Pith. sign in

REVIEW 5 major objections 5 minor 5 references

A Functional Human Liver Tissue Model: 3D Bioprinted Co-culture Discoids

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Thin 3D-bioprinted liver discs with endothelial cells in a microgel bed keep secreting albumin and urea and metabolizing drugs for 3-4 weeks without perfusion.

desk verdict Solid functional validation of a microgel-printed liver discoid platform; the 'no perfusion needed' claim rests on an oxygen-transport assumption that the paper doesn't directly test. read the letter →

arxiv 2501.00086 v1 pith:KXHHCSMF submitted 2024-12-30 cond-mat.soft physics.bio-phq-bio.TO

Vignesh Subramaniam , Carolina Abrahan , Brett R. Higgins , Steven J. Chisolm , Baleigh Sweeney , Senthilkumar Duraivel , Leandro Balzano-Nogueira , Glyn D. Palmer
show 1 more author
Thomas E. Angelini
This is my paper · ORCID
classification cond-mat.softphysics.bio-phq-bio.TO
keywords livertissuemodeldiscoid3DbioprintingmicrogelsupportmediumcellcultureADMEgeneexpressionhepatocyte-HUVECco-culturedrugmetabolismassay
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that a 3D-printed liver model can be made simple enough for routine drug screening without giving up liver-like function. The tissues are 200-micrometer-thick discs (discoids) of primary hepatocytes (liver cells) plus vessel-lining HUVECs, printed into a bed of packed spherical PEG microgels; the same bed that shapes the print then acts as the culture medium. Over 3-4 weeks the discs keep producing albumin and urea at stable rates, express more than 100 genes involved in absorbing, distributing, metabolizing, and excreting drugs (ADME) at levels comparable to human liver tissue, and convert probe drugs into metabolites through CYP and UGT enzymes. The authors report that the process is fast (about 100 tissues per hour) and dimensionally precise (about 4% diameter error), and that tissues can be assayed in standard 96-well plates without perfusion. If these results hold, the model offers a practical, reproducible platform for toxicity and metabolism testing.

What carries the argument

The central object is the packed-microgel support medium: spherical polyethylene glycol (PEG) beads about 6 micrometers in diameter, packed at 5% w/w into a yield-stress solid, a material that flows only when pushed hard enough, with a storage modulus around 100 Pa and a yield stress around 2 Pa. It holds printed structures in place during fabrication, yet its interstitial pores give it a permeability that follows Kozeny-Carman scaling, $k \sim d^2$, and is 15-40 times higher than irregular-particle packs. The second piece is the print path: a planar spiral that lays down a 200-micrometer-thick collagen-cell disc, a geometry that keeps the tissue within diffusion distance of the surrounding medium. The third piece is the H:Hu 2:1 cell ratio, whose cell-cell and cell-matrix contacts maintain hepatocyte phenotype.

What would settle it

Look for a hypoxic or necrotic core in a 2.5-mm H:Hu discoid at day 21: stain with pimonidazole or HIF-1alpha, or measure an oxygen gradient with a microelectrode; if the center is hypoxic while albumin and urea output remains high, diffusion alone is not the mechanism keeping the tissue alive.

Watch

Extended reading notes

Core claim

The core discovery is that geometry and support material, rather than perfusion, carry the model. A 200-micrometer-thick disc keeps every hepatocyte close to the culture medium, and the loosely packed spherical PEG microgels leave micron-scale pores that make the bed 15-40 times more permeable than the irregular-particle packs used in earlier embedded-printing media. In this environment, hepatocyte-HUVEC (H:Hu) co-culture outperforms hepatocyte-only, cholangiocyte-containing, and spheroid formats: it produces higher albumin and urea, holds cell numbers steady for 21 days, and its ADME transcriptome clusters closest to human liver tissue, closer than the established HepatoPac comparison. After 14 days the printed discs still form metabolites through CYP1A2, CYP2C9, CYP2D6, CYP3A4, and UGT1A1, and adding Y-27632 plus HGF roughly doubles albumin output, raises urea, and boosts UGT1A1 activity by about 2.5-fold.

