Optimizing monitoring strategies of trastuzumab induced cardiotoxicity: Development and application of a population pharmacodynamic model quantifying trastuzumab induced changes in cardiac function
J. G. Coen van Hasselt (1,2,3), Annelies H. Boekhout (1,3), Jos H. Beijnen (2), Jan H.M. Schellens (1), Alwin D. R. Huitema (2)
(1)Dep. of Clinical Pharmacology, The Netherlands Cancer Institute, Amsterdam, The Netherlands; (2)Dep. of Pharmacy and Pharmacology, Slotervaart Hospital/Netherlands Cancer Institute, Amsterdam, The Netherlands; (3)these authors contributed equally.
Trastuzumab is a humanized monoclonal antibody that selectively binds to the extracellular domain of HER2, and improves outcome in early and advanced HER2+ breast cancer [1-3]. Adjuvant trastuzumab treatment of HER2+ breast cancer consists of 3-weekly dosing during 1 year.
Trastuzumab treatment is associated with cardiac dysfunction mostly manifested by a decline in left ventricular ejection fraction (LVEF) [4,5]. Moreover, prior anthracycline therapy increases the risk for cardiac dysfunction. Therefore, cardiac function is monitored throughout trastuzumab treatment by repeated measurements of the LVEF. There is substantial uncertainty regarding the optimal clinical management of trastuzumab induced cardiotoxicity, for instance regarding the optimal interval of LVEF monitoring, the optimal time of recovery after cardiac toxicity, and the feasibility of patient-tailored monitoring strategies. In order to answer these clinically important questions it is essential to be able to make quantitative inferences on the sensitivity to, and recovery from trastuzumab-induced cardiotoxicity, the relationship with important determinants such as anthracycline therapy, and the magnitude of between-subject variability.
- To develop a population PD model for trastuzumab induced cardiotoxicity in terms of the LVEF, and to identify potential predictors of inter-patient variability on pharmacodynamic parameters.
- To develop and apply a simulation framework that simulates the LVEF-time profiles, incorporating treatment interventions, to quantitatively assess optimal cardiac monitoring strategies during adjuvant treatment of HER2+ breast cancer.
A) Model development
An unselected representative cohort of HER2+ primary and advanced breast cancer patients treated with trastuzumab, with and without prior anthracycline treatment, was included in this analysis. The data analysis was conducted using NONMEM. PK was described using dosing history, together with a previously published PK model. For the structural model, different effect compartment models were considered. The L2-method in NONMEM was used to simultaneously analyze LVEF values that were partially available as duplicate measurements.
Model predictions were evaluated using goodness-of-fit diagnostics and normalized prediction distribution errors (NPDE). Parameter precision was assessed using a non-parametric bootstrap.
B) Simulation framework
Simulation framework development
A simulation framework was developed to quantitatively assess the safety and efficacy of cardiac monitoring protocols during adjuvant trastuzumab treatment. The simulation framework comprised the following steps:
- Simulate individual LVEF-time profiles using the developed PK/PD model.
- Apply treatment pausing or termination, according a cardiac monitoring protocol.
- Calculate outcome measures to assess the performance of the cardiac management protocol.
The outcome measure calculated included dose intensity, deviation from true time of a cardiac event (CE), percentage of patients with a CE, severities of experienced CE's, and specificity/sensitivity and type I/II errors of the screening, where a CE is defined as as a drop in LVEF > 10%, with an absolute LVEF values <50% according to the summary of product characteristics (SPC). Dose intensity was used as the measure for therapeutic efficacy.
Currently, the cardiac safety of trastuzumab during adjuvant treatment is assessed according to the monitoring protocol defined in the SPC. Prior to treatment, a baseline LVEF is measured, and consequently the LVEF is monitored at a 3-monthly interval. If a patient experiences a CE, treatment should be paused, and the LVEF should be re-evaluated after 3 weeks. If LVEF has decreased again, treatment is terminated.
Simulation framework application
The following simulation scenarios were evaluated to assess the impact of changing different parts of the current SPC-defined cardiac monitoring protocol:
- Baseline scenario: Cardiac monitoring protocol in current clinical practice (according to SPC).
