evipedia.ai
← Back

Rentosertib for Health & Longevity

Evidence Review created on 09/11/2026 using AI4L / Opus 5

Also known as: ISM001-055, INS018_055

Motivation

Rentosertib is an experimental oral medication built to block a single enzyme that helps cells lay down scar tissue. It is the first compound whose biological target and chemical structure were both proposed by generative artificial intelligence and then carried all the way into human testing, which is why interest in it reaches well beyond the lung disease it was designed to treat.

That disease — a progressive stiffening and scarring of the lungs — begins for most people after the age of sixty, and the enzyme it targets sits on signaling routes that also govern tissue repair, inflammation, and the wear that accumulates in cells over a lifetime. That overlap turned the compound into a test case for a larger question: whether a medicine aimed at one age-related illness can also shift the biology of growing older.

This review examines what the published laboratory and human evidence shows about rentosertib: how it works, what it has achieved in the people who have taken it, what harm has been recorded, who produced the data, and how strong the underlying evidence is.

Benefits - Risks - Protocol - Conclusion

High-level sources that explain rentosertib, its molecular target TNIK (an enzyme that switches on scar-forming and inflammatory signaling), and the aging-biology claims attached to it.

Content from five of the six priority sources could not be found: Rhonda Patrick, Peter Attia, Andrew Huberman and Chris Kresser have published nothing on rentosertib, on the enzyme it blocks, or on the company developing it, and Life Extension’s site returned no editorial coverage. This is expected for an unapproved compound that has never been available outside a clinical trial. Lifespan.io is the only priority platform that has covered it.

Grokipedia

  • Rentosertib

    A dedicated encyclopedia entry covering the compound’s identities, its machine-generated discovery, the enzyme it targets, and the mid-stage trial outcome, with its sourcing exposed for checking.

Examine

No Examine article exists for rentosertib. Examine.com covers dietary supplements, foods and nutrition topics; it does not typically cover prescription-track medications, and rentosertib is an unapproved investigational drug that has never been sold as a supplement.

ConsumerLab

No ConsumerLab article exists for rentosertib. ConsumerLab tests and reviews commercially sold supplements, foods and consumer health products; it does not typically cover prescription medications, and rentosertib is an unapproved investigational drug with no retail product to test.

Systematic Reviews

Pooled analyses that place rentosertib against the other drug strategies tested in the same disease.

Only the benefit side of the trade-off is represented. No systematic review or meta-analysis covers rentosertib’s principal risk — drug-induced liver injury — or its gastrointestinal burden, because a single 12-week trial is too thin a base for one; that half of the evidence remains unrepresented at review level.

Mechanism of Action

TNIK (Traf2- and Nck-interacting kinase, an enzyme that attaches phosphate groups to other proteins, switching them on or off) is a serine/threonine kinase — a class no existing fibrosis drug targets. It sits where several scar-forming routes converge: Wnt/β-catenin (a developmental cascade re-activated during wound repair), TGF-β/SMAD (transforming growth factor beta, the dominant driver of scar deposition), Hippo–YAP/TAZ (a mechanical-sensing pathway controlling tissue growth) and NF-κB (a master inflammation switch) (Ren et al., 2025).

Rentosertib competes for the enzyme’s energy pocket: it occupies the adenosine triphosphate site, hydrogen-bonds to hinge residue Cys108, and pushes past gatekeeper Met105 into a hydrophobic back cavity — a binding mode unique among known inhibitors. Binding affinity is 4.32 nM, enzyme IC50 (the concentration giving half-maximal inhibition) 31 nM. In human lung cells it reverses TGF-β-driven epithelial-to-mesenchymal and fibroblast-to-myofibroblast transition, lowering alpha smooth muscle actin, fibronectin and collagen I.

A competing reading of the selectivity data: rentosertib also inhibits MAP4K4, ALK4/TGFBR1 and DDR1 at nanomolar concentrations, enzymes that themselves drive scarring, so how much of the effect is TNIK-specific remains open.

Pharmacologically: oral, half-life 7.4–11.9 hours, peak at 1–2 hours fasted, dose-proportional, no accumulation over seven days; a high-fat meal delays the peak to three hours and lowers it by 37%. Distribution favors the lung, where rodent concentrations run roughly 50-fold above blood. Two major metabolites are formed, INS018_063 and INS018_095; trial protocols bar strong or moderate CYP3A4 and CYP1A2 inhibitors and inducers (liver enzymes that break down most drugs), indicating those pathways dominate clearance.

