Reduced Risk of Serious Pneumococcal Infection Up to 10 Years After 7-Valent Pneumococcal Conjugate Vaccine in Arthritis Patients


 Background: To examine rates of serious pneumococcal infections up to 10 years after vaccination with 7-valent conjugated pneumococcal vaccine (PCV7) in patients with arthritis compared to non-vaccinated arthritis patients.Methods: In total, 595 adult arthritis patients (rheumatoid arthritis; RA=342, 80% women and spondylarthropathy; SpA=253, 45% women) received one dose of PCV7. Mean age/disease duration were 62/16 and 51/14 years, respectively. For each patient, 4 matched reference subjects were identified.At vaccination, 420 patients received bDMARDs (anti-TNF=330, tocilizumab=15, abatacept=18, anakinra=1, rituximab=56). Methotrexate was given as monotherapy (n=86) or in combination with bDMARD (n=220). 89 SpA patients received NSAIDs without DMARD. The Skåne Healthcare Register was searched for ICD-10 diagnostic codes for pneumococcal infections (pneumonia, lower respiratory tract infection, septicemia, meningitis, septic arthritis) between January 2000 and December 2018. Frequency of infections after vs before vaccination were calculated (relative risks). Relative risk ratio (RRR) and relative risk reduction (1-RRR) were calculated comparing patients vs non-vaccinated references. Kaplan-Meier and Cox regression were used to investigate time to first event and predictors of infections.Results: Among vaccinated RA and SpA patients, there was a significant relative risk reduction of pneumonia and all serious infections; 53% and 46%, respectively. There was no significant difference in time to first pneumonia or all serious infections after vaccination between patients and references. Higher age, RA diagnosis and concomitant prednisolone were associated with infections.Conclusion: One dose of pneumococcal conjugate vaccine may decrease risk of serious pneumococcal infection up to 10 years in patients with arthritis receiving immunomodulating treatment. Clinical trial registration number: EudraCT EU 2007-006539-29 and NCT 00828997


Introduction
Long-term protective effect following pneumococcal immunization is likely due to a combination of protective serum antibody levels, immunological memory and herd immunity. short-term immunogenicity and safety, but with impaired antibody response among patients with certain ongoing disease modifying anti rheumatic treatments (DMARDs) such as methotrexate and rituximab. among patients with IRD are scarce. (7) In the present study we present one of few evaluations of putative serious pneumococcal infection rates up to 10 years before and after administration of heptavalent conjugated pneumococcal vaccine (PCV7) in patients with in ammatory arthritis, compared to non-vaccinated arthritis patients.

Methods
Adult patients with rheumatoid arthritis (RA) and spondylarthropathy (SpA) including psoriatic arthritis (PsA) patients, at the Department of Rheumatology at Skåne University Hospital, Sweden with established diagnoses and regular follow-up were eligible for the study. The inclusion criteria partly consisted of 6 pre-set treatment groups. RA patients were to have ongoing treatment with 1) methotrexate (MTX), 2) MTX and bDMARD, or 3) bDMARD as monotherapy. SpA patients were to have ongoing treatment with 1) bDMARD, 2) MTX and bDMARD, or 3) NSAID only. All patients, apart from the NSAID treated SpA patients, could also have other concurrent ongoing sDMARD treatment and or glucocorticoids. Patients coming for their regular check-up and meeting the inclusion criteria were consecutively offered to enter the study. Between May 2008 and February 2012, 609 patients received one single intramuscular dose of PCV7. 595 patients were included in the present investigation, 13 patients moved from the area. Informed constent was obtained from all patients before inclusion in the study.
For each vaccinated patient, 4 reference subjects were identi ed in the Skåne Health Care Register including all ICD-codes from the entire health care system. The reference subjects (n=2379) were matched for arthritis diagnosis, age, sex, and residency in the same geographical area at least 3 years before and after the index date; that is the date of vaccination of their corresponding patient. References were presumed to be non-vaccinated for pneumococci with a conjugated pneumococcal vaccine, following the current national guidelines. The reference subjects were unaware of their participation and the ethical permission did not include review of their medical records.
The Skåne Healthcare Register was searched for ICD-10 diagnostic codes for serious putative pneumococcal infections between January 2000 and December 2018 for both patients and reference subjects. The outcome events included ICD-codes corresponding to community acquired pneumonia (J13.9, J18.0, J18.1, J18.9), lower respiratory tract infection (J22.9), septicemia (A40.3), meningitis ICD-codes corresponding to pneumonia (J13, J18) were calculated separately as well as together with ICD-codes corresponding to other serious pneumococcal infections. Relative risk (RR) of these two endpoints were calculated by comparing event rates after vs before vaccination. The ratio of relative risk (RRR) was calculated comparing the RR of vaccinated patients vs the non-vaccinated reference subjects.
Relative risk reduction corresponds to 1-RRR. A generalized estimated equation (GEE) was used to handle correlated data for several events in the same individual and included time of follow up as a covariate. Absolute risk reduction (ARR) and number needed to treat (NNT), were calculated when applicable.
Survival analysis (Kaplan-Meier) and Cox regression analysis were performed to investigate time to rst event before and after vaccination, time to death and predicting factors of contracting infection after vaccination. Values given as percentage or mean (SD), range. All treatment counts were ongoing at vaccination. RA, Rheumatoid Arthritis; SpA, spondylarthropathy; DAS28, Disease Activity Score using 28 joint counts; CRP, C-reactive protein; HAQ, Health Assessment Questionnaire; RF, rheumatoid factor; anti-CCP, anti-cyclic citrullinated peptide; HLA, human leucocyte antigen; MTX, methotrexate; bDMARD, biologic disease modifying anti rheumatic drug, including anti-TNF, abatacept, tocilicumab, rituximab, and one patient on anakinra; NSAID, non-steroidal antirheumatic drug.

