Page 133 - 2019_10 resto del Mondo_web

Meta-analysis of the impact of MRD in adult B-ALL
for patients with Ph-positive disease in particular, out- comes with standard combination chemotherapy are poor.3 The tyrosine kinase inhibitors imatinib, dasatinib, nilotinib, and ponatinib have shown efficacy in patients with Ph-positive disease, and there are some preliminary suggestions that they may also be efficacious in selected subsets of Ph-negative ALL.6–8 Moreover, several targeted therapies have been developed for patients with B-ALL, such as the antibody-based therapies rituximab, ino- tuzumab ozogamicin and blinatumomab, these latter demonstrating greater effectiveness than salvage chemotherapy in the very high-risk setting of relapsed/refractory ALL, including both Ph-positive and Ph-negative patients.9–11
The introduction of targeted therapies, alongside advances in diagnostic procedures, have improved out- comes for patients with B-ALL.12,13 However, despite a substantial proportion (74% to 91%) of patients achieving complete remission (CR), one-third or more will eventual- ly relapse because of the presence of submicroscopic lev- els of leukemic cells in the bone marrow.14–18 The presence of these remaining cancer cells is known as minimal resid- ual disease (MRD; alternatively termed ‘measurable resid- ual disease’).
MRD is increasingly being used in clinical practice as an independent prognostic marker for the duration of CR and long-term outcomes in patients with ALL, and for inform- ing treatment decisions.19–22 The European Society for Medical Oncology (ESMO) and National Comprehensive Cancer Network clinical practice guidelines for adult patients with ALL recommend the quantification of MRD whenever possible,6,23,24 but it is not yet clear whether MRD status alone is sufficient to predict prognosis, or whether it should be combined with other parameters, such as patient-related (e.g., age) or disease-related (e.g., cytogenetics) risk factors. In drug development, MRD response has been considered as an early marker of effica- cy in clinical studies, with potential use as a surrogate end- point in registration studies for accelerated drug approval.25,26 A full understanding of the prognostic signif- icance of MRD is needed to guide its use across this broad range of settings.
A large number of studies have shown that achievement of MRD-negative status correlates positively with CR duration, reduced risk of relapse, and HSCT success.22,27 The use of MRD testing is individualized within a given study protocol, and hence there is wide variation in the test method used, the timing and sensitivity of MRD assessment, the characteristics of the patients, and the treatments used before and after MRD was assessed. We conducted a systematic review to capture the evidence supporting the clinical significance of MRD on clinical out- comes and used this evidence to inform a meta-analysis with the aim of quantifying the impact of MRD status on relapse-free survival (RFS) and overall survival (OS). A recent meta-analysis suggested that event-free survival and OS were almost always better in patients with ALL who were MRD negative than in those who were MRD positive.28 This meta-analysis included studies in children and adults, and in B-cell and T-cell phenotypes. Our meta- analysis focused on adult patients with B-ALL and includes 23 studies, 18 of which were not in the previous meta-analysis, which enabled us to explore the impact of MRD status within clinically important subgroups in this population.
Methods
Systematic literature review
Evidence base
We systematically reviewed published studies according to a prespecified protocol, using a process compliant with the 2009 Preferred Reporting Items for Systematic Reviews and Meta- Analyses (PRISMA) guidelines.29 PubMed and Embase databases were searched (Online Supplementary Table S1) for studies in humans published in English between 1 January, 1995 and 1 March, 2016. Additional searches were conducted for congress proceedings [American Society of Clinical Oncology (ASCO), American Society of Hematology (ASH), European Hematology Association (EHA), and ESMO] published between 2012 and 2016.
Screening and data extraction
Titles or abstracts identified from the initial searches were eval- uated against the prespecified inclusion criteria (Online Supplementary Methods). The full texts were obtained for all studies deemed eligible, and for studies whose eligibility was unclear dur- ing the title/abstract screening. The full texts were independently screened by two reviewers to confirm which studies should be included. Any discrepancies were resolved by a third reviewer. Data extraction is described in the Online Supplementary Methods.
Meta-analysis
Evidence base
Studies identified from the systematic literature review were excluded from the meta-analysis if RFS was not reported separate- ly for patients who were MRD-negative and MRD-positive in sta- tus, or if insufficient data were reported to calculate a hazard ratio according to the methods outlined in Tierney et al.30 For studies that included patients with B-ALL and T-cell ALL (T-ALL), hazard ratios were calculated only for the patients with B-ALL, if it was possible to do so.
Outcomes
The primary outcome was RFS – alternatively termed disease- free, event-free or leukemia-free survival in some publications. In studies that provided RFS definitions, RFS was mostly measured from CR until relapse or death. Other studies that provided a def- inition measured RFS from HSCT31–33 or from the start of treat- ment.34–36 For the purposes of the meta-analysis, we used the meas- ure of RFS as provided in each study publication. The primary analysis set was based on studies that reported RFS according to MRD status. The secondary outcome was OS, determined from those studies in the primary analysis set that also reported OS according to MRD status.
Statistical analysis
The analysis of hazard ratios used unadjusted measures of treat- ment effect, where available. Meta-analysis was performed using the random effects model.37,38 Heterogeneity between studies was assessed statistically using the I2 and Cochran Q tests.39,40 Meta- regression was performed to investigate the relationships between covariates (Ph status, median follow-up time, MRD cut-off sensi- tivity threshold, post-MRD treatment, disease stage, sex and age) and study-level hazard ratios.38
Results
Search results
The systematic literature search identified 1,899 records (1,252 full papers and 647 congress abstracts) for
haematologica | 2019; 104(10)
2029