Systemic complement activation levels in Stargardt disease.

PURPOSE: Preclinical research provides evidence for the complement system as a potential common pathway in Stargardt disease (STGD1) and age-related macular degeneration (AMD) leading to retinal pigment epithelium (RPE) loss. However, systemic complement activation has not yet been assessed in STGD1 patients. We conducted a cross-sectional case-control study to assess systemic complement activation in STGD1 patients and its association with disease severity.
METHODS: Systemic concentrations of complement component C3 and its degradation product C3d were compared between 80 STGD1 patients and 80 controls that were frequency matched for age and sex. The C3d/C3 ratio was used as parameter of systemic complement activation. Within the STGD1 cohort, we additionally examined the association between the C3d/C3 ratio, demographic and behavioural factors (age, sex, smoking and BMI), and measures of disease severity (age at onset, visual acuity, and area of atrophy).
RESULTS: The C3d/C3 ratio did not significantly differ between patients (mean C3d/C3 ratio 3.5±1.4) and controls (mean C3d/C3 ratio 3.6±1.0), mean difference -0.156 (p = 0.804, independent samples t-test). The overall effect size was 8% (95% confidence interval, 3-15%). Elevated C3d/C3 ratios (>8.1) were found in three patients who all had a concomitant inflammatory condition at the time of blood draw. Within the patient cohort, C3 levels were associated with sex (mean difference -134, p = 0.001, independent samples t-test) and BMI (correlation coefficient 0.463, p<0.001, Spearman's Correlation).
CONCLUSIONS: Systemic complement levels were not elevated in STGD1 patients compared to age and sex matched controls and was not associated with STGD1 severity. Considering the continued absent proof of a systemic contribution of the complement system to RPE loss in STGD1 patients, we hypothesize that complement activation in STGD1 is more likely a local process. In light of upcoming complement-targeted therapies, further studies are needed that measure complement levels in the eye of STGD1 patients.

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

Participants

For this cross-sectional case-control study, 80 consecutive STGD1 patients and 80 controls without retinal pathology were recruited. Their demographics, blood samples and medical imaging were collected in the Radboud Biobank [28]. In a recent study performed by our group, the C3d/C3 ratio increased with 42% (95% confidence interval (CI) -17% - 103%) when comparing AMD patients with GA to controls. This effect size of 42% was derived from the log C3d/C3 ratio of 1.40 in controls versus 1.57 in AMD patients with GA and a population SD of 0.41 [26]. A sample size of 80 cases and 80 controls is appropriate to reach a statistical power of 80% at a significance level of 0.05, in case of a similar difference as was previously reported in AMD. Simulations for power in relation to patient numbers are provided in S1 Fig. With a cohort of 80 patients and 80 controls, the current study is not sufficiently powered to detect small effect sizes (<20%). Patients were included if they had a STGD1 phenotype defined as bilateral macular dystrophy with presence of yellow-white irregularly shaped flecks corresponding to hyperautofluorescent lesions on short wavelength autofluorescence (SW-AF) and at least one pathogenic ABCA4 variant.

The control cohort was actively recruited to match both age and sex of the STGD1 cohort. Controls were sampled to match the age and sex frequency of the cases. The control cohort consisted of employees of the Radboud University Medical Center and unrelated spouses of patients within in the eye clinic. To determine that control participants were indeed without retinal pathology, they underwent a full ophthalmic examination including either a fundus photo or spectral domain optical coherence tomography scan.

All participants provided written informed consent before inclusion in the study. Ethical approval was obtained by the CMO Arnhem-Nijmegen and the Radboud University Medical Center local institutional review board (project number: 2017–3535) and this study was conducted according to the tenets of the Declaration of Helsinki. The STROBE guidelines were used to ensure the reporting of this observational study [29].

Sample collection

Samples were collected according to the standard operating procedures (SOPs) of the Radboud Biobank [28]. Ethylenediaminetetraacetic acid (EDTA) tubes were centrifuged (10 minutes at 2000g at 4°C), immediately after blood draw. Serum tubes were first placed at room temperature for 1 hour to coagulate and then centrifuged (10 minutes at 2000g at 4°C). After centrifugation, the obtained plasma and serum samples were kept on dry ice until they were stored in a -80°C freezer within 2 hours after collection.

Complement measurements

The level of C3d was quantified in EDTA plasma using an ELISA, as previously described in detail with minor changes [30]. To determine C3d values, pooled activated EDTA plasma of healthy blood donors was used and set to 100%. This value was calibrated to mg/l using purified C3d containing 10 μg/ml protein. The reference value was determined in a group of 20 healthy volunteers (mean age ± SD; 45.8 ± 11.4) and set at <8.3 mg/l. The C3 concentration was determined in serum by turbidometry (Cobas 8000 platform, F. Hoffman-La Roche Ltd, Basel, Switserland). Reference values for C3 are 700–1500 mg/l. As the initial C3 concentration may influence the C3d level and because of intra-individual variations in the concentration of C3, we used the C3d/C3 ratio as the parameter for complement activation [31]. The reference value for the ratio C3d/C3 is <8.1. The investigators performing the C3 and C3d quantification were masked to patient information.

Clinical data

Clinical data including demographics (age, sex), use of medication and medical history including information about allergies, previous and current illnesses, surgeries, and abnormal results of physical exams and tests were obtained for all participants. Since inflammatory conditions result in increased complement activation, health complaints and fever in the week preceding blood draw were documented, as well as a history of any chronic inflammatory disease.

To evaluate the factors that might influence activation of the complement system within our STGD1 cohort, we provided the STGD1 patients with a previously used questionnaire [32]. BMI and smoking status (never/past/current and packyears) were documented, because these are known to influence activation of the complement system [15,33].

