Fanconi Anemia: Examining Guidelines for Testing All Patients with Hand Anomalies Using a Machine Learning Approach.

Christoph Wallner,Gill Smith,Anthony Barabás,Jane Hurst,Björn Behr,Mohammad Abu Tareq Rony

doi:10.3390/children9010085

Abstract

Background: This study investigated the questionable necessity of genetic testing for Fanconi anemia in children with hand anomalies. The current UK guidelines suggest that every child with radial ray dysplasia or a thumb anomaly should undergo further cost intensive investigation for Fanconi anemia. In this study we reviewed the numbers of patients and referral patterns, as well as the financial and service provision implications UK guidelines provide. Methods: Over three years, every patient with thumb or radial ray anomaly referred to our service was tested for Fanconi Anemia. CART Analysis and machine learning techniques using Waikato Environment for Knowledge Analysis were applied to evaluate single clinical features predicting Fanconi anemia. Results: Youden Index and Predictive Summary Index (PSI) scores suggested no clinical significance of hand anomalies associated with Fanconi anemia. CART Analysis and attribute evaluation with Waikato Environment for Knowledge Analysis (WEKA) showed no single feature predictive for Fanconi anemia. Furthermore, none of the positive Fanconi anemia patients in this study had an isolated upper limb anomaly without presenting other features of Fanconi anemia. Conclusion: As a conclusion, this study does not support Fanconi anemia testing for isolated hand abnormalities in the absence of other features associated with this blood disease.

Highlights

Fanconi anemia (FA) is a very rare, inheritable, primarily autosomal recessive anemia that leads to bone marrow failure
As a conclusion, this study does not support Fanconi anemia testing for isolated hand abnormalities in the absence of other features associated with this blood disease
The clinical features triggering referral for PB-chromosome breakage test (CBT) testing were examined for each patient tested, and the features of those patients testing positive for FA were analyzed

Summary

Introduction

Fanconi anemia (FA) is a very rare, inheritable, primarily autosomal recessive anemia that leads to bone marrow failure. Data from 1865 case reports in the literature suggest that the most frequent characteristic birth defects in FA The UK FA guidelines state that 71% of FA patients have skeletal anomalies, which include radial ray dysplasia (RRD), but without any emphasis on specific anatomical areas [3]. US guidelines indicate that approximately 50% of FA children have skeletal anomalies, and that 70% of these are upper extremity anomalies, most commonly affecting the thumb and radial border of the forearm [2]. The current UK guidelines suggest that every child with radial ray dysplasia or a thumb anomaly should undergo further cost intensive investigation for Fanconi anemia. Methods: Over three years, every patient with thumb or radial ray anomaly referred to our service was tested for Fanconi Anemia. CART Analysis and machine learning techniques using Waikato Environment for Knowledge Analysis were applied to evaluate single clinical features predicting Fanconi anemia

Methods

Results

Discussion

Conclusion