Authenticity, Safety, and Quality Assessment of Plant-Based Dietary Supplements Marketed for Liver Support

Abstract

Botanical dietary supplements are widely used, and many are marketed for liver support. Yet product-specific prevalence data are limited, and composition, standardisation, authenticity, and safety vary substantially. This focused narrative review integrates clinical, regulatory, product quality, and analytical evidence, using milk thistle, artichoke leaf, turmeric, and green tea as case studies. Unlike ingredient-centred reviews, it links clinical interpretation to product identity, formulation, batch characterisation, and analytical decision-making. Searches updated to 3 August 2026 prioritised controlled human studies, primary regulatory sources, and fit-for-purpose analytical literature. The selected evidence was dominated by small or short trials, heterogeneous formulations, surrogate outcomes, and incomplete lot characterisation. It did not establish disease-modifying efficacy for any reviewed preparation. Bioavailability-enhanced turmeric products and concentrated green tea extracts show that plausible benefits can coexist with liver-injury risk. Product substitution, inaccurate labelling, contamination, undeclared co-ingredients, and microbiological non-conformity can further alter exposure. Because no single method is sufficient, complementary chemical, microbiological, molecular, and palynological analyses are integrated into a staged, qualitative, risk-based workflow, from product definition through confirmatory testing and action. Product- and lot-specific characterisation, transparent trial reporting, adverse-event investigation, and surveillance are required before broad liver-support claims can be sustained.

1. Introduction

Dietary supplements are widely used and often self-selected. Because “natural” products may be perceived as inherently safe, adverse effects, interactions, and variable product quality can be underestimated [1,2,3].

Dietary supplements are generally intended to complement the usual diet by providing concentrated sources of nutrients or other substances with nutritional or physiological effects. In the United States, the Food and Drug Administration (FDA) defines dietary supplements as products intended for ingestion that contain one or more dietary ingredients. In contrast, the European Food Safety Authority (EFSA) describes food supplements as concentrated sources of nutrients or other substances with nutritional or physiological effects [4,5]. These products are marketed in diverse forms, including tablets, capsules, powders, and liquids, and may contain vitamins, minerals, amino acids, fatty acids, plant materials, or herbal extracts [1,4,5].

Within this broader market, products marketed for liver support commonly include milk thistle (Silybum marianum), artichoke (Cynara cardunculus), turmeric (Curcuma longa), green tea extract (Camellia sinensis), and other botanical ingredients associated with antioxidant, anti-inflammatory, choleretic, or metabolic claims [6,7,8,9,10]. However, the pharmacological relevance, composition, and clinical evidence for these products vary substantially by ingredient and formulation.

Nationally representative United States data from 2017–March 2020 estimated that 57.6% of adults had used a herbal or dietary supplement in the preceding 30 days and that 4.7% had used at least one of six botanicals selected for potential hepatotoxicity [11]. A six-country European consumer survey reported plant-based food supplement use among 18.8% of screened respondents in the preceding 12 months, but city-based and convenience sampling limit population inference [12]. Neither source estimates the prevalence of products marketed specifically for liver support, for which comparable global data remain unavailable.

The quality of dietary supplements may not consistently match the label claims. Inadequate levels of declared ingredients, botanical substitution, contamination, and undeclared pharmacologically active substances can compromise product quality and increase the risk of adverse effects or clinically significant interactions [13,14].

This review uses milk thistle, artichoke leaf, turmeric, and green tea extract as complementary rather than exhaustive case studies. Each illustrates a distinct product-level problem. Milk thistle highlights the gap between the extensive ingredient literature and product standardisation. Artichoke illustrates the limits of short-term surrogate evidence, turmeric shows formulation-dependent exposure and liver injury, and green tea distinguishes beverage preparations from concentrated extracts. The resulting product-risk framework applies to the wider category of plant-based supplements marketed for liver support.

This review examines how identity, formulation, dose, bioavailability, endpoint selection, and safety affect the clinical interpretation of evidence. It proposes a transferable product-level framework that integrates chemical, microbiological, molecular, palynological, documentary, and supply-chain data. By combining controlled human evidence with formulation and batch characterisation, authenticity assessment, safety surveillance, and fit-for-purpose analytical testing, it bridges ingredient-level findings with decisions on finished products.

Review Design and Evidence-Selection Approach

This focused narrative review comprised an initial conceptual synthesis and a targeted update completed on 3 August 2026. PubMed/MEDLINE and Europe PMC were searched from inception. The update also used primary sources from the European Medicines Agency (EMA), European Food Safety Authority (EFSA), United States Food and Drug Administration (FDA), European Directorate for the Quality of Medicines and HealthCare (EDQM), World Health Organisation (WHO), European Commission’s Rapid Alert System for Food and Feed (RASFF), and EUR-Lex. Publisher and digital object identifier (DOI) records were used to verify the metadata. Searching from database inception was intended to capture foundational clinical trials, regulatory assessments, and analytical standards, whereas the focused 2024–2026 update targeted recent safety signals, regulatory developments, emerging contaminants, and authentication technologies.

Searches covered: (i) controlled human and safety evidence for the four case studies; (ii) identity, standardisation, adulteration, contamination, and supplement-associated liver injury; and (iii) chemical, microbiological, molecular, and traceability methods. Botanical and ingredient terms were combined with concepts related to liver disease, trials, safety, supplements, authentication, adulteration, contaminants, chromatography, mass spectrometry, infrared spectroscopy, nuclear magnetic resonance, molecular authentication, microbiology, and palynology. Reference lists were checked.

