Tephrosin

Tephrosin at a Glance

Tephrosin is a rotenoid, and structurally it is one atom’s worth of difference from deguelin: it is 12aβ-hydroxydeguelin. It occurs in the same plants and the same commercial preparations — about 3.5% of cubé resin against deguelin’s 21% and rotenone’s larger share, the four together making up 77% of the resin’s weight ^{1,2Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →Reference 2Fang N · 1999In vitroCubé resin insecticide: identification and biological activity of 29 rotenoid constituents — in vitroView study →}.

That single hydroxyl group is the interesting thing about the molecule, because it pulls in two directions. On the respiratory-poison axis, 12aβ-hydroxylation is the body’s detoxification step: hydroxylated rotenoids are consistently weaker complex-I inhibitors than their parents, and it is the parent rotenoids, not their 12aβ-hydroxy metabolites, that produce the Parkinson’s-like syndrome in rats ^{3Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →}. On the NF-κB axis it goes the other way: among the phenolics of Amorpha fruticosa, tephrosin was the single most potent inhibitor at IC50 0.11 µM, ahead of 11-hydroxytephrosin at 0.19 µM and deguelin at 0.22 µM ^{4Reference 4Dat NT et al. · 2008In vitroPhenolic constituents of Amorpha fruticosa that inhibit NF-kappaB activation and related gene expression — in vitroView study →}.

It is still a rotenoid, still an insecticide, and still toxic to fish and to cultured cells — just less so than the compounds it sits beside ^{7,8Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}. Nothing here is a medicine, and nothing has been tested in a person.

Tephrosin Is Deguelin’s Hydroxylated Twin

Almost every claim made about tephrosin becomes legible once its relationship to deguelin is fixed, because the two are interconvertible in the laboratory and correlated in the plant.

In the plant and the product, they co-occur in fixed proportion. Cubé resin carries deguelin at about 21.2%, tephrosin at 3.5% and β-rotenolone at 3.0%, alongside rotenone ^{1Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →}; twenty-nine rotenoids in total have been catalogued from that resin, and when they were assayed side by side, the same structure–activity relationship governed complex-I inhibition, tumour-promotion inhibition and cytotoxicity alike ^{2Reference 2Fang N · 1999In vitroCubé resin insecticide: identification and biological activity of 29 rotenoid constituents — in vitroView study →}. Overall molecular conformation and the E-ring substituents decided all three.

In the flask, tephrosin is made from deguelin. The 2017 semi-synthesis opens rotenone with a zinc-mediated ring-opening to rot-2′-enonic acid, converts that to deguelin, and then installs the 12a-hydroxyl by a chromium-mediated oxidation that gives tephrosin as a single diastereoisomer ^{10Reference 10Russell DA et al. · 2017Stereocontrolled semi-syntheses of deguelin and tephrosinView study →}. Scalable and high-yielding, it is how both molecules are most practically obtained.

In the animal, the same transformation is metabolism. Human CYP3A4 and CYP2C19 hydroxylate rotenone and deguelin at the 12aβ position, with CYP2C19 the more selective of the two; the resulting 12aβ-hydroxyrotenoids are less active as complex-I inhibitors than the parents, and in the rat parkinsonism experiments the neuropathological lesions tracked brain levels of parent rotenoid rather than of these metabolites ^{3Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →}. Tephrosin, in other words, is what a body makes out of deguelin on the way to getting rid of it.

The practical consequence runs through the pesticide data: in a head-to-head comparison against the bruchid beetle Callosobruchus maculatus, tephrosin was less toxic than deguelin, which was less active than rotenone ^{7Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}. Same conclusion in the fish work — tephrosin, deguelin and rotenone were cytotoxic to trout gill cells at ≥50 nM, while the crude Tephrosia vogelii extract managed it at 5 nM ^{8Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}.

What It's Used For

Tephrosin’s only real-world role is as a component of botanical insecticides. Its laboratory literature is a small oncology set, all in vitro, in which the recurring result is ATP depletion — which is what a complex-I inhibitor does.

A minor partner in a real pesticide.One of the four active rotenoids of cubé resin at ~3.5% of the weight, and consistently the least potent of them against insects ^{1,2,7Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →Reference 2Fang N · 1999In vitroCubé resin insecticide: identification and biological activity of 29 rotenoid constituents — in vitroView study →Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}.

