Glucocorticoids selectively affect the memory T cell response to SARS-Cov2 spike in vaccinated and post-infected patients with systemic lupus erythematosus

Immune response to vaccines and pathogens remains unclear in patients with systemic lupus erythematosus (SLE). To investigate this, a single-center retrospective study was conducted with 47 SLE patients vaccinated against COVID-19, including 13 who subsequently developed an asymptomatic/mild disease. As compared to controls, post-vaccine response against Spike was reduced in SLE patients when considering both memory T-cells in a whole blood interferon gamma release assay (IGRA-S) and IgG anti-Spike antibody (Ab) responses. The SLE-associated defective IGRA-S response was associated with a serum albumin level below 40 g/L and with the use of glucocorticoids, while a defective IgG anti-Spike Ab response was associated with lower levels of anti-dsDNA and anti-SSA/Ro 52 kDa Abs. IGRA-S and IgG anti-Spike responses were independent from SLE activity and clinical phenotype, low complement, hypergammaglobulinemia, and lymphopenia. As compared to controls, SLE patients showed a rapid decay of anti-Spike T-cell memory and stable IgG anti-Spike Ab responses. In conclusion, both T cell and humoral anti-Spike responses were independently affected in our SLE patients cohort, which supports the exploration of both responses in the follow-up of SLE patients and especially in those receiving glucocorticoids.


Introduction
Patients with systemic lupus erythematosus (SLE) are characterized by an increased risk of infections as compared to the general population [1], and infections are a leading cause of mortality, morbidity, and hospitalization in this disease [2]. The high burden of infection in SLE is caused by the disease itself as well as by the use of immunosuppressive therapies resulting in altered B and T cell immune responses. The recent pandemic caused by SARS-Cov2 has further confirmed sensitivity of SLE patients to infections resulting in extended COVID-19 symptom duration, an increased risk of hospitalization and severe fatal outcome with an Odds ratio of 2.2 [3,4]. Part of the risk of severe or fatal covid -19 infection is related to the use of glucocorticoids, antimetabolite drugs (e.g. mycophenolate mofetil, methotrexate, azathioprine), and biotherapies (e.g. belimumab, rituximab) [5]. Consequently, vaccination represents a primary means of protection against severe COVID-19 in SLE patients to reduce SARS-Cov2 complications [6]. However, lessons from influenza, varicella-zoster and other vaccination campaigns have revealed a defect in both cellular and humoral immunity among SLE patients that is not solely related to therapeutics [7][8][9]. Accordingly, this study was designed to describe the T cell and humoral responses to COVID-19 vaccine and SARS-Cov2 infection in SLE patients, taking into account the clinico-biological characteristics of the disease and therapy.

Patients
Forty-seven unselected SLE patients immunized against COVID-19 in response to the vaccine were included in the study (Table 1). All SLE patients met the 2019 ACR/EULAR classification criteria [10]. Data regarding disease activity using a cut-off point ≥5 from the SLEDAI-2K score [11], clinical presentation, current treatment, number and time of covid-19 vaccine injections or infectious episodes were retrospectively collected from medical records. Laboratory data included lymphocyte count, albumin and gamma-globulin levels using serum protein electrophoresis, IgG anti-double stranded (ds)DNA and anti-chromatin Abs (Bioplex, Biorad, Hercules, CA), IgG anti-extractable nuclear Abs, antiphospholipid Abs such as anticardiolipin (aCL) and anti-beta2 glycoprotein I (aβ2-GPI) Abs, and the complement fractions C3, C4 and CH50 (The Binding Site, Birmingham, UK) [12][13][14][15].
Vaccinated (n = 64) and infected (n = 51) control groups comprised staff members from the medical laboratory of the University Hospital of Toulouse (CHU de Toulouse, Occitania, France), blood bank donors (EFS Toulouse, Occitania, France), and non-autoimmune patients recruited for COVID-19 infection follow-up at the Internal Medicine Department of the University Hospital of Toulouse. Some of them were described previously [16].
Blood from SLE patients was collected during a routine care visit in the medical departments to control for immunization after vaccination or infection (see Table 1), to advise patients on the risk of severe infection and opportunity of a new booster vaccine. Participants were informed and gave their consent, the related cohort ESSAi obtained approval from the ethics committee (CPP) in Paris Ile de France I under reference 2021-A03236-35.
For analysis, value from the negative control tube was subtracted from the signal obtained after stimulation with recombinant proteins. IGRA-S and IGRA-Nuc thresholds for positivity were fixed at 0.040 IU IFN-γ/mL. The test is recorded as indeterminate when the negative control is > 8 IU IFN-γ/mL or when the mitogen control <0.5 IU IFN-γ/mL, but such cases were not observed in this study.