Botanicals were selected purposively as those used in products marketed for liver support, that have relevant human clinical or safety evidence, and that illustrate a distinct product-level problem: incomplete standardisation, reliance on surrogate outcomes, formulation-dependent exposure, or extract-associated liver injury. Sources were included when they directly informed controlled human efficacy or safety, regulatory status or surveillance, product identity, adulteration or contamination, or fit-for-purpose analytical assessment. Duplicate publications, promotional sources, evidence unrelated to the selected cases or wider liver-support product category, and analytical papers without a defined matrix, target, or decision question were excluded. Mechanistic evidence was used to establish biological plausibility and not as evidence of clinical efficacy.

Controlled human studies were prioritised when preparation, dose, comparator, duration, and liver-related outcomes were extractable. Regulatory assessments and surveillance informed safety, claims, and quality, and method papers were retained when they addressed a defined analytical question. Biochemical and imaging surrogate outcomes, histological endpoints, and hard clinical outcomes were distinguished. Table S1 is a purposive comparison of ten reports, not a systematic inventory. Table S2 records source-specific roles. The reports were selected to represent all four botanicals, materially different formulations and doses, contrasting endpoint levels, and both efficacy and safety roles; selection was not based on the direction or statistical significance of the findings. Other clinical evidence is considered through the cited evidence syntheses and regulatory assessments rather than being exhaustively tabulated. No protocol was preregistered, and no duplicate screening, quantitative pooling, formal risk-of-bias assessment, or certainty grading was performed, limiting completeness and causal inference.

2. Plant-Derived Ingredients in Liver-Support Supplements

2.1. Terminology and Interpretation of Clinical Evidence

Current European clinical practice guidelines use the terms metabolic dysfunction-associated steatotic liver disease (MASLD) and metabolic dysfunction-associated steatohepatitis (MASH) in place of non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH) within the broader steatotic liver disease nomenclature [15]. MASLD requires hepatic steatosis in the presence of at least one cardiometabolic risk factor and is therefore not fully interchangeable with the historical NAFLD definition at the individual-patient level [15]. In this review, the original diagnostic terminology from earlier clinical studies is retained when describing study populations, eligibility criteria, or reported outcomes. However, the findings are interpreted within the current MASLD/MASH framework to facilitate comparison with contemporary clinical and regulatory literature. This approach recognises both the continuity of the evidence base and the differences introduced by the updated diagnostic criteria.

Outcomes were grouped into the biochemical markers alanine aminotransferase (ALT) and aspartate aminotransferase (AST), imaging measures, histology, and hard clinical events. Biochemical and imaging changes are surrogate signals; histology more directly assesses disease activity and fibrosis, whereas decompensation, transplantation, hepatocellular carcinoma, and liver-related mortality are most consequential but were not established by the reviewed studies. MASLD may coexist with normal aminotransferases despite clinically significant steatohepatitis or advanced fibrosis [15]; improvement in ALT or AST alone cannot validate a hepatoprotective claim.

2.2. Milk Thistle (Silybum marianum) and Silymarin

2.2.1. Composition, Bioavailability, and Evidence Boundaries

Silymarin, obtained from the fruit of Silybum marianum, is a mixture dominated by silybin A and B, isosilybin A and B, silychristin, and silydianin; taxifolin is a flavonoid rather than a flavonolignan [7,16,17,18]. Commercial preparations differ in plant material, extraction, drug-to-extract ratio, flavonolignan profile, and declared content, so the designation ‘milk thistle’ alone does not establish equivalence to a standardised medicinal extract [17,18].

Proposed mechanisms include free-radical scavenging, inhibition of lipid peroxidation, modulation of inflammatory signalling, and antifibrotic effects observed mainly in experimental systems [7,16]. These mechanisms support biological plausibility but do not establish clinical efficacy or equivalence among finished products.

Oral exposure to silymarin, particularly silybin, is limited by poor solubility, incomplete absorption, and presystemic metabolism. Phospholipid or lipid systems, solubilising agents, and particle-size reduction may increase exposure, but have not been shown to ensure greater clinical benefit [16,19,20]. Product comparison therefore requires formulation and batch-specific compositional data, not marker content alone [17,18,19,20].

Evidence syntheses report modest, heterogeneous biochemical and metabolic effects [15,21]. In two 48-week histology-based randomised trials, silymarin did not meet the primary endpoint: a ≥30% NAFLD activity score reduction in 99 adults receiving 700 mg three times daily, or a ≥2-point improvement in 78 participants receiving 420 or 700 mg three times daily [22,23]. Modest samples, incomplete batch characterisation, and the absence of hard clinical outcomes limit inference from these studies; neither trial established disease-modifying efficacy, and applicability to commercial supplements remains undemonstrated [17,21,22,23].

2.2.2. Safety Considerations and Product-Related Variability

Conventional milk-thistle preparations are generally well-tolerated; adverse effects are mainly mild gastrointestinal symptoms, headache, or allergic reactions, and Asteraceae hypersensitivity is a contraindication [3,24,25]. Human evidence for clinically important interactions is limited despite in vitro effects on enzymes and transporters. However, medication review remains prudent for multi-ingredient products, polypharmacy, or medicines with a narrow therapeutic index [19,25]. Safety has not been established during pregnancy, lactation, or in children and adolescents [25].

Ingredient-level tolerability does not cover hazards introduced by variable composition, adulteration, inaccurate labelling, pesticide or mycotoxin residues, or microbial contamination. Safety assessment must address the final product and lot [18,24].

2.3. Artichoke Leaf Extracts (Cynara cardunculus L. [Syn. Cynara scolymus L.])

2.3.1. Botanical Identity, Chemical Composition, and Standardisation

Medicinal artichoke leaf is derived from Cynara cardunculus L. (syn. Cynara scolymus L.); marketed products may use either synonym, while the European monograph concerns the leaf rather than the edible flower head or other plant parts [8,26,27].

