Yellow fever (YF) re-emergence is of major concern in sub-Saharan Africa and tropical South America.1 Its lethality averages 50% in symptomatic patients.2 The YF vaccine, strain-17D (YF17D), a live attenuated virus vaccine, is the only protection against YF and displays high efficacy: seroconversion rate is 99% within 30 days after immunization, and although WHO recommends booster immunization at 10-year intervals, protection may persist for 30 years after a single dose.3 It is generally well tolerated. However, cases of YF vaccine–associated neurotropic disease and, more recently, deaths related to YF vaccine–associated viscerotropic disease have been reported,4 especially in the elderly and in patients with a history of thymus disease, which raises concerns about the consequences of an unrestricted policy toward YF immunization.
Evaluation of YF immunization in HIV-infected patients is necessary, not only among travelers in countries of YF endemicity but also among populations of tropical sub-Saharan Africa, where HIV infection is highly prevalent and where YF outbreaks occur.5 In immunocompromised patients, immunization with the live strain-17D poses the risk of visceral disease or encephalitis.6 YF primary and booster immunizations are thus recommended for patients with a CD4 T-cell count >200 cells per cubic millimeter.7 Seroconversion rates and/or duration of protection could be lower than in immunocompetent subjects. In sub-Saharan Africa, HIV infection may lead to loss of protection induced by previous YF immunization(s). Data about the efficacy and tolerance of YF immunization in HIV-patients are scarce and based on a limited number of patients.8–13 Seroconversion rate and antibody titers assessed in 78 patients from the HIV-Swiss cohort within 1 year of immunization were lower than in HIV-negative subjects, but no risk factor for failure of YF immunization could be identified.13 There are no available data on the magnitude and durability of antibody response for more than 1 year after immunization. Additionally, the persistence of antibodies after HIV infection among patients immunized before their diagnosis of HIV infection has not been yet investigated.
Therefore, we conducted a study to analyze the immunogenicity of YF immunization in a large cohort of HIV-infected patients. The primary objective of this study was to assess the proportion of patients with a nonprotective YF antibody titer, according to whether patients were immunized before or after the diagnosis of HIV infection. The secondary objectives of the study were to investigate risk factors associated with a nonprotective YF antibody titer and to assess parameters associated with the magnitude of antibody response.
PATIENTS AND METHODS
A cohort study was conducted in the Department of Infectious Diseases at Saint-Antoine Hospital, Paris, France, between July 2007 and October 2008. All consecutive HIV-infected patients visiting the outpatient clinic were investigated about previous YF17D immunization. Patients who received at least 1 injection of YF17D vaccine were subsequently prescribed a serological assay to measure YF17D-specific neutralizing antibody titers (NT) and were enrolled in the cohort study. All data were extracted from the database of the infectious diseases unit, which received approbation from the local ethics committee. Patients also provided informed consents to participate in the study.
Serological analysis was performed at Laboratoire Cerba (Saint-Ouen l'Aumone, France), using the plaque-reduction neutralization test, which is the standard technique for assessing humoral response to YF17D immunization.14 Briefly, this assay measures the NT, which is the highest serum dilution able to induce an 80% plaque reduction in cells infected by the YF17D strain. A YF virus NT ≥1:10 is accepted as a serologic surrogate of clinical protection, and subjects were considered to have seroconverted if they had a NT above this threshold.13,15 All sera were assayed twice in parallel in 2-fold dilutions ranging from 1:10 to 1:80.
The following data were systematically extracted from the computerized outpatient database: age (at immunization and at time of NT determination), gender, whether the patient originated from a region endemic for YF or not, date of diagnosis of HIV infection, date of YF17D immunization (if a patient has received more than one injection during his or her life, the date of the last injection was recorded), whether the injection was a primary immunization or a booster type, CD4 T-cell counts (/mm3), plasma HIV RNA (copies/mL and log copies/mL) within 3 months before or after immunization and the time of NT determination, nadir of CD4 T-cell count (/mm3) before immunization and before NT determination, and date and result of YF-antibody NT determination. Dates of YF17D immunizations were extracted from the database of our center of international vaccinations or from the patient's international certificate of vaccination against YF.
Descriptive statistics were expressed using standard methods. Comparisons were performed using the χ2 test for categorical variables or the Mann–Whitney U test for continuous variables.
