Type 1 diabetes — disease-modifying immunotherapy¶
TL;DR — Immune intervention can change T1D tempo but has not reliably produced permanent immune tolerance. Teplizumab is the clearest stage-2 result: one 14-day course delayed median stage-3 diagnosis from 24.4 to 48.4 months (HR 0.41, 95% CI 0.22–0.78) in 76 relatives (Herold 2019, PMID 31180194; NCT01030861). In new-onset disease, rituximab, abatacept, low-dose ATG, teplizumab, and verapamil preserved C-peptide to varying degrees, but surrogate preservation does not automatically yield less insulin, better HbA1c, or durable remission (Pescovitz 2009, PMID 19940299; Orban 2011, PMID 21719096; Haller 2018, PMID 30012675; Ramos 2023, PMID 37861217). The frontier is biomarker-directed sequence and combination therapy with acceptable long-term immune risk.
Endpoint map¶
| Disease point | Typical primary endpoint | Why it is hard |
|---|---|---|
| Stage 1 | Time to dysglycemia or stage 3 | Long follow-up; variable tempo |
| Stage 2 | Time to stage 3 | Delay may not equal prevention |
| New-onset stage 3 | Stimulated C-peptide AUC | Surrogate may not change daily outcomes |
| Established disease | Severe events, insulin dose, HbA1c/TIR | Little remaining β-cell mass |
Mixed-meal stimulated C-peptide is the accepted β-cell-function endpoint because it is reproducible and biologically closer to endogenous secretion than insulin dose alone (Palmer 2004, PMID 14693724).
Intervention evidence¶
| Agent/pathway | Setting | Quantitative result | Interpretation |
|---|---|---|---|
| Teplizumab, anti-CD3 | Stage-2 relatives | Median diagnosis 48.4 vs 24.4 months; HR 0.41 (0.22–0.78) | Proven delay, not proven permanent prevention |
| Teplizumab | New-onset children/adolescents | C-peptide difference 0.13 pmol/mL (0.09–0.17) at week 78 | Preserved secretion; key clinical secondary endpoints not different |
| Rituximab, anti-CD20 | New onset | Higher 1-year C-peptide AUC; lower HbA1c and insulin dose | B cells contribute; effect waned with reconstitution |
| Abatacept, CTLA4-Ig | New onset | C-peptide AUC 59% higher at 2 years; estimated 9.6-month delay in decline | Costimulation blockade modifies tempo |
| Abatacept | Stage 1 | HR 0.702 (0.452–1.09), P=0.11 for dysglycemia/stage 3 | Immune/metabolic effects without primary-endpoint success |
| Low-dose ATG | New onset | C-peptide AUC 0.646 vs 0.406 nmol/L at 1 year | Signal with systemic immune toxicity considerations |
| Oral insulin | At-risk relatives | Primary stratum negative | Antigen-specific prevention not established |
| Verapamil | New-onset children | 30% higher C-peptide at 52 weeks | β-cell stress target, not classic immunosuppression |
Sources: (Herold 2019, PMID 31180194; Ramos 2023, PMID 37861217; Pescovitz 2009, PMID 19940299; Orban 2011, PMID 21719096; Russell 2023, PMID 36920087; Haller 2018, PMID 30012675; TrialNet Oral Insulin Study Group 2017, PMID 29164254; Forlenza 2023, PMID 36826844).
Teplizumab as proof of disease modification¶
TN-10 randomized 76 high-risk relatives, 44 to teplizumab and 32 to placebo. Stage 3 developed in 43% versus 72% during the initial report; annualized diagnosis rates were 14.9% versus 35.9%, with rash and transient lymphopenia expected (Herold 2019, PMID 31180194; NCT01030861).
