Introduction

Early achievement of optimal glycemic control in type-2 diabetes mellitus (T2DM) is associated with improved long-term outcomes, including reduced microvascular and cardiovascular complications and a greater likelihood of attaining near-normal glycemia or remission. However, the traditional stepwise treatment approach often fails to maintain durable glycemic control and adequate management of weight and cardiometabolic risk factors. Intensive multifactorial intervention has proven beneficial, and current guidelines emphasize meaningful weight reduction to improve insulin sensitivity and β-cell function. Tirzepatide, a dual glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 receptor agonist (GLP-1RA), has demonstrated robust HbA1c reduction, substantial weight loss, improvements in insulin sensitivity and cardiometabolic parameters, and a higher likelihood of achieving normoglycemia, particularly in early T2DM. 

Aim

The SURPASS-EARLY trial evaluated whether early tirzepatide initiation would provide superior and more durable glycemic control, as compared to intensified conventional care (ICC). 

Patient Profile

• Adult patients (age ≥ 18 years) with a confirmed T2DM diagnosis within the past 4 years. 
• The glycated hemoglobin (HbA1c) levels ranged from 7.0% to 9.5%, body mass index (BMI) ranged from 25 to 45 kg/m² and had stable weight (±5%) during the 90 days before screening. 
• Patients were on a stable treatment with metformin with no other antihyperglycemic medication for at least 90 days before screening. 

Methods

Study Design

• A randomized, open-label, parallel-group, phase 4 study currently ongoing at 78 sites across 10 countries.
• The study consists of a 3-week screening period, followed by a 208-week treatment period, and a 4-week safety follow-up period. 
• This analysis reports the results for the primary time point and week 104.

Treatment Strategy

• Patients were randomized 1:1 to receive tirzepatide or ICC. 
• Tirzepatide was initiated at 2.5 mg once weekly and escalated by 2.5 mg every 4 weeks to a target of 15 mg or the maximum tolerated dose (MTD). Flexible dose reduction and re-escalation were permitted to manage intolerance or safety concerns. Participants unable to tolerate 2.5 or 5 mg due to gastrointestinal (GI) adverse events (AEs) were required to discontinue therapy permanently. 
• Concomitant antihyperglycemic agents were restricted and allowed only as rescue treatment for persistent severe hyperglycemia or after permanent discontinuation of study medication.
• Intensified conventional care comprised guideline-supported glucose-lowering therapies other than tirzepatide, individualized to achieve near-normoglycemia while maintaining safety. Treatment adjustments or switches due to intolerance were permitted and not considered study discontinuation, allowing flexible therapy optimization according to clinical needs and investigator judgment.
• Participants in both groups were advised to maintain their usual physical activity and follow a healthy diet throughout the study. To minimize confounding effects on weight outcomes, participants were not permitted to initiate any new intensive diet or exercise program aimed at weight reduction during the study period.

Outcomes

Primary Outcome

Key Secondary Outcomes

Additional Secondary Outcomes

• Change from baseline in fasting serum glucose (FSG) level; the proportion of patients achieving HbA1c targets (<7%, ≤6.5%, or <5.7%) and weight loss thresholds (≥5%, ≥10%, or ≥15%); change from baseline in BMI; the proportion of patients achieving a composite of HbA1c of ≤ 6.5% and weight loss of ≥ 10% without clinically significant (blood glucose concentration <54 mg/dL) or severe hypoglycemia; change from baseline in lipid concentrations; homeostatic model assessment 2 (HOMA 2) of beta-cell function and of insulin resistance (calculated with C-peptide or fasting insulin); fasting C-peptide, intact proinsulin, and insulin concentrations; and subgroup analyses of change in HbA1c and weight by baseline duration of diabetes, BMI category, and HbA1c. 
• Post hoc analyses included change in HbA1c and weight from baseline to week 104 for the tirzepatide group vs. ICC group receiving injectable GLP-1RAs.