The following candidate variables were analyzed as putative risk factors for NT <1:10: age, gender, country of birth, receipt of YF injection before or after diagnosis of HIV infection, number of YF17D injections, immunovirological status at the time of immunization and serology, delay between HIV diagnosis and immunization, and between immunization and serology. A simple logistic regression analysis was used to quantify the association between these variables and lack of seroconversion, expressed as odds ratios (ORs) and 95% confidence intervals (95% CIs). A multivariate logistic regression model was adjusted for variables associated with NT <1:10 at a significance of P <0.1 in univariate analyses.
Parameters associated with the magnitude of NT, expressed as log10 of the reciprocal NT value, were also investigated. The Mann–Whitney U test or Spearman rank test was used to test the association between the mean log10 of the reciprocal NT value and candidate variables, whenever appropriate. General linear models were fitted to study the association between covariates and log10 of the reciprocal NT value at a level of 0.1 in univariate analyses. Resulting β parameters were expressed at a 95% CI.
All results with P values <0.05 were considered to be statistically significant. Statistical analysis was performed using STATA IC version 10 software.
A total of 364 HIV-infected patients who received at least 1 injection of YF17D vaccine and had visited the clinic during the study period were evaluated (Table 1).
Overall, 24 patients (7%) had an antibody titer <1:10 after a mean delay of 8.4 years after immunization (range: 0.9–23.3 years). The proportion of patients with an antibody titer ≥1:10 was 98% (44 of 45) when assayed within 1 year after immunization and remained at 92% (66 of 72) when assayed at >10 years after immunization (range: 10–52 years).
Patients immunized before (n = 124) and after (n = 240) the diagnosis of HIV infection differed significantly in factors such as age at immunization, frequency of origin from an area of YF endemicity, number of injections of vaccine, delay between immunization and NT determination. Therefore, stratification was carried out on whether patients were immunized before or after HIV diagnosis. Parameters associated with NT ≥1:10 and with the magnitude of antibody response were investigated for both groups of patients.
Patients Immunized After HIV Diagnosis (n = 240)
Nine subjects among 240 immunized after HIV diagnosis had a NT <1/10 (Table 1).
They were at higher risk of having an uncontrolled HIV replication at immunization [OR = 6.04, 95%CI (1.19 to 30.6), P < 0.03]. In these patients, HIV RNA load at immunization was significantly higher than in patients with NT ≥1/10 [mean = 3.6 (SD 1.3) vs. 2.4 (SD 1.0), P = 0.003], with a 2.1-fold increase of risk per log10 of HIV load (95% CI: 1.2 to 3.78).
The magnitude of NT response was determined in these 240 patients. In univariate analysis, lower values of log10 reciprocal NT were associated with female gender [1.45 vs. 1.56 in males, OR = 1.12, 95% CI: (1.01 to 1.24)], younger age at the time of immunization [ρ = 0.25, OR = 1.008 per 1 year of increase of age, 95% CI: (1.003 to 1.01)] and at the time of the serologic assay [ρ = 0.18, OR = 1.01 per 1 year of increase of age, 95% CI: (1.005–1.02)], shorter delay between immunization and diagnosis of HIV infection [ρ = 0.2, OR = 1.01 per 1 year of increase of delay, 95% CI: (1.005 to 1.02)], longer delay between immunization and NT determination [ρ = –0.18, OR = 0.98 per 1 year of increase of delay, 95% CI (0.96 to 0.99)], higher plasma HIV RNA at immunization [ρ = −0.31, OR = 0.89 per log10 of viral load, 95% CI (0.85–0.93)], noncontrolled plasma HIV RNA at immunization [1.37 vs. 1.60 in patients < 400 copies/mL, OR = 0.79, 95% CI (0.71 to 0.88)], and noncontrolled plasma HIV RNA at NT determination [1.41 vs. 1.54 in patients with <400 copies/mL, OR = 0.88, 95% CI (0.78 to 0.99)]. A borderline association was noted between lower values of NT and originating from an YF endemic area [1.48 vs. 1.57, OR = 1.09, 95% CI (0.98 to 1.22)]. No association between values of NT and the number injections of YF17D vaccines or CD4 T-cell counts (either at immunization or at NT determination) could be found. After multivariate analysis (General Linear Models), lower values of NT remained independently associated with younger age at the time of immunization [adjusted OR = 1.008 per one-year of increase of age, 95% CI (1.003–1.014), P = 0.005] and noncontrolled plasma HIV RNA at immunization [adjusted OR = 0.83, 95% CI (0.74–0.93), P = 0.001].
