Safety and Immunogenicity of Yellow Fever 17D Vaccine in Adults Receiving Systemic Corticosteroid Therapy: An Observational Cohort Study


  • Agents and Chemotherapy, September 17–20, 2011, Chicago, Illinois.

CIC de Vaccinologie Cochin Pasteur, Hôpital Cochin, 27 rue du faubourg Saint Jacques, 75679 Paris Cedex 14, France. E-mail:



To assess the safety and immunogenicity of live attenuated yellow fever (YF) 17D vaccine in adults receiving systemic corticosteroid therapy.


All adult travelers on systemic corticosteroid therapy who had received the YF17D vaccine in 24 French vaccination centers were prospectively enrolled and matched with healthy controls (1:2) on age and history of YF17D immunization. Safety was assessed in a self-administered standardized questionnaire within 10 days after immunization. YF-specific neutralizing antibody titers were measured 6 months after vaccination in patients receiving corticosteroids.


Between July 2008 and February 2011, 102 vaccine recipients completed the safety study (34 receiving corticosteroids and 68 controls). The median age was 54.9 years (interquartile range [IQR] 45.1–60.3 years) and 45 participants had a history of previous YF17D immunization. The median time receiving corticosteroid therapy was 10 months (IQR 1–67 months) and the prednisone or equivalent dosage was 7 mg/day (IQR 5–20). Main indications were autoimmune diseases (n = 14), rheumatoid arthritis (n = 9), and upper respiratory tract infections (n = 8). No serious adverse event was reported; however, patients receiving corticosteroids reported more frequent moderate/severe local reactions than controls (12% and 2%, respectively; relative risk 8.0, 95% confidence interval 1.4–45.9). All subjects receiving corticosteroids who were tested (n = 20) had neutralizing antibody titers >10 after vaccination.


After YF17D immunization, moderate/severe local reactions may be more frequent in patients receiving systemic corticosteroid therapy. Immunogenicity seems satisfactory. Large-scale studies are needed to confirm these results.


Yellow fever (YF) 17D vaccine is either strongly recommended or mandatory for people traveling to endemic regions of Africa and Latin America ([1]). Reactions to YF17D vaccine are typically mild ([2, 3]). Headaches, myalgia, low-grade fevers, or other minor symptoms can occur for 5–10 days after immunization. As is the case with other live attenuated vaccines, its administration is contraindicated in patients with immunosuppression because of the potential for severe complications. In patients taking systemic corticosteroids, the amount of systemically absorbed corticosteroids and the duration of administration needed to suppress the immune system are not well defined. US guidelines state that corticosteroid therapy is not usually a contraindication to live attenuated vaccines when administration is either short term (i.e., <14 days) and/or a low to moderate dosage (<20 mg of prednisone or equivalent per day for adults) ([1]). British and French guidelines recommend that for long-term treatments (i.e., <14 days), adults should not be vaccinated above the prednisone or equivalent dosage of 10 mg per day ([4, 5]). Therefore, the recommendation between 10 and 20 mg/day is not consensual.

There are few data in the literature on the safety and immunogenicity of the YF17D vaccine in patients with immunosuppression. Most reports focus on people living with human immunodeficiency virus (HIV) ([6-8]). In patients immunized with theYF17D vaccine after HIV diagnosis, primary and long-term immune responses are impaired compared to healthy individuals. Based on these limited data, there is no evidence of increased incidence of severe adverse reactions in patients infected with HIV with a CD4 cell count ≥350 cells/mm3. However, there is no specific study on the use of YF17D vaccines in people receiving systemic corticosteroids, and national guidelines are mostly based on expert advice. We therefore conducted an observational cohort study to assess the safety and immunogenicity of the YF17D vaccine in adult travelers receiving systemic corticosteroid therapy.

Box 1. Significance & Innovations

  • Based on our data, corticosteroid therapy is not a contraindication to the live attenuated yellow fever 17D vaccine when administration is either short term (i.e., <14 days) and/or a low to moderate dosage (<20 mg of prednisone or equivalent per day for adults); however, the sample size is too small to categorically exclude a low risk of a severe adverse event.
  • Patients should be warned of possible occurrence of moderate/severe local reactions after immunization, even for short-term therapies (i.e., <14 days).
  • Immunogenicity seems satisfactory in patients taking corticosteroids, since all subjects (n = 20) had mounted a yellow fever–specific antibody response by 6 months after immunization.


