xPeerd (AI review)

October 1, 2026

AI Review (xPeerd — DAReview)

Summary

This literature review examines Vernonia amygdalina as a candidate antimalarial agent, with particular focus on phytochemical composition and the hypothesis that some constituents may inhibit heme polymerization. The manuscript’s main strength is that it is generally cautious about overclaiming: the abstract and conclusion correctly state that direct evidence for β-hematin/hemozoin inhibition is limited, and that current support for heme-related action remains largely indirect. The review also usefully integrates ethnopharmacological context, phytochemistry, in vitro and in vivo antiplasmodial findings, and a forward-looking research agenda.

Its principal weakness is methodological and evidentiary: the review is framed around a mechanistic hypothesis, yet the manuscript itself acknowledges that only a single-concentration β-hematin assay exists for one extract and that no IC50 values were identified for fractions or isolated constituents. The narrative synthesis is clear, but the literature-search and study-selection process is only briefly described, making the review vulnerable to selection bias. Several interpretive sections rely on plausibility arguments from structural analogy rather than direct evidence. There is also some internal tension between presenting the plant as a “candidate antimalarial agent” and repeatedly showing that potency, selectivity, mechanism, and translational relevance remain insufficiently established.

Potential Major Revisions

1. Tighten the central claim and title framing to better match the actual strength of the evidence.

The title presents Vernonia amygdalina “as a Candidate Antimalarial Agent,” while the abstract concludes that “Heme binding by V. amygdalina constituents therefore remains a hypothesis” and that “Direct evidence of heme polymerization inhibition is limited to a single-concentration beta-hematin screen of one methanol leaf extract.” This is a reasonable scientific caution, but it also reveals a mismatch between the title’s confidence and the body’s evidentiary base.

The manuscript further states in the conclusion that “potent or specific inhibition has not been demonstrated” and that cytotoxicity and modest selectivity are “important caveats for any development pathway.” For a review centered on mechanism, this substantially limits the force of the “candidate agent” framing. I recommend revising the title and opening framing toward a more evidence-proportional formulation, for example emphasizing “current evidence and mechanistic uncertainty” rather than candidacy alone.

2. Strengthen review methodology and transparency of study selection.

The Methods section states that “A literature search was conducted in PubMed, Scopus, Web of Science, and Google Scholar,” that “53 records” were identified and “33 were included,” and that “The evidence is synthesized narratively, and this review was not conducted as a systematic review.” This is acceptable for a narrative review, but the current description is too sparse for reproducibility.

Specifically, the manuscript does not adequately specify:

• exact search strings,
• inclusion/exclusion criteria beyond broad source types,
• language restrictions,
• date limits applied per database,
• duplicate-removal procedures,
• criteria for deeming reports “authoritative,”
• how disagreements or borderline cases were resolved,
• whether reference-list snowballing was used.

Because the argument hinges on the claim that evidence for heme polymerization inhibition is minimal, the reader needs stronger assurance that the search process was sufficiently comprehensive to support that negative conclusion. At minimum, provide a fuller search appendix or a transparent study-selection flow description.

3. Separate clearly what is directly demonstrated from what is chemically plausible.

A recurring interpretive issue is the movement from structural plausibility to mechanistic suggestion. For example, the manuscript states that “Flavonoids are the most plausible heme-interacting class,” and that alkaloids “are candidates for heme-binding studies,” but also concedes that these possibilities “have not been tested” for the compounds actually reported from V. amygdalina.

Similarly, the introduction says that whether V. amygdalina constituents act on the hemozoin pathway “has largely been inferred from chemical structure rather than measured directly.” This is an important admission, but the manuscript would benefit from a more formal evidentiary hierarchy throughout:

• directly demonstrated in V. amygdalina,
• demonstrated in related compounds/species,
• hypothesized from structural analogy,
• speculative future directions.

Without this separation, readers may overread plausibility as evidence.

4. Address translational limitations of the antiplasmodial evidence more critically.

The manuscript does acknowledge that “Rodent P. berghei models do not replicate human P. falciparum infection, and the doses used in vivo (tens to hundreds of mg/kg) are high.” This is one of the most important limitations in the review and should be expanded. High-dose suppression in rodent models is not strong evidence of therapeutic promise unless linked to pharmacokinetics, tolerability, and chemical standardization.

