The significance of peripheral blood monocytes is increasingly recognized in the pathogenesis of inflammatory joint diseases, such as rheumatoid arthritis (RA) and ankylosing spondylitis (AS) [1
]. Monocytes are divided into the predominant classical CD14hi
CD16− population and CD16+ monocytes, which branch into intermediate CD14hi
CD16+ and nonclassical CD14dim
CD16+ subsets [4
]. Treatment of arthritis aims to reduce tissue infiltrating monocytes, which are osteoclast precursors and sources of inflammatory cytokines [5
]. CD16+ cells are increasingly present in both blood and synovium of RA patients, where they may express autoantigens correlating with the severity of joint damage [6
]. Puchner et al. recently demonstrated in murine models that nonclassical monocytes are crucial in the development of arthritis, although literature also attributes a role for classical subsets in osteoclast differentiation and joint damage [7
]. Monocyte-derived TNF-α may be a signaling molecule involved in this process and is a hallmark of inflammation [9
]. When encountering immune complexes, nonclassical monocytes are selective producers of TNF-α and IL-1β [10
] which have been investigated in numerous clinical trials as an effective method to reduce inflammatory disease activity [11
]. CD16+ monocytes may fail to cross endothelial barriers at the onset of inflammation, while differentiating in peripheral blood and tissues due to perturbed cytokine production from classical CD14hi
CD16− subsets [13
]. Therefore, monocyte adhesion and activation markers are other emerging fields of interest. CD11b is involved in migration and adhesion. Polymorphisms of CD11b are associated with systemic lupus erythematosus and other immune complex mediated diseases [14
]. Murine studies of targeted anti-CD11b antibodies demonstrate a suppressive effect on processes shaping arthritis [5
]. To the best of our knowledge, peripheral blood monocyte heterogeneity has not been extensively studied in patients with RA or AS receiving anti-TNF-α treatment. Therefore, our goal was to evaluate the effect of anti-TNF-α treatment on peripheral blood monocyte subpopulations, their functional properties, and to compare the effect of treatment in high disease activity, and in conventional treatment-refractory AS and RA patients.
The main finding of the present study is that in RA and AS patients, a decline in classical (CD14hi
CD16−) monocytes was observed during anti-TNF-α treatment in comparison to placebo, while the amount of CD14dim
CD16+ monocytes increased. This increase was associated with lower integrin expression on the surface of monocytes. A decline in the number of classical monocytes was previously reported in a small unrandomized study of 10 (5 RA, 5 AS) patients treated with infliximab, where a rapid decrease in classical monocyte subsets was reported [15
]. Aeberli et al. suggested that TNFi treatment might restrict monocyte recruitment into inflamed tissues and TNFi-induced apoptosis in peripheral blood monocytes, which was also reported earlier [15
]. Our observation that TNFi treatment increased circulating nonclassical (CD14dim
CD16+) monocyte fractions to a greater extent than conventional treatment is also in line with the change seen in response to infliximab, although this phenomenon was unclear in AS patients (p
= 0.045 at study endpoint) and mostly observed in RA patients [15
]. This could lead to an interpretation that the TNFi effect on monocyte populations is closely determined by disease character. In our study, however, we provide evidence that the TNFi-mediated effect is present in both RA and AS. Moreover, we expand these observations to several TNF-α-inhibiting agents including etanercept, a soluble TNF-α receptor, and monoclonal antibodies other than infliximab. In response to TNF-α, murine models showed an increase in Ly6Chigh
monocytes, corresponding to human classical CD14hi
CD16− monocytes, and a lesser increase in Ly6Clow
subsets, similar to human nonclassical monocytes [17
]. Subsequent treatment with adalimumab reduced circulating Ly6Chigh
but not Ly6Clow
, which is consistent with our findings.
Kawanaka et al. previously reported that elevated CD16 expressing monocyte subsets are characteristic for RA and the active disease state [18
]. The authors suggested that overflow of cytokines from the inflamed joint promotes the differentiation of CD16+ monocytes. This subpopulation may further exacerbate the disease with increased chemokine receptor expression, subsequent infiltration, and pro-inflammatory cytokine production. When comparing RA patients to healthy controls, intermediate CD14hi
CD16+ monocyte fractions are elevated, while nonclassical CD14dim
CD16+ fractions are decreased [19
]. In the present study, the TNFi-mediated shift towards nonclassical subsets and control of the intermediate fraction may resemble a change towards a “healthy” subset distribution. In AS, classical CD14hi
CD16− monocyte fractions are increased and nonclassical CD14dim
CD16+ are decreased when compared to healthy controls [20
]. Similarly, the TNFi-induced rise in nonclassical subsets may result in regression from the monocyte “disease” distribution. The significance of this effect remains to be established.