Load-bearing premise

The entire no-perfusion claim rests on the assumption that a 200-micrometer-thick, 1.5-2.5-millimeter-wide cell disc stays within oxygen and nutrient diffusion limits for 21 days, because the study reports stable DNA-based cell counts and function but no direct oxygen, hypoxia, or central-necrosis measurements.

Editorial extensions

If this is right

  • Because the microgel bed doubles as the culture medium, tissue fabrication, culture, and assaying happen in the same 96-well plate, removing a barrier to industrial adoption.
  • The size comparison shows 1.5-mm discoids make the most urea; cell density and tissue diameter are therefore controllable inputs for matching a desired metabolic output.
  • HUVEC co-culture is the composition that works best; adding cholangiocytes suppresses albumin and urea, so cell-ratio optimization is a direct route to tune the model.
  • Y-27632 plus HGF supplementation increases albumin about 2-fold, urea about 1.5-fold, and UGT1A1 metabolite formation about 2.5-fold, giving a chemically simple way to boost function.
  • The permeability result, $k \sim d^2$ with spherical beads 15-40 times more permeable than irregular ones, means the support medium can be adjusted if future designs need perfusion or larger constructs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper never directly measures oxygen or hypoxia; a day-21 hypoxia stain of the largest (2.5-mm) discs would settle whether the no-perfusion claim holds at the core or only on the outer shell.
  • The HUVEC benefit is left mechanistically open; an immediate test the authors do not run is whether HUVEC-conditioned medium alone reproduces the albumin and urea boost, which would separate soluble-factor signaling from direct cell-cell contact.
  • If the ADME profile really sits within the human-liver range, the model should be challenged with a blinded panel of known hepatotoxic and non-toxic drugs to see whether metabolite formation and albumin/urea decline predict toxicity.
  • The Kozeny-Carman scaling the authors fit implies that larger microgels could support thicker or denser tissues; this is a design rule they observe but do not exploit, and it is directly testable by printing the same discoid in different bead sizes.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The manuscript describes a 3D bioprinting approach for producing human liver tissue models ("discoids") — 200 µm thick and 1–3 mm in diameter — printed into a packed PEG microgel support medium. The authors report precise and accurate printing (≈4% diameter error, ~100 tissues/hour), stable albumin and urea synthesis over 3–4 weeks for hepatocyte/HUVEC co-cultures, size- and geometry-dependent functional differences, ADME gene expression clustering close to human liver controls, and LC-MS-measured CYP/UGT enzyme activities after 14 days. The paper argues that the high hydraulic permeability of the spherical microgel pack allows long-term culture without perfusion.

Significance. If the claims are substantiated, the system would provide an accessible, static-culture, millimeter-scale liver model suitable for pharmaceutical ADME and toxicology screening. The strengths include direct LC-MS-based measurement of five enzyme activities, benchmarking against freshly thawed hepatocytes, liver biopsy tissue, and HepatoPac, and a systematic comparison of co-culture compositions. The printing precision statistics are also a useful contribution to embedded bioprinting. However, the central claims of perfusion-free culture and functional equivalence to human liver rest on indirect or incomplete evidence (no oxygen/hypoxia data, DNA-based viability, limited ADME statistics), so the current version does not yet establish these claims at the level asserted.

