Answers are grounded in this paper and cite their sources.
To the Editor: Single nucleotide polymorphisms (SNPs) of CD33 are predictive for relapse in children with acute myeloid leukemia (AML) receiving frontline therapy with a regimen containing the anti-CD33 antibody-drug conjugate gemtuzumab ozogamicin (GO).1 Similarly, SNPs of ABCB1 influence GO-mediated calicheamicin accumulation in preclinical leukemia models, and studies suggest that these SNPs may also impact the risk of relapse in children treated with GO.2 However, the impact of CD33 and ABCB1 SNPs on response and the risk of relapse with GO-based therapies in adults is less clear. One large study suggested no predictive impact of the rs12459419 CD33 SNP in a cohort of adult patients with AML treated with frontline intensive chemotherapy with or without GO.3 In light of these contradictory reports, we sought to evaluate the impact of CD33 and ABCB1 SNPs in patients with AML, higher-risk myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML), or primary myelofibrosis (PMF) who received decitabine plus GO in a prospective clinical trial (NCT00882102), or who received the same regimen outside of a clinical trial.4 One-hundred and thirteen patients received decitabine 20 mg/m2 daily on days 1-5, plus GO at 3 mg/m2 on day 5 as induction, followed by up to five consolidation cycles of decitabine plus GO, then decitabine alone.4 As previously reported, this regimen did not improve overall survival compared to historical controls. One-hundred and four patients (92%) had stored samples available for retrospective analysis of CD33 and ABCB1 SNPs, and are the subject of this report (on protocol, n = 101; off protocol n = 3). Genotyping of the following SNPs was performed and correlated with clinical outcomes: ABCB1 rs1128503, ABCB1 rs1045642, CD33 rs2455069, CD33 rs35112940, CD33 rs12459419, CD33 rs1803254, CD33 rs61736475, and CD33 rs201074739. An aggregate “6 CD33 SNP score” was determined based on genotypes of the six CD33 SNPs, as previously described.5 A higher score is anticipated to be associated with increased anti-leukemia response to GO. Leukemic cell surface expression of CD33 by mean fluorescence intensity (MFI) was also evaluated. The characteristics of the 104 evaluable patients are shown in Table S1. Forty-nine patients were frontline (AML, n = 36; other, n = 13) and 55 patients had relapsed/refractory disease (AML, n = 46, other, n = 9). The median CD33 expression was 82.8% (range, 0-100%) and median CD33 MFI ratio was 29.8 (range, 1.0-180.4). Frequency of CD33 and ABCB1 SNPs are shown in Table S2. Fifty-one patients (49%) had a 6 CD33 SNP score < 0, and 53 patients (51%) had a score ≥ 0. There was no significant association between median MFI and any individual SNP or the CD33 SNP score. For the entire cohort, 33 patients (32%) achieved any response (overall response rate, ORR), defined as a composite of complete remission [CR] + CR with inadequate hematologic recovery + morphologic leukemia-free state. Fifteen patients (14%) achieved CR. The ORR was 47% and 18%, and CR rate was 22% and 7% for the frontline and relapsed/refractory cohorts, respectively. With a median follow-up of 107 months, median relapse-free survival (RFS) and overall survival (OS) were 4.7 and 10.0 months in the frontline, and 4.4 and 6.7 months in the relapsed/refractory cohorts, respectively. The 1- and 2-year OS in the frontline cohort were 43% and 12%, and in the relapsed/refractory cohort were 29% and 13%, respectively. For the whole cohort, CD33 MFI was associated with CR in univariate analysis (odds ratio (OR) 1.02 (95% CI, 1.00-1.03); P = .01) and remained significant in multivariate analysis adjusting for age, diagnosis (AML vs others), disease status (frontline vs relapsed/refractory) and cytogenetics (P = .01). No individual CD33 or ABCB1 SNP was predictive for ORR or CR; however, there was a trend towards higher CR rate in patients with a 6 CD33 SNP score ≥ 0 (21% vs 8%, OR 3.08 [95% CI, 0.91-10.40]; P = .07), but not for ORR (P = .94). Neither CD33 MFI nor the 6 CD33 SNP score were significantly associated with cumulative incidence of relapse (CIR), RFS, or OS (Figure 1A-F). In the global population, only the CD33 rs1803254 SNP was associated with CIR and RFS, although it was not associated with OS (Figure S1A-C in Appendix S1). The median RFS was 5.7 months vs 2.5 months (hazard ratio [HR] 4.54 [95% CI, 1.19-17.25]; P = .03), and the CIR was 73% vs 100% (Gray's test, P = .02) for patients with GG vs GC/CC genotypes, respectively. However, these findings were driven by only three patients in the CG/CC group who responded and were evaluable for these analyses. In