Measures of disease progression were collected in order to investigate its potential association with systemic complement activation. On the day of the blood draw, visual acuity was measured after refraction using Snellen charts, and a fundus autofluorescence (FAF) image was taken using the Heidelberg Spectralis HRA-OCT (Heidelberg Engineering, Heidelberg, Germany). Moreover, patient records and imaging databases were reviewed to extract the earliest VA measurement and FAF image. To assess the area of RPE atrophy on the FAF images, areas with definitely decreased autofluorescence (DDAF), as defined by Kuehlewein et al. [34], were analysed using the semiautomated RegionFinder software (Heidelberg Engineering, version 2.6.3) according to the method previously described by Schmitz-Valckenberg et al [35]. We used a simplified model of disease progression by considering VA decline and DDAF growth linearly: the difference between the earliest retrospective measurement and the study measurement on the day of blood collection was divided by the number of years between those two measurements.

Statistical analysis

Statistical analysis was performed using the SPSS statistics package for Windows; version 22 (SPSS IBM, New York, USA). First, C3 levels, C3d levels, the C3d/C3 ratio, age and gender were compared between patients and controls using an independent samples t-test or a chi-square test.

Because of the normal distribution of all complement data in both cases and controls, parametric independent t-tests were used when comparing the complement levels between cases and controls.

Second, Spearman’s correlation, ANOVA and independent samples t-test were used to examine associations between the complement levels (C3, C3d and C3d/C3 ratio) and variables related to disease progression (age at onset, VA, DDAF area, VA decline, DDAF growth) as well as demographic and behavioural variables (age, sex, smoking, BMI). The more general non-parametric Spearman’s correlation was used to test for associations in order to take the possibility of a non-linear relationship into account. A p-value of <0.05 was considered significant in this exploratory study.

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).

Table 1

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

Variable	STGD1 (n = 80)	Controls (n = 80)	Mean Difference (95% CI)	P Value
Age in years, mean (SD)	46 (18)	45 (18)	0.712 (-4.948–6.373)	0.804a
Female sex, n (%)	44 (55%)	44 (55%)	not applicable	1.00b
C3d in mg/l, mean (SD)	3.7 (1.3)	3.7 (0.9)	-0.033 (-0.373–0.308)	0.851a
C3 in mg/l, mean (SD)	1099 (196)	1057 (155)	42.5 (-12.6–97,5)	0.130a
C3d/C3 ratio, mean (SD)	3.5 (1.4)	3.6 (1.0)	-0.156 (-0.537–0.225)	0.419a

aIndependent samples t-test

bchi-square test. C3, complement component C3; C3d, complement degradation product C3d; CI, confidence interval; SD, standard deviation; STGD1, Stargardt disease.

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.

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.

Table 2

Correlation analysis of all parameters within Stargardt disease patients.

Parameter	STGD1 patients (n = 80), Mean (SD) or n (%)		C3d (mg/l)	C3 (mg/l)	C3d/C3 ratio
Cross-sectional parameters related to STGD1 pathology
Age at onset in years	33 (19)	Spearman’s CorrelationP value	-0.1270.262	-0.0440.700	-0.1170.303
Disease duration in years	13 (12)	Spearman’s CorrelationP value	-0.0610.589	0.0200.863	-0.0300.789
VA in decimals, average of both eyes	0.40 (0.40)	Spearman’s CorrelationP value	-0.1300.249	-0.0200.858	-0.1270.261
DDAF area in mm2, average of both eyes	10 (18)	Spearman’s CorrelationP value	-0.0090.938	-0.0110.929	0.0190.873
Retrospective parameters related to STGD1 progression (available for 50 patients)
VA decline in decimals/year, average of both eyes	0.02 (0.08)	Spearman’s CorrelationP value	0.1380.311	-0.0100.941	0.1340.324
DDAF growth in mm2/year, average of both eyes	0.94 (1.53)	Spearman’s CorrelationP value	-0.0140.926	-0.2300.108	0.0940.517
Demographic and behavioural factors
Age in years	46 (18)	Spearman’s CorrelationP value	-0.1680.135	-0.0370.744	-0.1500.183
Sex	Males, 36 (45%)Females, 44 (55%)	Level in males, mean (SD)Level in females, mean (SD)Mean Difference (95% CI)T-test, P value	3.7 (1.1)3.7 (1.4)-0.712 (-0.634–0.492)0.110	1025 (112)1159 (228)-134 (-212 - -56)0.001 a	3.6 (1.2)3.3 (1.2)0.302 (-0.315–0.920)0.396
BMI	25 (4)	Spearman’s CorrelationP value	0.1240.273	0.463 <0.001a	-0.1730.124
Smoking 3 categories	Current smokers, 11 (14%)Past smokers, 22 (27%)Non-smokers, 47 (59%)	Level in current smokers, mean (SD)Level in past smokers, mean (SD)Level in non-smokers, mean (SD)ANOVA, p value	4.2 (2.0)3.4 (0.8)3.7 (1.2)0.273	1122 (232)1104 (193)1091 (192)0.894	4.0 (2.6)3.1 (0.6)3.5 (1.2)0.211
Smoking in packyears	5 (11)	Spearman’s CorrelationP value	0.0210.855	0.0330.774	-0.0540.631

ap value <0.05. C3, complement component 3; C3d, complement component 3d; CI, confidence interval; SD, standard deviation; STGD1, Stargardt disease.

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.

Simulations for power in relation to patient numbers.

Panels display different prevalences of an above normal C3d/C3 ratio (>8.1) in controls; (A) 1%, (B) 3%, (C) 5%, (D) 10%. Graphs display the number of cases in relation to the statistical power of the study, assuming the 1:1 matched design of the current study. Each graph contains 4 lines with different effect sizes (5%, 10%, 15%, 20%). Very large cohorts are needed to detect small effect sizes.

(TIF)

Click here for additional data file.

Correlation plots of all parameters within Stargardt disease patients.

Correlation plots of respectively C3d, C3 and C3d/C3 ratio with (A) cross-sectional parameters related to STGD1 pathology, (B) retrospective parameters related to STGD1 progression and (C) demographic and behavioral factors.

(TIF)

Click here for additional data file.