Caffeoylquinic acids, including chlorogenic acid and cynarin (also spelt “cynarine”), luteolin glycosides, and sesquiterpene lactones are prominent constituents. However, no marker represents the whole preparation [8,27]. Source, harvest, plant part, extraction, drug-to-extract ratio, and standardisation alter composition; pharmacopoeial requirements for chlorogenic acid content in dried leaf do not establish equivalence among commercial extracts [26,27].

2.3.2. Bioavailability, Claimed Effects, and Clinical Evidence Boundaries

Caffeoylquinic acids and flavonoids are absorbed and extensively metabolised [28]. Antioxidant, lipid-modulating, and choleretic actions have been proposed [8,26,27,28], but links between exposure to constituents or metabolites and meaningful liver outcomes remain uncertain. These mechanisms therefore provide plausibility rather than proof of clinical hepatoprotection; EMA recognition for traditional relief of dyspeptic symptoms is not an indication for chronic liver disease [26,27].

Short trials using 600, 2600, or 2700 mg/day reported changes in aminotransferases, lipids, controlled attenuation parameter, or ultrasound measures [29,30,31], and a meta-analysis suggested reductions in ALT and AST [32]. Preparations, populations, comparators, and endpoints differ; follow-up is brief, and the artichoke studies reviewed here provide neither histological nor hard clinical evidence. In the prebariatric SteatoChoke pilot, imaging improved, but AST increased, and a liver-reduction diet was introduced. These findings neither establish disease modification nor product interchangeability [29,30,31,32].

2.3.3. Safety, Biliary Contraindications, and Product Variability

Artichoke leaf is generally well-tolerated, with mainly gastrointestinal or allergic reactions; Asteraceae hypersensitivity is a contraindication [26,27]. Due to choleretic action, medicinal preparations are contraindicated in bile-duct obstruction, cholangitis, gallstones, and hepatobiliary disorders requiring supervision. Interaction data are limited, and safety has not been established during pregnancy, lactation, or in children under 12 years [26,27]. Clinical suitability also depends on identity, extraction, drug-to-extract ratio, marker standardisation, storage, co-ingredients, and the consumer’s hepatobiliary and medication context [26,27].

2.4. Turmeric (Curcuma longa)

2.4.1. Botanical Identity, Curcuminoid Composition, and Standardisation

Turmeric products range from powdered or comminuted Curcuma longa rhizome and tinctures covered by the European monograph to dry extracts, curcuminoid-enriched preparations, isolated curcumin, and multi-ingredient formulations [9,33,34]. Principal constituents include curcumin, demethoxycurcumin, bisdemethoxycurcumin, and volatile turmerones [9,35,36,37]. Experimental antioxidant and anti-inflammatory activities do not make these formulations chemically or clinically interchangeable.

Cultivar, processing, extraction, drug-to-extract ratio, marker standardisation, dosage form, and bioavailability enhancers affect composition and exposure; EMA-reviewed data show batch variability in curcuminoids [33,34,38]. Artificial dyes, undeclared diluents, and synthetic curcumin have been documented, with adulteration patterns varying by product category and geographic market [39]. A turmeric or curcumin label therefore establishes neither equivalence nor product safety.

2.4.2. Bioavailability, Claimed Effects, and Clinical Evidence Boundaries

Conventional oral curcumin has low bioavailability because of poor solubility, limited absorption, instability, and rapid metabolism. Experimental studies attribute possible effects to the modulation of oxidative and inflammatory pathways, but these pleiotropic mechanisms have not been validated as product-level mediators of liver benefit [9,35,36,37,38]. Phospholipid, micellar, lipid, nanoparticle, and piperine-containing systems can increase exposure but may also alter interaction and safety profiles without ensuring greater clinical benefit [34,38].

An 8-week amorphous-dispersion trial reported improvements in ultrasound liver fat, aminotransferases, and metabolic measures [40]. A 12-month diabetes/MASLD trial reported improved inflammatory and FibroScan measures, but tumour necrosis factor (TNF) was the primary endpoint, and hepatic measures were secondary [41]; a meta-analysis suggests heterogeneous reductions in ALT and AST [42]. Differences in formulation, dose, co-ingredients, and endpoints preclude product-level generalisation, and neither selected turmeric trial assessed histology or hard clinical outcomes [15,40,41,42].

2.4.3. Safety, Hepatotoxicity, and Product Variability

The EMA recognises turmeric rhizome for the traditional relief of digestive symptoms, not chronic liver disease [33]. Conventional preparations mainly cause mild gastrointestinal effects; hypersensitivity is a contraindication, and safety has not been established during pregnancy or lactation or in people younger than 18 years. Choleretic activity also warrants avoidance in bile-duct obstruction, cholangitis, gallstones, or hepatobiliary disease requiring supervision [33].

Rare but potentially severe hepatocellular injury has been reported with turmeric or curcumin supplements. The final EMA addendum noted that causality could not be established because reported cases commonly involved concomitant medicines or supplements and incomplete information on dose or formulation. Nevertheless, the available data suggest a hepatotoxicity risk with high-bioavailability formulations, high doses of curcumin or curcuminoids, or both; no liver toxicity had been reported at the posologies listed in the EU herbal monograph [34]. Drug-Induced Liver Injury Network cases typically appeared within one to four months, were associated with human leukocyte antigen (HLA)-B*35:01, and included acute liver failure [43].