The manuscript also notes that some isolates in Ejiofor et al. were weak by the WHO criteria cited in the source itself: “an IC50 above 5 µg/mL is classed as inactive,” yet the cited isolates were 10.55–12.56 µg/mL, “although the authors concluded that the findings support the traditional use of the plant.” This is exactly the kind of source-level overinterpretation that the review should interrogate more strongly. I recommend a dedicated subsection distinguishing:

• in vitro potency against human malaria parasites,
• in vivo rodent suppression,
• transmission-blocking findings,
• cytotoxicity/selectivity constraints,
• mechanistic evidence.

5. Expand the critique of safety and selectivity.

The review appropriately notes that “Cytotoxicity is the main safety concern in the primary data,” and that vernolide, vernodalol, extracts, and fractions showed “considerable cytotoxicity,” while one ethanolic extract had “a selectivity index of only 6.14.” This is not a minor caveat; it is central to assessing whether the plant is a realistic lead source.

Because the article is positioning the plant within a drug-discovery context, the safety discussion should be more analytical. For example:

• explain what selectivity thresholds would generally be considered promising;
• distinguish crude-extract toxicity from isolated-compound toxicity;
• state whether reported active concentrations overlap meaningfully with toxic ranges;
• comment on whether the same electrophilicity invoked to explain bioactivity may also plausibly underlie nonspecific toxicity.

6. Improve critical appraisal of evidence quality across cited source types.

The manuscript includes “primary research articles, reviews, and authoritative reports,” and retains “Nine earlier publications” because they were “the original source of a finding or assay.” However, not all cited evidence appears equally robust. For example, Omoregie et al. (2010) is cited as a conference abstract in the references. Some conclusions rely partly on sources of limited detail or peer-review depth.

A more rigorous review should explicitly indicate the evidentiary weight of:

• full experimental papers,
• conference abstracts,
• review articles,
• institutional reports,
• in silico-only studies.

This matters especially where the manuscript discusses “n-hexadecanoic acid” as having “multistage, multitarget potential” based on docking plus mouse data summarized in Table 1. Docking-led claims should be moderated unless experimentally validated against the proposed mechanism.

7. Add a fraud-risk / coherence note: low suspicion overall, but watch for overextension from sparse data.

I do not see strong paper-mill or fabrication signals in the manuscript itself. The review is generally coherent, field-appropriate, and notably cautious in several places, especially when stating that heme polymerization inhibition remains unproven. That said, the manuscript does show a milder risk pattern common in literature reviews: a mechanistic narrative built around very limited direct evidence. This is not fraud, but it does create a vulnerability to rhetorical overextension. The authors should guard against that by making uncertainty more visible and by assigning evidentiary weight more explicitly.

Potential Minor Revisions

1. Typographic, formatting, and style issues

• The title uses an en dash before “A Literature Review,” but the surrounding typography is slightly uneven in the extracted text. Check consistency of dash type and spacing throughout the title page.
• In the abstract and methods, the manuscript alternates between “beta-hematin” and “β-hematin.” Standardize usage, or specify one style for prose and one for formal assay naming. Compare the abstract with the Methods and later evidence sections.
• The manuscript uses both “heme polymerization” and “hemozoin formation” in close proximity. These are related but not identical expressions in every context; a short definitional clarification early in the review would improve precision.
• Taxonomic naming is mostly careful, but because the manuscript states that “The accepted name in World Flora Online is now Gymnanthemum amygdalinum,” it would help to standardize whether the synonym is introduced once and then abbreviated, rather than alternating conceptually between accepted and retained usage.
• The page transition around the botanical description appears to interrupt the sentence flow in extraction, suggesting possible formatting crowding in the original layout. Check page breaks, table placement, and orphaned headings.

2. Minor content clarifications

• In the Introduction, the statement that the 2025 World Malaria Report “was devoted to the threat of antimalarial drug resistance” is strong and should be checked for exact wording against the cited WHO document title to avoid paraphrastic overstatement.
• The sentence “These surveys rely on self-reported use and different recruitment methods, so they show patterns of use rather than efficacy” is strong and appropriate, but could be moved earlier in the ethnopharmacology section to prevent readers from inferring medicinal effectiveness from frequency of citation.
• In the section on isolates, the phrase “support the traditional use of the plant” is attributed to prior authors despite weak potency by the criteria quoted. It would help to state explicitly that ethnomedical relevance and pharmacological potency are different evidentiary categories.
• The statement that saponins, tannins, and phenolics “often produce nonspecific effects in biochemical assays” is plausible, but a citation specifically supporting that methodological caution would strengthen the point.

