|
|
|
|  |
||||||||
 |
   
| |||||||||
| ||||||||||
| Investigator's Handbook A Manual for Participants in Clinical Trials of Investigational Agents
|
|||||||||
Abstract
Expeditious clinical development and approval of new drugs that are beneficial to patients are matters of high priority. There has been a great deal of discussion within the oncology community about what should constitute evidence of effectiveness of new anticancer agents for purposes of drug approval. This commentary is intended to illustrate a variety of end points that can lead to approval of new anticancer agents for specific clinical situations. Although the ultimate hope of antineoplastic therapy is prolongation of life, there are other effects of anticancer drugs that constitute clear clinical benefit and represent evidence of effectiveness. The guiding principle is that the beneficial effects obtained from a new drug should sufficiently outweigh the adverse effects such that the potential risk: benefit ratio achieved by an individual patient is favorable. The assessment of a new drug should flexibly evaluate safety and efficacy in the context of the specific clinical condition being treated. Early discussions with the Food and Drug Administration (FDA), http://www.fda.gov/, and the National Cancer Institute (NCI), http://cancer.gov, are recommended to identify prospectively the end points and trial designs needed to demonstrate effectiveness of a new drug. The general principles discussed will likely apply to the drug approval process for other medical disciplines as well.
Introduction
Expeditious clinical development and approval of new anticancer drugs that are beneficial to patients are matters of high priority. There has been a great deal of discussion within the oncology community about what should constitute evidence of effectiveness of new anticancer agents for purposes of drug approval. This commentary represents the efforts of a Working Group composed of Food and Drug Administration (FDA) staff in the:
It is intended to illustrate a variety of end points that can lead to approval of new anticancer agents for specific clinical situations.
Assumptions
The precise labeling indication being sought by the sponsor strongly influences whether comparative trials are needed before new drug approval. The new therapy must be compared with standard treatment when the labeling indication sought by the sponsor refers to a patient population where a standard therapy produces cures or significantly prolongs life. In such cases, comparative trials should be done to assess the risks of giving the new therapy, since it may be less effective than standard therapy.
For labeling indications that refer to the treatment of tumors for which there is no life-prolonging or curative therapy, approval of a new agent may not require comparative trials. Comparative trials of significant medical interest can and should be performed in the postmarketing period. For example, it is often possible to establish the usefulness of an agent in refractory or intolerant patients with relatively small studies well before evaluating the role of the agent as front-line treatment. The expeditious path to marketing a new agent may be for sponsors to focus initially on labeling indications for refractory or intolerant patients for whom there is no life-prolonging or curative therapy. This approach was taken in developing etoposide, ifosfamide, carboplatin, and fludarabine. Comparative efficacy trial conducted in the postmarketing period can then address other medically important questions and support other labeling indications.
Approaches to the Assessment of Net Benefit to the Treated Population
The following are acceptable end points for demonstrating effectiveness.
Survival Benefit:
Effectiveness is clearly shown when a new agent imparts a clinically significant survival advantage to the treated population. Such a therapeutic effect is generally best demonstrated in randomized, controlled trials (RCTs), although the effects may even be apparent through comparison with historical controls (e.g., etoposide for the second-line and ifosfamide for the third line treatment of metastatic germ cell tumors).
Time-to-Treatment Failure and Disease-Free Survival:
Improvement in time-to-treatment failure (TTF) or disease-free survival (DFS) in the adjuvant setting is generally best shown in a RCT and can be a reasonable basis for concluding that an agent is effective. For example, in breast cancer a large fraction of recurrences are symptomatic, making improved DFS a valid surrogate for improved quality of life (QOL).
Complete Response Rate:
For a variety of malignancies, an increased rate of durable complete responses (CRs) correlates with increased rates of survival and cure. Thus, reproducible, carefully documented CR rates of reasonable duration are persuasive indicators of effectiveness, particularly if there are no other effective options. On the other hand, when curative, standard therapy exists, it is important to ensure that the CR rate of the new drug translates into a comparable effect on survival.
Response Rate:
It may be argued that virtually any drug should be considered effective if it produces a response rate above some arbitrary level (e.g., 20% for may solid tumors). If one were to accept such thresholds, it seems clear that the level should be a function of the tumor type and stage. For example, the threshold response rate for previously untreated indolent lymphoma might be considerably higher than that for renal cell carcinoma or melanoma. In general, however, it is not meaningful to consider response rate without also considering duration of response, the general level of toxicity, and effects on QOL.
