Adv Ther (2019) 36:950–961 https://doi.org/10.1007/s12325-019-0873-7 ORIGINAL RESEARCH Matching-Adjusted Indirect Comparison of Blinatumomab vs. Inotuzumab Ozogamicin for Adults with Relapsed/Refractory Acute Lymphoblastic Leukemia Jinlin Song . Qiufei Ma . Wei Gao . Ze Cong . Jipan Xie . Zachary Zimmerman . Laura Belton . Janet Franklin . Stephen Palmer Received: December 10, 2018 / Published online: February 13, 2019 Ó The Author(s) 2019 ABSTRACT 2? prior salvage therapies from TOWER were excluded because such patients were not inclu- Introduction: In the absence of head-to-head ded in INO-VATE-ALL. To ensure balance in the trials, this analysis aimed to provide a fair remaining patients, baseline characteristics for indirect comparison of the efficacy between the TOWER patients were weighted to match the blinatumomab and inotuzumab ozogamicin average baseline characteristics in INO-VATE- (InO), two treatments for adult patients with ALL, including sex, age, race, performance status, relapsed or refractory acute lymphoblastic leu- bone marrow blast, previous salvage therapy, kemia (R/R ALL) who received no more than previous allogeneic transplantation, complete one prior salvage therapy, by adjusting for cross- remission with complete hematologic recovery trial differences. (CR) to most recent induction therapy, and Methods: Patient-level data from the Phase 3 duration of first remission. Overall survival (OS), blinatumomab trial TOWER and published including median and restricted mean survival aggregated data from the Phase 3 InO trial INO- time (RMST) at 12 and 20.7 months, and CR were VATE-ALL were used to conduct matching-ad- estimated and compared. justed indirect comparisons. Patients with Results: A total of 310 patients in TOWER were included (blinatumomab, n = 203; standard of Enhanced Digital Features To view enhanced digital care chemotherapy, n = 107). After matching features for this article go to https://doi.org/10.6084/ the listed baseline characteristics, the median m9.figshare.7571138. OS was 9.3 months for blinatumomab and 7.7 months for InO (weighted log-rank test J. Song Á J. Xie p = 0.4). The relative RMST at 12 months was Analysis Group, Inc., Los Angeles, CA, USA 1.6 months longer for blinatumomab than for Q. Ma Á Z. Cong Á Z. Zimmerman Á J. Franklin InO [95% CI (0.1, 3.2); p = 0.04]; at 20.7 months Amgen Inc., Thousand Oaks, CA, USA the RMST was not significantly different. The W. Gao (&) CR rates were similar [anchor-based differ- Analysis Group, Inc., Boston, MA, USA ence = - 2.8%, 95% CI (- 17.5%, 11.9%); e-mail: [email protected] p = 0.71]. L. Belton Conclusions: After adjusting for cross-trial dif- LB Biostatistics, London, UK ferences, blinatumomab demonstrated a similar CR rate and potential OS benefit versus InO S. Palmer among adult patients with R/R ALL who Centre for Health Economics, University of York, York, UK received no more than one prior salvage Adv Ther (2019) 36:950–961 951 therapy. Further studies are suggested to con- precursor B cell R/R ALL by the United States firm this finding. Food and Drug Administration (FDA) in July Funding: Amgen. 2017 (accelerated approval was granted in 2014) [11], and by the European Medicines Agency in Keywords: Blinatumomab; Complete remis- November 2015 (conditional approval) [12]. sion; Inotuzumab ozogamicin; Matching- A Phase III trial (TOWER) compared blinatu- adjusted indirect comparison; Overall survival; momab with standard of care (SOC) Relapsed or refractory acute lymphoblastic chemotherapy among patients with Ph- R/R leukemia B-cell ALL [13–16], and reported that the median OS was significantly higher for patients receiving blinatumomab (7.7 months) versus SOC INTRODUCTION chemotherapy (4.0 months), with a hazard ratio (HR) of 0.71 and a 95% confidence interval (CI) Acute lymphoblastic leukemia (ALL) is a rare of 0.55–0.93 (p = 0.01). Furthermore, 33.6% of but aggressive type of leukemia characterized by patients receiving blinatumomab achieved the proliferation of immature lymphoid cells in complete remission with full hematologic the bone marrow, peripheral blood, and other recovery (CR) within 12 weeks after the initia- organs [1]. Approximately 20% of ALL cases are tion, compared with 15.7% of patients receiving in adults (aged C 18 years) [2]. In Europe, the SOC chemotherapy (p \ 0.001). overall prevalence of ALL was approximately 1 InO is a humanized anti-CD22 antibody in 10,000 people in 2014 [3]. conjugated to calicheamicin (a potent cytotoxic Current treatments for ALL include antibiotic), which received FDA approval for chemotherapy, often with the intent of adult R/R B-cell precursor ALL in August 2017 hematopoietic stem cell transplant (HSCT). [17]. In the Phase III trial INO-VATE-ALL, the There are usually three phases to chemother- CR rate in the intention-to-treat population was apy: induction, consolidation, and mainte- 33.5% in the InO arm compared to 16.0% in the nance. More than 