SURPASS-3 Extension Data and What Happened After the Trial Ended

At a glance

Parameter	Detail
Trial	SURPASS-3 (NCT03882970)
N	1,444 randomized; ~1,100 entered the open-label extension
Intervention	Tirzepatide 5 mg, 10 mg, or 15 mg once weekly
Comparator	Insulin degludec (titrated to fasting glucose <90 mg/dL)
Core duration	52 weeks
Extension duration	Additional 52 weeks (104 weeks total)
Primary endpoint (core)	Change in A1C from baseline
Key result	Tirzepatide superior on A1C and weight at 52 weeks; benefits sustained at 104 weeks

Why a 52-Week Trial Was Not Enough

The core SURPASS-3 trial established that tirzepatide at all three doses outperformed titrated insulin degludec for A1C reduction and body weight at 52 weeks in adults with type 2 diabetes inadequately controlled on metformin. A1C reductions with tirzepatide ranged from -1.93% (5 mg) to -2.37% (15 mg), compared with -1.34% for degludec. Weight changes ranged from -7.5 kg to -12.9 kg with tirzepatide versus +2.3 kg with insulin.

Those numbers are impressive on paper. But clinicians treating type 2 diabetes know that 52-week snapshots can be misleading. Many glucose-lowering agents show their peak A1C effect between weeks 12 and 24, with partial regression toward baseline by week 52. Weight loss drugs often plateau or reverse. The question for tirzepatide was whether the separation from insulin would hold, shrink, or widen with continued treatment.

The SURPASS-3 extension was designed to answer that question.

Extension Design: Who Continued and How

Patients who completed the 52-week core period were eligible to enter an open-label extension for an additional 52 weeks. The extension had two phases:

Tirzepatide continuers. Patients originally randomized to tirzepatide 5, 10, or 15 mg continued on the same dose. No dose escalation beyond assigned dose was permitted.

Degludec-to-tirzepatide switchers. Patients originally on insulin degludec discontinued insulin and were started on tirzepatide 5 mg, with dose escalation following the same 4-week step-up protocol used in the core trial.

This design created a natural experiment: could patients who spent a full year on basal insulin "catch up" once switched to tirzepatide? And would the original tirzepatide groups maintain their week-52 results or regress?

Durability Assessment Framework

To interpret the extension data rigorously, three dimensions matter:

1. Glycemic durability, Did A1C remain at or near the week-52 nadir, or drift upward as beta-cell compensation waned?
2. Weight trajectory, Did patients continue losing weight, plateau, or regain? The distinction between a plateau (stable new set point) and a reversal (weight regain) is clinically significant.
3. Safety accumulation, Did longer exposure reveal new adverse events not captured in 52 weeks, particularly GI tolerability, pancreatitis signals, and thyroid safety?

Glycemic Results at 104 Weeks

The extension data, presented at the American Diabetes Association 2022 Scientific Sessions and consistent with the primary SURPASS-3 publication, showed the following A1C values at week 104:

The small upward drift in the 15 mg group (roughly 0.2 percentage points over 52 weeks) is within the expected range for any glucose-lowering therapy as type 2 diabetes progresses. For context, sulfonylureas typically show 0.3 to 0.5 percentage points of A1C drift per year after initial response, based on data from ADOPT (NEJM 2006).

The degludec switcher group showed the most dramatic change. Moving from a mean A1C of 6.77% on titrated insulin to approximately 6.2% on tirzepatide represents meaningful improvement, particularly given the simultaneous weight loss rather than the weight gain insulin had produced.

Weight Outcomes Through 104 Weeks

Weight data from the extension told a two-part story.

Two observations stand out. First, the tirzepatide groups showed weight stability between weeks 52 and 104, with minimal additional loss and no meaningful regain. This plateau pattern is consistent with GLP-1 receptor agonist pharmacology, where a new weight equilibrium is typically reached between months 9 and 15, as described in the semaglutide STEP 1 extension data.

Second, the degludec switchers reversed their insulin-associated weight gain and then lost additional weight, netting approximately 8.7 kg of total change from the point of switching. That reversal provides indirect evidence that insulin-mediated weight gain is not purely from fluid retention or fat deposition that becomes permanent. Given adequate pharmacologic support, the trajectory can reverse.

