Organoids versus hypertension: an epithelial readout

What the work claims

This is a completed observational cohort study with results posted, sponsored by the University of Florida with the National Heart, Lung, and Blood Institute as a collaborator. Recruitment targeted adults scheduled for elective colonoscopy at the University of Florida, classified as hypertensive or normotensive by the 2017 ACC/AHA definition, meaning systolic blood pressure of at least 130 mmHg and/or diastolic of at least 80 mmHg. The protocol planned 60 participants and enrolled 35: 16 in the high blood pressure group and 19 in the reference group, all ages 18 to 80, with healthy volunteers accepted¹.

The two registered primary outcomes map onto the two halves of each biopsy. The first is differential gene expression in gut epithelium, measured by RNA sequencing at the time of colonoscopy. The second is the growth rate of gut organoids derived from the same biopsies, assessed within days to weeks of culture using immunohistochemistry and fluorescence-activated cell sorting, which separates cells by surface markers as they pass one at a time through a laser. The study ran from 2021-03-05 to completion on 2024-04-21, with results first posted on 2025-09-29¹.

The posted results are minimal. For differential gene expression, the record lists two counts, 29 and 6, in units of "differentially expressed gene", with no gene identities, no direction of change, no fold-changes, and no statistics. For the organoid growth outcome, the results module contains a heading and no data at all. Baseline characteristics are posted: mean age 62 in the hypertensive group versus 55 in the reference group, and a sex split of 10 female and 6 male versus 15 female and 4 male. The study fell short of its own recruitment target, enrolling 35 of a planned 60¹.

How it works

The scientific premise is that hypertension is not purely a vascular disease: gut epithelial barrier function, immune tone, and microbial metabolites have all been implicated, and disentangling cause from correlation in a human gut is nearly impossible in vivo. The organoid arm of this study is an attempt at controlled disentangling. A biopsy taken during a clinically indicated colonoscopy is material that would otherwise be discarded; growing it as a three-dimensional epithelial culture lets the investigators observe the epithelium's intrinsic behavior, its growth dynamics and cell-type proportions, away from the live gut's immune infiltrate, vascular tone, neural input, and microbial cargo.

The exclusion criteria show how deliberately the design isolates that variable. Participants are excluded for antibiotic treatment within two months, for medications known to modify the gut microbiota including acid suppressants and proton pump inhibitors, for unwillingness to stop probiotics for two weeks, and for any history of inflammatory bowel disease, celiac disease, intestinal surgery, or chronic inflammatory conditions. In effect, the protocol scrubs the microbiome and immune-activation axes from both arms so that any residual epithelial difference between hypertensive and normotensive groups can be attributed to the epithelium itself or to systemic factors stable enough to persist in culture. That is a genuinely thoughtful use of organoids as a de-confounding instrument rather than a disease replica.

Where a skeptic should push

The load-bearing assumption is that an epithelial-intrinsic signature, if it exists, will survive culture and remain attributable to blood pressure status. The posted baseline table undermines the second half of that assumption before any organoid is grown: the hypertensive group averages 62 years old against 55 in the reference group, and the sex ratios differ too. Hypertension status is therefore partially confounded with age in a sample of 35, and no posted analysis adjusts for it. Gene expression differences between these groups could track age or medication exposure, which the record does not tabulate, as readily as blood pressure itself.

The deeper problem is that the hypothesis this study is built to test may be unanswerable in an epithelium-only system. The leading mechanistic story linking gut and blood pressure runs through microbial metabolites such as short-chain fatty acids, through immune cell traffic, and through vascular and neural compartments. A colon organoid deliberately stripped of microbes, immune cells, vessels, and nerves is a clean instrument precisely because it excludes the hypothesized mediators. It can detect an epithelial-intrinsic program, which is real information, but a null result would tell you little about the microbiome-driven hypothesis the study's own title invokes, and a positive result would not establish which in vivo compartment drives it. Add the sample size: 35 participants, 16 per arm, planned 60 and missed, split a colonoscopy population by a blood pressure threshold, is an exercise in low power for any expression signal with realistic effect sizes.

Finally, the results posting itself deserves skepticism as a practice. The registry marks this study as having results, but the posted content is two unexplained gene counts and an empty organoid endpoint. Any reader, including this one, is left unable to verify the direction, magnitude, or statistical standing of the central claim. An honest reading stops at what the record supports: the study was completed, the groups were small and age-confounded, and the posted results do not currently permit any scientific conclusion.

Organoids for diseases that are not cancer

The non-obvious implication is about market and methodology, not hypertension. The overwhelming majority of clinical organoid studies are oncology trials, because cancer drugs are expensive and patient selection pays. This record shows a second viable lane: chronic, non-cancer systemic diseases where the organoid is a cheap epithelial biosensor fed by tissue that clinics already discard. Colonoscopy programs, bariatric surgery specimens, and routine endoscopies are vast, already-consented-adjacent tissue sources. A platform that reads epithelial state from such byproducts could attach functional data to epidemiological cohorts at a cost oncology will never fund. The explicit de-confounding protocol here, scrubbing antibiotics, probiotics, and acid suppressants before sampling, is directly reusable and would raise the evidentiary value of any organoid study that compares patient groups.

The threat is the reductionist drift that the design makes visible. As organoid platforms get sold into drug discovery for cardiometabolic and inflammatory indications, the easiest model to industrialize is exactly this one: epithelium only, no immune system, no microbiome, no vessels. Companies will read efficacy in that system as efficacy in the disease. But if the therapeutic hypothesis involves microbial metabolites or immune trafficking, an epithelial monoculture is structurally blind to the mechanism, and it can produce clean, publishable, wrong negative results that kill programs for the wrong reason. The results-posting practice is the second threat: a registry that certifies "results posted" for two unexplained gene counts lets underpowered studies borrow the credibility of completed trials. For a field trying to convince payers and regulators that organoid readouts mean something, the gap between the badge and the content is corrosive.

The bottom line

Established by the record: a University of Florida team completed a 35-participant study deriving gut organoids and RNA-seq profiles from colonoscopy biopsies of hypertensive and normotensive adults, with deliberate exclusions that scrub microbiome-modifying exposures, and posted results consisting of two unexplained differential-expression counts and no organoid growth data. Asserted, not demonstrated: that the gut epithelium carries a hypertensive signature, since the posted data carry no direction, effect size, or statistics, and the hypertensive group is seven years older on average. What would confirm the claim is a full results disclosure or a peer-reviewed paper with adjusted analyses and per-gene data. What would break it is a properly powered replication finding that the apparent expression differences track age or medication rather than blood pressure status.