80% of adults with ALL SOC chemotherapy arm [14]. There was not a achieve complete remission with intensive statistically significant improvement in the induction chemotherapy [2, 4–8]. Patients who median OS for InO compared to the chosen fail to respond to induction therapy are con- chemotherapy regimen. The median OS as of sidered to have refractory ALL. More than half January 5, 2017 was 7.7 (95% CI 6.0–9.2) of the patients who initially responded will months for InO versus 6.2 (4.7–8.3) months for ultimately relapse [2, 4–8], i.e., leukemia cells SOC chemotherapy patients, with a HR of 0.75 reappear in the bone marrow or peripheral (97.5% CI 0.57–0.99; p = 0.04, above the pre- blood after attaining complete remission. The specified two-sided significance level boundary prognosis of relapsed or refractory (R/R) ALL for of p = 0.0208) [16, 18]. adult patients is very poor, with a median Currently, there are no randomized head-to- overall survival (OS) of 3–6 months [9]. For head trials that have compared the effects of patients receiving first-line salvage chemother- blinatumomab versus InO. These two Phase III apy, 49% were alive after 6 months, 26% at trials have substantial differences in design and 1 year, and just 11% at 3 years [9]. patient characteristics, which make direct Blinatumomab and inotuzumab ozogamicin comparisons of treatment efficacy challenging. (InO) are two recently approved treatment For example, TOWER included patients with options recommended by the latest National more than one previous salvage therapy while Comprehensive Cancer Network guidelines for INO-VATE-ALL did not. In addition, higher adult patients with R/R ALL [10]. Blinatu- proportions of patients in the TOWER popula- momab is a bi-specific T cell engager antibody tion had C 50% blasts in the bone marrow and that targets the CD19 antigen present on B had previous HSCT compared to the INO-VATE- cells. It received full approval for the treatment ALL population. Based on the literature, all of of Philadelphia-chromosome-negative (Ph-) these three disease characteristics are important 952 Adv Ther (2019) 36:950–961 factors impacting patients’ outcomes [8, 19, 20]. R/R precursor B-cell ALL (untreated first relapse Therefore, conclusions based on a naı¨ve com- with first remission duration \ 12 months, parison without adjusting for patient charac- untreated second or greater relapse, or relapse at teristics differences between trials would be any time after allogeneic HSCT) [11]. Patients biased and can be misleading. Additionally, the were randomized at a 2:1 ratio to receive bli- two trials assessed complete remission differ- natumomab or SOC chemotherapy regimens. ently. CR, complete remission with partial Patients in the blinatumomab arm received two hematologic recovery (CRh) and complete induction cycles, with up to three consolidation remission with incomplete hematologic recov- cycles and up to 12 months of maintenance, if ery (CRi) were assessed as key secondary efficacy patients achieved a bone marrow response endpoints in TOWER, but only CR and CRi were (B 5% bone marrow blasts) or CR/CRh/CRi. assessed in INO-VATE-ALL. Aggregated data for the InO and SOC arms of Matching-adjusted indirect comparison the Phase III trial INO-VATE-ALL as of January (MAIC) is a credible and widely-used method 5, 2017 were extracted from the literature for comparing outcomes of interventions in the [14, 16, 18]. INO-VATE-ALL was a randomized, absence of head-to-head trials, and uses indi- open-label study of InO compared to SOC vidual patient data from trial(s) of one treat- chemotherapy in adult patients with R/R CD22- ment to match the baseline summary statistics positive ALL. The trial enrolled 326 patients reported from trial(s) of a comparator [21]. After who were randomized at a 1:1 ratio to receive matching, treatment outcomes are compared InO or SOC chemotherapy. Patients in the InO across balanced trial populations using an arm could receive up to 6 cycles with no approach similar to propensity score weighting. maintenance phase. It is one of the population-adjusted indirect As this was a post hoc analysis of previously comparison approaches recognized by the Uni- published data, no institutional board review ted Kingdom’s National Institute for Health and was required. This analysis followed the princi- Care Excellence and is well accepted across ples of the Declaration of Helsinki. various health technology assessment (HTA) bodies [22, 23]. To adjust for heterogeneity in Sample Selection enrollment criteria and baseline characteristics between the trial populations and to provide a The study populations of the INO-VATE-ALL fair comparison between blinatumomab and and TOWER trials were adults (aged C 18 years) InO, this analysis used MAIC to indirectly with R/R ALL, Eastern Cooperative Oncology compare efficacy (OS and CR) between these Group (ECOG) performance status B 2, and therapies for R/R ALL.