Utility of serum complement factors C3 and C4 as biomarkers during therapeutic management of giant cell arteritis

Objective There is a strong unmet need for biomarkers in giant cell arteritis (GCA), as C-reactive protein (CRP) may be unreliable in patients treated with Tocilizumab (TCZ). We aimed to assess whether C3 and C4 are useful biomarkers in GCA patients, particularly in those treated with TCZ. Method We retrospectively enrolled all patients who underwent C3 and C4 measurement at baseline. All patients were evaluated at 3, 6, 12, and 24 months after diagnosis, as part of routine follow-up. Two assessments after the end of the observational period, in case of further relapses, were also included. Results At baseline, mean ± sd levels (mg/dL) of C3 (133 ± 28.99) and C4 (25.9 ± 9.04) were within normal ranges. During follow-up, C3 and C4 decreased in patients attaining remission (107.07 ± 19.86, p = 0.0006; 19.86 ± 10.27, p = 0.01, respectively) and sustained remission (95.85 ± 18.04, p = 0.001; 15.61 ± 9.75, p = 0.006). In TCZ-treated patients, even stronger decreases in C3 (83.11 ± 19.66, p = 0.001) and C4 (8.26 ± 3.83, p < 0.0001) were observed, and their values were not correlated with CRP or erythrocyte sedimentation rate. Conclusion C3 and C4 do not seem useful in the diagnosis of GCA, as normal values do not rule out active vasculitis. However, C3 and C4 correlate with disease activity. As the low C4 levels found in TCZ-treated patients are not correlated with CRP, C4 should be evaluated as a potential biomarker of disease activity and treatment response.

Giant cell arteritis (GCA) is a large-vessel vasculitis (LVV) involving the aorta and its major branches. Unlike Takayasu arteritis (TAK), GCA affects almost exclusively patients older than 65 years (1) and is strongly related to polymyalgia rheumatica (PMR), an inflammatory condition which may herald the onset of GCA in a considerable percentage of cases (2).
Clinical presentation may be variable: while the involvement of large vessels may result in non-specific symptoms, contributing to a longer diagnostic delay (3), cranial GCA often has a dramatic presentation, with visual loss being the most incisive manifestation. Visual loss, occurring in 15-20% of patients (4), is irreversible if not treated promptly.
GCA classically displays an overall good response to medium to high doses of glucocorticoids (GCs); nevertheless, 40-80% of patients are refractory to the treatment or relapse after GC tapering (5), thus requiring the concomitant administration of GC-sparing agents, also in order to minimize the overall GC toxicity (6).
In this regard, the interleukin-6 (IL-6) receptor inhibitor tocilizumab (TCZ) was shown to improve disease control and also to reduce GC demand (7), and is recommended by the European League Against Rheumatism (EULAR) for use in GCA patients with relapsing disease or in patients who present with comorbidities related to the use of GC or who have a high risk of developing these (8).
Conversely, uncertainty exists about the optimal biomarkers to be used in GCA patients, particularly those treated with IL-6 inhibitors; in this regard, several molecules have been extensively and variously proposed (9,10), but, to date, none of them has provided sufficiently robust evidence to be used in routine clinical practice (11). Therefore, we aimed to assess whether complement factors C3 and C4, representing a crucial target in several autoimmune conditions, may be considered reliable biomarkers in GCA, particularly during follow-up and in those patients on treatment with IL-6 inhibitors.
Thus, the primary aim of the study was to assess C3 and C4 levels in GCA patients in active disease, in remission, and during a relapse, as well as before and during TCZ treatment. Secondary targets were to assess the correlation of C3 and C4 with CRP levels and with imaging findings at follow-up.

Study design and patients
We retrospectively analysed all consecutive patients diagnosed with GCA at the Medius Kliniken Vasculitis Center between January 2009 and April 2021.
Inclusion criteria were a clinical diagnosis of GCA and the availability of a defined set of routine clinical, laboratory, and imaging data in the electronic database. Patients without measurement of C3 and C4 levels at the time of diagnosis (baseline) and patients with other types of LVV such as TAK or isolated aortitis were excluded. For a subgroup analyses, complement levels were analysed in patients before, during, and, if available, after treatment with TCZ.
Data from patients who were enrolled in blinded interventional clinical studies were included only before the administration of the experimental drug and after the end of the trial.