Results
342 RA patients and 253 SpA patients were included. Mean age was 62 and 51 years, 80 % vs 45% were women and mean disease duration was 16 and 14 years, respectively.
The number of events, relative risk (RR), ratio of relative risk (RRR), relative risk reduction (1-RRR), absolute risk reduction (ARR) and number needed to treat/vaccinate (NNT) in order to prevent one event, are all summarized in Table 2.
In the patient group (RA+SpA), the relative risk of pneumonia as well as serious putative pneumococcal infection was signi cantly reduced, 53% and 46% respectively. Among the vaccinated RA patients there was a signi cant relative risk reduction of pneumonia of 46%. The relative risk reduction of pneumonia among vaccinated SpA patients was 70%, however not statistically signi cant. The SpA patients had an absolute risk reduction of both endpoints; 2% vs 3%, and NNT of 50 and 34 respectively, see Table 2.
Time to rst event before vaccination was signi cantly shorter among the vaccinated patients for both pneumonia (p= 0.004), and for all serious pneumococcal infections (p=0.05). After vaccination there was no longer a statistically signi cant difference in time to rst event for either pneumonia (p=0.12) nor for all serious pneumococcal infections (p=0.09) including adjustment for number of the same type of event before vaccination. (Figure 1.A-D).
During the time of follow-up after vaccination, 140 vaccinated patients (24%) and 100 reference subjects (4%) deceased. Time to death was signi cantly shorter among the vaccinated cohort (p <0.0001). 15 patients and 10 reference individuals deceased within 90 days after a registered serious infection, whereof 8 patients and 8 references within 30 days. Table 3 summarizes results of uni-and multivariate Cox regression models with regard to the time after vaccination. In the univariate predictor analysis several signi cant risk factors of contracting a serious pneumococcal infection despite vaccination was seen; higher age, RA diagnosis, longer disease duration, higher disease activity at vaccination (DAS28CRP), and treatment with rituximab or prednisolone at vaccination. RA patients were on average 10 years older than the SpA patients and were to a larger extent treated with prednisolone. Gender, smoking at vaccination, treatment with methotrexate or bDMARD other than rituximab was not associated with higher risk of infection after vaccination.
When multivariate analysis was applied, only age, RA diagnosis, use of prednisolone and higher prednisolone dose at vaccination remained signi cant risk factors.