Historical Context & Evolution

The target came first, and it came from cancer research. TNIK was described as a Wnt-pathway component essential for colorectal cancer growth in 2010 (Shitashige et al., 2010), and inhibitors such as NCB-0846 and KY-05009 were built to exploit that, with Masuda et al., 2016 reporting that blocking it stripped colorectal tumors of their stem-like cells. None reached approval; the class stalled on poor selectivity and toxicity. A separate strand linked the same enzyme to scarring, showing it controls how collagen precursors are trafficked in liver fibrosis (Buchl et al., 2022).

Rentosertib arrived by a different route. Insilico Medicine’s target-ranking software placed TNIK first among kinase candidates drawn from lung-fibrosis multi-omics data, its generative chemistry system produced the leads, and the candidate was nominated roughly eighteen months later — with the company reporting that the same target had independently scored across multiple hallmarks of aging, which is the origin of the longevity interest (Ren et al., 2025).

That “discovered by artificial intelligence” framing is a claim about process, not about efficacy, and it is contested. Independent commentators note that TNIK already carried a decade of published Wnt and fibrosis evidence and that conventional medicinal chemistry did the lead optimization, so machine contribution is a matter of degree (Namboothiri et al., 2026). Both readings rest on the same record; the question of how much credit the software earns has not been settled by any experiment.

Expected Benefits

High 🟩 🟩 🟩

No benefit reaches High: every efficacy result rests on a single 12-week phase 2a trial, so no human clinical endpoint and no validated clinical surrogate has yet been reproduced in a second trial.

Medium 🟩 🟩

Preserved or Improved Lung Capacity

Forced vital capacity (FVC, the total volume a person can forcibly exhale, and the accepted surrogate for lung-fibrosis progression) rose on the highest dose while falling on placebo over 12 weeks. The mechanism proposed is suppression of the scar-forming transitions described above. Evidence basis: one randomized, double-blind, placebo-controlled phase 2a trial of 71 patients, all Chinese, 90% men, sponsored and co-authored by the developer. Gas-transfer and expiratory-volume measures did not move, which the authors attribute to small numbers and measurement variability (Xu et al., 2025).

Magnitude: Mean 12-week change of +98.4 mL (95% confidence interval, the range within which the true value probably lies, 10.9 to 185.9) at 60 mg once daily, against −20.3 mL (−116.1 to 75.6) on placebo; in patients taking no background antifibrotic drug (a medicine that slows scar formation, meaning nintedanib or pirfenidone) the change was +187.8 mL (68.6 to 306.9).

Low 🟩

Reduced Cough Burden ⚠️ Conflicted

Cough-related quality of life improved on the highest dose in one modeled analysis, but the trial calls its quality-of-life results largely inconclusive, with wide variation and no separation on walking distance (Xu et al., 2025). Net reading: one borderline result amid null patient-reported outcomes is not yet a benefit.

Magnitude: Direction favors rentosertib at 60 mg once daily only, where the modeled cough-questionnaire score rose relative to placebo (two-sided P = 0.0495, P being the probability that a difference this large would arise by chance alone); the trial reports no effect-size figure for this endpoint.

Speculative 🟨

Lower Predicted Biological Age on Protein-Based Clocks

Six independent blood-protein aging clocks all read lower after treatment, the split twice-daily schedule most consistently. Basis: exploratory re-analysis of 42 trial participants using unvalidated biomarkers, not measured health outcomes (Zhavoronkov et al., 2026).

Reduced Circulating Scar-Protein Signature

Treatment moved 326 blood proteins (Zhavoronkov et al., 2026); scar-associated ones including MMP10, LTBP2 and collagen I fell at higher doses (Xu et al., 2025). Basis: exploratory analysis; none is a validated target.

Lower Senescence-Associated Protein Burden

Proteins tied to cellular senescence (worn-out cells that stop dividing but keep emitting inflammation) rose on placebo, fell on treatment. Basis: exploratory blood-protein analysis; no senescent cells were counted (Zhavoronkov et al., 2026).

Antifibrotic Effect in Kidney and Skin

Oral and topical dosing reduced collagen, hydroxyproline and scarred area in rodent kidney-obstruction and skin-fibrosis models. Basis is animal work alone; no human study has examined rentosertib outside the lung (Ren et al., 2025).

Broad Anti-Inflammatory Effect

In a rodent lung-injury model it cut interleukin-1β, interleukin-6, interleukin-7 and tumor necrosis factor alpha more effectively than dexamethasone. Basis: animal work; no anti-inflammatory endpoint has been measured in people (Ren et al., 2025).