Further analysis focused on the 79 patients who received YF17D vaccination as a primary injection after diagnosis of HIV infection (Table 2). Patients with a NT <1:10 differed from those with a NT ≥1:10 by having a higher mean HIV plasma viral load and a lower mean duration of undetectable HIV plasma viral load at immunization and by having a lower mean CD4 T-cell count at NT determination. They were more likely to have a noncontrolled viral load >400 copies per milliliter at immunization. In multivariate analysis, a higher mean HIV plasma viral load at immunization was the unique independent risk factor for NT <1:10. There was also a nonsignificant trend toward a lower CD4 T-cell count in patients with a NT <1:10.
In these 79 patients, lower values of NT were associated with higher plasma HIV RNA at immunization, noncontrolled plasma HIV RNA at immunization, and a shorter duration of undetectable plasma HIV RNA at immunization (Table 3). A lower value of NT tended to be associated with a longer delay between immunization and NT determination. No association was found between values of NT and other variables [age at time of immunization or NT determination, origin from an endemic area, number of injections of YF17D vaccine, CD4 T-cell counts (either at immunization or at NT determination), plasma HIV RNA at NT determination]. In multivariate analysis (general linear model), lower values of NT remained independently associated with a shorter duration of undetectable plasma HIV RNA at immunization and higher plasma HIV RNA.
Patients Immunized Before the Diagnosis of HIV Infection (n = 124)
Among the 124 patients immunized before diagnosis of HIV infection (Table 1), the delay before immunization and subsequent diagnosis of HIV infection was shorter in subjects with NT <1:10 when compared with patients with NT ≥1:10 [−3.7 years (IQR: 3.9) vs. −6,9 (IQR: 7.2), P = 0.04], translating in a 1.12-fold increased risk of vaccine failure per year of delay (95%CI: 0.98 to 1.28).
Patients Immunized While Having CD4 T Cells <200/mm3
At time of YF immunization, 14 patients vaccinated after HIV diagnosis had a CD4 count below 200 cells per cubic millimeter (range: 16–198/mm3). Among these, 6 had an HIV viral load >400 copies per milliliter at the time of immunization. Most were males (n = 10), originating from sub-Saharan Africa (n = 11). Most had received a booster injection of vaccine (n = 7), whereas 4 received primary vaccination (data regarding the number of injections were missing for three patients). All of them exhibited a NT ≥1:10 after a mean delay of 4.6 years (range: 0.2–11.2 years) after immunization. They had no serious adverse events.
This study of a large series provides new data about the immunogenicity of YF immunization in HIV patients.
Overall, YF immunization is effective in HIV patients. The proportion of patients with an antibody titer ≥1:10 was especially high within 1 year after immunization (98%), and remained elevated despite a marginal decrease after 10 years (92%). These results are consistent with previous reports of smaller series,8,11–13 except 1 conducted in a child population in the Ivory Coast.9 In HIV patients, YF immunization seems to be more immunogenic than vaccines against hepatitis A16 or hepatitis B.17 In HIV-negative patients, seroconversion rate is 99% within 30 days after YF immunization,3,15 and protection may persist for 30 years after a single dose.3 Furthermore, seroconversion rates assessed in patients in the HIV-Swiss cohort were lower than in HIV-negative subjects from an historic control group, both within the first year (83% vs. 97%) and within 1–10 years after immunization (77% vs. 88%).13 Immunogenicity of YF immunization seems, therefore, to be slightly impaired in HIV patients compared with HIV-negative subjects.
Immunization After Diagnosis of HIV Infection
Among patients immunized although being HIV-infected suppressed plasma HIV viral load at baseline was the unique determinant for persistence of levels of antibodies ≥1:10 and was independently associated with higher values of NT. Surprisingly, higher values of NT were also correlated with older age at the time of immunization.