Safety Outcomes

Results

• Among 1,235 individuals screened, 794 were randomized to tirzepatide (n=398) or ICC (n=396), with 792 receiving at least 1 dose of the assigned treatment. At 104 weeks, 89.3% completed the follow-up (90.2% with tirzepatide vs. 88.4% with ICC). 
• Rescue medication use was lower with tirzepatide (0.5% vs. 2.5%), while treatment discontinuation was mainly due to AEs in the tirzepatide group and due to participant withdrawal in the ICC group.
• The patient demographics and characteristics were similar in both the study groups. The mean age of the study population was 53.5 years and 49.4% of them were females. The mean duration of diabetes was 2.6 years; mean HbA1c was 7.81%; mean weight was 100.1 kg; mean BMI was 35.4 kg/m²; and mean WC was 113.9 cm.
• At week 104, 96.5% of tirzepatide-treated patients had received therapy for ≥ 12 weeks, with most (84.1%) on the 15 mg dose; 4.4%, 3.5%, 3.5%, and 4.4% remained on 5, 7.5, 10, and 12.5 mg, respectively. 
• In the ICC group, 88.2% used one, 9.8% used two, and 1.2% of the patients used three glucose-lowering agents. GLP-1RAs were the most commonly used medications (85.3%), followed by sodium glucose cotransporter (SGLT)-2 inhibitors (16.7%), with ≥98% of participants maintained on therapy for at least 12 weeks.
• Patients treated with tirzepatide had a greater mean change in HbA1c from baseline to week 104, as compared to those treated with ICC (-2.24; 95% CI: -2.34, -2.14 vs. -1.45 95% CI: -1.55, -1.35 percentage points).

Fig. 1: Change in HbA1c from baseline to week 104 (efficacy estimand)

• Results of post hoc analyses, in terms of HbA1C reduction were similar for patients treated with tirzepatide vs. those treated with injectable GLP-1Ras in ICC group.
• A greater proportion of patients treated with tirzepatide vs. those treated with ICC achieved the HbA1c goals of < 7%, ≤6.5%, and < 5.7% (Table 1). Importantly, patients maintained the achieved targets throughout the 2 years of treatment. 

Table 1: Efficacy outcomes at week 104 (HbA1c; efficacy estimand)

Outcome at week 104		Tirzepatide 15 mg or MTD N=398			ICC N=396
		MBE (SE) [95% CI]		P-value	MBE (SE) [95% CI]	P-value
Participants achieving HbA1c target†, n (%)
<7.0%	n/N (%) [SE]	378/398 (95.1) [1.19]		-	309/396 (77.9) [2.19]	-
	Risk difference vs. ICC (95% CI)	17.1 (12.26, 21.96)		<0.001		-
≤6.5%	n/N (%) [SE]	362/398 (91.0) [1.56]		-	260/396 (65.5) [2.46]	-
	Risk difference vs. ICC (95% CI)	25.5 (19.82, 31.18)	<0.001		-	-
<5.7%	n/N (%) [SE]	272/398 (68.8) [2.43]	-		107/396 (26.7) [2.31]	-
	Risk difference vs. ICC (95% CI)	42.2 (35.68, 48.70)	<0.001		-	-

Data are model-based estimate (MBE) (SE) [95% CI] unless stated otherwise for randomized participants (efficacy analysis set).

• At 104 weeks, tirzepatide achieved greater FSG reduction than ICC (mean decrease: −60.1 vs. -42.7mg/dL; treatment difference: -17.4 mg/dL; 95% CI: -21.4, -13.40Consequently, FSG was lowered to 98.0 mg/dL with tirzepatide compared with 115.4 mg/dL with ICC at week 104.
• Patients treated with tirzepatide achieved a greater weight loss at week 104, as compared to those treated with ICC, demonstrating superiority of tirzepatide over ICC (Fig. 2).

