Monday, May 27, 2013

The Size of Safety Database In Drug Development Program

In planning the clinical program for drug approval, in addition to the efficacy assessment, the adequate size of the safety database is usually required by the regulatory agencies. Unlike the efficacy assessment while the sample size can be calculated, the size of safety database is a little bit vague. The size of safety database is often included in the pre-IND  meeting with FDA because the sponsor would like to understand the expectation from FDA regarding the size of safety database. There could be situations that the sample size for demonstrating the efficacy is relatively small and the overall clinical development (cost) would largely depend on the size of safety database.

The reference guidelines regarding the size of safety database are mainly the ICH E1 (The Extent of Population Exposure to Assess Clinical Safety for Drugs Intended for Long-term Treatment of Non-Life-Threatening-Conditions) and FDA guidance “Premarketing Risk Assessment”. The guidance is pretty specific in terms of the size of safety database “For products intended for long-term treatment of non-life-threatening conditions, (e.g., continuous treatment for 6 months or more or recurrent intermittent treatment where cumulative treatment equals or exceeds 6 months)

the ICH and FDA have generally recommended that 1500 subjects be exposed to the investigational product (with 300 to 600 exposed for 6 months, and 100 exposed for 1 year).  For those products characterized as chronic use products in the ICH guidance E1A, FDA recommends that the 1500 subjects include only those who have been exposed to the product in multiple dose studies, because many adverse events of concern (e.g., hepatotoxicity, hematologic events) do not appear with single doses or very short-term exposure. Also, the 300 to 600 subjects exposed for 6 months and 100 subjects exposed for 1 year should have been exposed to relevant doses (i.e., doses generally in the therapeutic range)
 
In product label, the corresponding sample database description is provided in FDA guidance “Adverse Reactions Section of Labeling for Human Prescription Drug and Biological Products — Content and Format
The data described below reflect exposure to drug X in [n] patients, including [n] exposed for 6 months and [n] exposed forgreater than one year. Drug X was studied primarily in placeboand active-controlled trials (n = __, and n = ___, respectively), and in long-term follow up studies. The population was [age range], [gender distribution], [race distribution] and had [diseases/conditions]. Most patients received doses [describerange, route of administration, frequency, duration, as appropriate].
 
However, for products intended for short-term or acute use and for products intended to treat life-threatening diseases, the guidance did not provide the specific requirement. The requirement of the size of safety database is usually based on the discussion with the review devision of the regulaory agency. FDA guidance “Premarketing Risk Assessment” did provide some general guidelines:

“Safety databases for products intended to treat life-threatening diseases, especially in circumstances where there are no alternative satisfactory treatments, are usually smaller than for products intended to treat diseases that are neither life-threatening nor associated with major, irreversible morbidity. A larger safety database may be appropriate if a product’s preclinical assessment or human clinical pharmacology studies identify signals of risk that warrant additional clinical data to properly define the risk. The appropriate size of the preapproval safety database may warrant specific discussion with the relevant review division. For instance, 21 CFR 312.82(b) (subpart E) provides that for drugs intended to treat life-threatening and seriously debilitating illnesses, end-of-phase 1 meetings can be used to agree on the design of phase 2 trials “with the goal that such testing will be adequate to provide sufficient data on the drug’s safety and effectiveness to support a decision on its approvability for marketing.”
 “For products intended for short-term or acute use ( e.g., treatments that continue for, or are cumulatively administered for, less than 6 months), FDA believes it is difficult to offer general guidance on the appropriate target size of clinical safety databases. This is because of the wide range of indications and diseases (e.g., acute strokes to mild headaches) that may be targeted by such therapies. Sponsors are therefore encouraged to discuss with the relevant review division the appropriate size of the safety database for such products. Because products intended for lifethreatening and severely debilitating diseases are often approved with relatively small safety databases, relatively greater uncertainty remains regarding their adverse effects. Similarly, when products offer a unique, clinically important benefit to a population or patient group, less certainty in characterizing risk prior to approval may be acceptable.”
 
It is no surprise if FDA provides the following guidance to the sponsor “FDA has generally accepted a minimum of 300-400 subjects as the safety database to support an indication in the intended use population(s) to rule out with 95% confidence an adverse event that occurs with a frequency of 1%. “

Apparently, the above statement requires statistical calculation (or beyond the statistical calculation). If we use the SAS Proc Freq, we can actually see how the number 300-400 is played out.

data samplesize;
  input scenario Event $ count;
  datalines;
  1 AE 1
  1 no 294
  2 AE 1
  2 no 299
  3 AE 1
  3 no 554
  4 AE 1
  4 no 559
run;

proc freq data=samplesize;
  weight count;
  tables Event /  binomial (p=0.01) alpha=0.05 cl; 
       **p=0.01 option indicates the standard rate to compare with,
         here we assume the AE rate of 1%;
       ** Alpha=0.05 to obtain two side 95% confidence interval;
  exact binomial;  *Obtain the exact p-value;
  by scenario;
run;