- Optimization of frequency of LVEF monitoring.
- Feasibility of adaptive monitoring frequency based on patient characteristics.
- Optimization of time of recovery before re-initiation of trastuzumab.
The outcome of this simulation study is used to propose optimization opportunities for cardiac monitoring protocols.
A) Model development
A total of 1651 LVEF values from 240 patients were available. The data were best described by an effect-compartment model with recovery, in conjunction with an Emax model.
The population recovery half-life after trastuzumab treatment (T1/2rec) was estimated at 49.7 days. A full variance-covariance matrix for between-subject variability (BSV) could be estimated. BSV on T1/2rec and EC50 were high, with 79.4 CV% and 103 CV% respectively. The cumulative anthracycline dose was a significant determinant of the EC50, causing a 45.9% increase in sensitivity (EC50) at the maximum cumulative anthracycline dose. Goodness-of-fit plots, the NPDE, and the bootstrap analysis indicated adequate performance of the model.
B) Simulation framework
The baseline scenario, according to the SPC, indicated that 10% of patients treated prior with the maximum dose of anthracyclines have an expected dosing intensity (DI) of <83%, while patients without prior anthracycline therapy have a higher DI (DI<94% for 5% of patients).
The sensitivity and specificity for detection of a true CE in anthracycline naive patients were 82% and 98% respectively, with a percentage of false-positive LVEF measurements of 54%, and 8% false-negative measurements.
Increasing the monitoring frequency clearly decreases the frequency of extreme absolute values in LVEF (<35%). Moreover, sensitivity increases up to 94% for 1-monthly monitoring, while specificity drops to 91%.
Implementation of adaptive monitoring strategies based on i) the pre-treatment LVEF, ii) the magnitude of LVEF decline at the first 3 months evaluation, and iii) the prior cumulative anthracycline dose, seems to be feasible. The risk of CE's given these three variables was determined. For instance, anthracycline naïve patients with a baseline LVEF of >65%, drop in LVEF at 3 months <13%, have a CE risk of < 0.47%. The risks were also visualized using a response surface.
A population PK-PD model describing the exposure-response relationship for trastuzumab induced cardiotoxicity in a representative patient cohort was developed. The cumulative anthracycline dose was a significant determinant for between-subject variability on EC50.
Subsequently, an adaptive simulation-framework was successfully applied to investigate optimal cardiac monitoring strategies.
Further work will focus on prospective validation of optimized monitoring protocols in a clinical study.
. E.H. Romond et al. "Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. The New England journal of medicine, vol. 353, Oct. 2005, pp. 1673-84.
. D.J. Slamon et al, "Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2," The New England journal of medicine, vol. 344, 2001, pp. 783-792.
. I. Smith et al. "2-year follow-up of trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer: a randomised controlled trial.," Lancet, vol. 369, 2007, pp. 29-36.
. L. Gianni et al. "Role of anthracyclines in the treatment of early breast cancer," J Clin Oncol, vol. 27, Oct. 2009, pp. 4798-4808.
. E. Tan-Chiu et al. "Assessment of cardiac dysfunction in a randomized trial comparing doxorubicin and cyclophosphamide followed by paclitaxel, with or without trastuzumab as adjuvant therapy in node-positive, human epidermal growth factor receptor 2-overexpressing breast can," J Clin Oncol, vol. 23, Nov. 2005, pp. 7811-7819.
. M.C. Pinder et al. "Congestive heart failure in older women treated with adjuvant anthracycline chemotherapy for breast cancer.," Journal of clinical oncology : official journal of the American Society of Clinical Oncology, vol. 25, Sep. 2007, pp. 3808-15.
. S.L. Beal et al. "NONMEM user guides," 1988.
. R. Bruno et al. "Population pharmacokinetics of trastuzumab in patients with HER2+ metastatic breast cancer," Cancer Chemotherapy Pharmacology, vol. 56, 2005, pp. 361-369.
. M. Karlsson et al. "Three new residual error models for population PK/PD analyses," J Pharmacokinet Biopharm., vol. 23, 1995, pp. 651-72.