Using M&S to Evaluate Oncology Drug Dosing

PK sampling following single- and multiple-dose administration in early development—first-in-human (FIH) studies and late-stage Phase 2-3 studies—are critical in characterizing the PK of the drug and will help in the evaluation of exposure-response relationships across dose levels.

The objective of PK sampling is to understand drug exposure over a range of dose levels and across the duration of dosing. For a small molecule, this can often be done with a few days of sampling at various times over the drug administration period. The samples are frequently close together. For large molecules, like monoclonal antibodies, the duration of PK sample collection stretches over multiple weeks, although the samples may be collected several days apart.

Characterizing the area under the time-concentration curve, drug clearance, elimination half-life, peak/trough concentrations, and other rudimentary PK parameters will provide more informed decision making during the dose escalation process and a better understanding of the exposure-response profile of the drug. Once this is understood, less frequent (sparse) sampling can be employed in later studies. These are analyzed using PopPK modeling from late-stage Phase 2-3 studies, using broader patient populations, allowing investigation of the effects of intrinsic and extrinsic variables of the drug’s PK and link to safety and efficacy biomarkers.

Data from the PK sampling helps to determine the correct dose and factors that might impact it (body weight, age, sex etc.). It also helps to determine the conditions of dosing (e.g. with food or on an empty stomach).

In addition to PK sampling, PK/PD relationships and molecular markers of response can be explored by measuring PD markers in clinical trials. Although molecular markers of response are rarely directly correlated to clinical response, understanding PK/PD relationships can help to support dose justifications, validate mechanisms of action, and provide proof of concept for an investigative drug. They may assist an investigator in identifying patients more likely to benefit from a particular drug. An example of where evaluating PD markers were used for dose justification is highlighted in a study on BCR-ABL small molecule kinase inhibitors that were approved for treatment of chronic myeloid leukemia. The investigators found a PK/PD correlation that predicted a better response. Patients with more time above the IC50 CD34+cells had a better prognosis than patients with shorter times above the IC50. The study data indicated that the PK/PD is correlated with molecular response at 3 months, which suggests that assessment of PD markers,² included in larger clinical trials, can bolster dose justification arguments. This is particularly useful in cases where the dose- or exposure-response relationships can be distinguished.

Exposure-response analysis—which relies on PK sampling across a broad range of doses —is a valuable tool to identify a range of concentrations associated with increasing responses or increasing toxicities. Understanding the impact of drug exposure, rather than administered drug dose on response and toxicity, can aid drug developers and regulators to develop dose adjustments in cases of altered PK.

In one example, both physiologically-based pharmacokinetics (PBPK) and PK/PD M&S were conducted to predict exposure alterations for one cancer drug. Ibrutinib is a cytochrome P450 3A4 (Cyp3A) substrate. The impact of inhibiting CYP3A on Ibrutinb exposure was predicted in using a PBPK model. Ibrutinib administered with ketoconazole (a strong CYP3A4 inhibitor) had decreased Ibrutinib metabolism, resulting in increased Ibrutinib drug exposure. The impact of co-administration with rifampin, (a strong CYP3A4 inducer) resulted in decreased Ibrutinib exposure. The results, based on the use of PK/PD modeling and known exposure-response relationships of Ibrutinib in the presence of moderate and weak CYP3A4 inhibitors and inducers, were used to determine dose appropriate modifications for the concomitant use of Ibrutinib with CYP3A4 modulators without the need for additional dedicated clinical studies The combination of well-characterized exposure-response with robust dose ranging and M&S can support optimized precision dosing for labeling.³

The decision to use either flat-fixed or body weight-based dosing depends on the effect of body weight on a drug’s clearance, the volume of distribution, the drug’s PD, and its therapeutic margin for safety and efficacy. An effective dosing strategy reduces interpatient PK variability and should maximize therapeutic outcomes.

Across therapeutic areas, the majority of drugs are given as fixed doses in their final, commercial forms. Most begin initial development as body-weight adjusted doses, but as exposure data are accrued, fixed dose administration becomes supportable. Modeling may support that body weight is not a significant covariate impacting drug exposure and does not need to be considered in dose selection. This makes it possible to deliver doses as fixed amounts. Simulations may demonstrate that it is feasible to attain drug concentrations within an efficacious exposure range, and to exclude exposures associated with toxicity across a range of body weights, allowing fixed dose to be determined.

Overall, the dosing strategy for Phase 3 trials should be determined based on assessing body size effect on PK/PD after the drug’s therapeutic window is established.

For orally-administered drugs, dosing in a fed or fasted state, or concurrent administration of acid reducing agents can affect the absorption of a drug (bioavailability) and hence impacts both the safety and efficacy. Having an early understanding of factors that change drug absorption can define the right dosing conditions and decrease variability of the drug effects (both positive and negative).