13 Apr 2021

PONE-D-21-08972

Systemic complement activation levels in Stargardt disease

PLOS ONE

Dear Dr. den Hollander,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Two reviewers found this a very interesting study and made suggestions to improve it. We look forward to the revised manuscript !

Please submit your revised manuscript by May 28 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Demetrios G. Vavvas

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for including your ethics statement:  "Ethical approval was obtained by the local institutional review board (project number: 2017-3535). All participants provided written informed consent before inclusion in the study. ".

Please amend your current ethics statement to include the full name of the ethics committee/institutional review board(s) that approved your specific study.

Once you have amended this/these statement(s) in the Methods section of the manuscript, please add the same text to the “Ethics Statement” field of the submission form (via “Edit Submission”).

For additional information about PLOS ONE ethical requirements for human subjects research, please refer to http://journals.plos.org/plosone/s/submission-guidelines#loc-human-subjects-research.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: In this manuscript, Dhooge et al. evaluate the systemic complement activation profiles of patients with Stargardt disease compared to controls without retinal pathology. This topic is clinically of interest since there are preclinical studies, predominately in animal models, that suggest that complement dysregulation may play a role in the pathogenesis of Stargardt disease. Furthermore, there is a human study, as the authors cite, investigating the use of the C5 inhibitor avacincaptad pegol in patients with Stargardt disease. As such, this is a timely study that has the potential to enhance our understanding of the role that complement dysregulation may play in Stargardt disease and to inform the development of future clinical trials. The authors are encouraged to address the following concerns.

Major concerns

1. Line 90: Selection of controls is a very important consideration for any case-control study. Can the authors provide further details regarding how these control participants were recruited? The manuscript mentions that they are “without retinal pathology.” How was this determined? To what extent did the authors investigate whether the control patients had any other medical/ophthalmic co-morbidities (e.g., self-report, ophthalmic examination, or otherwise)? If they were clinic-/hospital-based controls, the authors should consider the possibility of selection bias (Berkson paradox) and how this may have influenced their findings. Further details by the authors would enable a reader to assess the possibility of bias; limitations should also be detailed further in the Discussion.

2. Lines 117-119, Lines 168-172: The authors acknowledge the important point that other factors (e.g., concurrent infection, autoimmune disease) can influence systemic complement activation profiles. This reviewer is curious why the authors did not exclude participants with symptoms suggestive of underlying infection or chronic autoimmune disease, as inclusion of these participants may complicate understanding of the relationship between complement activation and Stargardt disease. Is it known whether cases or controls were on any immunosuppressive medications that could alter systemic complement activation? This is a major limitation of the present study that complicates interpretation of the central findings. Would it be possible for the authors to perform a sensitivity analysis of omitting the participants with such features to determine whether their results remain the same? This should also be further elucidated in the Discussion.

3. As the authors mention, complement activation is heavily influenced by other factors, including age, sex, BMI, and smoking. Despite the fact that the controls were age- and sex-matched to cases, there may be residual confounding that still warrants controlling for (https://www.bmj.com/content/352/bmj.i969), especially since there were differences based on sex in the cases (Line 187-188). The fact that there was no statistically significant difference between age or sex between cases and controls does not rule out the possibility that they may be influencing the results. As such, the authors are encouraged to repeat their analysis but control for the covariates of age, sex, BMI, and smoking, for example, with general linear models. This is not likely to change the ultimate conclusions given that the authors do not find a significant difference but is still worth considering for the sake of completeness and rigorous statistical methodology.

Minor concerns:

Line 83: Given that the patients were not followed longitudinally, I think it would be more appropriate to characterize this study as a cross-sectional study here and elsewhere.

Line 105: Were the investigators performing C3/C3d quantitation masked to patient information?

Line 128-135: Could the authors provide more specific information of what is meant by “first available” VA/FAF image? Does this mean the most recent or the earliest? Was the interval similar from patient to patient? This may affect the accuracy of the authors’ simplified model of assuming linear decline.

Line 138-144: Throughout the manuscript, could the authors provide more specific rationale for their selection of statistical testing. In some cases, they have used parametric independent t-tests but, elsewhere, they have used the non-parametric Spearman correlation test to evaluate similar data. Fisher’s exact test is often used instead of the chi-square test for small samples sizes; was there a specific reason that this was chosen?

Table 2: Please show the correlation plots of the relationships described in Table 2, which would enhance the ability of the reader to visualize the relationships or lack thereof.

Reviewer #2: # Overview

I would like to thank the authors for working on a very interesting question with particularly exciting therapeutic implications. Not being a subject expert, I can only comment on the epidemiological and methodological merits of this study.

This study uses a case-control design of 80 confirmed Stargardt cases and 80 age and sex matched controls to identify no clinically or statistically significant differences in C3d/C3 between the two.

This manuscript correctly uses a case-control design to study a rare disease and commendably presents power calculations. However, it also suffers from (a) the absence of more elaborate power analyses necessitated by the null findings of this study, (b) the lack of clarity in how controls were identified and sampled and (c) the lack of robust associational analyses between multiple potential risk factors and C3d/C3.

Overall, the epidemiological/methodological aspects of this manuscript require a major revision before it can meaningfully contribute to the biomedical literature.

------------------------------------------------------------------------

# Major concerns

1. **Power analysis.** Methods, Lines 85-87. The authors should be congratulated for including a power analysis. This power analysis is extraordinarily important in this article because it did not identify a statistically significant difference between cases and controls. As such, we need a few more details. First, we need to know what effect size this power refers to - merely stating that this is a "similar difference as was previously reported in AMD" is inadequate. Some numbers are mentioned in the Discussion (line 220), but we need the effect size, not the numbers (e.g. an increase of 20% in C3d/C3 ratio between controls and cases). Second, what was the assumed significance level? I presume this was 0.05, but it has to be clarified. Third, please provide a figure that consists of 4 panels of graphs. Each graph should represent a different prevalence of "high" C3d/C3 ratio in the controls to cover 4 realistic scenarios (e.g. 1%, 3%, 5%, 10% prevalence). The horizontal axis of each graph should be the number of cases, the vertical axis should be the power and each graph should contain 4 lines referring to 4 different realistic effect sizes (e.g. 5% vs. 10% vs. 15% vs. 20% increase). These graphs would assume the 1:1 matched design of the current study. Such a figure would help us understand how small an effect size this study was powered to detect and what we would need to do if we wanted to power a study for an even smaller effect size (which seems to be the case in this study).

