Issue Analysis Summary

Date: 2007-08-31

Help on accessing alternative formats, such as Portable Document Format (PDF), Microsoft Word and PowerPoint (PPT) files, can be obtained in the alternate format help section.

---

Contact Policy

---

The use of progression-free survival as the efficacy endpoint for approval of targeted and chemotherapeutic agents for advanced cancer

Oncology Division
Bureau of Metabolism, Oncology and Reproductive Sciences (BMORS)
Therapeutic Products Directorate
August 2007

---

Table of Contents

1. Issue
2. Purpose/Objective
3. Background
4. Consultation
5. Issue Analysis
6. Recommendations
7. References

---

1. Issue:

Health Canada has not issued a guideline on acceptable endpoints for the approval of oncology therapies due to difficulty with resource allocations. However the Oncology Division concurs with guidance documents put forth by other regulatory agencies and uses them as reference.^1,2,3 Endpoints to assess the efficacy of cancer therapies include overall survival, endpoints based on tumour assessments (disease-free survival, objective response rate, time to progression and progression-free survival), and endpoints based on symptom assessment.

Overall survivalis considered a precise and easily measured endpoint that is clinically meaningful and not subject to bias. Limitations of survival analyses include long follow-up periods in large trials, and the effect of subsequent therapies may confound the analysis of survival.

In the past, progression free survival (PFS) was not used as primary surrogate endpoint for market authorization. Nevertheless, a number of submissions have recently been filed with the Therapeutic Products Directorate (TPD) for advanced disease, and some received conditional marketing authorization based on PFS.^4,5,6 Conditional approvals were granted pending overall survival data as phase IV commitments.

PFS is a lucrative endpoint because it captures events of progressions and death, both of which are important, plausible endpoints in cancer therapy. Due to this definition, stable disease is captured as a benefit of therapy. It requires a smaller number of patients enrolled in clinical trials, and shorter follow-up when compared with survival studies. PFS is not affected by crossover or subsequent therapies, and events of progression are based on objective quantitative assessment.

The issue is whether an improvement in PFS represents a direct measure of clinical benefit or a surrogate for clinical benefit.

The following issues with PFS have been identified during the review of those submissions:

• PFS is a non-validated endpoint in many of those indications (i.e. data are lacking that longer PFS delivers longer survival or better quality of life).
  
  
• PFS is subject to assessment bias.
  
  
• The assessment intervals are arbitrary, making it difficult to compare median PFS results across different trials.⁷ There is no consensus among investigators what should be the standard radiologic surveillance interval for a particular disease, or for certain types of (cytotoxic vs non cytotoxic) drugs.
  
  
• Some of the pivotal studies were open label using different routes of administration of drugs and different administration schedules to that of the comparative arm. Due to the unblinded nature of those studies, biases cannot be ruled out.
  
  
• The definition of PFS and tumor progression criteria may vary among studies even for the same indication. As well, there are no standard regulatory criteria for defining progression.
  
  
• As an endpoint, PFS is not precisely measured. Although most of the time tumour measurements are verified by independent central review board that are blinded to study treatment, significant discrepancy may occur between the central and on-site investigator's assessments as definitions of progression may be different. This fact questions the reproducibility of PFS measurements, and leaves the regulatory agency at choice which data to take into consideration.
  
  
• There are different methodologies for handling missing data.
  
  
• There is no consensus on censoring methods.
  
  
• There is no consensus on sensitivity analyses (how many, why) to assess the robustness of the PFS data.
  
  
• Health Canada's input on design, acceptable endpoints etc. are rarely sought. In cases where Health Canada's advice is given, it may not be taken by the pharmaceutical company due to binding agreements with major worldwide regulatory agencies.
  
  
• Studies with statistically significant median PFS values may allow crossover before an established clinical benefit, such as survival, is demonstrated.^4,5,6,8,9

Analysis of overall survival is influenced by the long term risk-benefit profile of the drug and subsequent therapies the patient goes on to receive. Early stopping and crossover creates limitations for survival assessments and for drawing a final risk-benefit from the new treatment compared to that of the standard. If these PFS data lead to market authorization, the impact on further clinical studies aiming long-term benefit will be grave as patients do not want to be randomized to the perceived "less effective" therapy. Thus drug development may never reach its endpoint of real clinical benefit (i.e. on overall survival).

2. Purpose / Objective

• To present to the SAC-OT the issues that TPD is facing when reviewing drug submissions for oncological indications using PFS as the primary efficacy endpoint.
  
  
• To establish regulatory consistency in Canada for chemotherapeutic and targeted agents applying for approval using improvement in PFS as the primary efficacy endpoint.
  
  
• To endorse the approach of TPD towards these issues by the Canadian oncology community.

