Evidence base
What has actually been shown, and with which parameters
Every number carries its source. Where sources disagree, we show the conflict instead of quietly picking one value.
On this page
Who and what
| Origin | Michael P. Sheetz (mechanobiology), University of Texas Medical Branch, Galveston. Sheetz has since died. |
|---|---|
| Company | Mechanobiologics, Inc. (Santa Fe, NM): "Health & rejuvenation through ultrasound technology"; mechanical treatments for cancer, aging and injury. Presented at the 7th TimePie Longevity Forum (Shanghai, Sept 2026). |
| Key paper | Kureel SK et al. Rejuvenation of Senescent Cells, In Vitro and In Vivo, by Low-Frequency Ultrasound. Aging Cell 2025, doi:10.1111/acel.70008 [S1]. Preprint: bioRxiv 2022.12.08.519320, v1–v4 [S2]. |
| Related | LFU reverses insulin resistance and diabetes-induced muscle transcriptome changes in aged mice, AJP Endocrinol Metab [S6]. Ultrasound-mediated mechanical forces and tumor-cell apoptosis (same group) [S9]. |
| IP | US 2024/0001155 A1 · EP 4561696 A1 (pending) · US 12,714,885 "Systems and Methods for Immersion Mechanotherapy" (granted Aug 2026) · 3 more pending. See IP rights. |
Consolidated parameters
| Parameter | Aging Cell 2025 | Patent app. US20240001155 | OA trial NCT06562374 | Aging RCT NCT07168525 | TimePie 2026 slide (mice) |
|---|---|---|---|---|---|
| Frequency | 32.2 kHz optimal; 33 kHz > 39 kHz | 32.249 kHz; claimed 20–500 kHz | 33 kHz | "low frequency" | 33 kHz |
| Pressure | 4 kPa (hydrophone); preprint v2: "4 Pa" | 6–7 kPa; pulses 170 Pa–4.5 kPa+; 1–450 mW/cm²; <3 W/cm² | "4–10 Pascals" | not stated | 4–8 kPa |
| Duty cycle | 1.5 s on / 1.5 s off | 1.5/1.5 s (also 2–60 s) | 1.6 s / 1.6 s | — | 1.5 s |
| Session | 30 min | 2 min–2 h (example 30 min) | 30 min | 45 min | 30 min = 600 pulses |
| Schedule | mice: daily / every 2nd / every 3rd day; 1×, 1.3×, 2× power | 10 sessions in 1 month | every 2–3 days | 3×/week, 8 weeks | — |
| Geometry | target 7–10 cm above transducer; 4 L glass beaker; cylinder 13 cm h × 15.2 cm Ø; mesh | 9–10 cm; frequency sweep / beam steering vs standing waves | "generators encased in epoxy within the baths" | bathtub fitted with device | transducer at bottom; mice on wire mesh |
| Water | degassed; 35 °C cells, 32–35 °C mice | degassed, 32–35 °C | — | — | — |
| Drive | not stated | not stated | — | — | "varying power input 135–275 V" |
| Measurement | "calibrated needle hydrophone (ONDA MCT-2000)" — but the MCT-2000 is a cavitation meter, see below | — | — | — | — |
Pa vs kPa, peak vs RMS
Plane-wave intensity in water is I = p²/(ρc) for RMS pressure (or p²/2ρc for peak amplitude), with ρc ≈ 1.5 × 10⁶ Rayl.
| Quoted pressure | If peak amplitude | If RMS | Plausible? |
|---|---|---|---|
| 4 Pa | ≈ 5 × 10⁻¹⁰ W/cm² | ≈ 1 × 10⁻⁹ W/cm² | no below ambient noise; almost certainly a unit typo |
| 4 kPa | ≈ 0.53 mW/cm² | ≈ 1.1 mW/cm² | yes inside patent's 1–450 mW/cm² |
| 8 kPa | ≈ 2.1 mW/cm² | ≈ 4.3 mW/cm² | yes |
| 10 kPa | ≈ 3.3 mW/cm² | ≈ 6.7 mW/cm² | yes |
- Wavelength at 33 kHz in water ≈ 4.5 cm. A beaker or bathtub is only a few wavelengths across, so standing waves and hot/cold spots are the main reproducibility risk. The patent itself mentions sweep and beam steering.
- Mechanical index ≈ 0.02–0.04, far below diagnostic-imaging limits (1.9). The target is sub-cavitation mechanical stimulation, but at 33 kHz gas nuclei in non-degassed water can cavitate, so degassing is part of the dose.
- Energy: 30 min at a 50% duty cycle gives ~900 s of sonication, about 0.5–4 J/cm². Heating is negligible, and the bath temperature dominates.
- Three independent arguments for kPa. First, 4 Pa implies particle displacements of ~13 picometres and nanowatt acoustic powers — below ambient noise, and unmeasurable by a meter built to see 10–300 kPa; 4 kPa implies ~13 nm, matching the sister lab's interferometry. Second, the same group's cancer work runs at 39–166 mW/cm², i.e. 34–70 kPa, so a rejuvenation dose at 1/10 of that is plausible while 1/10,000 is not. Third, the millivolt field map matches an underwater hydrophone seeing kilopascals.
