Therapeutic Products Directorate
Scientific Advisory Committee on Oncology Therapies (SAC-OT) (2007/09/13)

Date: 2007-09-13

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 Bureau Enquiries

Record of Proceedings

Note: These minutes have been edited for confidential information in accordance with the Treasury Board Manual - Policy & Guidelines on Security; the relevant Sections of the Access to Information Act are indicated in [ ]where information has been removed.

Panel Members Present: Dr. Alexander Paterson (chair), Mr. Harley Ast, Dr. Allan Donner (via teleconference), Dr. Alwin Jeyakumar, Dr. Kong Khoo, Dr. Kara Laing, Dr. Ron McCormick, Dr. Lynne Nakashima, Dr. Maureen Trudeau, and Dr. Lesley Seymour

Regrets: Dr. Rick Abbott, Dr. Joseph Ayoub, Dr. Charles Butts, Dr. Mark Levine, Dr. Wojciech Morzycki, Mrs. Deanna Silverman, Dr. Ian Tannock

Sanofi-Aventis Inc.: Eight representatives from sanofi-aventis were present and two consultants

Health Canada (HC) Representatives: Monique Chaine (SAB Coordinator), Bob Li (Science Advisor), Barbara Rotter (BMORS Director), Martha Harczy (Bureau WG Lead), Roxana Alexa, Andrea Bell, Ama Bhojani, Robyn Blom, Eduardo del Campo, Jean-Francois Ethier, Karen Johns, Andree-Marie Leroux, Helen Mao, Jaigi Mathai, Tricia Newell, Elitza Palazov, Bruce Randall, Kathy Soltys

* Abbreviations used in this record

AI = Aromatase Inhibitor
BMORS = Bureau of Metabolism, Oncology and Reproductive Sciences
COI = Conflict of Interest
DFS = Disease Free Survival
FDA = United States Food and Drug Administration
HC = Health Canada
NOC = Notice of Compliance
NOC/C = Notice of Compliance With Condition
ORR = Overall Response Rate
OS = Overall Survival
PFS = Progression Free Survival
S-A = Sanofi-aventis
SAC-OT = Scientific Advisory Committee on Oncology Therapies

1. Opening Remarks & Welcome

The Chair welcomed and thanked everyone for attending the meeting. He noted that in the past year there have been some changes in committee membership. Some members completed their terms and chose not to renew while others were recruited to ensure that we have the required expertise on the committee. They were welcomed and asked to briefly introduce themselves.

2. Review of the Agenda and Conflict of Interest (COI) declaration

The agenda was accepted as established.

The Terms of Reference of the SAC-OT were reviewed for all and the administrative committee processes were also quickly discussed. A general Conflict of Interest declaration by each member followed, and none was declared. Members were advised that prior to each new topic discussion, a brief COI would be completed to ensure that nothing was missed by the general declaration at the beginning of the meeting.

Representatives of sanofi-aventis were invited to participate in the first portion of the meeting where questions referring to Eloxatin were discussed.  They were invited to present their views on the questions. Their presentation followed the Bureau's presentation.

3. Disease-Free Survival (DFS) as an Endpoint in Adjuvant Colon Cancer Clinical Trials

Health Canada presented data submitted to them for the first time in 2007 on the MOSAIC trial.  Six year disease-free survival (DFS) and overall survival (OS) data was available. This trial, which had 2,246 patients randomized either to 5-fluorouracil/folinic acid and Oxaliplatin or to 5-fluorouracil/folinic acid alone showed a significant DFS advantage at 3 years (this was the protocol-defined primary endpoint) but at 5 years, overall survival showed only a trend in favour of the 5-F-U/folinic acid/Oxaliplatin group which was not statistically significant. Six year overall survival, however, was significant for the sub-group of stage III patients.

Presentations were made by the sanofi-aventis team outlining the sequential improvements in overall survival with succeeding generations of adjuvant therapy. Clinical Practice Guidelines from Ontario and other organizations were presented and the MOSAIC study was further reviewed. Rationale supporting disease-free survival as a primary independent endpoint in adjuvant colon cancer trials was presented, including data from a meta-analysis of adjuvant colorectal cancer trials (Sargent et al). This rationale included the proposal that a DFS endpoint provides more rapid completion and earlier reporting of trials, thereby allowing promising agents to benefit patients more quickly; that it is not affected by subsequent therapy for recurrent disease and that it is a more biologically relevant measure of a new treatment's impact. In short, for these reasons it was concluded that it has widely agreed clinical merit.