major comments (5)
  1. [Section 1, Section 2.4, Fig. 5c] The central claim that 200-µm-thick discoids up to 2.5 mm in diameter can be cultured without perfusion is not supported by the presented data. The authors attribute the design to diffusion limits but provide no oxygen gradient measurements, hypoxia markers (e.g., HIF-1α, pimonidazole), or evidence of central necrosis; the only viability-related metric is total DNA content. Moreover, the size-dependence in Fig. 5c — urea synthesis significantly lower in 2.0 and 2.5 mm discs than in 1.5 mm discs at D7, D14, and D21 — is consistent with an emerging diffusion limitation, yet this interpretation is not tested. Compaction of H:Hu constructs to ~1.5-fold smaller area by D10 (Fig. 4d) may increase thickness and effective diffusion distance, and no post-print thickness data are reported. The claim that the microgel medium 'does not need to be perfused' therefore requires direct transport evidence or at minimum regional live/dead analysis.
  2. [Section 4.5, Fig. 4a, Fig. 5d] Cell numbers are estimated from PicoGreen DNA content, which cannot distinguish live from dead cells. Stable DNA-based cell counts therefore do not establish that viable cell numbers remain constant for 21 days; dead or lysed cells can contribute to the DNA signal, and proliferation can mask death. The conclusion in Section 2.3 that 'cell number within each group did not significantly change over the culture duration' and the related 'stable function' claims require direct viability measurements (live/dead staining, LDH release, ATP content, or apoptosis markers). Without them, the 21-day functionality claim rests on supernatant activity that could be supported by a viable rim.
  3. [Abstract, Section 2.4, Fig. 5d-f] The abstract's claim that the discoids are 'outperforming spheroid tissue models' is not supported as stated. The comparison is only against spheroids printed from the same bioink in the same microgel medium, not against established spheroid culture models (e.g., primary hepatocyte spheroids or commercial systems). The sphere constructs also had ~4-fold higher cell numbers, and their lower per-cell albumin/urea synthesis and 46% cell-number decrease by D14 are from DNA-based measurements (see above). The claim should be limited to 'outperforming printed sphere constructs of the same bioink' or supported by a direct comparison with a recognized spheroid model.
  4. [Section 2.5, Fig. 6, Methods 4.7] The claim that the tissues express 'more than 100 genes associated with ADME at levels within the range of human liver' is not demonstrated. Fig. 6a shows relative expression for only a subset of genes; the heatmap (Fig. 6b) is based on hierarchical clustering of 115 probes from n=2 independent experiments, with batch correction and normalization. Clustering 'closer to liver controls than HepatoPac' is not a statistical equivalence test, and no criterion is given for 'within the range of human liver.' Quantitative thresholds (e.g., within 2-fold of donor liver for a defined gene set) and per-gene variability are needed to support this claim.
  5. [Supplementary S1, Discussion] The permeability measurements show that packed spherical PEG microgels are 15–40x more permeable to pressure-driven flow than irregular microgel packs, but this Darcy-scale hydraulic permeability does not directly measure diffusive transport of oxygen or nutrients under static conditions. Static-culture oxygen delivery is governed by the diffusion coefficient and the geometry of the tissue, not by the hydraulic permeability of the surrounding support. The argument that high permeability 'side-steps the need to perfuse' (Discussion) is therefore a non-sequitur unless a link between microgel porosity and effective diffusivity is demonstrated.
minor comments (5)
  1. [Section 2.2] The displayed equations for the spiral diameter are garbled (e.g., '2 2 44 mm cA d cdπ' and '22 m cAd π= +'); these need proper typesetting and clear definition of variables, including the calibration factor f = 1.1.
  2. [Methods 4.8] The probe substrate concentrations are given as 100 mM, 25 mM, 50 mM, 15 mM, and 10 mM; these are almost certainly intended as µM and should be corrected, since mM concentrations would be cytotoxic and outside normal assay ranges.
  3. [Supplementary Fig. S2 caption] The caption lists 'Y-2732', which is a typo for Y-27632, as used in the main text (Section 2.6).
  4. [Methods 4.7] The formula for 'Theoretical Copy #: 2(40-CT value of target gene)/10K' is unclear; define the denominator and explain how copy numbers are normalized and then scaled.
  5. [Section 2.6, Fig. 7b] The 'data collapse' rescales each curve by its own fitted slope; this does not by itself demonstrate a linear scaling law. The raw concentration-vs-time data with linear fits would be more informative to the reader.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor fitted calibration factor in the diameter-accuracy metric; central liver-function claims are externally benchmarked and self-contained.

  1. fitted input called prediction [Section 2.2 (Equation for expected diameter de and Fig. 3e)]
    "To compare dm to expected diameter, de, we account for the width of the feature in a simple formula for a spiral, given by ... where φ is the total accumulated polar angle corresponding to a spiral of diameter de, and f = 1.1 is an empirically determined calibration factor. Plotting dm versus de for different spiral diameters and replicates, we found an average error in diameter of 4.2% with a standard deviation in the errors of 0.82%."