2. **Lack of control description.** Methods, Lines 90-91. (a) It is unclear whether this refers to individual matching (e.g. a 40 year old male STGD1 patient was matched to a 40 year old male control) vs. frequency matched (e.g. the controls were sampled to match the age and sex frequency of the cases). This matters because the two strategies necessitate different analytic approaches. I have assessed this manuscript assuming that frequency matching was used. Please clarify the type of matching here and in the abstract. (b) We need a lot more information about how controls were chosen. After recall bias, the second most important thread to the validity of a case-control study is the choice of controls. Where were these controls found? How were they sampled? What were the eligibility criteria?

3. **Poorly interpretable associational analysis.** Results, Lines 183-190. These results are very difficult to interpret using such analyses. If the authors wish to retain this work, I recommend that they replace it with a univariable regression analysis (i.e. use a regression to identify the associational effect of each variable on C3, C3d and their ratio) and supplement it with a multivariable regression analysis. Both analyses should also include a feature for whether each individual is a case or control to respect the design of this study. The authors can also consider excluding this analysis altogether as it does not contribute to the scientific importance of this work and its main hypothesis of interest.

4. **Need to use the STROBE reporting guidelines.** It is extremely important that the authors use the STROBE reporting guidelines (https://www.equator-network.org/wp-content/uploads/2015/10/STROBE_checklist_v4_combined.pdf) to make sure that they are reporting on all aspects of their study. As can be seen throughout my comments, there is a number of missing pieces. Please also cite the STROBE guidelines to promote trust in your work.

5. **Need to make all non-PII data and code available.** As per PLOS One guidelines, please make all data with non-personally identifiable information available on repositories like Zenodo. Similarly, please make all of your code available on repositories like Zenodo, GitHub or OSF (Open Science Framework).

------------------------------------------------------------------------

# Minor concerns.

1. Abstract, Line 16. Please report the overall effect size and its confidence interval, in addition to the means and variance within the two groups.

2. Abstract, Lines 19-20. Please report the mean difference and correlation in addition to the p-values.

3. Methods, Line 83 and Line 122. What does the manuscript mean with "prospective case-control study"? Does it mean that the exposures (i.e. C3d, C3) and risk factors (e.g. BMI, smoking, etc.) were recorded from medical records that predate the development of disease? If yes, then this is indeed a prospective case-control study, otherwise, this is a retrospective case-control study. In Methods Line 122, the use of a questionnaire suggests that this is probably a retrospective case-control study. Note that the reason we distinguish between retrospective and prospective case-control studies is one of the major caveats of such studies, namely "recall bias". Prospective studies use exposures that were recorded before the disease status was known, hence are unlikely to suffer from recall bias, whereas retrospective case-control studies use questionnaires to identify the status of exposures, hence are likely to suffer from recall bias.

4. Methods, Line 116. What medical history was obtained? This does not seem to be mentioned or used anywhere in the manuscript, yet it is important in understanding how similar the cases are to controls.

5. Methods, Line 140. It does not make sense to use a t-test and a Spearman's correlation. Spearman's correlation is used to avoid certain parametric assumptions, which are made by the t-test. As such, either use a non-parametric equivalent of the t-test (e.g. Mann-Whitney U Test) and Spearman's correlation, or the t-test and a parametric equivalent of the Spearman's correlation (e.g. Pearson's correlation). Unless given a good reason, this study should be using the t-test and Pearson's correlation, which are more powerful tests.

6. Results, Line 149. The Fisher's exact test was not mentioned in the statistical analysis. As per my above comment, please stick to parametric tests unless there is a good reason not to. Also, please add another column to this table to report the mean difference between the two groups and its 95% confidence interval - these are the numbers we really care about.

7. Results, Lines 151-152. Please report comparisons for ALL variables collected, not just the age, sex and C3d/C3 (e.g. smoking status, BMI, etc.).

8. Results, Lines 153-157. These dot plots are very helpful, thank you. Please superimpose these dots with a box plot to illustrate also the first and third quartiles.

9. Results, Lines 168-181. This paragraph confirms that the cases may indeed come from a qualitatively different "population' than the controls. As noted above, it is extremely important to describe how cases and controls were sampled, from where, etc.

10. Discussion, Line 196. Please clarify that by "disease progression" this manuscript refers to "rate of progression".

11. Discussion. Please include a limitations section here. Please state at least the three most important limitations and how robust this study may be to these limitations.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Stylianos Serghiou

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

25 May 2021

We would like to thank the editor and the reviewers for their additional comments to improve our manuscript. We uploaded a point-by-point response to their additional comments named Response to Reviewers. Please refer to this file which is easier to read due to layout. We hope you now find our manuscript suitable for publication in PLOS ONE.

Suggestion, Question, or Comment from Reviewer #1

1. Line 90: Selection of controls is a very important consideration for any case-control study. Can the authors provide further details regarding how these control participants were recruited? The manuscript mentions that they are “without retinal pathology.” How was this determined? To what extent did the authors investigate whether the control patients had any other medical/ophthalmic co-morbidities (e.g., self-report, ophthalmic examination, or otherwise)? If they were clinic-/hospital-based controls, the authors should consider the possibility of selection bias (Berkson paradox) and how this may have influenced their findings. Further details by the authors would enable a reader to assess the possibility of bias; limitations should also be detailed further in the Discussion.

We provided additional details about the control cohort in the methods part.

Because we did not include clinic-/hospital-based controls, we do not expect a selection bias. We did provide a limitation paragraph in the discussion section, however as we do not expect a selection bias, this was not included in this limitation paragraph.

