Role of Biomarkers in the Different Stages of Drug Development

Stages in the development of MTAs include basic science studies for target discovery and validation, translational studies for drug screening and validation in tumor models, and clinical evaluations of the safety and efficacy in patients (see figure below). Critical to successful transition from target identification to therapy are selection of the right agent, determination of the right dose and schedule, and identification of the right patients. Use of PD and predictive markers has the potential to improve the efficiency of these tasks and accuracy of the decisions. Specifically, PD markers can be used to confirm target engagement, define the dose PK-PD relationship and obtain early proof of principle for the biological effect downstream of the target inhibition. Predictive markers can be used to enrich or stratify patients for evaluation of the antitumor activities. In order to properly use the biomarker data to inform the drug development, it is important to recognize not only the potential values of various biomarkers but also their limitations.

Stages of Drug Development and Role of Biomarkers

80 Role of Biomarkers in Clinical Development of Cancer Therapies

Use of PD Markers in Phase I and Early-Stage Proof of Principle Studies

The goals of phase I trials are to evaluate the pharmacokinetics and safety of the agent and to make recommendations for the dose that will be used in phase II efficacy evaluation. For a drug to reliable test the hypothesis of the target, it is essential to ensure that the drug hits the intended target, and that the selected dose is optimal for efficacy and tolerability. The traditional models based on PK and toxicities have worked well for cytotoxic chemotherapies, whose target toxicities on proliferative tissues (bone marrow suppression or diarrhea) often follow a similar dose-effect relationship as in tumor cells. Some MTAs are also associated specific and quantifiable target-related toxicities (e.g., skin rash after EGFR inhibition), and these adverse events are useful indicators of the target effects. For other MTAs, the target effects on host tissues can be low or nonspecific, in which case, measurements of molecular changes in tumor tissues before and after drug exposure would be critical to verifying the target engagement and estimating the required dose for efficacy. Although PK and safety endpoints should remain the primary objectives of phase I trials, PD markers can provide important ancillary information when properly used and interpreted. Various PD markers have been used in modern day early clinical trials (Table 3.2).

Role of PD markers in verifying target engagement

For agents without target toxicities or nonspecific target toxicities, PD markers would be the only means to confirm the target engagement and the presumed mechanisms of action. An excellent example is the PARP inhibitors, which only affect cells in the context of DNA repair deficiency (e.g., tumors with BRAC1/2 deficiency) and therefore had no expected target toxicities. PD markers played an essential role in the early development of veliparib and olaparib, providing confirmation of target inhibition. In contrast, another "PARP" inhibitor, BSI-201, progressed to phase II studies without clear demonstration of target inhibition. Although the agent in combination with chemotherapy demonstrated a significant improvement in response rate and progression free survival in patients with triple negative breast cancers,⁵ subsequent preclinical

Role of Biomarkers in the Different Stages of Drug Development 81

Table 3.2 Types of post-treatment markers

studies with validated PAR assays revealed that the agent did not have any significant effect on the intended target, PARP (Ji et al., AACR 2011). While BSI-201 may have antitumor activity and warrant further development, the clinical trial strategy would need to be modified based on different mechanisms of action. Many MTAs are associated with very non-specific side effects (e.g., mTOR inhibitors, multi-target TKIs or proteasome inhibitors). Although MTD can often be reached, inhibition of the intended targets may or may to be achieved and requires PD measurements to confirm the target effects.

Role of PD markers in decisions on the recommended phase II dose: value and limitations

Because MTAs exerts their anti-tumor effects through modulation of the intended targets, it is conceivable that the optimal dose can be estimated based on the degree of post-treatment changes in the PD markers. PD marker played a critical role in the dose escalation and dose selection for bortezomib, a proteasome inhibitor. Because PK assay is not feasible, downregulation of proteasome 20 was used

82 Role of Biomarkers in Clinical Development of Cancer Therapies

as the PD marker for guidance on dose. Preclinical studies defined that proteasome decrease by 80% is required for antitumor activity and that inhibition of more than 80% is lethal. Successful use of this marker in phase I trials also required development of a reliable assay and the validation that peripheral blood mononuclear cells (PBMC) is a surrogate of tumor cells for this marker. Phase I trials were conducted based toxicities and PD endpoint of achieving 80% reduction of the target in PBMC.^6,7

