The Biochemical and Genetic Architecture of Geographic Atrophy: The Role of the FHL-1/CFH Axis and the Paradigm of RNA Interference Therapeutics

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is an interesting review on FHL-1/CFH axis in age related macular degeneration. It is generally well written. The major concern is the low number of references in this setting, given it is a review, it would be appropriate to cover a larger percentage of the work from the literature.

https://pubmed.ncbi.nlm.nih.gov/?term=%22factor+H+related%22+gene+and+age-related+macular+degeneration&sort=date&size=200

https://pubmed.ncbi.nlm.nih.gov/?term=CFH+gene+and+age-related+macular+degeneration%2C+geographic+atrophy&sort=date&size=200

There are other mechanisms and strategies to propose for GA other than siRNA. How about bolstering CFH or FHL-1 gene regulation? This part of the review seems quite limited.

 

Author Response

I thank the reviewer for a careful and constructive reading, and for the assessment that the manuscript is generally well written. Both points raised have been addressed substantively. Changes in the manuscript are shown as tracked changes; page and section references below refer to the clean version.

 

Comment 1

The major concern is the low number of references in this setting, given it is a review, it would be appropriate to cover a larger percentage of the work from the literature.

(with reference to PubMed searches for “factor H related” gene and AMD, and CFH gene and AMD / geographic atrophy)

Response. I accept this criticism. On re-checking the reference list against the reviewer’s searches I found not only that it was too short, but that it contained several errors — an entry that had merged two unrelated papers, an entry with no authors, incorrect journal abbreviations and volume/year mismatches, and three in-text citations to reference numbers that did not exist. These have all been corrected (see also Reviewer 2, point 5).

The reference list has been expanded from 26 to 37 entries, and every retained entry has been individually verified against the source record and corrected where necessary.

The new citations are drawn from precisely the literature the reviewer’s searches identify, and were chosen to strengthen the CFH/CFHR evidence base rather than to increase the count:

• Fritsche et al., Nat Genet 2016 (ref. 27) — the large AMD genome-wide association study, now cited to support the effect-size claim for Y402H and, importantly, to frame AMD as polygenic (see Reviewer 2, point 3).
• Cipriani et al., Nat Commun 2020 (ref. 28) — circulating FHR-4 and the rs10922109 association; this is now the principal source for the FHR arguments and replaces two of the three broken citation numbers.
• Clark et al., Front Immunol 2017 (ref. 29) — compartmentalisation of complement regulation by Bruch’s membrane, which underpins the spatial argument and the FHL-1 diffusion data.
• Armento, Ueffing & Clark, Cell Mol Life Sci 2021 (ref. 6) — a current review of the complement system in AMD.
• Grigsby et al., IOVS 2023 (ref. 7) — AAV augmentation of FHL-1 versus truncated CFH, now cited properly and with authors.
• Refs 30–36 — the clinical trial literature for the approved and failed complement-directed agents (OAKS/DERBY, GATHER1/GATHER2, GALE, CHROMA/SPECTRI, the pegcetacoplan safety report, and the GT005/PPY988 programme), added in response to Reviewer 2’s request for a clinical section.
• Refs 25 and 37 — peer-reviewed sources on dual-targeting siRNA, replacing a website citation.

I should be transparent about one point of scope. Rather than broadening the review towards a general survey, I have made its focus explicit: a new Scope and Methods section states that this is a narrative, focused review restricted to the FHL-1/CFH axis and the applicability of RNA interference to it, sets out the search strategy and databases, and notes that comprehensive reviews of AMD pathogenesis and of the wider GA pipeline already exist and are not duplicated here. My reasoning is that a further general review of complement in AMD would add little, whereas a bounded treatment of this particular axis and this particular modality does. If the reviewer and editor would prefer wider coverage instead, I am willing to reconsider.

 

Comment 2

There are other mechanisms and strategies to propose for GA other than siRNA. How about bolstering CFH or FHL-1 gene regulation? This part of the review seems quite limited.

Response. This is a fair criticism and, on reflection, an important structural gap: RNA interference can only reduce a transcript, and the defect being described is one of insufficient regulatory activity. A review that discussed only silencing therefore misrepresented the therapeutic landscape of its own subject.