- But the authors' own artwork says Pa. The preprint methods, the figure axes and the trial registration all say Pa; only the final journal version says kPa, with no erratum. This 1000× ambiguity sits at the centre of the replication — question A1.
What the paper's supplementary material actually shows
The supplementary bundle of the Aging Cell paper contains the setup schematic, Figure S1A — essentially the same slide shown at TimePie. It settles several questions and opens others.
Figure S1A — the setup
- A generic bench generator, a beaker, and a vacuum flask on a magnetic stirrer labelled "Vacuum": degassing is by vacuum plus stirring, not boiling.
- Two beakers: mice on a mesh at the waterline, and a culture dish near the surface. In both, the transducer at the bottom is drawn as a truncated-cone potted puck.
- Printed settings: varying power input 135–275 V · 32,248 Hz · 1.5 s duty cycle · 30 min (600 pulses) · 7–10 cm above transducer · output power 4–8 Pa.
- The 135–275 V is the electrical drive amplitude, not mains voltage: 275/135 ≈ 2.0, which maps exactly onto the 1× / 1.3× / 2× power groups of the mouse study.
Figure S1B — the field map
- The field across the treatment plane was mapped in millivolts of meter output, from "<20 mV" to "~180 mV".
- That is a 9× spread across the sample plane. The quoted "4 kPa" is a point value inside a strongly non-uniform field.
- An underwater-acoustics hydrophone (≈28 µV/Pa) reading 20–180 mV corresponds to roughly 0.7–6.4 kPa — which independently supports the kPa reading.
The optimisation experiments are equally informative. Growth was compared at 33 vs 39 kHz across pressures labelled 2.2 / 3.8 / 7.2 / 10 "pa", then at 33 kHz across 4 / 5.2 / 6.4 / 8 "pa". Duty cycle was varied by ±0.25 s around 1.5 s, and 1.5/1.5 won. The optimum is sharp at the ±17% level in duty cycle, and narrow in both frequency and amplitude — a strong claim for a mechanism nobody has identified, and the most replication-critical detail after pressure.
The measuring instrument is not what the paper implies. The ONDA MCT-2000 is a cavitation meter used with a cleaning-tank hydrophone probe, not a calibrated needle hydrophone. Its sibling's datasheet describes the output as a time-averaged total pressure over 1–60 s, and states that "kPa units require self-calibration to an absolute reference". So the published "4 kPa" is a broadband, time-averaged reading — neither peak nor RMS in the classical sense — and if that self-calibration was never done, the reading is in arbitrary units. That fits the field map being labelled in millivolts. A replication using a calibrated hydrophone and an oscilloscope will not automatically measure the same quantity.
The hardware, from the sister papers
No version of the senescence paper ever names the transducer, generator or amplifier. The same group's cancer and stem-cell papers do, and they are copyable:
| Branch | Apparatus |
|---|---|
| Sheetz / Singapore (Tijore et al. 2025, cancer) | Beijing Ultrasonic PZT4 piezoceramic rings, 25×10×4 mm and 16×8×4 mm, with an aluminium cone to widen the field to about 5 cm and "attenuate the amplitude to form plane waves", epoxy and silicone-rubber coated, glued to the tank bottom; generator and amplifier outside; samples 8 cm above; tank in a 37 °C incubator; degassed deionised water. |
| Tijore lab, IISc Bangalore (2026) | APC International Langevin transducer APC 90-4050, epoxy-bonded to a steel container on a 3-D-printed platform with height adjustment, driven by a PiezoDrive PDU 210, calibrated with an impedance analyser, pressure measured with a hydrophone into a Digilent Analog Discovery USB oscilloscope, vacuum-degassed water. The desired pressure is set by adjusting the drive voltage. |
The same hardware family is used by the group's cancer work, at 20–70× the intensity, to kill tumour cells rather than rejuvenate senescent ones. That comparison gives us the dose ladder and a biological cross-check on calibration: see the cancer branch.
That second rig is directly copyable, and its papers publish a pressure-to-displacement calibration measured by laser interferometry: 1 kPa → 11 nm, 5 kPa → 65 nm, 10 kPa → 118 nm, 20 kPa → 236 nm of dish-glass displacement. In the same work at 39 kHz, 2–5 kPa promotes stem-cell growth while 10–20 kPa kills — independent evidence of a narrow window in the low kPa range.
Biological claims
In vitro (one 30-min session)
- Senescent cells (>90% SA-β-gal+) resume growth: EdU, divisions, population doublings up to passage 24.
- Reversed: SA-β-gal, p16, p21, cell size, 5mC, H3K9me3, γH2AX, nuclear p53, ROS/mitoROS.
- Telomere length increases.