3. Table and presentation issues

Table 1 is useful, but it would be stronger if it added columns for:

• assay type,
• species/strain,
• potency metric,
• dose route,
• selectivity/cytotoxicity,
• evidence type for heme-related mechanism.

At present, the note that “potencies are not directly comparable across studies” is important, but the table still encourages side-by-side comparison without enough methodological normalization.

4. AI content analysis for this post-2021 manuscript

Estimated percentage of AI-generated content: 18–32%

This estimate is necessarily approximate and should not be treated as proof. My basis is stylistic rather than forensic: the prose is highly even, consistently formal, lexically polished, and structurally repetitive in a way that can be consistent with light-to-moderate AI assistance or strong editorial smoothing. At the same time, the manuscript also contains domain-specific caution, source-linked nuance, and locally tailored synthesis that argue against naive full-generation.

Highlighted sections with elevated AI-likeliness:

• The abstract, especially the compact sequence of database names, record counts, phytochemical classes, and cautious concluding formula, reads highly templated. Example: “Of 53 records identified, 33 were included… Heme binding by V. amygdalina constituents therefore remains a hypothesis.”
• The Methods section contains standard review-language scaffolding with generic phrasing such as “Titles and abstracts were screened for relevance, and full texts were read for records that appeared eligible.”
• The “Research Gaps and Future Directions” section is well organized but formulaic in progression: quantitative gap, composition–activity gap, variability gap, safety gap, computational prioritization gap. This may simply reflect competent academic writing, but it also resembles AI-assisted structuring.

Assessed epistemic impact:

• Low to moderate, if AI assistance was used only for drafting, polishing, or structural organization.
• Higher, if AI contributed unsupported claims, source mischaracterization, or mechanistic inferences. In this manuscript, the main epistemic risk is not hallucinated jargon but over-smoothing of uncertainty around sparse direct evidence. The authors should therefore manually verify every comparative claim, every mechanistic inference, and all record-count/search-method statements.

Recommendations

1. Reframe the manuscript more explicitly as a critical evidence review of a promising but unvalidated mechanistic hypothesis, rather than as an already credible candidate-agent case. This would improve title-to-content alignment.

2. Expand the Methods section substantially. Include search strings, screening logic, inclusion/exclusion rules, treatment of duplicates, and a clearer rationale for source weighting. Even for a narrative review, this will improve credibility and reproducibility.

3. Introduce an evidence-grading framework throughout the manuscript. Label findings as direct, indirect, analogous, or hypothetical. This is especially important for flavonoid-, alkaloid-, and sesquiterpene-based mechanistic discussion.

4. Add a dedicated subsection on translational barriers, integrating potency, selectivity, standardization, species-model limitations, and mechanism uncertainty. The current material exists, but it is distributed across sections and should be synthesized more critically.

5. Rework Table 1 to better distinguish assay comparability and evidence strength. Without this, readers may overinterpret heterogeneous data as cumulative proof.

6. Make the safety critique more central. The observed cytotoxicity and modest selectivity are not peripheral observations; they materially constrain the manuscript’s translational optimism.

7. Perform a final source audit for accuracy of paraphrase, consistency of terminology, and exactness of mechanistic claims. This is particularly important where the review is drawing negative conclusions from absence of evidence, because such conclusions depend heavily on search completeness and precise source interpretation.

XSci (AI review)

October 1, 2026

Referee Report

Manuscript: Vernonia amygdalina Delile (Asteraceae) as a Candidate Antimalarial Agent: Phytochemical Composition, Ethnopharmacological Use, and Heme Polymerization Inhibition Potential – A Literature Review

Recommendation: Major revision

Summary and Recommendation

This narrative review asks whether inhibition of hemozoin formation can account for the antimalarial activity attributed to Vernonia amygdalina Delile, a plant used across sub-Saharan Africa for malaria and febrile illness. Searching four databases and retaining 33 of 53 identified records, the authors describe the plant's sesquiterpene lactones, flavonoids, alkaloids, saponins, tannins, and phenolics; set out hemoglobin digestion and heme crystallisation in the digestive vacuole of Plasmodium falciparum; and tabulate ten primary studies spanning in vitro antiplasmodial testing, rodent efficacy, transmission-blocking activity, cytotoxicity, and a single cell-free β-hematin screen. Their central finding is a negative one, honestly stated: direct evidence of heme polymerization inhibition is confined to one single-concentration screen of one methanol leaf extract, no IC₅₀ values exist for fractions or isolated constituents, and the mechanism remains hypothetical.