A new section, “Alternative Strategies: Augmenting Rather Than Silencing the Regulatory Arm”, has been added before the concluding section. It covers:

• AAV-mediated gene augmentation. Subretinal CFI delivery (GT005/PPY988), including the FOCUS ocular biomarker data showing that local augmentation of a soluble regulator is achievable, and its discontinuation for futility in 2023. Grigsby et al. are discussed in detail, since their finding that AAV-FHL-1 and truncated CFH differentially restore regulation in the ocular versus plasma compartment bears directly on the spatial argument made earlier in the review — the two constructs are not interchangeable.
• Construct design. Why full-length CFH is problematic for AAV packaging, why truncated constructs and engineered “mini-FH” proteins (SCR 1–4 fused to SCR 19–20) have been pursued, and why FHL-1 is a particularly attractive payload for the macula given its size, lack of glycosylation and diffusion through Bruch’s membrane.
• Protein supplementation, and the trade-off it reintroduces: repeated intravitreal dosing, which is the burden that both RNAi and gene therapy are intended to avoid.
• Transcriptional upregulation of endogenous CFH (CRISPR activation, modulation of the inflammatory signalling that governs CFH expression in RPE), together with the specific logical objection that in a Y402H carrier this would simply produce more of a poorly anchoring protein — which is the argument for allele-selective correction by base editing or allele-specific silencing instead.
• Combination logic. Local augmentation and systemic FHR removal act on opposite sides of the same binding equilibrium at Bruch’s membrane and are complementary rather than competing.

The section closes by stating plainly that the one clinical test of ocular complement augmentation conducted to date did not succeed, and that the reasons — target choice, expression level, timing relative to established atrophy, or the anatomical endpoint itself — remain unresolved.

Related additions elsewhere in the manuscript also broaden the mechanistic discussion: a new clinical section covers the approved C3 and C5 inhibitors and the failure of the Factor D inhibitor lampalizumab, and Table 1 now compares FH, FHL-1, FHR-1, FHR-4, FHR-5, C3, C5 and Factor D side by side, including a column for therapeutic targeting status that distinguishes clinically validated targets from hypothesis-stage ones.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses an important and timely topic: the role of the CFH/FHL-1 axis in geographic atrophy and the therapeutic potential of RNA interference-based complement modulation. The mechanistic sections on FHL-1, Bruch’s membrane vulnerability, Y402H-mediated impairment, and the danger of direct CFH/FHL-1 silencing are scientifically valuable and relevant. However, the manuscript requires major revision before it is suitable for publication:

1. The manuscript would benefit from a short methods section describing the literature search strategy and the databases searched.
2. The manuscript should include a dedicated clinical section on currently approved complement inhibitors, including pegcetacoplan and avacincaptad pegol.
3. The author should present FHL-1/CFH dysregulation as a major mechanistic axis within a multifactorial disease, rather than as a fully sufficient explanation for GA pathogenesis.
4. The FHR discussion should distinguish genetic association, circulating protein association, tissue-level evidence, and causal therapeutic evidence.
5. The reference list appears incomplete and contains formatting/citation errors. In-text citations include numbers such as 29, 36, and 37, but the reference list shown in the manuscript ends at 26. Some references are malformed or incomplete; for example, reference 6 appears to merge different papers, reference 7 lacks normal bibliographic details, and reference 25 is a website.
6. The manuscript would be much easier to read if the authors reduced rhetorical language such as “monumental,” “catastrophic,” “massive,” “unyielding,” and “fiercely pro-inflammatory.”
7. The paper would benefit from one or two figures: one summarizing the FH/FHL-1/FHR architecture, and one showing the proposed therapeutic logic: direct CFH/FHL-1 silencing is harmful, while C5/FHR targeting may be beneficial.
8. There are several typographical or encoding issues in the PDF, such as broken words and strange characters in “attack” and other terms. Please correct them.
9. A table comparing FH, FHL-1, FHR-1, FHR-4, FHR-5, C3, C5, and Factor D would improve the clarity and help the reader have a more focused comparison between them.

Comments on the Quality of English Language

The Quality of the English Language can be improved.

Author Response

I thank the reviewer for a detailed and exacting review. The nine points are well made, and I have acted on all of them. The criticisms regarding evidential over-claiming (points 3 and 4) and the reference list (point 5) were particularly useful, and correcting them has improved the manuscript considerably. All changes are shown as tracked changes.

 

Point 1 — Methods section

The manuscript would benefit from a short methods section describing the literature search strategy and the databases searched.