Beneficial Effects on Disease-Related Symptoms and/or Quality of Life (QOL):
QOL may be improved if treatment decreases symptoms of disease, as measured directly or with validated QOL instruments. Parameters of QOL could include improvement in tumor-related signs or symptoms, improved physical or psycho logic function, decreased reliance on medical support, gain in lean body mass, etc. The usefulness of QOL instruments in demonstrating patient improvement is a matter of considerable interest. Improved QOL may be demonstrated in two kinds of studies:
Comparison with control therapy in a Randomized, Controlled Trial(RCT):
For a new drug with substantial toxicity, net clinical benefit can be best shown through a RCT. If the improvement in QOL achieved with the new treatment is at least equivalent to a control treatment also shown to enhance QOL, then the new treatment can be considered effective. On the other hand, the new drug would have to provide superior QOL if compared with a control therapy that did not result in improved QOL. In situations where life-prolonging standard therapy exists, survival would also have to be assessed, as one would not usually sacrifice survival for superior QOL. However, once the comparative impact on survival is clear, individual patients and physicians can consider the relative importance of symptom relief and QOL versus survival.
Comparison of patient's post treatment status with pretreatment status:
In patients who have well-defined symptoms or signs that confidently can be attributed to their tumors, disappearance or amelioration of the symptoms may be a meaningful result. This design has been used recently for interferon alpha in the treatment of hairy cell leukemia; reduction in transfusion requirements and infection rates were used to document treatment effectiveness. Another example is the somatostatin analogs, which reduce symptoms from hormone secretion in islet cell/carcinoid tumors. This trial design is also suitable for evaluating QOL end points in the treatment of rare, symptomatic tumors, where accrual to larger, randomized studies would take a very long time. Because this approach can be confounded by drug toxicity, it is best suited to the evaluation of relatively well-tolerated regimens. This design could not detect a treatment-induced slowing of clinical deterioration because it lacks a comparison group.
Illustrative Examples
It is assumed in these examples that
These examples show that the expected benefit from a treatment effect depends on the clinical circumstances in which it occurs. In all cases, the guiding principle is that the beneficial effects obtained should sufficiently outweigh the adverse effects, such that the potential risk: benefit ratio achieved by an individual patient is likely to be favorable. The assessment of a new drug should flexibly evaluate safety and efficacy in the context of the specific clinical condition being treated.
Examples Demonstrating Effectiveness in Clinical Settings where Life-Prolonging or Curative Standard Therapy Exists
Example 1:
A new drug (use singly or in replacing a drug in a standard combination) for front-line treatment of advanced testicular cancer produces identical complete response (CR) rates compared with standard therapy. Data are not yet available comparing survival experience. Toxicity with the new drug is less.
Comments:
To conclude that the new drug is effective for this indication when standard treatment is known to be curative, adequate assurance of comparable 3-year survival and freedom from relapse are required. If the similar response rates correlated with similar survival in one study, conclusions from the confirmatory study could be based on CR rates alone.
Example 2:
Drug B is an analog of parent drug A (such as cisplatin). It is being developed for the treatment of the same tumors for which drug A is used, as well as for others. It differs from A in that some toxicities are reduced. Drug B may or may not be completely cross resistant with drug A.
Comments:
Comparative trials of A and B may or may not be necessary. If A (the parent) is clearly effective in improving survival in a particular tumor type and stage, and B (the analog) is being developed as a substitute for the parent for the same type/stage of disease, comparative trials are necessary. Results should show similar effects on survival or well-established surrogate end points for survival. A somewhat poorer result for the new drug, still yielding evidence of a survival benefit, might support (with accurate labeling) availability for people who are not willing to accept the side effects of the parent drug. For people intolerant of the parent drug or who have a medical condition that precludes the use of the parent drug, evidence of effectiveness of the analog could be shown in noncomparative studies in drug A-intolerant patients. Acceptable end points of effectiveness would include a good rate of durable CRs or improvement in QOL.
