Page 136 - 2019_10 resto del Mondo_web

R. Bassan et al.
Europe or East Asia, and almost all studies prospectively assessed MRD; only two studies included patients exclu- sively in second CR (CR2) or later, and the rest were in patients in first CR (CR1) at the time of MRD assessment (of these, three studies included a minority of patients in CR2 or later in combination with patients in CR1, and these were categorized as CR1 for the purposes of the meta-analysis). Approximately half were in Ph-positive patients (n=11), and these patients had typically received chemotherapy plus a tyrosine kinase inhibitor before their MRD assessment (Online Supplementary Table S2). Most patients with Ph-negative ALL received chemotherapy, with or without targeted agents. As mentioned, there were only two studies in which all patients were in CR2 or later, and these patients had received treatment with blinatumomab or inotuzumab ozogamicin, with or with- out chemotherapy, before MRD was assessed.36,43 Two studies, in Ph-positive patients, examined the impact of MRD status following transplantation.35,52
Treatment received after MRD assessment varied across, and within, the studies; in most studies (n=14), pooled survival outcomes were reported for the mix of post-MRD treatments (targeted therapy, chemotherapy and HSCT) (Table 1). Six studies reported outcomes sepa- rately for patients who received an HSCT after their MRD assessment; a further two studies reported outcomes sep- arately for post-MRD treatment with a targeted agent.
Minimal residual disease assessment methodology and timing
MRD was most commonly assessed at a central labora- tory (n=10); other studies used a local reference laboratory (n=7) or did not specify whether centralized assessment was performed (n=6). A range of methodologies were used to assess MRD, although most (n=17) were poly- merase chain reaction (PCR) based (Table 1). All but one of the studies of patients with Ph-negative ALL used PCR- based methodologies; three specified that they looked for immunoglobulin or T-cell receptor gene rearrangements (Online Supplementary Tables S2 and S3). The remaining study used multiparameter flow cytometry (Online Supplementary Table S2). For patients who had Ph-positive ALL, all studies (n=11) used BCR-ABL as a marker of MRD; most (n=9) used real-time quantitative PCR, one study used molecular testing and one study used fluores- cent in-situ hybridization. The most commonly used sen- sitivity limit was 1 in 104 cells for all methodologies, and sensitivities of 1 in 105 cells were only reported for PCR- based detection in studies of patients with Ph-positive sta- tus (Table 1). The three studies that included patients with mixed Ph status or did not report Ph status all used multi- parameter flow cytometry (22%) (Online Supplementary Tables S2 and S3). The number of colors used ranged from three to eight, and no trends in the number of colors used were observed over time in the studies.
MRD measurements were generally taken within 3 months of induction treatment (n=14) (Online Supplementary Tables S2 and S3).
Clinical outcomes according to minimal residual disease status
Relapse-free survival
Figure 2 shows the meta-analysis results for the 23 stud- ies included in the primary analysis. The overall results show improved RFS for patients who achieved MRD neg- ativity [random effects hazard ratio (HR)=2.34; 95% con-
fidence interval (CI): 1.91–2.86], and the effect was seen consistently across all studies.
The I2 value was 59%, indicating moderate-to-high het- erogeneity between the studies in the primary analysis. This effect was anticipated given the known variations between studies in the design, patients’ characteristics and treatments received. For this reason, a predefined set of subgroups was examined to explore whether MRD nega- tivity had a consistent effect in important subgroups of patients.
All subgroups showed an improved RFS in patients who achieved MRD negativity; no significant differential sub- group effects were observed (Figure 3). Although it should be noted that for some subgroups there were very few studies available, which contributed to overlapping confi- dence intervals, some trends were seen. The prognostic value of MRD appeared stronger in patients who received chemotherapy than in those who received targeted thera- py before MRD assessment (HR=2.98; 95% CI: 2.12–4.20 and HR=0.90; 95% CI: 1.53–2.36, respectively), but the confidence intervals overlap. The subgroup that received HSCT before MRD assessment has a hazard ratio of 5.19 (95% CI: 1.95–13.8), but it should be noted that this is based on one study, and the confidence intervals are very wide. The prognostic value of MRD also seemed to be stronger in patients who received chemotherapy after their MRD assessment (HR=6.52; 95% CI: 2.43–17.5) than in those who received HSCT after MRD assessment (HR=1.73; 95% CI: 1.27–2.37), but again it should be noted that the data for chemotherapy after MRD assess- ment are from only one study.
The subgroup analysis was repeated separately accord- ing to Ph status and timing of MRD assessment (Online Supplementary Figures S1 and S2). There was no difference in RFS improvement for patients who achieved MRD neg- ativity between studies in the Ph-positive (HR=2.04; 95% CI: 1.53–2.73) and Ph-negative (HR=2.46; 95% CI: 2.02– 2.98) groups. Hazard ratios in all subgroups favored MRD negativity, regardless of Ph status, with no apparent differ- ence between Ph-positive and Ph-negative cases in any subgroup (Online Supplementary Figure S1). Similarly, there was no difference in RFS improvement for patients who achieved MRD negativity at early timepoints (within 3 months from induction: HR=2.60; 95% CI: 2.05–3.31) and later timepoints (more than 3 months from induction: HR=2.23; 95% CI: 1.67–2.97). Hazard ratios were more favorable in most subgroups at the early MRD timepoint compared with the later MRD timepoint group, but the subgroup confidence intervals overlapped between the early and later MRD timepoints (Online Supplementary Figure S2).
Overall survival
The meta-analysis results for the 14 studies included in the secondary (OS) analysis are summarized in Figure 4. The overall results show improved OS for patients who achieved MRD negativity (HR=2.19; 95% CI: 1.63–2.94). Individually, the results were consistently in favor of MRD negativity across all the studies with the exception of that by Bachanova et al.,31 who reported a hazard ratio of 0.94, but with a confidence interval that crossed the null value (95% CI: 0.65–1.35). Four other studies also had confi- dence intervals that crossed the null value.36,43,44,56
The I2 value was greater for the OS analysis (67%) than for the RFS analysis (59%), again confirming the extent of
2032
haematologica | 2019; 104(10)