Response. A Scope and Methods section has been added immediately after the Introduction. It states three things.

First, that this is a narrative, focused review and not a systematic review, with a scope defined in advance and restricted to two questions: the role of the FHL-1/CFH axis at the RPE–Bruch’s membrane interface, and whether RNA interference can be applied rationally to that axis.

Second, the search strategy: PubMed/MEDLINE and Web of Science from January 1990 to May 2026, with the search terms listed in full; ClinicalTrials.gov and the EU Clinical Trials Register for interventional studies of complement-directed agents in GA; and hand-searching of reference lists. Eligibility and the preference for human genetic, human donor tissue and clinical data over animal data are stated.

Third — and I think this matters more than the search strategy itself — that article selection was made by one author on relevance grounds without a formal quality-appraisal instrument, and that the synthesis therefore carries the selection biases inherent to narrative review. To mitigate this, claims are graded throughout by evidence type, which is the mechanism used to answer point 4.

A short Limitations section has also been added before the conclusions.

 

Point 2 — Clinical section on approved complement inhibitors

The manuscript should include a dedicated clinical section on currently approved complement inhibitors, including pegcetacoplan and avacincaptad pegol.

Response. A section, “Complement Inhibition in Current Clinical Practice”, has been added before the RNA interference material. It covers pegcetacoplan (mechanism, OAKS/DERBY, the GALE extension out to 36 months including the up to 32% reduction in lesion growth and the microperimetry finding, FDA approval, and the absence of EMA authorisation) and avacincaptad pegol (mechanism, GATHER1/GATHER2, FDA approval, and the point that GATHER2 excluded foveal-centre-involving lesions so the two programmes are not directly comparable).

Consistent with the focused scope described under point 1, I have kept this deliberately compact and have not attempted a full appraisal of trial design, health economics or real-world use, which are covered thoroughly elsewhere. What the section does do is draw out the three features of the approved class that motivate interest in nucleic acid approaches: reliance on an anatomical endpoint without demonstrated acuity benefit; indefinite monthly or every-other-month intravitreal dosing; and the class adverse effects, including new-onset exudative AMD in roughly 5–7% of monthly-treated patients and post-marketing retinal vasculitis with pegcetacoplan at approximately one per 4,000 first injections.

The section also records the field’s failures — lampalizumab in CHROMA and SPECTRI, and the discontinuation of CFI gene augmentation for futility — and states that repeated failure of genetically validated targets is a caution against inferring efficacy from mechanism, which applies to the strategies this review itself proposes.

 

Point 3 — FHL-1/CFH as one axis within a multifactorial disease

The author should present FHL-1/CFH dysregulation as a major mechanistic axis within a multifactorial disease, rather than as a fully sufficient explanation for GA pathogenesis.

Response. Accepted; the original framing was overstated. Three changes address this.

A new paragraph in the Introduction states that AMD is polygenic and multifactorial, that genome-wide analyses identify more than thirty independent loci, and that the second-largest effect — ARMS2/HTRA1 at 10q26 — lies at a locus containing no complement gene at all. Ageing, smoking, lipid handling, mitochondrial and lysosomal dysfunction in RPE, oxidative stress, matrix remodelling and choriocapillaris dropout are named as contributors whose relative weight differs between patients. The paragraph closes by stating the review’s claim in bounded form: that this axis is dominant and pharmacologically tractable in a substantial subset of patients, not that it is a sufficient explanation for GA in all of them.

Several individual over-claims have been corrected, including: “entirely dependent on” the FH/FHL-1 axis, now qualified and set alongside the other ageing-related processes at that interface; the alternative pathway described as “the primary driver” of GA, now “the pathway most strongly implicated”; “the most biologically validated, fundamentally sound strategy”, now “among the best genetically supported strategies”; and the claim that FHR deletion confers “near-total immunity to AMD”, now stated as an association with reduced risk that is partial rather than complete.

The new Limitations section states explicitly that the review does not attempt a balanced survey of non-complement mechanisms.

 

Point 4 — Distinguishing tiers of evidence for the FHRs

The FHR discussion should distinguish genetic association, circulating protein association, tissue-level evidence, and causal therapeutic evidence.