Benefit-Modifying Factors

  • Background antifibrotic therapy: The single largest modifier. Patients on no other antifibrotic gained roughly twice the lung capacity of the pooled group, while those already taking nintedanib or pirfenidone showed no significant change at all.

  • Baseline lung function: Enrolment required 25–80% of predicted gas-transfer capacity and at least 40–45% of predicted lung volume. Nothing is known about response in people outside that band, in either direction.

  • Baseline blood markers: Scar-associated serum proteins such as LTBP2 and MMP10 fell most in responders, and the sponsor proposes this panel as a response signature; it has not been tested prospectively as a selection tool.

  • Genetic variation in drug metabolism: Clearance runs through CYP3A4 and CYP1A2, so common variants — including the CYP1A2*1F allele that makes the enzyme strongly inducible — plausibly shift exposure and therefore effect. No pharmacogenetic analysis has been published.

  • Sex: No sex-specific efficacy estimate exists. The trial was 90% male, mirroring who gets lung fibrosis, so the benefit figures above are effectively a male dataset.

  • Age: Trial participants averaged 67 years and the phase 3 study admits only people aged 40 and above. Whether the aging-biology signal differs in healthy people decades younger is untested.

  • Pre-existing health conditions: Obstructive airway disease, meaningful pulmonary hypertension and recent respiratory infection were all excluded, so benefit in those groups is unknown rather than absent.

Potential Risks & Side Effects

High 🟥 🟥 🟥

No risk reaches High: no adverse event has been documented in more than one controlled trial — patient exposure is confined to a single 12-week phase 2a study, the two phase 1 studies having been run in healthy volunteers.

Medium 🟥 🟥

Liver Injury and Abnormal Liver Function ⚠️ Conflicted

Liver-related events were the leading reason patients stopped treatment, defined as ALT or AST (liver enzymes released when liver cells are damaged) above three times the upper limit of normal with doubled bilirubin, or ten times alone. Most affected patients also took nintedanib, itself liver-toxic, so rentosertib’s own contribution is unresolved; no such clustering appeared with pirfenidone, and neither phase 1 study in healthy volunteers showed a liver signal (Xu et al., 2025; Ren et al., 2025). Net reading: a real, dose-related hepatic signal, amplified by co-medication.

Magnitude: Protocol-defined liver injury in 2/18 (11.1%) at 30 mg twice daily and 1/18 (5.6%) at 60 mg once daily, against 0/17 on placebo; 7 of 54 rentosertib recipients (13.0%) discontinued for liver toxicity, four of those seven while also on nintedanib.

Diarrhea

A clean dose-response with no placebo cases at all, consistent with the compound’s activity on gut epithelial renewal through the Wnt pathway it suppresses. Together with liver events it accounted for most treatment discontinuations. Severity was generally manageable, but the trial ran only 12 weeks, so tolerability over the 52-week exposure now being tested is unknown (Xu et al., 2025).

Magnitude: 2/18 (11.1%) at 30 mg once daily, 3/18 (16.7%) at 30 mg twice daily and 5/18 (27.8%) at 60 mg once daily, against 0/17 (0%) on placebo.

Low Blood Potassium

Hypokalemia (potassium below the normal blood range, which can disturb heart rhythm and muscle function) was the single most frequent adverse event overall. It occurred on placebo too, so part of it reflects the underlying illness and its treatment, but rates ran higher on every rentosertib arm. It matters mainly because low potassium lengthens the cardiac repolarization interval the phase 3 protocol monitors (Xu et al., 2025).

Magnitude: 3/18, 5/18 and 3/18 in the 30 mg once-daily, 30 mg twice-daily and 60 mg once-daily arms respectively, against 2/17 on placebo.

Acute Worsening of the Underlying Lung Disease

Sudden severe deterioration occurred more often on the highest dose than on placebo. The authors propose that immune dampening may blunt the local response to infection, a known trigger. Numbers are far too small to separate signal from chance (Xu et al., 2025).

Magnitude: 3/18 (16.7%) at 60 mg once daily against 1/17 (5.9%) on placebo; all three treated cases were hospitalized for a mean of 23.3 days, while the placebo case was not hospitalized.

Low 🟥

Speculative 🟨

Increased Susceptibility to Infection

It suppresses interleukin-1β, interleukin-6 and tumor necrosis factor alpha in animals, and the trial authors propose blunted immunity as a mechanism. Basis: mechanism and isolated reports; no infection endpoint measured (Ren et al., 2025).

