Staging Disease with PROTEGE 4: Example of Multiple Myeloma

Staging Disease with PROTEGE 4: Example of Multiple Myeloma

Eric BRILLETa,[1], Valérie BERTAUD GOUNOTa, Olivier DECAUX b,
RégisDUVAUFERRIERa,JérémyLASBLEIZa

aUnitéInserm U936, IFR 140IFR 140, Faculté de Médecine,
University of Rennes 1, France

b Department of Internal Medicine CHU of Rennes, France

Abstract.The goal of this study is to create an applicative ontology of the Multiple myeloma. We have extracted a multiple myeloma sub-ontology from the NCI Thesaurus.Secondly,using Protégé 4 owl2, we have addedthe missing classes and quantitative data propertiesto the ontology to improve the multiple myelomarepresentation. The obtained ontology has been checked by a reasoner (Fact++) and tested on twenty real medical reports.The twenty patients were classified at the correct stages according to the report findings.

Keywords. NCI Thesaurus, OWL, Biomedical ontologies, Ontology, Myeloma.

1.Introduction

The goal of this study is to create an ontology of the myeloma with quantitative data properties to obtain an applicative ontology that will permit to stage patients.To obtain a multiple myeloma representation, we have automatically extracted from the NCIThesaurus [1]a sub-ontology of 281 class, 17 relations and 25 restrictions. The extractor retrieved all ancestors up to the top, all children down to the leaves, all target concepts connected by a restriction to PLASMA_CELL_NEOPLASM or its children.To improve the sub-ontology, we have used Protégé [2] a free open source ontology editor and knowledgebase Framework. The language we uses is owl2, because it can deal with quantitative data properties, not taken into account in NCIT owl1.

In 2003, the International Myeloma Working Group have established criteria for the diagnosis of plasma cell proliferative disorders [3]. They have distinguished monoclonal gammopathy of undetermined significance (MGUS), which is a premalignant disease and is defined by an intact immunoglobulin < 3 g/100 ml, < 10% bone marrow plasma cells and absence of end-organ damage. End-organ damage includes hypercalcemia, renal failure, anemia and bone lesions that are related to a plasma cell proliferative disorder and not explained by another unrelated disease or disorder. Symptomatic MM is differentiated from MGUS and Smoldering MM based on the presence or absence of end-organ damage attributable to the underlying plasma cell proliferative process.

Those quantitative concepts missing in the sub-ontology and needed to stage patients, have been added with U936- as prefix in the ontology. For examples:

• Disease stages: U936_Symptomatic_Myeloma ; U936_Durie/Salmon_Plus
• Biological examination: U936_IgA_serum_Measurement U936_Urinary_Lambda_Or_Kappa_Measurement
• Findings:U936Focal_Myeloma_Lesions_in_FDG-PET_scan

To translate definition from literal into formal, we have used new relations as U936_Has_Value and U936_Has_Unit. Those relations are essential to describe quantitative data properties.

2.Results

We have used a OWL reasoner, Fact++ that check consistency of the ontology in terms of formal definitions of classes. They can classify an instance when the instance contains all the necessary and sufficient conditions. 20 patients, with different variants or stages of myeloma, records or exit letters have been analyzed. Each patient has been added as a class in the ontology to allow the reasoner to work. All patients have been reaffected into myeloma different stages.

Figure 1 Patient classified after the run of Fact++.

3.Conclusion

The ontology we built is one of the first to deal with quantitative data properties. It permits to classify “real life patients”, with for example biological findings. Physicians, at patient bedside, can use this tool to stage disease and take therapeutic decisions.

References

[1]Fragoso G, de Coronado S, Haber M, Hartel F, Wright L. Overview and utilization of the NCI thesaurus, Comp Func Genomics 5(8) (2004), 648–54.

[2] access January 2011

[3]The international Myeloma Working group. Criteriafor the classification of monoclonal gammopathies, multiple myeloma and related disorders: a report of the international Myeloma Working group. Br J Haematology 2003; 121: 749-57

[1] Corresponding Author