Baseline evaluation
Baseline evaluation (T0) included the minimum core set of parameters for observational studies in GCA according to the 2018 EULAR recommendations (12). In addition, the fulfilment of American College of Rheumatology (ACR) classification criteria (13) and GiACTA trial inclusion criteria for GCA (14), C3 and C4 serum levels, and the results of temporal artery biopsy and imaging [positron emission tomography/ computed tomography (PET/CT) and colour Doppler ultrasonography (CDUS)], expressed dichotomously, were included in the baseline data.

Follow-up visits
Further evaluations at 3 (T1), 6 (T2), 12 (T3), and 24 (T4) months after first diagnosis were included in our retrospective analysis. Data at the time of relapse within the 24 month period were also recorded. In addition, up to two further visits (T5 and T6) in case of relapse after the end of the 24 month observational period were included.
At every follow-up visit, C3, C4, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) serum levels, PET and CDUS findings, disease activity, and therapeutic modifications were recorded.
Imaging findings were expressed on a 0-3 qualitative scale (no activity, improved, unchanged, and worsened), while disease activity was categorized according to EULAR consensus definitions (8). According to the 2018 EULAR definitions, sustained remission was defined as remission for at least 6 months plus achievement of the individual target GC dose (8). GCfree remission was defined as sustained remission plus discontinued GC therapy (other immunosuppressive therapy was allowed) (8).
A relapse was defined as a recurrence of active disease leading to a change in treatment. This included patients with a recurrence of a clinical symptom, including cranial, extracranial, and polymyalgia rheumatica symptoms, and/or an elevation of markers of inflammation attributed to active LVV and/or worsening of imaging findings (e.g. progression of peripheral artery stenosis).
Current immunosuppressive medications, as well as any significant modification of treatment at the time of follow-up visit, were recorded.

C3 and C4 complement measurement
Serum was collected from inpatients and outpatients within routine care and immediately analysed in the onsite immunology laboratory of the vasculitis centre. C3 and C4 levels were measured by quantitative turbimetry using a Siemens Dimension EXL analyser.

Statistical analyses
All analyses were performed using SAS version 11. Normality of data was assessed using the Kolmogorov-Smirnoff test. Statistically significant differences for all comparisons were calculated using the Student's t-test. Correlations between C3 and C4 and other biomarkers such as CRP levels, as well as imaging findings, were calculated with Pearson's test. For all tests, p-values less than 0.05 were considered statistically significant.

Ethics
Ethics approval for this retrospective analysis was obtained by the Ethical Committee of Eberhard-Karls University of Tübingen (protocol number 207/ 2020BO2). The procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation and with the Declaration of Helsinki.

Baseline characteristics
We identified 139 out of 300 (Supplementary file 1) consecutive patients with LVV in whom C3 and C4 levels had been measured in our laboratory at the time of first diagnosis (baseline, T0).
Out of 139 patients with a clinical diagnosis of GCA, 28 were classified as having cranial GCA, 61 large-vessel (LV)-GCA, and 50 cranial plus LV-GCA. The baseline features are summarized in Table 1.
Laboratory findings displayed elevations of both CRP and ESR and a decrease in serum haemoglobin. Conversely, both C3 and C4 were within normal ranges.

