Efficacy of Influenza Vaccination in HIV-Infected Persons: A Randomized, Double-Blind, Placebo-Controlled Trial FREE

Influenza vaccination is recommended for patients with HIV-1 infection (1), but no studies have measured its efficacy in this population. Patients infected with HIV-1, especially those with low CD4⁺ cell counts, often have minimal antibody responses to influenza vaccine (2 - 3), leading some researchers to conclude that vaccination is not cost-effective (4). In addition, vaccination has been associated with transient increases in HIV replication (5 - 8) and small decreases in percentage of CD4 cells (8); both of these findings are of undetermined clinical relevance.

To better estimate the potential benefit of routine influenza vaccination in persons infected with HIV, we conducted a randomized, controlled trial of influenza vaccine in a sample of HIV-1-infected patients during one influenza season.

The study site, a hospital-based clinic affiliated with the U.S. military, provides care to 400 to 500 patients infected with HIV-1. Posters invited patients who visited the clinic from October to December 1995 to refer themselves for enrollment. All HIV-infected patients who had not yet received their annual influenza vaccine were eligible to participate. All patients gave informed consent, and the hospital's institutional review board approved the study. Blood was drawn for measurement of CD4⁺ cell counts, influenza antibody titers, and plasma quantitative HIV-1 RNA level. Using a random computer-generated sequence, which was kept sealed until the end of the study, pharmacy staff prepared syringes that were filled with saline or with influenza vaccine purchased by the hospital for clinical use (whole virion containing 15 µg each of A/Johannesburg/33/94 [H3N2], A/Texas/36/91 [H1N1], and B/Harbin/07/94 [Pasteur Mérieux Connaught, Lyon, France]. Clinic staff could not distinguish vaccine from placebo.

At 1 and 3 months after vaccination, patients returned for follow-up measurement of CD4⁺ cell count and HIV-1 RNA level and for serologic examination for influenza. Patients returned for viral cultures only if respiratory symptoms developed. At the conclusion of the study, all patients were interviewed about respiratory illness (broadly defined as any rhinitis, pharyngitis, or cough) experienced during the study period.

Samples for T-cell subset analysis were processed according to standard methods. The variation of percentage of CD4 cells on replicate testing was ±1%. Plasma samples for measurement of HIV-1 RNA level were drawn in acid citrate-dextrose tubes, separated at 2000 rpm for 15 minutes, and stored at −70 °C. Serum samples were stored similarly. Quantitative HIV RNA reverse transcriptase polymerase chain reaction assays (Amplicor Monitor, Roche, Nutley, New Jersey) were done in batches by Quest Diagnostics at Nichols Institute, San Juan Capistrano, California; all specimens from individual patients were tested simultaneously. The lower limits of quantification were 400 RNA copies per mL of plasma, and the intra-assay coefficient of variation of 32.3% to 45.3%.

Serum samples taken before and 1 month after vaccination were tested for antibody responses to each of the three individual vaccine components by microtiter hemagglutination inhibition; egg-grown virus antigens starting at a dilution of 1:4 were used (9). Samples below this threshold were assigned a value of 1:2. Serum samples obtained 1 month and 3 months after vaccination were tested against the A/Johannesburg/33/94 and A/Nanchang/33/95 (H3N2) virus strains; the latter is antigenically similar to the A/Wuhan/95-like viruses that were circulating in the United States in 1995 and 1996. All of the patients who responded to A/Johannesburg/94 (those who demonstrated a fourfold or greater increase in titer) also responded to A/Nanchang/95. Because both A/Johannesburg-like and A/Wuhan-like viruses circulated in the study sample, an antibody response to A/Nanchang/95 occurring between 1 and 3 months after vaccination was defined as evidence of influenza A infection. Viral cultures were performed by using standard laboratory methods.

Statistical analysis was done by using Statview 4.5 software (Abacus Concepts, Inc., Berkeley, California). Changes in antibody titer, plasma HIV-1 RNA level, and CD4⁺ cell count were analyzed by using the unpaired t-test to compare vaccine and placebo recipients. Illness rates were compared by using the chi-square test. The percentage of protective efficacy of vaccination was calculated as 1 − (risk ratio comparing vaccine and placebo recipients) × 100.

A total of 102 patients enrolled in the study. Fifty-five patients were randomly assigned to receive vaccine, and 47 patients were randomly assigned to receive placebo. Table 1 summarizes the participants' characteristics. Because the study was conducted in fall 1995, antiretroviral regimens were limited to single or double reverse transcriptase inhibitor combinations. No patients were taking protease inhibitors. All patients had received at least one previous influenza vaccination.

Prevaccination and postvaccination paired serum samples from 49 vaccine recipients and 43 placebo recipients were tested against the three vaccine constituents (Table 2). Although the aggregate responses to vaccination were statistically significant, only 12%, 29%, and 36% of vaccine recipients had a fourfold or greater response to the A/H1N1, A/H3N2, and B components, respectively.