Risk assessment requires the documentation of curcuminoid dose, bioavailability enhancers, co-ingredients, concomitant medication, and authenticity [34,39]. For products expected to produce high exposure, clear disclosure of the daily curcuminoid dose, formulation, and bioavailability enhancers would support risk communication. Suspected injury warrants discontinuation and evaluation. Routine monitoring is not supported for all consumers. However, trials and the supervised use of high-exposure products require prespecified liver-safety monitoring, particularly for people with pre-existing liver disease or taking potentially hepatotoxic medication [34,43].

2.5. Green Tea Extract (Camellia sinensis)

2.5.1. Composition and Standardisation

Green tea from Camellia sinensis contains catechins, particularly epigallocatechin-3-gallate (EGCG). Conventional beverages and concentrated extracts differ in catechin content, extraction, formulation, co-ingredients, and daily exposure and require separate assessment [10,44].

Catechins can influence oxidative, inflammatory, and metabolic signalling in experimental models; however, these mechanisms do not predict efficacy or safety across beverage and extract matrices. Exposure to epigallocatechin-3-gallate, fasting conditions, formulation, and host susceptibility may all modify the clinical response [10,44].

2.5.2. Evidence Boundaries

Two small controlled studies reported short-term biochemical or imaging signals: a three-arm beverage trial assigned 17 participants to high-catechin, low-catechin, or placebo beverages, and an extract trial included 80 participants [45,46]. Neither assessed histology or long-term outcomes; incomplete batch characterisation and different catechin matrices prevent generalisation across infusions, extracts, and multi-ingredient supplements [10,44,45,46].

2.5.3. Hepatotoxicity and Product Risk

Rare but clinically significant hepatocellular injury has been linked to concentrated green-tea extracts [10,44,47,48]. In the Minnesota Green Tea Trial, 843 mg/day epigallocatechin-3-gallate increased ALT and AST; 26 out of 513 extract recipients developed grade ≥2 liver-test abnormalities, with dechallenge–rechallenge evidence [47]. EFSA could not identify a universally safe extract dose and considered ≥800 mg/day concerning, although injury has occurred at lower exposure [10]. Association with HLA-B*35:01 supports an idiosyncratic component [48]. Dose, formulation, chemical characterisation, and surveillance are therefore central [10,44,47,48].

Beverages and extracts are also not interchangeable exposures when assessing interactions. In healthy volunteers, a specific aqueous green-tea extract markedly reduced fexofenadine exposure, supporting an interaction specific to the formulation and substrate rather than a universal class effect [49]. Green-tea use, therefore, warrants documentation and review when reduced medicine exposure could compromise treatment.

Supplementary Table S1 provides a purposive, rather than exhaustive, comparison of ten controlled human study reports across the four botanicals, including efficacy and safety roles, endpoint level, product characterisation, and principal limitations; its principal cross-case implications are synthesised in Section 2.6 and Section 3.4.

2.6. Cross-Case Synthesis: From Ingredient-Level Evidence to Product-Level Claims

The four cases expose different limits. Among the selected studies, milk thistle is the only intervention with histology-based trials, neither of which met its primary endpoint [22,23]. Artichoke remains supported mainly by short-term surrogates [29,30,31,32], whereas turmeric findings are formulation-dependent and accompanied by rare liver injury [34,40,41,42,43]. Green tea highlights the distinction between beverage matrices and concentrated-extract risk [10,44,45,46,47]. Across Table S1, biochemical and imaging surrogate outcomes predominate; both histological silymarin trials were negative on their primary endpoints, and none of the selected reports established benefit for decompensation, transplantation, hepatocellular carcinoma, or mortality.

Ingredient names and nominal product doses cannot substitute for product characterisation. Across the studies summarised in Table S1, no consistent cross-study dose–efficacy relationship can be inferred. Nominal product doses did not represent equivalent phytochemical exposures because plant part, extraction, drug-to-extract ratio, marker content, formulation, bioavailability enhancers, co-ingredients, study populations, treatment durations, and endpoint selection differed across studies [18,27,34,39,44]. Among the selected reports, the only within-study comparison of two silymarin doses did not show a significant dose-dependent improvement in the primary endpoint: response occurred in 19% of participants receiving 420 mg three times daily, 15% receiving 700 mg three times daily, and 12% receiving placebo (p = 0.79) [23]. No selected report provided a complete lot-specific chemical fingerprint; equivalence to a studied preparation must therefore be demonstrated rather than assumed.

3. Product-Level Authenticity, Safety, and Analytical Assessment of Plant-Based Liver-Support Supplements

3.1. Authenticity, Safety Risks, and Regulatory Context

The following sections apply these case-study lessons to the wider category of plant-based supplements marketed for liver support. Unless a botanical is named, the risks concern products rather than only the four ingredients.

3.1.1. Adulteration, Mislabelling, Undeclared Pharmaceuticals, and Botanical Substitution

Relevant non-conformities include undeclared drugs, substituted or uncertain botanicals, inaccurate quantities, poorly standardised extracts, undisclosed co-ingredients, and incomplete labels [14,18,39,50,51,52].

Table 1. Illustrative quality non-conformities across selected supplement categories.

Supplement Category	Relevance to Liver-Support Assessment	Illustrative Non-Conformities and Evidence
Selected botanicals used in liver-support supplements	Directly relevant to the case studies	Botanical substitution [53]; variable marker content and incomplete labelling [18]; synthetic curcumin, artificial dyes, Pb-containing adulterants, and other undeclared materials in turmeric-derived products [39]; As contamination in botanical supplements [13].
Weight-loss supplements	Contextual comparator	Sibutramine pharmacology [54]; undeclared sibutramine [52,55] and other appetite suppressants, stimulants, or laxatives [14,52].
Sexual-performance supplements	Contextual comparator	Phosphodiesterase type 5 inhibitors and analogues [14,50,52,56].
Bodybuilding supplements	Contextual comparator	Anabolic steroids, prohormones, or stimulants [14,50,52].