Study design and population

The Amarcor study was conducted between July 2008 and February 2011 in 63 vaccination centers of metropolitan France. Recruitment and followup were based on telephone calls and e-mails to participating centers and patients during the study period. Physicians were invited to prospectively include all adults receiving systemic corticosteroid therapy, with no restrictions on daily dose or duration, who were planning to travel to an area with risk of YF transmission. The study was strictly observational and the decision to vaccinate the patient was taken by the physician alone, according to his or her usual practice. The study protocol was approved by the Ethical Review Committee CPP Ile de France III, Paris, France. For each patient enrolled, the practitioner completed a questionnaire collecting data on the patient's age, sex, place of birth, previous administration of the YF17D vaccine and date of immunization, duration of corticosteroid therapy, daily dose at the start of therapy and at enrollment, as well as underlying conditions. If the physician decided not to vaccinate the patient, reasons were documented in the questionnaire. If the physician administered the vaccine, the patient was invited to participate in the study by signing informed consent and attending prospective followup. After being vaccinated, participants were asked to record, on a self-administered questionnaire supplied by the study, all local (at the injection site) or systemic signs and symptoms occurring in the first 10 days after vaccination. This group of patients constituted the corticosteroids group.

Adverse event (AE) assessment and immunogenicity

Safety data were collected according to a safety toxicity grading scale adapted from the recommendations of the Food and Drug Administration (FDA) ([9]). Both local (pain, tenderness, and erythema/redness) and systemic AEs (fatigue, joint pain/myalgia, headaches, nausea/vomiting, and fever) were documented. The timing and duration of each reaction were collected and the intensity was quoted by the patient as mild, moderate, or severe, based on the impact on daily activities. Serious AEs were defined according to the FDA definition ( as those causing death, hospitalization, disability/permanent damage, or congenital anomaly/birth defect. These were reported according to a serious AE reporting procedure in the 6 months following immunization.

Each participant in the corticosteroids group was matched to 2 healthy travelers on age (±10 years) and history of previous YF17D vaccination. Safety was assessed in the same way in these participants (reported as controls) within 10 days after vaccination.

Participants in the corticosteroids group were also offered to be tested for YF17D-specific neutralizing antibody titers in the 6 months after vaccination. Neutralizing antibody titers were measured in a centralized laboratory (Cerba, Saint-Ouen l'Aumône, France) using the plaque reduction neutralization test, which is the standard technique for assessing humoral responses to YF17D immunization ([2, 10]). Dilutions ranged from 1:10 to 1:80, and neutralizing antibody titers ≥1:10 were defined as protective.

Statistical methods

Results are expressed as the median (interquartile range [IQR]) for continuous variables and the number (percentage) for categorical variables. The effects of covariates were studied using the Wilcoxon or Fisher's exact test for continuous or categorical variables, respectively. The primary outcome was the reporting of at least one moderate or severe reaction (local or systemic). Relative risks (RRs) and their 95% confidence intervals (95% CIs) were estimated by Miettinen's method for individually matched series ([11]). Statistical analyses were performed using the R 2.11.1 statistical package (R Development Core Team, R Foundation for Statistical Computing []).



Between July 2008 and February 2011, 131 participants were included in 24 vaccination centers (117 vaccinated and 14 not vaccinated). Among the 117 vaccinated participants, 102 (87%) completed followup (34 and 68 in the corticosteroids and controls groups, respectively) (Figure 1) and 20 (59%) in the corticosteroids group completed the immunogenicity assessment. The median age in both groups was 55 years (IQR 45–60 years), 60% of the participants were women, and 77 (75%) were born in metropolitan France. Overall, 45 patients of the 102 participants had been previously vaccinated against YF. This proportion did not differ between the corticosteroids and controls groups, since participants were matched on age and history of previous YF17D vaccination (Table 1). The main indications for corticosteroid therapy were rheumatoid arthritis (n = 9), chronic inflammatory diseases (n = 14), and upper respiratory tract infections (4 acute sinusitis and 4 pharyngitis). The remaining 3 patients were treated for Bell's palsy, dental extraction, and allergy. All patients with upper respiratory tract infections except 1 had been treated with corticosteroids for <7 days (median 3.5 days) with prednisone or equivalent dosages of ≥20 mg/day (median 30 mg/day, IQR 20–60). In the category of long-term therapies (≥15 days, n = 24), all but 1 were treated for either chronic inflammatory disease (n = 14) or rheumatoid arthritis (n = 9). All were receiving prednisone or equivalent of <20 mg/day (21 of 24 were receiving prednisone or equivalent of <10 mg/day). No participants in either the corticosteroids or controls group were treated with other immunosuppressive drugs.

