• vaccination
• COVID-19
• biological therapy
• methotrexate
• arthritis, rheumatoid

Clinical trials leading to approval of the COVID-19 vaccines did not include immunocompromised individuals. Concerns have been raised that immunogenicity of the vaccines may be impaired in patients with autoimmune and inflammatory rheumatic disease (AIIRD) on immunomodulatory drugs. In December of 2020, our clinic recommended adjustments to the dosing/timing of select drugs to attempt to optimise vaccine response. The ACR COVID-19 Vaccine Clinical Guidance Task Force first provided recommendations in February 2021.1 Our recommendations were similar but differed in some key aspects (online supplemental table 1). In contrast, EULAR guidance states that ‘pausing or reducing immunosuppression may increase the risk of flare, and therefore it is generally advised not to, or only temporarily, interrupt or decrease your medication for this purpose…’2 As there is limited published data on the effects of adjustments to dosing and/or timing of immunomodulatory drugs in patients with AIIRD,3 we sought to characterise antibody responses to the SARS-CoV-2 spike protein in those who adjusted and did not make adjustments to their drug therapy in a community based rheumatology practice in Charlotte, North Carolina.

We measured SARS-CoV-2 IgG antibody levels to the receptor-binding domain of the S1 spike antigen using a semiquantitative assay (Siemens Atellica; QUEST) in which a negative result is <1.00 and a positive result is reported as 1.00–20 or >20.4 This assay had a Pearson correlation coefficient of 0.8 when compared with viral neutralisation titers.5 All tests were performed by day 14 or later following completion of the two vaccine series using BNT162b2, mRNA-1273 or single dose Ad26COV2.S vaccine. Patients were excluded if they self-reported a prior clinical history of COVID-19 infection or if labs were drawn more than 6 months after their vaccine course was completed. We compared the number of antibody positive (Ab+) versus antibody negative(Ab−) and mean Ab levels in Ab+ patients for those who did or did not make adjustments to dosing/timing of methotrexate (MTX), JAK inhibitors (JAKi), rituximab (RTX) and/or abatacept (ABA). Continuous variables were analysed using a non-parametric Wilcoxon rank-sum test and categorical variables using either χ² or Fisher’s exact test for association. Patients were not selected systematically but came to our attention during the course of routine clinical care. Patients with internet capability had been notified regarding vaccine dosing adjustments or when seen by their provider.

We studied 198 patients with AIIRD between 17 March and 14 May 2021. The mean age was 69 years and 77% were female. About 97% received the two dose mRNA vaccine series. About 77% of patients had RA. Additional baseline characteristics are provided in online supplemental table 2. There were no significant differences in age, gender, vaccine type or underlying disease for those who did or did not adjust their medication.

Findings related to percentage of total patients on MTX, JAKi, RTX and ABA with absent antibody response and comparison of antibody response with or without adjustment in dosing and/or timing are presented in table 1.

The mean time from completion of the vaccine series to measurement of antibodies was 50 days. Only 13% of all patients on RTX and 52% of all patients on ABA were Ab+. In contrast, fully 91% of patients on JAKi were Ab+ whereas 80% of patients on methotrexate alone were Ab+. Adjustments to the dose of MTX or JAKi did not significantly alter the immunogenicity as measured by IgG antibodies to the S1 spike antigen. Adjustments to the timing of vaccine in relationship to last RTX dose and last vaccine injection in relation to next RTX infusion failed to increase the likelihood of an antibody response. Adjustments to the timing of ABA (either intravenous or subcutaneous) did not significantly increase antibody response. These findings were not influenced by age of the patient, time from last vaccine dose to measurement of antibody levels, percentage of patients on steroids or the mean steroid dose (table 2).

Only 62% (123/198) of the patients reported in this cohort were seropositive for SARS-CoV-2 spike protein. This is significantly lower than the 90%–95% range of seroconversion seen in other studies6 7 Our patient population was also much older and had a significantly greater number of patients on rituximab and abatacept. We have previously shown that vaccine response is age dependent and in a multiple regression analysis declines by 5% with every advancing decade.8 Low rates of seropositivity are by now well recognised for patients on rituximab.7 9 10 Our patients on rituximab had a mean age of 68. The mean time from last rituximab infusion to first vaccine dose was 146 days (range: 106–211 days; SD: 35.34) in the 11 patients who adjusted rituximab and 53 days (range: 10–96 days; SD: 23.83) in the 27 patients who did not adjust. It is now apparent that holding vaccine dose until peripheral B cells repopulate (at least 9 months and often up to 12 months)11 12 is often required before seroconversion occurs. Information is more limited for abatacept.11 We did see a lower rate of seroconversion for intravenous abatacept (45%) versus subcutaneous abatacept (71%) which may in part be due to the older age in the intravenous group(69 vs 56, respectively).

Our initial recommendation to reduce the dose of methotrexate to 7.5 mg per week for 2 weeks following each vaccine dose differs from the ACR recommendation that methotrexate should be held for 1 week after each vaccine dose. This is based on our reading of the Park study which indicated that a dose of ≤7.5 mg methotrexate per week for 2 weeks following the influenza vaccine resulted in comparable vaccine response to the four influenza antigens studied as did holding methotrexate for 2 weeks (see Park et al, figure 3).13 As the COVID-19 mRNA vaccines required two doses, we also reasoned that the lower dose of methotrexate versus stopping methotrexate for 2 weeks would be less likely to result in disease flare.

This study has a number of limitations. We did not measure antibodies to the nucleocapsid protein but relied on the patients’ self-reported history of COVID-19 infection. The study was not randomised as to who adjusted and did not make adjustments to their medication although this did not appear to affect outcomes. We did not control for patients on RTX and ABA who were on additional immunosuppressive therapies aside from prednisone and sample size was relatively small for several of the drugs studied. Finally, we do not provide any data on cellular immune responses.

This study suggests that holding JAKi, lowering the dose of MTX or altering the timing of the vaccine in relation to RTX or ABA administration did not alter immunogenicity as measured by SARS-CoV-2 Abs to the S1 spike antigen. Although these findings may assist in future recommendations for dosing and timing of immunomodulatory therapies in the setting of COVID-19 vaccinations including boosters in patients with AIIRD, conclusive evidence for this question can only be provided by well controlled prospective randomised clinical trials.