The manuscript has real strengths. It keeps mechanism and activity conceptually distinct, insists that convergent traditional use "is not evidence of efficacy or safety," observes that a positive cell-free result "is a lead to be confirmed, not proof of mechanism," and confronts the isolates of Ejiofor et al. (2020) with the 5 µg/mL activity criterion those authors themselves quoted. The recognition that the α-methylene-γ-lactone moiety might explain cytotoxicity as readily as antiplasmodial activity is an astute piece of chemical reasoning, and these are not common virtues in reviews of this plant. The difficulty is that this critical posture rests on an evidence base that is unevenly appraised, a phytochemistry assembled entirely from secondary compilations, and a mechanistic argument neither pursued far enough nor set against its alternatives, while the human clinical literature on the plant is absent altogether. Major revision is recommended; none of what follows requires new laboratory work.

Major Concern Category 1: The Evidence Base and Its Appraisal

1.1 The search is not reproducible, yet the central claims are absence claims

The Methods occupy a single paragraph reporting the arithmetic of a systematic review while disclaiming systematic-review status. Readers receive a precise yield and a breakdown by source type, but no eligibility criteria, no database-specific search strings, no per-database yields, and no screening flow. The strategy omits Gymnanthemum amygdalinum, the accepted name the authors themselves report in the Taxonomy section, and no African regional database was searched despite the subject matter. A yield of 53 records from four databases including Google Scholar appears implausibly low. This matters because the review's headline conclusions are absence claims, which are only as strong as the search behind them. The authors should either report the strategy in full and defend the yield, or restate those conclusions as claims about the sources reviewed rather than about the literature.

1.2 Tiers of evidence are not distinguished, and potency criteria are applied unevenly

Omoregie et al. (2010) supplies three load-bearing figures, an IC₅₀ of 9.83 µg/mL, a selectivity index of 6.14, and 82.3% in vivo suppression, which recur across the abstract, four body sections, Table 1, and the Conclusion. The reference list identifies this source as a conference abstract; the body never does. Such a source cannot report assay format, replicate structure, concentration range, or the CC₅₀ underlying the selectivity index, yet the figures are given to three significant figures and tabulated alongside full research papers without annotation. The potency criterion is applied with similar unevenness: the 5 µg/mL cutoff is invoked against the Ejiofor isolates at 10.55–12.56 µg/mL but not against the Omoregie extract at 9.83 µg/mL, which sits above the same threshold yet is called "most active." Zulkefli et al. (2025) report the most potent values in the review, 1.131–1.973 µg/mL, and the only ones clearing that cutoff, yet receive one sentence, no reconciliation with the weaker P. falciparum data, and no selectivity index despite Table 1 recording that Vero cells were used, an omission that contradicts the review's own stated requirement.

1.3 The in vivo evidence carries no information bearing on internal validity

Six of the ten tabulated studies are rodent experiments supplying most of the manuscript's quantitative content, and for none does the review report group size, allocation, blinding, counting method, variance, or statistical test, while percentages are given to two decimal places that microscopy on small groups cannot support. Studies therefore cannot be weighed against one another, and this is not hypothetical: Table 1 records that the best combination of Angupale et al. (2024) achieved 91.0% chemosuppression against an artemisinin–lumefantrine control of 87.5%. Either the positive control underperformed, which would call the experiment into question, or an aqueous herbal combination matched an artemisinin-based therapy, which would be a finding of the first importance; reporting it in a parenthesis serves neither possibility. Each study's design characteristics should be extracted and appraised with a recognised risk-of-bias instrument.