Preliminary discussion between the Committee and sanofi-aventis (S-A):

Eloxatin had been approved in over sixty countries around the world. The drug was approved for stage IV colon cancer in Canada in June 2007. Common clinical practice in Canada currently is to use Oxaliplatin-based adjuvant chemotherapy in stage III colorectal cancer patients, but it is also used in a selected group of high-risk stage II patients.

[Section 20(1)]

In conclusion, the Sponsor felt that the Folfox-4 treatment had demonstrated improvement over 5-Fluorouracil and Leucovorin. The overall survival results may also have been affected by cross-over therapy in the control group on relapse.

At this point, members of the sanofi-aventis team left the room.

There then followed discussions on four specific questions put forward by Health Canada.

Question 1:
For MOSAIC, why is DFS not a surrogate of OS in the overall population? Consider safety profile; dilutional effect of Stage II patients; subsequent therapies.

Some general statements were put forward by the committee members with respect to DFS as a surrogate for overall survival.

If a trial is designed so that all of the patients in the control group receive the experimental treatment on relapse, then it may be hard to show an overall survival difference. This makes knowledge and reporting of treatment on relapse (with both the experimental agent(s) and other salvage therapies) critical when reviewing a clinical trial from the perspective of OS. Some trials are done in countries where the experimental treatment is not given to the control group and in those instances overall survival may be easier to demonstrate.

Different time frames for follow-up may be required for different diseases (for example, breast cancer versus colon cancer) and also for different modalities of therapy (such as hormone therapy versus chemotherapy).

[Section 20(1)]

The meta-analysis of adjuvant colorectal cancer trials (Sargent, et al.) supported the link between DFS and OS in 5-FU based studies, but to assume that this relationship can always be applied to other disease sites and other treatments requires further evidence. While it was likely that evidence of a link between DFS and OS might be demonstrated with longer follow-up (for example, in the adjuvant treatment of breast cancer), it cannot necessarily be applied as a generalization to all malignant diseases with all treatments. Each trial has to be viewed individually, with a balanced consideration of a number of interacting factors including the
magnitude of the DFS benefit, the toxicities encountered in the adjuvant setting (which, if chronic, may negatively impact OS), and the availability and quality of salvage therapies on relapse.

Question 2:
Should DFS still be considered as a surrogate of OS in the adjuvant therapy of colon cancer?

A lengthy discussion was held on this issue.

[Section 20(1)]

The magnitude of the DFS difference in any clinical trial was of critical importance. A large difference in DFS was more likely to favourably impact overall survival even with the availability of good salvage therapies. If there was only a modest increment in DFS with good available salvage therapy, then an overall survival benefit may not be seen.

If a DFS advantage was relatively small and the regimen under review caused significant morbidity and possibly mortality, then it would be important to demonstrate clearly that an overall survival advantage or other clinical benefit was occurring.

In trials where OS is a secondary endpoint and there is a significant DFS advantage, and OS analyses fail to show statistical significance, these results should be viewed as "inconclusive" rather than "negative".

Consensus Summary

Disease-free survival, when considered in general, often predicts for an overall survival advantage but does not necessarily do so in all malignancies and with all therapies. Where there is a large DFS gain for the experimental arm, this may predict for OS, but small benefits may not predict. Where salvage therapies on relapse are good, OS may be impacted. Chronic toxicities from the adjuvant therapy may negatively impact OS.

DFS is a reasonable endpoint and is likely a surrogate for OS when the difference between the experimental and control arms is large. For experimental treatments that have significant toxicity and where a DFS advantage is small though statistically significant, longer follow-up and supplementary evidence for benefit in addition to DFS may be required in the absence of a demonstrated OS benefit.

Question 3:
HealthCanadaoften receives submissions with early data. If HC had received only the 3 yr DFS data from MOSAIC, Eloxatin may have been approved for both the Stage II and III indications. In light of the long data from MOSAIC, what other parameters should HC look at?

