Computational Analysis Model Applied to Spanish Gaucher Registry Data

Introduction

Type 1 Gaucher disease (GD)(OMIM # 230800), has a pan-ethnic distribution, in Spain its prevalence is about 1/100,000. Since more than twenty years the impact of therapies in the awareness of the disease is changing the characteristics and expectations of patients. The Spanish Gaucher disease Registry (SGDR) is working since 1993 and compiled demographic, clinical, genetic, analytical and imaging data about 360 type 1 GD Spanish patients. The application of new high computing capacity and powerful network analysis to analyze the registered data could provide a visualization tool and allows to extract knowledge from complex and very numerous relationships. The objective of this analysis is to discover useful ideas and new correlations to predict the risk of developing late complications and to extract knowledge of complex and very numerous relationships.

Patients & Methods

From 416 patients included in the SGDR we have selected 358 with more than 70% of data and follow-up. GD type 2 patients have been excluded. The variables included in the database at diagnosis: demographic data and the clinical, analytical and imaging information at diagnostic and during the treatment follow-up as well as comorbidities. With Kampal Data Solutions, a spin-off company of the University of Zaragoza dedicated to the development of computer applications and advanced data analytics, this company has experience in the homogenization of information and in the elaboration of classic and advanced statistics projects, as well as in the visualization of said information complex network techniques and the relationship between variables and model designs.Variables:Birthdate, age at diagnosis, gender, death date, severity category of disease (mild, moderate, severe), concomitant diseases, Parkinson disease in relatives, liver volume, spleen volume, spleen removal, bone disease, S-MRI, DEXA, chitotriosidase, CCL18/PARC, lyso Gb1, B12 vitamin level, iron concentration, ferritin, cholesterol, triglycerides, HDL-cholesterol, LDL-cholesterol, AST, ALT, GGT, acid phosphatase, bilirubin, hemoglobin concentration, Hsc, WBC count, platelets count, serum gammaglobulin fraction, IgG, IgA, IgM, GBA activity, GBAgenotype, CHIT1genotype, age to start therapy, type of therapy (ERT, SRT), new bone crisis or joint replacement, development of malignancies or Parkinson disease.

Results: The 358 subjects were mostly GD1 (340 vs 18 GD3), 18% were splenectomized and 39% have advanced bone disease and bone complications. Most of the patients have a complex heterozygous genotype (81% vs 19% homozygous). 47% of patients were diagnosed before 2,000 and 10% die before this study. Most of them are receiving ERT (54%). About comorbidities, 4% of patients developed MGUS or Parkinson disease and 6% malignant neoplasias. The main results have founded a significant correlation between skeletal complications and impaired and spleen removal (p=0.0005); this fact confirm previous reports. In this study a low IgA serum level shows a significant correlation with severe bone disease(p=0.0000), and with the incidence of new bone crisis during long term ERT. A IgG increase was related to the development of neoplasia. There were two more important factors that we have found; the age at diagnosis and the age to start therapy.

Comments:

Registries are key resources to help increasing timely and accurate diagnosis, improving patient's management, tailoring treatments, facilitating clinical trials, supporting healthcare planning and speeding up research. This is the very first attempt to establish a correlation network among different biomarkers and clinical characteristics in a national base cohort. As has been hypothesized seems that the impairment of immune system has a strong impact in long term complications, in our study the humoral immunity dysfunction pops up as an important factor. Despite of our short cohort, the quality of data is accurate and can reflect the own Spanish Gaucher disease characteristics. Currently we are working in the design of algorithms to help to predict patient outcomes.