Members of the Working Group did not agree entirely on how to apply these concepts to the approval process in diseases for which standard therapy produces profound degrees of cytoreduction, high remission rates, impressive patient benefit, modest improvements in DFS and overall survival, and a small (10% - 20%) cure rate. Examples include previously untreated advanced ovarian cancer and acute myeloblastic leukemia (AML). A majority of the group favored the position that new agents (whether or not they are analogs) labeled for treatment of such conditions should demonstrate equivalent prolongation of survival or equivalence of accepted surrogates for survival (e.g., durable pathologic CR rates in ovarian cancer). The approval criteria for new drugs for these situations should then differ from approval criteria for other conditions, such as estrogen receptor-negative metastatic breast cancer and extensive small-cell lung cancer, where standard therapy perturbs the natural history of the disease to a much less extent and rarely, if ever, produces cures. In these latter diseases, comparative trials would not be required to demonstrate effectiveness for an active new agent, if that agent produced a significant durable CR rate or improved QOL.
A minority of the group favored the position that if the outcome of standard therapy for diseases such as previously untreated advanced ovarian cancer and AML was sufficiently poor, comparative trials should not be required. If a significant number of durable CRs (comparable with the number achieved with standard therapies) could be demonstrated in previously untreated patients in single-arm studies, a new drug could be considered effective. These members would accept new agents for front-line treatment of these conditions based on reproducible and substantial activity and safety in Phase 2 trials in previously untreated patients.
These are differing views of the issue, "How good are the standard therapies for advanced ovarian cancer and AML?" For cancers where standard therapies are strictly palliative, comparative trials are generally not needed to show effectiveness of a new drug or analog. On the other hand, for cancers where standard therapy is largely curative (e.g., testicular cancer and Hodgkin's disease), there is general agreement that comparative efficacy trials should be done. For clinical conditions between these two extremes (e.g., standard therapy cures a small minority of patients, provides a modes improvement in survival, or improves QOL to a limited extent), the threshold for deciding that a standard therapy is "good enough" to require comparative studies is a matter of judgment and ongoing discussion.
However, there was strong agreement that for clinical conditions where standard therapy clearly cures, equivalence of a new drug should be demonstrated in comparative trials before approval. Because such studies can require many years to complete, approval of a promising new drug can also be sought for the treatment of refractory cancers based on Phase 2 studies.
If B (the analog) can induce CRs of good duration or improve tumor-related symptoms in patients whose tumors have progressed on A (the parent), then B would be considered effective for second-line treatment.
If B is studied for conditions not known to be responsive to A (whether A has or has not been studied), comparisons with A are not needed.
Example 3:
A new drug is compared with the combination of fluorouracil (5-FU) and levamisole for adjuvant treatment of stage III colon cancer. Toxicity is moderate and is similar for the two arms. DFS and point estimates of survival appear equivalent, but at present there are insufficient data to assess equivalence or inequivalence of survival at a statistically significant level. Twenty-five percent of the expected deaths have occurred on the control arm.
Comments:
The new drug cannot be considered effective at present. Equivalence of survival between the new drug and 5-FU/levamisole has not been established at a statistically significant level. Since the combination of 5FU-levamisole has been shown to increase long-term survival compared with no treatment, it should not be replaced with a treatment that might give inferior survival. In general, DFS is an acceptable surrogate for survival provided it reproducibly correlates with improvement in survival for a particular patient population. To date, few studies have shown survival improvement in stage III colon cancer; therefore, it is not yet possible to assess whether DFS can be used as a surrogate for survival. If DFS could be shown to be a reliable surrogate for survival in this disease, then DFS equivalence would be a basis for concluding that the new drug is effective.
If, in this example, the new drug produced a superior reduction in recurrence rates and an improvement in DFS that were statistically significant, the new drug would be considered effective. It could be assumed that survival with the new drug would be at least equivalent to the combination of 5-FU/levamisole. Follow-up of survival would need to be continued and reported after marketing.
Examples Demonstrating Effectiveness in Clinical Settings where No Standard Life-Prolonging or Curative Therapy Exists
Example 4:
An antiestrogen produces a response rate of 30% in previously untreated patients with post-menopausal, estrogen receptor-positive (ER+), metastatic breast cancer (compared with 50% to 60% for tamoxifen). The response rate is 10% to 15% in patients who have previously responded to and then progressed on, all other hormonal therapy. Toxicity is minimal.
Comments:
If the labeling indication centered on hormonally refractory patients, these results would support the demonstration of net clinical benefit if patients benefit could be shown. There may be other instances in which the magnitude, quality, and/or duration of the objective responses are remarkable and the demonstration of symptomatic improvement may not be necessary, but that is not the case for a 10% to 15% response rate. However, for a labeling indication referring to previously untreated ER+ patients, comparison with tamoxifen in an RCT should be required. The 30% response rate in the previously untreated population would not prove effectiveness by itself. Indeed, the 30% response rate suggests an unacceptable risk: benefit ratio for this indication, since the new drug appears significantly less effective than the established agent and has no other advantages.