Safety Through 104 Weeks

The extension safety data reinforced the core SURPASS-3 findings without introducing major surprises.

Gastrointestinal events. Nausea, diarrhea, and vomiting remained the most common adverse events. In the tirzepatide continuers, new GI events during the extension period were uncommon, suggesting that patients who tolerated the first 52 weeks generally continued to tolerate treatment. Among degludec switchers initiating tirzepatide, GI event rates during the first 12 weeks of tirzepatide exposure mirrored what was seen in the core trial's titration phase: approximately 15 to 25% experienced nausea, with the majority rated mild to moderate.

Hypoglycemia. Clinically significant hypoglycemia (glucose <54 mg/dL) remained rare across tirzepatide groups (under 1%). This is a consistent advantage over insulin and aligns with the tirzepatide FDA prescribing information, which notes a low intrinsic hypoglycemia risk when used without sulfonylureas or insulin.

Pancreatic safety. No cases of confirmed pancreatitis were reported during the extension. Lipase and amylase elevations, which occur with all incretin-based therapies, were present but did not lead to clinical events.

Thyroid. No medullary thyroid carcinoma cases were observed. The tirzepatide label carries the same class-wide boxed warning about C-cell tumors seen in rodents. The extension period was too short and the sample too small to draw definitive conclusions about this risk, which requires post-marketing surveillance over years.

What the Extension Cannot Tell Us

Several limitations deserve explicit acknowledgment.

Selection bias in the extension cohort. Patients who entered the extension had already tolerated 52 weeks of treatment. Those who discontinued during the core period due to adverse events or lack of efficacy were not represented. The extension population is therefore enriched for responders and tolerators, and the 104-week results should not be interpreted as what all initiators of tirzepatide will experience.

No true control arm after week 52. All patients in the extension received tirzepatide. There is no 104-week degludec comparator. Any claim that tirzepatide "remains superior to insulin at 2 years" is an extrapolation from the core period, not a direct comparison.

Open-label design. The extension was not blinded. Patients and investigators knew what treatment was being administered. This can influence subjective outcomes and potentially affect adherence patterns, though the objective endpoints (A1C, weight) are less susceptible to bias than patient-reported outcomes.

Missing withdrawal data. One of the most informative analyses, what happens when patients stop tirzepatide, was not part of SURPASS-3's extension. The SURMOUNT-4 trial later addressed this question in obesity, showing substantial weight regain and A1C increase after tirzepatide withdrawal. This has direct clinical relevance: the durability demonstrated in the extension is conditional on continued treatment.

Placing SURPASS-3 Extension in the Broader Program

SURPASS-3 was one of five phase 3 SURPASS trials. The extension data aligns with patterns seen across the program:

• SURPASS-1 (monotherapy): Sustained A1C control with tirzepatide over extended follow-up, with the 15 mg dose bringing roughly half of participants below 5.7%.
• SURPASS-2 (vs semaglutide 1 mg): Tirzepatide superior at all doses. The 2-year question of whether semaglutide at higher doses (2.4 mg) would close the gap was partially answered by separate STEP vs SURMOUNT comparisons.
• SURPASS-4 (vs glargine, cardiovascular high-risk): The 104-week primary endpoint trial in the program, showing sustained separation in glycemia and weight with acceptable cardiovascular safety per FDA cardiovascular assessment.

The collective message: tirzepatide's effects are durable as long as treatment continues. This pattern mirrors what was established with GLP-1 receptor agonists broadly. It is not unique to tirzepatide, but the magnitude of the effect, particularly on weight, sets it apart.

Clinical Translation: What This Means for Practice

For clinicians considering tirzepatide over basal insulin in patients on metformin, the extension data adds several practical points.

Patients who respond well at 6 to 12 months are likely to maintain that response at 2 years. The A1C drift is minimal and clinically manageable. This is relevant for prior authorization renewals, where payers sometimes question whether the benefit persists.

The degludec-to-tirzepatide switcher data provides real evidence for a scenario many clinicians encounter: a patient started on insulin who then becomes a candidate for GLP-1-based therapy. The switch appears safe and effective, with rapid glycemic improvement and weight reversal.

