Burden of hospitalization in acute lymphoblastic leukemia patients treated with Inotuzumab Ozogamicin versus standard chemotherapy treatment

Abstract Background Inotuzumab Ozogamicin (INO), has demonstrated an improvement in overall survival, high rate of complete remission, favorable patient‐reported outcomes, and manageable safety profile vs standard of care (SoC; intensive chemotherapy) for relapsed/refractory (R/R) acute lymphoblastic leukemia (ALL) in the phase 3 INO‐VATE trial. With a one‐hour weekly dosing schedule, INO might be associated with lower healthcare system burden. This study analyses hospitalizations for INO vs SoC. Methods All patients receiving study treatment in the INO‐VATE trial were included. The days hospitalized during study treatment was calculated. Due to different treatment durations for INO and SoC (median of 3 vs 1 cycles), number of hospital days was mainly reported per observed patient month. Hospital days per patient month were analyzed for different treatment cycles, subgroups, and main reasons for hospitalization. Differences between treatments were analyzed by the incidence rate ratio (IRR). Results Overall, 82.9% and 94.4% INO and SoC patients experienced at least one hospitalization. The mean hospitalization days per patient month was 7.6 and 18.4 days for INO and SoC (IRR = 0.413, P < .001), which corresponds to patients spending 25.0% and 60.5% of their treatment time in a hospital. Main hospitalization reasons were R/R ALL treatment (5.2 (INO) vs 14.0 (SoC) days, IRR = 0.368, P < .001), treatment toxicities (1.4 vs 2.8 days, IRR = 0.516, P < .001) or other reasons (1.0 vs 1.6 days, IRR 0.629, P < .001). Conclusions Inotuzumab Ozogamicin treatment in R/R ALL is associated with a lower hospitalization burden compared with SoC. It is likely this lower burden has a favorable impact on healthcare budgets and cost‐effectiveness considerations.


| BACKGROUND AND OBJECTIVES
Acute lymphocytic leukemia (ALL) is a life-threatening diagnosis. 1 Current therapies for adults with newly diagnosed Bcell ALL are associated with rates of complete remission (CR) as high as 60%-90%. [2][3][4][5][6][7][8][9] However, many of the patients with CR experience a relapse. [2][3][4][5][6][7][8][9] For these patients the estimated 5-year survival rate is less than 10%. The prognosis of adults with relapsed or refractory B-cell ALL (R/R ALL) depends on several parameters, including response to prior salvage therapy, duration of first remission, patient age, and disease burden at time of relapse. 10 The only curative option consists of achieving a second CR by salvage therapy followed by an allogenic hematopoietic stem cell transplantation (HSCT), but less than half of the patients achieve a second CR and only a limited subset of patients are eligible for this procedure. 6,[11][12][13] Standard chemotherapy regimens for adults with R/R ALL are associated with rates of CR of 31% to 44% when they are the first salvage therapy administered after an early relapse, and 18%-25% when they are the second salvage therapy. 10,11,14,15 So, as CR is generally considered a prerequisite for subsequent HSCT, these low rates of CR mean that few adults with R/R ALL proceed to HSCT; a potential curative option.
A Phase III trial confirmed that Inotuzumab Ozogamicin (INO), an anti-CD22 antibody conjugated to calicheamicin, results in better outcomes in patients with R/R ALL than standard of care (SoC) chemotherapy, with a manageable safety profile. In the INO-VATE ALL trial, INO was associated with higher rates of CR/CRi within the ITT218 population than SoC (80.7% vs 29.4%, P < .001). The estimated HR for the second primary endpoint of OS was 0.770 (97.5% CI, 0.578-1.026), with one-sided P = .0203 in favor of INO over control therapy based on the stratified analysis, indicating an overall 23% reduction in the risk of death in favor of INO. The survival probability at 24 months was 23% (95% CI, 16%-30%) in the INO arm and 10% (5%-16%) in the control arm. 16 Treatment of R/R ALL is associated with a significant burden for both patients and health care systems, the latter mainly because of frequent and lengthy hospitalizations of patients. High rates of hospitalizations in this patient group can be explained by limited effectiveness of standard chemotherapy, potential toxicity of that treatment, and inconvenient chemotherapy dosing schedules. 17 Several previous studies reported that R/R ALL patients undergoing chemotherapy spent about 50% of their treatment time, defined as time between first and last administration of a dosage, in hospital. [17][18][19][20] Due to its superior efficacy and manageable safety profile as well as a convenient one-hour weekly dosing schedule, 16 INO might be associated with lower health care system burden, especially because of lower hospitalization frequency during treatment periods. Data in this respect have not been published so far. That is why, the main objective of this study was to analyze hospitalization frequency of R/R ALL patients who received either INO or SoC chemotherapy, based on the data collected in the INO-VATE ALL trial.