Note: Weight-loss, sexual-performance, and bodybuilding supplements are contextual comparators only. These entries illustrate adulteration types and relevant analytical capabilities; they do not estimate adulteration prevalence in liver-support products.

distinguishes evidence from botanical products relevant to the case studies from evidence in contextual comparator categories.

Sibutramine is an anti-obesity medicine [54]; when undeclared in weight-loss supplements, it illustrates intentional pharmaceutical addition [14,52,55]. Botanical substitution may instead reflect misidentification, admixture, economic fraud, or inadequate authentication [53]. Both can alter exposure, interaction risk, and the interpretation of clinical or adverse-event data.

No recurrent pattern of pharmaceutical adulteration has been established specifically for liver-support products. A 2024 workflow based on high-resolution mass spectrometry detected illegal drugs in 41 out of 50 selected batches of suspicious functional foods [56]. Because this sample was not specific to liver-support products, it demonstrates analytical capability but does not estimate prevalence. Current concerns more often involve identity, label accuracy, marker variability, co-ingredients, and contamination [14,50,52]. Finished-product specifications need to cover identity, purity, strength, composition, and contaminants, supported by documented manufacturing controls [51].

3.1.2. Raw-Material Variability, Supply-Chain Control, and Traceability

Raw-material identity, origin, cultivation, harvest, processing, extraction, and storage influence chemical and microbiological quality. Supplier qualification combines authenticated identity, acceptance specifications, and records linking source material, certificates of analysis, processing, batch, and final product [51,53,57,58,59,60]. The final 2025 European Medicines Agency Revision 1 guideline on good agricultural and collection practice (GACP) provides a useful benchmark for herbal starting materials. Its stated scope is herbal substances used to prepare herbal medicinal products rather than food supplements generally [61].

3.1.3. Chemical Contaminants, Residues, and Environmental Exposure

Cultivation, processing, and storage can introduce metals and metalloids, pesticide residues, mycotoxins, residual solvents, and toxic plant constituents [13,62,63,64]. A 2024 survey measured 79 natural contaminants, including mycotoxins and pyrrolizidine and tropane alkaloids, with totals up to 5.3 mg/kg. Because the sample was not specific to liver-support products, the findings illustrate the breadth of possible contamination rather than global prevalence [65]. Testing needs to reflect the source, extraction, dose, and formulation; combine specifications and targeted panels; and confirm emerging hazards when surveillance or unexpected signals justify it [51,62,63,64]. Applicable EU maximum contaminant and pesticide-residue levels define legal constraints for the relevant food category but do not replace product-specific, risk-based monitoring [63,64].

3.1.4. Microbiological Quality and Biological Contamination

Bacteria, yeasts, and moulds may arise from raw material, processing, storage, or handling. Effective controls integrate hygienic production, drying, storage, packaging, representative sampling, viable counts, and specified-organism testing appropriate to product and consumer risk [57,58,59,60,62]. Culture assesses viability; mycotoxins require separate chemical analysis even when fungal contamination is suspected [58,59,60,62].

3.1.5. Product-Related Adverse Effects, Herb–Drug Interactions, and Causality Assessment

Botanical supplements can cause gastrointestinal or allergic reactions, interactions, cardiovascular or neurological effects, and hepatotoxicity [3,24,66]. Attribution is difficult in multi-ingredient products because the cause may be the declared botanical, a concentrated extract, a co-ingredient, an adulterant, a contaminant, or a combined exposure.

Suspected supplement-associated liver injury is a diagnosis of exclusion. Clinical management involves stopping non-essential suspect products and documenting the product, manufacturer, lot, formulation, dose, timing, concomitant medication, other supplements, and alcohol. Laboratory assessment includes ALT, AST, alkaline phosphatase (ALP), bilirubin, and international normalised ratio (INR). Calculate the R value; exclude viral, autoimmune, ischaemic, and biliary alternatives as indicated; and follow the course after withdrawal [24,66,67]. Retaining the container and the remaining product from the consumed lot, along with chain-of-custody records, permits targeted analysis, causality assessment, reporting, and possible recall [50,51,66,68].

Older age alone is not a universal susceptibility factor [67], but explicit medication and supplement reconciliation is especially important because older adults often use medicines and supplements together [69]. Pregnancy, lactation, childhood, pre-existing hepatobiliary disease, and polypharmacy warrant product-specific precaution because safety and interaction data are often incomplete [25,26,33,67]. Human data on green tea further show that interaction findings are specific to the preparation and substrate, not a universal class effect [49].

Table 2. Practical investigation of a suspected adverse event associated with a botanical supplement.

Step	Minimum Action	Escalation or Decision
1. Stabilise	Stop non-essential suspect products; assess symptoms, jaundice, encephalopathy, bleeding, and haemodynamic instability [67].	Urgent specialist or hospital assessment for acute liver failure, rising INR, severe jaundice, or rapid deterioration [67].
2. Reconstruct exposure	Record the exact product, manufacturer, lot, formulation, daily dose, start/stop dates, prior exposure, all medicines/supplements, and alcohol [24,66,67].	Build a dated exposure–event timeline; do not infer identity from the front label alone [24,66].
3. Characterise injury	Measure ALT, AST, ALP, bilirubin, and INR; calculate R value; exclude viral, autoimmune, ischaemic, and biliary causes as indicated [67].	Classify pattern and severity; trend tests after withdrawal; avoid intentional rechallenge in routine care [67].
4. Retain evidence	Secure the consumed container and remaining lot; photograph the label and record storage and chain of custody [50,51,66,68].	Compare the retained product with the declared composition, supplier records, and, where possible, an unopened comparator lot [50,51,66].
5. Test by hypothesis	Select identity, adulterant, contaminant, microbiological, or non-targeted analyses from the suspected hazard and matrix [50,57,62,70].	Confirm consequential screening signals with orthogonal, quantitative, and validated methods [50,57,62,70].
6. Integrate and report	Combine clinical course, alternative causes, analytical findings, formulation, and supply-chain evidence [50,51,66,67].	Report serious events; consider targeted surveillance, withdrawal, or recall; state residual uncertainty [66,67,68].

summarises a practical six-step investigation.

