Daniel Bending 1, Venkatesh Pilla Reddy
1 Eli Lilly (, )
Introduction/Objectives
Chemotherapy-induced neutropenia remains a significant dose-limiting toxicity, and granulocyte colony-stimulating factor (G-CSF) prophylaxis represents a cornerstone of supportive care optimisation. Semi-mechanistic models of myelosuppression, exemplified by the Friberg transit compartment framework, have proven valuable for characterising neutrophil dynamics following conventional cytotoxic chemotherapy and have been successfully extended to incorporate G-CSF effects¹². However, antibody-drug conjugates (ADCs) present a different pharmacological challenge: the sustained release of cytotoxic payload following internalisation and lysosomal processing produces an exposure-response relationship that differs from the short, high-peak exposures characteristic of conventional chemotherapy³. While Friberg-type models have been successfully applied to ADC-induced myelosuppression in monotherapy settings⁴⁵, their extension to combination regimens incorporating both ADCs and conventional chemotherapy, with or without G-CSF support, has not been systematically evaluated. The present work investigated the feasibility of combining established literature models for ADC myelosuppression, chemotherapy-induced neutropenia, and G-CSF pharmacodynamics into a unified framework capable of informing G-CSF dosing optimisation in ADC-chemotherapy combinations⁶⁷.
Methods
Published semi-mechanistic models were identified from the literature for three components: ADC-induced myelosuppression driven by unconjugated payload exposure⁴⁵, carboplatin-induced neutropenia with G-CSF support², and standalone G-CSF pharmacodynamics¹. Model integration was attempted through several approaches: additive drug effects on the proliferating compartment, sequential parameter substitution, and hierarchical model linking. Simulations were conducted to evaluate model behaviour under combination scenarios over multiple treatment cycles, comparing predicted neutrophil dynamics for carboplatin plus ADC with and without G-CSF prophylaxis. Structural adequacy was assessed by examining the physiological plausibility of predicted nadir depth, nadir timing, and recovery dynamics. Sensitivity analysis was performed on the feedback-related parameters to characterise the dependence of model behaviour on these empirically derived values when extrapolated beyond their original estimation context.
Results
Integration of published literature models revealed fundamental structural incompatibilities when combining ADC and chemotherapy drug effects within Friberg-type frameworks. While individual models adequately described their respective monotherapy scenarios, the combined model structure produced non-physiological predictions under multi-cycle simulation conditions. When ADC effects were incorporated into an established carboplatin-G-CSF model, simulations produced a counterintuitive result: carboplatin plus ADC without G-CSF yielded higher neutrophil nadirs than carboplatin plus ADC with G-CSF support. This divergence became increasingly pronounced over repeated treatment cycles; by cycle ten, predicted nadirs in the G-CSF-containing regimen were greater than 10⁷-fold lower than those without G-CSF, a phenomenon termed “colony collapse.” This arises from the interaction between G-CSF effects on maturation and the absence of an explicit stem cell compartment. G-CSF increases the maturation rate out of the proliferating compartment, causing an initial reduction in proliferative cell mass. In Friberg-type models, recovery depends entirely on the endogenous feedback mechanism acting on the remaining proliferative pool; however, when this pool is sufficiently depleted by the combined cytotoxic and maturation effects, insufficient proliferative capacity remains to generate meaningful feedback-driven regeneration. Sensitivity analysis confirmed that model behaviour was highly dependent on the feedback parameters, which lacked mechanistic grounding outside their original estimation context.
Conclusions
The integration of established Friberg-type literature models for ADC myelosuppression, chemotherapy-induced neutropenia, and G-CSF pharmacodynamics into a unified framework suitable for G-CSF dosing optimisation appears limited by structural assumptions inherent to the semi-mechanistic approach. The absence of an explicit stem cell compartment means that recovery depends entirely on feedback acting on the proliferative pool, which becomes insufficient when depleted below a critical threshold by combined cytotoxic and G-CSF-mediated maturation effects. Future efforts toward G-CSF dosing optimisation in ADC-containing regimens will likely require mechanistic granulopoiesis models that explicitly represent haematopoietic stem cell dynamics⁶, providing an exogenous source of proliferative capacity independent of the feedback mechanism.
References:
1. Friberg LE et al. J Clin Oncol. 2002;20(24):4713-4721. doi:10.1200/JCO.2002.02.060
2. Quartino AL et al. Cancer Chemother Pharmacol. 2014;73(2):319-331. doi:10.1007/s00280-013-2359-4
3. Liao MZ et al. Xenobiotica. 2024;54(8):543-551. doi:10.1080/00498254.2024.2351044
4. Fostvedt LK et al. Clin Pharmacol Ther. 2019;106(5):1006-1017. doi:10.1002/cpt.1500
5. Ait-Oudhia S et al. AAPS J. 2017;19(5):1436-1448. doi:10.1208/s12248-017-0113-5
6. Craig M et al. J Theor Biol. 2016;399:115-133. doi:10.1016/j.jtbi.2016.03.034
7. Henrich A et al. J Pharmacokinet Pharmacodyn. 2017;44(4):285-302. doi:10.1007/s10928-017-9515-4
Reference: PAGE 34 (2026) Abstr 12229 [www.page-meeting.org/?abstract=12229]
Poster: Drug/Disease Modelling - Oncology