Response. This was the most useful comment in the review, and I have adopted the reviewer’s four-way distinction directly. A new subsection, “Grading the Evidence Implicating FHR Proteins”, treats each tier in turn:

• Genetic association — the CFHR3–1 deletion and rs10922109; noted as replicated and strong, but as associations at a locus where several genes are in linkage disequilibrium, so they do not identify which gene product is causal.
• Circulating protein association — elevated plasma FHR-4 (with FH not elevated), and raised FHR-1 and FHR-5 in advanced disease; noted as cross-sectional, unable to establish direction of causality, and using systemic concentration as an imperfect proxy for concentration within Bruch’s membrane.
• Tissue-level evidence — FHR-4 in choriocapillaris, Bruch’s membrane and drusen; FHR-1 in the sub-RPE space engaging mononuclear phagocytes; in vitro competition with FH/FHL-1 for C3b. Identified as the tier on which this review’s therapeutic argument principally rests, and as observational evidence from eyes with established atrophy.
• Causal and interventional evidence — stated plainly to be empty. No FHR protein has been targeted clinically in GA, no interventional data exist in any ocular indication, and there is no Mendelian randomisation analysis for FHR-4 and lesion growth.

The subsection concludes that FHR silencing is a mechanistic hypothesis supported by genetic and tissue observation rather than a validated strategy, and should be read as such in a field where lampalizumab and CFI augmentation both failed after preclinical rationale of comparable apparent strength.

This grading is also applied to all eight proteins in the new Table 1 (point 9), where the evidence column is broken out by tier so that the reader can compare the strength of support for each proposed target directly.

 

Point 5 — Reference list

The reference list appears incomplete and contains formatting/citation errors. In-text citations include numbers such as 29, 36, and 37, but the reference list shown in the manuscript ends at 26. Some references are malformed or incomplete; for example, reference 6 appears to merge different papers, reference 7 lacks normal bibliographic details, and reference 25 is a website.

Response. The reviewer is correct on every count, and checking the list revealed further errors beyond those noted. The list has been rebuilt: every entry verified against the source record, and expanded from 26 to 37.

The three broken in-text citations. Citations to 29, 36 and 37 have been remapped to real entries: 29 → ref. 27 (Fritsche, for the Y402H effect-size claim), and 36 and 37 → ref. 28 (Cipriani 2020, for FHR ligand binding and competitive displacement).

Reference 6. This had merged Grigsby et al. with Mulfaul et al., the latter also appearing separately as reference 15. The underlying problem was that reference 6 is cited throughout the text for general complement biology, whereas the entry itself was a specific research paper — the citation and the entry did not correspond. Reference 6 is now Armento, Ueffing & Clark (Cell Mol Life Sci 2021), a current review that matches how the citation is used; the duplicated Mulfaul text has been deleted, and reference 15 is retained.

Reference 7. This lacked authors entirely. It is now Grigsby et al., IOVS 2023, with full bibliographic details, and it is discussed substantively in the new augmentation section.

Reference 25. The website has been replaced by two peer-reviewed sources on dual-targeting siRNA: Belgrad et al., Nucleic Acids Res 2024 (ref. 25) and Tiemann et al., RNA 2010 (ref. 37).

Further errors corrected. “van Lookeren CM” → van Lookeren Campagne M; “Popperlaars F” → Poppelaars F; “Liebreman J” → Lieberman J; “Suahria H” → Sahrai H; “N Eng J Med” → N Engl J Med; “J Path” → J Pathol; reference 2 dated 2017 against volume 125(4), which is 2018; reference 4 lacking a year; a missing space in reference 16; a stray comma in the author list of reference 20; and inconsistent volume/issue/page formatting throughout, now standardised.

References 27–37 are new, covering the GWAS and FHR-4 literature, Bruch’s membrane compartmentalisation, and the clinical trial record for the approved and discontinued agents.

 

Point 6 — Rhetorical language

The manuscript would be much easier to read if the authors reduced rhetorical language such as “monumental,” “catastrophic,” “massive,” “unyielding,” and “fiercely pro-inflammatory.”

Response. Agreed. Approximately sixty such constructions have been revised. The specific terms cited are all gone, along with “devastating”, “staggering”, “astonishing”, “relentlessly”, “exhaustively”, “revolutionary”, “unequivocally”, “perpetual state of peril”, “spirals out of control”, “immune storm”, “shrouded”, “elegant”, “daunting”, “formidable”, “lethal”, “aggressively”, “unprecedented”, “dangerous”, and the section heading “The Extreme Danger of Direct FHL-1 Silencing”, now “Why Direct FHL-1 Silencing Is Contraindicated”.