Methods, participants, line 90, following text was added:

The control cohort was actively recruited to match both age and sex of the STGD1 cohort. Controls were sampled to match the age and sex frequency of the cases. The control cohort consisted of employees of the Radboud University Medical Center and unrelated spouses of patients within in the eye clinic. To determine that control participants were indeed without retinal pathology, they underwent a full ophthalmic examination including either a fundus photo or spectral domain optical coherence tomography scan.

2. Lines 117-119, Lines 168-172: The authors acknowledge the important point that other factors (e.g., concurrent infection, autoimmune disease) can influence systemic complement activation profiles. This reviewer is curious why the authors did not exclude participants with symptoms suggestive of underlying infection or chronic autoimmune disease, as inclusion of these participants may complicate understanding of the relationship between complement activation and Stargardt disease. Is it known whether cases or controls were on any immunosuppressive medications that could alter systemic complement activation? This is a major limitation of the present study that complicates interpretation of the central findings. Would it be possible for the authors to perform a sensitivity analysis of omitting the participants with such features to determine whether their results remain the same? This should also be further elucidated in the Discussion.

We did not exclude participants with symptoms suggestive of underlying infection beforehand, because we wanted to give a complete overview of our STGD1 cohort as we are dealing with a rare disease.

We are well aware that this complicates the understanding of the relationship between complement activation and STGD1. This is why we elaborated on the concomitant diseases with a full paragraph in the results section (lines 168 – 181).

Neither cases, nor controls were on any immunosuppressive medication at time of inclusion. The patient with breast cancer is being treated with anastrozole, a hormone therapy that does not modulate the immune system.

We performed a sensitivity analysis omitting patients with symptoms suggestive of underlying infection (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 did not show a significant difference between these means (p = 0.077). The sensitivity analysis does not change our ultimate conclusions and confirms that systemic complement levels are indeed not elevated in STGD1 patients.

Results, line 170, text added: Neither patients, nor controls were on any immunosuppressive medication.

Discussion, line 199, text added: 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.

3. As the authors mention, complement activation is heavily influenced by other factors, including age, sex, BMI, and smoking. Despite the fact that the controls were age- and sex-matched to cases, there may be residual confounding that still warrants controlling for (https://www.bmj.com/content/352/bmj.i969), especially since there were differences based on sex in the cases (Line 187-188). The fact that there was no statistically significant difference between age or sex between cases and controls does not rule out the possibility that they may be influencing the results. As such, the authors are encouraged to repeat their analysis but control for the covariates of age, sex, BMI, and smoking, for example, with general linear models. This is not likely to change the ultimate conclusions given that the authors do not find a significant difference but is still worth considering for the sake of completeness and rigorous statistical methodology.

We agree that general linear models or regression analysis would complete the statistical methods. However, there are 3 important reasons why we did not include these in the manuscript.

1) With the current data, it is only possible to control for the covariates age and sex, and not for BMI and smoking (when comparing STGD1 to controls), because we did not collect BMI and smoking variables for our control cohort. It is also not possible to do a multivariable regression analysis with a feature indicating if an individual is a case or control. The more is asked from control participants, the less likely they are willing to cooperate. Therefore we asked as little as possible from the control group. At the moment of data collection we were only interested in the additional factors that might influence complement activation within the STGD1 cohort and limited the BMI and smoking question to our STGD1 cohort (together with parameters related to STGD1, which are not present in controls). We now see that this was not clearly written in the manuscript and we have changed the text accordingly.

Since we cannot correct for all possible covariates, we did not add this calculation to the manuscript because it wouldn't give a complete picture.

2) Repeating the association analysis with general linear models or multivariate regression for the variables in table 2 is difficult, because the exact role of the complement system within STGD1 is still not clear. When using regression or linear models you need to define cause and consequence. However, we are not sure if atrophy leads to higher complement levels, or that, if a person has higher complement levels from other causes (BMI/sex), he/she also develops more atrophy. We hypothesize that increased complement activation leads to increased damage to the RPE. This increased damage will lead to lower visual acuity and larger areas of geographic atrophy. This leads to the following sequence:

Age/Sex/BMI/Smoking � Complement levels � RPE damage � VA & DDAF

Complement level is an intermediary variable in this sequence. Therefore we cannot include all variables in Table 2 in one model. Correcting for age/sex/BMI/smoking when evaluating whether systemic complement levels are associated with disease parameters or disease progression is not possible in this way.

3) Using general linear models or regression analysis, one assumes that relationships are linear. We used Spearman’s correlation to determine more general relationships between values. Since we don’t find any correlation between C3d/C3 ratio and the values in table 2, it is unlikely that we find an association with general linear models or univariate regression. Adding general linear models or univariate regression analysis would therefore not change the ultimate conclusion in this manuscript. We prefer to use Spearman’s correlation to look at more general relationships between values.

This study is the first to investigate the potential association between systemic complement activation and disease progression and can therefore be seen as a pilot study. Prospective studies in which complement levels are measured at several time points during the disease course are necessary to establish the effect of covariates on complement levels and to establish the effect of complement levels on disease progression. This is also stated in the discussion section.

Methods, Clinical Data, lines 119-122: ‘In addition to …

through a questionnaire.’ was changed to ‘To evaluate the factors that might influence activation of the complement system within our STGD1 cohort, we provided the STGD1 patients with a previously used questionnaire. [31] BMI and smoking status (never/past/current and packyears) were documented, because these are known to influence activation of the complement system.[15, 32]’

Discussion, line 256, limitations paragraph added:

Our study has several limitations … Third, complement levels are known to be influenced by other factors including age, sex, BMI and smoking. [15, 32] 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 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.

4. Line 83: Given that the patients were not followed longitudinally, I think it would be more appropriate to characterize this study as a cross-sectional study here and elsewhere.

We agree and changed prospective case-control study to cross-sectional case-control study.

Abstract line 6: ‘We conducted a prospective case-control study’ was changed to: We conducted a cross-sectional case-control study.