The most potent NF-κB inhibitor in its plant.IC50 0.11 µM in TNF-α-stimulated reporter assays — ahead of deguelin, and far ahead of the amorfrutins from the same shrub ^{4Reference 4Dat NT et al. · 2008In vitroPhenolic constituents of Amorpha fruticosa that inhibit NF-kappaB activation and related gene expression — in vitroView study →}.

Its cancer-cell activity reads as energy failure.Combined with the glycolysis blocker 2-deoxy-D-glucose it accelerates ATP depletion, drives AMPK activation and mTOR shutdown, and — unusually — blunts the autophagy that normally rescues starved cells ^{12Reference 12Choi Y · 2011In vitroThe combination of tephrosin with 2-deoxy-D-glucose enhances the cytotoxicity via accelerating ATP depletion and blunting autophagy in human cancer cells — in vitroView study →}.

No human data, no animal pharmacokinetics, no dose.Everything below is cell culture, insects, fish or a single 1993 mouse skin study ^{5Reference 5Konoshima T et al. · 1993AnimalStudies on inhibitors of skin tumor promotion, XIIView study →}.

Evidence by IndicationStrength of support · Beta

A small, entirely in-vitro set, and the interesting thread is that it keeps arriving at energy failure rather than at a signalling target. In A549 lung cancer cells, tephrosin arrested the cell cycle at G2/M and killed by autophagy rather than apoptosis — no PARP cleavage at 24 h, but acidic vesicular organelles and LC3-I→LC3-II conversion, alongside ROS generation and reduced Hsp90 expression ^{11Reference 11Li J et al. · 2010In vitroTephrosin-induced autophagic cell death in A549 non-small cell lung cancer cells — in vitroView study →}.

Combined with the glycolytic inhibitor 2-deoxy-D-glucose, it accelerated intracellular ATP depletion, drove AMPK activation and mTOR inactivation, and suppressed the eEF-2-kinase-dependent autophagy that 2-DG normally induces — removing the cell’s escape route ^{12Reference 12Choi Y · 2011In vitroThe combination of tephrosin with 2-deoxy-D-glucose enhances the cytotoxicity via accelerating ATP depletion and blunting autophagy in human cancer cells — in vitroView study →}.

In paclitaxel-resistant ovarian cancer cells it resensitised to paclitaxel, inhibiting FGFR1 and its adapter FRS2 (but not EGFR) and lowering XIAP, without affecting P-glycoprotein expression or function ^{13Reference 13Kim HS et al. · 2023In vitroTephrosin Suppresses the Chemoresistance of Paclitaxel-Resistant Ovarian Cancer via Inhibition of FGFR1 Signaling Pathway — in vitroView study →}.

In multidrug-resistant breast and gastric lines it lowered the IC50 of daunorubicin and mitoxantrone without reducing P-gp or ABCG2 transcription ^{14Reference 14Darzi S et al. · 2021In vitroImprovement of cytotoxicity of mitoxantrone and daunorubicin by candidone, tephrosin, and bavachinin — in vitroView study →}.

Gap: no animal tumour model, no pharmacokinetics and no human data. A compound that kills by starving cells of ATP is not selective for tumour tissue, and the two studies that looked for a transporter mechanism found none — leaving general metabolic poisoning as the parsimonious reading of the chemosensitisation ^{13,14Reference 13Kim HS et al. · 2023In vitroTephrosin Suppresses the Chemoresistance of Paclitaxel-Resistant Ovarian Cancer via Inhibition of FGFR1 Signaling Pathway — in vitroView study →Reference 14Darzi S et al. · 2021In vitroImprovement of cytotoxicity of mitoxantrone and daunorubicin by candidone, tephrosin, and bavachinin — in vitroView study →}.

Mechanisms

Target / pathway	Effect	Relevant to
Mitochondrial complex I	inhibition — weaker than deguelin or rotenone, since 12aβ-hydroxylation is the detoxifying modification	insecticidal · piscicidal · cytotoxic activity
Cellular ATP / AMPK / mTOR	ATP depletion with AMPK activation and mTOR shutdown, amplified by concurrent glycolytic blockade	antiproliferative activity2-DG synergy
eEF-2 kinase	inhibited, blunting the protective autophagy a starved cell would otherwise mount	why the 2-DG combination is more than additive
Hsp90	expression reduced in A549 cells	autophagic cell death in lung cancer models
FGFR1 / FRS2	phosphorylation suppressed (EGFR unaffected); XIAP down	resensitisation of paclitaxel-resistant ovarian cells
NF-κB	reporter activity blocked at IC50 0.11 µM; DNA binding and target-gene expression suppressed	inflammatory signalling (in vitro only)

The first two rows are one row. Everything downstream — AMPK, mTOR, autophagy, the drug-resistance reversal — is what a cell does when its respiratory chain is inhibited and its ATP falls.

Pharmacokinetics

None published. There is no absorption, distribution, metabolism or half-life study for tephrosin in any species. What is known runs the other way: tephrosin is itself a metabolite, the 12aβ-hydroxy product that human CYP3A4 and CYP2C19 generate from deguelin, and that route is a detoxification — the hydroxylated rotenoids are weaker complex-I inhibitors than the parents and did not accumulate in rat brain at the levels the parents reached ^{3Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →}. The related molecule deguelin has a rat pharmacokinetic profile; tephrosin has nothing.

Clinical Trials

None. Tephrosin has not been administered to human subjects.

Completed	Planned	Terminated	Preclinical
—	—	—	In-vitro oncology; insect, fish and mouse-skin models

Last checked: September 2026.

Isolate vs. Plant Studies

The oncology work is all isolate in cell culture, with no animal or human layer to transfer it to. The pesticide work is the reverse and more informative: whole Tephrosia vogelii extract and cubé resin are what people actually use, and in the aquatic-toxicity study the crude extract was ten times more potent than any purified rotenoid in it, cytotoxic at 5 nM against 50 nM for tephrosin, deguelin and rotenone individually ^{8Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}. At the same time the chemotype comparison shows the rotenoids genuinely carry the pesticidal effect, since a chemotype without them has none ^{7Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}. Both facts hold: the rotenoids are the active principles, and the mixture is more toxic than the sum of the parts as isolated.

Neither source plant has a medicinal tradition that would generate human exposure to tephrosin, so there is no plant-side clinical literature to weigh against the isolate work.

Prevalence in Nature

Tephrosin is a legume rotenoid. It takes its name from Tephrosia, a genus of roughly 400 species in which rotenoid content varies enormously between species and even between strains of one species ^{9Reference 9Zhang P et al. · 2020ReviewPlants in the Genus Tephrosia: Valuable Resources for Botanical Insecticides — reviewView study →}. T. vogelii — the fish-poison bean — is the important one: two chemotypes exist, one making rotenoids and one not, and only the first has pesticidal activity ^{7Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}.

Its richest measured source is cubé resin, the root extract of Lonchocarpus utilis and L. urucu, at about 3.5% of the resin — a distant third behind rotenone and deguelin, and roughly level with β-rotenolone ^{1Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →}. Twenty-nine rotenoids have been catalogued from that resin, among them four isomers of 4′,5′-dihydro-4′,5′-dihydroxytephrosin, some of which are likely artefacts of extraction and processing rather than plant products ^{2Reference 2Fang N · 1999In vitroCubé resin insecticide: identification and biological activity of 29 rotenoid constituents — in vitroView study →}.

It also occurs in Derris species and in false indigo bush (Amorpha fruticosa), whose leaves, seeds and roots carry it beside deguelin, 11-hydroxytephrosin, amorphigenin, dalbinol and amorphispironone ^{4,5,6Reference 4Dat NT et al. · 2008In vitroPhenolic constituents of Amorpha fruticosa that inhibit NF-kappaB activation and related gene expression — in vitroView study →Reference 5Konoshima T et al. · 1993AnimalStudies on inhibitors of skin tumor promotion, XIIView study →Reference 6Li L et al. · 1993In vitroAntitumor agents, 138View study →}. As with all rotenoids, the plant concentrates it in roots and seeds, where a respiratory poison defends immobile, high-value tissue.

Discovery & Synthesis

Tephrosin belongs to the 1930s rotenoid chemistry that grew up around commercial fish-poison and insecticide plants, and its name records the genus it came from. The precise first-isolation report predates the indexed literature and is not asserted here.

The modern route is a semi-synthesis from rotenone, published in 2017: zinc-mediated ring-opening of rotenone hydrobromide to rot-2′-enonic acid, conversion of that to deguelin, and then a chromium-mediated hydroxylation that delivers tephrosin as a single diastereoisomer — the authors propose an Étard-like mechanism for the stereochemical outcome. The sequence is operationally simple, scalable and high-yielding, and it supersedes earlier approaches ^{10Reference 10Russell DA et al. · 2017Stereocontrolled semi-syntheses of deguelin and tephrosinView study →}.

Tephrosin is not manufactured as a product. It reaches the world inside rotenone-based botanical insecticides, and reaches laboratories as an isolate or a semi-synthetic sample.

Patents: not yet researched.

Toxicity & Safety

Tephrosin is a rotenoid and carries the class hazard, moderated by the hydroxyl that makes it the weakest of its group.

Safety at a Glance

Toxic

A respiratory poison, at reduced potency. Complex-I inhibition is the mechanism; 12aβ-hydroxylation weakens it relative to deguelin and rotenone but does not remove it ^{3,7Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}.
Non-selective cytotoxicity. The antiproliferative results are ATP depletion measured in cancer cells, with no demonstrated margin over normal tissue ^{11,12Reference 11Li J et al. · 2010In vitroTephrosin-induced autophagic cell death in A549 non-small cell lung cancer cells — in vitroView study →Reference 12Choi Y · 2011In vitroThe combination of tephrosin with 2-deoxy-D-glucose enhances the cytotoxicity via accelerating ATP depletion and blunting autophagy in human cancer cells — in vitroView study →}.
Aquatic toxicity. Cytotoxic to trout gill epithelium and harmful to zebrafish embryos from 50 nM; the crude Tephrosia extract it comes in is toxic at 5 nM ^{8Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}.
Travels unmeasured in pesticide products. Only rotenone is counted in the legal residue determination for rotenone-based formulations, though the congeners are present and active ^{1Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →}.

Full Safety & Interactions Detail

Class hazard, and where tephrosin sits in it. Rotenone has been withdrawn or restricted in many countries over its link to Parkinson’s disease, and the rat work that established that link tested deguelin alongside it — deguelin produced the same nigrostriatal degeneration at roughly half rotenone’s potency ^{3Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →}. Tephrosin was not tested in that model. What the same study establishes is directional: 12aβ-hydroxyrotenoids are less active on complex I than their parents, they did not reach the brain levels the parents did, and the neuropathology tracked the parents ^{3Reference 3Caboni P et al. · 2004AnimalRotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal modelView study →}. Tephrosin is therefore expected to be the least neurotoxic of the group — an inference from metabolism and potency data, not a measured result, and not a safety clearance.

Measured toxicity. Tephrosin was cytotoxic to the RTgill-W1 trout gill epithelial line and caused deformity and mortality in zebrafish embryos at concentrations ≥50 nM ^{8Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}. Against stored-product beetles it is the least potent of rotenone, deguelin and tephrosin ^{7Reference 7Belmain SR et al. · 2012Highly variable insect control efficacy of Tephrosia vogelii chemotypesView study →}. In cell culture it kills by autophagy with ROS generation and G2/M arrest, at micromolar concentrations, in lines of several tissue origins ^{11Reference 11Li J et al. · 2010In vitroTephrosin-induced autophagic cell death in A549 non-small cell lung cancer cells — in vitroView study →}.

Interactions. No interaction study exists. The one mechanistically relevant observation is that tephrosin sensitised multidrug-resistant cells to daunorubicin and mitoxantrone without reducing P-glycoprotein or ABCG2 expression, and without affecting P-gp function in a separate ovarian model ^{13,14Reference 13Kim HS et al. · 2023In vitroTephrosin Suppresses the Chemoresistance of Paclitaxel-Resistant Ovarian Cancer via Inhibition of FGFR1 Signaling Pathway — in vitroView study →Reference 14Darzi S et al. · 2021In vitroImprovement of cytotoxicity of mitoxantrone and daunorubicin by candidone, tephrosin, and bavachinin — in vitroView study →} — so the effect is not transporter inhibition, and it does not predict a pharmacokinetic drug interaction in a person.

What is not known. There is no human toxicology for tephrosin: no poisoning case reports naming the isolated compound, no genotoxicity or repeat-dose package, no reproductive-toxicity study, and no occupational data separating it from rotenone. Human exposure occurs as part of rotenoid pesticide mixtures.

Dosage

There is no human dose, no traditional preparation and no supplement form. The figures below are the concentrations at which tephrosin has actually been studied:

Setting	Form	Figure
Cubé resin composition	Root-extract resin	~3.5% tephrosin by weight (deguelin ~21.2%) ^{1Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →}
Residue on treated olives at harvest	Field application	0.06 mg/kg (rotenone 0.08 mg/kg) ^{1Reference 1Cabizza M et al. · 2004Rotenone and rotenoids in cubè resins, formulations, and residues on olivesView study →}
Trout gill cells and zebrafish embryos	Water exposure	Toxic from 50 nM ^{8Reference 8Said AH et al. · 2020In vitroIsolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitroView study →}
NF-κB reporter inhibition	In-vitro culture	IC50 0.11 µM ^{4Reference 4Dat NT et al. · 2008In vitroPhenolic constituents of Amorpha fruticosa that inhibit NF-kappaB activation and related gene expression — in vitroView study →}
Cancer cell lines	In-vitro culture	Low micromolar, varying by line ^{11,12Reference 11Li J et al. · 2010In vitroTephrosin-induced autophagic cell death in A549 non-small cell lung cancer cells — in vitroView study →Reference 12Choi Y · 2011In vitroThe combination of tephrosin with 2-deoxy-D-glucose enhances the cytotoxicity via accelerating ATP depletion and blunting autophagy in human cancer cells — in vitroView study →}

These are research and pesticide figures, not a personal recommendation, and none of them supports an internal human dose.

References

Cabizza M, Angioni A, Melis M, et al. (2004). Rotenone and rotenoids in cubè resins, formulations, and residues on olives. J Agric Food Chem. https://pubmed.ncbi.nlm.nih.gov/14733510/
Fang N, Casida JE (1999). Cubé resin insecticide: identification and biological activity of 29 rotenoid constituents — in vitro. J Agric Food Chem. https://pubmed.ncbi.nlm.nih.gov/10552508/
Caboni P, Sherer TB, Zhang N, et al. (2004). Rotenone, deguelin, their metabolites, and the rat model of Parkinson’s disease — animal model. Chem Res Toxicol. https://pubmed.ncbi.nlm.nih.gov/15540952/
Dat NT, Lee JH, Lee K, et al. (2008). Phenolic constituents of Amorpha fruticosa that inhibit NF-kappaB activation and related gene expression — in vitro. J Nat Prod. https://pubmed.ncbi.nlm.nih.gov/18841906/
Konoshima T, Terada H, Kokumai M, et al. (1993). Studies on inhibitors of skin tumor promotion, XII. Rotenoids from Amorpha fruticosa — mouse model. J Nat Prod. https://pubmed.ncbi.nlm.nih.gov/8350086/
Li L, Wang HK, Chang JJ, et al. (1993). Antitumor agents, 138. Rotenoids and isoflavones as cytotoxic constitutents from Amorpha fruticosa — in vitro. J Nat Prod. https://pubmed.ncbi.nlm.nih.gov/8326318/
Belmain SR, Amoah BA, Nyirenda SP, et al. (2012). Highly variable insect control efficacy of Tephrosia vogelii chemotypes. J Agric Food Chem. https://pubmed.ncbi.nlm.nih.gov/22970736/
Said AH, Solhaug A, Sandvik M, et al. (2020). Isolation of the Tephrosia vogelii extract and rotenoids and their toxicity in the RTgill-W1 trout cell line and in zebrafish embryos — in vitro. Toxicon. https://pubmed.ncbi.nlm.nih.gov/32454059/
Zhang P, Qin D, Chen J, et al. (2020). Plants in the Genus Tephrosia: Valuable Resources for Botanical Insecticides — review. Insects. https://pubmed.ncbi.nlm.nih.gov/33096762/
Russell DA, Freudenreich JJ, Ciardiello JJ, et al. (2017). Stereocontrolled semi-syntheses of deguelin and tephrosin. Org Biomol Chem. https://pubmed.ncbi.nlm.nih.gov/28134391/
Li J, Wang XL, Fang YC, et al. (2010). Tephrosin-induced autophagic cell death in A549 non-small cell lung cancer cells — in vitro. J Asian Nat Prod Res. https://pubmed.ncbi.nlm.nih.gov/21061222/
Choi Y, Lee JH (2011). The combination of tephrosin with 2-deoxy-D-glucose enhances the cytotoxicity via accelerating ATP depletion and blunting autophagy in human cancer cells — in vitro. Cancer Biol Ther. https://pubmed.ncbi.nlm.nih.gov/22123175/
Kim HS, Bae S, Lim YJ, et al. (2023). Tephrosin Suppresses the Chemoresistance of Paclitaxel-Resistant Ovarian Cancer via Inhibition of FGFR1 Signaling Pathway — in vitro. Biomedicines. https://pubmed.ncbi.nlm.nih.gov/38137377/
Darzi S, Mirzaei SA, Elahian F, et al. (2021). Improvement of cytotoxicity of mitoxantrone and daunorubicin by candidone, tephrosin, and bavachinin — in vitro. Mol Biol Rep. https://pubmed.ncbi.nlm.nih.gov/34564803/