While both diseases differ substantially in clinical features and blood monocyte heterogeneity, we demonstrated that in comparison to placebo, effective treatment with TNFi reduces disease activity and is paralleled by a shift in peripheral blood monocytes. In line with our findings, AS patients treated with anti-TNF-α agents showed a decline in circulating classical macrophages (M1) and a shift towards nonclassical macrophage (M2) subsets, corresponding with improvement in BASDAI index [21
]. The M2/M1 shift positively correlated with joint destruction and negatively with inflammatory activity. This may explain why, even with effective control of disease activity by TNFi, erosive joint changes can still progress in some patients. This outcome is of high clinical importance. We hypothesize that the TNFi-induced shift toward nonclassical monocytes is sufficient to prevent elevation in the “inflammatory” intermediate phenotype and therefore reduces cytokine burden. However, the rise in the nonclassical subset, which has been demonstrated to be a mediator of tissue damage and correlates with joint destruction [8
], can be seen as detrimental and counterintuitive. It should be noted that the “inflammatory” subpopulations of nonclassical and intermediate monocytes are observed to transmigrate less efficiently than classical subsets [13
], but may lead to tissue and organ dysfunction [23
]. We provide evidence that TNF-α-inhibiting agents may further reduce this capability by decreasing the expression of adhesive molecules, such as CD11b. Reduced monocyte tissue infiltration via TNFi-mediated effects on endothelium was previously proposed [15
]. Enhanced CD11b expression occurs in response to chemotactic factors [24
], while CD11b positive monocytes are able to differentiate into osteoclasts under TNF-α and Il-17 [25
], at the same time as being inhibited by infliximab. Taken together, a significant decrease in CD11b expression in response to TNFi in all monocyte subpopulations, most prominent in nonclassical subsets (p
< 0.001), indicates a reduced ability of these “disease-driving” cells to adhere to the endothelium and transmigrate. Indeed, antibodies targeting CD11b inhibit inflammatory cell recruitment, synovial infiltration, and suppress arthritis development in murine models [5
]. The difficulty of nonclassical monocytes in crossing endothelial barriers may be amplified by TNFi-mediated effects, including CD11b downregulation. Finally, the increase in the “tissue-destructive” reservoir of nonclassical CD14dim
CD16+ monocytes is sequestered in peripheral blood and limited from entering inflamed joints. This hypothesis is still preliminary and requires further study.
It was previously reported that in active AS (BASDAI ≥ 4), corresponding to our study subjects, greater CD11b expression was observed in patients with high disease activity compared to those with low disease activity [20
]. Decreased expression of CD11b on monocyte subsets may stem from effective reduction of disease activity due to TNFi treatment, paralleled by a reduction in activation markers characteristic of active disease. CD11b is described as crucial in myeloid lineage differentiation, with TNF-α inhibition decreasing its expression on myeloid cells [26
]. However, in RA, expression of CD11b is higher than in healthy controls and decreases after glucocorticoid treatment, which suggests that CD11b is an indicator of disease activity, rather than selectively tied to TNF-α inhibition [27
]. We observed increased expression of CD45RA after TNFi treatment in our study. CD45RA, a monocyte activation marker, shows higher expression in healthy patients when compared to RA, particularly for intermediate subsets [19
]. Response to adalimumab monotherapy has been associated with monocyte CD11c expression; future TNFi responders had higher basal CD11c expression than control and non-responders [28
]. A shift towards the “healthy expression pattern” of monocyte markers was seen in clinical responders but was absent in non-responders. We observed a decrease in CD11c expression exclusively on CD14dim
CD16+ cells after TNFi treatment.
It has been suggested that the heterogeneity of inter-study findings may relate to diversity of treatment profiles (e.g., glucocorticoid use leading to decreased CD16+ expressing monocytes) [29
]. Glucocorticoids were also shown to induce a phenotype change from classical CD14hi
CD16− to CD16+ [30
]. Whether MTX-naïve or MTX-treated patients differ with respect to monocyte distribution heterogeneity also remains unclear [31
], though this may depend on whether the patient is MTX-responsive or MTX-refractory. To eliminate confounding factors in the present study, concomitant medication use, including glucocorticoids and MTX, did not differ between the TNFi and placebo groups and dosage was unaltered from baseline.
The limitations of the study have to be addressed. Although the baseline monocyte subpopulations did not differ between the TNFi treatment and placebo group, a baseline discrepancy between activation markers was observed. Entirely homogenous populations were likely undifferentiated due to the small size of the placebo group (n = 12), limitations due to high disease activity, and other strict inclusion criteria. We approached this limitation by investigating three consecutive time points using repeated-measures ANOVA.
In summary, we observed a shift towards nonclassical CD14dimCD16+ monocytes in peripheral blood due to TNFi treatment, and this was seen in both AS and RA patients. To the best of our knowledge, this is the first placebo-controlled study investigating the effects of TNFi on monocyte subsets in AS and RA subjects. Our study contributes to a better understanding of the complex effects of TNFi in humans and could lead to the development of possible cellular biomarkers for TNFi monitoring.