the frontline population (n = 49) the CD33 rs1803254 SNP was associated with CIR (median 1.8 vs 10.1 months for CC/CG and GG genotypes, respectively, P < .01), RFS (median 2.2 vs 6.8 months, respectively, P < .01), and OS (median 6.6 vs 11.4 months, respectively, P = .04). (Figure S2A-C in Appendix S1). Interestingly, the CD33 rs1803254 CG/CC genotype was more frequent among black patients compared to non-black patients (38% vs 13%), and black patients had worse CIR and RFS (Table S3 and Figure S3A-C in Appendix S1). No other CD33 or ABCB1 SNP was significantly associated with CR, ORR, CIR, RFS or OS in the global cohort or in any of the subgroup analyses. Survival outcomes according to CD33 MFI and CD33 6 SNP score. (A) Overall survival according to CD33 MFI and (B) CD33 6 SNP score. (C) Relapse-free survival according to CD33 MFI and (D) CD33 6 SNP score. (E) Cumulative incidence of relapse according to CD33 MFI and (F) CD33 6 SNP score. Abbreviations: MFI, mean fluorescence intensity; SNP, single nucleotide polymorphism In frontline pediatric AML studies, the CD33 rs12459419 SNP has been shown to influence long-term outcomes with the combination of chemotherapy plus GO.1 In contrast, our study did not show a significant impact of this SNP on RFS or OS in patients with AML, or other advanced myeloid malignancies treated with decitabine plus GO. Although different from what has been reported in children, this finding is consistent with two prior analyses evaluating this SNP in the context of regimens containing GO, or the anti-CD33 antibody drug conjugate SGN-CD33A.3, 6 We did however observe higher relapse rates in both the global and frontline populations among patients with the CC or CG genotypes of CD33 rs1803254, which were cumulatively present in 16% of the patients. A previous study in children also showed poorer outcomes with GO-based therapy for patients with these genotypes, although in this pediatric study the inferior outcome was due to increased treatment-related mortality rather than increased relapse rate, as we observed in our study.7 It should also be noted that we observed a higher prevalence of the CC or CG genotypes of CD33 rs1803254 in black patients, a group that had higher relapse rates compared to non-black patients. As numbers were too small to perform multivariable analysis, we cannot fully assess the relative contributions of genotype and race to the poorer outcomes in these groups. It should be noted that our study population generally had very poor-risk disease and was relatively heterogeneous. It is possible that this heterogeneity could have limited our ability to detect differences based on CD33 or ABCB1 genotypes that may have been evident in a more homogeneous population. Furthermore, as decitabine plus GO produced a relatively low global ORR rate (32%), only a minority of patients were evaluable for the CIR and RFS analyses. Previous pediatric studies have suggested that CD33 SNPs influence relapse risk but not initial response. Therefore evaluation of these SNPs in the context of more effective GO-based therapeutic strategies that yield higher response rates (eg, novel regimens incorporating venetoclax [NCT03390296] or frontline induction with GO), may help to further characterize the predictive impact of these SNP in adults. In conclusion, in this cohort of high-risk adult patients with AML or other advanced myeloid malignancies treated with decitabine plus GO, the CD33 rs1803254 CG/CC genotype was associated with significantly worse CIR and RFS. However, no other CD33 or ABCB1 SNP was predictive for clinical outcomes. Future studies of the impact of CD33 and ABCB1 SNPs in the context of more effective GO-based regimens are needed. The authors report no conflicts of interest. N.J.S., G.R-C., J.K.L. and N.D. designed the study, analyzed the data and wrote the manuscript. R. K-S., K.P., and I.P. performed flow cytometry and sequencing, M.K., G.B., F.R., C.D.D., T.M.K. and H.K. treated and enrolled patients. All authors reviewed and edited the manuscript and approved the final version. Supported by the MD Anderson Cancer Center Support Grant CA016672, the Charif Souki Fund, and the Dick Clark Immunotherapy Research Fund. N.J.S. is supported by the K12 Paul Calabresi Clinical Oncology Scholar Award and the American Society of Hematology Junior Faculty Scholar Award in Clinical Research. Jatinder Lamba is supported by the Leukemia Lymphoma Society, Translational Program Award. Appendix S1: Supporting information Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
No takes yet. Share an insight, caveat, or question.
Short et al. (2020) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: