Introduction
Soft tissue sarcomas (STS) are a group of rare tumours of mesenchymal origin, accounting for approximately 1% of all adult cancers [
1‐
3]. For most histological subtypes, the standard management of localised disease consists of complete surgical resection with or without radiation. Despite optimal management, however, high-risk patients will develop recurrent locally advanced inoperable or metastatic disease. The outcome for patients with advanced inoperable/metastatic STS is poor with a median overall survival (OS) in the range of 12–18 months [
4‐
8]. There are few treatment options available, and these have historically included doxorubicin with or without ifosfamide. Over the last few years, a number of other drugs have emerged including gemcitabine/docetaxel, trabectedin, pazopanib, and eribulin [
6,
9‐
12].
Olaratumab is a recombinant human immunoglobulin G1 monoclonal antibody to platelet-derived growth factor receptor alpha (PDGFRα) [
13]. A randomised phase 2 trial of doxorubicin with or without olaratumab in patients with advanced STS demonstrated a significantly longer median OS for the combination of doxorubicin and olaratumab compared to doxorubicin alone (26.5 and 14.7 months, respectively, hazard ratio [HR] 0.46,
p = 0.0003) [
14]. The increase in progression-free survival (PFS) was also significant (6.6 months and 4.1 months, respectively, HR 0.67,
p = 0.0615) and the combination of olaratumab with doxorubicin led to a slight increase in toxicity but remained well-tolerated. Based on this phase 2 STS trial, olaratumab was granted accelerated/conditional approval by a number of regulatory agencies.
A matched case–control analysis [
15] performed on the phase 2 PFS and OS survival data stratified by quartiles of olaratumab serum exposure indicated that patients in the lowest quartile may not have received optimal level of clinical benefit [
14]. A population pharmacokinetic (PopPK) analysis subsequently performed using PK data combined from four phase 2 studies, including that in STS, indicated that the dose of 15 mg/kg administered on Days 1 and 8 of a 21-day cycle yields olaratumab serum levels likely to achieve full target saturation [
16]. In light of these findings, it seems necessary to better define the therapeutic window of olaratumab and determine whether the dose of 15 mg/kg used in the phase 2 study represents the optimal dose to be used in combination with doxorubicin in STS patients. The aim of this study was therefore to characterise the exposure–response relationship of olaratumab in combination with doxorubicin for PFS, OS, and safety for patients with advanced STS.
Discussion
The objective of this analysis was to characterise the exposure–response relationship of olaratumab for survival outcomes and safety when combined with doxorubicin in patients with advanced STS. The combined exposure–response information for efficacy and safety was then used to optimise the dosing strategy of olaratumab and better target its therapeutic window in the ongoing confirmatory phase 3 study (NCT02451943) after accelerated approval by Food and Drug Administration and the conditional approval by the European Medicines Agency.
The exposure–response relationship of olaratumab was first characterised for survival outcomes. OS in the study was best described by a model with a constant baseline hazard, and a sigmoidal relationship for the effect of olaratumab. The Cmin1-based model yielded an ECmin150 estimate (66 µg/mL) corresponding to the 25th percentile of the Cmin1 distribution in the study, and a Hill coefficient indicative of a steep exposure–response relationship. The EMAX estimate corresponded to a maximum decrease in the HR of approximately 75%, and was reached within the range of olaratumab serum concentration achieved in the study. Importantly, a similar exposure–response relationship was identified with the Cavg-based model: the baseline hazard, EMAX, and Hill coefficient estimates were similar to those obtained in the Cmin1-based model, and the ECavg50 estimate (134 µg/mL) also corresponded to the 25th percentile of the Cavg distribution in the study. Both PK variables therefore seem to be similarly predictive of the effect of olaratumab on OS.
These findings indicate that a small increase in
Cmin1 or
Cavg in the vicinity of the EC50 is expected to lead to a dramatic change from low to near maximal OS benefit. Since the EC50 estimates correspond to the 25th percentile of olaratumab exposure in the study population, the dose of 15 mg/kg, administered on Days 1 and 8 of a 21-day cycle, therefore provides the majority of the study population with near maximum OS benefit. This is consistent with results from the PopPK analysis, where the linear clearance used to describe the disposition of olaratumab suggested that the dose of 15 mg/kg achieves serum levels leading to full target saturation [
16]. This is also in line with the results of a previously published matched-case control (MCC) analysis on the same data [
14] which indicated that: (1) patients in the upper three C
min1 and C
avg quartiles showed an improvement in OS; (2) there was no consistent difference in OS benefit across the upper three C
min1 and C
avg quartiles; and (3) HR values observed in the upper quartiles were in line with the model-predicted E
MAX. It should also be pointed out that the ECOG PS and the number of prior lines of treatment were found to have a significant influence on OS, which is in line with the current understanding of clinical prognostic factors in STS and further supports the validity of our findings.