1.4 The ethnobotanical evidence is summarised rather than analysed

Having articulated the right principle, that surveys "show patterns of use rather than efficacy," the review does not apply it to its own data. V. amygdalina is reported as cited 13 times by Oladeji et al. (2020) and "among the more frequently mentioned species," without the ranking: the four leading species in that survey of ninety informants were cited 42, 38, 28, and 27 times, and V. amygdalina does not appear among those ranked by the source authors' measure of perceived effectiveness, nor is their own caution about cultural factors relayed. The fidelity level of 100% from Tugume et al. (2016) is given without its denominator or any explanation that the index measures use specificity rather than effectiveness. Most importantly, the confound peculiar to this species goes unmentioned: V. amygdalina is a staple leafy vegetable, so citation frequency measures culinary ubiquity as well as perceived efficacy.

Major Concern Category 2: The Mechanistic Argument and Its Foundations

2.1 The single available data point is not interrogated

The title and framing foreground heme polymerization, yet the section justifying them discusses one experiment in a single paragraph and leaves three questions unasked. What does the inhibition index measure, on what scale, in what direction, with what variance? A reader cannot tell whether 0.12 against chloroquine's 0.17 is a meaningful gap or noise. Second, the review reports that the Vargas et al. (2011) protocol has a positive predictive value of only 50% for extracts, then does not apply that statistic to a panel in which 19 of 20 extracts were positive, which is precisely the pattern that predictive value predicts and which suggests a near-saturated assay. Third, the review notes elsewhere that saponins and tannins "often produce nonspecific effects in biochemical assays" without connecting this to the result: saponins are surfactants, and a crude methanol extract tested in a detergent-mediated crystallisation assay is a textbook setting for a mechanism-independent positive. Drawing these connections would strengthen the authors' sceptical conclusion, not weaken it.

2.2 The phytochemistry rests on secondary compilations, and two attributions need verification

Every chemical statement is referenced to Degu et al. (2024), Ugbogu et al. (2021), or Alara et al. (2017), all themselves reviews; no primary isolation paper appears anywhere. The inventory is purely qualitative, with no content or yield for any constituent in any plant part, making it impossible to judge whether luteolin and apigenin could reach the concentrations at which the comparator flavonoids were active. It also conflates analytical registers, listing oxalates, phytates, and cyanogenic glycosides, measured in food-science antinutrient screening, alongside compounds characterised by preparative isolation. Two attributions should be checked against primary sources. Cryptolepine, isocryptolepine, and neocryptolepine are the characteristic indoloquinolines of Cryptolepis sanguinolenta (Apocynaceae) and are not a known feature of the Asteraceae; if no primary isolation from V. amygdalina exists, the Alkaloids section loses its only chemical argument for alkaloids as heme-binding candidates. Separately, vernodalol has been reported as a possible solvolysis artefact of vernodalin formed during methanolic extraction, a disputed claim that nonetheless bears on the recommendation to derivatise that compound.

2.3 Computational and chromatographic claims are reported uncritically

The statement that phytol and n-hexadecanoic acid are the "most abundant phytochemicals" is sourced at second hand, through a 2021 report as cited in a 2024 paper, so the primary finding has not been consulted. Both are ubiquitous in solvent extracts of green leaf tissue, palmitic acid being a primary metabolite and phytol a chlorophyll hydrolysis product, and both are among the most frequently reported "major constituents" in gas chromatography–mass spectrometry surveys for reasons unrelated to pharmacology. More seriously, the finding that n-hexadecanoic acid forms favourable complexes with targets from four parasite life-cycle stages is reported as supportive, when a flexible amphiphilic C16 fatty acid scoring well against multiple unrelated targets exhibits the classic signature of a non-specific docking hit. Nor does the review reconcile 89.74% chemosuppression at 100 mg/kg with the fact that whole extracts achieved less at ten times that dose.

2.4 Alternative mechanisms and the resistance liability of the target are unexamined

No mechanism other than hemozoin inhibition is surveyed, although the review assembles the materials for stronger alternatives without using them. Having identified the α-methylene-γ-lactone as an electrophile that reacts readily with thiols, it does not note that the same digestive vacuole contains the falcipains, papain-family cysteine proteases whose catalysis depends on an active-site thiol and which are established antimalarial targets. A thiol-reactive Michael acceptor delivered into a compartment full of cysteine proteases is a more direct proposition than heme coordination by a neutral lactone, and would also account for the reported cytotoxicity, since mammalian cathepsins share that chemistry. Host-mediated activity is likewise unconsidered, although the work underlying the review's most-cited source also documented antioxidant and immunomodulatory effects. Finally, the review twice attributes chloroquine resistance to PfCRT, a digestive-vacuole transporter, without drawing the implication that any agent depending on high vacuolar concentrations is exposed to the mechanism that defeated the quinolines.