Figure 1.

Flow chart of study participants.

Table 1. Characteristics of the study participants*
 Corticosteroids (n = 34)Controls (n = 68)Pa
  1. Values are the median (interquartile range) unless otherwise indicated.
  2. aBy the Wilcoxon or Fisher's exact test.
  3. bGraves', Behçet's, Sharp's, or Crohn's disease; hemorrhagic rectocolitis; giant cell arteritis; polymyalgia rheumatica; sarcoidosis; lupus; multiple sclerosis; and systemic sclerosis.
Age at inclusion, years55 (43–59)55 (46–61)0.69
Women, no. (%)22 (65)39 (57)0.53
Region of origin metropolitan France, no. (%)24 (71)53 (80)
Previous yellow fever vaccination, no. (%)15 (44)30 (44)1
Time since last vaccination, years15 (13–18)13 (11–21)0.40
Indication for corticosteroid therapy, no. (%)   
Rheumatoid arthritis9 (27) 
Chronic inflammatory diseasesb14 (41) 
Upper respiratory tract infections8 (24) 
Others3 (9) 
Dose and duration of corticosteroids   
Time receiving corticosteroids, months10 (<1 to 67) 
Long-term (≥15 days) therapies, no. (%)24 (71) 
Prednisone or equivalent dosage, mg/day7 (5–20) 


No serious AEs were reported. The most frequent AEs reported by the participants were fatigue (n = 23), joint pain/myalgia (n = 23), headaches (n = 19), and pain at the site of vaccine injection (n = 13). Local AEs started a median of 1 day (IQR 0–1 day) after vaccination and lasted for a median of 2 days (IQR 1–3 days). For systemic AEs, the median onset was 3 days (IQR 1–6 days) and the median duration was 2.5 days (IQR 2–7 days). Neither onset nor the duration of local and systemic AEs differed between the corticosteroids and controls groups (P = 0.19 and 0.26, respectively).

The overall report of at least one local or systemic AE did not differ between the groups (RR 1.3, 95% CI 0.9–1.7) (Table 2). However, patients receiving corticosteroids reported more frequent moderate/severe local AEs than controls (12% and 2%, respectively; RR 8.0, 95% CI 1.4–45.9). The frequency of moderate/severe systemic AEs did not differ between the groups (32% and 22%, respectively; RR 1.5, 95% CI 0.9–2.5). Participants vaccinated for the first time did not report significantly more moderate/severe AEs compared to others (3 [6%] of 55 and 1 [2%] of 45 for local reactions, respectively; P = 0.62 and 17 [31%] of 55 and 8 [18%] of 45 for systemic reactions, respectively; P = 0.17). The report rate of moderate/severe local AEs in the corticosteroids group according to duration and daily dose of steroids as compared to the controls group is shown in Figure 2. Both local and systemic AEs were more frequently reported in the subgroup of short-term/high-dose therapies, but the difference was not significant (P = 0.79).

Table 2. Safety of the yellow fever 17D vaccine*
 Corticosteroids (n = 34)Controls (n = 68)RR (95% CI)
  1. Values are the number (percentage) of participants reporting adverse events (AEs) unless otherwise indicated. RR = relative risk; 95% CI = 95% confidence interval.
Local AEs   
≥1 local AE9 (27)13 (19)1.4 (0.7–2.6)
≥1 moderate/severe local AE4 (12)1 (2)8.0 (1.4–45.9)
Systemic AEs   
Joint pain/myalgia   
≥1 systemic AE15 (44)29 (43)1.0 (0.7–1.6)
≥1 moderate/severe systemic AE11 (32)15 (22)1.5 (0.9–2.5)
Figure 2.