- SASP blocked: IL-6, IL-8, IL-10, IL-15, TNF-α, IFN-γ, VEGF, MIP-1α.
- RNA-seq: 50 genes up ≥2×, 140 down (incl. IGF2, IGFBP2, FGF7, C1QTNF7).
- Proposed mechanism: Ca²⁺ entry → autophagy ↑, mTORC1 ↓, SIRT1 moves from nucleus to cytoplasm, mitochondrial fission ↑.
In vivo (C57BL/6J, 22–25 months at start)
- SA-β-gal area in kidney/pancreas: ~70% (sham) vs 10–20% (LFU); p16/p21 staining down.
- Wheel running 7–10× sham at 29 and 32 months; treadmill improved in all groups; inverted cling in 3 of 5.
- Weight and fur maintained, while sham mice lost weight and developed bald spots.
- Lifespan: D2+D3 combined ≈50% alive at 1000 days; individual groups (n = 7–8) not significant.
- No tumors at autopsy (small n, very old animals).
- Dose–response is non-monotonic: the best survivors received the lowest doses, every 2nd or 3rd day. More is not better.
- A third paper reports ~4× lower liver burden after a Salmonella challenge in treated aged mice, with chemokines restored toward young levels [S10].
The second mouse paper is largely negative, and that matters. In aged mice made insulin-resistant by diet plus streptozotocin, LFU for 30 min twice a week for 4 weeks (32.248 kHz, 4 kPa, 1.5 s on/off, 37 °C degassed water) did not improve body composition, blood glucose control or physical function. What changed was the muscle transcriptome: 260 of 503 diet-induced gene changes were reversed, with the shared signature dominated by immune and B-cell genes plus Socs1/Socs3. The authors conclude that LFU acts on inflammation and immune-cell function in muscle, not on glycaemic control or performance. Any claim that LFU "reverses diabetes" is not supported by the paper that tested it [S6].
Human trials
NCT07168525 · Healthy aging recruiting
UT Health San Antonio (Barshop), PI Blake Rasmussen, XPRIZE collaborator.
- n = 20, age ≥ 70, randomized, participant-blinded sham (tub without ultrasound).
- 45 min × 3/week × 8 weeks in an ultrasound bathtub.
- Primary: knee-extension strength (Biodex).
- Secondary: SPPB, TUG, 6MWT, blood senescence (IHC/biochemistry/RNA-seq), immune aging (epigenetics), cognition (face-name memory), DEXA lean and fat mass.
- Estimated primary completion: Oct 2026.
NCT06562374 · Knee osteoarthritis suspended by IRB
UTMB Galveston, PI Thomas Blackwell.
- n = 10 (actual), single arm, open label; "non-significant-risk device".
- 33 kHz, "4 or 10 Pa" (sic), 1.6 s on/off, 30 min, every 2–3 days; transducers potted in epoxy within the bath.
- Primary: participant-reported safety; secondary: survey-based joint response.
- Suspended in March 2026. The reason is not public.
Strength of evidence: our reading
- Single lab and no independent replication. Small mouse groups, and lifespan is not significant per group.
- The only human safety pilot was suspended by its ethics board. The RCT (n = 20) has not reported.
- The dose is ambiguous across sources (Pa/kPa, peak/RMS, 1.5/1.6 s, 30/45 min). Nobody outside the group can currently reproduce it. That makes an open, calibrated device a genuine scientific contribution.
- The in-vitro phenotype is true reversal with proliferation and telomere lengthening, the category with the highest theoretical oncogenic concern. See risks.
- Conflicts are disclosed and material: several authors are inventors on the patents, and two have financial interests in the company planning to market the devices.
- The scale-up is unvalidated. Everything published uses a single transducer under a 4 L beaker with half a mouse in the water, or a dish 7–10 cm above the source. A human in 200–300 litres with many sources is a different acoustic problem.
- A deflationary hypothesis nobody has tested: at 33 kHz the wavelength is 4.6 cm and the beaker is 15.2 cm across, so the "optimum" could be a standing-wave mode of that particular beaker rather than a property of the cells.
Sources
- [S1] Kureel et al., Aging Cell 2025. PMC12151899 · Wiley
- [S2] bioRxiv 2022.12.08.519320. v4
- [S3] US20240001155A1
- [S4] EP4561696A1
- [S5] mechanobiologics.com
- [S6] AJP Endocrinol Metab, doi 10.1152/ajpendo.00470.2024
- [S7] NCT06562374
- [S8] NCT07168525 · UT Health XPRIZE news
- [S9] Tijore et al., Ultrasound-mediated mechanical forces activate selective tumor cell apoptosis, Bioeng Transl Med 2025 · Saigaonkar et al., ACS Nanosci Au 2026 — the APC Langevin + PiezoDrive rig
- [S10] Zangoui et al., LFU reduces susceptibility to Salmonella infection in aged mice, J Bacteriol 2025
- [S11] Onda Corporation — MCT-series cavitation meters and HCT cleaning-tank hydrophones