PROTECT randomized 328 recently diagnosed children/adolescents. Two 12-day courses increased stimulated C-peptide at week 78 by 0.13 pmol/mL (95% CI 0.09–0.17); 94.9% versus 79.2% retained peak C-peptide ≥0.2 pmol/mL, but insulin dose, HbA1c, TIR, and clinically important hypoglycemia did not significantly differ (Ramos 2023, PMID 37861217; NCT03875729).
| Safety/domain | Observed or operational issue |
|---|---|
| Infusion | Consecutive daily intravenous dosing |
| Immune effects | Transient lymphopenia; cytokine-release symptoms |
| Skin | Rash is common |
| Infection screening | Requires protocolized assessment and monitoring |
| Benefit uncertainty | Individual delay cannot be predicted exactly |
| Access | Screening, staging, infusion capacity, and payer coverage are prerequisites |
Lessons from partial successes¶
Rituximab’s four-dose course partially preserved secretion, with B cells returning to 69% of baseline by 12 months; this links waning pharmacodynamic effect and disease recurrence without proving B cells are the sole driver (Pescovitz 2009, PMID 19940299; NCT00279305).
Abatacept produced a 59% higher adjusted C-peptide AUC at two years and an estimated 9.6-month delay in decline in new-onset T1D, yet stage-1 abatacept did not significantly delay its composite primary endpoint (Orban 2011, PMID 21719096; Russell 2023, PMID 36920087; NCT00505375; NCT01773707).
Low-dose ATG alone outperformed placebo for one-year C-peptide, whereas adding GCSF did not meet the prespecified threshold, a warning that mechanistic combinations can dilute rather than add efficacy (Haller 2018, PMID 30012675; NCT02215200).
Oral insulin failed in the 389-participant primary stratum, although a secondary subgroup signal generated hypotheses. Subgroup findings after a negative primary endpoint require prospective confirmation (TrialNet Oral Insulin Study Group 2017, PMID 29164254).
Why trials disagree¶
- Stage, age, baseline C-peptide, autoantibody pattern, and rate of decline differ.
- C-peptide assays and mixed-meal timing can alter sensitivity.
- A transient pharmacodynamic effect may shift curves without inducing tolerance.
- Trial enrollment enriches relatives and selected ancestries.
- Background insulin technology changes HbA1c and hypoglycemia independently.
- Multiple comparisons and responder analyses can overstate subgroup signals.
- Safety tolerance differs when treating asymptomatic stage 1 versus symptomatic stage 3.
Combination strategies¶
The biological rationale is to pair complementary functions: suppress active autoreactivity, restore regulation, reduce β-cell stress, and provide antigen-specific tolerance. The clinical-trial registry includes rituximab followed by abatacept in new-onset T1D (NCT03929601), precision ATG with or without verapamil (NCT06455319), and a planned platform comparison of teplizumab with ATG in stage 2 (NCT07216391). Registry status is not evidence of efficacy.
| Combination goal | Example | Principal risk |
|---|---|---|
| Sequential depletion + costimulation blockade | Rituximab → abatacept | Compound infection/vaccine effects |
| Immune modulation + β-cell protection | ATG + verapamil | Attribution and additive adverse effects |
| Repeat anti-CD3 | Additional teplizumab courses | Cumulative toxicity; uncertain marginal benefit |
| Immune therapy + antigen tolerance | Depletion/modulation then insulin antigen | Timing and responder selection |
Regulatory and evidentiary boundary¶
An approved delay therapy changes clinical practice without answering lifetime benefit. Required post-approval evidence includes duration distribution, repeat-course effects, real-world safety, quality of life, access by ancestry and income, and whether delayed diagnosis reduces cumulative complications.
