Inhibitor formation is the most serious complication of FVIII replacement therapy for hemophilia A. The long-standing "danger theory" posits that inhibitors may form as part of collateral damage from immune response to a primary challenge such as an infection or vaccination. Innate immune signaling, for example through Toll-like receptors (TLRs), could be one way of triggering or reinforcing unwanted immune responses. However, the danger hypothesis has been contradicted by recent reports showing no increase in inhibitor formation in boys and animal models with hemophilia when FVIII was administered concurrently with vaccines. The aim of this study was to elucidate the influence of TLR9 signaling on FVIII inhibitor formation in hemophilia A mice.

Hemophilia A (F8e16-/-) B6/129 mice were co-injected IV with FVIII (1.5 IU) and ODN-1826 (a class B CpG oligodeoxynucleotide, 50 µg), which is a TLR9 agonist. Control mice were naïve or received FVIII only. Blood samples and spleens were collected for Bethesda assay and flow cytometry analysis 3h, 24h, 2, 3 and 7 days or 2, 4, 6 and 8 weeks after a single or repeated once-weekly injections.

After four weeks, mice co-injected with FVIII and ODN-1826 (n=4) showed ~15-fold higher inhibitor titers (median 2667 BU/mL) than mice injected with FVIII only (median 181.4 BU/mL; n=15). We also found significantly higher T follicular helper (Tfh; CD4+CXCR5+PD1+Bcl6) [F (7, 96) = 9.801, p<0.0001] and Germinal Center (GC) B cell numbers (CD19+GL7+CD95+) [F (6, 92) = 11,53, p<0.0001] in the spleen after 4, 6 and 8 weeks of co-injections, with the Tfh numbers being 1.3-, 12.2- and 1.65-fold, while GC B cell numbers being 2.9-, 5.3- and 1.1-fold higher, respectively, compared to mice injected with FVIII only. GC B cell numbers correlated with inhibitor titers [r(31)=0.52, p=0.002].

We also investigated dendritic cell (DC) responses after a single injection of FVIII or co-injection with ODN-1826. We found increased numbers of monocyte-derived DCs (moDCs; CD11chighMHCII+CD11b+ CD64+MAR-1+) in the co-injected group at all time points (3, 24, 48, 72 hours and 1 week after a single injection; n=4-5 per group) with the peak number (~10-fold higher compared to naïve mice; p=0.0002) being reached 72 hours and remaining similarly elevated at 1 week. However, levels of activation markers CD86 and MHCII on moDCs in both injected groups were not significantly elevated. The moDC numbers were also ~2.5-fold higher in the FVIII only group than in naïve animals at one week post-administration, but the difference was not statistically significant. Conversely, exposure to FVIII with or without ODN-1826 did not significantly affect the number of CD8α+CD11b- DCs, but this subset showed >2-fold upregulation of both CD86 and MHCII in the co-injected group 3h and 24h after injection (n=5-11, p<0.01). We further found increased numbers of plasmacytoid DCs (pDCs; CD11c+CD11bPDCA+) but the difference was significant only in the co-injected group at d1 and d7 after injection (~2-fold higher than the naïve group; n=4-16 per group, p<0.05). This DC subset also showed upregulated CD86 and MHCII levels ~2-fold in the co-injected group, with difference between CD86 levels showing significance (p=0.0024). We next examined an interferon signature in pDCs by intracellular cytokine staining and found that the cells from co-injected mice had ~2-fold higher levels of IFNβ (p=0.0001, n=5 per group).

We propose that TLR9 stimulation enhances FVIII inhibitor formation in hemophilia A mice through recruitment and/or activation of conventional CD8α+ and plasmacytoid dendritic cells. The narrowing differences in Tfh and GC B cell numbers between the FVIII only and ODN-1826 co-injected groups at week 8 suggest that TLR9 stimulation accelerates GC formation in response to FVIII, which otherwise may reach a similar magnitude, but it takes longer. Also, our results suggest that it may be premature to put the danger theory to rest. The outcome of an immune challenge may vary depending on which innate immune receptor is concurrently stimulated. Notably, all vaccinations tested in the animal study showing no link to inhibitor formation were against pathogens with genomes composed of single-stranded RNA (measles, mumps, rubella and influenza viruses), which does not activate TLR9. Therefore, the impact of vaccination against single-stranded DNA pathogens that can stimulate TLR9, such as VZV, may be worth further investigation.


Herzog:Takeda Pharmaceuticals: Patents & Royalties.

Author notes


Asterisk with author names denotes non-ASH members.