This observation highlighted the necessity for tightening requirements for "pharmaco-vigilance" and a requirement for follow-up of chronic toxicities. Performance status and quality of life studies were to be encouraged.
While companies are required to submit a pharmaco-vigilance Plan, this plan of follow-up will only document the known toxicities -- the unknown toxicities may be missed. It was recommended that clinicians must continue to be encouraged to report any unusual toxicities. The grass-roots clinician has the prime responsibility for observing and reporting all unusual toxicities.

Question 4:
Is the clinical benefit of Eloxatin from the MOSAIC study sufficiently robust to warrant an indication as proposed by the sponsor?

[Section 20(1)]

Question 5:
Should DFS be considered as a surrogate in adjuvant colorectal cancer?

Further discussion was held on this question and a request for clarification was received from the floor.

Generally, the opinion of the Committee was that DFS was "not inconsistent" with being a surrogate for overall survival. Where significant benefits in DFS are seen, DFS on its own might be acceptable with supportive data on survival (not necessarily "statistically significant") and toxicity. However, DFS cannot automatically be accepted as a surrogate of overall survival and depends on the specific disease and type of treatment being tested. A robust clinical benefit with full consideration of acute and chronic toxicities (which may impact on subsequent survival) and supportive overall survival data was also required.

DFS might be satisfactory for conditional approval, but follow-up data on OS with full consideration of the clinical benefit and chronic toxicities would be required for full approval. One problem is that Canada is a small regulatory jurisdiction and if larger jurisdictions do not ask for conditions for full approval, it is unlikely that these requirements would be carried out.

In summary, DFS as an endpoint might be acceptable for conditional approval, but full approval would require supportive data on survival with overall clinical benefit outweighing acute and chronic toxicities.

4. DFS in Hormone Response of Early Breast Cancer

A review of the adjuvant aromatase inhibitor (AI) studies, including ATAC, BIG-198, IES and MA-17.

Discussions then ensued on specific questions set by Health Canada.

Question 1:
If the observation is made that DFS does not result in a survival benefit, what is the most likely explanation for this?
a. OS are not yet sufficiently mature.

It was agreed that for colorectal cancer DFS, Relapse-Free Survival, Time to Recurrence, Time to Treatment Failure, etc., the definitions given in the article: "Endpoints in Adjuvant Treatment Trials: The Systematic Review of the Literature in Colon Cancer and Proposed Definitions for Future Trials" (Punt, C.J.A., et al. Journal of the National Cancer Institute, Vol. 1999- Issue 13, page 998-1003, July 4, 2007) was a reasonable starting point. Modification would be required for other disease sites such as breast cancer. It was emphasized that for breast cancer adjuvant trials, long follow-up was important.

In the absence of an overall survival benefit for extended adjuvant use of AI or AI as alternative to tamoxifen, the strategy of delaying initiation of AI therapy until recurrence of disease is reasonable. This allows a substantial proportion of patients who may not require adjuvant therapy to avoid acute and chronic toxicities. This choice revolves around the question: "Is it preferable to treat many people early for the benefit of a few who might recur, when you can treat only the few that actually recur and still attain similar advantages in these few, had they been treated earlier?" Obviously, this strategy allows many patients to avoid the potential toxicity of early adjuvant treatment.

It was pointed out that in the adjuvant aromatase inhibitor trials, at five years the DFS advantage is 3-4%, and this is matched by a 3-4% increase in risk of fracture compared to treatment with Tamoxifen.

The explanation for the absence of a clear OS advantage in the adjuvant A.I. trials may well be that follow-up is not sufficiently mature. The Oxford meta-analyses tend to show that with follow-up beyond 10-15 years, DFS does appear to emerge as a surrogate for OS, although this is not always the case.

b. DFS is not a good surrogate for OS

Distant DFS may well be a surrogate for OS in the adjuvant therapy of breast cancer but more research and follow-up is required to confirm this. If a patient suffers a distant recurrence, the patient will die of the disease (in the absence of other earlier non-cancer mortality, such as a cardiac event) whereas for local and/or regional recurrences, this may not be the case. Good salvage therapies exist for local and regional occurrences in breast cancer, and this may be one reason why DFS is not a clear surrogate for early OS in the setting of hormone therapy.