Heart-Rhythm Effects

The phase 3 protocol excludes prolonged QTcF (a heart-rate-corrected measure of cardiac electrical recovery), bars drugs that lengthen it, and schedules eight electrocardiograms. Basis: sponsor precaution; no rhythm signal reported (NCT07687459).

Consequences of Long-Term Wnt-Pathway Suppression

The pathway rentosertib blocks also sustains intestinal, skin and blood-forming stem cells. Basis: target biology rather than human observation; longest continuous human exposure is 12 weeks (Li et al., 2025).

Risk-Modifying Factors

  • Existing liver disease or fatty liver: The dominant modifier. Chronic liver disease of any Child-Pugh class (a severity grade for liver disease), fatty liver, and liver enzymes at or above 1.5 times normal are phase 3 exclusions.

  • Concurrent nintedanib: Four of the seven liver-related withdrawals occurred in people also taking nintedanib, itself a drug toxic to the liver. No comparable clustering appeared with pirfenidone.

  • Gilbert’s syndrome: This common inherited variant of the UGT1A1 enzyme (which conjugates bilirubin for excretion) raises bilirubin harmlessly and is a phase 3 exclusion, because it would mask or mimic the injury stop-rule.

  • Genetic variation in clearance enzymes: CYP3A4 and CYP1A2 activity varies widely between individuals, and poor clearance raises exposure and therefore hepatic and gastrointestinal risk. No pharmacogenetic data have been published.

  • Baseline blood markers: Gamma-glutamyl transferase at or above three times normal, alkaline phosphatase or bilirubin at or above 1.5 times normal, and estimated glomerular filtration rate below 60 mL/min/1.73 m² each disqualify a candidate.

  • Sex: No sex-specific safety estimate exists; the trial was 90% male. The only sex-differentiated parameter is the cardiac interval threshold, set at 450 ms for men and 470 ms for women.

  • Age: Participants averaged 67 years, an age band with reduced hepatic reserve and heavy co-medication, so the observed liver and electrolyte signals arise in exactly the group most vulnerable to them.

Key Interactions & Contraindications

  • Nintedanib (caution, close monitoring): Additive liver toxicity and additive diarrhea; most liver-related withdrawals occurred on this combination. Mitigation: intensified liver-enzyme monitoring, or separation of the two agents.

  • Pirfenidone (caution): Additive gastrointestinal upset, though no liver clustering was seen with this pairing. Mitigation: dose with food for tolerability only if exposure loss is accepted, or reduce pirfenidone.

  • Strong or moderate CYP3A4 inhibitors (absolute contraindication in trial; ketoconazole, itraconazole, clarithromycin, ritonavir): Raised exposure and therefore hepatic injury risk. Mitigation: a two-week washout before dosing is the protocol requirement.

  • Strong or moderate CYP3A4 inducers (absolute contraindication in trial; rifampicin, carbamazepine, phenytoin): Collapsed exposure and loss of effect. Mitigation: substitute a non-inducing agent, with a two-week washout.

  • Strong or moderate CYP1A2 inhibitors and inducers (absolute contraindication in trial; fluvoxamine, ciprofloxacin, omeprazole, tobacco smoke): Unpredictable exposure in both directions. Mitigation: avoid; current smoking is itself a trial exclusion.

  • QT-prolonging medications (absolute contraindication in trial; amiodarone, sotalol, macrolides, fluoroquinolones, ondansetron): Additive risk of dangerous heart rhythms, worsened by the drug’s own potassium depletion. Mitigation: substitution plus electrocardiogram monitoring.

  • Warfarin (absolute contraindication in trial): Excluded outright from the phase 3 protocol; the shared clearance route makes anticoagulation control unpredictable. Mitigation: use an anticoagulant not routed through the same enzymes.

  • Immunosuppressive medications (absolute contraindication in trial): Additive immune dampening on top of the compound’s own cytokine suppression, against a background signal of acute disease worsening. Mitigation: none identified short of avoidance.

  • Grapefruit, pomelo and Seville orange (absolute contraindication in trial): All inhibit CYP3A4 in the gut wall and raise exposure. Mitigation: the protocol requires abstinence for at least 48 hours before the first dose and throughout.

  • St John’s wort (caution): A potent over-the-counter CYP3A4 inducer that would reduce exposure as effectively as a prescription inducer. Mitigation: discontinue two weeks before starting.

  • Paracetamol and high-dose non-steroidal anti-inflammatories (caution): Additive hepatic load on a compound whose main dose-limiting toxicity is liver injury. Mitigation: cap paracetamol well below maximum and avoid chronic use.

