Milk thistle

View my plan →

The short answer

Liver-disease trials are heterogeneous and do not establish routine detoxification or treatment benefit.

Evidence and scope: Mixed evidence
This status does not grade the ingredient for every goal, outcome, or person.
Important limitation: Trials are inconsistent or small, so a reliable clinically important benefit is not established.
Before deciding: Key concern: Cancer therapies, diabetes medicines, CYP interactions, ragweed allergy, pregnancy, and liver symptoms require review.

01 / PRACTICAL GUIDE

Dose, absorption, and combinations

These are indication-specific reference or studied ranges—not a personalized prescription. Start with the objective, then check the form, label, medicines, and health conditions.

Reference range by objective

Meeting usual needsAbout 420 mg/day silymarin in divided doses in many trials; no established wellness dose

Reference or studied range—not a personalized prescription. The product label or a clinician may specify a different dose.

Dose source ↗

More is not automatically better. A clinician or product label may specify a different amount.

Absorption / solubilityFormulation-dependent

Not usefully classified as a fat- or water-soluble vitamin; absorption depends on the salt, extract, delivery system, and meal.

How it is commonly taken

Take with food; standardization and formulation matter.

Do not combine casually

Cancer therapies, diabetes medicines, CYP interactions, ragweed allergy, pregnancy, and liver symptoms require review.

Dose and use source ↗

Dose and use links lead to primary or official source material. Search links are discovery tools, not proof that every returned paper applies.

02 / EVIDENCE

What the research says

We keep supportive, mixed, negative, safety, and contextual findings together. The headline never replaces the underlying papers.

1research papers in this ledger

8unique primary studies represented

—unique participants represented

These are audited counts for this JIVANA collection, not search-result totals. Participants are counted once when multiple papers use the same cohort; review papers are not added again to participant totals.

Open the research timelineResearch timeline

Entries are ordered by publication year. A study period appears only when the source reports it.

2021

Mixed or qualifiedPMID 33418491

Silymarin in nonalcoholic fatty liver disease

Transaminase levels fell more with silymarin, but the review identified quality flaws and could not establish that enzyme changes correspond to histologic improvement. It does not demonstrate prevention of cirrhosis, liver failure, transplantation, or death.

Published: 20218 primary studies represented

Study design and methods

Design: Systematic review and meta-analysis of 8 randomized clinical trials
Population: People with nonalcoholic fatty liver disease; the abstract did not report a pooled participant total
Intervention: Silymarin versus placebo
Outcome: Alanine and aspartate aminotransferases, body mass index, and liver histology

Open research record ↗

Unscreened discovery appendix—not evidence29 candidate publicationsSearch run: Aug 21, 2026

Complete for the displayed PubMed query at the search timestamp—not for all databases or every possible synonym. Candidates can be irrelevant, duplicate the same study, or report negative findings. They are never summed as evidence or participants until screened.

03 / SOURCES

Start with the source material

Official evidence overviewPubMed bounded human-evidence search Live PubMed discovery searchNational Library of Medicine

The live search is for discovery only. Search-result counts change and are not included in the audited totals above.

Recent PubMed discovery recordsDirect links to the first 12 records returned by the reproducible search. These are discovery candidates—not reviewed summaries or a count of supportive findings.

04 / EXPERT CONTEXT

Expert context—not evidence

Expert positions are dated context—not scientific evidence. They never raise the evidence status or become consensus by repetition.

No direct, source-verified endorsement is currently recorded. Research authorship or discussion alone is not treated as endorsement.