The real impact of modeling and simulation is best illustrated through real-world case studies. The following examples highlight how M&S approaches are applied in practice.

Antibody–drug conjugates pose unique challenges due to their complex structure and narrow therapeutic index. A QSP platform model developed with HER2-targeted ADCs (T-DM1 and Enhertu) successfully predicted clinical pharmacokinetics, tumor payload distribution, and hematological toxicities. Virtual clinical trial simulations replicated observed response differences between HER2-high and HER2-low populations, enabling researchers to identify the most sensitive parameters and anticipate therapeutic index limits. This work demonstrated how ADC models can integrate preclinical data to inform safe and efficacious dosing strategies in humans.

For bispecific antibodies, such as ATG-101 (a PD-L1/4-1BB bispecific), mechanistic modeling provided insights into the bell-shaped dose-response relationship often seen with receptor crosslinking agents. The model was used to predict the “sweet spot” where both trimer formation and checkpoint blockade were maximized, guiding a first-in-human dose and differentiating ATG-101 from competitor molecules. Similarly, multi-specific T-cell engagers were modeled to establish MABEL-based starting doses by linking in vitro trimer formation to in vivo efficacy and cytokine release, ensuring a rational and safe clinical entry strategy.

For CAR-T therapies, M&S approaches captured the living-cell kinetics that distinguish these products from traditional drugs. Mechanistic PK/PD models reproduced the hallmark expansion–contraction–persistence dynamics of CAR-Ts in lymphoma patients, while sensitivity analyses highlighted tumor characteristics (e.g., antigen density, division time) that could strongly influence outcomes.

Building on this, Certara’s QSP software, Certara IQ™, for multiple myeloma CAR-T therapies integrated clinical data from anti-BCMA and anti-GPRC5D trials, enabling virtual patient simulations to explore dosing strategies, sequence of combination regimens, and the potential of bi-specific CAR-T constructs.

For more information, watch the on-demand webinar or read the full publication.

Mechanistic PK/PD modeling has also supported dose justification for checkpoint inhibitors. For cosibelimab (CK-301), simulations compared tumor receptor occupancy against approved PD-L1 inhibitors, demonstrating that dosing at 800–1200 mg every two or three weeks could achieve >99% PD-1 receptor occupancy. This gave regulators and developers confidence that cosibelimab could deliver efficacy comparable to established therapies.

Novel modalities such as molecular glue degraders (MGDs) also benefit from modeling. In vitro and in silico models of lenalidomide and pomalidomide degradation pathways accurately reproduced degradation kinetics of transcription factors Ikaros and Aiolos. These tools allowed researchers to scan binding affinities, turnover rates, and ligase expression to identify tunable properties for optimizing degraders while avoiding off-target effects.

M&S methods like QSP modeling have been instrumental for small molecule oncology drugs. In collaboration with a large pharma company, a QSP model was developed for their molecule, a KRAS-G12C inhibitor, to assess performance against a leading competitor in the market. The QSP model successfully predicted superior tumor responses compared with the leading competitor, despite lower systemic exposure, due to its higher binding affinity. Virtual clinical trials supported confidence in advancing our partner’s molecule to the clinic as a potential best-in-class compound. Similarly, PK/PD and PK/TGI models have been widely applied to small molecules, scaling preclinical tumor inhibition data to humans, and helping identify safe and efficacious first-in-human dosing strategies.

Finally, QSP modeling has extended to oncolytic viruses, which selectively replicate in tumors while stimulating systemic immune responses. For a myxoma virus–based platform, models predicted systemic cytokine exposure following intravenous administration, incorporating virus kinetics, transgene-driven cytokine production, and secondary immune responses. Simulations indicated that even at high doses, predicted cytokine levels would remain within known safety limits, providing confidence to proceed with planned clinical dosing. This approach illustrates how modeling can bridge preclinical and translational data in an emerging therapeutic class.

The US FDA Prescription Drug User Fee Act (PDUFA) for fiscal years 2018–2022 reflects the agency’s goals to expedite bringing safer therapies to patients. It also reflects and incorporates the advances in regulatory science and decision-support tools, such as M&S, to support drug development and decision-making.

M&S tools have demonstrated their importance in dose selection, especially in special populations, dose optimization and dosing regimen, characterizing exposure-response, predicting drug-drug interactions, and more. As M&S tools are increasingly used in clinical studies, its importance will be reinforced in advancing drug development.

Our modeling and simulation services can help you transform drug development with the power to predict and make data-driven decisions, even when data is sparse or absent. Find out more about our modeling and simulation services.