Modeling of hypertension in response to anti-angiogenic therapy
Ron J Keizer (1), Anubha Gupta (2), Mendel Jansen (2), Jantien Wanders (2), Jos H Beijnen (1,4), Jan HM Schellens (3), Mats O Karlsson (5), Alwin DR Huitema (1)
Slotervaart Hospital / the Netherlands Cancer Institute
Objectives: Hypertension is a common toxicity for anti-angiogenic drugs targeting the VEGF pathway, such as tyrosine kinase inhibitors (e.g. sorafenib [1]) and monoclonal antibodies (bevacizumab [2]). In most cases, hypertension can be controlled by dose reductions, dose delays, or prescription of antihypertensive therapy. Our aim was to construct a general pharmacokinetic-pharmacodynamic (PK-PD) model for hypertension toxicity following treatment with anti-angiogenic agents. This model may guide clinical interventions and provide treatment recommendations for this compound and other drugs in this class.
The novel multiple tyrosine kinases inhibitor E7080 has shown anti-tumor activity in preclinical testing [3,4] and phase I clinical trials. [5-7] Although generally well tolerated in phase I clinical studies, commonly observed toxicities with E7080 were hypertension and proteinurea. From data collected in these phase I trials a PK-PD model was developed describing the increase in BP in response to exposure with the drug. Using simulations, we evaluated several intervention options for management of hypertension toxicity.
Methods: Plasma concentration and blood pressure (systolic: BPsys, diastolic: BPdia) data were obtained from 67 patients enrolled in a two-center phase I study, investigating qd dosing of E7080 at increasing dose levels. BP data were recorded weekly, over a mean period of 21 weeks (range 1-77 weeks). PK and BP data were modeled sequentially with NONMEM software, using the FOCE-I method. The PK model was developed previously, and consisted of a two-compartment model, combined zero- and first-order absorption, and linear elimination. Several models for BP, baseline BP, and drug effect on BP were evaluated including effect-compartment models and indirect effect models or combined effects. [8] Prescription of anti-hypertensive (AH) medication was accounted for in the model as a negative effect on BP using ‘defined daily dose equivalents' (DDDE), i.e. the ratio of daily dose/defined daily dose, cumulative over all AH medications. Model fits were evaluated using the likelihood ratio test, individual plots and visual predictive checks.
Using the final model and parameter estimates, several simulation studies were performed evaluating interventions for management of hypertension. Simulations were performed for 500 patients. To be able to estimate precision, this was repeated 500 times with parameter sets drawn from the full covariance matrix obtained for the final model. In the simulation studies, efforts were made to mimic the clinical protocol as best as possible: when presented with hypertension in the clinic, the BP measurements were repeated at least 2 times during the next hour(s). This was implemented in the intervention model used in the simulations by averaging multiple drawn observations on each occasion when hypertension occurred, assuming that 50% of the variation in BP was due to inter-day variability and 50% to inter-measurement variability.
Results: The final BP model consisted of two separate indirect-effect models for BPsys and BPdia. Baseline BPsys and BPdia were estimated at 126 and 77 mmHg (RSE <5%) respectively, with 10% (RSE 12%) inter-individual variability (IIV) for both BPsys and BPdia, and 70% (RSE 15%) correlation between systolic and diastolic baseline. Input rate for the indirect effect model was 0.35 mmHg.hours-1 (RSE 20%) for both BPsys and BPdia. The effect of E7080 plasma concentration on the input rate of the indirect effect PD model was 0.497 (RSE 21%) and 0.88 (RSE 13%) ng-1.ml for BPsys and BPdia respectively, leading to an approximate similar absolute effect size (in mmHg). IIV on drug effect was estimated at 72% (RSE 40%) for BPsys and 26% (RSE 50%) for BPdia input rates, with a correlation of 42% (RSE 62%) between them. The effect of AH medication on BP input rates could not be estimated separately for BPsys and BPdia, and was 0.036 per DDDE for both. The overall exponential residual error in BP was 12.9% (RSE 3 %), with 53% (RSE 3%) correlation between systolic/diastolic errors.