3. Background

Canadian Drug Approval Process
In Canada, drugs may be issued a notice of compliance with conditions (NOC/c) to provide patients suffering from serious, life-threatening or severely disabling diseases or conditions, accelerated access to promising new therapies.¹⁰ Approval under the NOC/c Policy is based on the demonstration of promising clinical evidence that the drug has an effect on a surrogate or clinical endpoint that is reasonably likely to predict clinical benefit. For full approval (NOC) in Canada, a therapy should demonstrate direct evidence of clinical benefit (eg increased overall survival or improvement of quality of life) or improvement in an established surrogate for clinical benefit. For a therapy approved under the NOC/c Policy, the post-approval clinical trial commitments should demonstrate the actual clinical benefit of the therapy.

Consistent with the FDA, Health Canada historically approved cancer therapies based on objective response rate, which was not always a validated surrogate for clinical benefit, but was felt to indicate pharmacological activity of traditional cytotoxic oncology therapies. The trend then became the approval of cancer therapies based on direct evidence of clinical benefit such as an improvement in survival, improvement in a patient's quality of life, improved physical functioning or improved tumour-related symptoms.

Approvals based on surrogate endpoints were done recognizing the "unmet medical need" and the available suboptimal therapeutic options for the disease in question. Commonly accepted surrogate variables have been used for a number of indications in which they are believed to be reliable predictors of clinical benefit. The International Committee on Harmonization (ICH) has outlined issues and considerations for the use of surrogate endpoints.¹¹ Firstly, the surrogate endpoint may not be a true predictor of clinical benefit. Secondly, a proposed surrogate endpoint may not yield a quantitative measure of clinical benefit that can be weighed directly against adverse effects in formulating a risk-benefit analysis of the proposed therapy. Thirdly, experience in the validation of surrogate endpoints is limited. Of note, relationships between clinical and surrogate variables for one treatment do not necessarily apply to a product with a different mode of action for treating the same disease. As the ICH describes, the strength of the evidence for surrogacy depends upon:

• the biological plausibility of the relationship
  
  
• the demonstration in epidemiological studies of the prognostic value of the surrogate for the clinical outcome
  
  
• evidence from clinical trials that treatment effects on the surrogate correspond to effects on clinical outcome

Chemotherapeutic (cytotoxic) agents
The traditional surrogate endpoint for new chemotherapeutic agents applying for market authorization in advanced cancer is objective response rate (ORR), in other words, tumour shrinkage. ORR encompasses complete responses (CR) and partial responses (PR) and does not include a measure of stable disease. ORR is considered direct evidence of pharmacologic activity of the drug. ORR has been a surrogate of overall survival in only a few malignancies. In those malignancies, where increased ORR did not result in survival advantage, it was assumed that tumour shrinkage may lead to a decrease in tumour-associated symptoms, and hence, to clinical benefit. This is a biologically plausible assumption, although it has been rarely supported by convincing evidence.

If a chemotherapeutic agent increases PFS without significantly increasing ORR that leaves questions with regard to the pharmacological activity of the agent for the disease. Hence, improvement in PFS on its own is not sufficient evidence to grant market authorization for chemotherapeutic agents.

Targeted (cytostatic) agents in oncology therapy ^9,12,13,14
As proposed by the ICH, the biological plausibility of the relationship should be considered when assessing whether a surrogate endpoint is a measure of clinical benefit.¹¹

Targeted therapies are agents that inhibit a specific molecular target deemed to be crucial for cancer cell proliferation, resulting in tumour stasis (cytostatic response). While cytotoxic agents affect highly proliferative tumour and non-tumour cells, targeted agents block receptors over-expressed on cancer cells or bind molecular entities involved in tumour growth such as growth factor receptors and signal transduction pathways in the tumour stroma, without causing direct cytotoxic activity. In other words, they may not shrink tumours but instead halt their progression. Thus, PFS (which registers only progression and death) is a plausible endpoint for targeted therapies, which act mainly to halt tumour progression

Early stopping of randomized clinical trials
Clinical trials should provide interpretable results while preserving the safety of trial participants.¹⁵ Interim analyses often lead to early stopping of a trial based on pre-specified stopping rules that relate to toxicity and outcome. If a statistically significant benefit is observed, the trial is stopped early, and the patient is typically offered the opportunity to receive the therapy that is perceived to be superior. The early stopping rule has the potential to minimize harm and to maximize benefit for participants in a randomized trial.

A leading FDA biostatistician disagrees with crossover justification based on PFS: "The rationale assumes the experimental treatment impacts overall survival positively, i.e. it assumes exactly what it was supposed to (and failed to) demonstrate." ⁸
Trials stopped early for benefit are becoming more prevalent.^8,16 An issue is that the use of a composite endpoint, such as PFS, to monitor the trial may lead to decisions to stop the trial driven by the least patient-important outcome that makes up the composite endpoint, and as a result, few events important to patients, such as survival, accrue. As well, because of the limited follow-up in these truncated trials, the lack of adequate safety data may affect the perceived and actual risk-benefit ratio by overestimating the benefit and under-estimating the risk. By employing surrogate markers as therapeutic endpoints, the trial is often stopped before the point at which it might otherwise have defined real benefit to the patient.¹⁷ Montori et al. (2005) suggest that consideration should be given to the plausibility of the treatment effect, the planned sample size, the number of interim analyses that were performed, and the statistical methods used to monitor the trial in terms of assessing the results of a trial stopped early for perceived benefit.