C3 and C4 levels in active GCA versus remission
During our 2 year observational period, 27 patients attained remission and underwent complement measurements at all time-points. Out of these 27 patients, seven and four, respectively, were considered to be in sustained remission for at least 6 months, or in GC-free remission.
All patients received high-dose GC induction therapy with subsequent taper. Nine of them were also treated with TCZ, 12 with methotrexate, and six with other conventional immunosuppressants.
Both C3 and C4 values decreased significantly between baseline (T0) and the first follow-up evaluation at remission (Table 2). C4 levels showed a stronger predictability for the state of disease compared to C3 using a binary multivariate regression model (C4: β-coefficient 0.71, p = 0.005; C3: βcoefficient 0.009, p = 0.141).
Patients were then further stratified according to the different subtypes of remission (Table 3). There was a statistically non-significant trend towards lower C3 and C4 serum levels in patients with sustained remission and GC-free remission compared to patients who had a shorter duration of remission and/or needed higher GC doses (Table 3).
In patients in remission, CRP levels correlated strongly with C3 and C4 levels (Supplementary file 2). A correlation between CRP and C3 levels was also found in active disease patients at baseline, while C4 levels were not associated with CRP levels at baseline or at relapse. In a binary regression model, CRP and ESR showed significantly stronger associations with the state of disease than complement factors C3 and C4.

C3 and C4 levels in GCA at relapse
During the course of our observational period, 23 of the 139 GCA patients relapsed, and C3 and C4 levels at the time of relapse were available in 13 of these patients. Relapse occurred 3 and 6 months after diagnosis in three patients and one patient, respectively, in four patients after 1 year, in three after 2 years, and in the other four after a mean ± sd of 41.5 ± 7.1 months. All relapses were classified as minor relapses and all of them were treated with lowdose oral GCs and/or conventional immunosuppressants at the time of relapse, but none with anti-IL-6 agents.
A statistically significant difference between remission and relapse findings was found for CRP, ESR, and C3, but not for C4 serum levels (Table 4).  In four subjects, PET/CT showed no vascular inflammation, while in nine, vascular FDG uptake was decreased compared to baseline. Imaging was unchanged in two patients and worsened in another two patients.
At the same time, no statistically significant differences between group means, as determined by one-way analysis of variance (ANOVA), were found for C3, C4, ESR, or CRP.

C3 and C4 levels according to different CDUS findings at follow-up
CDUS of the temporal arteries and/or axillary arteries was performed in 97 patients at baseline and in 71 of them the findings were consistent with a diagnosis of GCA. During follow-up, repeat CDUS was performed in nine patients, and complement levels were obtained at the same time-point. In six of these nine patients, CDUS was normal, while in one patient and two patients CDUS findings were unchanged and worsened, respectively.
Pairwise comparison with post-hoc Tukey's test revealed no difference in C3 and C4 levels between patients with normal and those with worsened CDUS findings.

C3 and C4 levels in GCA patients treated with TCZ
In nine of our GCA patients treated with TCZ, complement serum levels were measured before and during treatment. Before the administration of TCZ, C3 and C4 levels were normal while ESR and CRP were elevated. At the first visit after TCZ start, all patients were considered in clinical remission and displayed normal ESR and CRP levels ( Table 4). Serum levels of C3 and particularly C4 were statistically significantly decreased compared to baseline (Table 5, Figure 1). No significant correlations were found between C3 and CRP (r = −0.04, p < 0.9) or between C4 and CRP (r = 0.4, p = 0.28) (Supplementary file 3). In the two patients who tapered or discontinued TCZ, C3 and C4 increased, almost within the normal range (105 and 7.7 mg/dL and 91 and 17.9 mg/dL, respectively).
Finally, a statistically significant difference was assessed for C3 and C4 serum levels between patients who attained remission with and without TCZ (Supplementary file 4).

