ITPP (Myo-Inositol Trispyrophosphate)

What the molecule is

The compound travels under two names, and they denote one molecule: OXY111A, the development code used throughout the clinical literature, and myo-inositol trispyrophosphate. A single PubChem record covers both, carrying OXY111A in its synonym list: CID 10439981, molecular formula C6H12O21P6, molecular weight 605.99, InChIKey HEDKSUBRULAYNO-UHFFFAOYSA-N. The FDA Global Substance Registration System files the same substance as UNII 116EYZ0PPX against CAS 802590-64-3, and ChEMBL1182171 carries the identical InChIKey. Structurally it is the myo-inositol ring with three pyrophosphate bridges closed across adjacent hydroxyl pairs. The NCATS regulatory aggregator, keyed to that same UNII, lists no approval year and no regulatory designation of any kind.

Three CAS numbers circulate under this name on catalogue and aggregator pages, and they are not interchangeable. Number 802590-64-3 is the free acid and the one the FDA registry carries. Number 23103-35-7 resolves to PubChem CID 10439980, the hexasodium salt of the cyclic form, formula C6H6Na6O21P6 and molecular weight 737.88. Number 623552-11-4 resolves to CID 71771705, formula C6H12Na6O24P6 and molecular weight 791.93, a hexasodium record whose connectivity is drawn open rather than cyclic and which differs from the cyclic salt by three molecules of water. A page pairing the weight 605.99 with the registry number 623552-11-4 has combined two records that PubChem holds separately.

Selectivity for the red cell is the property that separates this molecule from its parent. Inositol hexakisphosphate binds haemoglobin but does not cross the erythrocyte membrane. Duarte and colleagues reported in 2010 that entry of the trispyrophosphate was inhibited by DIDS and by NAP-taurine, implicating band 3, the erythrocyte anion exchanger; the report does not state the species of the red cells used. Ion chromatography put the maximum intracellular concentration at 5.5 millimolar, falling to the detection limit in NAP-taurine-treated cells, and the dissociation constant for binding to red cell ghosts at 1.72 x 10-5 molar. Band 3-mediated uptake is what makes the distribution red-cell-directed rather than general, and that characterisation rests on one paper from the originating laboratory.

What was measured in the cuvette

The founding experiment was a synthesis screen. Fylaktakidou, Lehn and colleagues prepared nine inositol tripyrophosphate salts in 2005 and measured their behaviour against free haemoglobin and against whole blood. Every compound bound free haemoglobin, and each crossed the red cell membrane to some degree; the oxygen-haemoglobin dissociation curves shifted towards higher oxygen partial pressures for free and for intracellular haemoglobin alike. No animal was involved. The report established that the chemistry does what its design intended in a cuvette, which is a narrower statement than the one usually attached to it further downstream.

Evans and colleagues tested in 2021 whether that shift survives the pH dependence haemoglobin already has. Ten-millilitre samples of red blood cells, whose species the report does not state, were incubated at 37 degrees for one hour across concentrations from zero to 240 millimolar, and oxygen affinity was measured in triplicate on a Clark electrode at pH 6.8, 7.4 and 7.6. Rising concentration and falling pH each raised P50, with p below 0.0001 for both. The effect appeared at pH 7.4 and at 6.8 and was absent at 7.6. Read as the authors read it, the Bohr effect was conserved: the compound added to the affinity drop already produced by tissue acidosis without altering affinity under conditions resembling the pulmonary microenvironment.

Koj and colleagues published a phosphorus-31 NMR study in 2025 that followed the 2,3-diphosphoglycerate resonances, whose positions report intracellular pH and oxygenation state, and described lowered red cell internal pH alongside increased oxygen release (PMID 39828634); that paper declares no competing interests while carrying affiliations to the company holding the patent and to the originating chemistry institute. Between a P50 shift in a cuvette and a reoxygenated tumour in a living animal there are several steps. The animal literature is where those steps were tested, and it is also where the results stop agreeing with one another.