3.1.6. Regulatory Frameworks, Health Claims, and Post-Market Surveillance

In the European Union, Directive 2002/46/EC governs food supplements, Regulation (EU) No. 1169/2011 sets general food-information requirements, and national approaches to botanicals remain partly heterogeneous [71,72,73]. Marketing terms such as ‘liver support’ or ‘detoxification’ may imply a health benefit and should therefore be assessed case by case under Regulation (EC) No. 1924/2006 and the applicable food-information rules [74].

RASFF supports the European exchange of food-risk alerts and coordinated withdrawal or recall [75]. United States supplements are not approved pre-market as medicines; manufacturers remain responsible for the quality and reporting of serious adverse events [4,51,68]. Under Section 413(a)(2) of the Federal Food, Drug, and Cosmetic Act, when the statutory criteria apply, a manufacturer or distributor must submit a new dietary ingredient notification at least 75 days before introducing the product into interstate commerce. The FDA’s March 2024 final guidance describes the procedure and contains non-binding recommendations except where it cites statutory or regulatory requirements. Even an acknowledgement without objection is not an independent FDA finding that the ingredient or supplement is safe or not adulterated [76].

3.2. Complementary Analytical Approaches for Quality Assessment and Traceability

These methods form a transferable, risk-based framework rather than a mandatory panel. In this review, risk-based denotes qualitative prioritisation of analytical effort according to the suspected hazard, product matrix and exposure, the plausibility or likelihood suggested by the label, supply chain, or surveillance information, and the clinical or regulatory consequence of an incorrect decision. The workflow is conceptual and does not constitute a validated quantitative risk score. The decision question—identity, composition, contamination, exposure, microbiological quality, or traceability—should determine sampling and method selection. No single technique resolves all attributes. Representative sampling, lot identification, and documented chain of custody are essential when the results may affect quality, safety, or causality conclusions [50,53,57,59,62,70,77]. Applied to the four case studies, this logic targets silymarin flavonolignans, artichoke caffeoylquinic acids and flavonoids, turmeric curcuminoids and potential synthetic colourants, or green-tea catechins. Orthogonal identity and contaminant tests are then selected according to the matrix and suspected non-conformity [17,18,26,27,34,39,44]. Figure 1 summarises the workflow.

3.2.1. Chemical Profiling, Marker Quantification, and Contaminant Screening

High-performance thin-layer chromatography (HPTLC), high-performance liquid chromatography (HPLC) with diode-array detection (HPLC-DAD), and ultra-performance liquid chromatography (UPLC) fingerprints are used to compare authenticated materials and lots. At the same time, targeted assays quantify markers [78,79]. HPLC-DAD, HPLC with ultraviolet (UV) detection, and liquid chromatography-tandem mass spectrometry (LC-MS/MS) address predefined compounds. Liquid chromatography-high-resolution mass spectrometry (LC-HRMS) adds suspect and non-targeted screening [50,70,79]. Examples include illegal-drug detection [56] and an LC-Orbitrap-HRAMS study that combined untargeted metabolomics with machine learning to classify geographical origin, variety, tissue identity, and adulteration in turmeric and ashwagandha samples [80]. Gas chromatography-mass spectrometry (GC-MS) is preferable for volatile or selected non-polar analytes. Solution-state nuclear magnetic resonance (NMR), including quantitative NMR (qNMR), provides structure-rich fingerprints and marker measurements but has lower trace sensitivity than liquid chromatography-mass spectrometry (LC-MS) and does not independently establish species identity [70,81]. Accurate mass or a library match should remain tentative; consequential identities should, where feasible, be confirmed against a reference standard using concordant retention time, isotope pattern, and diagnostic MS/MS fragmentation, with validation aligned to the intended decision [50,56].

Contaminant testing follows source and formulation risk. Inductively coupled plasma mass spectrometry (ICP-MS) or inductively coupled plasma optical emission spectrometry (ICP-OES) can quantify elements. In contrast, LC-MS/MS or gas chromatography-tandem mass spectrometry (GC-MS/MS) can assess pesticides. Multi-analyte LC-MS/MS can measure mycotoxins and relevant plant alkaloids, and headspace gas chromatography with flame ionisation detection (GC-FID) or GC-MS can assess residual solvents [13,50,62,63,64,65,82]. Complex matrices require the validation of recovery, precision, selectivity, matrix effects, and reporting limits. Panels should be revised when suppliers, surveillance, or adverse events reveal new hazards [62,82,83].

3.2.2. Spectroscopic and Molecular Authentication

Fourier transform infrared spectroscopy (FTIR) provides rapid, non-destructive fingerprints with little preparation. It can screen for lot similarity or atypical products when suitable references and validated chemometric models are available [79,83,84,85,86]. Overlapping bands, moisture, particle size, presentation, and preprocessing can change classification; representative training sets, predefined criteria, and external validation are therefore required. FTIR is a comparative screening tool, not a substitute for confirmatory identification or quantification [79,83,84,85,86].