  • Liver-toxic supplements (caution; high-dose green tea extract, kava, ashwagandha, comfrey, high-dose niacin): Additive liver injury risk. Mitigation: stop before starting and re-check liver enzymes if any is resumed.

  • CYP3A4-inhibiting supplements (caution; berberine, high-dose curcumin with piperine, quercetin, resveratrol): Raised exposure through the same route as prescription inhibitors. Mitigation: separate or discontinue; these are not captured by drug-interaction checkers.

  • Potassium-lowering supplements (caution; licorice root/glycyrrhizin, high-dose caffeine): Additive hypokalemia on top of the compound’s own signal, compounding rhythm risk. Mitigation: avoid, or track potassium and supplement it.

Populations who should avoid rentosertib:

  • Anyone with chronic liver disease of any severity (Child-Pugh Class A, B or C) or with fatty liver disease, including its inflamed form
  • Anyone with screening ALT or AST at or above 1.5 times the upper limit of normal, bilirubin at or above 1.5 times, gamma-glutamyl transferase at or above 3 times, or alkaline phosphatase at or above 1.5 times
  • Anyone with Gilbert’s syndrome
  • Anyone with an estimated glomerular filtration rate below 60 mL/min/1.73 m²
  • Anyone with a heart-rate-corrected QT interval above 450 ms (men) or 470 ms (women)
  • Anyone who has had an acute worsening of lung fibrosis within the previous 6 months
  • Current smokers, and anyone testing positive for cotinine (a nicotine breakdown product used to verify non-smoking)
  • Anyone with an active or suspected cancer, or any cancer within the previous 5 years
  • Anyone with human immunodeficiency virus infection or clinically significant viral hepatitis
  • Women who are pregnant or breastfeeding, and anyone unwilling to use highly effective contraception until 90 days after the last dose
  • Anyone with known hypersensitivity to serine/threonine kinase inhibitors

Risk Mitigation Strategies

  • Liver panel before and during treatment: ALT, AST, bilirubin, gamma-glutamyl transferase and alkaline phosphatase at baseline, then weeks 2, 4, 8 and 12, then quarterly — the cadence that detects the compound’s leading toxicity before it becomes injury.

  • Hard stop rule for liver enzymes: The trial ended treatment at ALT or AST of 3 times normal with bilirubin at twice normal, or at 10 times normal alone — its own definition of liver injury.

  • Avoidance of concurrent nintedanib: Four of seven liver-related withdrawals occurred on that pairing. Running rentosertib without background antifibrotic therapy also produced roughly double the lung-capacity gain.

  • Lowest starting dose with delayed escalation: 30 mg once daily produced no protocol-defined liver injury at all, against 11.1% at 30 mg twice daily; escalation to 60 mg follows 4 weeks of clean liver enzymes.

  • Potassium held at 4.0–4.5 mmol/L: Measured at baseline and with each liver panel, and corrected before any dose escalation, because hypokalemia was the most common adverse event and lengthens the cardiac recovery interval.

  • Two-week washout of interacting drugs: The protocol clears strong or moderate CYP3A4 and CYP1A2 inhibitors and inducers, and removes grapefruit, pomelo and Seville orange at least 48 hours ahead, preventing exposure spikes that drive hepatic injury.

  • Electrocardiogram before and periodically during treatment: Rules out pre-existing QT prolongation above 450 ms (men) or 470 ms (women) and catches additive rhythm risk from potassium loss and co-medication.

  • Early management of diarrhea: Oral rehydration, temporary dose reduction, or a split schedule at the first grade-2 episode prevents the dehydration-driven electrolyte loss that compounds hypokalemia.

  • Dosing held during respiratory infection: Acute worsening of lung fibrosis was numerically more common on the highest dose, and infection is a recognized trigger for it.

Therapeutic Protocol

No practitioner prescribes rentosertib: it is unapproved everywhere, and every regimen below is a trial protocol rather than clinical practice.

  • Standard regimen: 60 mg orally once daily as a tablet, the dose selected for the 52-week phase 3 study after producing the largest lung-capacity gain in phase 2a.

  • Competing regimen: 30 mg twice daily. It produced a smaller lung-capacity change but the most consistent movement on protein-based aging clocks, so the two regimens separate on which readout the evidence favors.

  • Competing regimen: 30 mg once daily. Lowest exposure, no protocol-defined liver injury, and no lung-capacity benefit — the arm where tolerability and efficacy diverged most sharply.