However, solely replying on PD markers for dose selection can be misleading. In the phase I trial a specific VEGFR tyrosin kinase inhibition, PTK787,⁸ vascular permeability (defined as Ki) derived from dynamic contrast enhanced MRI (DCE-MRI) was used as a PD marker. Target effect was confirmed by a significant decrease in Ki following drug administration. It was also observed that the degree of Ki decreases at the end of cycle 1 appeared to correlate with stable disease. While stable disease status in a single arm trial could be the results of the natural history of the tumor with or without the drug effect, a "desired degree” of Ki decrease was defined based on "stable disease” and used for selection of the PTK787 dose (1250 mg QD) for subsequent phase III trial in combination with FOLFOX. While the same trials also indicated a short half-life of the drug and decrease in drug exposure in cycle 2, the observation did not impact the dose selection. Both phase III trials failed to meet the primary endpoint, and further exploration of the dose and schedules had to be pursued. Among the many potential reasons for the trial failure, over-interpretation of the PD marker is probably the one that could have been avoided. This PD marker has not been validated in either preclinical or clinical settings for correlation with efficacy and should be used as an ancillary, hypothesis-generating, rather a decision-making endpoint.

A pre-requisite for PD markers in phase I trials is the availability of reliable assays. It is also important to obtain sufficient information from preclinical models for guidance for the timing of the biopsies and the degree/duration of target inhibition that should be used to define as the desirable PD response. However, it is important to recognize that due to the heterogeneity between tumors as well as the intrinsic difference between models and patients, the PK-PD modeling derived from one or a few preclinical models may not always accurately predict the optimal dose in patients.

Role of Biomarkers in the Different Stages of Drug Development 83

Use of distal PD markers to measure the biological and molecular consequences of target inhibition

The hypothesis of molecularly targeted therapies assumes that if a valid target is inhibited, the downstream molecular and biological processes would be perturbed, leading to growth inhibition or death of the tumor cells. Although objective responses are the most direct and rapid readout of the antitumor activity, some potentially useful MTAs are cytostatic and requires randomized trials to demonstrate the therapeutic effect. It is hoped that PD markers distal to the target and reflective of the biological consequences may provide earlier and more sensitive indicators for the proof of principle. Examples of markers forbiologicalactivitiesincludethoseforcellular proliferation (e.g., Ki67), apoptosis (TUNEL), among others (Table 3.2). Molecular and functional imaging methodologies have also been developed to monitor the biological effects in tumors, including 18F-FDG-PET for metabolic status (glycolysis), 18F-FLT-PET for proliferation and ¹²⁴I-annexin V scans for apoptosis. Because imaging modalities are noninvasive and feasible for serial measurements at multiple time points, they have been increasingly used in clinical trials with MTAs. Utility of functional imaging in drug development however has not been validated. For some targeted agents, such as inhibitors of the AKT-mTOR pathway, which is directly involved in glucose homeostasis, changes in FDG-PET may simply reflect modulation of the target, rather than impact on cellular viability and proliferation.⁹

The post-treatment tumor tissues also offer the opportunity to explore additional molecular changes outside the putative downstream pathways and biological effects. Such studies have the potential to uncover resistance and escape mechanisms or unanticipated mechanisms of action.

Continued effort in developing and qualifying markers for intermediate biological effect are warranted. At the current time, use of these markers in the development of investigational agents should be considered explorative. Because biological consequences of target inhibition depend on the biological relevance of the target, "distal" PD marker status can vary with the molecular context of the tumor being treated. Similar to clinical evaluation of efficacy, studies for markers of biological will be more efficient in patients with tumor of similar molecular background.

84 Role of Biomarkers in Clinical Development of Cancer Therapies

In summary, PD markers should be considered in early clinical evaluation of MTAs, and the value and limitations should be recognized in their application. To date, PD markers have been successfully used to validate the MOA and confirm target engagement, and lack of or weak PD effects should be a clear warning sign that the agent or the dose is inadequate. However, what constitutes an optimal or sufficient dose is difficult to predict due to lack of validated preclinical models. We recommend that for most agents, it would be prudent to escalate the dose to MTD and that decision on the phase 2 dose should take into account all factors including PK, toxicity as well as PD effects. When in doubt, testing two potential dose levels in expansion cohorts could be valuable for additional assessment of the PD markers and biological activities, before finalizing the dose selection for definitive trials.