PFS in the study was also best described by a model with a constant baseline hazard and a sigmoidal relationship for the effect of olaratumab. The ECmin150 and ECavg50 estimates corresponded, respectively, to the median Cmin1 and Cavg in the study population, and the predicted EMAX was lower than that for OS. In addition, the HR for PFS was not predicted to improve until the Cmin1 or Cavg reaches values corresponding to the 25th percentile of their distribution in the study. These findings are line with the lower activity on PFS compared with OS previously reported for olaratumab and with the previous MCC analysis on PFS which suggested that patients in the lowest exposure quartile tend to experience disease progression within the first two to three cycles of treatment.
Patients who received olaratumab in combination with doxorubicin did experience an increase in the rate of TEAEs when compared to doxorubicin alone, consistent with the toxicity profile of doxorubicin. There was, however, no apparent additional increase in the rate of TEAEs with increasing olaratumab serum levels, regardless of the TEAEs examined in the patients examined thus far. The exposure–response relationship of olaratumab for toxicity is thus very shallow, so that an increase in clinical benefit may be achieved without an increase in serious (high grade) TEAEs. The findings from the safety assessment should be interpreted with caution due to the limited number of patients that experienced TEAEs. Safety data from the ongoing confirmatory phase 3 study will provide more conclusive results.
Altogether, our analysis indicates that olaratumab has a wide therapeutic window, characterised by a steep exposure–response relationship for efficacy and a shallow exposure–response relationship for toxicity. It also indicates that the therapeutic window of olaratumab was effectively targeted by the dose of 15 mg/kg tested in the randomised phase 2 study where approximately 75% of the population were exposed to olaratumab serum levels associated with OS benefit and maximum OS benefit was potentially reached. Finally, our analysis suggests that an olaratumab Cmin1 of 66 µg/mL or Cavg of 134 µg/mL may represent a minimum threshold for delaying disease progression and providing OS benefit in STS.
This hypothesis was used to further optimise the dosing strategy for the ongoing randomised phase 3 study of olaratumab combined with doxorubicin (ANNOUNCE). Simulations using the PopPK model previously developed for olaratumab indicate that the use of a loading dose of 20 mg/kg on Days 1 and 8 of Cycle 1 would achieve olaratumab serum levels comparable to those observed at steady state with 15 mg/kg. The dosing strategy for the randomised phase 3 study therefore consists of a loading dose of 20 mg/kg of olaratumab during Cycle 1 followed by 15 mg/kg in ensuing cycles. This dosing strategy is expected to better target the therapeutic window of olaratumab by (1) minimising the number of patients whose Cmin1 falls below 66 µg/mL at the start of treatment; (2) replicating olaratumab steady-state serum levels associated with OS benefit; and (3) preserving the positive benefit–risk ratio of olaratumab by maintaining olaratumab serum levels with the same total range as in the randomised phase 2 olaratumab trial.
Conclusions
The exposure–response relationship of olaratumab for PFS and OS are best described by time-to-event models with exponential hazard functions, and the effects of olaratumab on PFS and OS were well-characterized by inhibitory EMAX functions with Hill coefficients. Both PK endpoints, Cmin1 and Cavg, were equally predictive of the effect of olaratumab on OS and PFS. The model estimated maximum OS benefit was achieved by 75% of the patients in the trial, whereas only 50% of the patients are estimated to have achieved maximum benefit in PFS. The therapeutic window of olaratumab appears wide as increasing olaratumab serum concentration was not associated with increased incidence rate of TEAEs. Based on the analysis presented here and the evidence previously reported on this clinical study, a loading cycle of 20 mg/kg of olaratumab was incorporated into the confirmatory phase 3 study with the aim to prevent early disease progression and increase the number of patients that could potentially experience OS benefit.
Compliance with ethical standards