Despite advances in chemotherapy and novel cellular therapies, allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains a well-established curative therapy for defined subsets of hematologic malignancies (1, 2). In its initial historical application decades ago, allo-HSCT was associated with substantial risks, including graft-versus-host disease (GVHD) and graft failure (GF) (3, 4). Multiple randomized trials demonstrated that anti-thymocyte globulin (ATG) could reduce the incidence of both severe acute (aGVHD) and chronic GVHD (cGVHD) post-transplant (5–7), while some randomized studies reported no significant improvement in cGVHD-free survival in specific cohorts with ATG (8, 9). Collectively, these findings provide a compelling rationale for the incorporation of ATG into allo-HSCT to prevent GVHD.

ATG is a polyclonal antibody that could deplete a variety of immune cells, while the primary mechanism of GVHD prophylaxis is T-cell depletion (10). Historically, three main types of ATG products have been available for clinical use. The first ATG formulation was horse-derived ATG (ATGAM^®, Pfizer, USA) (11). ATGAM^® is not typically used for the indication of allo-HSCT, as two prospective trials failed to demonstrate its efficacy in prophylaxis of aGVHD (12, 13). The other two ATGs, Thymoglobulin^® (ATG-T, Sanofi, France) and Grafalon^® (formerly known as ATG-Fresenius, ATG-F, Neovii, Germany), are both derived from rabbits. Although most of these products are commercially available, ATG-T remains the most commonly used ATG preparation in clinical practice (14, 15). Consequently, this review will focus on the investigations into optimizing the dosage of ATG-T. It is important to note that there is no universally accepted bioequivalent dosing between ATG-T and ATG-F, special caution should be exercised when switching between the two ATG preparations in clinical practice (16–18).

Pharmacological studies of ATG found that the immunological effects of ATG are critically influenced by its concentration (19–26). Therefore, optimizing ATG dose in allo-HSCT to maximize its GVHD prophylaxis effect and minimize its potential side effects is crucial for improving transplant outcomes (14, 27, 28). Early studies explored the optimal ATG dose using body weight-adjusted dosing strategy (6, 8, 29). However, due to ATG pharmacokinetics being influenced by body weight of recipients, lymphocyte count and timing of ATG administration, the inter-individual heterogeneity is considerable (30–32). As such, the optimal ATG dose in allo-HSCT has not yet been determined. Given the ATG pharmacokinetic heterogeneity among transplant recipients, individualized ATG dosing may be a potential solution and has garnered significant research interest. Recent pharmacological studies have found that optimal ATG exposure is associated with lower incidence of GVHD and virus reactivation, and may even lead to improved non-relapse mortality (NRM) and overall survival (OS) (33–35). Importantly, achieving optimal ATG exposure through individualized dosing can reduce adverse events in allo-HSCT and improve health-related quality of life (19, 36, 37).

This review aims to provide a comprehensive summary of the advances of individualized ATG dosing in allo-HSCT and its effect on transplant outcomes.

ATG-T is a heterologous polyclonal immunoglobulin G (IgG) that targets over 40 antigens (14, 15). These antigens are classified into two categories based on their biological function: immune cell response antigens and adhesion/cell-trafficking molecules ( Figure 1 ) (14, 15, 38, 39). ATG-T mediates its immunomodulatory effects primarily by targeting T cells and other immune effector cells. It targets key T-cell antigens, including CD2, CD3, CD4, CD5, CD6, CD8, CD28 and HLA class I molecules, leading to T-cell depletion via complement-dependent lysis and T-cell activation-induced apoptosis (38, 40). Additionally, ATG-T also contains antibodies against B-cell surface proteins CD5, CD19, CD20, CD30, CD38, CD40, CD80, CD95, CD138 and HLA-DR, triggering caspase- and cathepsin-dependent B cell apoptosis (38, 40, 41). Furthermore, ATG-T could inhibit dendritic cell (DC) maturation and migration by targeting CD1a, CD4, CD11a, CD11b, CD29, CD32, CD51/61, CD86, MHC I and MHC II (38, 42). In vitro studies have demonstrated its capacity to expand CD4+ CD25+ regulatory T cells (Tregs) by targeting CTLA-4, FOXP3, GITR (43, 44). Finally, ATG modulates leukocyte-endothelial interactions by targeting integrins (VLA-4, LPAM-1), chemokine receptors (CXCR4, CCR5, CCR7), and leukocyte adhesion molecules (ICAM-1, ICAM-2, ICAM-3), thereby disrupting leukocyte adhesion to endothelia (38, 45).