The relationship between suppressed plasma HIV viral load and success of immunization has been previously evidenced for immunization against hepatitis B,18,19 hepatitis A,16 Japanese encephalitis,20 varicella,21 influenza,22 and pneumococcus,23 but not yet for YF. In our cohort, the duration of undetectable viral load at immunization was an independent factor associated with NT ≥1:10 and higher values of NT among recipients of primary YF17D vaccine. To our knowledge, this parameter has never been evaluated before in previous studies that have assessed the immunogenicity of vaccines in HIV patients. Induction of memory responses might be improved among patients with a longer delay of control of HIV infection at the time of immunization. In the HIV Swiss cohort,13 undetectable HIV RNA at baseline tended to be related to higher NT, but it was not predictive of success of YF immunization. However, there might have been a lack of statistical power due to the size of their study population. In the same cohort study, undetectable HIV RNA at NT determination was associated with higher NT, but this was not observed in our population, in which all 16 patients with HIV RNA of >400 copies per milliliter at the time of serology had a NT of ≥1:10.
There is a conflicting influence of CD4 T-cell count on the immunogenicity of YF immunization. Higher CD4 T-cell counts at baseline and at NT determination were associated with a higher magnitude of antibody response in the report of the HIV Swiss cohort.13 In our study, no association was found between antibody response and CD4 T cell or its nadir. Despite the small number of patients with a CD4 T-cell count <200 cells per cubic millimeter at baseline (n = 14), it is noteworthy that all of them exhibited an antibody level ≥1:10 after a mean delay of 4.6 years after immunization. The median CD4 T-cell count at immunization was 451 cells per cubic millimeter, which might have prevented us identifying a low CD4 T-cell count as a risk factor for NT <1:10. Nevertheless, similar findings have been previously reported with immunization of HIV patients against influenza,22 hepatitis A,16 hepatitis B,18 and varicella.21
A more rapid decline of antibody titer has been suggested in HIV-infected compared with HIV-negative patients.13 Among the 9 patients of our cohort with NT <1:10, 7 had been immunized less than 5 years before. Additionally, although most patients in our cohort had been immunized <10 years before NT determination, a wider delay between immunization and serologic testing was associated with a lower NT, but the multivariate analysis did not confirm this association. Booster immunization is recommended every 10 years in immunocompetent patients.7 Our data and those from the HIV Swiss cohort suggest that immunogenicity of YF immunization might wane more quickly in HIV patients. Control of the YF antibody titer might thus be warranted more frequently if the patient is likely to travel to an endemic YF area.
Immunization Before Diagnosis of HIV Infection
To our knowledge, immunogenicity of YF immunization and its durability have not been investigated among subjects who have been immunized before diagnosis of HIV infection. In many countries of sub-Saharan Africa, immunization is expected to be provided during childhood or in mass campaigns during YF outbreaks. HIV infection status is usually not known in these circumstances and noninfected patients at the time of immunization are at risk of subsequent acquisition of HIV. In our study population, a shorter delay between immunization and diagnosis of HIV infection was a risk factor for NT <1:10. In the 124 patients of our cohort which were immunized before the diagnosis of HIV infection, YF immunization was performed either in patients not yet infected or in patients already infected but without knowledge of their HIV serological status and therefore without highly active antiretroviral therapy. We hypothesize that as the time between YF immunization and HIV diagnosis increases, the risk of actually having HIV infection with replicative HIV at the time of YF immunization decreases. Therefore, this result is consistent with the concept that vaccine response is better when HIV is controlled.
First, we were not able to assess the influence on immunogenicity of the number of injections and the level of NT before revaccination as follows: the former because of missing data and the latter because serology could not be performed on stored plasma. Indeed, we aimed to perform serologic testing on stored samples, but only stored plasma and not serum samples were available. Despite the fact that the NT determination was indifferently performed on plasma or serum in the study from the Swiss HIV cohort,13 reproducibility of results between both kinds of samples was not achieved in our hands.
Second, log10 transformation of antibody titers and linear regression analysis of factors associated with the magnitude of antibody titer have been conducted in previous studies13,24,25 and in ours. However, to our knowledge, there are no data supporting the hypothesis that level of protection against YF is correlated with the level of antibodies (ie, response to immunization is dichotomous and is not a dose–response type). Furthermore, when assessing the determinants of the “magnitude” of immunogenicity by linear regression, authors statistically postulated that an increase in NT from 1:10 to 1:20 is equivalent to an increase from 1:20 to 1:40, which is questionable. The association between NT level and its determinants must therefore be considered with caution.