Major Concern Category 3: Translation, Safety, and the Research Agenda

3.1 The human clinical evidence is absent

The review moves from survey data directly to in vitro and rodent pharmacology and never addresses whether V. amygdalina has been tested in people with malaria. It has: a clinical trial of an infusion of fresh leaves in uncomplicated malaria reported an adequate clinical response at day 14 in 67% of cases, complete parasite clearance in only 32% of responders, and recrudescence in 71% of those cleared, with no significant toxicity (Challand & Willcox, 2009). The omission changes the review's evidential architecture: the Limitations section discounts the rodent data because P. berghei does not replicate human falciparum infection, yet a human dataset bearing on exactly that point exists and is unused. It also changes interpretation, since chemosuppression of 60% to 82% reads differently once one knows that a traditional preparation achieved clinical but not parasitological cure, a pattern more consistent with antipyretic or immunomodulatory action than with potent direct killing.

3.2 There is no pharmacokinetic reasoning, and the accumulation requirement is left undrawn

The manuscript contains no discussion of absorption, distribution, bioavailability, or exposure. Its only approach is the observation that in vivo doses were "tens to hundreds of mg/kg," which understates a range spanning 10 to 2,000 mg/kg and is never translated. Standard body-surface-area conversion places a 600 mg/kg mouse dose near 49 mg/kg in humans, roughly three grams of extract daily for a 60 kg adult, which at typical extraction yields represents tens of grams of dried leaf, a figure worth setting beside the decoctions the ethnobotanical sources describe. The more serious omission concerns compartmental accumulation. The review reports that bromoquine accumulates in the digestive vacuole at about 1,000 times its medium concentration (Kapishnikov et al., 2019) and infers nothing from it. Quinolines achieve this because they are diprotic weak bases trapped in an acidic compartment, and that accumulation is what makes crystal capping possible. Sesquiterpene lactones, flavone aglycones, steroid glucosides, and fatty acids are not weak bases, so a constituent active at 0.69 mg/mL in a cell-free system faces a demanding requirement inside the vacuole. This argument, built from a figure the review already cites, sharpens its own conclusion considerably.

3.3 The safety appraisal does not extend beyond cytotoxicity

The review is right to foreground the cytotoxicity reported by Abay et al. (2015) and the selectivity index of 6.14, but a plant eaten as a vegetable across a continent, proposed for febrile children, and suggested as a chloroquine adjunct requires a broader analysis. Its well-documented hypoglycaemic activity, noted twice as a traditional indication, is never connected to the fact that hypoglycaemia is a recognised and dangerous complication of falciparum malaria in young children, the population the Introduction identifies as bearing most of the mortality burden. The toxicology data also deserve a different reading: a dose-dependent fall in erythrocyte count with a concurrent dose-dependent rise in bilirubin is the laboratory signature of haemolysis, reported for a plant proposed to treat a haemolytic disease, and reassurance that values stayed within historical reference ranges is weak. Finally, the apparent potentiation of chloroquine is presented without any formal assessment of synergy and without considering that a pharmacokinetic interaction would potentiate cardiotoxicity along with efficacy.

3.4 Novelty is not established, and the research agenda is a list rather than a programme

A review must say what it adds. This one draws its phytochemical content from three recent reviews of the same plant, one of which the authors call comprehensive, without stating what those established or what question remains. The distinctive contribution is the mechanistic appraisal, which yields a null result; that is valuable, but it must be framed as such, and the title promises more than two paragraphs of mechanistic content deliver. Reframing around the actual finding, that a widely repeated assumption rests on almost no direct evidence, would produce a sharper claim. The Research Gaps section has the opposite problem, being comprehensive without being specific: five activities are listed with no sequencing, no assay performance parameters such as concentration range, replicate structure, mediator choice, or controls, and no criteria for progression or termination. Two recommendations also rest on questionable premises, since docking algorithms are parameterised for protein cavities rather than crystal faces, and a campaign against HDP is proposed without acknowledging that the protein-catalysed and membrane-interface nucleation models cited in adjacent sentences are competing accounts.