The rodent oncology record, and the models where nothing happened

Aprahamian and colleagues reported in 2011 on an orthotopic syngeneic rat hepatocellular carcinoma model, comparing weekly intravenous administration against untreated animals and against doxorubicin. P50 shifted by 30 per cent, HIF-1 alpha activity fell, VEGF expression was down-regulated, caspase-3 was induced in tumour and surrounding liver, and microCT follow-up showed growth inhibition with long-term survival in almost all treated animals. Haematological parameters were unchanged. Kieda and colleagues in 2013 reported raised oxygen tension and blood flow in melanoma and breast models with selective activation of endothelial PTEN (PMID 23471434). Raykov and colleagues in 2014 reported restricted primary tumour growth and inhibited metastasis in rat and mouse pancreatic models (PMID 24214898). Limani and colleagues extended the work to colorectal liver metastasis in mice in 2016 and to metastatic colon cancer in 2017 (PMID 28742687).

Several models produced nothing, and one produced a worse outcome than radiation alone. Iyengar and Schwartz implanted syngeneic GL261 glioblastoma cells intracranially in C57Bl/6 mice and treated with hypofractionated radiation, with the compound, or with both. Radiation alone prolonged survival, the compound alone did not, and in combination it appeared to reduce the effectiveness of radiation. Foernvik and colleagues, indexed as Foernvik with the diaeresis restored in the reference list, implanted RG2 glioblastoma cells intracranially or subcutaneously in Fischer 344 rats and administered the compound intraperitoneally, intravenously, or by both routes together, reporting no prolonged survival by any route. Tran, Cao-Pham and colleagues found reoxygenation in all six tumour models they tested, no improvement when radiotherapy was combined with treatment in rhabdomyosarcoma, and a split result in 9L-glioma where some tumours were cured and others were not.

The radiotherapy-combination record is not uniformly null. Grgic and colleagues reported in 2021, from the same Zurich group that ran the human trial, that a high-dose bolus given before a single high-dose fraction of ionising radiation reduced tumour growth in multiple preclinical mouse models, immunocompromised and immunocompetent, while the compound alone did not affect growth; gamma-H2AX foci indicated increased initial and residual radiation-induced DNA damage within initially hypoxic regions, scheduling experiments showed the effect only when it preceded radiation, and irradiation-induced vascular damage was prevented by the combination. Tran and colleagues reported in 2025 that oxygen unloading measured by photoacoustic imaging in mouse CT26 and 4T1 tumours tracked with effector T cell frequencies, and that combination with immune checkpoint blockade improved tumour control and survival in both models (PMID 40594309).

A 2026 zebrafish xenograft study is the most recent entry and the most explicit about its own null result. Abou Khouzam and colleagues used pancreatic ductal adenocarcinoma xenografts in wildtype and transgenic lines, including a double transgenic showing vessels and erythrocytes together, treating by immersion at 5 millimolar for three or four consecutive days. Within that window more xenografts had erythrocyte-positive tumour-associated vessels, with no measurable change in vessel architecture, and apoptosis rose in vivo without cytotoxicity in vitro. The authors could not detect major effects and attributed that to the absence of a hypoxic microenvironment in the model. Reoxygenation is the part of this file that replicates across laboratories and techniques, including by electron paramagnetic resonance oximetry in mice (PMID 37167239) and alongside immune-infiltrate changes (PMID 33624446). Downstream tumour control and radiosensitisation did not replicate: positive in some models (Grgic 2021, PMID 33587991) and null or adverse in others (Iyengar 2017, PMID 28314273; Foernvik 2016, PMID 27793896; Tran 2019 rhabdomyosarcoma, PMID 30575283).

The one trial in people

NCT02528526 ran at a single centre in Zurich and is the only clinical study of this compound published anywhere. Schneider, Linecker and colleagues reported it in 2021: an unrandomised, open-label, three-plus-three dose escalation in 28 patients with advanced hepatopancreatobiliary malignancy or colorectal liver metastases, recruited between 27 April 2015 and 6 July 2018. Eighteen were male and ten female, median age 65. Diagnoses were pancreatic ductal adenocarcinoma in ten, colorectal liver metastases in eight, cholangiocarcinoma in seven and hepatocellular carcinoma in three. Twenty-five of the 28 had already received extensive prior anti-tumour therapy. Each patient was scheduled for nine eight-hour infusions across three weeks, over eight dose levels from 1,866 to 14,500 mg/m2 per infusion, followed by chemotherapy.