Discussion
In this report we investigate the effectiveness of one dose of PCV7 among patients with in ammatory arthritis and ongoing anti-rheumatic treatments by assessing the number of registered putative pneumococcal infections up to ten years before and after vaccination, comparing event rates with nonvaccinated reference subjects. The main results were relative risk reductions of pneumonia as well as all serious pneumococcal infections of 53% and 46%, respectively, among all vaccinated arthritis patients.
The RA patients, who were 10 years older than the SpA patients and affected by the majority of events, showed a relative risk reduction of the two endpoints of 46% and 37% respectively. The relative risk reduction of pneumonia among SpA patients was 70%, however not statistically signi cant, probably due to a low number of events. The incidence of events among the vaccinated SpA patients was higher than their references before vaccination, but lower after vaccination resulting in the absolute risk reduction of all serious pneumococcal infections of 3%, and NNT of 34. Time to rst event before vaccination was signi cantly shorter among the patients, however no signi cant difference was seen after vaccination.
Treatment with rituximab and higher disease activity at vaccination were associated with contracting infection after vaccination, but after adjustment for age, these factors were no longer signi cant predictors, possibly due to lack of power. As expected, higher age predicted a signi cant increase in risk of infection, as well as the diagnosis RA, use and dose of Prednisolone, adjusted for age.
Several studies have evaluated the antibody levels before and shortly after different kinds of pneumococcal immunizations among arthritis patients, concluding that most bDMARDs including anti-TNF does not signi cantly hamper the antibody reaction, whereas MTX and RTX usually does. (7,9) Previous . Studies indicate however cross-reactivity after both PPV and PCV between vaccine serotypes and nonvaccine serotypes. (14)(15)(16)(17) Although this phenomenon seems to decrease with age, (18) it could be suggested to be part of the protective effect of immunization seen in our study. While the live attenuated tuberculosis vaccine (BCG) has been shown to induce a nonspeci c cross-protection against other pathogens through so called trained innate immunity (19), this has not been shown in pneumococcal vaccines to this date. One could however argue that preventing one infection regardless of the pathogen, reduces the risk of contracting another. Even though this study was performed before the covid -19 pandemic, it could also be worth mentioning the signi cant inverse correlation shown between rates of pneumococcal vaccination and rates of SARS-CoV-2 infections and death. (20) Our study is not without important limitations. The calculations of event rates based on ICD-codes from a diagnostic register have weaknesses. Apart from pneumococcal infections, the registered events could obviously be representing illness caused by alternative pathogens. However, this possibility of misclassi cation is estimated to be alike in both patients and reference subjects.
Before inclusion started in 2008, 6 different treatment groups were prede ned in order to ensure the inclusion of arthritis patients with a variety of ongoing treatment (see methods). Previous studies have shown reduced antibody response among patients with ongoing methotrexate and rituximab (21), therefore we wanted to include patients with an expected bene t of an immunization. All patients meeting the inclusion criteria and concurrently tting in to one of the treatment groups were consecutively offered to enter, when arriving for their regular follow-up at the Department of Rheumatology in Lund. However, RA patients in remission without DMARD, on only salazopyrine, hydroxychloroquine or glucocorticoids were not offered to be included, nor were patients not showing up to their doctor's appointments due to low or absent disease activity. 75% of the included patients were on treatment with bDMARD at the time of vaccination, in comparison to on average 25% in Swedish RA patients at the time. (22)(23) The reference subjects however, were included strictly on the matching criteria (age, sex, arthritis diagnosis) and were to be registered as living in Skåne county 3 years before and 3 years after index date, for the purpose of obtaining their events in the Skåne health care register, implying residual confounding (confounding by indication).
The study protocol did not enable review of the medical records of the reference subjects, limiting the comparisons possible on a treatment group level between vaccinated and non-vaccinated. The number of reference subjects receiving a polysaccharide pneumococcal vaccine (PPV) or a conjugated pneumococcal vaccine (PCV) of any kind during the course of the study is unknown. In 2011 the EULAR recommendations included pneumococcal vaccine in the guidelines (24) but based on previous assessments of pneumococcal vaccine coverage during this time and later, (25)(26)(27) and the results presented in this study, we would argue that it probably was a minor proportion.
Overall, we interpret the differences between the groups regarding burden of infectious disease before vaccination date and mortality rate after vaccination date despite the matching on arthritis diagnosis, age, sex and geographical area as a consequence of confounding by indication, including the reference subjects having to be alive at least 3 years after index date (immortal time bias).
Before vaccination, the incidences of events were higher, and the time to rst event was shorter among the vaccinated patients compared to the references. The time to rst event after vaccination did not differ between the groups on a statistically signi cant level but the estimates still suggested a tendency of 8-9 months shorter time for the vaccinated. The mortality rates were higher and time to death was shorter in the vaccinated group. Speci c information of causes of death was not available to our group, but only a minor number of patients (n=8) and references (n=8) deceased within 30 days following a registered serious infection. The calculations are adjusted for a shorter time of follow-up among the vaccinated patients, due to this difference in mortality. Despite the mismatch however, the relative risk reduction of infection was signi cant among the vaccinated patients. The differences at baseline are clearly a weakness of the study but could be argued to strengthen the result and could possibly underestimate the vaccine e cacy.
In summary, our study showed signi cant relative risk reductions of both endpoints; pneumonia and all serious putative pneumococcal infections up to ten years after immunization in RA and SpA patients on different antirheumatic treatments despite a higher burden of disease at baseline among the vaccinated patients compared to non-vaccinated reference subjects with arthritis. Higher age, RA diagnosis, and use as well as higher dose of prednisolone were associated with increased risk of infection after vaccination, whereas treatment with methotrexate, rituximab or bDMARD were not.

Declarations
Ethical Approval and Consent to participate Ethical approval was obtained from the Regional Ethical Review Board at Lund University, Lund, Sweden. (LU 513-01) Oral and written consent of participation was obtained from all patients receiving vaccination before inclusion in the study.
Consent for publication Oral and written consent for publication was obtained from all patients receiving vaccination before inclusion in the study.
Data availability statement For reasons related to the legal framework governing the raw data used for this study, individual-level data cannot be freely shared. For reasonable requests for study data, please contact the corresponding author.
Disclosure statement Prevenar7 vaccine was provided by Wyeth Pharmaceuticals. The sponsors had no role in study design, data collection, data interpretation or writing of the report. The authors declare no con icts of interest.