Simulations showed that continuous dosing of E7080 at the maximum tolerable dose (MTD) of 25 mg qd without AH medication, would result in an increase of 16.2 / 17.7 mmHg (systolic/diastolic, RSE 20%) for a typical patient, with half of the increase attained within 2.5 weeks. At this dose level, dose limiting hypertension (defined as an increase of ≥20 mmHg in BPdia, corresponding to CTC grade ≥2) would be expected in 56.7% (RSE 15.2%) of the patients. In patients that developed hypertension, BPdia could be reduced by a median 2.6 mmHg (RSE 84%) after 12 weeks of treatment with AH medication at 1 DDDE. Intensification of AH therapy to 2 DDDE when hypertension could not be controlled sufficiently, could reduce BPdia by 4.8 mmHg (RSE 39%). This resulted in 63.7% and 68.8% (RSE 13%) of patients respectively being able to continue treatment with E7080. The effect of a dose reduction to 50% of the MTD (12.5 mg/day) resulted in a median decrease of 7.0 mmHg (RSE 29.3%), resulting in 82.1% (RSE 7%) of patients being able to continue treatment.
Conclusions: A PK-PD model was developed that was able to capture the effect of daily treatment with E7080 on BP, which showed a clear exposure-response relation. We were able to account for the use of anti-hypertensive medication in the model. Using simulations we evaluated several interventions for management of hypertension toxicity. Data from upcoming studies with different regimens, will aid in defining the relationship between exposure and hypertension and the effects of AH therapy in more detail. Incorporation of a model to describe proteinurea toxicity is currently ongoing, which may allow further treatment optimization. The current model will aid in further clinical development of E7080, and can serve as a template model for analyzing hypertension toxicity in treatment with this class of drugs.
References:
[1] D.A. Sica, "Angiogenesis inhibitors and hypertension: an emerging issue," Journal of clinical oncology : official journal of the American Society of Clinical Oncology, vol. 24, Mar. 2006, pp. 1329-31.
[2] M.A. Cobleigh et al., "A phase I/II dose-escalation trial of bevacizumab in previously treated metastatic breast cancer," Seminars in oncology, vol. 30, Oct. 2003, pp. 117-24.
[3] J. Matsui, et al., "E7080, a novel inhibitor that targets multiple kinases, has potent antitumor activities against stem cell factor producing human small cell lung cancer H146, based on angiogenesis inhibition," International journal of cancer. Journal international du cancer, vol. 122, Feb. 2008, pp. 664-71.
[4] J. Matsui et al., "Multi-kinase inhibitor E7080 suppresses lymph node and lung metastases of human mammary breast tumor MDA-MB-231 via inhibition of vascular endothelial growth factor-receptor (VEGF-R) 2 and VEGF-R3 kinase," Clinical Cancer Research: An Official Journal of the American Association for Cancer Research, vol. 14, Sep. 2008, pp. 5459-65.
[5] K. Yamada et al., "Phase I dose escalation study and biomarker analysis of E7080 in patients with advanced solid tumors," J Clin Oncol, vol. 26, May. 2008.
[6] J. Nemunaitis et al., "Phase I dose-escalation study of E7080, a multikinase inhibitor, in patients with advanced solid tumors," J Clin Oncol, vol. 26, May. 2008.
[7] H. Glen et al., "A phase I dose finding study of E7080 in patients (pts) with advanced malignancies," J Clin Oncol (Meeting Abstracts), vol. 25, Jun. 2007, p. 14073.
[8] N.L. Dayneka, V. Garg, and W.J. Jusko, "Comparison of four basic models of indirect pharmacodynamic responses," Journal of pharmacokinetics and biopharmaceutics, vol. 21, Aug. 1993, pp. 457-78.
