Cell therapies during machine perfusion may modify inflammation, tissue injury, and metabolism in donor organs, but evidence of sustained benefit after transplantation remains incomplete. Alemi, Edwards, and Papalois reviewed 31 studies spanning kidney, liver, lung, and heart perfusion. Their central finding is a gap between promising molecular or histological changes outside the body and consistent improvement in transplanted organ function.
Machine Perfusion as a Treatment Window
Ex vivo machine perfusion provides a controlled period in which a donor organ can receive oxygenated perfusate, undergo functional assessment, and potentially receive targeted treatment before implantation. The review examines interventions delivered during this period, rather than cell therapy administered to a donor or recipient in vivo.
Normothermic and hypothermic approaches were eligible. Machine perfusion itself can influence organ recovery, so the additional effect of a cellular product must be distinguished from the effects of perfusion, the preservation solution, and the underlying injury model.
What Counts as a Cell-Based or Cell-Derived Therapy?
Living-cell treatments included mesenchymal stromal cells (MSCs), multipotent adult progenitor cells, and organ-specific progenitor populations. Cell-derived products included extracellular vesicles (EVs) and secreted factors without intact viable cells. Organoid-based approaches used three-dimensional cellular structures representing selected tissue functions.
These categories are biologically different and should not be treated as interchangeable. MSCs may act through signals released into the surrounding environment rather than by replacing damaged tissue directly. EVs carry cell-derived signals, while organoid strategies may aim for more sustained cellular engraftment or structural repair.
How the Review Was Conducted
The protocol was prospectively registered on PROSPERO (CRD420261358704). Two reviewers independently screened studies and extracted data from MEDLINE, Embase, and Web of Science searches through April 6, 2026. Only English-language full-text publications were eligible.
The authors used narrative synthesis rather than pooling effect sizes because organ models, products, delivery methods, perfusion conditions, comparators, and outcomes varied substantially. The review therefore does not provide a single overall treatment effect or identify one best cell product.
| Organ | Included studies | Studies with a transplant phase | Reported post-transplant pattern |
|---|---|---|---|
| Kidney | 10 | 2 | Both classified as mixed |
| Liver | 11 | 3 | Two beneficial; one mixed |
| Lung | 9 | 2 | One beneficial; one mixed |
| Heart | 1 | 1 | Beneficial findings over a short observation period |
Cell Therapies During Machine Perfusion: Organ-Specific Findings
Kidney
Kidney studies frequently reported changes in injury biomarkers, inflammatory signaling, or cellular metabolism. Physiological improvement was less consistent: increased urine production in one model did not establish a general improvement in filtration. Both studies that continued to transplantation had mixed findings. Evidence of viable implanted cells or organoids also does not by itself demonstrate improved recipient outcomes.
One mechanistic experiment found that silencing CD73 removed an EV-associated protective effect, supporting a role for adenosine-related signaling in that experimental setting. This is stronger mechanistic evidence than a simple association, but it remains insufficient to establish a clinical treatment.
Liver
Liver models explored whole-cell therapy, EVs, organoids, and indirect delivery through a bioreactor. Some reported better injury markers, bile production, tissue appearance, or mitochondrial measures. Three studies extended to transplantation, with two classified as beneficial and one as mixed.
Six of the eleven liver publications came from one research group. Related models and overlapping investigators should not be counted as six independent replications. The review also notes uncertainty about whether some control cohorts were independently generated. These questions limit the strength of repeated positive findings without establishing that data were duplicated.
Lung
Lung studies repeatedly illustrated the difference between reduced inflammatory signals and better organ function. Several treatments changed cytokines or injury markers without consistently improving oxygenation, compliance, or vascular resistance. Other models showed clearer functional signals, including restoration of alveolar fluid clearance.
Of two studies continuing to transplantation, one showed improved tissue injury measures without better physiological function; the other reported better oxygenation and compliance and fewer primary graft dysfunction events in a porcine model. Delivery timing and the underlying injury model also influenced results.
Heart
Only one cardiac study was included. It used human umbilical cord MSC-derived exosomes during perfusion of rat donation-after-circulatory-death hearts, followed by heterotopic transplantation and 90 minutes of in vivo reperfusion. Mechanical function and several injury measures improved, but the short follow-up and experimental transplant configuration cannot establish durable graft recovery or human benefit.
Why Better Perfusion Biomarkers Are Not Enough
Only eight of the 31 studies included a transplantation phase. Four were classified as beneficial and four as mixed. These classifications combined different models, outcomes, and follow-up periods, so the split must not be interpreted as a 50% clinical response rate.
A cytokine reduction, a lower enzyme concentration, or improved tissue staining may indicate biological activity without proving sustained filtration, gas exchange, contractility, survival, or freedom from rejection. Conversely, a small experiment with a nonsignificant functional result may simply lack sufficient power. The relevant next step is adequately powered transplant follow-up with predefined functional outcomes.
Relevance for Perfusion Practice and Research
The review positions cell-based and cell-derived treatments as experimental adjuncts to machine perfusion. It does not support adopting a specific cellular product, dose, route, or treatment duration into routine clinical practice. Findings in discarded human organs should also be distinguished from outcomes in transplanted human recipients.
For perfusion research, reporting should separate delivery feasibility, biological activity, organ function during perfusion, and post-transplant performance. Perfusion conditions and appropriate perfusion-only controls are essential to understanding what a treatment adds. Independent replication and standardized functional measurements remain priorities, especially in cardiac perfusion, where the evidence rests on a single study.
Limitations and Disclosures
The evidence was dominated by animal and ex vivo experiments with heterogeneous methods and often very small groups. Sequence generation was unclear in 30 of 31 studies, allocation concealment in 29, and the review reported substantial uncertainty in other bias domains. Poor reporting does not prove poor conduct, but it limits confidence in the findings.
The reviewers applied an adapted SYRCLE risk-of-bias framework to human ex vivo organ studies as well as animal experiments, deviating from their original plan. That framework has not been specifically validated for isolated human organs. English-language restriction, inconsistent outcome reporting, related research groups, and commercial or intellectual-property relationships in some underlying studies add uncertainty.
The review authors report no external funding and no conflicts of interest. The review does not establish clinical safety or efficacy of the experimental interventions it summarizes.
Original Article
Alemi M, Edwards Z, Papalois V. Cell-Based and Cell-Derived Therapies Delivered During Ex Vivo Machine Perfusion for Solid Organ Reconditioning Prior to Transplantation: A Systematic Review. Journal of Clinical Medicine. 2026;15:7698. doi:10.3390/jcm15197698.
This independently written educational summary accompanies the unchanged open-access source PDF, published under CC BY 4.0. The cover is an AI-generated editorial illustration.