Example 5:
A new drug for symptomatic chronic lymphocytic leukemia (CLL) refractory to standard therapy produces a response rate of 50%, with 15% CRs. The duration of response is approximately 1 year for the CRs and 6 months for the partial responses (PRs). Considerable Phase 2 experience also shows the new drug to have a reproducible 80% response rate, with 25% CRs in previously untreated patients with CLL and with response durations comparable to those achieved with standard therapies. Toxicity is mild.
Comments:
Such a drug would be considered effective for treatment of CLL refractory to standard treatment, because of the significant rate of durable CRs. In addition, because of the reproducible, high, durable response rates in previously untreated patients, and the mild toxicity (comparable to treatment with standard therapy), this new drug would also be considered effective for front-line treatment of CLL without the need for comparative efficacy trials. Standard treatment for CLL is not curative and is not known to prolong survival, but is believed to have a favorable impact on QOL. Therefore, reasonable assurance that a new drug imparts comparable net patient benefit is a legitimate basis for demonstrating effectiveness. Comparative efficacy trials of medical interest can and should be conducted in the postmarketing period. Had the new drug been associated with significant toxicity and not had any obvious advantage over standard therapy, however, a comparative trial would usually be needed to assess potential differences in QOL and survival.
Example 6:
A new drug has a reproducible, 40% to 50% response rate in adequately sized trials in previously untreated metastatic breast cancer. There are 10 to 15% unequivocal CRs; median duration of the CRs is 10 months. The patient population studied is representative of the general population of patients with metastatic breast cancer. Toxicity is modest; moderate nausea and vomiting and grade 3 myelosupression in 80% of patients. Patients who respond experience improvement in their disease-related symptoms; nonresponders are not adversely affected to any significant degree.
Comments:
This example is potentially controversial because it depends on one's view of the adequacy of available treatment for metastatic breast cancer. In general, however, this new drug appears to confer net clinical benefit based on the durable CR rate and modest toxicity, and because the response rate and CR rate are not notably inferior to alternative standard therapies. Standard combination chemotherapy has not been clearly shown to increase survival, and cures are not expected. Therefore, comparative trials with standard regimens for metastatic breast cancer would not necessarily be required for approval. These comparative efficacy trials, however, are of significant clinical interest and should be conducted along with pivotal marketing trials, or immediately thereafter. On the other hand, if this new drug were associated with more severe, potentially life-threatening toxicity, or if the CRs were equivocal or less durable, comparison with a standard regimen would usually be needed to assess whether treatment resulted in inferior QOL or survival.
Example 7:
A new drug produces a 20% response rate (all PRs) in metastatic renal cell carcinoma, with a median duration of 4 months; no PR lasts over 6 months. Treatment-related toxicity includes severe, refractory nausea and vomiting, lasting several days after each dose.
Comments:
It is unlikely that the data would support a claim of net patient benefit. If patients with symptomatic liver, lunch, or brain disease experienced decreased symptoms, perhaps net clinical benefit could be shown. It seems more likely, however, that the short response durations and severe nausea and vomiting would preclude a determination that the risk: benefit ratio is favorable.
Example 8:
A new drug has a reproducible 30% response rate in metastatic renal cell carcinoma, including 10% to 15% durable CRs (median duration of the CRs is 12 months). Toxicity is modest.
Comments:
Effectiveness is clear. The durable CRs strongly suggest that patient benefit has resulted from the new drug. In light of the modest toxicity, the CRs provide a basis for concluding that there is a net benefit.
Example 9:
A cytotoxic agent has activity against Hodgkin's disease refractory to mechlorethamine, vincristine, procarbazine, and prednisone/doxorubicin, bleomycin, vinblastine, and dacarbazine (MOPP/ABVD). The associated response rate is approximate 35%. (PRs only). The median duration of response is 5 months, with no response lasting longer than 8 months. Toxicity is moderate: nausea and vomiting last for 2 to 4 hours in 60% to 70% of patients, grade 3 myelosupression in 75%, and urticaria in about 7%.
Comments:
The drug would be considered to have a favorable risk: benefit ratio for use as salvage therapy if the data show net clinical benefit. Comparison with standard therapy in the salvage setting would not necessarily be required, since there is no established standard therapy for this patient group. Data should show that the 30% response rate, with the associated toxicities, translates into overall benefit for the treated population. One way to show this would be to demonstrate improved symptom/sign control in responders compared with pretreatment status, without strong adverse effects in the nonresponders.
