Dynavax's (DVAX) HEPLISAV had an FDA Advisory Committee on November 15, 2012. The Advisory Committee voted 13 to one that HEPLISAV data adequately demonstrated immunogenicity. Additionally, the Committee voted eight to five with one abstention that there were insufficient data to adequately support the safety of HEPLISAV. To follow-up on the safety issues discussed in the recent review performed by Theodore Cohen, I would like to make the following comments:
Safety database is comparable to Engerix-B's
No significant differences in safety between HEPLISAV and Engerix-B
As the FDA briefing documents state:
The overall safety evaluation across studies did not reveal significant imbalances in rates of clinically important adverse events. No significant differences in ANA titers, ANCA or anti-dsDNA levels were detected between recipients of HEPLISAV or Engerix-B. Subgroup analysis for AEs did not identify subgroups of subjects that exhibited higher rates of AEs than other groups. The CBER clinical reviewers concluded that analyses of the two pivotal studies and the integrated summary of safety did not reveal any clinically significant differences in safety between HEPLISAV and its active comparator, Engerix-B.
Not enough patients may have been treated with HEPLISAV to detect rare adverse events
The briefing documents further state: "The safety database for HEPLISAV may not have sufficient power to detect rare adverse events."
There were three adverse events identified as of most concern, all could be considered to be autoimmune events, and could be considered to be related to vaccine administration:
One case each of vasculitis in the HEPLISAV treatment arm (cyotoplasmic-ANCA [c-ANCA] positive Wegener's granulomatosis) and Engerix-B treatment arm (perinuclear-ANCA [p-ANCA] positive vasculitis) and one case of Guillain-Barré syndrome in the HEPLISAV arm.
The most conservative approach would be to consider that all these events are vaccine related. In the overall safety database for the trials, the incidence rate of such AEs would be: 3/5845 = 0.00051; in the HEPLISAV arm 2/4425 = 0.00045 and in the Engerix-B arm, 1/1420 = 0.00070.
Because of the temporal association with a subsequent influenza vaccine: "The subject's Guillain-Barré Syndrome was assessed by the investigator as being severe and 'probably not related' to study treatment but, instead, related to the influenza vaccine the subject received 5 days prior to symptom onset." If, this patient is discounted from the HEPLISAV arm, the incidence of autoimmune events for HEPLISAV drops to 1/4425 = 0.00026.
A much weaker argument could be made for dropping the patient from the Engerix arm because she might have had preexisting disease, a mixed connective tissue disease (MCTD), exacerbated by vaccine treatment, as opposed to de novo manifestation of disease as in the patient with Wegners.
Patient sample sizes necessary to demonstrate equivalence for rare adverse events
How could one demonstrate equivalence in safety for rare events? One issue is to decide upon an acceptable margin by which the two treatment groups might differ. Typically, in non-inferiority clinical trials, such a margin might be 25%, with α = 0.05 and β = 0.20 (i.e., 80% power, the probability of not making a false negative decision). In this case, we would hypothesize that there is equivalence if the rate of autoimmune adverse events is 0.0005 +/- 0.000125. The sample size to demonstrate such equivalence would be n = 395016 per group. If one were willing to tolerate that one group has twice the number of adverse event as the other, i.e., that there is equivalence with a rate of 0.0005 +/- 0.0005, then the sample size is n = 24688 per group.
Should the frequency of rare adverse events be assessed pre or post-approval?
It is unclear whether the panel members who voted for more safety data on HEPLISAV wanted these safety data to be provided pre or post-approval. However, in the event that such data would be required pre-approval, it would have a very chilling effect on vaccine development. Vaccines are responsible for the greatest medical advances from the standpoint of public health, e.g., the eradication of smallpox and the near eradication of polio. As it stands, many biopharmaceutical companies do not get involved in vaccine development because the financial returns may be lower than with therapeutics. Indeed, vaccines may confer lifelong immunity, while therapeutics typically necessitate repeat administration.
FDA likely to approve HEPLISAV without more clinical data
To require a properly sized clinical trial prior to approval to determine the incidence of rare adverse events, particularly in the absence of any evidence of imbalance in the safety data between treatment groups, would not only adversely affect Dynavax but would set a precedent for future vaccine development. In addition to routine pharmacovigilance, Dynavax has proposed an open-label, prospective, observational study to assess the incidence of medically significant adverse events, including autoimmune disease, in 5000 individuals initiating vaccination with HEPLISAV versus a concurrent population of 5000 individuals initiating vaccination with Engerix-B. One possibility would be to increase the number of patients enrolled in such a trial to ensure the proper statistical power to better evaluate the relative frequency of these rare events.
Disclosure: I am long DVAX.