Passive immunity for the treatment of influenza: quality not quantity

Sanjat Kanjilal, Michael J Mina

Lancet 2019; 7: 922-923

The use of convalescent plasma to treat influenza infection was first described during the 1918 influenza pandemic.¹ The concept is based on the assumption that patients who have recovered from an infection did so partly because they had effective antibody responses against the virus. However, studies into the clinical benefit of this concept have been inconclusive. Two randomised controlled trials by John Beigel and colleagues² and Richard Davey and colleagues³ in The Lancet Respiratory Medicine tested whether serotherapy enriched for anti-influenza antibodies would confer clinical benefit to hospital inpatients with influenza. These studies report mostly null effects of high-titre plasma or high-titre intravenous immunoglobulin (hIVIG). The studies raise key questions and highlight crucial gaps in the understanding of naturally derived antibody-mediated immunotherapies, and they provide important clues that can advance the care of patients hospitalised with influenza and other infections.

The trial led by Beigel and colleagues² randomly assigned 140 US patients of all ages admitted to hospital with influenza A infection between 2016 and 2018 to receive immune plasma containing high titres (≥1:80) or low titres (≤1:10) of haemagglutination inhibition antibodies specific to contemporaneous seasonal vaccine strains. The FLU-IVIG trial led by Davey and colleagues³ enrolled 329 adults across nine countries who had influenza A or influenza B infection and compared hIVIG with saline placebo. Both trials concluded that infusion of products containing anti-influenza antibodies as determined by haemagglutinin inhibition assays provides no overall benefit to patients with influenza.

The results report the most convincing evidence to date that absolute haemagglutination inhibition antibody titres to influenza virus might not be reliable laboratory correlates for protection from disease. One of the first clinical uses of this measure was in the first influenza vaccine trial, led by Salk in 1943,⁴ in which increasing titres were associated with protection but infections still occurred in people whose plasma had high haemagglutination inhibition titres (≥1:1024). Similarly, other studies investigating vaccine failure also note a discordance between titres and protection.5, 6 Together with the two new studies, these point towards a potential fundamental inadequacy of haemagglutination inhibition titres as a predictor of protection.

In the study by Davey and colleagues,³ a beneficial effect of hIVIG in patients with influenza B, but not A, was detected, although hIVIG lots contained considerably higher haemagglutination inhibition titres for influenza A than for B. The binding affinity of antibodies to influenza B viruses was estimated to be ten-times stronger than binding to A viruses. Although this finding could be interpreted as supporting the use of antibody affinity over haemagglutination inhibition as a marker for protection,⁷ antibody affinity does not necessarily correlate with protection or neutralising capacity. Many studies that map antibody binding sites to various viruses show the wide variety of antibodies produced during infections or vaccinations;8, 9 only a fraction have any role in protection and some even have adverse effects.¹⁰

In early studies, particularly those from 1918, convalescent serum was drawn from donors shortly after natural infection, and therefore was likely to be enriched with protective antibodies to the contemporaneous circulating strain. By contrast, the hIVIG and high-titre plasma used in the two trials discussed here were derived from healthy donors. In the absence of a very recent infection, which can result in neutralising antibody titres many orders of magnitude higher than baseline, products with high haemagglutination inhibition titres from healthy donors probably have different neutralising capacities and have been produced in response to previous infections or vaccines. This antibody composition is particularly likely, given that the haemagglutination inhibition assays in both studies used contemporaneous vaccine viruses, and not circulating viruses during years when vaccine efficacy has been estimated to be as low as 19%.

Additionally, haemagglutination inhibition titres are usually considered protective against new infection if they are higher than 1:40 or 1:80. However, treatment is not equivalent to prevention, and therapy is likely to require much higher titres. By the time that a patient becomes symptomatic with influenza, viral loads are orders of magnitude higher than the infecting dose. Therapy thus requires titres that are orders of magnitude higher, usually exceeding 1:50 000 and can be as high as 1:1 000 000. Thus, in the early studies in 1918, which used convalescent serum from recently recovered patients, the antibodies were potentially both more specific and present at much higher concentrations than those in the blood products used by Davey, Beigel, and colleagues. Therefore, whether truly hyperimmune plasma drawn from a carefully selected donor pool could still be an effective therapeutic in patients with severe influenza is still unclear.

Although the present studies2, 3 provide strong evidence that blood products containing antibody titres considered to be protective do not work against influenza A viruses, the beneficial effects against influenza B show that protection depends on the context. For influenza viruses, it is more probable that the quality and not the quantity of the antibodies will drive efficacy. We consider that these carefully done studies provide the strongest argument yet for the development of new correlates of protection against influenza (including new assays to identify and quantify truly protective antibodies), for redefining protective versus therapeutic antibody titres, and for leveraging these definitions to develop new classes of influenza therapeutics.