Discussion
In our study, which represents the first large cohort of GCA patients undergoing complement evaluation, we were able to significantly link C3 and C4 levels to the state of disease at baseline, remission, and relapse, even Data are shown as mean ± sd. CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; GC, glucocorticoid. †Non-significant trend (p < 0.1); *significant (p < 0.05) vs remission of < 6 months' duration. though this link was much weaker compared to CRP and ESR. C3 and C4 serum levels were normal in patients with active newly diagnosed disease. Remarkably, only seven out of 139 GCA subjects displayed C3 values above the upper limit, while in all of them C4 was within the normal range. Therefore, serum C3 and C4 levels cannot be used to support a diagnosis of GCA.
A study from 1986 reported that sera from patients with active GCA caused deposition of C3 and C4 in an indirect immunofluorescence test on rat kidney medullary structures (15). This reaction was not observed with sera from GCA and PMR patients in remission and healthy controls (15). Since then, the role of complement as a potential biomarker in GCA has not been studied, while only one recently published study found increased serum levels of C3 in patients affected by active TAK (16).
In GCA patients treated with TCZ, C3 and particularly C4 levels were even lower than in patients in remission who were on treatment with GCs and other immunosuppressants. This represents the first evidence of reduced complement serum levels in a cohort of GCA patients treated with biological agents, as no reallife study no clinical trial (7,17,18) has assessed C3 and C4 serum levels. Nevertheless, in line with this observation, reduced serum complement levels have also been reported following treatment with TCZ in a cohort of patients with rheumatoid arthritis (19). The strict correlation between hypocomplementemia and drug-mediated IL-6 inhibition, and not only with the mere disease activity, was confirmed when we compared C3 and C4 serum levels of TCZ-treated patients with those who had been treated with conventional immunosuppressive drugs. The pharmacological mechanisms whereby TCZ reduces C3 and C4 serum levels are unknown. We may hypothesize that they are related to the inhibition of IL-6 (20) and CRPmediated (21) complement activation or to the inhibition of CRP-mediated C3b inactivation (22,23), thus leading to aberrant C3 and C4 consumption; nevertheless, the normal levels of circulating immunocomplexes assessed in RA patients treated with TCZ seem to rule out the latter hypothesis (20). Conversely, a direct link between complement activation and IL-6 release has already been described (24), thus suggesting that a more complex cross-talk exists between them.
It is worth mentioning that the reduction in C3 and C4 below normal values in patients treated with TCZ did not lead to significant adverse events or to severe fungal infections, although C3 represents an essential element of resistance against Aspergillus fumigatus (25).
In patients in remission, CRP levels correlated strongly with C3 and C4 levels, reflecting the crosstalk existing between acute-phase reactants and complement: CRP activates the classical complement pathway (21) and, on the other hand, prevents an aberrant immune response through the inactivation of C3b to iC3b (22,23) and the blockage of the alternative pathway (26).
We found that this strong correlation between complement levels and CRP was completely absent in GCA patients treated with TCZ. The strong decrease, particularly in C4 levels, and the lack of association with CRP levels may render complement a potential candidate biomarker for TCZ-treated patients. Since our retrospective study did not include patients relapsing on TCZ, further studies are required to study the potential role of C4 as a serum biomarker of GCA activity in patients treated with TCZ, where CRP and ESR were found to be unreliable (17). Nevertheless, preliminary data from a small cohort (27) showed that C4 may increase in refractory or relapsing GCA patients treated with TCZ; this may suggest that the reduction in C4 is strictly related to disease activity and not only to the pharmacological inhibition of IL-6, confirming the potential reliability of C4 in GCA patients treated with TCZ.
No significant association was found between C3 and C4 and imaging findings of active GCA in patients in clinical remission. However, it is unclear whether residual findings on imaging represent true inflammatory activity. Several studies, focusing on imaging procedures during follow-up, have shown persistent magnetic resonance angiography abnormalities (28) or FDG uptake (29) in GCA patients considered to be in clinical and serological remission. Therefore, the correlation between subclinical imaging findings and serum biomarkers is still challenging.
The main limitation of this study lies in its retrospective design and real-life nature: since complement measurement is not part of the routine follow-up laboratory evaluation of patients with GCA, C3 and C4 levels were available in only 27 out of the 139 patients with complement data at baseline. Since no patient in our cohort relapsed during treatment with TCZ, no valid conclusion regarding the role of complement levels as a biomarker of GCA can be drawn at this stage. However, the observation of a significant decrease in C4 serum levels that is not correlated with CRP levels in TCZ-treated GCA patients suggests that C3 and C4 complement levels should be considered in future prospective studies investigating novel biomarkers for GCA.

Conclusion
In summary, the results of our study suggest that serum levels of C3 and C4 are significantly associated with disease activity in GCA. Importantly, we observed that the low C4 complement levels found in TCZ-treated GCA patients are not correlated with CRP levels. More research is needed to validate serum complement factors C3 and C4 as biomarkers for disease activity and treatment response in patients with GCA treated with TCZ.