Pasteurized Akkermansia: Muscle Promise, Cancer Questions

Pasteurized Akkermansia muciniphila produced statistically significant gains in selected leg-strength measures in a 12-week placebo-controlled trial of older adults. That is unusual evidence in a supplement category built largely on biomarkers, mice, and optimistic label copy. It is also incomplete evidence: the study did not show a grip-strength benefit, did not report absolute torque changes, and enrolled a group that was 88% female.

Before treating it as a sarcopenia supplement, buyers should also understand the cancer question. A 2025 systematic review found that live A. muciniphila can behave as a pathobiont—an ordinarily harmless microbe that contributes to disease under certain conditions—in inflamed, dysbiotic mouse guts. That finding does not show that a pasteurized supplement causes colorectal cancer in humans. No human trial or prospective cohort has established that risk. Still, the people most interested in preserving muscle mass are also more likely to have histories of antibiotic exposure, gut disease, or colorectal pathology that deserve more study than a supplement label can provide.

The muscle result is real, but narrow

In a 2024 randomized, double-blind, placebo-controlled trial, researchers assigned 100 South Korean adults with low SARC-CalF scores to either 1×10¹⁰ cells per day of pasteurized A. muciniphila HB05P or microcrystalline-cellulose placebo for 12 weeks. Participants averaged 65.03 years old, and 88% were women. The per-protocol analysis included 92 participants: 47 in the HB05P group and 45 in the placebo group.

Compared with placebo, HB05P significantly improved left-leg extensor peak torque (p=0.0103) and combined extensor peak torque (p=0.049). Torque relative to body weight also favored HB05P for the left leg (p=0.0052) and combined legs (p=0.0346). Serum follistatin rose from baseline in the HB05P group while it declined in the placebo group. Follistatin regulates pathways related to muscle growth, but a biomarker shift is not proof that the supplement prevents disability, falls, hospitalization, or frailty.

The trial did not report confidence intervals or absolute torque changes in newton-meters, so readers cannot judge the size of the improvement or compare it fairly with resistance training, creatine, adequate protein, or leucine. Statistical significance answers whether the observed group difference is compatible with chance under the study model; it does not answer whether the difference changes daily life.

The result also depended heavily on the left leg. Right-leg extensor measures did not independently reach statistical significance, and grip strength showed no significant between-group difference at any time point. Grip strength is a widely used clinical measure in sarcopenia research, so its null result limits claims about whole-body function. The study offers a useful proof of concept, not a validated treatment for clinical sarcopenia.

The trial can support this claim: pasteurized HB05P improved selected isokinetic leg-strength outcomes over 12 weeks in a mostly female group of older adults with mild or subclinical sarcopenia. It cannot tell us how many newton-meters of strength a buyer should expect, whether the effect applies to men, or whether it improves meaningful physical function.

Why the cancer story depends on form and context

A. muciniphila lives in the intestinal mucus layer and can break down mucin. In a healthy gut, that activity may fit into normal mucus turnover. In an inflamed or disrupted gut, it may contribute to thinning of the protective mucus barrier. The 2025 systematic review of 34 studies identified the concerning evidence in live-bacteria animal models, including colitis-associated AOM/DSS models, Apc^min/+ models, and germ-free settings after antibiotic-related dysbiosis. These experiments linked high-density live bacteria, excessive mucin degradation, barrier disruption, and tumor-promoting conditions.

Those are biologically serious findings, but they are not evidence that swallowing pasteurized cells causes cancer. The HB05P intervention used heat-killed bacteria, not organisms able to colonize and keep degrading mucin. Pasteurization at 70°C for 30 minutes is intended to eliminate viability while retaining some heat-stable bacterial components. Research on pasteurized preparations identifies retained bioactive proteins, including P5, under that processing condition in metabolic models; higher-temperature killing abolished the reported activity in that work (Niu et al.).

That distinction should cool both marketing and alarm. It is reasonable to infer that a nonviable preparation differs biologically from live bacteria colonizing an inflamed mouse colon. It is not reasonable to call pasteurized supplementation anti-cancer in humans on that basis.

Preclinical anti-tumor findings do not settle human safety

Several laboratory and animal studies point in the opposite direction from the live-bacteria tumor models. In colorectal cancer models, the surface protein Amuc_1434 reduced PD-L1 expression on cancer cells and enhanced CD8+ T-cell-mediated anti-tumor immunity (Zhu et al.). In mice exposed to chronic stress, A. muciniphila-derived outer-membrane vesicles attenuated colorectal tumor growth (Jin et al.). A separate cell-based study reported inhibition of colorectal cancer-cell proliferation through ferroptosis, an iron-dependent form of cell death (Han et al.).

These studies help explain why the organism cannot be reduced to either "good probiotic" or "cancer microbe." They do not establish that a commercial pasteurized product prevents cancer, treats cancer, or is safe for people with cancer. Cell lines and mouse tumors answer mechanistic questions under controlled conditions; they do not substitute for clinical outcomes in people.

Human microbiome observations further complicate the story. Across three cross-sectional datasets summarized in the pathobiont review—including studies of people with polyps, adenomas, and colorectal cancer—A. muciniphila abundance decreased as colorectal cancer stage advanced. One dataset included 455 people and stratified colorectal cancers by mismatch-repair status. These associations contradict a simple claim that higher abundance drives human colorectal cancer, but they cannot establish protection either. Cancer, diet, treatment, inflammation, and the tumor microenvironment can all alter the microbiome after disease develops.

The unanswered safety question

The evidence gap is specific. No human intervention study has tested pasteurized A. muciniphila in people with active inflammatory bowel disease, active colitis, recent major antibiotic-related dysbiosis, previous colorectal cancer, or established colorectal lesions. Reviews of the field describe host context—not merely bacterial identity—as the determinant of whether A. muciniphila may help or harm the intestinal barrier (Zhang et al.; Mierlan et al.).

Strain identity adds another complication. A. muciniphila includes several candidate species and phylogroups with differing characteristics, so results from HB05P should not be assumed to apply to every product sold under the species name. Commercial products also do not generally screen buyers for inflammatory bowel disease, prior colorectal cancer, active mucosal inflammation, or tumor mismatch-repair status.

For a healthy older adult without known inflammatory bowel disease or colorectal pathology, pasteurized A. muciniphila is a defensible experimental addition—not a replacement for progressive resistance training, adequate calories, protein, and medical evaluation of muscle loss. Expectations should stay modest because the only strength trial leaves the most useful effect-size question unanswered.

People with active IBD or colitis, substantial recent antibiotic disruption, a personal history of colorectal cancer or advanced polyps, or ongoing evaluation for gastrointestinal symptoms should discuss it with a gastroenterologist or treating clinician before using it. The current evidence does not show pasteurized A. muciniphila causes cancer in these groups. It also does not show that it is safe there.