It is important to note that the immunomodulatory effects of ATG depended critically on its concentration. Specifically, a low dose of ATG (e.g., 1 mg/kg) is sufficient to induce antibody-dependent cell-mediated cytotoxicity (ADCC) against activated T cells in blood circulation. However, this ATG concentration is inadequate for depleting lymphocytes (T cells, B cells and NK cells) and antigen-presenting cells residing within secondary lymphoid tissues (22, 46). Additionally, B cells (CD20+) and NK cells (CD16+/CD56+) may only be affected at higher doses (> 5mg/kg) of ATG-T (22). Lower-dose ATG selectively depleted activated T cells while preserving the function of B and NK cells, thereby mitigating systemic immunosuppression (14). Although the effects of ATG are dose-dependent on various cell types, it needs special caution to adjust the dose of ATG for individuals to improve the efficacy of HSCT.

The concentration of ATG, often labeled on the vial, generally refers to the total ATG. Total ATG levels in patient samples could be quantified by enzyme-linked immunosorbent assay (ELISA) (47, 48). The component capable of binding to human lymphocytes was defined as active ATG. Despite comprising only 10% of total ATG, active ATG significantly affects aGVHD, immune reconstitution and post-transplant lymphoproliferative disorder (PTLD) (19–21, 49). The quantification of active ATG remains challenging (50, 51), with flow cytometry being the most widely utilized method for its detection (21, 51). In 2020, liquid chromatography-mass spectrometry (LC-MS) was employed for the first time to quantify the active fraction of ATG in plasma (52). This technique offers superior precision; however, its application remains limited due to restricted accessibility. The establishment of ATG detecting methods enables its pharmacokinetic and pharmacodynamic evaluation in allo-HSCT ( Table 1 ).

The complex immunomodulatory mechanisms of ATG underlies the significant interindividual heterogeneity in its pharmacokinetics (27, 48, 51, 53–55). Waller, et al. (48) reported that the clearance of ATG was relatively slow, and serum total ATG remained detectable up to 90 days post-transplant. The calculated half-life of active and total ATG were 7 days and 14 days, respectively. The study further demonstrated that the time for active ATG levels decreasing to sub-therapeutic levels (1 μg/mL) in the 6 mg/kg group (17 days) was significantly shorter than 10 mg/kg group (45 days; P = 0.002). Similarly, when using 16-20mg/kg ATG-T, the median time for active ATG level to decline to less than 2.0μg/ml was 45.5 days (51). The clearance time of the 16–20 mg/kg ATG group was not significantly longer than that of the 10 mg/kg group, suggesting that a higher dose (> 10 mg/kg) of ATG-T may not be necessary. An Austrian study by Seidel, et al. found that the half-life of ATG-T was consistent when the ATG-T dose within the range of 7.5-20mg/kg, with a linear correlation between the dose and maximum serum concentration (C_max). However, when the ATG-T dose was 30–40 mg/kg, the active fraction of ATG-T accumulated in the body, leading to a sharp increase in C_max and resulting in ATG overexposure (55).

Weight-based ATG dosing induces marked interindividual variability in ATG exposure, arising from recipient-specific and regimen-related determinants ( Figure 2 ). Body weight and absolute lymphocyte count (ALC) constitute principal recipient-specific determinants of ATG clearance. Pharmacokinetic analyses demonstrate that pediatric HSCT recipients with higher body weight and lower ALC exhibit over exposure to active ATG (31, 56).

Both graft source and timing of ATG administration significantly modulate ATG exposure (27, 32). Compared to G-CSF-mobilized peripheral blood stem cells (G-PBSC), bone marrow and cord blood grafts contain fewer memory T cells and more naïve T cells, contributing to delayed post-transplant T-cell reconstitution. This necessitates ATG dose reduction in bone marrow or cord blood HSCT to promote T-cell recovery (32, 57–59). The timing of ATG administration is also important. Early ATG administration (between days -9 and -5) demonstrated reduced ATG exposure and accelerate T-cell reconstitution compared to later administration (between days -5 and 0) (32, 60).