Results line 83: For this case-control study was changed to for this cross-sectional case-control study.

5. Line 105: Were the investigators performing C3/C3d quantitation masked to patient information? Yes, these investigators were masked.

We added this information to the manuscript.

Methods, complement measurements, line 113, text added: The investigators performing the C3 and C3d quantification were masked to patient information.

6. Line 128-135: Could the authors provide more specific information of what is meant by “first available” VA/FAF image? Does this mean the most recent or the earliest? Was the interval similar from patient to patient? This may affect the accuracy of the authors’ simplified model of assuming linear decline.

In this case, ‘first available’ means ‘the earliest’.

The used interval differs between patients and depends on the age at diagnosis and disease duration. Therefore, the simplified model gives a value for the progression per year.

For atrophy measurements it is known that atrophy radius enlarges at a constant rate of approximately 0.163 mm/year, independent of time.(PMID: 3104722) VA progression is not linear in STGD1 and different intervals may indeed affect the accuracy of the simplified model. However, until now there is no easy way to assess the VA progression using one variable. As we state in the discussion, 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.

Methods, clinical data, line 128: ‘patient records and imaging databases were reviewed to extract the first available VA measurement and FAF image’ was changed to: patient records and imaging databases were reviewed to extract the earliest VA measurement and FAF image.

Methods, clinical data, line 134: ‘the difference between the first retrospective measurement…’ was changed to: the difference between the earliest retrospective measurement.

7. Line 138-144: Throughout the manuscript, could the authors provide more specific rationale for their selection of statistical testing. In some cases, they have used parametric independent t-tests but, elsewhere, they have used the non-parametric Spearman correlation test to evaluate similar data. Fisher’s exact test is often used instead of the chi-square test for small samples sizes; was there a specific reason that this was chosen?

Our current data show a normal distribution for all the complement data in both cases and controls (C3, C3d and C3d/C3 ratio). Because of the normal distribution of the data, we can use parametric independent t-tests when comparing the complement levels between cases and controls (Table 1).

For Table 2 we specifically used non-parametric tests. The complement values are normally distributed, but the other tested values (VA,DDAF) are often not normally distributed. Besides, we assume that the relationship between all tested values in Table 2 might not be linear. The parametric Pearson’s correlation only evaluates linear relationships, while the non-parametric Spearman’s correlation is more general. We therefore used Spearman’s correlation to evaluate more general relationships between values.

Fischer’s exact was only used to compare sex between cases and controls. We agree that chi-square test is a better fit for small samples sizes. We therefore changed the analysis accordingly. When comparing sex between cases and controls with Chi-square instead of Fisher’s exact, p value remains 1.00.

We adjusted the statistical analysis section in methods to clarify our rationale.

Methods, Statistical Analysis, lines 138-144 was changed to: Statistical analysis was performed using the SPSS statistics package for Windows; version 22 (SPSS IBM, New York, USA). First, C3 levels, C3d levels, the C3d/C3 ratio, age and gender were compared between patients and controls using an independent samples t-test or a chi-square test. Because of the normal distribution of all complement data in both cases and controls, parametric independent t-tests were used when comparing the complement levels between cases and controls. Second, Spearman’s correlation, ANOVA and independent samples t-test were used to examine associations between the complement levels (C3, C3d and C3d/C3 ratio) and variables related to disease progression (age at onset, VA, DDAF area, VA decline, DDAF growth) as well as demographic and behavioural variables (age, sex, smoking, BMI). The more general non-parametric Spearman’s correlation was used to test for associations in order to take the possibility of a non-linear relationship into account. A p-value of <0.05 was considered significant in this exploratory study.

Results, lines 148-149: ‘There was .. both groups’ was changed to There was indeed no difference in age (p=0.804, t-test) or sex (p=1.00, chi-square test) between both groups.

Table 1, line 151: ‘Fischer’s exact’ was replaced by chi-square test.

8. Table 2: Please show the correlation plots of the relationships described in Table 2, which would enhance the ability of the reader to visualize the relationships or lack thereof.

We provided correlation plots in supporting Fig S2.

Supporting Fig S2 was added to the manuscript.

Suggestion, Question, or Comment from Reviewer #2

1. **Power analysis.** Methods, Lines 85-87. The authors should be congratulated for including a power analysis. This power analysis is extraordinarily important in this article because it did not identify a statistically significant difference between cases and controls. As such, we need a few more details. First, we need to know what effect size this power refers to - merely stating that this is a "similar difference as was previously reported in AMD" is inadequate. Some numbers are mentioned in the Discussion (line 220), but we need the effect size, not the numbers (e.g. an increase of 20% in C3d/C3 ratio between controls and cases). Second, what was the assumed significance level? I presume this was 0.05, but it has to be clarified. Third, please provide a figure that consists of 4 panels of graphs. Each graph should represent a different prevalence of "high" C3d/C3 ratio in the controls to cover 4 realistic scenarios (e.g. 1%, 3%, 5%, 10% prevalence). The horizontal axis of each graph should be the number of cases, the vertical axis should be the power and each graph should contain 4 lines referring to 4 different realistic effect sizes (e.g. 5% vs. 10% vs. 15% vs. 20% increase). These graphs would assume the 1:1 matched design of the current study. Such a figure would help us understand how small an effect size this study was powered to detect and what we would need to do if we wanted to power a study for an even smaller effect size (which seems to be the case in this study).

The effect size in our recent study performed by our group in AMD was 42%. (PMID: 32176267) The assumed significance level was 0.05.

We clarified the power calculation accordingly.

We provided the requested figure as supporting Fig S1.

Methods, Participants, line 85-87: ‘With a sample … in house study’ was changed to In a recent study performed by our group, the C3d/C3 ratio increased with 42% (95% confidence interval (CI) -17% - 103%) when comparing AMD patients with GA to controls. This effect size of 42% was derived from the log C3d/C3 ratio of 1.40 in controls versus 1.57 in AMD patients with GA and a population SD of 0.41.[26] A sample size of 80 cases and 80 controls is appropriate to reach a statistical power of 80% at a significance level of 0.05, in case of a similar difference as was previously reported in AMD. Simulations for power in relation to patient numbers are provided in supporting Fig S1. With a cohort of 80 patients and 80 controls, the current study is not sufficiently powered to detect small effect sizes (<20%).