Where a rhetorical term was carrying an evidential claim rather than merely colour, the replacement states the evidence instead of softening the adjective. For example, “unequivocally proven by sophisticated in vitro knockdown models” is now “supported by in vitro knockdown models”, and the closing assertions that FHL-1 is “the crucial shield” whose destruction “accelerates the disease” are now expressed as expectations conditional on the mechanism rather than as established fact.

 

Point 7 — Figures

The paper would benefit from one or two figures: one summarizing the FH/FHL-1/FHR architecture, and one showing the proposed therapeutic logic.

Response. Both figures have been produced to the reviewer’s specification and are supplied as separate files in addition to being embedded in the manuscript.

Figure 1 shows the domain architecture of the whole family at the RCA locus: FH with all 20 SCRs, FHL-1 with SCR 1–7 and the SFTL tail, and FHR-1 to FHR-5. Colour coding separates the regulatory module (SCR 1–4), the heparan sulfate site carrying Y402H (SCR 7), and the sialic acid recognition site (SCR 19–20), and shows which FHR domains are homologous to which FH domains. The figure is designed to make the central structural point visible at a glance: the FHRs possess the ligand-binding domains but no counterpart of SCR 1–4, which is the structural basis of competitive de-regulation.

Figure 2 shows the therapeutic logic as a two-panel contrast, from a shared premise that the pathology is regulator insufficiency rather than regulator excess. The left panel traces CFH silencing through loss of FI cofactor and decay-accelerating activity to increased MAC deposition and NF-κB activation, with the supporting knockdown evidence listed. The right panel traces CFHR1/CFHR5 and C5 silencing through to restored functional dominance of endogenous FHL-1, with the supporting genetic and tissue evidence listed. A footer states the unresolved issues: hepatic silencing does not act on locally synthesised complement, and no FHR-directed RNAi agent has entered clinical testing.

 

Point 8 — Typographical and encoding issues

There are several typographical or encoding issues in the PDF, such as broken words and strange characters in “attack” and other terms. Please correct them.

Response. I should report an unexpected finding here. On inspecting the submitted .docx at the XML level, the affected words are intact in the source file: “membrane attack complex” and “Membrane Attack Complex” appear correctly at all five occurrences, and the only non-ASCII characters in the entire document are ordinary typographic dashes, µ, κ and ü.

The corruption the reviewer saw therefore appears to have been introduced during PDF conversion rather than being present in the manuscript. The likely cause is font substitution: the document carries “Google Sans” as its East Asian font throughout, and the superscript citation markers additionally carry grey colouring at a non-matching point size — both artefacts of the manuscript having passed through Google Docs. I have flagged this to the editorial office, since it may affect other submissions processed the same way, and I would ask that the regenerated PDF be checked.

Genuine textual errors were found and corrected: “halting the progressive of geographic atrophy” → “progression”; a sentence in the alternative pathway description that ended without a full stop and without naming its product, now completed as “converting intact C3 into the biologically active hydrolysed form, C3(H2O).”; and “Epidemiological data suggests” → “suggest”. The reference-list errors are listed under point 5.

 

Point 9 — Comparative table

A table comparing FH, FHL-1, FHR-1, FHR-4, FHR-5, C3, C5, and Factor D would improve the clarity and help the reader have a more focused comparison between them.

Response. Table 1 has been added, covering all eight proteins as specified, with columns for: protein and gene; structure; intrinsic regulatory activity; principal site of action in the macula; evidence linking the protein to AMD/GA; and therapeutic targeting status.

The evidence column is deliberately structured by the four tiers requested in point 4, so that the table also serves as the summary of evidential strength across the whole review. This makes several contrasts explicit that were previously buried in the text — for instance that C3 and C5 have interventional evidence from phase 3 trials whereas FHR-1, FHR-4 and FHR-5 have none; that FHR-5 rests on weaker evidence than FHR-1 and FHR-4, which the earlier text had not distinguished; and that Factor D, despite comparable genetic and tissue support, failed in two phase 3 trials and stands as the field’s principal cautionary precedent.

I am grateful for the thoroughness of this review. Points 3, 4 and 5 in particular identified problems that materially affected the reliability of the manuscript, and I hope the revision now meets the standard required.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for taking into consideration all the recommendations and improving the manuscript.