The scenarios 1 and 2 are based on the calculation of the asymptotic CI:
For N=295 and event=1, the 95% CI =(0.0000, 0.0100), which meant with 295 subject safety database, it would not be able to rule out an AE that occurs with a frequency of 1%;
For N=300, and event=1, the 95%CI=(0.0000, 0.0099), which meant with 300 subject safety database, it would be able to rule out an AE that occurs with a frequency of 1%

The scenarios 3 and 4 are based on the calculation of the exact CI:
For N=555 and event=1, the 95% CI =(0.0000, 0.0100), which meant with 555 subject safety database, it would not be able to rule out an AE that occurs with a frequency of 1%;
For N=560, and event=1, the 95%CI=(0.0000, 0.0099), which meant with 560 subject safety database, it would be able to rule out an AE that occurs with a frequency of 1%

Even though the exact confidence interval is more appropriate when dealing with such low frequency of AE. It may be inpractical to impose a safety database of 560 subjects treated with the testing product.

The size of safety database may depend on the specific area in development. For example, FDA has its requirement for safety database on vaccine product. Below is the specific requirement from FDA guidance “Guidance for Industry: Clinical Data Needed to Support the Licensure of Seasonal Inactivated Influenza Vaccines

“Safety data must be collected from subjects enrolled in pre-licensure clinical trials intended to support the accelerated approval of a new seasonal inactivated influenza vaccine (21 CFR 312.23, 312.32, 312.56, 312.60 and 312.62). The monitoring of these subjects should follow the outline for safety evaluations described in Section III.A.3. above. A total safety database large enough to rule out a serious adverse event that occurs at a rate of 1 in 300 may be sufficient when a sponsor has adequate marketing and safety experience with the same manufacturing process for a seasonal vaccine licensed outside the United States and these data are presented in the BLA and assessed as such. For example, the upper limit of the two-sided 95% CI of the true serious adverse event rate is 0.0032 ( less than 1 in 300) when no serious adverse event is observed among 1150 subjects who received vaccine in clinical trials, using the Clopper-Pearson method. However, the size of the pre-licensure safety database, especially for seasonal influenza vaccines manufactured using novel processes such as cell-culture and for seasonal influenza vaccines that contain novel adjuvants, would be influenced by factors such as the nature of the new manufacturing process and available pre-clinical and clinical data, and should be discussed with CBER.
Moreover, if a serious adverse event is present in a safety database of about 1,000 subjects, and there is concern that it may be vaccine-related, additional safety data may be needed. Safety data to support use in pediatric populations would also be needed and should be submitted either as part of the BLA, or as a clinical efficacy supplement at a later time, if pediatric studies are deferred under PREA (see Section III.C.4. - Pediatric Research Equity Act).”
 
Safety Database versus Safety Population

These two terms can sometimes cause confusion. Safety Database includes only the subjects who exposed to the testing drug and not control. Safety database is usually used for discussion of the entire clinical program (multiple studies). Safety Population is the term used in clinical protocol or statistical analysis plan pertinent to a specific protocol. Safety population defined as the subjects who received any dose of the study medication (includes both testing drug and control).

Monday, May 20, 2013

Biomarkers versus Surrogate Endpoints

“Biomarkers” and “Surrogate endpoints” are closely related, although a biomarker can serve as a surrogate endpoint, the terms of biomarkers and surrogate endpoints are not synonymous.

FDA Guidance for Industry “Qualification Process for Drug Development Tools” provided the clear definitions for biomarker and surrogate endppoint.

A biological marker or biomarker is defined as a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or biological responses to a therapeutic intervention.4 A biomarker can define a physiologic, pathologic, or anatomic characteristic or measurement that is thought to relate to some aspect of normal or abnormal biologic function. Changes in biomarkers following treatment may predict or identify safety problems related to a drug candidate or reveal a pharmacological activity expected to predict an eventual benefit from treatment. Biomarkers may reduce uncertainty in drug development and evaluation by providing quantitative predictions about drug performance.

A surrogate endpoint is defined as a biomarker intended to substitute for a clinical efficacy  endpoint. Surrogate endpoints are expected to predict clinical benefit (or harm, or lack of benefit or harm). A clinical endpoint is defined as a characteristic or variable that reflects how a patient feels, functions, or survives.
 
In the subsequent FDA public workshop “ Measurement in Clinical Trials: Review and Qualification of Clinical Outcome Assessments”. a glossary included the slight different definition for biomarker and surrogate endpoint.