Supporting Fig S1 was added to the manuscript.

2. **Lack of control description.** Methods, Lines 90-91. (a) It is unclear whether this refers to individual matching (e.g. a 40 year old male STGD1 patient was matched to a 40 year old male control) vs. frequency matched (e.g. the controls were sampled to match the age and sex frequency of the cases). This matters because the two strategies necessitate different analytic approaches. I have assessed this manuscript assuming that frequency matching was used. Please clarify the type of matching here and in the abstract. (b) We need a lot more information about how controls were chosen. After recall bias, the second most important thread to the validity of a case-control study is the choice of controls. Where were these controls found? How were they sampled? What were the eligibility criteria?

Controls were sampled to match the age and sex frequency of the cases.

We provided additional details about the control cohort in the Methods section.

Please see our previous answer to reviewer 1 point 1. Abstract, line 10: ‘80 controls matched for age and sex’ was changed to 80 controls that were frequency matched for age and sex.

3. **Poorly interpretable associational analysis.** Results, Lines 183-190. These results are very difficult to interpret using such analyses. If the authors wish to retain this work, I recommend that they replace it with a univariable regression analysis (i.e. use a regression to identify the associational effect of each variable on C3, C3d and their ratio) and supplement it with a multivariable regression analysis. Both analyses should also include a feature for whether each individual is a case or control to respect the design of this study. The authors can also consider excluding this analysis altogether as it does not contribute to the scientific importance of this work and its main hypothesis of interest.

Indeed, our main objective is to compare the systemic complement levels between cases and controls. Our second objective is to evaluate whether systemic complement levels are associated with disease parameters or disease progression. This study is the first to investigate this potential association. In our opinion, the correlation analysis in Table 2 is scientifically important and prefer not to exclude this analysis.

Please see our previous answer to reviewer 1 point 3.

4. **Need to use the STROBE reporting guidelines.** It is extremely important that the authors use the STROBE reporting guidelines (https://www.equator-network.org/wp-content/uploads/2015/10/STROBE_checklist_v4_combined.pdf) to make sure that they are reporting on all aspects of their study. As can be seen throughout my comments, there is a number of missing pieces. Please also cite the STROBE guidelines to promote trust in your work.

We have checked our work with the STROBE reporting guidelines and cited the guidelines in the Methods section.

Methods, line 94, sentence added: The STROBE guidelines were used to ensure the reporting of this observational study.[29]

5. **Need to make all non-PII data and code available.** As per PLOS One guidelines, please make all data with non-personally identifiable information available on repositories like Zenodo. Similarly, please make all of your code available on repositories like Zenodo, GitHub or OSF (Open Science Framework).

We deposited the data in the Radboud Repository. Our data files will first be checked by our institutional Radboud Information Systems Servicedesk. After the check, the dataset will be submitted to the DANS EASY archive and will be publicly available. This process takes some time. We will provide the DOI as soon as possible.

6. Abstract, Line 16. Please report the overall effect size and its confidence interval, in addition to the means and variance within the two groups.

We reported these values. Effect size was also added to the discussion section.

Results, line 16: ‘The C3d/C3 … controls (3.6±1.0)’ was changed to The C3d/C3 ratio did not significantly differ between patients (mean C3d/C3 ratio 3.5±1.4) and controls (mean C3d/C3 ratio 3.6±1.0), mean difference -0.156, p=0.804, independent samples t-test. The overall effect size was 8% (95% confidence interval, 3-15%).

Discussion, line 219, sentence added: 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]

7. Abstract, Lines 19-20. Please report the mean difference and correlation in addition to the p-values.

We reported these values.

Abstract, Results, lines 16-21: ‘The C3d/C3 ratio … smoking status’ was changed to The C3d/C3 ratio did not significantly differ between patients (mean C3d/C3 ratio 3.5±1.4) and controls (mean C3d/C3 ratio 3.6±1.0), mean difference -0.156 (p=0.804, independent samples t-test). The overall effect size was 8% (95% confidence interval, 3-15%). Elevated C3d/C3 ratios (>8.1) were found in three patients who all had a concomitant inflammatory condition at the time of blood draw. Within the patient cohort, C3 levels were associated with sex (mean difference -134, p=0.001, independent samples t-test) and BMI (correlation coefficient 0.463, p<0.001, Spearman’s Correlation).

8. Methods, Line 83 and Line 122. What does the manuscript mean with "prospective case-control study"? Does it mean that the exposures (i.e. C3d, C3) and risk factors (e.g. BMI, smoking, etc.) were recorded from medical records that predate the development of disease? If yes, then this is indeed a prospective case-control study, otherwise, this is a retrospective case-control study. In Methods Line 122, the use of a questionnaire suggests that this is probably a retrospective case-control study. Note that the reason we distinguish between retrospective and prospective case-control studies is one of the major caveats of such studies, namely "recall bias". Prospective studies use exposures that were recorded before the disease status was known, hence are unlikely to suffer from recall bias, whereas retrospective case-control studies use questionnaires to identify the status of exposures, hence are likely to suffer from recall bias. We agree that prospective case-control study is not the correct term.

Patients were included if STGD1 diagnosis was already established. Blood was drawn to collect C3d and C3 exposures and we used a questionnaire to record the other exposures (BMI, smoking, medication) at the time of blood draw. Imaging was performed at the same day as the blood draw. In conclusion, all variables were collected at one given point in time. The patients were not followed longitudinally.

We therefore changed the characterization of our study to a cross-sectional case-control study.

Abstract line 6: ‘We conducted a prospective case-control study’ was changed to: We conducted a cross-sectional case-control study.