Fig. 2: Change in weight from baseline to week 104

Table 2: Proportion of patients achieving weight reduction threshold at week 104 (efficacy estimand) 

Outcome at week 104		Tirzepatide 15 mg or MTD N=398		ICC N=396
		MBE (SE) [95% CI]	P-value	MBE (SE) [95% CI]	P-value
Participants achieving weight loss threshold†, n (%)
≥5%	n/N (%) [SE]	311/398 (77.9) [2.11]	-	195/396 (49.6) [2.56]	-
	Risk difference vs. ICC (95% CI)	28.3 (21.81, 34.76)	<0.001	195/396 (49.6) [2.56]	-
≥10%	n/N (%) [SE]	243/398 (60.7) [2.46]	-	102/396 (26.0) [2.22]	-
	Risk difference vs. ICC (95% CI)	34.7 (28.20, 41.14)	<0.001	-	-
≥15%	n/N (%) [SE]	175/398 (43.7) [2.49]	-	47/396 (11.9) [1.63]	-
	Risk difference vs. ICC (95% CI)	31.8 (25.98, 37.65)	<0.001	-	-
Participants achieving composite endpoint of HbA1c ≤6.5%, ≥10% weight loss, and no hypoglycemic event†,‡, n (%)	n/N (%) [SE]	217/398 (54.2) [0.37]	-	83/396 (21.1) [0.25]	-
	Risk difference vs. ICC (95% CI)	33.1 (26.90, 39.38)	<0.001	-	-

Data are model-based estimate (MBE) (SE) [95% CI] unless stated otherwise for randomized participants (efficacy analysis set).

†Proportion of participants achieving the HbA1c targets, weight thresholds, and composite endpoint are from logistic regression with multiple imputation. Percentage of participants was calculated by combining proportion of participants achieving target in imputed datasets using Rubin's rule. Treatment comparisons (tirzepatide vs ICC) are shown as risk difference (95% CI).

‡No hypoglycemia was defined as no occurrence of hypoglycemic event defined as blood glucose <54 mg/dL or severe hypoglycemia from randomization to week 104.

