Baliakas P, Mattsson M, Stamatopoulos K, Rosenquist R
J. Intern. Med. 279 (4) 347-357 [2016-04-00; online 2015-12-28]
The remarkable clinical heterogeneity in chronic lymphocytic leukaemia (CLL) has highlighted the need for prognostic and predictive algorithms that can be employed in clinical practice to assist patient management and therapy decisions. Over the last 20 years, this research field has been rewarding and many novel prognostic factors have been identified, especially at the molecular genetic level. Whilst detection of recurrent cytogenetic aberrations and determination of the immunoglobulin heavy variable gene somatic hypermutation status have an established role in outcome prediction, next-generation sequencing has recently revealed novel mutated genes with clinical relevance (e.g. NOTCH1, SF3B1 and BIRC3). Efforts have been made to combine variables into prognostic indices; however, none has been universally adopted. Although a unifying model for all groups of patients and in all situations is appealing, this may prove difficult to attain. Alternatively, focused efforts on patient subgroups in the same clinical context and at certain clinically relevant 'decision points', that is at diagnosis and at initiation of first-line or subsequent treatments, may provide a more accurate approach. In this review, we discuss the advantages and disadvantages as well as the clinical applicability of three recently proposed prognostic models, the MD Anderson nomogram, the integrated cytogenetic and mutational model and the CLL-international prognostic index. We also consider future directions taking into account novel aspects of the disease, such as the tumour microenvironment and the dynamics of (sub)clonal evolution. These aspects are particularly relevant in view of the increasing number of new targeted therapies that have recently emerged.