Serological tests
The serological tests were carried out on serum and the level of IgG antibodies to SARS-Cov2 Spike mammalian cell-expressed recombinant protein was assessed by using the SARS-CoV-2 IgG II Quant assay (Abbott Laboratories, IL, USA). ELISA total values are expressed in BAU/ mL, and with an assigned cutoff at 7.14 BAU/mL, as previously described [19,20]. The SARS-CoV-2 IgG assay (Abbott Nuc) was used to detect anti-Nuc antibodies using a threshold fixed at 1.4 [21].

Statistics
Quantitative data are presented as mean ± standard deviation (SD) or as median and interquartile (IQ) 25th-75th percentile when analyzed using non-parametric assays. Receiver operating characteristic (ROC) curves were generated to determine the area under the curve (AUC), the optimal cut-off values were chosen using Youden's index, and the Cohen's kappa was used to compare techniques. Categorical data were analyzed using Fisher's exact test. Statistics were conducted using GraphPad Prism 9.2 (La Jolla, CA) software, and p-values<0.05 were considered significant.

Cellular and humoral responses to COVID-19 are altered in SLE
Memory T cell response against Spike (IGRA-S) was first evaluated in SLE patients and controls following COVID-19 vaccination and infection ( Fig. 1A-C). Quantitative (p = 0.0007) and qualitative (p < 10 − 4 ) differences were retrieved following COVID-19 vaccine response for IGRA-S with a positive response in 18/34 (52.9%) SLE patients versus 61/64 (95.3%) in controls. In contrast, no difference was observed between 13 SLE patients and 51 controls infected with SARS-Cov2 when exploring IGRA-S and IGRA-Nuc levels and/or positivity. When considering the humoral response, IgG anti-Spike Ab level was reduced among the SLE COVID-19 vaccine sub-group (p < 10 − 4 ), and IgG anti-Spike Ab positivity was 28/34 (82.4%) in SLE patients versus 64/64 (100%) in controls (p = 0.001). Within the SLE COVID-19 infected sub-group, IgG anti-Spike Ab and IgG anti-Nuc Ab were similar to the control group.
The level of agreement between T cell and humoral responses against   Spike was further evaluated showing an almost perfect Cohens' kappa (K) coefficient between IGRA-S and IgG anti-Spike Ab in controls (K = 0.91) as compared to a substantial agreement in the 47 SLE patients (K = 0.13) (Fig. 1D). Such agreement between T cell and humoral responses against Nuc was fair in both SLE patients (K = 0.59) and controls (K = 0.56), which reflects a higher sensitivity for the humoral assay. Altogether this supports the idea that the cellular and humoral responses against SARS-Cov2 Spike are altered in SLE after covid-19 vaccination.

Serum albumin level can be used as a surrogate for IGRA-S reponse
As presented in Fig. 2A, the 47 SLE patients were characterized as compared to controls by a lower lymphocyte count (p < 10 − 4 ), presence of hyper-gammaglobulinemia (p = 7 × 10 − 4 ), a lower serum albumin level (p < 10 − 4 ), and a higher CRP level (p < 10 − 4 ). Of note, T cell memory PHA response reflecting the functional integrity of T cells tended to be lower in SLE patients (p = 0,06). As these factors are known to influence IGRA and ELISA responses [22][23][24], their capacity to affect the IGRA-S and IgG anti-Spike vaccine response in SLE patients was evaluated (Fig. 2B). From such univariate analysis, only the serum albumin level was observed to be associated with IGRA-S response (p = 0.007), and the optimal threshold point using maximum Youden's index from the ROC curve analysis (AUC = 0.795) was fixed at 40 g/L with a sensitivity of 77.8% (95% CI: 58.8-91.0%) and a specificity of 85.2% (95% CI: 67.5-94.1%). This threshold may be used as a surrogate to predict a negative IGRA-S response (Fig. 2C).

Factors associated with a defective immune response in SLE
We further considered demographic parameters, clinical factors, complement fractions, and SLE-associated autoantibodies in order to test their associations with T cell activation in response to PHA used as mitogen (non specific) or Spike (specific) and to IgG anti-Spike positivity.

Drug exposures and immune response
As reported in Fig. 4, many drug combinations were used in SLE patients and among them glucocorticoid therapy was associated with IGRA-S negativity (p = 0.005). No association was retrieved with the other drug classes: hydroxychloroquine, antimetabolites, and biotherapies that included belimumab or rituximab. To determine the cutoff point of glucocorticoid dose that affects IGRA-S response, a ROC curve (AUC = 0.725, p = 0.01) was done and the greater sensitivity and specificity threshold was obtained at 8.75 mg/day of prednisone with a sensitivity of 47.4% (95% CI: 27-68%) and a specificity of 89% (95% CI: 73-96%). A defective IGRA-PHA response (<10 IFNγ IU/mL) and IgG anti-Spike Ab negativity were not associated with any drug exposure.   3.6. Accelerated T cell response decline, but not humoral response, following immunization Finally and to complete the global analysis, the cellular and humoral responses against Spike were compared between SLE patients and controls according to time elapsed since last immunization in order to compare both the post-vaccine immune response (<50 days postimmunization), and the decline of the immune response (≥50 days) (Fig. 5). We further took advantage of data from 10 SLE patients and 6 controls that had been tested twice within a delay ranging from 6 to 14 months.
Regarding memory T cell response against Spike, results revealed similar IGRA-S IFNγ levels between SLE and controls in response to the immunization (<50 days, p = 0.541), and, next, an accelerated decline was reported in SLE patients as compared to controls (>50 days, p = 0.001). The follow-up of SLE patients and controls further confirmed a significant decline of IGRA-S with time in SLE patients (p = 0.004), which was not the case in controls. Of note, such effect was conserved both when excluding from the analysis SLE patients receiving glucocorticoids and when excluding SLE patients with a serum albumin level <40 g/L (data not shown).
When considering IgG anti-Spike Ab responses (<50 days), IgG anti-Spike Ab levels were reduced in SLE patients as compared to controls (p = 0.05). After 50 days post-immunization, we observed a stable humoral response in SLE patients while there was a fast decline in controls (p = 0.02).
Altogether this supports that the reduction of the global anti-Spike immune response in SLE patients (Fig. 1) results both from an accelerated decline of the T cell immune response following a normal immunization, and from a weaker humoral response (<50 days) counterbalanced by a sustained humoral response with time.

Discussion
In this monocentric and retrospective study aimed at depicting the cellular and humoral response to SARS-Cov-2 in a cohort of SLE patients, we found strong evidence that the capacity to produce IgG anti-Spike Abs as well as maintaining a specific anti-Spike T-cell memory response are altered in SLE patients on various treatments. Our study further supports that COVID-19 mRNA does not protect from COVID-19 in 13/47 (27.7%) SLE patients but provides a protective cellular and/or humoral response after 2-4 vaccine injections against severe COVID-19. One should note, many of these lupus patients had a symptomatic COVID-19 infection during the era of the delta or omicron variant but none were severe enough to lead to hospitalization or death. Last but not least, better knowledge of the immune response following vaccination may help to better protect this population at high risk of infections.
In patients with SLE, several studies have established that mRNA COVID-19 vaccines are well tolerated, not associated with an increased risk of flares, but associated with a lower level of IgG anti-Spike Abs in association with a reduction of naïve B cell precursors at baseline [27][28][29]. The latter point is consistent with our report that the early stage of the humoral response is affected, which provides an argument to interrupt anti-metabolite drugs in the 2 weeks following COVID-19 vaccination as recently suggested [30]. Moreover, IgG anti-Spike Ab positivity with the presence of anti-dsDNA, anti-SSA/Ro 52 kDa, and anti-CL Abs together with sustained immune response after 50 days post immunization may reflect a post-vaccine expansion and survival of plasmablasts as previously reported in SLE [31,32]. In our study, immunosuppressive drugs did not significantly affect anti-Spike humoral response, which is in agreement with previous reports regarding glucocorticoids, hydroxychloroquine and belimumab [27,33]. Mycophenolate mofetil, methotrexate and rituximab are known to alter the post-vaccine humoral response but these drugs were infrequently used in our study [34]. Regarding glucocorticoids, a reduced vaccine humoral response is reported at high doses (>20 mg/day), which was prescribed in only 3 patients in our cohort [35].
Our results are consistent with previous reporting on T cell responses following influenza, varicella-zoster, and COVID-19 vaccine immunization in SLE [7,36,37]. These studies reported a protective but reduced memory T cell response in the majority of SLE patients, while the others remained unprotected. The diminished memory T cell vaccine response in SLE was associated in part with the use of glucocorticoids, but results were discordant between studies with regards to the effect of anti-metabolites, biotherapies, and hydroxychloroquine. The actual consensus is that disease flares, disease manifestations and changes in the level of SLE-autoantibodies and complement fractions are independent from the post-vaccine cellular immune response [7,36]. Our study further supports that the anti-Spike memory T cell reduction observed in SLE patients (Fig. 1, global analysis) results more from defective maintenance in the circulation rather from a defective expansion based on the report that IGRA-S levels are similar to controls in the first 50 days post-immunization. Such observation needs further prospective studies in a larger cohort [38]. Another point that deserves to be further explored is the difference observed between immunization in response to C0VID-19 vaccine and SARS-Cov2 infection in SLE patients, which supports a lower immunization of the vaccines as compared to the virus.
A decreased production of circulating IFNγ-producing T cells in response to mitogen stimulation is a hallmark of SLE patients and associated with active disease, autoantibody production, and complement activation as reported by others and us [25,[39][40][41][42]. Indeed it has been suggested that type II IFNγ dysregulation in T cells starts at the preclinical stage and precludes auto-Abs accrual (e.g. anti-chromatin, anti-Sm/RNP), and subsequently the type I IFNα signature occurs at SLE onset [43]. Type I IFNα is produced mainly by macrophages and dendritic cells and elicits a large spectrum of effects on the immune system including tissular T cell attraction and (autoreactive)-plasma cell differentiation, a process that can be reversed directly in the presence of anifrolumab and indirectly by belimumab [44,45]. Accordingly, it has been proposed that IGRA-PHA reduction reflects the IFN-dependent capacity of autoreactive memory T cells to migrate from peripheral blood to the inflamed tissues [44]. Another option is related to the polarization or exhaustion of memory T cells in SLE patients [46][47][48].
In conclusion, our study shows that the combined analysis of both humoral and T cell vaccine responses may help to depict vaccine protection against COVID-19 in SLE and other immunocompromised patients. In the previous years, protection against severe COVID-19 was provided by the infusion of monoclonal antibodies in immunosuppressed patients showing no humoral vaccine response. However, new variants are less sensitive to the more affordable monoclonal antibodies. Therefore, one could propose a new strategy of prevention of severe COVID-19 considering not only the humoral, but also the cellular response after vaccine or infection. It may be pertinent in high risk immunosuppressed patients to boost an altered cellular response, especially when no significant antibody response has been detected and when possible to promote quick tapering of glucocorticoids under 10 mg/day.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.