Quantitative evidence matrix¶
| Intervention/stage | Design | β-cell or progression result | Clinical/safety boundary |
|---|---|---|---|
| Teplizumab, stage 2 | 76-person RCT | HR 0.41 (95% CI 0.22–0.78); median 48.4 vs 24.4 months | Rash and transient lymphopenia; relative-enriched cohort |
| Oral insulin, stage 1 | 389-person primary stratum | HR 0.87; P=0.21 | Negative primary endpoint; positive 55-person secondary stratum |
| Abatacept, stage 1 | 212-person RCT | HR 0.702 (95% CI 0.452–1.09) | Immune/metabolic effects without primary progression benefit |
| Low-dose ATG, new onset | 89-person three-arm RCT | One-year AUC C-peptide 0.646 vs 0.406 nmol/L | GCSF did not enhance preservation |
| Verapamil, new onset | 88-person pediatric RCT | Adjusted AUC difference 0.14 pmol/mL (0.01–0.27) | β-cell-directed, not tolerance-inducing |
| Tight AID control, new onset | 113-person pediatric RCT | C-peptide difference −0.01 (−0.11 to 0.10) | TIR +16 points without preservation |
Oral insulin illustrates subgroup risk. The primary stratum had 58/203 versus 62/186 diagnoses and no significant delay; a 55-person secondary stratum had HR 0.45 and P=0.006, but was not the primary confirmatory population (Krischer 2017, PMID 29164254). A subgroup signal should generate a prospectively stratified trial, not reverse a negative primary conclusion.
Low-dose ATG and the failed add-on¶
At one year, low-dose ATG produced mean mixed-meal AUC C-peptide 0.646 versus 0.406 nmol/L with placebo (P=0.0003), while ATG+GCSF was 0.528 and missed the prespecified one-sided threshold (P=0.031). Both active groups lowered HbA1c, but GCSF added no C-peptide benefit (Haller 2018, PMID 30012675).
At two years, ATG alone retained a strong C-peptide difference (P=0.00005), whereas ATG+GCSF again missed its statistical criterion (P=0.032); ATG increased regulatory-to-conventional CD4 ratios and PD-1+CD4 cells (Haller 2019, PMID 30967424). This is evidence of durable partial preservation, not immune tolerance or insulin independence.
A 2025 six-person open-label pilot combined a deliberately subtherapeutic allogeneic-islet dose with short-course ATG, mTOR inhibition, and GCSF as an “immune education” strategy. Median stimulated C-peptide AUC was 91–100% of baseline at 12 months but 44–56% at five years; two participants retained stable secretion. The absence of a control group and the waning group trajectory mean this does not establish durable tolerance (Piemonti 2025, PMID 41438972).
B-cell depletion reveals network compensation¶
Rituximab depleted B cells and slowed first-year C-peptide loss, but autoantibody effects were antigen-specific: 40% (19/48) of IAA-positive treated participants became IAA-negative versus 0/29 placebo, while GADA, IA-2A, and ZnT8A changed less (Yu 2011, PMID 21831969). Autoantibody suppression is therefore not a uniform pharmacodynamic surrogate.
Systems analysis found that transiently elevated heterogeneous T-cell populations after rituximab correlated with weaker pharmacodynamic activity, more islet-antigen proliferation, and faster C-peptide loss (Linsley 2019, PMID 29925930). The suggested rituximab-plus-T-cell strategy is mechanistic rationale, not proven combination efficacy.
Preserved C-peptide versus patient-experienced benefit¶
In T1DAL, higher four-hour C-peptide AUC predicted fewer subsequent major hypoglycemic events and correlated with lower HbA1c and glucose variability, regardless of alefacept assignment. The post hoc analysis included 49 participants and found no treatment-group interaction (Pinckney 2016, PMID 27209482). This supports biological value of secretion while falling short of validating C-peptide as a universal surrogate.
Across 19 immunotherapy trials (1,852 participants), a 2025 meta-analysis estimated modest C-peptide preservation (SMD 0.221, 95% CI 0.069–0.373) without HbA1c improvement (SMD 0.033, −0.070 to 0.136). Effects attenuated with time, and leukocyte-clearance approaches had more serious events than T-regulatory enhancement approaches (Biswas 2025, PMID 41267047). Heterogeneous agents and endpoint timing limit pooled causal interpretation.
