRA-associated hypereosinophilic syndrome.

Abstract

Background: FIP1L1-PDFGRA (F/P)-associated Hypereosinophilic Syndrome (HES) is a rare condition. F/P fusion gene testing is one of the first-line investigations in patients with unexplained eosinophilia and yields poor diagnostic performance.

Objective: To build and validate the FIP-Score: a set of weighted criteria warranting testing for the F/P fusion gene.

Methods: We merged data from 151 patients with F/P-associated HES and 320 patients with either F/P-negative HES (n=279) or hypereosinophilia of undetermined significance (n=41). Training and validation cohorts (comprising respectively 90% and 10% of all patients) were randomly dichotomized. Variables with a p-value <0.20 in univariate analysis were included in the multivariable forward-backward logistic regression model to assess their independent contribution to testing positive for the F/P fusion gene. Beta coefficients from multivariable logistic regression were used to assign points for the construction of the score.

Results: Age under 66 years, male sex, splenomegaly, lymphomatoid papulosis, absence of gastrointestinal involvement, high serum vitamin B12, high serum tryptase and normal serum immunoglobulin E levels were the eight variables retained in the model. The best cutoff value was greater than 48. The model yielded a sensitivity, specificity, positive predictive value, negative predictive value and area under the curve respectively of 88.3%, 93.7%, 87.1%, 94.4%, and 0.962 in the training dataset and of 85.7%, 97.0%, 85.7% and 97.0%, 0.986 in the validation dataset.

Conclusion: The FIP-score highlights the need for closely selecting patients with hypereosinophilia for whom F/P fusion gene testing should be performed, resulting in medical time reduction and substantial cost-savings.