  • Who established these: All three regimens come from Insilico Medicine’s phase 2a program, run by Zuojun Xu’s group at Peking Union Medical College Hospital, Beijing; no independent group has proposed an alternative.

  • Time of day: Trials specify once-daily dosing without a fixed clock time. Morning dosing is the practical default, since it aligns the 1–2 hour peak with waking hours for symptom monitoring.

  • Half-life: 7.4 to 11.9 hours across single and repeated dosing, with no accumulation over seven days — short enough that once-daily dosing produces a genuine trough between doses.

  • Single versus split dosing: The half-life supports both. Once daily gave the better lung-capacity result; twice daily gave the better blood-protein result and a higher rate of liver discontinuation.

  • Food: Fasted administration. A high-fat meal delays the peak from one hour to three and cuts peak concentration by roughly 37%, and the trials dosed under fasted conditions.

  • Genetic polymorphisms: CYP3A4 and CYP1A2 variants plausibly shift exposure several-fold, and the CYP1A2*1F allele is strongly induced by smoke. No pharmacogenetic dosing guidance exists; none of the trials genotyped participants.

  • Sex: No sex-based dose adjustment has been studied. The trial population was 90% male, so female exposure and tolerability data are thin rather than reassuring.

  • Age: Every trial has set a minimum age of 40, and participants averaged 67. No dose reduction for older age is specified, despite the hepatic and electrolyte signals.

  • Baseline biomarkers: Transaminases, bilirubin, gamma-glutamyl transferase, alkaline phosphatase, potassium, kidney filtration rate and a cardiac interval measurement all gate entry; failing any of them means the regimen does not start.

  • Pre-existing conditions: Background nintedanib or pirfenidone must be stable for 12 weeks before starting, or absent for 8 weeks — mixing an unstable antifibrotic regimen with rentosertib is not a studied scenario.

Discontinuation & Cycling

  • Intended duration: Open-ended. The disease it treats is progressive and incurable, the phase 3 study runs 52 continuous weeks, and no stopping point has been defined for responders.

  • Withdrawal effects: None reported. Sixteen of 71 phase 2a participants stopped early, many abruptly, and no withdrawal syndrome, rebound or discontinuation event appears in the safety record.

  • Tapering: Not used and not required. The 7–12 hour half-life clears the compound within roughly two days, and every trial discontinuation to date has been abrupt.

  • Rebound of disease: Untested. Whether the lung-capacity gain is retained, plateaus or reverses after stopping has never been measured, because no trial has followed patients off treatment.

  • Cycling: Not recommended and never studied. The mechanism is continuous suppression of scar-forming signaling, and the phase 3 design assumes uninterrupted daily dosing.

  • Stopping for toxicity: The one established reason to stop. Liver injury by the protocol threshold, or intolerable diarrhea, prompted immediate permanent discontinuation in the trial rather than dose interruption.

Sourcing and Quality

  • No legitimate consumer source exists: Rentosertib is unapproved in every jurisdiction. The only pharmaceutical-grade material reaches people through enrolment in a sponsor trial; there is no prescription, compounding or named-patient route.

  • Research-chemical suppliers are not a substitute: Vendors such as MedChemExpress and MedKoo list the compound (CAS 2828567-39-9, formula C27H30FN7O) explicitly for laboratory use only, with no human-grade manufacturing, no release testing and multi-month lead times.

  • What to look for if material is obtained anyway: A certificate of analysis from an independent laboratory covering identity by nuclear magnetic resonance or mass spectrometry, purity above 98% by chromatography, and residual-solvent and heavy-metal testing.

  • Formulation matters: Every human trial used an oral solid dose of defined strength — capsules in phase 2a, tablets in phase 3. Raw powder cannot be dosed accurately at 30–60 mg without an analytical balance, and solubility was a known formulation problem.

  • An inhaled formulation is in development: A nebulized solution has cleared investigational-drug review, exploiting roughly 50-fold higher lung than blood concentrations in animals. It is not available outside that development program.

  • Counterfeit exposure is high: A widely publicised unapproved compound with no legal supply chain is a standard target for substitution and adulteration, and no third-party verification program covers it.

Practical Considerations

  • Time to effect: Blood-protein changes appear by week 4 and largely plateau there. The lung-capacity difference was measured at 12 weeks; no earlier functional readout has been reported.

  • Pitfall — combining with nintedanib: The most consequential error. It roughly halved the lung-capacity gain while accounting for most liver-related withdrawals, turning a favorable trade-off into an unfavorable one.

  • Pitfall — dosing with food: A high-fat meal cuts peak exposure by around 37%, silently converting a 60 mg dose into something closer to the 30 mg arm that showed no functional benefit.