Results line 83: For this case-control study was changed to for this cross-sectional case-control study.

9. Methods, Line 116. What medical history was obtained? This does not seem to be mentioned or used anywhere in the manuscript, yet it is important in understanding how similar the cases are to controls.

We obtained a personal medical history including information about allergies, previous and current illnesses, surgeries, and abnormal results of physical exams and tests for both cases and controls. We specified this in the Methods section

Besides, we specifically asked participants about health complaints and fever in the week preceding blood draw as well as a history of any chronic inflammatory disease since inflammatory conditions result in increased complement activation. This was specified in lines 117-119.

Methods, lines 116-117:‘Clinical data including .. obtained for all patients’ was changed to Clinical data including demographics (age, sex), use of medication and medical history including information about allergies, previous and current illnesses, surgeries ,and abnormal results of physical exams and tests were obtained for all participants.

10. Methods, Line 140. It does not make sense to use a t-test and a Spearman's correlation. Spearman's correlation is used to avoid certain parametric assumptions, which are made by the t-test. As such, either use a non-parametric equivalent of the t-test (e.g. Mann-Whitney U Test) and Spearman's correlation, or the t-test and a parametric equivalent of the Spearman's correlation (e.g. Pearson's correlation). Unless given a good reason, this study should be using the t-test and Pearson's correlation, which are more powerful tests.

Please see our previous answer to reviewer 1 point 7.

11. Results, Line 149. The Fisher's exact test was not mentioned in the statistical analysis. As per my above comment, please stick to parametric tests unless there is a good reason not to. Also, please add another column to this table to report the mean difference between the two groups and its 95% confidence interval - these are the numbers we really care about.

We added another column to Table 1 that includes mean difference and the 95% confidence interval.

Indeed, Fisher’s exact was not mentioned in the statistical analysis. We agreed with reviewer 1 that chi-square test is a better fit for small samples sizes. We therefore changed the analysis accordingly.

We adjusted the statistical analysis section in the methods section and included the chi-square test.

Please see our previous answer to reviewer 1 point 7.

A column with mean differences and 95% confidence interval was added to Table 1.

12. Results, Lines 151-152. Please report comparisons for ALL variables collected, not just the age, sex and C3d/C3 (e.g. smoking status, BMI, etc.).

Please see our previous answer to reviewer 1 point 3 (reason 1).

Unfortunately, we cannot report additional comparisons between cases and controls. We did not collect BMI and smoking variables for our control cohort. The other values in Table 2 are related to STGD1 progression and it makes no sense to measure and report these for the control cohort.

13. Results, Lines 153-157. These dot plots are very helpful, thank you. Please superimpose these dots with a box plot to illustrate also the first and third quartiles.

We superimposed the dot-plots with a box plot.

Figure 1 was changed.

14. Results, Lines 168-181. This paragraph confirms that the cases may indeed come from a qualitatively different "population' than the controls. As noted above, it is extremely important to describe how cases and controls were sampled, from where, etc.

Please see our previous answers to reviewer 1 point 1 and point 2. And please also review the answer to reviewer 2 point 2.

15. Discussion, Line 196. Please clarify that by "disease progression" this manuscript refers to "rate of progression".

Text was changed accordingly.

Discussion, line 196: ‘Moreover, systemic complement levels were not associated with disease parameters or disease progression’ was changed to Moreover, systemic complement levels were not associated with disease parameters or the rate of disease progression.

16. Discussion. Please include a limitations section here. Please state at least the three most important limitations and how robust this study may be to these limitations.

We added a paragraph to the discussion section that describes the limitations of our study.

Discussion, line 256, limitations paragraph added:

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, 32] 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 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.

Journal Requirements

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

We checked our work with the PLOS ONE’s style requirements.

2. Thank you for including your ethics statement: "Ethical approval was obtained by the local institutional review board (project number: 2017-3535). All participants provided written informed consent before inclusion in the study. ".

Please amend your current ethics statement to include the full name of the ethics committee/institutional review board(s) that approved your specific study. Once you have amended this/these statement(s) in the Methods section of the manuscript, please add the same text to the “Ethics Statement” field of the submission form (via “Edit Submission”).

Ethics statement was adjusted accordingly.

Methods, Participants, line 92: ‘Ethical approval ..was board (project number: 2017-3535)’ was changed to Ethical approval was obtained by the CMO Arnhem-Nijmegen and the Radboud University Medical Center local institutional review board (project number: 2017-3535)

Other adjustments

1. Accidently swapped values in manuscript: When recalculating the mean difference, we noticed that p-values were swapped for the correlation analysis of sex with the complement levels. There was a significant difference in C3 between males and females (and not in C3d/C3 ratio as currently written in the manuscript).

Table 2 was changed accordingly.

Abstract, Results, lines 19-21: ‘Within the patient cohort … smoking status’ was changed to Within the patient cohort, C3 levels were associated with sex (mean difference -134, p=0.001, independent samples t-test) and BMI (correlation coefficient 0.463, p<0.001, Spearman’s Correlation).

Results, lines 187-188: ‘Complement levels … (p=0.001, independent samples t-test).’ was changed to 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).

2. Reference was added: We added the recently published manuscript (PMID 33849814) as reference 46

Reference 46. Dhooge PPA, Mulders TWF, Hoyng CCB. Commentary on “Evidence of complement dysregulation in outer retina of Stargardt disease donor eyes”. Redox Biology. 2021:101957. was added (and reference 45 was changed to 46).

Submitted filename: Response to Reviewers PONE-D-21-08972_AdH.docx

Click here for additional data file.

11 Jun 2021

Systemic complement activation levels in Stargardt disease

PONE-D-21-08972R1

Dear Dr. den Hollander,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Demetrios G. Vavvas

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have satisfactorily addressed all of my comments. I congratulate them on this excellent work!

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

18 Jun 2021

PONE-D-21-08972R1

Systemic complement activation levels in Stargardt disease

Dear Dr. den Hollander:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Demetrios G. Vavvas

Academic Editor

PLOS ONE