Genomic data in prognostic models—what is lost in translation? The case of deletion 17p and mutant TP53 in chronic lymphocytic leukaemia Benjamin Chin-Yee 1 , Bekim Sadikovic 2 and Ian H. Chin-Yee 2,3 1 Department of Medicine, University of Toronto, Toronto, ON, Canada, 2 Department of Pathology and Laboratory Medicine, Schulich School of Medicine & Dentistry, Western University, London, ON, Canada and 3 Department of Medicine, Division of Hematology, Western University, London, ON, Canada Summary Genomic technologies are revolutionizing the practice of haematology-oncology, leading to improved disease detec- tion, more accurate prognostication and targeted treatment decisions. These advances, however, have also introduced new clinical challenges, which include problems of prognostic underdetermination and its attendant risks of over- and undertreatment. Genomic data is generated from different technologies, from cytogenetics to next-generation sequenc- ing, which are often interpreted interchangeably and in a binary fashion—as the presence or absence of a given chro- mosomal deletion or mutation—an oversimplification which may lead to mistaken prognosis. We discuss the clinical use of one such prognostic marker, represented by sequence and copy number alterations in TP53, located on chromosome 17p. Mutations in TP53 are strongly linked to poor progno- sis in a variety of haematological malignancies, including chronic lymphocytic leukaemia (CLL). We review studies in CLL which utilize the 17p deletion or TP53 mutations for prognostic stratification with specific focus on the technolo- gies used for detection, the thresholds established for clinical significance, and the clinical contexts in which these alter- ations are identified. The case of CLL illustrates issues arising from simplistic, binary interpretation of genetic testing and highlights the need to apply a critical lens when incorporat- ing genomics into prognostic models. Keywords: chronic lymphocytic leukaemia, genomics, prog- nostic models, cytogenetics, next-generation sequencing. Genomic technologies are revolutionizing the practice of haematology-oncology. From improved disease detection and prognostication, to mutation-targeted therapies with companion diagnostics, genomic technologies are catalysing rapid changes in how we care for patients with haematolog- ical malignancies. These advances, however, have also intro- duced a new set of challenges in the clinical setting, which include problems of prognostic underdetermination and its attendant risks of over and undertreatment. Prognostic risk scores incorporating a range of clinical and laboratory vari- ables have become ubiquitous in haematology-oncology, and are routinely used to inform treatment decisions. Genomic data increasingly dominate prognostic risk scores and shape treatment algorithms, adding new layers of com- plexity to clinical management. Although these data have been useful—for example, by allowing for the identification of ‘driver’ mutations and potential treatment targets—inter- pretation of genetic testing in the clinic can be complex and context-specific. For haematological malignancies, genomic data is gener- ated from a variety of technologies from traditional kary- otyping and interphase fluorescence in situ hybridisation (FISH), to next-generation sequencing (NGS) (Table I). In addition to their variable operating characteristics, informa- tion produced by these distinct technologies differs in clinical significance. Despite this, in the clinical setting, genomic data produced by different methodologies are often interpreted interchangeably and in a binary fashion—for example, as the presence or absence of a given chromosomal deletion or mutation—an oversimplification that has the potential to lead to mistaken prognosis and mismanagement. In this article, we discuss the clinical use of one such prognostic marker, represented by sequence and copy num- ber alterations in TP53 located at chromosome 17p. Muta- tions in TP53 are the most common genomic alterations in cancer and are strongly linked to poor prognosis in many cancer subtypes including haematological malignancies such as chronic lymphocytic leukaemia (CLL) and multiple mye- loma. We review the literature on the use of 17p deletion or Correspondence: Dr. Ian Chin-Yee, Department of Pathology and Laboratory Medicine, Schulich School of Medicine & Dentistry, Western University, Victoria Hospital, 800 Commissioners Road East, London, ON, Canada N6A 5W9. E-mail: ian.chinyee@lhsc.on.ca annotation First published online 5 March 2019 doi: 10.1111/bjh.15827 ª 2019 British Society for Haematology and John Wiley & Sons Ltd British Journal of Haematology, 2020, 188, 652–660