Jerri B Perkins, MD
Perkins & Perkins, Inc.
This question has been posed to me on numerous occasions.
As a former FDA Medical Officer, I will give an insider's perspective
on reviewing an NDA. I have been consulting for both the pharmaceutical
and medical device industries for the past 10 years. Prior to
that, I was a Medical Officer for the FDA for eight years.
Key sections for the medical reviewer are the summaries of safety,
efficacy, and risk/benefit. Were the studies well conceived,
well designed, well written, and well presented to the Agency?
Only sound scientific data will suffice.
The Label
The label is where a company should begin and end. Do the
studies match the label? What is the product and its intended
function? Is it new, a breakthrough product, or a "me-too?"
What is the significance of this product? Will it save lives
(these are rare)? Is it more convenient (therefore, more compliance
might be expected?) Is it safer than an alternative therapy?
The key question here is, should this product be on the market?
Does the benefit outweigh the risk?
Safety First
Perhaps one of the best illustrations of FDA safety concerns is illustrated in a clip from The Washington Post, February
3,1995 (below). This emphasizes the importance of risk/benefit.
It is obvious that the FDA's primary goal is to keep a potentially
unsafe product off the market rather than withdraw it later because
of a question of safety.
What are the safety issues with the products? It is rare that
there are no safety concerns. The most appropriate question here
is, what is the risk/benefit allowable with this product? For
a product with marginal benefit, little, if any, risk would be
tolerable. Whereas, a lifesaving product may have more allowable
risk.
Adverse Events (AEs)
What do the AEs look like? It is very difficult to second guess
the physician on site caring for the patient. If the physician
believes the adverse event (conversely is not) product-related,
that opinion is very carefully evaluated. Conversely, if each
individual physician sees no correlation between the product
and AEs, the company may see trends that might only be apparent
overall and not on a case-by-case basis.
Were there any statistically significant AEs? What are they?
How often do they occur? Are they dose-related? In what patient
population are they seen? If not statistically significant, are
there trends? Are there drug/drug interactions? Are there laboratory
values outside the normal ranges (and there should be, otherwise
the question arises, is thc data too perfect)? Finally, are there
sufficient numbers of patients studied that would detect unacceptable AEs?
Adverse events cannot be over-emphasized. How were they originally
defined in the study? What was the investigator's interpretation
and the company's interpretation? Are there differences? If so,
explain. The key question for a company should be, can the AEs
be reduced by cutting back the dose and maintaining efficacy?
Dangerous Medicine
Between 1970 and 1992, 56 drugs were removed from the market
in France, Germany, Britain or the United States because of safety
problems.
COUNTRY
|
DRUGS MARKETED AND WITHDRAWN 1970-1992
|
| France |
31 |
| Germany |
30 |
| Britain |
23 |
| United States |
9 |
NOTE: Some of the 56 drugs were withdrawn in more than one
country.
SOURCE: Public Citizen Health Research Group
Efficacy Essential
What studies are available to document efficacy? Are there two*
"adequate and well controlled" trials, and were they
well conducted? Was a placebo used? If not, what was the comparison
study? How many of the trials were pivotal? How many studies?
How many patients? Is there a dose response? Are there hard end
points? Most importantly, is there statistical significance?
The above elements should be easily readable in a good summary
of the submission.
The next focus is on the studies conducted. Did the study protocol
follow FDA current guidelines and current FDA policy? Was the
protocol well-designed and strictly followed? Was the study a
randomized, double blind, placebo, prospectively controlled trial?
If not, why not? AIDS, oncology, and anti-infective drugs are
examples of studies that are not placebo trials. What patient
populations were studied? Were women, pediatric, and geriatric
populations included? If not, why not? Is the statistical design
adequate?
Were the studies well conducted and adequately monitored? For example, if 50 percent of the patients dropped out, why? Was the
study poorly conducted? Was the drug toxic, or were the dropouts
actually failures?
Data Presentation
Were the clinical and laboratory data presented in a straight-forward,
scientific manner? Was there an attempt to present data with
less than scientific presentation?
Looking at the data for both safety and efficacy, is the dosage
proposed appropriate for the intended population? What is the
relevance of data outliers? Are the data presented consistent
with the medical literature, and is it consistent throughout
the entire NDA submission? Were the most current FDA guidelines
used? If not, your NDA may well fail. Companies who use minimum
requirements run the risk of rejection. Since medicine and policies
are constantly changing, companies using more stringent requirements
are more likely to be the winners.
The entire submission should be clear and easily readable
and understandable. The data must stand on its own merit. Data
tables, charts, etc., should be self-explanatory. Limited cross
references should be used It is counterproductive for the reviewer
and, therefore, the company, to have the reviewer searching for
relevant data in several obtuse areas in the submission. It should
be clear, readable, and persuasive.
Problem Areas
It is not possible to conduct any clinical trial without some problems.
Address any and all problems up front. Problems exist
and they should be explained.
Summary
This discussion has been a brief overview which, hopefully,
will offer companies a guide to follow when beginning their clinical
trials. I have seen the FDA reject both a well written NDA without
scientific data, and, conversely, good data poorly presented.
Companies should be forewarned. Only sound scientific data that
is well presented to the Agency will suffice.
In conclusion, I would like to quote from an excellent article
published REGULATORY AFFAIRS in the Spring, 1989 issue,
by Dr. Gloria J. Troendle (at that time Deputy Director; Division
of Metabolism and Endocrine Drug Study Products, Office of Drug
Evaluation, FDA).
"In summary, if you think like a reviewer, you should
come up with the information and the displays that a reviewer
will find most helpful. However; do not get the idea that a reviewer
will not be interested in some of the details. Most reviewers
really do look at mountains of detail, and if they are provided
with maps of the interesting points along the way, it might be
possible to prevent them from getting lost."1
1 Troendle, Gloria J., "Preparing
the Clinical Section of a New Drug Application," REGULATORY
AFFAIRS, Vol.1, 1989, pp.49-54.
Jerri B. Perkins, M.D. is President of Perkins & Perkins,
Inc.
UPDATE NOTE: There are now circumstances in which the
FDA requires only one (1) clinical trial.