Percentage of participants who reported at least one moderate/severe local or systemic reaction within 10 days after yellow fever vaccination, according to duration (<15 or ≥15 days) and daily dosage (<10 or ≥10 mg/day) of corticosteroids.


Twenty participants from the corticosteroids group were tested for neutralizing antibody titers within a median of 65 days (IQR 55–118 days) after immunization. Their median age was 55 years (IQR 38–60 years), 11 (55%) had a history of previous YF17D immunization, and 17 (85%) had been treated with corticosteroids for >15 days (median 35 months). Indications of corticosteroids were rheumatoid arthritis (n = 9), chronic inflammatory diseases (n = 8), and upper respiratory tract infections (n = 3). All had neutralizing antibody titers ≥1:10. The geometric mean reciprocal neutralizing antibody titer did not differ between participants who had a history of previous YF17D vaccination and those who did not (56.6 UI/ml [95% CI 43.8–73.1] and 51.5 UI/ml [95% CI 33.8–78.4], respectively; P = 0.94).

Patients receiving corticosteroids and not vaccinated

Fourteen patients receiving corticosteroids were not vaccinated. Their median age was 42 years (IQR 35–58 years), 7 were women, 4 were born in metropolitan France, and 4 were born in Sub-Saharan Africa. Five had a history of previous YF17D vaccination. The median time receiving corticosteroids was 42 months (IQR 2–39 months) and the median dosage of prednisone or equivalent was 12.5 mg/day (IQR 5.6–23.8). Nine patients were followed for a chronic inflammatory disease, 3 had solid organ transplants, and 1 had received corticosteroids for laryngitis and another for otalgia. Nine were planning to travel to Sub-Saharan Africa and 5 to South America. In 7 patients (50%), vaccination had been denied because of concomitant administration of an immunosuppressant drug (mycophenolate mofetil, n = 3; azathioprine, n = 2; methotrexate, n = 2; cyclosporin, n = 2; and tacrolimus, n = 1). For 6 other patients, the corticosteroid therapy itself was the main reason for vaccination being denied. In these cases, the dosages of corticosteroids were between 5 and 60 mg/day.


In this observational cohort study of 34 patients receiving systemic corticosteroids and 68 controls, no severe adverse reaction to the YF17D vaccine was observed. However, moderate/severe local AEs were experienced 8 times more frequently in participants receiving corticosteroids. No difference was observed for systemic AEs. By 6 months after vaccination, all participants receiving corticosteroids had neutralizing antibody titers above the protective level.

In healthy adults, reporting of local symptoms after YF17D vaccination varies widely between studies from 3–40% ([3, 12, 13]), which is consistent with our findings in the control group (n = 13 [19%]). To the best of our knowledge, there are no data in the literature on the safety of the YF17D vaccine in people specifically taking corticosteroids. Few data exist for other categories of immunocompromised patients, such as people living with HIV ([6-8, 14-17]) or those treated with immunosuppressive therapies to prevent graft rejection ([18]) or for chronic rheumatic diseases ([19, 20]). Apart from one fatal case of myeloencephalitis following YF vaccination in an adult with HIV infection and low CD4 counts ([21]), no serious AEs following YF vaccination have been observed in these populations. Moreover, local reactions do not seem to be more frequently reported in these retrospective studies; however, AEs had not been prospectively asked about in a standardized questionnaire. Safety of other live or inactivated vaccines in terms of local events is generally satisfactory in patients with impaired immunity, being comparable to that in healthy individuals ([22]). To our knowledge, there was only one recent report of more frequent local transient reactions at the site of the injection in patients treated with immunosuppressant drugs (methotrexate, prednisone, and/or tumor necrosis factor α blockers) for rheumatic diseases, compared to healthy subjects after administration of the MF59-adjuvanted H1N1 virus monovalent influenza vaccine ([23]). On the other hand, some severe systemic AEs have been observed after administering live attenuated vaccines in immunocompromised people caused by disseminated vaccinal disease ([24-27]).

