Introduction

Stargardt disease (STGD1) is a Mendelian inherited disease in which vision of both eyes progressively deteriorates, a process that is irreversible to date [1,2]. STGD1 is caused by mutations in the adenosine triphosphate binding cassette transporter A4 (ABCA4) gene. This gene encodes a transmembrane protein which eliminates retinal from photoreceptor disks and endolysosomes in the retinal pigment epithelium (RPE) [3,4]. In the absence of a functional ABCA4 protein, metabolic byproducts of the visual cycle react to form bisretinoids, the most important of which is N-retinylidene-N-retinylethanolamine (A2E). A2E accumulates in the RPE as a component of lipofuscin, ultimately leading to irreversible degeneration of both RPE and photoreceptors [5–7].

Clinically, there is a substantial heterogeneity in STGD1 phenotypes, age at onset and disease progression [1,2,8]. Even amongst siblings carrying the same ABCA4 variants, functional outcomes differ considerably [9]. This leads to the question whether other mechanisms, besides the ABCA4 genotype, are involved in STGD1 pathogenesis. Recent preclinical studies highlight a role for the complement system in the pathophysiology of STGD1. The complement system is part of the innate immune system and protects the human body from pathogens. In cultured RPE cells, photo-oxidation of A2E was shown to activate the complement system and alter the transcription of genes involved in stress response, apoptosis and immune response [10–14]. Moreover, RPE cells of abca4-/- mice showed higher levels of complement activation products and increased inflammatory markers as compared to wildtype mice [13,15,16]. Increased levels of C3d, the degradation product of complement component C3, were detected in serum of Abca4-/- mice concomitantly to the increase of systemic C3d levels, indicating a relationship between systemic and local complement activation [16]. When treating Abca4-/- mice with a targeted gene therapy that increases expression of a complement negative regulatory protein (complement receptor 1-like protein, Crry) in the eye, bisretinoid accumulation and photoreceptor degeneration were reduced [17]. Equally important, the first direct evidence of complement dysregulation in RPE of STGD1 donor eyes was reported recently by immunohistochemical analysis [18]. Taken together, several preclinical studies suggest that complement dysregulation is a contributing factor to photoreceptor degeneration in STGD1.

In age-related macular degeneration (AMD), a disease with similar features as STGD1, the role of the complement system in the pathophysiology is well-established [19]. In contrast to STGD1, AMD is a multifactorial disease including both genetic and behavioural risk factors [20]. The genetic risk for AMD is co-determined by variants in genes encoding components of the complement system (C3, C9, CFB, CFH, and CFI) [21,22]. In AMD patients, activation of the complement system was not only detected within the eye [23,24], but also systemically in serum and plasma [15,25,26]. The processing of complement factor 3 (C3), measured by the ratio of C3 and its degradation product C3d (C3d/C3), has repeatedly been associated with AMD [27]. Moreover, C3d/C3 levels have been correlated to AMD disease stage [26].

In contrast to extensive research in AMD patients, there is little published data on the contribution of complement dysregulation to the pathogenesis of STGD1. Nevertheless, the results of the described preclinical studies have already lead to the initiation of a phase 2b clinical trial with a pharmacological agent that targets the complement system by an intravitreal injection with a complement factor C5 inhibitor (avacincaptad pegol/Zimura, NCT03364153). Further understanding of the role of the complement system in STGD1 may shed light on whether complement activation is a local disease process or whether it extends to the systemic circulation, and may contribute to the design of therapeutic approaches for STGD1. Therefore, the current study aimed to assess systemic activation of the complement system in STGD1 patients and investigates whether complement levels are correlated with STGD1 severity.

Methods

Results

The C3 and C3d levels, and the C3d/C3 ratio of 80 patients, 44 females and 36 males, with a mean age of 46±18 years (range 17–84), were compared to 80 age- and sex-matched controls. There was indeed no difference in age (p = 0.804, t-test) or sex (p = 1.00, chi-square test) between both groups. The mean levels of C3d and C3, and the C3d/C3 ratio were similar in both groups (Table 1 and Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Complement activation levels in Stargardt disease (STGD1) patients and controls matched for age and sex.

