Skip to main content

Innovative risk assessment for individualizing treatment in chronic lymphocytic leukemia

Periodic Reporting for period 1 - CLLassify (Innovative risk assessment for individualizing treatment in chronic lymphocytic leukemia)

Reporting period: 2016-04-01 to 2018-03-31

Personalized medicine (PM) is the customization of healthcare, with medical decisions and practices tailored to the individual characteristics of a given patient and one of the great challenges for the 21st century. Chronic lymphocytic leukemia (CLL) is a malignancy of B lymphocytes and the most common adult leukemia in the West, mainly affecting the aged population and characterized by remarkable clinicobiological heterogeneity. The great majority (~85%) of patients with CLL do not require treatment upon diagnosis, however may later experience disease evolution and develop a need for therapy. Presently, it is not possible to predict with accuracy when this may happen, if at all. Hence, it is not paradoxical that CLL patient management has largely adopted a ‘watch and wait’ approach. For those patients developing a need for therapy, several therapeutic options are available, with varying efficacy and cost. On these grounds, essential pre-requisites for realizing PM in CLL are two-fold:
(i) robust prognosis at the time of diagnosis,
(ii) individualized therapeutic approach for each patient based on personalized prediction.

The potential impact of addressing these problems is multidimensional. (a) The prevalence of CLL in Europe is on the rise not least due to the aging of the population. Thus, inoculating novel statistical concepts for addressing both prognosis and personalized prediction will have significant impact in the context of CLL management and European health, in general. (b) From the healthcare professional’s perspective, robust prognostication may assist in the design of the follow-up strategy and the selection of the optimal treatment, increasing clinical efficacy while significantly reducing unwanted toxicities and costs, thus benefiting society at large. (c) Prognostic information can also assist the patient and his/her family in seeing the big picture and making the best decisions, hence improving their quality of life.

The overall objectives of CLLassify were:
(i) To improve prognostic assessment of CLL patients at the time of diagnosis. This may guide the physician to define the risk group the patient belongs to and, consequently, efficiently manage the follow-up.
(ii) To match individual patient characteristics with particular outcomes at any stage during the disease course, thus enabling personalized management of CLL. This relates to the ability to efficiently predict the pattern of disease course of the individual patient, including his/her response to different treatment modalities. Efficient prediction could reasonably be anticipated to empower clinical decision-making with various benefits related to prevention, efficient disease management, economy, etc.
The remarkable clinical heterogeneity observed in CLL complicates therapeutic decisions while standard clinical staging systems cannot discriminate patients with markedly different clinical course and outcome. This underscores the need for alternative risk stratification approaches while also highlighting the importance of factors which could be used for optimizing both the follow up policy and the therapeutic approach for each individual patient.
To this end, various factors have been evaluated, including both host-related (e.g. age, gender, physical status) and also tumor-related i.e. pertaining to the biology of the tumor cells. Until now, the major focus has been on genomic aberrations, especially within the TP53 gene; and, immunogenetic features, in particular the somatic hypermutation (SHM) status of the IGHV genes. These factors assist in both determining the overall prognosis at diagnosis and predicting the response to particular treatments, thus empowering therapy decisions.
Within this frame, CLLassify has three major research pillars. The first refers to refined stratification of early-stage CLL patients in risk groups, based on established and emerging prognostic factors measured at diagnosis. Advanced clustering techniques have been applied and combined, resulting in two different prognostic indices, separately developed based on the two categories of the SHM status. The results support that compartmentalizing CLL might improve prognostication in early-stage CLL.
The second is related to evaluating the impact inflicted by these factors on how the risk for CLL progression evolved over time, accounting in parallel for the SHM status. The patterns of risk evolution exhibited significant differences: this supports the notion that the SHM status is more than a simple prognostic/predictive factor, and that segregation of CLL patients based on SHM might aid to detect important time effects on risk evolution within specific genomic CLL subgroups.
The third is dedicated to personalized prediction i.e. the ability to predict at a specific time point the future need for treatment of an individual patient but also his/her response. Within this pillar, patients’ follow-up data was recorded in a time series format; then, several statistical predictive models were applied, including standard and novel approaches in order to evaluate the ability to predict disease evolution. Current results provide encouraging evidence indicating that certain crucial characteristics of future behavior could be predicted: this is extremely important, since it could influence patient management. Further investigation is underway to realize the extent of efficient prediction.
The main results achieved so far have been submitted for publication to high-impact scientific journals and already presented in prestigious international conferences, indicatively, the 58th and the 59th ASH Annual Meeting & Exposition, ASH being the premium world event in hematology; the abstract in the 58th was awarded the ASH Abstract Achievement Award for high-scoring abstract.
All the project’s updates have been systematically published in the social media of CLLassify (Facebook, website and Twitter).
Most prognostic models for CLL apply the same prognostic factor indiscriminately on all patients, overlooking its biological complexity and clinical heterogeneity. Since prognostic factors may have different prognostic impact within distinct subgroups of patients, obviously this policy is suboptimal. In addition, concerns are raised by the usage of factors that may differ over time.
CLLassify progressed beyond the state of the art in that a compartmentalized approach was followed to account for the immunogenetic features of the patients, revealing pronounced differences between major patient categories regarding the prognostic impact of the factors considered: from a broader perspective, such compartmentalized approaches might prove relevant to other B cell malignancies as well. Regarding the time effect, the risk for CLL progression was for the first time investigated over time within specific genomic subgroups and accounting for the SHM status. The results support the usage of over time risk analysis, since latent information might be revealed with potential application in clinical practice. Finally, the novel idea of utilizing individual information from the patient follow-up towards personalized prediction has been applied for the first time and shown great potential. Eventually, efficient prediction, the main focus of CLLassify, could have significant impact on CLL patient management with obvious benefits for the patients and their carers, the healthcare system and society at large.
CLLassify logo