The FDA’s Internal SOP On How To Do An NDA Safety Review Part 1

Sep 18, 2012
Bart Cobert

Pharmacovigilance, Drug Safety and Regulatory Affairs Author & Expert

One of the lesser known aspects of the FDA is that their SOPs are in the public domain and are available on-line.  They are found in the Manual of Policies & Procedures (MOPPs) of CDER.

Although some are as routine and mundane as anyone else’s SOPs a few are gems and are well worth reading and understanding.  We’ll talk about on set, in particular, in Bart’s Corner over the next several weeks.  This is MOPP 6010.3R: Good Review Practice: Clinical Review Template.

It consists of two sections.  The first is 6010.3R and is called Attachment A: Annotated Clinical Review Template (last revised 12/14/2010) and covers the review of an NDA or BLA by the FDA reviewers.  It touches on efficacy, safety, risk-benefit, and the other key parts of these sponsor submissions. It is not really just a template but a detailed 29 page review and instruction manual.

The second, which we will talk about today, is an 84-page manual on how to do a safety review called “Attachment B: Clinical Safety Review of an NDA or BLA. “ Interestingly, it is nearly three times as long as the clinical review (efficacy) part noted above.

The manual reads like a textbook on how to critically review the safety data in an NDA/BLA.  It is frank, thoughtful and full of insights and advice for industry (maybe this was the FDA’s intent!).

We will go through this SOP in detail covering the points that are critical or that you may not be aware of or think about.  Direct quotations from the SOP are in quotation marks (” “). Bart’s, my, comments are in italics unless otherwise specified.

The table of contents covers: high level concepts, methods, adequacy of the safety assessments, major safety results, supportive safety results (labs), other safety explorations (drug interactions), pediatric and pregnancy use and much more.  There are sample tables in the appendix.  This document should be read by anyone doing drug safety whether in the clinical research or post-marketing areas.

Bart’s Corner will summarize this document starting today.  But you should definitely read it yourself.

This SOP gives detailed instructions to the FDA reviewer (usually a physician) on how to critically look at and evaluate the safety data submitted by a pharmaceutical company (called the sponsor or the applicant).  The FDA states clearly for the reviewer (but less so in publications for the general public) that “no drug is safe in the sense of being entirely free of adverse events…the safety of a drug for the uses recommended in labeling has been interpreted as meaning that the benefits of a drug outweigh its risks for those uses.”

High Level Views

The SOP starts with a goal statement for the reviewer: to assess “the adequacy of the applicant’s safety evaluation” and identify and assess the “significance of the adverse events (AEs) reported in studies/clinical trials (controlled or uncontrolled). The clinical reviewer should collaborate with the FDA statisticians to evaluate event rates, estimation of risk over time, subgroup differences and measures of uncertainty.


The document gives various definitions known to us all but adds one we don’t hear much about: “Adverse dropout” meaning a subject who did not complete the study because of an AE whether considered drug related or not and includes subjects on the study drug, comparator or placebo.

The message here: The FDA will scrutinize subjects who dropped out for safety reasons. Don’t ignore analyzing these (and all) dropouts when you prepare your NDA.

Reviewer Tasks

The reviewer’s primary tasks are four-fold:

  1. Evaluate the adequacy of the data to support the safety analysis and to identify the limitations of the data.  Is the exposure at the relevant doses adequate?
  2. Identify SAEs that may signal important problems severe enough to prevent use of the drug altogether, limit its use or require special risk management.
  3. Identify and estimate the frequency of common non-serious AEs likely to be drug related.
  4. Identify the unresolved safety concerns needing attention before approval or after approval; for example the absence of data from high-risk populations or potential interactions.

In addition, the reviewer should:

  1. Identify factors that predict the occurrence of the adverse reactions (e.g. age, ethnicity, illness, comorbidities, drug dose, duration of exposure).
  2. Identify ways to avoid adverse reactions (ARs) and how to manage them when they do occur.
  3. For a drug that is about to be approved, provide a comprehensive evaluation of risk to support a factual label summary.

Do you do all of this when you prepare your NDA/BLA (or PSUR or DSUR)?  Do you put yourself in a (critical) FDA reviewer’s shoes?

The Approach to Safety is Different from the Approach to Efficacy

This section describes the way safety data is collected and evaluated in trials.  Most of us in safety know this but not everyone else in the company quite realizes the differences since they tend to focus in phase 2 and 3 studies on efficacy.  If the drug doesn’t work, it doesn’t matter whether it is safe of not in some senses.

The FDA notes that the applicant designs these trials identifying critical efficacy endpoints in advance and setting up the study with sufficient sample size to get an adequate assessment of effectiveness/efficacy.  Calculations are done to minimize type 1 error for the key endpoints with adjudication committees often set up to make sure the data is tight.  The FDA notes that these studies are “not designed to test safety hypotheses nor to measure or identify ARs with any prespecified level of sensitivity.”  Thus statistical analyses of safety data with significance levels (p values) is rarely done; rather sponsors just provide listings.

Nonetheless, the reviewer and sponsor can explore and estimate rates looking at individual studies and pooled studies. These are a “critical and essential part of a safety evaluation”.  But caution must be used as these analyses can “of course, lead to false conclusions”.

Assessing SAEs

The FDA talks of two kinds of SAEs.  The first (they call them SARs as they are due to the drug) and are “consequences or at least potential consequences of treatment…because they would be unusual in the population under study.”  The second are SAEs that “are not so readily attributed to the drug because they can occur even without the drug” such as heart attacks, strokes in an elderly population and could therefore represent intercurrent illness.

The FDA notes that the atypical or unexpected AEs are very hard to pick up and attribute to a drug giving several examples including “retroperitoneal fibrosis with methylsergide (Sansert)” and drugs for heart failure (e.g. beta agonists, flosequinan) which caused increased rates of death seen with the underlying disease such as worsening heart failure or arrhythmias.

The FDA then gives some specific classes of drugs that are hard to evaluate including drugs acting on the immune or other impaired systems in patients with cancer or HIV, drugs for seizure disorders and drugs for schizophrenia.  Comment: It looks like drugs for these disorders will get heightened scrutiny at the FDA.

The FDA’s Advice to the Reviewer

Reviewing safety is hard!  “There is no simple answer…but…using the prepared mind [the FDA’s italics] approach of close examination of all subjects who die or who leave a study prematurely because of an AE (whether or not thought to be drug-related)” should be done.

Sometimes surrogate markers are useful such as increased transaminases and bilirubin together may predict the occurrence of severe liver disease; visual field defects may portend irreversible peripheral vision loss; substantial QT prolongation may predict torsade de pointes ventricular tachycardia.

It is difficult to discover that “a drug causes a modestly increased rate of serious events that are relatively common in the population” and “only large controlled studies/clinical trials can provide a satisfactory answer”.  The “reviewer needs to consider whether such studies/clinical trials are needed.” [italics mine].

The bottom line here: 

The FDA is telling reviewers and companies to pay attention to certain key safety issues which companies may not focus on (e.g. dropouts).  Even if you don’t, the FDA reviewer will.

In the next Bart’s Corner we will continue looking at this SOP and move into the methodology the reviewer uses.