Minor Issues

1. Transcription error in Table 1. The Omoregie et al. (2010) row gives the cytotoxicity model as "V. coli cells," whereas the main text correctly identifies Vero cells and no organism of that name exists. Since the selectivity index of 6.14 is one of the review's few quantitative safety anchors, the cell line should be stated consistently.

2. Dated terminology throughout. Hemozoin is a hydrogen-bonded assembly of heme dimers linked by reciprocal iron–carboxylate bonds, not a covalent polymer, and the field now prefers "heme crystallization." The manuscript describes the dimeric structure while retaining "polymerization" in title, abstract, keywords, and headings, and omits "hemozoin" from the keywords entirely.

3. Geographical mischaracterisation. The Oladeji et al. (2020) survey is described as covering "four communities in southwestern Nigeria," but Omu Aran lies in Kwara State in the North Central zone, and the source describes a study area spanning two geopolitical zones.

4. Selectivity indices undefined and unquantified. The index is called "moderate" without being defined as a ratio of cytotoxic to antiplasmodial concentration, without a threshold for adequacy, and without CC₅₀ values for the cytotoxicity described as "considerable." Vero cells are also a weak surrogate for hepatotoxicity.

5. No dispersion estimates anywhere. Not one IC₅₀, chemosuppression percentage, or inhibition index carries a standard deviation or confidence interval, even where the sources report them. Table 1 should include these where available and indicate where they are not.

6. Reporting gaps in Table 1. Route of administration appears for only two in vivo studies; extraction yields and solvent ratios are absent, making cross-study comparisons uninterpretable; concentrations are given variously in µg/mL and mg/mL; plant provenance and voucher deposition go unrecorded although demanded of future work; and the table is split across two blocks with a duplicated header.

7. Inconsistencies in the pharmacological reporting. Text and table disagree on the Iwalokun (2008) chloroquine regimen, whose 5 mg/kg combination arms are subtherapeutic. The Babylon et al. (2026) prophylactic result of 91.67% at 200 mg/kg sits beside a treatment result of 60.78% at 600 mg/kg without noting that the designs differ, and the Abay et al. (2015) transmission percentages lack denominators and significance tests.

8. Chemical presentation and consistency. The genus is said to comprise about 1,000 species immediately before the transfer to Gymnanthemum is reported. The comparators of Abu-Lafi et al. (2020) are flavonols and a flavan-3-ol active only at 0.1 mg/mL, whereas luteolin and apigenin are flavones lacking the 3-hydroxyl generally implicated in metal coordination, so the analogy is weaker than presented.

9. Presentation and sourcing. The heading hierarchy has collapsed, leaving empty parent headings; there are no figures, although a digestive-vacuole schematic and a structure panel would aid the argument; species names appear in mathematical markup; the WHO burden figures are cited via a partner briefing rather than the report itself; and the secondary citation to Afolayan and Abdulkareem (2021) cannot be located from the reference list.

10. Preparation chemistry. V. amygdalina leaves are conventionally washed and squeezed before culinary use specifically to remove the bitter saponins and sesquiterpene lactones identified here as candidate actives, whereas medicinal decoctions retain them; whether the preparation chemistry matches that of the extracts tested should be addressed.

Missing and Inadequately Integrated Literature

Cited but Inadequately Integrated

Several sources already cited contain what is needed to resolve the concerns above. Vargas et al. (2011) supplies the 50% positive predictive value for extracts that is never carried into the interpretation of the review's only direct mechanistic evidence. Abiodun and Oladepo (2018) carries the entire mechanistic claim yet is summarised in four sentences, without the assay format, the definition and scale of the inhibition index, the controls, the replicate structure, or the criterion by which six extracts advanced to IC₅₀ determination. Combrinck et al. (2015) is recommended as the cell-based confirmation step with no account of why cellular heme fractionation discriminates genuine intracellular inhibition from cell-free artefact. Kapishnikov et al. (2019) supplies the thousand-fold accumulation figure whose weak-base implications go undrawn, and Blasco et al. (2017) the PfCRT mechanism whose cross-resistance implications are never assessed. Njan et al. (2008) is read as a null toxicology finding when its dose-dependent haematological and bilirubin changes constitute a signal, and Iwalokun (2008) is interpreted pharmacodynamically without considering a pharmacokinetic interaction. Ejiofor et al. (2020) is appraised against the authors' own activity criterion, commendably, but without noting that the material tested was stem bark whereas documented traditional use is of the leaf. Among the surveys, Oladeji et al. (2020) is used without its comparative ranking, Tugume et al. (2016) without defining the fidelity level, and Omagha et al. (2021) without the implication that single-species studies do not model the combination recipes it documents. Finally, Jani et al. (2008) and Kapishnikov et al. (2021) are cited as though complementary when they represent competing accounts of hemozoin nucleation.