Biomarker — A patient characteristic that is measured as an indicator of biologic processes; normal, pathogenic, or response to a therapeutic intervention. Patient characteristics that are classified as biomarkers are those that are not significantly influenced by rater judgment or patient motivation and effort (e.g., not a measure of a patient’s volitional performance of some defined procedure). Measurements that can (and therefore, should) have a sufficiently well defined procedure for measurement so that minor aspects of rater or technician involvement have no impact on the measurement are included within the category of biomarkers. Biomarkers are not psychomodulated measures.

Surrogate endpoint — An indirect outcome measure that is used as a substitute for a direct measurement of how a patient feels or functions. All biomarkers used as a clinical trial outcome assessment are surrogate endpoints or proposed as surrogate endpoints. The acceptability of a surrogate endpoint is dependent upon a demonstration that it can be used to reliably infer treatment benefit. The term is also sometimes applied to indirect psychomodulated measures to emphasize that they are indirect, and a substitute (replacement) for a direct measure of how a patient feels or functions.
 
In a paper by Strimbu and Tavel “What are biomarkers?”, biomarker and surrogate endpoint were defined as following:

Biomarkers are by definition objective, quantifiable characteristics of biological processes. They may but do not necessarily correlate with a patient's experience and sense of wellbeing, and it is easy to imagine measurable biological characteristics that do not correspond to patients' clinical state, or whose variations are undetectable and without effect on health. It is also even easier to imagine measurable biological characteristics whose variance among populations is so great as to render them all but useless as reliable predictors of disease or its absence.

In contrast, clinical endpoints are variables that reflect or characterize how a subject in a study or clinical trial “feels, functions, or survives”. They are, in other words, variables that represent a study subject's health and wellbeing from the subject's perspective.

When used as outcomes in clinical trials, biomarkers are considered to be surrogate endpoints; that is, they act as surrogates or substitutes for clinically meaningful endpoints. But, not all biomarkers are surrogate endpoints, nor are they all intended to be. Surrogate endpoints are a small subset of well-characterized biomarkers with well-evaluated clinical relevance. To be considered a surrogate endpoint, there must be solid scientific evidence (e.g., epidemiological, therapeutic, and/or pathophysiological) that a biomarker consistently and accurately predicts a clinical outcome, either a benefit or harm. In this sense, a surrogate endpoint is a biomarker that can be trusted to serve as a stand-in for, but not as a replacement of, a clinical endpoint.
FDA Guidance for Industry “Qualification Process for Drug Development Tools” provided the detail descriptions for the biomarkers.
Biomarkers include measurements that suggest the etiology of, susceptibility to, activity levels of, or progress of a disease. Alterations in biomarker measurements indicate responses  (favorable or unfavorable) related to an intervention. The biomarker may reflect biological processes closely related to the mechanism of action or processes substantially downstream. Biomarkers may assess many different types of biological characteristics or parameters, including
    • genetic composition (e.g., BRCA, HER2)   
    • receptor expression patterns
    • radiographic or other imaging-based measurements (e.g., progressive free survival (PFS), CT lung densitometry, thrombolysis through arteriogram,...)
    • blood composition measurements (e.g., serum enzyme levels, prostate specific antigen) electrocardiographic parameters
    • organ function (e.g., creatinine clearance, pulmonary function tests, cardiac ejection fraction)

Biomarkers can be categorized into three categories:

Prognostic biomarker
a baseline patient or disease characteristic that categorizes patients by degree of risk for disease occurrence or progression. A prognostic biomarker informs about the natural history of the disorder in that particular patient in the absence of a therapeutic intervention.
Predictive biomarker
a baseline characteristic that categorizes patients by their likelihood for response to a particular treatment. A predictive biomarker is used to identify whether a given patient is likely to respond to a treatment intervention in a particular way. It may predict a favorable response or an unfavorable response (i.e., adverse event).
pharmacodynamic (or activity) biomarker
a dynamic assessment that shows that a biological response has occurred in a patient after having received a therapeutic intervention. A pharmacodynamic biomarker may be treatment-specific or more broadly informative of disease response. Examples include blood pressure, cholesterol, HbA1C, intraocular pressure, radiographic measures, and C-reactive protein.



  • Surrogate endpoints are a subset of pharmacodynamic biomarkers; it is likely that only a few biomarkers will be appropriate for use as surrogate endpoints. In other words, only pharmacodynamic biomarker may be qualified as surrogate endpoint.  
  • Prognostic biomarker can be very useful in sub-group analyses.
  • Predictive biomarker can be very useful in designing the enrichment designs.

In Summary,          
  • Surrogate endpoints are biomarkers
  • Not all biomarkers can serve as surrogate endpoints
  • Surrogate endpoints are a subset of pharmacodynamic biomarkers;
  • Surrogate endpoint should predict/be correlated with clinical endpoint, but correlation alone is insufficient