Peak influenza activity in San Diego County in 1995 and 1996 occurred from mid-January through early February and was predominately caused by the A/Johannesburg (H3N2) virus strain; some A/Wuhan (H3N2) virus was also present (San Diego Public Health Department. Personal communication). Sixteen of 55 vaccine recipients (29%), compared with 23 of 47 placebo recipients (49%), reported symptoms of respiratory illness during the fall and winter of 1995 and 1996 (P = 0.04). No patients were hospitalized or treated for pneumonia during the study period.

Twenty of 39 patients (12 placebo recipients and 8 vaccine recipients) who reported respiratory illness returned to the clinic for viral respiratory culture. Influenza A (H3N2) virus was isolated from 3 patients, and influenza B virus was isolated from 2 patients, all of whom received placebo. Serum samples taken 1 and 3 months after vaccination were available for 86 patients (43 vaccine recipients and 43 placebo recipients). One vaccine recipient and 10 placebo recipients had serologic evidence of influenza A infection (P = 0.004). During the 1995-1996 influenza season, 13 of 47 placebo recipients had laboratory-documented influenza A infection (by culture or serologic examination) compared with only 1 of 55 vaccine recipients (P < 0.001) (calculated protective efficacy, 93% [CI, 69% to 100%]).

The one vaccinated patient who had serologic evidence of influenza infection had advanced HIV infection with a CD4⁺ count of 36 cells/mm³ at baseline. Although our study included too few patients with CD4⁺ cell counts less than 200 cells/mm³ to make firm conclusions, only one of five vaccine recipients (20%) reported respiratory symptoms during the study period compared with four of eight placebo recipients (50%) (P > 0.2).

The one vaccinated patient with serologic evidence of infection did not report symptoms of respiratory illness during the study period. Ten placebo recipients and no vaccine recipients had symptomatic laboratory-confirmed influenza A (protective efficacy, 100% [CI, 73% to 100%]).

No significant differences were seen in changes in the percentage of CD4 cells or plasma HIV-1 RNA levels between the vaccine and placebo groups or between patients with and those without documented influenza infection. At 1 month, the plasma HIV RNA level increased 0.03 log (CI, −0.12 log to 0.18 log) in vaccinated patients, a value that is well within the anticipated variability (0.3 to 0.5 log) of the plasma HIV RNA level (10). Twenty-six percent of the study patients (14 vaccine recipients and 13 placebo recipients) changed their antiretroviral regimens during the study period.

Our results indicate that influenza vaccine is highly effective in HIV-infected patients. Calculated protective efficacy was 100% for symptomatic influenza A, and the lower bound of the CI around this estimate was 73%. This protection is comparable to the 70% efficacy described in immunocompetent adults (1). Although patients with AIDS were underrepresented, the subset of such patients showed nonstatistically significant decreases in respiratory illness that were nearly identical to those of the group as a whole.

Our use of serodiagnosis may have missed influenza infections in vaccinated patients who did not have follow-up titers or had antibody titers that were already high before the influenza season. However, the protective efficacy, as measured by serodiagnosis, was supported by a similar decrease in culture-positive illness and significant reductions in respiratory symptoms. Although excellent protection was seen against H3N2 influenza, which was circulating in 1995 and 1996, an inherent limitation of studying influenza is that other vaccine components may not be as efficacious against future epidemic influenza strains.

Although our study is the first to examine the efficacy of influenza vaccine in persons infected with HIV-1, many studies have measured the immunogenicity of influenza vaccine in this population (2 - 3,7,11 - 16). Reported response rates range from 0% to 85%. The wide variation reflects host factors, such as stage of illness, and variable antigenicity of different vaccine strains. Although the patients in our sample were relatively healthy, their antibody responses were similar to those previously reported in patients with AIDS. Nevertheless, our patients had excellent protection against H3N2 influenza, perhaps because of modest responses to vaccine that were below the threshold of a fourfold increase in titer.

Influenza vaccination has been associated with increases in plasma HIV-1 RNA level in some reports (5 - 8) but not in others (16 - 18). The clinical significance of such changes has not been determined. Although we found that immunization and infection did not affect plasma HIV-1 RNA levels or CD4⁺ cell counts, we may have missed transient increases in viral replication. Because approximately one third of the placebo recipients had documented influenza, the difference in immune stimulation between the two groups was minimized. In addition, more than 25% of the study participants changed antiretroviral regimens during the study; this may have masked potential effects of vaccination or illness on viral replication.

Healthy adults benefit from influenza vaccination. Decreases in illness and in days missed from work save almost $47 per dose (19). Each year, influenza is estimated to cause more than 300 excess hospitalizations per 10 000 HIV-1-infected women 15 to 64 years of age; this number is greater than those seen in any other high-risk group (20). The efficacy documented in our study provides evidence to clinicians and policymakers that routine influenza vaccination benefits patients with HIV-1 infection.