Long-duration natural-history data give the surrogate a clinical anchor: after mean 35 years, peak C-peptide >0.03 nmol/L associated with substantially less severe hypoglycemia, but not fewer advanced microvascular complications (Gubitosi-Klug 2021, PMID 33529168).
Responder biomarkers: promising and hazardous¶
Seven-year AbATE follow-up found less C-peptide loss and persistent PD-1-positive central-memory/anergic CD8 changes among participants classified as one-year teplizumab responders, but no lower HbA1c or insulin use (Perdigoto 2019, PMID 30569273). Because response was defined after treatment, this cannot select patients prospectively.
Routine clinical measures may support adaptive trials. A week-12 Beta2 score predicted one-year response across baricitinib and validation datasets from rituximab, abatacept, and ATG, with AUC 0.864 and 0.765; a 6.2% decline predicted HbA1c −0.6% and insulin −0.26 units/kg/day (So 2025, PMID 40424079). External prospective validation remains necessary.
Stage-1 abatacept reanalysis identified high baseline secretors with HR 0.46 (95% CI 0.25–0.84) and 15.8 (4.85–26.68) progression-free months gained; treatment interaction HR was 2.92 (1.23–6.96) (Galderisi 2026, PMID 41237315). This cannot override the parent negative primary endpoint without confirmation.
Trial design implications¶
| Design choice | Advantage | Failure risk |
|---|---|---|
| Factorial immune + β-cell protection | Tests interaction efficiently | Toxicity and underpowered interaction |
| Sequential depletion/modulation | Respects immune dynamics | Order and interval multiply arms |
| Response-adaptive platform | Moves allocation toward signal | Temporal drift and changing background technology |
| Enrichment by secretion/antibody | Raises event and response rates | Excludes slower endotypes and reduces transportability |
| Maintenance/re-dosing | Tests fading effects | Cumulative infection, malignancy, and burden |
Future trials need prespecified estimands for delay, durable stage-free survival, off-treatment persistence, severe events, patient burden, and cumulative toxicity. Enhanced designs and β-cell-death/imaging endpoints were proposed precisely because single-agent trials repeatedly produced transient C-peptide signals without restored tolerance (Ehlers 2016, PMID 27068279; Bone 2017, PMID 28534310).
β-cell protection is complementary, not interchangeable¶
Near-normal AID control increased TIR by 16 points but produced no C-peptide difference at 52 weeks (McVean 2023, PMID 36826834). Verapamil produced a 30% relative C-peptide advantage in the paired factorial trial. Albiglutide, by contrast, produced no appreciable 52-week preservation in a small phase-2 trial (Pozzilli 2020, PMID 32219329). “β-cell support” is a mechanistic category, not a class effect.
Open questions¶
- Can any regimen create a treatment-free plateau rather than a temporary right-shift in diagnosis? (Herold 2019, PMID 31180194)
- Which immune and metabolic biomarkers identify responders before exposure? (Russell 2023, PMID 36920087)
- Does C-peptide preservation translate into fewer severe events or complications over decades? (Ramos 2023, PMID 37861217)
- What sequence maximizes efficacy while avoiding cumulative immunosuppression? (Pescovitz 2009, PMID 19940299; Orban 2011, PMID 21719096)
- Can β-cell stress therapy complement immune therapy without masking progression? (Forlenza 2023, PMID 36826844)
Related pages¶
- screening and early detection — how candidates are found.
- immunopathogenesis — targets and response biology.
- biomarkers — response and progression measures.
- clinical trials landscape — registered pipeline.