| MATERIALS AND METHODS
The INO-VATE ALL trial has already been described in detail elsewhere. 16 In short, INO-VATE ALL was an openlabel, randomized, controlled phase 3 trial on adult R/R ALL patients who were scheduled to receive their first or second salvage treatment. Patients were randomly assigned, in a 1:1 ratio, to receive either INO or SoC (investigator's choice); no crossover between groups was allowed. Patients who achieved complete remission could undergo stem-cell transplantation at the investigator's discretion.
Patients in the INO group received the trial drug intravenously at a starting dose of 1.8 mg per square meter of bodysurface area per cycle; they received 0.8 mg on day 1 of each cycle and 0.5 mg on days 8 and 15. Cycle 1 lasted 21 days and the subsequent cycles each lasted 28 days; the patients received treatment for up to six cycles. Once a patient achieved complete remission or complete remission with incomplete hematologic recovery, the dose that was administered on day 1 of each cycle was reduced to 0.5 mg for the duration of the trial. Patients in the SoC group received the investigator's choice of one of the following three regimens: FLAG (fludarabine, cytarabine, and granulocyte colony-stimulating factor) therapy for up to four 28-day cycles, cytarabine plus mitoxantrone for up to four 15-to-20-day cycles, or high-dose cytarabine for up to two 12-dose cycles. Details are provided in the respective trial publication. 16 Admission for drug administration was neither recommended nor mandated, but was entirely based on physician judgment and local SoC.
All analyses presented in this paper were based on the January 2017 data cut of the INO-VATE ALL trial and consider the safety population in the INO-VATE ALL trial, which included all R/R ALL patients who were randomized to either INO or SoC therapy and received at least one dose of the respective therapy. With regard to comparison of hospitalization burden between treatments, number of patients with at least one hospitalization and the total number of days a patient was hospitalized from randomization until end of study treatment was calculated. Hospitalization days before randomization in the trial were excluded from this comparison, hospitalizations after end of study treatment were not documented in the INO-VATE ALL trial.
Due to different durations of observation for INO and SoC, number of hospital days of all patients were mainly reported per observed patient month. Number of patients with at least one hospitalization and hospital days per patient month were also analyzed for different treatment cycles as well as with regard to main reasons for hospitalizations as documented by study physicians.
Number of hospitalization days per patient month was compared between all patients who received either INO or SoC by calculating the incidence rate ratio (IRR). The comparison was repeated within several subgroups: males vs females, R/R ALL patients treated in different regions of the world, older vs younger patients (<55 years vs >55 years), short vs long duration of first remission (<12 months vs >12 months), salvage status (first vs second), and Philadelphia status (positive vs negative).
Finally, in a multivariable analysis influence of the following independent variables on hospitalization burden, defined as hospital days per observed patient month, was estimated: age, gender, salvage status, Philadelphia status, duration of first remission, INO vs SoC treatment. Generally, due to the categorical nature of the dependent variable, either a negative binominal regression or a Poisson regression model was considered for this analysis. [21][22][23] In order to identify the most suitable model, first, the requirement of overdispersion was tested. Afterward, fit of both types of models was assessed based on the Akaike's information criterion (AIC) as well as the Bayesian information criterion (BIC). The model with the best fit was chosen. To express coefficients of that model in a more interpretable form, we estimated the IRR related to observed hospitalization days per patient month for each independent variable in addition to the respective coefficients. In these estimates, the IRR can be interpreted as the ratio of two the number of hospitalization days per month in the INO vs the SoC arm.
All analyses were done with R 3.4.3 and Stata version 14.1 software. The INO-VATE ALL trial was approved by an ethics commission, 16 no separate approval for this analysis was obtained or deemed necessary. Some of the clinical data presented within this article has previously been presented at conferences. 24,25