    The reported accuracy metric (4.2% diameter error) is computed by comparing measured dm to expected de, but de itself contains f = 1.1, an 'empirically determined calibration factor.' If f was adjusted on the same printed structures, the residual 4.2% error is partly minimized by construction rather than being an independent test of the print-path model. The parameter appears only in the geometry-calibration step, not in the albumin/urea, ADME, or CYP measurements, so the circularity is limited to the manufacturing-precision metric and does not affect the liver-function claims.

full rationale

The paper's main claims — stable albumin/urea synthesis for 3-4 weeks, ADME expression within the range of human liver, and CYP/UGT metabolite formation — are supported by direct measurements compared against external controls (liver biopsy tissue, freshly thawed hepatocytes, HepatoPac, and suspension controls). No fitted parameter enters these comparisons: ADME profiling uses a TaqMan array with standard normalization, and enzyme-activity rates are linear fits to LC-MS metabolite concentrations. The only fitted parameter, f = 1.1 in Section 2.2, calibrates the expected spiral diameter de used to report print-diameter accuracy; the 4.2% error is therefore somewhat softened by construction, but this affects a manufacturing metric, not any liver-function conclusion. Self-citations to prior microgel work are used for rheological context and are not load-bearing for the hepatic-function claims; no uniqueness theorem or ansatz is imported from the authors' prior papers. Hence no significant circularity beyond the minor geometry-calibration issue.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central liver-function claims are benchmarked against external references (liver biopsy tissue, freshly thawed hepatocytes, HepatoPac, suspension controls), keeping the ledger light. The fitted and hand-chosen quantities are the print-geometry calibration factor f = 1.1, the 5% PEG formulation, the assumed porosity 0.32, and the fixed 2:1 cell ratio. The load-bearing assumptions are biological and conventional: marker-based inference of DILI utility, constant DNA per cell for normalization, representativeness of the primary hepatocyte lot(s), and diffusion-limited viability of 200 micrometer thick constructs without perfusion; the last is supported only indirectly.

free parameters (4)
  • f (spiral diameter calibration factor) = 1.1
    Section 2.2: empirically determined calibration factor in the expected-diameter formula for spiral print paths. It is fitted to the same printing data used to report the 4.2% diameter error, so the accuracy statistic is partly internal to the calibration.
  • PEG concentration of microgel printing medium = 5% (w/w)
    Section 2.1: hand-chosen from rheology data (G' = 100 Pa, yield stress about 2 Pa) to balance printability and support. A design choice for the support medium, not independently optimized in this paper.
  • Porosity epsilon in Kozeny-Carman permeability check = 0.32 (assumed)
    Supplementary S1: assumed porosity close to loose packing of spheres, used to show consistency between measured permeability and the Kozeny-Carman relation. Supports the permeability-advantage claim, which underlies the perfusion-free design.
  • Hepatocyte:HUVEC co-culture ratio = 2:1
    Section 2.3 and Methods 4.3: cell ratio chosen for co-culture experiments and not systematically optimized; cholangiocyte-containing ratios are likewise fixed. The H:Hu ratio drives the headline functional results.
assumptions (5)
  • domain assumption Albumin and urea synthesis, ADME gene expression, and CYP activities are valid surrogate markers for pharmacological usefulness and DILI relevance of a liver model.
    Sections 1 and 3 benchmark the model on these markers (citing ref 21), but the paper does not validate the model against known hepatotoxicants, so the clinical and regulatory utility claim is inferred rather than demonstrated.
  • domain assumption DNA content per cell stays constant across culture groups and time, so PicoGreen-based cell counts are unbiased for per-cell normalization.
    Methods 4.5: albumin and urea rates are normalized to cell number estimated from DNA. Hepatocyte binucleation or differential HUVEC survival would bias the rates; the paper notes only 'some differences between groups' at D3 and D7.
  • domain assumption A 200 micrometer thick, up to 2.5 mm diameter disc in packed spherical microgels remains within oxygen and nutrient diffusion limits for 21 days without perfusion.
    Section 1 and Section 2.4 use this thickness to avoid perfusion; Supplementary S1 shows high permeability, but no oxygen, hypoxia-marker, or necrosis measurements support the 21-day viability claim directly.
  • domain assumption The primary human hepatocyte lots used are representative of human liver function.
    Methods 4.3: commercially cryopreserved primary hepatocytes are used; the ADME array used a matched lot, but the number of donor lots across functional experiments is not stated. Hepatocyte lot-to-lot variability is a known confound in such models.
  • standard math Standard exponential drainage model h(t)=C exp(-t/tau) and the Kozeny-Carman relation describe flow through the packed microgels.
    Supplementary S1: used to extract permeability k from column drainage and to check the d^2 scaling. Both are standard porous-media tools, appropriately applied.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Functional Human Liver Tissue Model: 3D Bioprinted Co-culture Discoids." pith.science (2026). https://pith.science/paper/KXHHCSMF

