Page 98 - 2019_08-Haematologica-web

C. Fava et al.
161) overall estimated TFR was 62% (95%CI: 56;68). At 12 months, TFR was 68% (95%CI: 62;74) for ima- tinib, 73% (95%CI: 64;83) for second generation TKI. Overall median time to restart treatment was six months (IQR 4;11). No progressions occurred. Although our study has the limitation of a retrospective study, our experience within the Italian population confirms that discontinuation of imatinib and second generation TKI is feasible and safe in clinical practice.
Introduction
Chronic myeloid leukemia (CML) patients have reached a near-normal life expectancy thanks to tyrosine kinase inhibitors (TKI).1,2 These drugs, however, can cause several persistent low-grade side effects that affect quality of life, and can be associated in the long term with severe toxici- ties.3 For this lifelong disease, tolerance and adherence to treatment are an issue. Furthermore, thanks to the success of the therapies, patients grow older and accumulate comorbidities that require concomitant treatments that can possibly interfer with TKI. Younger patients have other problems because living with TKI interferes with family planning, availability for the job market, life insur- ance, and so on.4 Besides, as more and more patients are living with their disease, high treatment costs are becom- ing an important issue.5
Over recent years, several papers have reported on treat- ment discontinuation in CML patients in persistent deep molecular response (DMR).6-25 The majority of these stud- ies have reported on patients who had achieved a DMR with imatinib. There are fewer data reported on the dis- continuation of second generation TKI, and with a shorter follow up. The definitions of DMR and the criteria for treatment discontinuation, for molecular relapse, and for treatment resumption, varied among these studies. Therefore, the reported treatment-free remission (TFR) rate ranged widely (between 30% and 70%), with the first reports mostly showing a TFR rate of approximately 40%; more recent reports, which adopted less stringent criteria for treatment discontinuation and therapy resumption, showed a TFR rate of approximately 60%. Partly due to these different definitions, it is still difficult to identify the factors that may predict for the TFR rate, although some analyses have drawn attention to the predictive value of treatment duration, Sokal score, duration of molecular response (MR), and response to first-line TKI treatment. Prior studies were mostly academic or company-spon- sored; these were mostly prospective in nature, with restricted and carefully selected inclusion criteria. Nowadays, doctors and patients are willing and ready to introduce TKI discontinuation in clinical practice. Very few data are available on the effects and the outcome of treatment discontinuation outside prospective studies and without a central control of MR. We report here on 293 adult patients who discontinued TKI outside studies, as per clinical practice.
Methods
Study design and purpose
We designed a retrospective observational study of Italian patients with Philadelphia positive (Ph+) CML in chronic phase who had discontinued TKI treatment in DMR, with a follow up after discontinuation over one year. All hematology centers
belonging to the Italian Group for the Hematologic Diseases of the Adults (GIMEMA) were invited to participate; thirty-two centers contributed to this study. The primary end point of the study was the TFR rate after one year from TKI treatment dis- continuation. Secondary end points included: longer-term TFR status, safety (including the outcome after treatment resumption and disease progression), identifying factors associated with MR. Data on the main disease characteristics were collected for each patient. These were: all treatments before and after discon- tinuation, duration of each treatment, response to each treat- ment, and the reasons for discontinuation. The cutoff date for this analysis was February 2017. The observational retrospective study protocol was approved by the ethics committees of all centers taking part.
Response definitions and statistical analysis
Molecular response was assessed by quantitative polymerase chain reaction (qPCR) according to the standard methodology;26 all analyses were performed by the GIMEMA Laboratories Network (LabNet) for CML, expressed according to the International Scale. Major molecular response (MMR) was defined as a BCR-ABL1 ratio ≤0.1 with at least 10,000 ABL1 copies. Deep molecular response was defined as MR4 (BCR-ABL1 ratio ≤0.01% with at least 10,000 ABL1 copies), or MR4.5 (BCR-ABL1 ratio ≤0.0032% with at least 32,000 ABL1 copies), or MR5 (BCR-ABL1 ratio ≤0.001% with at least 100,000 ABL1 copies) confirmed at least three times before TKI discontinuation.26 In a few patients who discontinued TKI before the establishment of molecular stan- dardization, DMR was defined as a level of BCR-ABL1 transcript undetectable by qPCR or by qualitative PCR, confirmed in at least two controls. The cytogenetic response was assessed according to European LeukemiaNet (ELN) criteria.27
Treatment-free response was assessed using the Kaplan-Meier method, from the date of TKI discontinuation to the date docu- menting the restart of therapy regardless of the reason. In fact, since this is a retrospective study, criteria for treatment resumption have not been pre-established. TFR was estimated using a Kaplan- Meier curve and 95% confidence interval (CI). Deaths were con- sidered as censored events. For all the other patients, data were censored at the date of last qPCR.
Continuous data were expressed as medians with interquartile ranges (IQR, i.e. 25th and 75th percentiles) as a measure of variabil- ity. A Mann-Whitney U test was used for comparison of quantita- tive variables and χ2 or Fisher exact test was used for categorical variables as appropriate.
Clinical and biological variables at baseline were assessed as potential independent prognostic factors for MR by univariate analysis using Cox regression model. Variables were entered with- out any transformation or cut off.
For the multivariate analysis, a stepwise backward selection procedure was carried out.28 The non-linear effect of continuous covariates was modeled using a restrictive cubic spline function, and its significance was assessed using the Wald test; similar methods were used to check interactions.29 The best fitting model was chosen according to the Akaike information criterion.
P=0.05 was considered statistically significant. All analyses were carried out using R v.3.3.3 statistical software.30
1590
haematologica | 2019; 104(8)