Pasteur, Hôpital Cochin, Paris, and Unité de Maladies Infectieuse et Tropicales, Hôpital Raymond Poincaré, Garches. Our main exclusion criteria were: current febrile illness; immunodeficiency; recent receipt of blood, blood-derived products, or immunosuppressive treatment; current long-term systemic corticosteroid therapy; recent (previous 4 weeks) vaccination or vaccination planned in next month; and pregnancy. A history of travel to H5 endemic areas was not an exclusion criterion, and we did not restrict the travel of participants after enrolment. Our inclusion criteria were age 18–40 years on day of enrolment, provision of written informed consent, ability to attend all scheduled visits and to comply with all trial procedures, entitled to national social security, and registered in the French file of healthy volunteers in clinical trials. Women had to be unable to bear children or able to produce a negative urine pregnancy test, having used an effective method of contraception or been abstinent for at least 4 weeks before the first vaccination (to be continued for at least 4 weeks after the second vaccination).
The trial protocol and all relevant documents were approved by an ethical review committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale, Paris-Necker), and we notified the French health authority (Agence Française de Sécurité Santaire des Produits de Santé) of the study in accordance with French regulations. All eligible volunteers gave written informed consent before inclusion.
Procedures
The H5N1 influenza vaccine was a monovalent A/H5N1, inactivated, split-virion influenza vaccine manufactured by sanofi pasteur (Lyon, France). It was produced on a pilot scale in embryonated hens' eggs, using the licensed manufacturing process for the interpandemic vaccine Vaxigrip as previously described16 and adapted to the particular avian strain, according to biosafety guidelines for the production and quality control of vaccines for use during pandemics of human influenza.17
The vaccine strain was the influenza A/Vietnam/1194/2004/NIBRG14 (H5N1) reference strain prepared by the NIBSC, and is one of the reference viruses indicated as suitable for use in a mock-up vaccine by the CHMP.8 This modified virus contains modified haemagglutinin and neuraminidase from the highly pathogenic avian strain influenza A/Vietnam/1194/2004 and other viral proteins from influenza A/PR/8/34 (H1N1).15 The haemagglutinin was modified to remove the multibasic aminoacid at the cleavage site. Haemagglutination-inhibition tests confirmed that NIBRG-14 haemagglutinin was antigenically the same as that of the wildtype influenza A/Vietnam/1194/2004 virus. The NIBSC tested for absence of virulence in animal models, including ferrets. We selected the dose range to be assessed based on published experience with other pandemic vaccine candidates.9,10 We prepared prefilled 0·5 mL syringes (23 g 1 inch needle) of 7·5 μg, 15 μg, or 30 μg of haemagglutinin in a phosphate buffered saline solution without adjuvant. The neuramidase content of the vaccine was not measured. The non-adjuvanted formulations were ready to use. We prepared the three aluminium hydroxide-adjuvanted formulations on site (by the bedside) just before use, according to a reconstitution protocol, to contain a final adjuvant content of 600 μg aluminium hydroxide per injected dose: we transferred the total volume of influenza vaccine contained in the prefilled syringe and the total volume of separate prefilled syringes of adjuvant to a sterile vial and mixed for 10 s before drawing the solution into a new syringe (23 g 1 inch needle), gently swirling to homogenise the antigen-adjuvant suspension, and injecting intramuscularly. Three vaccine batches were used: S3977 for the 30 μg, the adjuvanted 30 μg, and the adjuvanted 15 μg formulations; S3976 for the 15 μg and the adjuvanted 7·5 μg formulations; and S3974 for the 7·5 μg formulation. We did a study mimicking on-site formulation of vaccine with adjuvant on the 30 μg, 15 μg, and 7·5 μg adjuvanted formulations; results indicated similar adsorption coefficients. The injected volume was 0·5 mL in all cases, except for the adjuvanted 30 μg formulation for which the volume was 1 mL. Each final dose also contained 45 μg of the preservative thiomersal.
We randomly allocated participants to one of six groups to receive 7·5 μg, 15 μg, or 30 μg of haemagglutinin, with or without adjuvant. Each participant received two intramuscular injections, 21 days apart, of the assigned formulation into the deltoid. We stratified randomisation lists by centre and created them with the block method, using decreasing block sizes of 18, 12, and six so that a similar number of people was enrolled into each group at any given time. The sequence was generated at sanofi pasteur by a statistician who was not involved in the rest of the trial.
Our aim was to describe the safety profile of, and immune response to, the vaccine 21 days after each dose. Participants attended three trial visits (on day 0, 21, and 42) for blood sampling, vaccination (day 0 and 21 only), and safety data collection. We kept individuals under observation for 30 min after vaccination, and gave them safety diaries, digital thermometers, and rulers to assess and record adverse events. From days 0 to 7, diaries included a list of solicited injection site and systemic adverse events, including those recommended for the assessment of interpandemic vaccines by the CHMP. The intensity of non-quantifiable reactions—eg, pain—was assessed with a severity scale: mild, noticeable but does not interfere with daily activities; moderate, interferes with daily activities; severe, prevents daily activities. At the next visit (day 21 and 42), we (J-LB, CP, OL, and clinicians at each study site) interviewed patients, transcribed adverse events onto case report forms, and assessed whether unsolicited adverse events were vaccine-related. By convention, we judged all solicited adverse events vaccine-related. We collected blood samples in dry tubes and processed them at each trial centre within 24 h. We then froze and shipped the samples to the Health Protection Agency Centre for Infection, London, UK, for centralised analysis.
We assayed serum samples obtained from individuals on day 0 (before first vaccination), 21, and 42 at the same time to detect haemagglutination-inhibition activity and neutralising activity against the vaccine seed virus (NIBRG14). Work with this virus was done at containment level 3 after local risk assessment, and in line with Health and Safety Executive guidelines. There was no mention of treatment group on either the samples or the accompanying listing provided to the laboratory—ie, we undertook immunogenicity analyses under blinded conditions.
We used a previously described modified haemagglutination-inhibition assay, using horse erythrocytes instead of turkey erythrocytes, because of their increased sensitivity for the detection of human antibodies against H5.12 Briefly, we treated sera for 16 h with receptor-destroying enzyme, which was then inactivated by incubation at 56°C. From the initial dilution of one in eight, we undertook two-fold dilutions of the serum samples and incubated with the antigen suspension (4 haemagglutination units/25 μL) for 1 h at room temperature. After adding 25 μ