Despite the demonstration that NT ≥1:10 conferring clinical protection in experimental models has not yet been established, neutralizing antibody response ≥1:10 has been considered as a surrogate marker of clinical protection in previous studies.13,15 However, protection against YF involves both innate and adaptive immune responses, such as cellular immunity mediated by cytotoxic CD8 T cells,26 or dendritic cells, which present antigen cells to CD4 T cells and influence subsequent cytotoxic and humoral responses. These mechanisms of immune responses to the 17DYF vaccine might also be impaired in HIV patients.
For Medical Practice
Our data, and those from previous studies, are reassuring concerning the safety of YF immunization in HIV-infected patients, including those with CD4T-cell counts <200 cells per cubic millimeter. Nevertheless, a case of fatal encephalitis after YF immunization has been reported in an HIV-infected patient with a CD4 T-cell count of ∼100/mm3,6 which suggests caution when administering the YF vaccine in deeply immunocompromised patients. Additionally, polymorphisms in CCR5 and RANTES genes were evidenced in an HIV-negative patient suffering from YF vaccine–associated viscerotropic disease,27 which raises the issue of the safety of YF vaccination in HIV-infected patients on maraviroc, an antiretroviral agent belonging to the emerging class of anti-CCR5.28
Overall, YF immunization is effective in HIV patients. Suppressed plasma HIV RNA, and a longer delay in undetectable viral load at immunization, improves the response to YF immunization, whereas a low CD4 T-cell count is not a determinant of failure of YF immunization. Therefore, control of NT needs to be performed after immunization in HIV-infected travelers without a combined antiretroviral therapy, otherwise combined antiretroviral therapy initiation may be considered before YF immunization.
In HIV patients, seroprotection may wane more quickly than in immunocompetent subjects. Thus, in cases where there is travel in an YF-endemic country, we suggest that NT determination be performed before departure, whatever the time since a previous immunization, especially in patients with noncontrolled HIV RNA at the time of travel or immunization, and in patients immunized before diagnosis of HIV infection. Therefore, patients with nonprotective NT might be revaccinated earlier than recommended by current guidelines, whereas revaccination might be avoided in patients with persisting YF antibodies.
We thank N. Desplanques for managing the logistics of the study, H. Bougouffa for her help in the data management, M. Debruyne for her technical assistance (NT determination), and A. Boyd for careful editing of the submitted manuscript. The authors are also grateful to the patients and to the clinical teams for their commitment to the study.
1. Robertson SE, Hull BP, Tomori O, et al.. Yellow fever: a decade of reemergence. JAMA. 1996;276:1157–1162.
2. Staples JE, Monath TP. Yellow fever: 100 years of discovery. JAMA. 2008;300:960–962.
3. Poland JD, Calisher CH, Monath TP, et al.. Persistence of neutralizing antibody 30–35 years after immunization with 17D yellow fever vaccine. Bull World Health Organ. 1981;59:895–900.
4. Khromava AY, Eidex RB, Weld LH, et al.. Yellow fever vaccine: an updated assessment of advanced age as a risk factor for serious adverse events. Vaccine. 2005;23:3256–3263.
5. Veit O, Hatz C, Niedrig M, et al.. Yellow fever vaccination in HIV-infected patients. HIV Ther. 2010;4:17–26.
6. Kengsakul K, Sathirapongsasuti K, Punyagupta S. Fatal myeloencephalitis following yellow fever vaccination in a case with HIV infection. J Med Assoc Thailand (Chotmaihet thangphaet). 2002;85:131–134.
7. Cetron MS, Marfin AA, Julian KG, et al.. Yellow fever vaccine. Recommendations of the Advisory Committee on Immunization Practices (ACIP), 2002. MMWR Recomm Rep. 2002;51(RR-17)1–11; quiz CE1–4.
8. ; Goujon C, Tohr M, Feuillie V, et al. Good tolerance and efficacy of yellow fever vaccine among subjects carriers of HIV [abstract 32]. Presented at: Program and abstracts of the 4th International Conference on Travel Medicine; 1995; Acapulco, Mexico..
