Systematic review and meta-analysis of epidemiologic data on vaccine response in relation to exposure to five principal perfluoroalkyl substances

By Lori Crawford, Scott A Halperin, Michael W Dzierlenga, Becky Skidmore, Matthew W Linakis, Shinichi Nakagawa, and Matthew P Longnecker
Environ Int
February 7, 2023
DOI: 10.1016/j.envint.2023.107734

Background

Epidemiologic studies of serum per- and polyfluoroalkyl substances (PFAS) and antibody response to vaccines have suggested an adverse association, but the consistency and magnitude of this association remain unclear.

Objective

The goal of this systematic review was to determine the size of the association between a doubling in perfluoroalkyl substances (PFAS) serum concentration and difference in log antibody concentration following a vaccine, with a focus on five PFAS: perfluorooctanoic acid (PFOA), perfluorooctane sulfonate (PFOS), perfluorohexane sulfonate (PFHxS), perfluorononanoic acid (PFNA), and perfluorodecanoic acid (PFDA).

Data Source

We conducted online searches of PubMed and Web of Science through May 17, 2022 and identified 14 eligible reports published from 2012 to 2022.

Study Eligibility Criteria, Participants, And Interventions

We included studies conducted in humans, including mother-child pairs, which examined serum PFAS concentration in relation to serum concentration of antibody to a specific antigen following a vaccine.

Study Appraisal And Synthesis Methods

We used the risk of bias assessment for non-randomized studies of exposure and certainty assessment method proposed by Morgan et al. (2019). Using a multilevel meta-regression model, we quantitatively synthesized the data.

Results

The 14 reports represented 13 unique groups of subjects; the frequency of studies of a given antibody was Tetanus (n = 7); followed by Diphtheria (6); Measles (4); Rubella (3); Haemophilus influenzae type b and Influenza A H1N1 (2 each); and Hepatitis A, Hepatitis B, Influenza A H2N3, Influenza B, and Mumps (1 each). There were approximately 4,830 unique participants included in the analyses across the 14 reports. The models of coefficients between antibody concentration and the five principal PFAS showed homogeneity of associations across antibody types for each principal PFAS. In the models with all antibodies treated as one type, evidence of effect modification by life stage was present for PFOA and PFOS, and for consistency, all associations were evaluated for all ages and for children. The summary associations (coefficients for difference in log[antibody concentration] per doubling of serum PFAS) with 95% confidence intervals that excluded zero ("statistical support"), and certainty of evidence ratings were as follows: for PFOA and all antibodies treated as one type in all ages, -0.06 (-0.10, -0.01; moderate) and in children, -0.10 (-0.16, -0.03; moderate); for Diphtheria in children, -0.12 (-0.23, -0.00; high); for Rubella in all ages, -0.09 (-0.17, -0.01; moderate), and for Tetanus in children, -0.12 (-0.24, -0.00; moderate). For PFOS the summary associations were, for all antibodies treated as one type in all ages, -0.06 (-0.11, -0.01; moderate) and in children, -0.10 (-0.18, -0.03; moderate); for Rubella in all ages, -0.09 (-0.15, -0.03; high) and in children, -0.12 (-0.20, -0.04; high). For PFHxS the summary associations were, for all antibodies treated as one type in all ages, -0.03 (-0.06, -0.00; moderate) and in children, -0.05 (-0.09, -0.00; low); and for Rubella in children, -0.07 (-0.11, -0.02; high). Summary associations for PFNA and PFDA did not have statistical support, but all PFAS studied tended to have an inverse association with antibody concentrations.

Limitations And Conclusions

Epidemiologic data on immunosuppression and five principal PFAS suggest an association, with support across antibodies against multiple types of antigens. Data on Diphtheria, Rubella, and Tetanus were more supportive of an association than for other antibodies, and support was greater for associations with PFOA, PFOS, and PFHxS, than for PFNA or PFDA. The data on any specific antibody were scarce. Confounding factors that might account for the relation were not identified. Nearly all studies evaluated were judged to have a low or moderate risk of bias.

View on PubMed

Topics: