In June 2021, a previously healthy woman in her 60s was referred with neutropenic fever. She had fatigue and malaise, starting 24 hours following the first dose of the COVID-19 vaccine mRNA-1273. On the sixth day post-vaccination, she developed fever (39°C), sore throat, coughing and vomiting. She denied taking non-steroidal anti-inflammatory drugs that could cause neutropenia (i.e. NSAIDs or metamizole). She is a non-obese, non-smoker and has no diabetes or hypertension. The only available blood count from four years earlier was normal. The physical examination was unremarkable. Specifically, it revealed no palpable lymph nodes or splenomegaly and no petechia, rash, or pharyngitis. The CT scans showed several marginally increased cervical lymph nodes without organomegaly, especially without splenomegaly. Systemic inflammation markers were very high with a C-reactive protein (CRP) of 285 mg/l (reference <5mg/l) and a ferritin peak level of 5276 ng/ml (reference <300ng/mL), compatible with macrophage activation ( Figure 1 ). Tests for SARS-CoV2 and other infections (blood cultures, multiplexed PCR panels from stool and nasal swabs, serology for hepatitis viruses, CMV, and Parvovirus B19) were negative. PET-CT revealed pronounced diffuse hypermetabolism of the hematopoietic bone marrow and spleen. Bone marrow biopsy showed infiltration of CD8⁺ T cells and hypercellularity with severely reduced granulopoiesis, consistent with pure white cell aplasia ( Figure 2A ). Interstitial T-cell lymphocytosis with some linearity and intravascular spread of the T cells was suggestive of T-LGL (19) ( Figure 2B ). Immunophenotyping of the lymphocytes in the peripheral blood and bone marrow identified a large CD8+ T cell population (~45% of all T cells) expressing CD5^dim, CD7, CD16^partial, TCRαβ, HLA-DR and that was negative for CD56, TCRγδ, CD30, CD10, CD25, and PD1 in the clinical routine lab, compatible with T-LGL. The T- and B-cell clonality and translocation analysis (20) showed a dominant TCRγ-rearrangement with a base pair length of 200 bp (type V-I) on a polyclonal background. There was no evidence of B-cell clonality or translocations (t(11;14), t(14;18)). Next-generation sequencing applying a custom lymphoma panel covering 68 genes commonly mutated in lymphomas (21, 22) identified no mutation, particularly not in STAT3 or STAT5b. A diagnosis of possible T-LGL leukemia with pure white cell aplasia was made.

Treatment with a broad-spectrum antibiotic and recombinant human G-CSF was initiated, but peripheral agranulocytosis persisted. Only after treatment with intravenous immunoglobulins (IVIG 30g; given in four doses) and dexamethasone (16mg/day) was started agranulocytosis and systemic inflammation instantly resolved ( Figure 1 ), and after two weeks, she recovered completely. During follow-up, the T-LGL population declined to about 34% of all T cells. No further treatment was initiated as she remained asymptomatic.

In January 2022, she was referred to our immunology clinic for counseling regarding completing the COVID-19 immunization series. She had a tonsillectomy for recurrent tonsillitis as a young adult. She never experienced an episode of cytopenia or lymphocytosis prior to the here described event, and had no history of autoimmunity or opportunistic infections. Detailed patient history revealed mild oral sicca and recurrent arthralgias but no arthritis. Anti-nuclear antibodies (ANA), anti-neutrophilic cytoplasmatic antibodies (ANCA), rheumatoid factor and anti-citrullinated protein antibodies (ACPA) were negative. She had been vaccinated according to the Scottish immunization plan and never experienced relevant side-effects from a vaccine. Her sister has rheumatoid arthritis and her brother retroperitoneal fibrosis.

At his time, the clinical exam was unremarkable. She had no lymphadenopathy or splenomegaly. The white blood count (WBC) and CRP were normal. Anti-SARS-CoV-2-N(ucleoprotein)-IgG/IgM were negative and anti-S(pike)-IgG/IgM at low levels (61.8 U/mL (<0.7); Roche Elecsys^® assay). The T-LGL clone was stable at about 800/ul. The CD8⁺ T cells expressed high levels of CD95 and CD57, typical for T-LGL ( Figure 2C ).

The patient’s bone marrow T cells at diagnosis ( Figure 3A ), but not those of a control bone marrow ( Supplementary Figure S1 ), and the patient’s peripheral blood immune cells during follow-up ( Figure 3B ), showed high spontaneous STAT3 phosphorylation (pSTAT3). In vitro stimulation of peripheral blood mononuclear cells of the patient with the mRNA-1273 vaccine resulted in distinct STAT3 protein phosphorylation in monocytes and T cells of the patient ( Figure 3C ). This was associated with mRNA-vaccine-induced IL-6 production in vitro ( Figure 3D ). pSTAT3 activation could be blocked by the anti-IL-6 blocking antibody siltuximab and the anti-IL-6 receptor blocker tocilizumab (TCZ) ( Figure 3E and Supplementary Figure S2 ), indicating IL-6 mediated pSTAT3 upregulation. The mRNA vaccine also activated T cells IL-6-dependently ( Figure 3F ).

The IL-6/STAT3 pathway can be activated by toll-like receptors, including TLR-3 (15). Virus-derived double-stranded RNA (dsRNA), but also endogenous mRNA, can act as activators of TLR3 (23). The novel mRNA vaccines were modified to dampen RNA-mediated TLR3 activation by reducing dsRNA formation and introducing nucleotides with less inflammatory properties (16). However, even modified mRNA vaccines activate innate immunity via TLR3 (16). We confirmed TLR3 activation by stimulating peripheral blood mononuclear cells with mRNA-1273 resulting in increased phosphorylation of TLR3 ( Figure 3G ). Combined, our data suggest that the mRNA-1273 activates (innate) immune cells via TLR to secrete IL-6, subsequently resulting in STAT3 phosphorylation.

We next tested for differences in the in vitro immune activation between different COVID-19 vaccines, specifically mRNA-1273 and the adenoviral vector vaccine Ad26.COV2.S. We observed lower pSTAT3 activation ( Figure 4A ), IFNγ production in CD8⁺ T-cells ( Figure 4B ), as well as interferon-stimulated genes (ISG) ( Figure 4C ) and IL-6 ( Figure 4D ) induction when stimulating the patients’ PBMC with Ad26.COV2.S compared to mRNA-1273. Additionally, direct ex vivo stimulation of whole blood with the mRNA vaccine resulted in strong neutrophil degranulation in the patient, which could not be recapitulated after polyclonal T-cell stimulation with agonistic anti-CD3/28 antibodies, suggesting a T cell-independent effect ( Supplementary Figure S3 ). Whether this in vitro observed degranulation occurs in vivo will need to be tested.

Based on the in vitro data, we speculated that a strong mRNA vaccine-mediated IL-6/STAT3 activation contributed to the adverse reaction by TLR-mediated innate immune cell activation and indirectly by T-cell activation. Since the vector vaccine showed weaker stimulation of these pathways in vitro, we completed her COVID-19 immunization series with a single dose of Ad26.COV2.S, which was well tolerated and caused only mild T-cell-intrinsic STAT3 phosphorylation ( Figure 4E ). Her blood counts, including neutrophil count, remained stable.