ISSUE 005 · Oct 7, 2026 · 8 min read

What if transplanted islet cells could signal distress before glucose control fails?

After stem-cell-derived islet transplantation, connect CGM, insulin use, C-peptide, immunosuppressant levels, and infection signals to test a clinical surveillance service that separates early graft decline from treatment toxicity.

The issue in one sentence

Diagram 01

From glucose drift to cause-specific action

01Islet infusion→

02CGM · insulin dose→

03C-peptide · drug levels→

04AI risk trajectory→

05Transplant-team review→

06Cause-specific workup→

07Function · safety

ANM proposed model. A glucose alert never becomes a treatment order by itself; multimodal signals and clinician review route patients to testing, medication review, or infection assessment.

01Clinical problem · Evidence

Reaching insulin independence may be easier than protecting the graft safely over time.

For a narrow group with recurrent severe hypoglycemia and impaired awareness, islet-cell replacement may restore endogenous insulin secretion. Yet portal-vein infusion and long-term immunosuppression carry infection, hematologic, renal, bleeding, and thrombosis risks, while worsening glucose may reflect graft decline, intercurrent illness, medication effects, or sensor error.

02Technology stack · Proposal

Turn CGM into a shared control plane for graft function and immunosuppression safety.

Join CGM trajectories and insulin requirements with mixed-meal C-peptide, HbA1c, immunosuppressant troughs, kidney function, blood counts, and infection symptoms. AI does not diagnose; it classifies change patterns and prioritizes transplant-team review.

03Research question · Hypothesis

Can surveillance detect graft decline earlier without provoking unnecessary immunosuppression?

Test whether multimodal surveillance plus clinician review advances action on adjudicated graft dysfunction by at least 14 days versus schedule-based follow-up, while keeping false alerts below one per patient-month and remaining noninferior for serious immunosuppression-related harm.

04First study · Proposal

Six transplant centers, 60 recipients, an 18-month pragmatic trial.

Randomize adults receiving allogeneic islet replacement for recurrent severe hypoglycemia and impaired awareness 1:1 to protocolized usual follow-up or multimodal surveillance with weekday transplant-team review. A blinded committee adjudicates dysfunction onset from stimulated C-peptide decline plus insulin requirement and time-in-range change. Co-primary outcomes are time from adjudicated onset to cause-specific action and false alerts per patient-month. Severe hypoglycemia, time in range, insulin dose, eGFR, infections, admissions, and patient burden are secondary outcomes.

05Defensible data · Hypothesis

The moat is not the glucose trace—it is the cell–drug–action linkage.

Link cell lot and dose, infusion details, CGM and insulin, C-peptide, immunosuppressant exposure, renal, hematologic, and infection markers, alerts, clinician decisions, actions, and outcomes on one patient timeline. This can reveal which workup is safe for which signal pattern.

06Venture thesis · Assumption

Start with customers who bear the highest cost of graft failure.

Initial customers are cell-therapy trial sponsors and islet-transplant centers. Bundle per-patient-year surveillance, medication-safety dashboards, alert-review audit trails, and regulatory-grade real-world evidence. Payer or outcomes-warranty models are later options only if prospective data show fewer admissions, retransplants, or staff hours.

Evidence map

The evidence gap between efficacy and safety burden

14 people

Recipients followed for at least 12 months in the 2025 stem-cell-islet interim analysis

10 of 12

Full-dose recipients not using exogenous insulin at day 365

3 people

Neutropenia, the most common serious adverse event in the same small study

57 planned

Expected enrollment in the phase 1/2/3 master protocol updated in August 2026

Evidence

Established evidence versus the untested venture thesis

Established: a sponsor-funded phase 1–2 interim analysis published in NEJM in 2025 found post-transplant C-peptide in all 14 recipients. All 12 full-dose recipients remained free of severe hypoglycemia through day 365 with HbA1c below 7% and time in range above 70%; 10 were insulin independent. However, the analyses were interim and not prespecified, with a small sample and short follow-up. Three cases of neutropenia and two deaths were also reported, so efficacy cannot be separated from the burden of immunosuppression. The registry has expanded into a phase 1/2/3 master protocol but still targets a narrow population with severe hypoglycemia and impaired awareness. FDA evidence for an approved deceased-donor islet product likewise confirms serious risks from chronic immunosuppression and portal-vein infusion. There is no evidence yet that AI surveillance lengthens graft survival or improves safety; the service and revenue model remain testable venture assumptions.

2025 NEJM phase 1–2 study Ongoing phase 1/2/3 registry FDA islet-cell approval record FDA label and safety warnings

Failure conditions and safety, ethics, and regulatory risks

Clinical failure condition

Do not scale active alerts if action is not advanced by at least 14 days, false alerts exceed one per patient-month, or serious harm increases.

Over-immunosuppression

The algorithm never orders a dose increase. Escalation requires infection assessment, drug levels, kidney function, confirmatory testing, and clinician authorization.

False reassurance and sensor error

Normal CGM does not exclude graft decline or infection. Missing data, acute illness, or discordant sensors trigger conventional testing and urgent pathways.

Rare events and overfitting

Few patients and graft-failure events create high overfitting risk. Before external validation, use explainable rules and conservative thresholds.

Fair access and selection

Prevent an expensive investigational therapy and its monitoring from being allocated by coverage, device literacy, or connectivity. Report non-enrollment reasons and subgroup alert performance.

Regulatory and business boundary

Software that specifically changes transplant or immunosuppression management may require medical-device review. Reject the per-patient-year business model if sponsors and centers do not reduce admissions, repeat testing, or audit burden.

Bibliography

References

Established facts are separated from research and venture hypotheses that still require validation.

01
Reichman TW, Markmann JF, Odorico J, et al. Stem Cell–Derived, Fully Differentiated Islets for Type 1 Diabetes. N Engl J Med. 2025;393(9):858–868.
A small, short-term, non-prespecified interim analysis: 10 of 12 full-dose recipients were insulin independent at day 365; three cases of neutropenia and two deaths were reported.PMID 40544428 · DOI 10.1056/NEJMoa2506549
02
ClinicalTrials.gov. A Safety, Tolerability, and Efficacy Study of VX-880 and VX-017 in Participants With Type 1 Diabetes.
A phase 1/2/3 master protocol in adults with impaired awareness and recurrent severe hypoglycemia; continuous CGM use is required.NCT04786262 · Updated Aug 27, 2026
03
U.S. Food and Drug Administration. FDA Approves First Cellular Therapy to Treat Patients with Type 1 Diabetes. June 28, 2023.
Evidence for a deceased-donor product: 21 of 30 single-arm participants were insulin independent for at least a year, while most had serious reactions related to infusion or immunosuppression.LANTIDRA · BLA 125734
04
U.S. Food and Drug Administration. LANTIDRA (donislecel-jujn) Prescribing Information. Revised June 2023.
Specifies permanent immunosuppression and monitoring for opportunistic infection, malignancy, anemia, bleeding, portal thrombosis, and graft rejection.BLA 125734

The next issue publishes every day at 9 AM KST.All issues