Level of (A) C3d, (B) C3, (C) C3d/C3 ratio in STGD1 patients and controls. Dotted lines represent the reference values. Red lines represents the mean. There were no significant differences in means between patients and controls as tested with an independent samples t-test. C3, complement component 3; C3d, complement degradation product C3d.

https://doi.org/10.1371/journal.pone.0253716.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Complement activation levels in Stargardt disease patients and controls matched for age and sex.

https://doi.org/10.1371/journal.pone.0253716.t001

Six patients (7.5%) had complement activation levels outside the reference window. Four of these patients had abnormal C3 levels (C3 > 1500mg/l) and two patients had abnormal C3d levels (C3d > 8.3mg/l). Both patients with a high C3d level also had a high C3d/C3 ratio. In addition, there was one patient with an elevated C3d/C3 ratio, with normal C3d and C3 levels. Only one individual in the control group (1.3%) had a marginally abnormal complement activation level of C3, namely a C3 of 1501mg/l (reference < 1500mg/l). None of the controls had an abnormal C3d/C3 ratio. With the C3d/C3 ratio as a parameter for complement activation, these findings point towards systemic activation of the complement system in three STGD1 patients and none of the controls.

Since systemic activation of the complement system can be caused by inflammatory conditions, we evaluated the questionnaire data in the participants of the study and specifically in the individuals with altered complement activity. Neither patients, nor controls were on any immunosuppressive medication. Overall, none of the participants reported having had fever in the week before the blood draw. There were 2 patients and 2 controls who reported flu-like symptoms (sore throat, coughing, or running nose). In addition, one STGD1 patient was treated for a urinary tract infection and another STGD1 patient was being treated for breast cancer in the week before the blood draw. Indeed, ten patients and only three controls had a medical history of a chronic inflammatory disease. These conditions included a rheumatic disorder (5 patients and 1 control), bronchitis (4 patients and 2 controls), and ulcerative colitis (1 patient). Of the five STGD1 patients who suffered from a rheumatic disorder, two had active arthritis during blood draw and both of these patients had an elevated C3d/C3 ratio. The third patient with a high C3d/C3 level was treated for a urinary tract infection at the moment of blood draw for this study. To conclude, all three STGD1 patients with an elevated C3d/C3 ratio had a concurrent inflammatory condition during the blood draw.

We next evaluated whether complement activation levels were associated with disease progression and with demographic and behavioral variables. Within the STGD1 cohort, complement levels were not associated with age at onset, VA, DDAF area, VA decline, or DDAF growth as measures of disease progression. The correlation coefficients and corresponding p-values are provided in Table 2 and correlation plots are provided in S2 Fig. C3 levels were associated with sex and BMI in STGD1 patients. Female patients had a higher C3 ratio compared to males (p = 0.001, independent samples t-test). Patients with higher BMI, had higher C3 levels (p<0.001, Spearman’s Correlation). Age, smoking status and the number of packyears were not associated with complement activation levels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Correlation analysis of all parameters within Stargardt disease patients.

https://doi.org/10.1371/journal.pone.0253716.t002

Discussion

Previous preclinical studies indicate a role for the complement system in the pathophysiology of STGD1 [10–17,36,37]. The results of this study did not detect a significant increase or decrease in systemic C3d/C3 ratio in STGD1 patients. Moreover, systemic complement levels were not associated with disease parameters or the rate of disease progression. In 3 of the in total 80 STGD1 patients that did show an increased C3d/C3 ratio a concurrent inflammatory condition most likely explains complement consumption. As we are dealing with a rare disease, we did not exclude participants with symptoms suggestive of underlying inflammation on beforehand in order to give a complete overview of our STGD1 cohort. However, this can be a limitation as it might complicate the understanding of the relationship between complement activation and Stargardt disease. When omitting the participants with flu-like symptoms, medial complaints during blood draw and a medical history of chronic inflammatory disease (thus excluding 12 patients and 5 controls) the mean C3d/C3 ratio was 3.28±0.96 in cases and 3.57±1.06 in controls. Independent samples t-test does not show a significant difference between these means (p = 0.077), which confirms that systemic complement levels are indeed not elevated in STGD1 patients. Therefore, the current study does not provide support for systemic complement activation in the pathophysiology of STGD1.

Previous studies in AMD did find elevated complement levels in patients compared to controls [15,25,26,38]. Moreover, complement levels were associated with disease stage, where systemic complement levels increased with consecutive AMD disease stages [26]. AMD patients with geographic atrophy tended to have relatively higher complement activation levels [15,26,38]. However, in the current study complement levels were not elevated in STGD1 patients, who also present with atrophy of the RPE. Yet, previous preclinical studies do provide evidence for a common role for the complement system, leading to RPE loss in both STGD1 and AMD [13,39]. In both diseases, the accumulation of vitamin A byproducts in RPE cells initiates a vicious cycle where chronic low-grade inflammation leads to increased complement activation which, in turn, leads to increased damage to the RPE [13,17,26,36,37]. Once RPE damage occurs, complement factors are exchanged between the blood and retina which contributes to worsening of the disease [16,38]. It is possible that this last step is more likely to occur in AMD patients as compared to STGD1 patients due to contribution of systemic complement levels, which are largely driven by genetic variants in the complement system. In addition, it was previously hypothesized that RPE atrophy acts as a local stimulus that further amplifies the complement response in AMD [40]. However, we did not find an association between systemic complement levels in STGD1 and RPE atrophy.

