Chronic myeloid leukemia (CML) is a model of targeted therapy for human malignancies. Over the past decade, a broad array of drugs designed to selectively inhibit protein tyrosine kinases (PTK) [i.e., tyrosine kinase inhibitors, (TKI)] have emerged as novel therapies for cancer patients. Hence, CML is an hematopoietic stem cell disorder in which a t(9;22) (q34;q11) reciprocal chromosomal translocation gives rise to Philadelphia chromosome (Ph) and generates the BCR-ABL1 fusion gene encoding a constitutively activated PTK. TKIs, such as imatinib by blocking BCR-ABL1 kinase activity, selectively eradicate CML cells and induce durable responses and prolong survival. CML patients treated with TKI are monitored by quantitative RT-PCR to detect leukemic BCR-ABL1 transcript performed from peripheral blood samples (1).

Since TKI treated CML patients have a near-normal life expectancy two important issues must be considered in the future:

1. the quality of life and ethical aspects of a lifetime treatment,
2. the budget impact for healthcare providers of treating patients during lifetime.

One of the best ways to consider these two points is to ask the question about stopping TKI in good responder patients. We first reported a pilot study where imatinib was withdrawn in 12 CML patients treated and maintained in complete molecular remission (CMR), defined by undetectable residual disease (with sensitivity of 4.5 log) on quantitative RT-PCR, for at least two years. Then, we demonstrated in a multicenter study entitled STIM trial that imatinib could be safely discontinued in patients with CMR for at least 2 years (2). All molecular relapsing patients were sensitive when imatinib was re-challenged (3). Around 40% of these patients remain in a prolonged treatment-free remission (TFR) after treatment cessation (4). Taking into account the cost of imatinib and the number of months without treatment in STIM trial, the savings in France were estimated to be 9 million €. However, since only 40 % of patients are in treatment free remission, a study, assessing the real budget impact of stopping TKI in the eligible population seems necessary as no published study has ever addressed this question in France.

Our aim is to assess the budget impact of discontinuing TKI treatment in patients with CML in deep molecular response since at least two years, compared with treatment during lifetime, from the French healthcare system point of view. This budget impact will be expressed as a "net benefit" and will be based on the difference between total costs incurred by this strategy and total costs avoided also.

One of the originality of our study is to raise the issue of treatment cessation in the context of a chronic disease from an economic point of view.

The other originality of this study is to use a decision model to perform this French budget impact analysis of TKI discontinuation, without setting up another trial. Besides the literature review and meta-analysis; the proposed probabilistic Markov model will use direct costs (including treatment costs and all health care related costs as well as costs related to relapse) extracted mainly from the French Health Insurance Databases.

Study period between 01/01/2005 and 12/31/2015 (cf. figure) including:

• Inclusion period: 01/01/2008 to 12/31/2014;
• Date of inclusion: date of the last TKI reimbursement before discontinuation which will be defined by the absence of any TKI reimbursement without the occurrence of death during at least 61 days following the 30-day coverage period of a TKI reimbursement;
• History: 3-year period;
• Follow-up: at least 1 year after inclusion or until death whatever occurs first.

Inclusion criteria for CML population

• Patients aged 18 years or over;
• Patients with LTD registration or hospitalization for CML (primary, associated and linked ICD-10 diagnosis code, i.e. C92.1 or C921) during the study period;
• Patients who discontinued their TKI treatment for the first time during the inclusion period;
• Patients treated with TKI during the inclusion period with a minimum of 3 year-period of TKI regular treatment before TKI discontinuation. A 3 year-period of TKI regular treatment will be defined on the presence of at least 10 TKI reimbursements per year during the 3 years preceding TKI discontinuation.

Exclusion criteria for CML population

• Patients who proceeded to allogeneic or autogenic hematopoietic stem-cell transplant (hospitalization ICD-10 code diagnosis Z94.80) in the 3 year-period prior to or in the month following the last TKI reimbursement identified before TKI discontinuation;
• Patients with HIV/AIDS (hospitalization ICD-10 code diagnosis B24) or chronic Hepatitis C or B (hospitalization ICD-10 code diagnosis B18) in the 3 year-period prior to or in the month following the last TKI reimbursement identified before TKI discontinuation;
• Recent (i.e. in the previous year) or ongoing pregnancy at TKI discontinuation date identified by an algorithm based on codes of hospitalization diagnoses and medical procedures.

Causes of TKI discontinuation will be further investigated in the cohort in order to conduct analyses only patients with a TKI discontinuation due to a complete molecular response.

A cohort of CML patients treated with TKI will be identified in the French nationwide claims and hospital database (SNIIRAM)