Cannistra (2004) has suggested that early stopping rules might compromise proper interpretation of clinical trial results and thereby jeopardize the patients they are designed to protect.¹⁵ It is argued that the early stopping of a study because of the possibility of a survival advantage may expose patients to potential harm by not fully elucidating the long-term risks and benefits of a therapy. It is suggested that the deficiencies of a study that is stopped early because of perceived clinical benefit in terms of PFS advantage should be acknowledged, but the treatment option should be discussed with the patient nonetheless. As well, it is argued that PFS should be considered as an early stopping criterion for those instances in which standard treatment is so unsatisfactory that any effect could be interpreted as being clinically meaningful, and no effective salvage therapy at relapse is known to exist.

4. Consultation

A literature search was conducted to determine the standpoints, concerns and recommendations of experts, regulatory agencies and international panels on clinically acceptable endpoints for the assessment of oncology therapies, with a focus on cytostatic agents. The guidance papers put forth by the FDA and EMEA on clinical trial endpoints for the approval of oncology therapies were consulted, with a particular emphasis on PFS.

5. Issue Analysis

1. Should approval be granted to a new chemotherapeutic agent based on a statistically significant increase in PFS compared with that of the standard therapy, but with weak evidence of efficacy based on ORR?

Current approach of TPD: If a chemotherapeutic agent increases PFS without significantly impacting on ORR, this questions the pharmacological activity of the agent for the disease. Traditionally, pharmacological activity (tumour shrinkage) was important evidence of efficacy. Between 1999-2002, more than 50% (34/62) of approved chemotherapeutic agents were granted market authorization based on ORR as the primary endpoint. Thus, demonstration of evidence of pharmacological activity on the cancer is important evidence of efficacy. It is appreciated that the effect of chemotherapy is not only confined to tumour cell killing (there may be evidence of immuno-modulatory, anti-angiogenic and cytostatic effects of certain chemotherapeutic agents). Given the above outlined issues with PFS, the likelihood cannot be ruled out that a potentially inactive drug be approved if PFS is the only evidence of efficacy for this chemotherapeutic agent.

2. For treatments of palliative intent, can a significant improvement in PFS on its own be the basis of full approval (NOC) of a targeted agent or should it be considered in combination with additional endpoints? What should those other endpoints be (ORR, survival, quality-of-life, and symptom improvement etc.)?

Current approach of TPD: ORR and/or PFS have been the basis of conditional approval of targeted agents, with the request for survival data to confirm the clinical benefit of the therapy. Survival data usually comes as long-term follow-up from the registration trial or from another trial for the same indication. TPD has not considered PFS as a direct measure of clinical benefit for treatments of palliative intent. So far, none of the targeted agents that received conditional marketing authorization based on either ORR or PFS have delivered conclusive evidence of survival benefit, and few have demonstrated palliative benefit (e.g. quality-of-life data). With early crossover of these trials based on a PFS benefit, the assessment of further efficacy data and the ability to demonstrate a survival advantage is compromised. Thus, the phase IV commitment of survival advantage becomes undeliverable. Given that PFS is dependent on multiple factors and settings (follow up interval; censoring methods; investigator claims etc), the ultimate benefit of those therapies in terms of palliation or survival is uncertain.

3. What should be the phase IV commitment of targeted agents registered with PFS as the primary endpoint, where patients were allowed to crossover to the new drug arm based on a statistically significant PFS benefit?

Current approach of Health Canada: Health Canada's Guidance for Industry Notice of Compliance with Conditions (NOC/C section 2.4) states:..."confirmation of either the efficacy or safety of the agent under question [require] trials which should be replicates of the pivotal trials or studies of different design where the outcomes are congruent with, and complementary to those of the original trial."¹⁰ In other words, evidence should come from similar studies for the same indication. It is acknowledged that it is difficult to conduct such trials following conditional approval of a new targeted agent. Patients want the newly approved drug and do not want to have a chance to be randomized to the standard "less effective" therapy. Health Canada may accept a study with evidence of efficacy on the same surrogate for the same disease, but for a less advanced stage (second-line versus third-line therapy). Following crossover, overall survival data may be compromised, and with a lack of ongoing confirmatory studies, clinical benefit may not be shown.

6. Recommendations

The issue analysis highlights the need for consultation between regulators and the oncology community regarding the issue of PFS as an efficacy endpoint for the approval of cytostatic oncology therapies. A consistent approach needs to be employed to carry out Health Canada's mandate of making safe and effective therapies available to Canadians in a timely manner. The advice of the panel will be used by TPD as guidance to apply consistency in approvals of cytostatic therapies that use PFS as the efficacy endpoint for the basis of approval.

7. References