Furthermore, ATG pharmacokinetics differ between preparations. Rabbit-ATG (Thymoglobulin^®) exhibits a longer half-life, with detectable plasma active ATG persisting for one month, whereas active horse-ATG (ATGAM^®) components decline within two weeks (53). As xenogeneic proteins, ATG preparations can induce anti-ATG antibodies. Early antibody formation (before day +22) mediates accelerated ATG clearance, substantially reducing post-transplant exposure (61).

The timing of ATG administration significantly influences allo-HSCT outcomes. Late ATG administration (closer to day 0) more effectively depletes donor T cells in the graft, while its effect on recipient T cells and antigen-presenting cells remain comparable with earlier dosing. Consequently, late ATG administration is often associated with reduced GVHD but carries an increased risk of viral reactivation compared to early dosing (before day -5) (70).

These timing-dependent effects are further supported by clinical studies. In severe aplastic anemia patients undergoing haplo-PBSCT, Wu, et al. demonstrated that shifting ATG dosing from early (days -9 to -7) to late (days -5 to -3) effectively controlled GVHD but led to increased rates of CMV reactivation and EBV-associated post-transplant lymphoproliferative disorder (EBV-PTLD) (71). Conversely, early ATG administration appears to facilitate T-cell reconstitution. Lindemans, et al. observed accelerated reconstitution of CD3+, CD4+, and naïve T cells in cord blood transplant recipients receiving early ATG (days -9 to -5) compared to later ATG (days -5 to 0) (32). Similarly, a Japanese study in adult PBSCT found that early ATG administration (1.25mg/kg on day -4), rather than the standard schedule (1.25 mg/kg, days -2 and -1), reduced post-transplant ATG exposure and accelerated CD4+ T-cell recovery (60). These findings indicate that ATG dosing should be adjusted according to timing of administration. A relatively increased dose may be necessary with early ATG dosing, whereas the dose could be reduced if ATG administered closer to day 0.

ATG exposure was quantified using the area under the concentration-time curve (AUC). Total ATG exposure was divided into pre- and post-transplant exposure using day 0 (graft infusion) as the reference point. ATG exposure better predicts outcomes in allo-HSCT than the concentration at designated timepoints. Several studies have assessed the association between ATG exposure and transplant outcomes, including GVHD, immune reconstitution, relapse, and survival ( Table 4 ) (19, 21, 31, 72–76).

Admiraal, et al. discovered that excessive exposure to active ATG post-transplant significantly decreases the rate of successful immune reconstitution (21). In subsequent study, they revealed that for every 10% increase in the post-transplant AUC of active ATG-T, the likelihood of successful CD4+ T cell immune reconstitution decreased by 26%. Lower post-transplant active ATG exposure (< 16 AU × day/mL) and successful CD4+ immune reconstitution were both associated with improved event-free survival (72). Additionally, they found that pre-transplant active ATG exposure ≥ 40 AU × day/mL significantly reduced the incidence of grade II-IV aGVHD, cGVHD, and graft failure (21). Similarly, Jamani, et al. (73) from Canada discovered that the lowest quintile of pre-transplant AUC and post-transplant AUC of active ATG were associated with higher aGVHD and worse cGVHD- and relapse-free survival (cGRFS) in myeloablative allo-HSCT. A multinational prospective study by Oostenbrink, et al. reported that prolonged ATLG exposure (active ATLG ≥ 1 AU/mL on day +16) significantly reduced the incidence of grade II-IV aGVHD (from 50% to 8.2%) (74). A study by Dabas, et al. from Canada showed that high pre-transplant active ATG exposure of MNC-binding (> 282.36 UE*hr/L) and CD33+ cells- binding (> 60.53 UE*hr/L) were associated with a lower risk of relapse and better RFS. Whereas higher post-transplant exposure of lymphocyte-binding (> 1022.42 UE*hr/L) was associated with higher risk of relapse and lower RFS (75).