Based on mean C3d/C3 ratios in STGD1 cases and controls, and the population SD of 1.21, the current study had an effect size of 8% (95% CI, -5% - 70%). In order to be able to demonstrate a difference in a subsequent study, 2325 participants are needed per cohort. The current study had sufficient power to detect effect sizes that were previously reported in AMD patients (effect size of 42%) [26]. We can thus infer that the effect size of C3d/C3 ratio in STGD1 patients is lower in comparison to AMD patients, or that an effect of C3d/C3 is absent in STGD1. This is likely explained by the difference in genetic predisposition between AMD and STGD1 patients. Genetic variants in the complement genes, most importantly the CFH gene, are associated with both increased AMD risk and systemic complement activation [26,40,41]. We did not analyze genetic variants in the complement system in STGD1 patients in the current study, but previous studies did not find an association between the CFH genotype and STGD1 and controls [42]. Since we do not detect systemic elevation of C3d/C3 in STGD1 patients, genetic variants in the complement system likely do not play a major role in STGD1, or only contribute to disease in a selected group of STGD1 patients.

To date, very little is known about the relationship between systemic complement activation and local complement activity in the eye. Serum contains the vast majority of complement factors, which are produced by the liver. However, the retina synthesizes several complement proteins locally [43]. Increased complement activity detected in STGD1 in preclinical studies can thus be of local origin, and therefore not measurable in serum. Recently, investigators examined complement activation in the context of RPE from donor eyes clinically diagnosed with STGD1 [18]. However, only one eye was genetically classified as STGD1. For the second donor eye, extracted DNA was of insufficient quality. The third donor eye harbors a PRPH2 mutation without a ABCA4 mutation and should therefore be diagnosed as pseudo-Stargardt instead of STGD1 [44–46]. Further studies on intra-ocular complement levels are needed to shed further light on the role of complement in the retina.

In AMD patients, studies on aqueous and vitreous humour have established elevated complement components in vivo within the eyes of patients [24,47]. Such studies have not been carried out in STGD1 patients yet, therefore, collecting aqueous and vitreous humor of STGD1 patients would be relevant for future studies. Nevertheless, sample collection is expected to be challenging since only very few STGD1 patients undergo vitrectomy. Alternatively, the ongoing phase 2b clinical trial with complement C5 inhibitor Avacincaptad Pegol (Zimura) [NCT03364153] might prove that local complement dysregulation is a causal factor in the pathogenesis of STGD1, if this therapy successfully rescues the phenotype.

Within the 80 STGD1 patients in this study, no association was detected between systemic complement levels and VA, DDAF area or age at onset. Therefore, systemic complement activation does not explain the clinical heterogeneity within our STGD1 cohort. Prospective studies in which complement levels are measured at several time points during the disease course are necessary to establish the effect of complement levels on disease progression prospectively.

Our study has several limitations. First, we present a cohort that was sufficiently sized to detect effect sizes as large as those previously detected in AMD patients with GA. However, this cohort is too small to detect lower effect sizes. Larger studies are needed to exclude the role of systemic complement activation in STGD1. Second, the inclusion of patients with a possible underlying inflammatory condition creates outliers which causes an extreme data bias. However, when omitting these patients, our results remained the same. Third, complement levels are known to be influenced by other factors including age, sex, BMI and smoking [15,33]. Even though cases and controls were matched on the basis of age and sex, there may be residual confounding and unfortunately, we did not collect BMI and smoking variables for our control cohort. Therefore, it is not possible to control for these covariates when analyzing our data. This study is the first to investigate systemic complement activation in STGD1 and its potential association with disease progression, and should therefore be considered as a pilot study. Ideally, complement activity should be monitored in individual patients as their disease progresses to establish the effect of the covariates on complement levels and to establish the effect of complement levels on disease progression.

In conclusion, this is the first study to analyze systemic complement activation in STGD1 patients. We did not find evidence that systemic complement activation is associated with STGD1. If the complement system contributes to STGD1 pathogenesis, it might only be locally involved as a response to local processes, or it might be involved on systemic level only in a selected patient group. Since complement inhibitors are being evaluated as a potential therapeutic strategy for STGD1, a better understanding of the role of the complement system in STGD1 patients, by measuring the complement levels within the eye and determining their effect on disease progression, needs to be established. Once we understand the role of the complement system in STGD1 pathogenesis, we can optimize patient selection and intervention timing to maximize the therapeutic effect of complement inhibitors.

Supporting information

Acknowledgments

The authors would like to thank Johannes Groenewoud for assisting in statistical analysis.