Toxicity concentrated in one finding. No dose-limiting event appeared in cohorts one through seven. In cohort eight, at 14,500 mg/m2, the third patient recorded a grade II hypercalcaemia of free ionised calcium and the fourth recorded grade II and grade IV events of the same kind, which set the maximum tolerated dose at the cohort seven level of 12,390 mg/m2. Of 56 adverse events recorded, 32 were judged at least possibly related to the compound. Hypercalcaemia occurred in 19 of 28 patients, 17 of them grade I. The investigators attributed it to the calcium chloride admixed to the infusion to offset the compound's calcium chelation, noting it began with infusion and normalised on cessation. Hypomagnesaemia occurred in five patients, hypophosphataemia in four, and grade I blood-pressure elevation in three.

Pharmacokinetics were measured across all eight cohorts. Peak plasma concentration rose with dose from 47.2 to 439.2 mg/L, reached at six hours from infusion start in every cohort, and fell rapidly afterwards. Half-life by terminal slope ranged from 1.3 to 3.3 hours and did not track dose. Concentration at the start of each infusion was below the detection threshold in every cohort, which the investigators read as clearance without systemic accumulation. Clearance, mean residence time and volume of distribution at steady state were similar across the dose range. These are the only human pharmacokinetic figures that exist for this molecule, and they describe an eight-hour intravenous infusion.

What the trial did not measure matters as much as what it did. Oxygen-haemoglobin dissociation was not assessed in any patient, and P50 does not appear in the report. Tumour hypoxia was not imaged and tumours were not routinely biopsied, because the responsible ethics committee precluded both, and the authors state that the trial is limited by the lack of direct evidence for anti-hypoxic action. Of 27 patients assessed for efficacy after monotherapy, follow-up imaging was unavailable for two; of the remaining 25, fourteen had stable disease by RECIST 1.1 and eleven progressed. After subsequent chemotherapy, 20 assessable patients returned 12 stable, six progressive and two partial responses. Median overall survival from the start of infusions was 206 days. The investigators raise the possibility themselves that the stabilisations reflect the natural course of disease in a small, heterogeneous, heavily pretreated group.

Exercise capacity, and the doping record

One paper supplies the entire performance literature. Biolo and colleagues reported in 2009 that intraperitoneal administration to normal mice at 0.5 to 3 g/kg produced a dose-related rise in P50, with a maximum increase of 31 per cent, and a dose-related rise in maximal exercise capacity, with a maximum increase of 57 plus or minus 13 per cent. In transgenic mice with severe heart failure from cardiac Gq-alpha overexpression the maximal increase was 63 plus or minus 7 per cent. Administration in drinking water raised P50, and raised maximal exercise capacity by 34 plus or minus 10 per cent, in normal and failing mice; no percentage is given for the oral P50 change. Myocardial contractility in isolated cardiac myocytes was unaffected, as was arterial blood pressure in vivo.

The cardiovascular strand extends past that single mouse study. Okninska and colleagues reported in 2021 on monocrotaline-induced pulmonary hypertension in male Wistar rats, measuring myocardial oxygen tension directly. Right ventricular pO2 was 32 plus or minus 5 mmHg in control animals and 18 plus or minus 9 mmHg in animals with pulmonary hypertension; a single intraperitoneal dose of 1.5 g per kilogram raised it to 26 plus or minus 5 mmHg, with five animals per group in that arm and 44 animals used across the study. Left ventricular pO2 did not change, and there was no effect on pO2 in control animals. The name is indexed as Okninska with the acute accent restored in the reference list.

No equivalent measurement exists in a person. A PubMed search combining the compound name with exercise, athlete, VO2 and endurance returned seven records in August 2026: six are doping-control assay development papers and the seventh is the 2009 mouse study. The phase Ib trial measured neither exercise capacity nor oxygen delivery, and its population was people with advanced cancer. Percentages quoted for human endurance gains are the mouse percentages, moved across species with no intervening experiment.

Detection chemistry developed faster than the clinical programme. Lam and colleagues gave 200 mg intravenously to a single Standardbred mare in 2014 and tracked elimination: plasma concentration peaked five minutes after administration and remained detectable for six hours, while urine peaked at 1.5 hours and stayed positive to 24 hours (PMID 23733541). Goergens and colleagues validated a hydrophilic-interaction Orbitrap assay for human urine the same year, with limits of detection of 15 ng/mL for screening and 1 ng/mL for confirmation, and reported stability in urine across the storage period mandated for doping-control laboratories. An ion-chromatography mass-spectrometry method covering this compound alongside ten bisphosphonates in equine plasma followed, published online in June 2024 and carried in the April 2025 issue of the same journal.