These studies highlight the importance of maintaining pre- or post-transplant ATG exposure within an optimal range, as both excessive and insufficient exposure compromise transplant outcomes. A Dutch retrospective analysis identified an optimal post-transplant active ATG exposure of 60–95 AU/mL/day. Sub-optimal exposure (< 60 AU/mL/day) increased grade III-IV aGVHD and NRM, while over-optimal exposure (> 95 AU/mL/day) increased relapse-related mortality (RRM). Only patients within the optimal range achieved the best 5-year event-free survival (EFS) and OS (31). A single-center prospective study from China found an optimal total active ATG exposure of 100 to 148.5 UE/mL/day in haplo-HSCT following Beijing Protocol. Interestingly, the optimal AUC group showed a significantly lower incidence of cytomegalovirus (CMV) reactivation and persistent CMV viremia compared to the non-optimal AUC group (total AUC < 100 or >148.5 UE/mL/day). While no significant difference in NRM and recurrence were observed between two groups, optimal AUC group showed a trend toward better 2-year OS (75.7% vs. 57.8%, P = 0.061) (19). A recent phase IV trial established an optimal post-transplant ATG exposure (55–75 AU/mL/day) for acute leukemia patients undergoing myeloablative haplo-cord HSCT. Compared to non-optimal range, patients within optimal range have lower 2-year relapse (38.9% vs. 15.2%), higher leukemia-free survival (LFS) (49.5% vs. 84.8%), superior OS (50.8% vs. 89.5%) and GRFS (39.1% vs. 71.3%), and reduced grade II-IV aGVHD (37.8% vs. 20.5%) (76).

This review discussed the challenges of optimizing ATG dosing in allo-HSCT to balance GVHD prophylaxis with immune reconstitution, while minimizing malignant disease recurrence and life-threatening infections. Extensive research has focused on weight-based ATG dosing regimens in allo-HSCT, yet this approach remains suboptimal in addressing the pharmacokinetic variability mediated by multiple parameters, including genetic polymorphisms (e.g., HLA compatibility and Fcγ receptor genotypes), timing of ATG, anti-ATG antibody development, and comorbidities. The weight-based ATG dosing approach failed to address the substantial pharmacokinetic variability among patients, thus attempting to establish body weight based optimal ATG dosing regimen will continue to prove futile. Pharmacodynamic studies demonstrated that lower ATG concentration was associated with increased risks of aGVHD and cGVHD, although its impact on TRM and relapse remains unclear (47, 62, 63). Notably, emerging evidence highlights the association between ATG exposure and transplant outcomes such as GVHD incidence, immune reconstitution, relapse, and OS (21, 31, 72, 75). Importantly, an optimal ATG exposure range has been identified, associated with reduced viral reactivation, accelerated immune reconstitution, and improved OS (19, 31).

TDM and pharmacogenomics (PGx) are fundamental approaches for achieving personalized dosing in clinical practice. Advances in understanding ATG-PGx, including drug-metabolizing enzymes, therapeutic targets, and drug transporters, will enable optimized balancing of ATG’s efficacy against treatment-related toxicity. Integrating TDM with PGx in ATG personalized dosing represents a promising strategy to improve outcomes of allo-HSCT. Recent studies have shown promising outcomes using individualized ATG dosing strategies based on ALC or TDM (36, 37). However, current individualized ATG dosing protocols are often derived from physiologically based pharmacokinetic (PBPK) and population pharmacokinetic (popPK) models (21, 56). These protocols exhibit inherent static limitations of failing to integrate real-time patient data and dynamic health trends. Model-informed precision dosing (MIPD) provides a potential solution for optimizing ATG dosing via mathematical modeling that integrates multidimensional data, including patient characteristics, drug properties, and disease status. Collaboration across clinicians, informaticians, clinical pharmacologists, and TDM specialists will establish ethical framework for data sharing, technology accessibility, and patient privacy, thereby facilitating clinical implementation of MIPD. Artificial intelligence (AI) and machine learning (ML) represent emerging tools for advancing MIPD in personalized medicine, but their clinical application remains experimental with unproven benefits for patient care (89–91). Robust clinical validation and technological innovation are essential to overcome inherent challenges, including data privacy and algorithmic bias, thereby enabling tangible patient benefits and facilitating clinical implementation (92). In conclusion, the ongoing development and optimization of individualized ATG dosing strategies are critical for enhancing the safety and efficacy of allo-HSCT, ultimately improving transplant outcomes.