@misc{pith2026250100086,
  author       = {Pith},
  title        = {Pith review of: A Functional Human Liver Tissue Model: 3D Bioprinted Co-culture Discoids},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KXHHCSMF}},
  note         = {Machine review of arXiv:2501.00086}
}
read the original abstract

To reduce costs and delays related to developing new and effective drugs, there is a critical need for improved human liver tissue models. Here we describe an approach for 3D bioprinting functional human liver tissue models, in which we fabricate disc-shaped structures (discoids) 200 {\mu}m in thickness and 1-3 mm in diameter, embedded in a highly permeable support medium made from packed microgels. We demonstrate that the method is precise, accurate, and scalable; up to 100 tissues per hour can be manufactured with a variability and error in diameter of about 4%. Histologic and immunohistochemical evaluation of printed discs reveal self-organization, cell cohesion, and key liver marker expression. During the course of 3-4 weeks in culture, the tissues stably synthesize albumin and urea at high levels, outperforming spheroid tissue models. We find the tissues express more than 100 genes associated with molecular absorption, distribution, metabolism, and excretion (ADME) at levels within the range of human liver. The liver tissue models exhibit enzymatic formation of metabolites after exposure to multiple test compounds. Together, these results demonstrate the promise of 3D printed discoids for pharmacological and toxicological applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

  1. [1]

    Polyelectrolyte scaling laws for microgel yielding near jamming

    Bhattacharjee T, Kabb CP, O’Bryan CS, Urueña JM, Sumerlin BS, Sawyer WG, et al. Polyelectrolyte scaling laws for microgel yielding near jamming. Soft Matter. 2018;14(9):1559-70

    work page 2018

  2. [2]

    Jammed Polyelectrolyte Microgels for 3D Cell Culture Applications: Rheological Behavior with Added Salts

    O'Bryan CS, Kabb CP, Sumerlin BS, Angelini TE. Jammed Polyelectrolyte Microgels for 3D Cell Culture Applications: Rheological Behavior with Added Salts. ACS Appl Bio Mater. 2019;2(4):1509-17

    work page 2019

  3. [3]

    Fluid flow through granular beds

    Carman PC. Fluid flow through granular beds. Chemical Engineering Research and Design. 1997;75:S32-S48

    work page 1997

  4. [4]

    Methods to Characterize Granular Hydrogel Rheological Properties, Porosity, and Cell Invasion

    Qazi TH, Muir VG, Burdick JA. Methods to Characterize Granular Hydrogel Rheological Properties, Porosity, and Cell Invasion. ACS Biomaterials Science & Engineering. 2022;8(4):1427-42

    work page 2022

  5. [5]

    Determining Rates of Molecular Secretion from Supernatant Concentration Measurements in a 3D- Bioprinted Human Liver Tissue Model

    Anjum MR, Subramaniam V , Higgins BR, Abrahan C, Chisolm SJ, Krishnaprasad KA, et al. Determining Rates of Molecular Secretion from Supernatant Concentration Measurements in a 3D- Bioprinted Human Liver Tissue Model. ACS Biomater Sci Eng. 2024

    work page 2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.