  • Pitfall — skipping liver monitoring: The toxicity that limits this compound is asymptomatic until it is advanced, and it was the single leading cause of treatment discontinuation in the only patient trial.

  • Regulatory status: Investigational everywhere; no approval, and therefore no off-label route either. It holds United States orphan drug designation for lung fibrosis and a Chinese breakthrough therapy designation.

  • Cost and access: Effectively infinite outside a trial, since no supply exists. Trial participation is the only access route, and current enrolment is concentrated in China with a smaller study in the United States.

  • Cost incentives around competing drugs: Approved antifibrotics are expensive branded products, so a successful third entrant would reduce, not raise, payer costs — meaning institutional payers have no structural incentive to suppress it, and no guideline body has yet taken a position for or against.

  • No organizational position exists on either side: No professional society, patient-advocacy organization or guideline panel has endorsed or rejected rentosertib, so no body with revenue tied to the answer has shaped the evidence base described here.

Interaction with Foundational Habits

  • Sleep: Direction is none, on present evidence. No sleep-related adverse event was reported at any dose, and the 7–12 hour half-life means once-daily morning dosing leaves little drug at bedtime. The target enzyme is abundant at brain synapses, which makes central effects mechanistically possible but entirely unobserved in the human trials.

  • Nutrition: Direct and substantial. A high-fat meal delays peak absorption from one hour to three and reduces it by roughly 37%, which is why every trial dosed fasted. Grapefruit, pomelo and Seville orange are prohibited outright because they inhibit the same gut enzyme that clears the compound. Potassium-rich foods help offset the hypokalemia signal.

  • Exercise: Direction is none, on present evidence. Six-minute walk distance did not separate from placebo over 12 weeks in either direction. The Wnt pathway the compound suppresses contributes to muscle satellite-cell repair, so blunted training adaptation is a theoretical concern with no supporting data and no study designed to detect it.

  • Stress management: Direction is indirect. No cortisol or stress-axis endpoint has been measured. The plausible link runs the other way: the compound broadly suppresses inflammatory cytokines in animals, and chronic psychological stress raises those same cytokines, so stress reduction is complementary rather than interacting.

Monitoring Protocol & Defining Success

Before any exposure, the gating tests are a full liver panel, serum potassium, kidney filtration rate, a 12-lead electrocardiogram and complete lung function testing, because each of these defines an exclusion threshold in the sponsor’s own protocol rather than merely a baseline. A screening test for bilirubin-handling variants matters too, since an inherited harmless elevation would confound the injury stop-rule. Once dosing starts, the trial cadence is the liver panel and potassium at weeks 2, 4, 8 and 12, then every 3 months; the electrocardiogram at weeks 2, 4, 8, 12, 26, 39 and 52; and lung function at weeks 4, 12, 26, 39 and 52. Success is defined not as improvement but as a slowed rate of lung-capacity loss against the person’s own pre-treatment trajectory, with liver enzymes holding below three times the upper limit.

Biomarker Optimal Functional Range Why Measure It? Context/Notes
ALT 10–25 U/L (men), 8–20 U/L (women) Earliest signal of drug-induced liver stress Conventional labs flag only above roughly 40–55 U/L; the trial stop-rule is 3 times the upper limit with bilirubin at twice the limit, or 10 times alone
AST 10–25 U/L Confirms liver-cell injury alongside ALT Conventional labs flag only above roughly 35–40 U/L; also rises with muscle damage, so pair with creatine kinase if elevated in isolation
Total bilirubin 0.3–1.0 mg/dL Separates a harmless enzyme rise from true liver injury Gilbert’s syndrome raises it harmlessly and was a trial exclusion; fasting and dehydration raise it transiently
GGT 10–25 U/L (men), 8–20 U/L (women) Most sensitive marker of bile-duct and drug-induced liver stress GGT is gamma-glutamyl transferase; the conventional upper limit is roughly 50–70 U/L; phase 3 excludes 3 times that limit; alcohol raises it for days
Alkaline phosphatase 55–90 U/L Distinguishes a bile-duct pattern of injury from a liver-cell pattern The conventional range runs roughly 40–130 U/L; phase 3 excludes 1.5 times the upper limit; draw fasting, and interpret against bone turnover in older adults
Potassium 4.0–4.5 mmol/L The most frequent adverse event, and low levels lengthen cardiac recovery time Conventional range is 3.5–5.2 mmol/L; a hemolyzed sample falsely raises the result, so redraw before acting on a high value
eGFR Above 90 mL/min/1.73 m² Kidney clearance capacity; below 60 disqualifies from every trial eGFR is the estimated glomerular filtration rate; a cystatin C-based estimate beats creatinine in older adults with low muscle mass
QTcF Below 430 ms (men), below 450 ms (women) Detects additive rhythm risk from co-medication and potassium loss Trial exclusion thresholds are looser at 450 and 470 ms; correct potassium and magnesium before interpreting
FVC, percent predicted No established on-treatment target; track change from the individual’s own baseline, where a 2–6% shift is the accepted meaningful difference The primary efficacy measure in lung fibrosis Use the same machine, same technician and same time of day; day-to-day variability easily exceeds the treatment effect
DLCO, percent predicted No established on-treatment target; track change from the individual’s own baseline Gas transfer across the lung membrane, the second pillar of disease staging DLCO is the diffusing capacity of the lung for carbon monoxide; it did not move in the trial; correct for hemoglobin, since anemia depresses the result