On the whole, it is premature to say that DFS is not a surrogate since there may well be a tail of long-term survivors. Cross-over from control to experimental arm in the clinical trials makes it difficult to show OS advantages. In the case of the adjuvant aromatase inhibitor studies, early un-blinding of the trials makes an OS analysis very difficult.

Indications for extended adjuvant therapy have been given with a qualifier of duration of effect. For example, the indication may be given with the qualifier that "evidence is for 26 months of follow-up" etc. Safety data in the indications are always updated.

Question 2:
If DFS is not a surrogate for OS or cannot be demonstrated to be such, should DFS be considered an endpoint reflecting a real benefit (being disease-free for longer) and thus, a basis for full, unconditional approval?

The members felt that DFS was still a valuable endpoint, despite the reservations indicated in the previous section. However, full information on acute and chronic toxicity was required. Quality of life assessments and health economics are also important considerations here. Health Canada does label clearly all the risks in language understood by patients. The final analysis includes all safety data.

Question 3:
Are there better endpoints available that reflect a direct clinical benefit? Would distant DFS be more reasonable given the different prognosis after local versus distant recurrence?

It was generally felt that "distant DFS" may be a better endpoint and surrogate for OS. DFS can be a useful endpoint in the absence of an OS benefit, but where possible, distant DFS should be available as well.

It was agreed that DFS is not a good surrogate for early OS, but longer term follow-up may show that it does predict.

5. Progression-Free Survival as an Efficacy Endpoint for Approval of Targeted and Chemotherapeutic Agents for Advanced Cancer

Conflict of Interest statements were signed, received and considered satisfactory.

The Bureau of Metabolism, Oncology and Reproductive Sciences (BMORS) made a presentation outlining problems associated with "progression-free survival" as an endpoint in trials of "targeted" therapeutic agents and chemotherapeutic agents for advanced cancer. Generally, the definition of progression-free survival ("PFS") has been time from randomization in the particular clinical trial to progression of disease or death. In other words, PFS is equivalent to "time to progression" plus deaths occurring.

Prior to 2001 no submissions were approved based on progression-free survival, but in the last two years, six submissions have been received, with progression-free survival as the primary endpoint. Progression-free survival is not affected by cross-over of the experimental treatment to the control group. It does capture events which are important in the trajectory of the cancer therapy.  The events are based on objective measurement but which include the entity of "stable disease".

Progression-free survival has not been statistically validated as a surrogate for OS and might be subject to assessment bias in open-label trials. Follow-up intervals vary and may even vary within the arms of a trial. There is no consensus on surveillance protocols for particular diseases or certain types of drugs, and definitions vary among studies. There are different methodologies for handling missing data and differing censoring methods.

Question 1:
Does progression-free survival constitute a direct measure of clinical benefit and/or is it a surrogate for clinical benefit?

One risk is the "self-fulfilling prophecy" -- that the justification for cross-over from the control to the experimental arm on progression of disease makes the assumption that OS is impacted positively.

Chemotherapeutic agents have traditionally been assessed by tumor shrinkage (response rate), indicating that there is pharmacological activity of the drug but "targeted agents" don't necessarily cause tumor shrinkage, although they may slow or halt their progression. PFS may be an endpoint reflecting an assumed mechanism of action.

Question 2:
Should approval be granted to a new chemotherapeutic agent based on a statistically significant increase in progression-free survival compared with that of the standard therapy but with weak evidence of efficacy based on an objective response rate (ORR)?

In discussion, the members agreed that this was a difficult topic. The paper "Early Stopping Rules: Who's Protecting Whom" discussed the issue well. (Cannastra, S.A., "Journal of Clinical Oncology". Volume 22, No. 9. May 1, 2004).