Truly Missing Papers

The most consequential omission is Challand and Willcox (2009, J Altern Complement Med 15:1231–1237), the only clinical trial of V. amygdalina in uncomplicated malaria. Omoregie and Pal (2016, Avicenna J Phytomed 6:236–247) is the full peer-reviewed publication corresponding to the conference abstract on which the review depends and additionally reports antioxidant and immunomodulatory activity, resolving both the source-tier problem and the question of alternative mechanisms. Bihonegn et al. (2019, SAGE Open Med 7:2050312119849766) tested the hydromethanolic extract and its solvent fractions against P. berghei, addressing the review's own stated gap while reporting crude-extract suppression roughly half that of Babylon et al. (2026) at the same doses. Abosi and Raseroka (2003, Br J Biomed Sci 60:89–91) and Masaba (2000, Trans R Soc Trop Med Hyg 94:694–695) are the foundational in vivo and in vitro reports for this species.

For the chemistry, Kupchan et al. (1969, J Org Chem 34:3908–3911) is the original isolation of vernodalin and vernomygdin, and Jisaka et al. (1992, Biosci Biotechnol Biochem 56:845–846; 1993, 57:833–834) report plasmodicidal activity of defined constituents as well as the vernodalol artefact claim. Pagola et al. (2000, Nature 404:307–310) determined the β-haematin structure and overturned the polymer model, while Egan (2008, Mol Biochem Parasitol 157:127–136) reviews the competing nucleation hypotheses. On methodology, Page et al. (2021, BMJ 372:n71), Cos et al. (2006, J Ethnopharmacol 106:290–302), Heinrich et al. (2020, J Ethnopharmacol 246:112230), Fidock et al. (2004, Nat Rev Drug Discov 3:509–520), and Reagan-Shaw et al. (2008, FASEB J 22:659–661) supply the reporting framework, proof-of-concept and phytopharmacological standards, rodent-model criteria, and dose translation; Rosenthal (2011, Adv Exp Med Biol 712:30–48) the cysteine-protease rationale; Friedman et al. (1986, J Ethnopharmacol 16:275–287) the origin of the fidelity level index; and Carter et al. (2010, Comb Chem High Throughput Screen 13:285–292) the detergent-mediated assay whose interference liabilities bear on the single mechanistic datum. All were verified as existing publications.

Conclusion and Path Forward

This is a carefully written and intellectually honest treatment of a literature more often handled promotionally than critically. The authors deserve credit for declining to convert weak data into a confident mechanistic claim, for applying a published activity criterion to a study whose own authors did not, for insisting that selectivity indices accompany every antiplasmodial measurement, and for recognising that the reactivity of the α-methylene-γ-lactone could explain cytotoxicity as readily as antiplasmodial effect. The research question is worth asking, and a revised version of this manuscript would be a useful contribution.

The revisions required are substantial but tractable, and none demands new laboratory work. The search and appraisal methodology should be reported so that readers can reproduce it and judge the absence claims that carry the conclusions, with evidence tiers disclosed, potency and selectivity criteria applied consistently, and the internal validity of the rodent studies examined rather than assumed. The phytochemistry should be rebuilt on primary isolation reports, presented quantitatively, and cleared of attributions that may have propagated through secondary compilations. The mechanistic argument should be pursued to its conclusion, which means interrogating the single available data point, drawing the accumulation inference the review's own citations make available, and setting hemozoin inhibition alongside the alternatives the plant's chemistry suggests. The clinical evidence should be incorporated and allowed to reshape both the interpretation of the preclinical results and the proposed agenda, and the safety discussion should extend to systemic pharmacology and interaction potential. Finally, the authors should consider reframing around what they have actually established, namely that a widely repeated mechanistic assumption about this plant rests on almost no direct evidence. That is a more interesting and more defensible claim than the one the title currently advertises, and one this manuscript is well placed to make.