Phellinus linteus for Cancer Adjunct Therapy
As of 2026, there is no completed human randomized trial or clinical meta-analysis showing that Phellinus linteus (sanghuang) helps as an add-on to cancer treatment. The evidence is too thin and conflicted to consider it effective, and it should never replace standard oncology care.
Why this grade7-layer evidence engine
This grade reflects how little human evidence exists. The only study with actual cancer outcomes is a single retrospective propensity-matched cohort in pancreatic cancer patients after surgery (PMID 30501431, n=217), where Phellinus use was the lone significant predictor of completing adjuvant chemotherapy (p=0.039) and was linked to better survival (p<0.05). But this design cannot rule out selection bias or healthy-user effects, so it falls far short of proof.
The rest of the file is preliminary or non-clinical. A 2020 randomized trial (PMID 32011496) is only a published protocol measuring NK-cell activity—a surrogate, not a cancer endpoint—and has reported no results years later. The 2024 hepatocellular carcinoma work (PMID 38240856) is mouse and cell-culture only, and a 2023 systematic review of its polysaccharides (PMID 36623622) concedes clinical applications are 'still very limited.'
Regulators and major clinics reinforce caution. The FDA lists a Phellinus product (Mesima) on DailyMed as an 'unapproved drug,' a stronger signal than a routine supplement warning, while the American Cancer Society notes dietary supplements 'have generally not been proven to help prevent or treat cancer or its symptoms, and might sometimes even cause harm.' Mayo, Cleveland Clinic, and Harvard do not address it at all.
Scoring transparency
All scores computed by a 7-layer evidence engine — fully auditable▸View the full decision path (audit trail)
- compute_raw_score — 加權公式: L2×0.30 + L3×0.25 + L5×0.25 + L11×0.10 + L1×0.10 = 0.427
- tier_from_score — 依分數區間映射至 tier letter
- apply_hec_rules — 無高階證據可裁決
- tier_strict_requirement_check — | C→U 因 scope.conflation_risk=true 且 L11 獨評較低 (B7-2 tier cap)
- detect_disputes — 偵測到 1 個 hard + 0 個 soft dispute
- decide_status — 依 tier + dispute 結果決定 status