Qualitative markers worth tracking alongside the laboratory values:

  • Cough frequency and how much it disrupts sleep and conversation — the one patient-reported measure that moved in the trial
  • Breathlessness during a fixed, repeatable task such as one flight of stairs
  • Stool frequency and consistency, since diarrhea is the most common reason people stop
  • Energy levels and exercise tolerance across the day
  • Appetite, nausea, right-upper-abdominal discomfort and any yellowing of skin or eyes, which are the late symptoms of the liver toxicity the blood tests are meant to catch first

Emerging Research

  • Phase 3 confirmatory trial: NCT07687459 randomises 320 patients across 47 Chinese centers to 60 mg once daily or placebo for 52 weeks, with annual lung-capacity decline as the primary endpoint. It is the first result that could raise any benefit above a single-trial grade.

  • Second phase 2a study outside China: NCT05975983 is recruiting 40 patients at sites including the United States, measuring adverse events plus concentrations of the parent compound and its two major metabolites. It addresses the phase 2a trial’s most cited limitation — an all-Chinese population.

  • Completed evidence base: The phase 2a trial NCT05938920 and the phase 1 study NCT05154240 are both complete and published; everything claimed for the compound in people traces back to these two records.

  • Aging clocks as trial endpoints: Zhavoronkov et al., 2026 argue that disease trials should carry aging readouts alongside clinical ones. The proposal is testable and could strengthen the longevity case, or dissolve it if the signal proves to be fibrosis improvement in disguise.

  • The confounder that could weaken the case: One scar-associated protein, LTBP2, was used by all six aging clocks and drove much of the apparent age reduction, so the clocks may be reading fibrosis rather than aging (Zhavoronkov et al., 2026).

  • Comparative positioning: The network meta-analysis of Xu et al., 2026 ranks rentosertib near the top for preserving lung capacity but rates the evidence low certainty on indirect comparison. Head-to-head trials against nintedanib would settle it and none is registered.

  • Inhaled delivery: A nebulized formulation has cleared investigational-drug review, aiming to exploit roughly 50-fold higher lung than blood concentrations seen in rats (Ren et al., 2025). If it works, the liver toxicity that limits the oral route may largely disappear.

  • Target biology beyond fibrosis: Ewald et al., 2024 map the same enzyme onto cancer, glucose and lipid handling, and neuronal signaling. That breadth is the source of both the longevity interest and the concern about suppressing it for years.

Conclusion

Rentosertib is an experimental oral medication that blocks an enzyme sitting at the junction of several scar-forming and inflammatory signaling routes. Its notability rests on two things: it was the first compound whose target and structure were both machine-generated and then tested in people, and the enzyme it blocks is tied to processes that also change with age.

The evidence is thin and young. One short, small trial in one country showed that people on the highest dose held or gained lung capacity while those on placebo lost it. Blood-protein analyses from the same trial read as younger and less scar-associated, but these are exploratory measures that have never been validated against any health outcome, and one scar protein drove much of that reading. Against this sits a clear, dose-related pattern of liver injury and diarrhea that ended treatment for a substantial minority, sharply worse when combined with an existing medicine that slows scarring.

Almost all of this evidence was produced and funded by Insilico Medicine, the company developing the compound, which does not make it wrong but does leave it unreplicated by any independent group. For someone weighing an unapproved medicine with a serious liver signal and no legitimate supply route, the honest position is that the longevity claim remains an unvalidated laboratory reading, and the one real clinical finding still rests on a single short study.

Top - Benefits - Risks - Protocol