Table 3: Results for additional secondary outcomes

Outcome at week 104		Tirzepatide 15 mg or MTD N=398		ICC N=396
		MBE (SE) [95% CI]	P-value	MBE (SE) [95% CI]	P-value
Body weight*, %	Baseline	100.9 (1.02) [98.90, 102.90]	-	99.3 (1.03) [97.3, 101.3]	-
	Change from baseline	-15.3 (0.48) [-16.2, -14.33]	<0.001	-6.4 (0.48) [-7.28, -5.42]	<0.001
	Estimated treatment difference vs. ICC (95% CI)	-8.9 (-10.24, -7.59)	<0.001	-	-
Waist Circumference (cm)	Baseline	114.3 (0.72) [112.9, 115.7]		113.5 (0.72) [112.1, 114.9]
	Change from baseline	-13.3 (0.47) [-14.22, -12.38]	<0.001	-6.1 (0.46) [-6.98, -5.17]	<0.001
	Estimated treatment difference vs. ICC (95% CI)	-7.2 (-8.51, -5.93)	<0.001	-	-
FSG*, mg/dL	Baseline	157.3 (2.245) [152.9,161.8]	-	159.5 (2.239) [155.1,163.9]	-
	Change from baseline	-60.1 (1.455) [-63.0, -57.3]	<0.001	-42.7 (1.438) [-45.5, -39.9]	<0.001
	Estimated treatment difference vs ICC (95% CI)	-17.4 (-21.4, -13.40)	<0.001	-	-
Fasting C-peptide§, μg/L	Baseline	3.01 (0.071) [2.87, 3.15]	-	3.02 (0.071) [2.88, 3.16]	-
	Change from baseline	-0.53 (0.054) [-0.64, -0.42]	<0.001	-0.15 (0.062) [-0.27, -0.03]	-
	Percent change from baseline	-17.57 (1.789) [-21.01, -13.98]	<0.001	-5.07 (2.060) [-9.03, -0.93]	0.017
	Estimated treatment difference vs. ICC (95% CI)	-13.17 (2.665) [-18.25, -7.77]	<0.001	-	-
Fasting Intact Proinsulin§, pmol/L	Baseline	10.57 (0.375) [9.83, 11.31]	-	10.45 (0.370) [9.72, 11.18]	-
	Change from baseline	-4.59 (0.131) [-4.84, -4.33]	-	-3.33 (0.159) [-3.64, -3.02]	-
	Percent change from baseline	-43.63 (1.247) [-46.02, -41.13]	<0.001	-31.67 (1.509) [-34.57, -28.64]	<0.001
	Estimated treatment difference vs. ICC (95% CI)	-17.50 (2.580) [-22.41, -12.28]	<0.001	-	-
Fasting Insulin§, mU/L	Baseline	16.04 (0.599) [14.86, 17.21]	-	16.07 (0.601) [14.89, 17.25]	-
	Change from baseline	-5.26 (0.403) [-6.05, -4.46]	-	-2.86 (0.493) [-3.82, -1.89]	-
	Percent change from baseline	-32.74 (2.510) [-37.49, -27.62]	<0.001	-17.79 (3.068) [-23.60, -11.54]	<0.001
	Estimated treatment difference vs. ICC (95% CI)	-18.19 (4.321) [-26.24, -9.25]	<0.001	-	-
HOMA2-B§ (calculated using Cpeptide)	Baseline	61.2 (1.88) [57.5, 64.8]	-	62.0 (1.88) [58.3, 65.7]	-
	Change from baseline	71.2 (2.91) [65.5, 76.9]	-	50.6 (2.43) [45.8, 55.3]	-
	Percent change from baseline	115.6 (4.72) [106.5, 125.1]	<0.001	82.1 (3.94) [74.5, 90.0]	<0.001
	Estimated treatment difference vs. ICC (95% CI)	18.4 (3.65) [11.45, 25.8]	<0.001	-	-
HOMA2-B§ (calculated using insulin)	Baseline	55.2 (2.07) [51.2, 59.3]	-	56.6 (2.10) [52.5, 60.8]	-
	Change from baseline	56.1 (2.83) [50.5, 61.6]	-	41.5 (2.44) [36.7, 46.3]	-
	Percent change from baseline	100.2 (5.07) [90.5, 110.4]	<0.001	74.2 (4.36) [65.9, 83.0]	<0.001
	Estimated treatment difference vs ICC (95% CI)	14.92 (4.094) [7.15, 23.2]	<0.001	-	-

Data are model-based estimate (MBE) (SE) [95% CI] unless stated otherwise for randomized participants (efficacy analysis set).

*Data for HbA1c, body weight, waist circumference, FSG, and BMI are from MMRM. Estimated treatment difference are MBE (95% CI); P-values for HbA1c, body weight and waist circumference are for superiority vs ICC, adjusted for multiplicity.

†Proportion of participants achieving the HbA1c targets, weight thresholds, and composite endpoint are from logistic regression with multiple imputation. Percentage of participants was calculated by combining proportion of participants achieving target in imputed datasets using Rubin's rule. Treatment comparisons (tirzepatide vs ICC) are shown as risk difference (95% CI).

‡No hypoglycemia was defined as no occurrence of hypoglycemic event defined as blood glucose <54 mg/dL (<3.0 mmol/L) or severe hypoglycemia from randomization to week 104.

§ Assessed in oral glucose tolerance test. Data are estimate (SE) [95% CI] from analysis of covariance with log transformation and then converted back to original scale (efficacy estimand). Estimated treatment difference are MBE (95% CI).

Conclusions

• Amongst patients with early T2DM and uncontrolled glycemia while receiving metformin, those treated with tirzepatide achieved superior reductions in HbA1c, weight, and WC, as compared to those treated with ICC. 
• A higher proportion of patients treated with tirzepatide, vs. those treated with ICC achieved normoglycemia after 2 years.
• These findings support the concept that early initiation of tirzepatide treatment could establish better and potentially more durable glycemic control than can be achieved with conventional care.

Ann Intern Med. 2026 Jul;179(7):917-928.