The 2026 World Anti-Doping Code Prohibited List, in force from 1 January 2026, prohibits artificially enhancing the uptake, transport or delivery of oxygen under M1.2. Its examples are perfluorochemicals, efaproxiral, voxelotor and modified haemoglobin products, introduced with the wording "including, but not limited to". This compound is not among the substances named anywhere in that document. In horse racing it has produced at least one recorded positive: the French laboratory LCH detected it in a urine sample from Akoya, winner of the Group 1 Prix Dragon for purebred Arabians at Longchamp in September 2019, with confirmation by a second laboratory in Mauritius, and the horse was disqualified. Racing trade press reported in January 2020 that the trainer denied wrongdoing and faced a one-year suspension; that account rests on trade coverage rather than on any primary regulatory document.

What is registered, and what is not

One registry record exists worldwide. ClinicalTrials.gov holds NCT02528526, sponsored by the University of Zurich, phase 1/2, single-group, estimated enrolment 69, start date February 2014. Its overall status reads unknown, its last known status recruiting, its estimated completion December 2016, and its most recent update was submitted on 18 August 2015. No results have been posted. The published report describes recruitment continuing to July 2018 and 28 patients enrolled, so the registry entry has been stale for roughly a decade and understates what happened. Searches of the EU Clinical Trials Register and ISRCTN for the compound name and for the OXY111A code returned nothing at all.

Two further human studies are asserted in the literature and appear in no registry. The 2016 protocol paper justifies its starting dose by reference to phase I data in healthy volunteers, attributed in parentheses to unpublished data obtained from the company developing the compound; the same source gives human-equivalent-corrected no-observed-adverse-effect levels of 15,000 mg/m2 in rats dosed once weekly for five weeks and 43,750 mg/m2 in minipigs also dosed once weekly. The 2021 trial report repeats the healthy-volunteer claim in its introduction. A 2022 review states that the compound is being tested in phase 2 trials in humans with various tumours, and the trial report itself closes by saying efforts for a phase 2 dose-extension are ongoing. Neither the healthy-volunteer study nor any phase 2 trial appears on ClinicalTrials.gov, the EU register or ISRCTN, and neither has been published.

Nothing here is approved for human use in any jurisdiction. The NCATS regulatory aggregator lists no approval year and no designation against the UNII, the single registered trial never posted results, and the one published clinical report is a dose-finding study in 28 people with advanced cancer that did not measure the mechanism it was testing. That is the whole of the human record.

What is not known

The human record is one open-label dose-escalation study of 28 people with advanced, heavily pretreated cancer, and it did not measure the mechanism under test. Oxygen-haemoglobin dissociation was not assessed in any patient, tumour hypoxia was not imaged and tumours were not routinely biopsied, because the responsible ethics committee precluded both; the authors state plainly that their trial lacks direct evidence of anti-hypoxic action. There is consequently no human measurement of P50, oxygen delivery, exercise capacity or tissue oxygenation at any dose. There is no controlled trial, no comparison against any active agent, no data in people without cancer, and no repeat-exposure or long-term safety data beyond the nine infusions and ten-day follow-up of that study. Pharmacokinetics are known only for an eight-hour intravenous infusion; no oral, sublingual or subcutaneous human pharmacokinetic data exist, and the only oral evidence in any species is mice given the compound in drinking water. The single registered trial has posted no results and its registry entry has not been updated since 2015. Preclinically, reoxygenation replicates across laboratories while what follows from it does not: a 2021 study in multiple mouse models reported enhanced radiation response after ITPP-induced oxygenation and a 2025 study reported improved control with checkpoint blockade, while two glioblastoma models produced no survival benefit, one suggested interference with radiotherapy, a rhabdomyosarcoma model showed no radiosensitisation, and a 2026 zebrafish xenograft study detected no major effect. Sample sizes are absent from most retrieved rodent reports. No study characterised reproductive or developmental effects, carcinogenicity or immunogenicity, and the rat and minipig no-observed-adverse-effect levels that set the human starting dose are attributed to unpublished company data that no one outside the sponsor has seen. Nothing here is approved for human use in any jurisdiction.

Questions