There are important advantages to performing RCTs in this setting, especially if the therapy is toxic. Comparing such a new drug with a second therapy believed to have comparable antitumor activity can aid the demonstration of net patient benefit by providing a basis for comparison of response rates, toxicity, and QOL measurements. Also, a comparative trial could detect a beneficial drug-induced slowing of symptom progression; this could not be definitively demonstrated in a single-arm trial.
Examples Demonstrating Effectiveness of New Agents Aimed at Reducing Treatment-Related Toxicities
Example 10:
A hematopoietic growth factor is shown in an RCT to reduce the duration of neutropenia in small-cell lung cancer (SCLC) patients treated with aggressive combination chemotherapy. The duration is reduced from an average of 7 days to an average of 4 to 5 days. This difference results in significantly fewer episodes of febrile neutropenia and hospitalization. Toxicity from the growth factor is mild (fevers, transient bone pain, skin rashes). There is no significant difference in response rates between the two arms.
Comments:
The shortened duration of neutropenia with consequent fewer episodes of febrile neutropenia and hospitalizations provides a basis for concluding that the growth factor is effective in preventing severe infectious complications of chemotherapy. Other patient benefit end points of interest include decreased need for antibiotics and/or shorter hospitalizations, decreased numbers of documented infections, and lessened mucositis.
Example 11:
The hematopoietic growth factor described above also allows delivery of a more dose-intensive treatment, i.e., a greater total dose per unit time or an improved ability to administer a fully intensive established dose. This is shown for extensive-stage SCLC patients in a randomized trial. The chemotherapy regimen is toxic but known to be effective in SCLC. The improved dose intensity is due to quicker recovery of the WBCs after chemotherapy and fewer cycle delays. Again, toxicity of the growth factor is mild. Patients receiving the more dose-intensive regimen have significantly higher CR rates (35% vs. 20%; durations are comparable), which historically have been shown to correlate with survival in this disease. Median TTF is 10 months for the growth factor arm versus 6 months for the chemotherapy-alone arm.
Comments:
The known correlation of CR rate with overall benefit-in this case, survival-supports the conclusion that the growth factor enhances the effectiveness of chemotherapy. Significant improvement in TTF or QOL would also be a justifiable basis for showing effectiveness.
Example 12:
In a randomized trial, a chemoprotector is shown to reduce the incidence of nephrotoxicity and ototoxicity of a known effective regimen in advanced ovarian cancer. Toxicity of the chemoprotector is mild. The two arms are equivalent in overall response rates and in pathologic CR rates.
Comments:
Reduced chemotherapy-induced nephrotoxicity and ototoxicity with no adverse effects on response rates or overall toxicity form the basis for the demonstration of effectiveness. Data on survival should be submitted during Phase 4 (postapproval).
Acknowledgements
The authors thank Dr. Louis Lasagna, Chairman of the National Committee to Review Current Procedures for Approval of New Drugs for Cancer and AIDS; Dr. Craig Henderson, chairman of the FDA Oncologic Drug Advisory Committee; and Dr. Fredrick Appelbaum and Ernest Borden of the FDA Biological Response Modifiers Advisory Committee for their review and thoughtful comments.
Appendix
This commentary was written as a white paper in February 1988 by the following members of NCI's DCT Board of Scientific Counselors (BSC) Ad Hoc Review Committee for FDA New Drug Approval: Drs. Martin Abeloff, Robert Capizzi, Lawrence Einhorn, B.J. Kennedy, John Laszlo, Brigid Leventhal, Franco Muggia, John Niderhuber (Chairman), Robert Wittes, and Robert Young. This initial white paper was approved by the full DCT BSC in February 1988 and became the basis for ongoing discussions between the NCI and the FDA. In 1990, this document was expanded and put into its final form by the NCI/FDA Working Group. Substantial contributions to this effort were made by Drs. Samuel Broder, Michael Grever, Michael Hawkins, Carl Peck, Jay Siegel, and Kathryn Zoon. A subcommittee of the DCT BSC consisting of Drs. Paul Carbone (Chairman), William Hryniuk, Loretta Itra, and Donald Kufe reviewed and revised the commentary, which was then approved by the full DCT BSC in February 1991. This commentary has also been reviewed by FDA's Office of Health Affairs and Center for Devices and Radiologic Health, http://www.fda.gov/cdrh/index.html.
    |
