Page 118 - 2020_01-Haematologica-web

M. Stahl et al.
the effect size in subgroup analysis, and led to a reduction in heterogeneity.
For ORR, HI-E and TI, the study with the largest influ- ence on the heterogeneity of these outcomes was the study by Xiao et al. examining the use of ATG + CsA.13 Removal of this study changed the ORR by 5.1% (from 42.5% to 37.4%) in the overall analysis and by 6.5% (from 35.2% to 28.7%) in the subgroup analysis of studies examining ATG + CsA. In addition, removal of this study led to a loss of heterogeneity in the overall analysis (I2=62%, Cochran’s Q statistic = 52.1, P=0.001) and in the subgroup analysis of studies examining ATG + CsA (I2= 0%, Cochran’s Q statistic = 3.4, P=0.64). Removal of the study by Xiao et al. also changed the HI-E by 3.9% (from 41.8% to 37.9%) in the overall analysis and 17.1% (from 44.8% to 27.7) in the subgroup analysis of studies exam- ining ATG + CsA. This led to a loss of heterogeneity in the overall analysis (I2 = 55%, Cochran’s Q statistic = 31.1, P=0.005) and in the subgroup analysis of studies examin- ing ATG + CsA (I2 = 0%, Cochran’s Q statistic = 0.17, P=0.68). For TI, removal of this study changed the out- come by 2.8% (from 33.4% to 30.6%) in the overall analy- sis and by 8.8% (from 28.4% to 19.6%) in the subgroup analysis of studies examining ATG + CsA. This also led to a loss of heterogeneity in the overall analysis (I2=35%, Cochran’s Q statistic = 18.6, P=0.1) and in the subgroup analysis of studies examining ATG + CsA (I2=40.6, Cochran’s Q statistic = 3.4, P=0.19).
The study with the largest influence on the AML pro- gression rate was that reported by Passweg et al.,18 the removal of which changed the AML progression rate by 0.5% (from 8.6% to 8.1%). In addition, removal of this study led to a loss of heterogeneity (I2 = 67%, Cochran’s Q statistic = 27.6, P=0.001).
Discussion
To our knowledge, this is the first systematic review and meta-analysis on IST in MDS, and included a total of nine prospective cohort studies and 13 clinical trials. The meta-analysis of these studies demonstrated an ORR of 42.5% (with a CR rate of 12.5%) and achievement of red blood cell transfusion independence in one-third of the patients.
Previous retrospective studies reported similar ORR and TI rates among MDS patients with IST. A recent uncon- trolled large international retrospective study in MDS patients treated with various different IST regimens demonstrated results very similar to our meta-analysis, with an ORR and TI rate of 48% (with 11.2% achieving a CR) and 30% of patients, respectively. In other large retro- spective studies, the ORR and TI rates ranged from 30% to 42% and from 31% to 41%, respectively.22,23,42 In our meta-analysis, ATG +/- CsA was the most commonly used IST regimen, similar to the finding in a recent large retrospective study.12 Importantly, while the National Comprehensive Cancer Network (NCCN; Version 2.2019) guidelines suggest the use of ATG +/- CsA as a treatment option for certain types of MDS,14 they do not suggest other IST regimens. However, in our meta-analysis, mul- tiple different IST regimens other than ATG +/- CsA were included, among them ATG + etanercept, CsA and several “other IST” regimens including etanercept +/- azacitidine, sirolimus, mycophenolate mofetil and alemtuzumab. It is
Figure 4. Rate of acute myeloid leukemia (AML) progression during study dura- tion. Forest plots of odds ratios (squares, proportional to study weights used in meta-analysis, 95% confidence intervals) for various forms of immunosuppres- sive therapy (IST), with the summary measure (center line of diamond) and asso- ciated confidence intervals (lateral tips of diamond) for rate of transformation to AML during the study period.
important to point out that these other IST regimens do not have completely overlapping mechanisms of action, tolerability, and expected response rates compared to ATG and CsA. Acknowledging these differences, we included these IST regimens in our analysis as they all provide an element of immunosuppression regardless of their specific mechanism of action.
While the identification of clinical and molecular mark- ers to predict response to IST would be of clinical impor- tance to optimize treatment of individual patients, data for several of these co-variates that had been proposed, such as younger age, shorter disease duration, hypocellu- lar bone marrow, or the presence of HLA DR15 and PNH clones, are controversial.12,21-23,42,43 Given the heterogeneity of the studies included in this meta-analysis, we were unable to address the utility of predictive biomarkers as they were only reported by a minority of studies. However, several studies, including the largest study to date by Stahl et al., were unable to validate any predictive biomarkers except of bone marrow hypocellularity.42-44
Based on available prospective data, the current National Comprehensive Cancer Network guidelines sug- gest IST with ATG +/- CsA as a treatment option for symptomatic anemia in low-risk, non-del(5q) MDS espe- cially for patients younger than 60 years of age, ≤5% blasts in the bone marrow, or with a hypocellular bone marrow, PNH clone positivity, or STAT-3 mutant cytotox- ic T-cell clones.14 However, further prospective studies are warranted to verify these predictive markers. It also remains to be shown how the wider availability of genetic testing, for example, by next generation sequencing, will impact individualized treatment decisions for MDS patients. Supporting a potential role in guiding manage-
108
haematologica | 2020; 105(1)