Can literature models describe the neutropenic time course produced by hyper-CVAD chemotherapy in non-Hodgkin lymphoma?

Background: Cytotoxic anticancer drugs methotrexate and cytarabine are incorporated into various high-dose multi-agent chemotherapy regimens (e.g. hyper-CVAD) for aggressive non-Hodgkin lymphoma (NHL). Methotrexate and cytarabine have narrow therapeutic windows and display large inter-patient variability on pharmacokinetic (PK) parameters thus creating difficulty in standardising doses within a population. Studies have shown that neutropenic events after chemotherapy result not only in toxic effects but may also improve long-term survival outcomes [1,2]. Chemotherapy-associated neutropenia is a useful pharmacodynamic (PD) marker for drug exposure. A semi-mechanistic myelosuppression model describing neutrophil time-course following anticancer drugs [3] has been published. However, it was developed from solid tumour data, it remains invalidated in haematological cancers, and it does not incorporate exogenous granulocyte colony stimulating factor (GCSF) as commonly used within high-dose chemotherapy NHL regimens (e.g. hyper-CVAD).

Objective: To simulate from literature sourced PKPD models and parameter values the impact of the B-cycle of hyper-CVAD chemotherapy with GCSF support on the neutrophil count in NHL patients.

Methods: Berkeley-Madonna software was used to develop the models using differential equations. Population PK parameters for high dose intravenous methotrexate and cytarabine were derived from the literature to simulate the concentration-time profile under the hyper-CVAD regimen. The semi-mechanistic myelosuppression model was combined with these literature PK values to describe the time course profile of the neutrophils. A previously published PK model for filgrastim was added to account for the effect of recombinant GCSF of circulating neutrophils. The simulated time course of neutrophils after the B-cycle of hyper-CVAD was adjusted against the observed neutrophil values of a sample population of NHL patients receiving this chemotherapy (n = 50).

Results: The final model comprised 13 differential equations. The neutrophil time course profile of the observed versus literature simulated followed a similarly steep drop in neutrophil counts and duration of nadir. However, the literature produced neutrophil profile simulated a nadir occurring 100 hours earlier than that observed. Adjustment using the parameter slider function in Berkeley-Madonna indicated a smaller transit rate constant for proliferating to circulating neutrophils (faster mean transit time between granulopoesis compartments) than reported in the literature giving a nadir at approximately 150-200 hours following the start of chemotherapy which was more consistent with the observations. The addition of GCSF into the simulated model improved the comparison further.

Conclusion: A constructed model based on literature PKPD models and parameters to simulate myelosuppression occurring after hyper-CVAD treatment can resemble observed neutropenia profiles. Furthermore, it provided greater understanding of the limitations of using literature models to compensate for unobtainable observations or incomplete data in the complex system of NHL cancer treatment and its neutrophil effects.

1. Lyman GH. Impact of chemotherapy dose intensity on cancer patient outcomes. JNCI J Natl Cancer Inst. 2009;7(1):99-108.
2. Mayers C, Panzarella T, Tannock IF. Analysis of the prognostic effects of inclusion in a clinical trial and of myelosuppression on survival after adjuvant chemotherapy for breast carcinoma. Cancer. 2001;91(12):2246-57.
3. Friberg LE, Karlsson MO. Mechanistic models for myelosuppression. Invest New Drugs. 2003;21:183-94.

Geeta Sandhu

  • University of Queensland