| RESULTS
In the INO-VATE ALL trial (safety population), 164 patients received INO and 143 received SoC (93 patients received FLAG, 33 Ara-C/Mito and 17 HIDAC). Mean age of the whole population was 46 years, 40.4% of patients were female, and about 48.5% of patients were treated at North American study sites. 66.8% received their salvage 1 therapy at date of study start, the remaining patients received salvage 2 therapy. Patients in the INO arm received a median of 3 cycles of study therapy compared to a median of 1 cycle for the SoC arm ( Table 1). The majority of patients (85.7%) were Philadelphia negative. During the treatment period, patients in the INO arm were observed significantly longer than SoC patients (mean: 85.0 days vs 45.2 days per patient). Table 2 describes observed overall hospitalizations for trial patients as well as hospitalizations per treatment cycle.
Overall, 136 of 164 (82.9%) INO patients and 135 of 143 (94.4%) SoC patients experienced at least one hospitalization. The mean hospitalization days per patient month in the INO arm was 7.6 days compared with 18.4 days in the SoC arm, resulting in an IRR of 0.413 (P < .001). During treatment cycle 1, the mean hospitalization days per patient month are 12.5 days in the INO arm and 18.9 days in the SoC arm (IRR: 0.661, P < .001). In cycles 2, 3, and 4 the hospitalizations in the INO arm are lower compared to the SoC arm; 6.1, 5.2, and 3.8 days per patient month vs 16.7, 14.1, and 8.8 days per patient month, respectively. The IRRs for cycles 2, 3, and 4 are 0.363 (P < .001), 0.365 (P < .001), and 0.424 (P = .0246).
Based on the overall observational period, Figure 1 describes the IRRs comparing the hospitalization days per patient month in the INO arm with the SoC arm within predefined subgroups. In all comparisons, hospital burden, defined as observed hospitalization days per patient month, was lower in the INO arm compared with the SoC arm. Highest differences could be observed in the subgroup of patients included in North America with an IRR (INO vs. SoC) of 0.233 (P < .001) and in patients with a duration of first remission at randomization of ≥12 months with an IRR ( Table 4 summarizes the results of the multivariable analysis of factors being associated with the number of hospitalization days per observed patient month. In this model, age, gender, a duration of first remission of <12 months, salvage stage as well as Philadelphia status were not significantly associated with the dependent variable. However, patients having been treated in Asia experienced a higher hospitalization burden (IRR = 2.210 compared with North America, P < .001). Finally, INO was associated with a significantly lower number of hospital days per patient month, in comparison to SoC (IRR = 0.404, P < .001).

| DISCUSSION
The INO-VATE ALL trial showed patients receiving INO vs SOC patients achieved higher response (80.7% vs 29.4%, P < .001), MRD-negativity rates (2.8-fold increase; P < .001), and prolonged progression-free survival and overall survival. Veno-occlusive disease (VOD) was a major nonhematologic toxicity. 16 Using the documented on-treatment hospitalizations of the INO-VATE ALL trial, the main aim of our analysis was to report treatment-associated hospital burden of R/R ALL patients and to compare this burden between patients who received either a treatment with INO or SoC.
We confirm the results of earlier studies that hospital burden is high in this patient population. 17 26 Finally, as hospitalizations were documented on treatment only in the R/R ALL trial, our analysis only refers to these periods. This implies that our current analysis does not capture the potential impact of INO-related VOD on hospitalizations, as VOD occurs within the HSCT period and not during treatment with either INO or SoC.
Whether there are other differences in hospitalization burden between INO and SoC patients after end of treatment is unknown. We conclude that INO treatment is not only more effective with a significantly greater CR/CRi rate than SoC in R/R ALL patients but is also associated with lower hospital burden. In line with this, in a previous analysis of the INO-VATE ALL trial data, Kantarjian et al found that, regardless of the treatment received, non-hospitalized patients exhibited improvements in measured patient-reported outcomes such as the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (EORTCQLQ-C30) and the EuroQol Group 5 Dimensions Questionnaire (EQ-5D), whereas hospitalized patients exhibited deteriorations or minimal changes from baseline in this respect. 27 Therefore, it is expected that the lower hospital burden of INO is likely to contribute to better patient reported outcomes, with patients being more ambulatory and realizing significantly less negative impact on their daily lives. Even if INO patients are generally treated longer and a higher percentage proceeds to SCT, we also conclude that the above lower hospital burden probably has a favorable impact on health care budgets and cost-effectiveness considerations. Collection of further data in the real-world setting is warranted to verify and further quantitate these findings.

CONFLICTS OF INTEREST
This study was sponsored by Pfizer Inc David I. Marks and Hagop Kantarjian were part of the advisory board and received honoria from Pfizer. Ilse van Oostrum and Sabrina Mueller are employees of Ingress-Health, who were paid consultants to Pfizer in connection with the development of this manuscript. Verna Welch, Erik Vandendries, Fausto R. Loberiza, and Sarah Böhme are employees at Pfizer. Yun Su is a former employee at Pfizer. Matthias Stelljes reports research support from Pfizer during the conduct of the study and other consulting fees and honoraria from Pfizer, Amgen, and Jazz Pharmaceuticals.