The Committee felt that for new chemotherapeutic agents associated with significant toxicity, progression-free survival on its own might not be compelling enough without some further evidence of a clinical benefit. The example of gemcitabine in carcinoma of the pancreas was discussed where a small improvement in time to progression and overall survival was documented, but -- of importance -- a clinical benefit in symptom relief was seen. For chemotherapeutic agents, unless the difference in progression-free survival is large, further evidence of clinical benefit (such as overall response rate, quality of palliation, quality of life, lack of toxicity, easier mode of administration) should be required for full approval.

Question 3:
For treatments of palliative intent, can a significant improvement in progression-free survival on its own be the basis of full approval (NOC) of a targeted agent, or should progression-free survival be considered as "promising evidence of clinical benefit" (NOC-C)?

If considered for conditional approval, what additional endpoints would confirm efficacy (overall response rate, survival, quality of life and symptom improvements, etc.)?

In specific circumstances where there is an "unmet medical need," progression-free survival evidence may be sufficient where an agent has limited toxicity. In other clinical situations where there are alternative treatments, the members felt that a tangible clinical benefit was preferred. For chemotherapeutic agents, an example might be an oral medication which has equivalent efficacy and toxicity to systemic intravenous medications, but has the obvious advantage of being given orally. For chemotherapeutic agents, progression-free survival on its own would be insufficient grounds for approval, unless the magnitude of the progression-free survival was truly impressive.

For targeted therapies, members of the Committee felt that with a small progression-free survival advantage being presented on its own and with no OS data, no quality of life data, no quality of palliation data or response data being presented - this may not be sufficiently compelling evidence for approval. Supporting data such as performance status, responses, quality of life, low toxicity, improved mode of administration, and quality of palliation, is preferred.

Members of the Committee pointed out that trials of targeted agents in advanced disease can be extremely difficult to assess. When a clinician has seen an individual patient respond with little toxicity, that in itself can be an impressive though limited piece of evidence.

If the progression-free survival is significant in the context of the disease being assessed, then an agent could be considered for an NOC/C, but follow-up would have to be requested to demonstrate that there was a significant clinical benefit to patients.

It was pointed out that there are examples of studies where if a targeted agent for a particular disease were to have been conducted in specific patient subgroups, then the response rate could be quite high, and that such patients would not have access to such treatments if that is not recognized. In diseases with no treatment available, then progression-free survival or even response rate may be reasonable to achieve approval, but in other diseases where there are more treatment options, progression-free survival may not be good enough and clear evidence of clinical benefit might be required. Generally, if the difference in progression-free survival is large enough, patients will feel better - and this is measureable.

It was pointed out that "targeted therapies" can be quite toxic and, in some instances, may be as toxic as chemotherapeutic agents.

Concern was expressed that the bar for approval may be getting lower and that a line may have to be drawn.

Question 4:
What should be the phase IV commitments of targeted agents registered with progression-free survival as the primary endpoint, where patients were allowed to cross-over to the new drug arm, based on a statistically significant progression-free survival benefit?

It is often difficult to replicate the pivotal study - recruitment to trials of this nature can be hard.

In general, small benefits in progression-free survival will not show an OS benefit (although non-small cell carcinoma of the lung might be an exception). Statistical significance does not necessarily mean clinical benefit, but if approval occurs in other jurisdictions, it is extremely difficult to do confirmatory trials.

In assessing agents for approval, the quality of the trial must be considered - good cooperative group studies with the data open to everyone's surveillance will likely be favored. In clinical situations where the action of a drug is biologically implausible or an unusual result occurs in the trial, then it is reasonable to demand a confirmatory study.

In summary, the acceptance of progression-free survival as the only endpoint depends on the disease being treated, the existence of alternative therapies, the magnitude of the PFS difference, and the types of patients being treated on the clinical trial (i.e. - are they end stage or are they early metastatic?). Where a small benefit in progression-free survival is seen, further evidence of clinical benefit will be required for approval.

5. Next Steps, Closing Remarks, and Adjournment of Meeting

There was a suggestion that if some members have ideas of possible subjects they'd like to address, they could forward these on to the Coordinator and these could be added to the Agenda. 

The date for the next meeting was discussed. It was agreed that mid to late March would be best.  Members will be canvassed to confirm the date.

Meeting adjourned.

Created: September 28, 2007
Revised: October 15, 2007

Approved by Chair: November 22^nd 2007