Leukocyte Dynamics Reveal a Persistent Myeloid Dominance in Giant Cell Arteritis and Polymyalgia Rheumatica

Abstract

Giant cell arteritis (GCA) and polymyalgia rheumatica (PMR) are inflammatory diseases requiring long-term glucocorticoid treatment. Limited data on dynamics in leukocyte counts before, during and after treatment are available. Leukocyte counts were measured, as cellular markers of inflammation, at fixed time points in our prospectively studied cohort of pre-treatment glucocorticoid-naive GCA (N = 42) and PMR (N = 31) patients. Values were compared with age-matched healthy controls (HCs; N = 51) and infection controls (N = 16). We report that before start of treatment monocyte and neutrophil counts were higher in GCA and PMR patients than in HCs, while NK- and B-cell counts were lower. C-reactive protein (CRP) levels correlated positively with monocyte counts in GCA, and negatively with B-cell and NK-cell counts in PMR. During glucocorticoid treatment, myeloid subsets remained elevated whereas lymphoid subsets tended to fluctuate. Interestingly, erythrocyte sedimentation rate (ESR) outperformed CRP as marker for relapses in GCA. We defined stable treatment-free remission groups in both GCA and PMR. GCA patients in treatment-free remission still demonstrated elevated monocytes, neutrophils, ESR, and platelets. PMR patients in treatment-free remission had normalized levels of inflammation markers, but did have elevated monocytes, lowered CD8+ T-cell counts and lowered NK-cell counts. Finally, we showed that low hemoglobin level was predictive for long-term GC treatment in PMR. Overall, leukocyte composition shifts toward the myeloid lineage in GCA and PMR. This myeloid profile, likely induced by effects of inflammation on hematopoietic stem cell differentiation, persisted during glucocorticoid treatment. Surprisingly, the myeloid profile was retained in treatment-free remission, which may reflect ongoing subclinical inflammation.

Keywords: giant cell arteritis; glucocorticoids; longitudinal cohort study; polymyalgia rheumatica; relapses; treatment-free remission; vasculitis.

Figure 1

Pre-treatment measurements in newly-diagnosed, treatment naive GCA and PMR patients. (A), Leukocyte counts in the blood for GCA and PMR as well as two control groups: HC and INF. The n is depicted in the figure and indicates the number of samples measured in the different groups. Data is expressed as median and interquartile range. Statistical differences by Mann Whitney U between groups are displayed if Kruskal Wallis testing indicated significant differences: *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001. (B), Stacked leukocyte subset counts show a clear shift to the myeloid lineage in GCA and PMR pre-treatment.

Figure 2

Correlations between three leukocyte subsets and the inflammatory marker CRP. Correlations between CRP and the leukocyte subset in pre-treatment GCA (closed circles) and PMR (open circles) patients. Spearman's R, the p-value of the correlation and the N are indicated in each graph for GCA and PMR. Regression line for GCA is shown as an uninterrupted line, for PMR as a dotted line. Correlations for neutrophils, CD4+ T-cells, and CD8+ T-cells are displayed in Supplementary Figure 2.

Figure 3

Smoothed median of leukocyte counts for GCA and PMR patients over time while on GC treatment. The smoothed median is calculated by taking the median of each new measurement and that of the 19 measurements before that point. This method enables to distinguish patterns over time that would be unnoticeable if each point is plotted separately. For interpretation the interquartile range of HC (green box, cross-sectional measurement) and the median of the INF (dotted line, cross-sectional) were added to the figures. Time point 0 indicates the pre-treatment sample. Also, the three different treatment phases are indicated.

Figure 4

Dynamics in cell population counts during follow-up in GCA (A) and PMR (B) patients. Counts are expressed in radar plots as median fold-change compared to healthy controls (n = 51) for the following groups: pre-treatment (GCA n = 42, PMR n = 31), treatment phase I (GCA n = 38, 69 measurements; PMR n = 25, 54 measurements), phase II (GCA n = 32, 43 measurements; PMR n = 23, 33 measurements), and phase III (GCA n = 29, 65 measurements; PMR n = 19, 56 measurements). Pre-treatment only includes the visit before start of treatment; treatment phase I includes follow-up visits at 2, 6 weeks, and 3 months; treatment phase II includes 6 and 9 months; treatment phase III includes 12 months and beyond. †: sign difference between HC and baseline, ×: sign difference between HC and treatment phase I, #: significant differences between HC and phase II, and ¤: significant differences between HC and phase III (Mann-Whitney U-test, P < 0.05).

Figure 5

Levels of inflammatory markers during treatment phase I, II, and III for GCA and PMR patients in remission and during relapse. The definition of remission and relapse was based solely on clinical signs and symptoms. Data is expressed as median plus interquartile range. The number of measurements is indicated by n. Statistical significance is expressed as follows: *p < 0.05, **p < 0.01, ****p < 0.0001 (Mann Whitney U-test).

Figure 6

Different leukocyte subset counts and inflammatory markers in treatment- free remission. Leukocyte subset counts pre-treatment and in treatment-free remission (A: GCA n = 13 patients, 17 samples and B: PMR n = 15 patients, 25 samples) were expressed as median fold-change compared to healthy controls. †: sign difference between HC and treatment-free remission. ×: sign difference between pre-treatment and treatment-free remission (Mann-Whitney U-test p < 0.05). (C) Inflammatory markers in HC, GCA treatment-free remission, and PMR treatment-free remission (Mann-Whitney U-test: ***p < 0.001,****p < 0.0001). Data is expressed as median and interquartile range. (D) Correlation between B-cell counts and ESR in treatment-free remission patients.

Figure 7

In PMR patients, long-term GC requirement is predicted by pre-treatment Hb levels, but not by CRP and ESR. The CRP, ESR, and Hb of PMR patients before treatment were split into low or high levels (based on the median) and were plotted in a Kaplan-Meier curve against time to GC-free remission. p-value and hazard ratio (HR; including 95% confidence interval) of the log rank test are depicted in the graphs.