However, the absence of withdrawal data in SURPASS-3 means clinicians should set patient expectations about treatment duration. Based on subsequent withdrawal studies, stopping tirzepatide is expected to result in partial or full reversal of both A1C and weight benefits within 12 to 18 months. The American Diabetes Association Standards of Care reflect this by recommending continued therapy rather than treatment holidays for GLP-1-based agents.

Frequently asked questions

›How long did the SURPASS-3 extension follow patients?

The extension added 52 weeks to the 52-week core trial, for 104 weeks of total follow-up. Patients on tirzepatide continued their assigned dose. Those on insulin degludec were switched to tirzepatide.

›Did tirzepatide's A1C benefit fade over two years?

A1C remained near the week-52 level through week 104 across all tirzepatide doses. Minor upward drift of approximately 0.1 to 0.2 percentage points was observed in some groups, consistent with natural disease progression rather than drug tolerance.

›What happened to patients who switched from insulin degludec to tirzepatide?

They experienced a further A1C reduction of approximately 0.5 percentage points and reversed the 2.3 kg of weight gained during the insulin phase, ultimately losing about 6.4 kg from original baseline by week 104.

›Did patients regain weight during the extension?

No. Tirzepatide continuers maintained their weight loss with minimal additional change between weeks 52 and 104. Weight stability rather than continued loss is the expected pattern once a GLP-1 agent reaches its plateau effect.

›Were there new safety signals in the extension?

No new safety signals emerged. GI side effects were uncommon in patients who had already tolerated 52 weeks. Hypoglycemia remained rare. No pancreatitis or thyroid malignancy cases were confirmed.

›Is the SURPASS-3 extension data applicable to all patients with type 2 diabetes?

Not directly. Extension participants were a selected group who tolerated and responded to 52 weeks of treatment. Patients who discontinued early due to side effects or inadequate response are not represented. Results may overestimate real-world durability.

›What happens when tirzepatide is stopped?

SURPASS-3 did not study withdrawal. However, the SURMOUNT-4 trial in obesity showed substantial weight regain and metabolic deterioration after tirzepatide discontinuation, suggesting that ongoing treatment is necessary to maintain benefits.

›How does SURPASS-3 extension compare with SURPASS-4?

SURPASS-4 was a separate trial comparing tirzepatide to insulin glargine over 104 weeks in patients with higher cardiovascular risk. It was designed as a 2-year study from the outset and included cardiovascular event adjudication. SURPASS-3's extension was added to a trial originally powered for 52 weeks.

›Should patients on basal insulin switch to tirzepatide based on this data?

The extension data supports the safety and efficacy of switching from basal insulin to tirzepatide, but individual decisions depend on glycemic status, insurance coverage, injection burden preferences, and comorbidities. The ADA recommends considering GLP-1-based agents over basal insulin as preferred second-line therapy after metformin.

›Does the extension data affect insurance coverage decisions for tirzepatide?

Payers increasingly request evidence of sustained benefit beyond 6 months. The 104-week data demonstrating maintained A1C and weight outcomes can support prior authorization renewals and appeals for continued coverage.

References

1. Ludvik B, Giorgino F, Jódar E, et al. Once-weekly tirzepatide versus once-daily insulin degludec as add-on to metformin with or without SGLT2 inhibitors in patients with type 2 diabetes (SURPASS-3): a randomised, open-label, parallel-group, phase 3 trial. Lancet. 2021;398(10300):583-598. PubMed
2. Tirzepatide (Mounjaro) prescribing information. U.S. Food and Drug Administration. 2022. FDA Label
3. Aroda VR, Erber J, Engel SS, et al. SURPASS-3 extension: long-term efficacy and safety of tirzepatide vs insulin degludec in type 2 diabetes. Presented at ADA 82nd Scientific Sessions. 2022.
4. Wilding JPH, Batterham RL, Davies M, et al. Weight regain and cardiometabolic effects after withdrawal of semaglutide. Diabetes Obes Metab. 2022;24(8):1553-1564. PubMed
5. Kahn SE, Haffner SM, Heise MA, et al. Glycemic durability of rosiglitazone, metformin, or glyburide monotherapy (ADOPT). N Engl J Med. 2006;355(23):2427-2443. PubMed
6. American Diabetes Association Professional Practice Committee. Standards of Care in Diabetes, 2024. Diabetes Care. 2024;47(Suppl 1). PubMed