Deoxyribonucleic acid (DNA) barcoding identifies individual botanicals, whereas metabarcoding surveys species in powders, capsules, or mixtures without diagnostic morphology. Performance depends on amplifiable DNA, extraction, curated databases, and contamination controls [53,77,87,88,89]. Results are not inherently quantitative, do not measure active constituents, and may fail after intensive processing or detect trace contaminants and excipients. Interpretation therefore combines DNA evidence with chemical fingerprints and marker assays [53,77,87,88,89].

3.2.3. Microbiological and Trace-Evidence Assessment

Culture-based methods estimate viable loads and investigate specified organisms; test design must reflect material, processing, intended use, consumer vulnerability, and sampling [57,58,59,60]. Polymerase chain reaction (PCR) and quantitative PCR (qPCR) can identify isolates or targets, but DNA detection does not establish viability. Molecular assays complement culture when viable counts or microbiological limits are required [57,58,59,60].

Palynology examines pollen and spores as trace evidence in raw, powdered, tea, capsule, and minimally processed mixed products. It can support questions about botanical association, unexpected plant material, environmental context, or processing contamination, but does not independently prove authenticity or provenance [90,91,92,93,94,95]. Specialist expertise, comparative material, representative sampling, and explicit uncertainty are required; low abundance, mixed assemblages, contamination, processing loss, and limited standardisation constrain attribution [90,91,92,93,94,95]. Conversely, the absence of diagnostic pollen does not exclude a declared extract, because extraction, purification, filtration, and formulation can remove or destroy palynomorphs [90,91,92,93,94,95].

Table 3. Comparative analytical methods for botanical-supplement quality assessment.

Method	Primary Application	Main Strength	Principal Limitation/Evidential Boundary	Ref.
HPTLC/HPLC/UPLC fingerprints; marker assays	Authenticated comparison, lot consistency, and marker quantification	Links overall pattern with defined quantitative markers	A marker alone does not prove species identity or absence of substitution; fingerprints require representative references	[78,79]
LC-MS/MS and LC-HRMS	Predefined markers, adulterants, or contaminants; suspect/non-targeted screening	High sensitivity and selectivity; broad acquisition using high-resolution mass spectrometry	Matrix effects and tentative library matches require controls and orthogonal confirmation	[50,56,70,79,80]
GC-based methods	Volatiles, selected non-polar analytes, pesticides, or solvents	Strong separation and mature libraries for amenable compounds	Requires volatility or derivatisation and is unsuitable for many polar or thermolabile analytes	[50,55,70]
NMR and qNMR	Chemical fingerprinting, structural elucidation, and marker quantification	Reproducible, structure-rich spectra; absolute quantification in suitable protocols	Lower trace sensitivity than LC-MS, spectral overlap, and high cost; chemical evidence does not independently establish taxonomic or species identity	[70,81]
FTIR/chemometrics	Rapid comparative screening and lot similarity	Fast, non-destructive, and requires little preparation	Sensitive to matrix and presentation; classifiers require representative external validation	[79,83,84,85,86]
ICP-MS or ICP-OES	Metals and metalloids	Sensitive multi-element quantification	Destructive and element-specific; speciation may require separate methods	[13,62,63,64,83]
Contaminant panels	Pesticides, mycotoxins, plant alkaloids, and residual solvents	Hazard-specific quantification against specifications	Fixed panels miss unexpected hazards; matrix validation and risk-based updates are essential	[62,63,64,65,82,83]
Culture and PCR/qPCR	Viable counts, specified organisms, and targeted identification	Culture estimates viability; molecular assays improve identification	DNA detection does not prove viability; sampling and contamination controls are critical	[57,58,59,60]
DNA-based methods	Support for species identity in raw materials, powders, or mixtures	Useful when diagnostic morphology is lost	Not quantitative for chemical exposure; processing, reference gaps, and trace DNA can confound results	[53,77,87,88,89]
Palynology	Trace botanical or environmental context in minimally processed material	Supports defined traceability questions	Processing loss, mixed assemblages, and limited standardisation prevent stand-alone attribution	[90,91,92,93,94,95]

compares the primary applications, strengths, and evidential limits of the methods discussed above.

3.3. Integrated Workflow and Evidential Interpretation

Analytical methods provide complementary, not interchangeable, evidence. Fingerprints and marker assays assess composition and consistency. Mass spectrometry addresses predefined or unexpected analytes but requires confirmation, whereas FTIR screens for similarity. Culture estimates viability, while molecular assays aid identification. DNA supports botanical identity but not chemical equivalence, and palynology supplies context mainly for minimally processed material [50,53,57,77,79,83,87,90]. Evidential weight depends on the question, matrix, processing, validation, reference materials, and decision thresholds.

Assessment begins by defining the decision question and documenting the product, lot, formulation, supplier records, sampling, and chain of custody. Screening combines label and documentary review with fit-for-purpose comparative methods; discordant or consequential findings trigger orthogonal confirmatory and quantitative testing. Figure 1 makes these stages and escalation points explicit [51,53,62,70,77].

Defensible conclusions integrate formulation, lot, supplier, and manufacturing records; certificates of analysis; storage history; and complementary laboratory findings. Suspected pharmaceutical adulteration, batch inconsistency, botanical substitution, microbial contamination, and environmental traceability require different tests. Retention of the container and remaining product from the consumed lot, with chain-of-custody documentation, is critical when findings may support causality, recall, or consumer protection [51,53,62,66,77].