References¶
- Palmer JP, et al. C-peptide is the appropriate outcome measure for type 1 diabetes clinical trials. Diabetes. 2004;53:250-264. PMID 14693724
- Pescovitz MD, et al. Rituximab, B-lymphocyte depletion, and preservation of beta-cell function. N Engl J Med. 2009;361:2143-2152. PMID 19940299
- Orban T, et al. Co-stimulation modulation with abatacept in recent-onset type 1 diabetes. Lancet. 2011;378:412-419. PMID 21719096
- TrialNet Oral Insulin Study Group. Effect of Oral Insulin on Prevention of Diabetes in Relatives. JAMA. 2017;318:1891-1902. PMID 29164254
- Haller MJ, et al. Low-Dose ATG Preserves β-Cell Function and Improves HbA1c. Diabetes Care. 2018;41:1917-1925. PMID 30012675
- Herold KC, et al. An Anti-CD3 Antibody, Teplizumab, in Relatives at Risk. N Engl J Med. 2019;381:603-613. PMID 31180194
- Russell WE, et al. Abatacept for Delay of Type 1 Diabetes Progression in Stage 1. Diabetes Care. 2023;46:1005-1013. PMID 36920087
- Ramos EL, et al. Teplizumab and β-Cell Function in Newly Diagnosed Type 1 Diabetes. N Engl J Med. 2023;389:2151-2161. PMID 37861217
- Forlenza GP, et al. Effect of Verapamil on Pancreatic Beta Cell Function. JAMA. 2023;329:990-999. PMID 36826844
- Haller MJ, et al. Low-Dose ATG: Two-Year Clinical Trial Data. Diabetes. 2019;68:1267-1276. PMID 30967424
- Yu L, et al. Rituximab selectively suppresses specific islet antibodies. Diabetes. 2011;60:2560-2565. PMID 21831969
- Linsley PS, et al. Elevated T cell levels predict poor response following rituximab. Genes Immun. 2019;20:293-307. PMID 29925930
- Pinckney A, et al. Hypoglycemia, variability, and C-Peptide after alefacept in T1DAL. Clin Ther. 2016;38:1327-1339. PMID 27209482
- Biswas S, et al. Immunomodulatory interventions in type 1 diabetes: systematic review and meta-analysis. BMC Endocr Disord. 2025;25:270. PMID 41267047
- Gubitosi-Klug RA, et al. Residual β cell function in long-term type 1 diabetes and hypoglycemia. J Clin Invest. 2021;131. PMID 33529168
- Perdigoto AL, et al. Teplizumab follow-up after 7 years from diagnosis. Diabetologia. 2019;62:655-664. PMID 30569273
- So M, et al. β-Cell Function Derived From Routine Measures Predicts Immunotherapy Response. Diabetes Care. 2025;48:1370-1376. PMID 40424079
- Galderisi A, et al. Baseline Insulin Secretion Determines Response to Abatacept in Stage 1. Diabetes. 2026;75:229-240. PMID 41237315
- Ehlers MR. Strategies for clinical trials in type 1 diabetes. J Autoimmun. 2016;71:88-96. PMID 27068279
- Bone RN, Evans-Molina C. Combination Immunotherapy for Type 1 Diabetes. Curr Diab Rep. 2017;17:50. PMID 28534310
- McVean J, et al. Tight Glycemic Control and Pancreatic Beta Cell Function. JAMA. 2023;329:980-989. PMID 36826834
- Pozzilli P, et al. Albiglutide Versus Placebo in Newly Diagnosed Type 1 Diabetes. J Clin Endocrinol Metab. 2020;105. PMID 32219329
- Mirmira RG, et al. Biomarkers of β-Cell Stress and Death in Type 1 Diabetes. Curr Diab Rep. 2016;16:95. PMID 27541297
- Galderisi A, et al. Trajectory of beta cell function and insulin clearance in stage 2 type 1 diabetes. Diabetologia. 2025;68:646-661. PMID 39560746
- Lin A, et al. Low-Dose ATG/GCSF in Established Type 1 Diabetes: Five-Year Follow-up. Diabetes. 2021;70:1123-1129. PMID 33632742
- Piemonti L, et al. Induction of immune education in type 1 diabetes through controlled allogeneic islet rejection at onset: a monocentric open-label pilot study. EClinicalMedicine. 2025;90:103685. PMID 41438972