There are very few published data on the immunogenicity of YF17D vaccines in immunosuppressed individuals and none at all on patients taking corticosteroids. Patients infected with HIV respond to the 17D vaccine with lower neutralizing antibody titers, more often demonstrating nonprotective neutralizing antibody titers, and may experience a more rapid decline in neutralizing antibody titers during followup than healthy individuals ([7]). Shorter duration of undetectable plasma HIV RNA and higher plasma HIV RNA levels at immunization have been identified as independent predictors of impaired response ([6]). In rheumatoid arthritis patients receiving infliximab therapy revaccinated by the 17D vaccine, postimmunization neutralizing antibody titers tended to be lower in patients than in controls, but due to the small number of patients, a formal statistical analysis was not performed ([20]).

Lack of other data in the published literature corroborating our findings, as well as intrinsic limitations of our study, must be taken into account when interpreting our results. First, the sample size of our study, as well as those of others, was too small to categorically exclude a low but relevant risk of severe AEs in these immunocompromised populations. Despite the multicenter design and active recruitment based on followup calls and e-mails to participating centers, only 40 vaccinated individuals receiving corticosteroids were identified in 2 years. Some potential participants may have been missed by the investigators. However, before participating in the study, vaccination centers were asked to declare the number of patients receiving corticosteroids vaccinated against YF in the past 12 months. Only 12 (38%) reported at least one vaccination, which is in perfect agreement with the rate of participation (24 [38%] of 63 centers included at least one patient). This suggests that our study sample size reflected the realistic annual rate of YF immunization in patients receiving corticosteroids in France. Second, the robustness of our findings on toxicity must be confirmed in larger cohorts, since only one conflicting report of an AE would have led to non–statistically significant results. Third, as in all nonrandomized observational studies, major local toxicity in people receiving corticosteroids could only be the consequence of an observation bias in this group. Assessment of local and systemic AEs during followup by a blinded clinician in all participants would have been preferable. However, it is likely that such bias would have affected the reporting of both local and systemic AEs, whereas we observed similar reported rates of systemic AEs. Finally, complete evaluation of immunogenicity would have required prevaccination testing of neutralizing antibody titers to identify true seroconversions, as well as additional serologic testing to assess maintenance of antibodies over time. Indeed, the single postvaccination antibody determination does not distinguish a boosted immune response, which presumably occurred in participants who had been previously vaccinated, and a primary immune response in recipients who had never received the vaccine.

Only 40 vaccinated individuals receiving corticosteroids were recruited in 2 years. Since the prevalence of systemic corticosteroid use is estimated to be approximately 1–3% in the general population ([28]), people receiving corticosteroids seem to be underrepresented in the population of travelers to YF-endemic countries. Even if these patients represent a minority of adults undergoing YF17D vaccination, they are likely to increase in the upcoming years due to aging populations and wider use of corticosteroids. Therefore, safety and immunogenicity of YF vaccines are of particular concern in these populations.

Our results are too preliminary to draw general recommendations on the use of YF17D vaccines in people undergoing systemic corticosteroid therapy. However, with a prednisone or equivalent dosage of 10 mg/day, the YF17D vaccine seems to be well tolerated. For higher doses, patients should be warned about possible major local AEs, even for short-term therapies (i.e., <2 weeks). Large-scale studies are needed to confirm these results. In the near future, an inactivated YF vaccine would represent a very valuable alternative in these situations ([29]).


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Kernéis had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Kernéis, Launay, Ancelle, van der Vliet, Wyplosz, Consigny, Hanslik.

Acquisition of data. Kernéis, Launay, Ancelle, Iordache, Naneix-Laroche, Méchaï, Fehr, Leroy, Issartel, Dunand, Wyplosz.

Analysis and interpretation of data. Kernéis, Launay, Ancelle, van der Vliet, Consigny, Hanslik.


The authors would like to thank all of the patients, nurses, and physicians in the investigation centers. The Amarcor Study Group included S. Kernéis (coordinator), T. Ancelle, P.-H. Consigny, J. P. Donne, T. Fehr, C. Goujon, T. Hanslik, O. Launay, D. van der Vliet, and B. Wyplosz.