Three applications illustrate the decision logic. For suspected turmeric-associated liver injury, the high clinical consequence justifies stopping exposure, retaining the consumed lot, documenting enhancers and co-ingredients, quantifying curcuminoids, investigating adulterants or contaminants, and integrating analytical findings with clinical causality evidence [34,39,43,66,67]. For routine milk-thistle lot release, identity, chemical fingerprint, marker content, contaminants, microbiological specifications, and lot consistency are assessed, with escalation only when screening or documentation identifies a consequential discrepancy [18,51,57,58,59,60,61,62,78,79]. For a high-dose green-tea extract, epigallocatechin-3-gallate exposure and formulation are verified, product conformity and co-ingredients are assessed, and safety evaluation is escalated when symptoms, abnormal liver tests, or a high-exposure formulation raise concern [10,44,47,48].

3.4. Research Gaps and Reporting Standards

Research priorities include authenticated reference materials and transparent marker panels. Reporting needs to cover plant part, extraction, drug-to-extract ratio, marker content, and bioavailability enhancers. Fingerprinting and contaminant assays require matrix-specific validation, while DNA and palynological workflows require reproducible sampling and interpretation [78,79,87,90]. The Orbitrap-HRAMS/chemometric approach is promising. However, its transferability beyond the study’s defined sample set remains to be established through independent external validation across additional origins, harvests, processing states, laboratories, and finished-product matrices [80].

Clinical studies need to treat the defined product—not a generic botanical name—as the intervention. Reports require taxonomy and plant part, extraction solvent and ratio, source, lot, expiry and storage, complete formulation, enhancers, dose, markers, and preferably batch-specific validated fingerprints. MASLD/MASH trial protocols need prespecified outcome hierarchies that distinguish biochemical and imaging surrogate outcomes from histological endpoints and long-term clinical outcomes [15,78,79,96].

Table 4. Minimum reporting checklist for product-characterised controlled clinical studies of botanical supplements marketed for liver support. Adapted from and extending CONSORT guidance for herbal interventions [96].

Reporting Domain	Minimum Information to Report
Botanical identity and source material	Scientific name, plant part, voucher or authentication method, geographic source, where available [78,79,96].
Extraction and formulation	Extraction solvent(s), drug-to-extract ratio, extraction process, dosage form, excipients, and any bioavailability-enhancing agents [78,79,96].
Manufacturing and batch traceability	Manufacturer, batch/lot number, expiry date, storage conditions, and whether the same batch was used throughout the trial [51,78,96].
Chemical characterisation	Validated quantitative marker panel, chemical fingerprint, analytical method, reference standards, and batch-specific results [70,78,79,80,81,96].
Product purity and safety status	Testing or certificates of analysis addressing relevant contaminants, microbial quality, and undeclared pharmacologically active substances [51,57,58,59,60,61,62,63,64,65,96].
Dose and exposure	Daily dose of product and marker compound(s), schedule, administration conditions, and co-intervention restrictions [96].
Comparator or placebo	Composition, matching strategy, and evidence that the placebo does not introduce botanical or analytical confounding [96].
Clinical population and diagnostic framework	Eligibility criteria, MASLD/MASH or historical NAFLD/NASH definition, metabolic criteria, baseline fibrosis or imaging status [15,96].
Outcomes and safety monitoring	Prespecified primary outcome; distinction between biochemical and imaging surrogate outcomes, histological endpoints, and long-term clinical outcomes; adverse-event collection and liver-safety plan where appropriate [15,24,67,96].
Data availability	Batch-specific analytical documentation and key product-characterisation data are provided in the Supplementary Materials or an accessible repository [96].

provides a minimum reporting checklist.

Table S1 shows why this matters: studies differ in formulation, dose, duration, endpoint hierarchy, safety role, and characterisation. Without batch data, discordant results may reflect non-equivalent interventions rather than biological disagreement. Table 4 extends CONSORT guidance for herbal interventions so that product identity, analytical characterisation, efficacy endpoints, and safety reporting are reproducible [96].

3.5. Limitations

This focused narrative review has several limitations. We selected case studies and reports purposively rather than through a systematic-review process. No protocol was preregistered, and duplicate screening, quantitative pooling, formal risk-of-bias assessment, and certainty grading were not performed. The controlled evidence was heterogeneous and dominated by short-term surrogate outcomes, while incomplete formulation and lot characterisation constrained product-level comparability. Because searches were last updated on 3 August 2026, later sources may alter specific interpretations. Finally, the proposed workflow is conceptual and qualitative. It has not been validated as a quantitative risk score and should not replace validated analytical methods, regulatory requirements, or clinical causality assessment.

4. Conclusions

Plant-based supplements marketed for liver support do not form a uniform therapeutic class. Clinical assessment requires identifying the exact product and dose, reviewing concomitant medications and hepatobiliary contraindications, and distinguishing surrogate changes from disease modification. If injury is suspected, retain the remaining product from the consumed lot where possible, and report serious or otherwise reportable events through the applicable system. Manufacturers and trialists need to disclose and verify the identity, plant part, extraction, drug-to-extract ratio, marker profile, enhancers, co-ingredients, lot, and storage. Regulators and laboratories can then link claims and supply-chain records to representative sampling, screening, confirmatory analysis, and traceable batch data.

Across milk thistle, artichoke, turmeric, and green tea, the purposively selected controlled human evidence remains dominated by short-term surrogate outcomes. The histology-based silymarin trials did not meet their primary endpoints, and no selected report established that a defined product reduced hepatic decompensation, hepatocellular carcinoma, the need for transplantation, or liver-related mortality. Turmeric or curcumin products and concentrated green tea extracts also show that a plausible benefit can coexist with liver-injury risk. Until defined products demonstrate clinically meaningful benefit, communication needs to remain product-specific, uncertainty explicit, and quality assurance and surveillance prioritised over broad ‘liver support’ claims.