Model-based dosing for better survival after transplantation

In this issue of Blood Advances, Dvorak et al¹ enhanced our understanding of the impact of rabbit antithymocyte globulin (rATG) exposure on outcomes after allogeneic hematopoietic cell transplantation (HCT). They investigated estimated rATG pharmacokinetic (PK) exposure using a validated population PK model in a large cohort of pediatric patients with leukemia after α/β T-cell receptor–depleted allogeneic HCT and related it to key outcomes. In line with previous pharmacodynamic analyses of rATG in T-cell–replete and T-cell–depleted settings, they found an optimal rATG exposure in which patients with a pre-HCT exposure of >50 AU × d/L and post-HCT exposure of <12 AU × d/L had the best outcomes.^2,3 Their study had additional observations that are worth highlighting: (1) nonnegligible rejection rates after chemotherapy-based conditioning; (2) low rates of chronic graft-versus-host disease (GVHD) associated with high pre-HCT rATG exposure; and (3) HCT with sibling haploidentical donors was associated with worse survival.¹ What can we learn from retrospective model-based simulations such as Dvorak et al’s study, and most importantly, how can this help us design future trials?

Among all patients, the day 100 rejection rate was 9.3%, most notably for patients who received various chemotherapy-based conditioning regimens, for whom the rates were ∼15% (for total body irradiation-based, it was only 1.5%). For a malignant indication, this is a remarkably high rate. Interestingly, the authors did not relate the rATG exposure to rejection but correlated the absolute lymphocyte count (ALC) with rejection. They found that those with a pre-rATG ALC >1.2 were more likely to experience graft rejection. It should be noted that the start date of rATG was not consistent in their study, which predicted exposure and may have had an impact on rejection rates. For these reasons, we question why the authors did not specifically evaluate the association of rATG exposure and rejection rates, which would have been of interest. These high rejection rates are distinct from what Lakkaraja et al described in a large (N = 554) ex vivo CD34⁺ T-cell depletion cohort, in which total body irradiation- and chemotherapy-based conditioning regimens were both associated with very low rejection rates (∼1%-2%).³ The difference in this cohort was that patients receiving chemotherapy-based conditioning largely received a standard regimen containing busulfan, melphalan, and fludarabine, and the majority received fully matched donors. Factors other than rATG may have played a role as well, such as variability in busulfan and melphalan exposures in a multicenter setting.⁴ For busulfan, it was shown in large pediatric/young adult analyses that lower exposure using a population PK model was associated with higher rejection rates (and relapse).⁵ Furthermore, lower fludarabine exposure may have had an impact as shown in prior studies.⁶ For future analyses, it will be important to consider the PK exposure of all these drugs.

We concur with the authors that higher exposure pre-HCT rATG exposure may give deeper depletion of recipient antigen-presenting cells, resulting in less antigen presentation available for GVHD. Furthermore, timely CD4⁺ reconstitution may have played a role. Understanding that their study involves a multicenter cohort, collection of immune reconstitution data may be challenging. For the proposed prospective validation trial, we would strongly encourage the investigators to include mandatory immune monitoring parameters at consistent and frequent time points after HCT. Numerous studies, both in the T-replete and T-depleted settings, have observed that attaining CD4⁺ cells >50/μL within 100 days is an excellent predictor of superior outcomes including survival.^7-9 Limiting post-HCT rATG exposure through model-based dosing will ensure better and more predictable immune reconstitution. Intriguingly, Lakkaraja et al showed that high post-HCT rATG exposure (as independent predictor) was associated with higher GVHD rates and speculated that this was most likely due to lack of adequate immune reconstitution, resulting in more infections and more tissue injury (and more antigen presentation), which may trigger GVHD.³ 

Dvorak et al appropriately propose a prospective validation trial in the α/β T-cell–depleted allogeneic HCT setting.¹ We and others have previously shown that model-based rATG dosing (PARACHUTE trial [NL4836] and PRAISE-IR [NCT04872595]) is feasible and affects outcomes.^2,10,11 Moreover, model-based rATG dosing was associated with attaining CD4⁺ reconstitution (>50/μL) in most of the patients, whereas previous weight-based dosing results in 30% to 50% of patients attaining that value.^2,10,11 With model-based rATG dosing, the range in rATG dosing is a factor 5 ranging from 2 to 10 mg/kg, depending on bodyweight and ALC. This nicely illustrates that with ALC and weight (up to 40 kg only) as predictors for clearance, in older patients (>10 years) with a low ALC before dosing, high post-HCT rATG exposure is a serious risk with potentially life-threatening consequences. This may also explain, as mentioned by the authors, why the patients receiving a haplo sibling, who were older, were overexposed with rATG, resulting in higher nonrelapse mortality and relapse. However, it cannot be excluded that parental haplo donors exert an augmented antileukemic effect, with age being a predictor for GVHD.¹ 

The analysis from Dvorak et al is yet another excellent example that we can learn from retrospectively simulating drug exposures, in either previously conducted trials or in well-defined retrospective cohorts. It adds to the growing data validating improved HCT outcomes when model-based rATG dosing is applied in clinical practice as seen in the prospective PARACHUTE and PRAISE-IR trials. Furthermore, in large “real-world” data analyses (on different continents), these results have consistently held.^10,12 Similarly, for busulfan, a frequently used myeloablative agent in HCT, 2 large retrospective analyses in pediatric and young adult patients identified an optimal PK exposure.^5,13 These results were the basis for guidelines (European Society for Blood and Marrow Transplantation handbook 2024) and prospective clinical trials. Similarly, in other cellular therapy settings, such as CD19 chimeric antigen regimen T-cell cohorts, immune effector cell simulation of fludarabine observed an optimal drug exposure for improved survival.^14,15 An international, randomized controlled clinical trial (INFLUENCE) comparing model-based fludarabine dosing vs standard body-surface area dosing will start later this year in the United States and Europe.

In summary, the authors once again confirm that weight-based rATG dosing is suboptimal, and model-based rATG dosing (based on weight and ALC-based) should be routinely used. Their study demonstrates that using validated population PK models can help us better understand historical trials and retrospective cohort results. In some respect, these studies challenge us to rethink the way we administer our preparative regimens, allowing for us to further advance the field by developing individualized, evidence-based conditioning regimens that are simple, inexpensive to enact, and easily modifiable. Future analyses should probably not focus on single drugs but combine all the currently validated population PK models available, including busulfan, fludarabine, melphalan, rATG, and clofarabine, etc. This will help us to rationally design future prospective validation trials.

Conflict-of-interest disclosure: J.J.B. has received compensation for consulting from Sanofi, Sobi, Merck, and SmartImmune; is Data Monitoring Committee chair or member with honoraria from CTI and Advanced Clinical; and has received research funding from Sanofi. M.S. served as a paid consultant for McKinsey & Company, Angiocrine Bioscience, and Omeros; received research funding from Angiocrine Bioscience, Omeros, and Amgen; served on ad hoc advisory boards for Kite Pharma and Miltenyi Biotec; received honoraria from i3Health, Medscape, and Cancer Network for continuing medical education–related activity; and also declares honoraria from IDEOlogy.