9. Sibailly TS, Wiktor SZ, Tsai TF, et al.. Poor antibody response to yellow fever vaccination in children infected with human immunodeficiency virus type 1. Pediatr infect Dis J. 1997;16:1177–1179.
10. Receveur MC, Thiebaut R, Vedy S, et al.. Yellow fever vaccination of human immunodeficiency virus-infected patients: report of 2 cases. Clin Infect Dis. 2000;31:E7–E8.
11. Tattevin P, Depatureaux AG, Chapplain JM, et al.. Yellow fever vaccine is safe and effective in HIV-infected patients. AIDS. 2004;18:825–827.
12. Pistone T, Verdière CH, Receveur MC, et al.. Immunogenicity and tolerability of yellow fever vaccination in 23 French HIV-infected patients. Curr HIV Res. 2010;8:461–466.
13. Veit O, Niedrig M, Chapuis-Taillard C, et al.. Immunogenicity and safety of yellow fever vaccination for 102 HIV-infected patients. Clin Infect Dis. 2009;48:659–666.
14. Niedrig M, Lademann M, Emmerich P, et al.. Assessment of IgG antibodies against yellow fever virus after vaccination with 17D by different assays: neutralization test, haemagglutination inhibition test, immunofluorescence assay and ELISA. Trop Med Int Health. 1999;4:867–871.
15. Lang J, Zuckerman J, Clarke P, et al.. Comparison of the immunogenicity and safety of two 17D yellow fever vaccines. Am J Trop Med Hygiene. 1999;60:1045–1050.
16. Overton ET, Nurutdinova D, Sungkanuparph S, et al.. Predictors of immunity after hepatitis A vaccination in HIV-infected persons. J Viral Hepat. 2007;14:189–193.
17. Laurence JC. Hepatitis A and B immunizations of individuals infected with human immunodeficiency virus. Am J Med. 2005;118(suppl 10A)75S–83S.
18. Overton ET, Sungkanuparph S, Powderly WG, et al.. Undetectable plasma HIV RNA load predicts success after hepatitis B vaccination in HIV-infected persons. Clin Infect Dis. 2005;41:1045–1048.
19. Bailey CL, Smith V, Sands M. Hepatitis B vaccine: a seven-year study of adherence to the immunization guidelines and efficacy in HIV-1-positive adults. Int J Infect Dis. 2008;12:E77–E83.
20. Puthanakit T, Aurpibul L, Yoksan S, et al.. Japanese encephalitis vaccination in HIV-infected children with immune recovery after highly active antiretroviral therapy. Vaccine. 2007;25:8257–8261.
21. Levin MJ, Gershon AA, Weinberg A, et al.. Immunization of HIV-infected children with varicella vaccine. J Pediatr. 2001;139:305–310.
22. Evison J, Farese S, Seitz M, et al.. Randomized, double-blind comparative trial of subunit and virosomal influenza vaccines for immunocompromised patients. Clin Infect Dis. 2009;48:1402–1412.
23. Hung CC, Chang SY, Su CT, et al.. A 5-year longitudinal follow-up study of serological responses to 23-valent pneumococcal polysaccharide vaccination among patients with HIV infection who received highly active antiretroviral therapy. HIV Med. 2010;11:54–63.
24. Hepburn MJ, Kortepeter MG, Pittman PR, et al.. Neutralizing antibody response to booster vaccination with the 17d yellow fever vaccine. Vaccine. 2006;24:2843–2849.
25. Roukens AH, Vossen AC, Bredenbeek PJ, et al.. Intradermally administered yellow fever vaccine at reduced dose induces a protective immune response: a randomized controlled non-inferiority trial. PloS One. 2008;3:e1993.
26. Miller JD, van der Most RG, Akondy RS, et al.. Human effector and memory CD8+ T cell responses to smallpox and yellow fever vaccines. Immunity. 2008;28:710–722.
27. Pulendran B, Miller J, Querec TD, et al.. Case of yellow fever vaccine–associated viscerotropic disease with prolonged viremia, robust adaptive immune responses, and polymorphisms in CCR5 and RANTES genes. J Infect Dis. 2008;198:500–507.
28. Roukens AH, Visser LG, Kroon FP. A note of caution on yellow fever vaccination during maraviroc treatment: a hypothesis on a potential dangerous interaction. AIDS. 2009;23:542–543.