5-year Simulation of Diabetes-related Complications in People treated with Tirzepatide or Semaglutide vs. Insulin Glargine

Abstract

Objective

This study compared the 5-year incidence rate of macrovascular and microvascular complications for tirzepatide, semaglutide, and insulin glargine in individuals with type 2 diabetes, using the BRAVO diabetes simulation model.

Research Design and Methods

This study was a five-year SURPASS-2 trial extrapolation, with an insulin glargine arm added as an additional comparator. The one-year treatment effects of tirzepatide (5 mg, 10 mg, or 15 mg), semaglutide (1 mg) and insulin glargine on HbA1c, SBP, LDL, and body weights were obtained from the SUSTAIN-4 and SURPASS-2 trials. We used the BRAVO model to predict 5-year complications for each study arm under two scenarios: the one-year treatment effects persisted (optimistic) or diminished to none in five years (conservative).

Results

When compared with insulin glargine, we projected a 5-year risk reduction in cardiovascular adverse events (MACE) (rate ratio [RR]0.64, 95% Confidence Interval [CI] 0.61–0.67) and microvascular composite (RR 0.67, 95% CI 0.64–0.70) with 15 mg tirzepatide, and 5-year risk reduction in MACE (RR 0.75, 95% CI 0.72–0.79) and microvascular composite (RR 0.79, 95% CI 0.76–0.82) with semaglutide (1 mg) under an optimistic scenario. Lower doses of tirzepatide also had similar, albeit smaller benefits. Treatment effects for tirzepatide and semaglutide were smaller but still significantly higher than insulin glargine under a conservative scenario.

The 5-year risk reduction in diabetes-related complication events and mortality for the 15 mg tirzepatide compared to insulin glargine ranged from 49% to 10% under an optimistic scenario, which was reduced by 17%−33% when a conservative scenario was assumed.

Conclusion:

With the use of BRAVO diabetes model, Tirzepatide and semaglutide exhibited potential to reduce the risk of macrovascular and microvascular complications among individuals with type 2 diabetes, compared to insulin glargine in a five years window. Based on the current modeling assumptions, tirzepatide (15 mg) may potentially outperform semaglutide (1 mg). While BRAVO model offered insignts, the long-term cardiovascular benefit of tirzepatide should be further validated in a prospective clinical trial.

Introduction

People with type 2 diabetes have 2 to 4 times higher risks of cardiovascular disease (CVD) and CVD-related death compared to the general population.¹ Better control of risk factors, including glycated hemoglobin (HbA1c), systolic blood pressure (SBP), low-density lipoprotein (LDL), and body mass index (BMI), has been noted to be associated with lower risks of diabetes-related complications among patients with type 2 diabetes.^2,3 Major guidelines recommend preferentially prescribing medications that reduce HbA1c, but also lower cardiovascular risk as a consideration (or at least ensure cardiovascular safety).^4,5 Thus, newer glucose-lowering drugs, including glucagon-like peptide1(GLP-1) receptor agonists (e.g., semaglutide), and sodium-glucose co-transporter-2 (SGLT2) inhibitors, have gained popularity due to their clinical efficacy and favorable cardiovascular outcomes.^6–11

Tirzepatide is a dual glucose-dependent insulinotropic polypeptide (GIP) and GLP-1 receptor agonist. It requires once-weekly subcutaneous administration to treat type 2 diabetes.¹² Previous trials (phases 1 and 2) have demonstrated that tirzepatide has comparable glucose-lowering and weight loss effects to those of established GLP-1 receptor agonists in individuals with type 2 diabetes.^12–14 Five completed phase 3 clinical trials (SURPASS 1–5 trials) also showed that trizepatide resulted in significant reductions in glucose and body weight and resulted in normoglycemia (<5.7%) in about 1/3 of the individuals with type 2 diabetes inadequately controlled with diet and exercise alone, or with medications (i.e. metformin, or insulin glargine).^15–19 Besides, tirzepatide was also associated with decreased lipid levels, blood pressure, and several biomarkers of inflammation.^13,20 Despite the favorable effects of tirzepatide on a range of cardiovascular risk factors, only the SURPASS-4 trial reported its cardiovascular safety among people with type 2 diabetes and elevated cardiovascular risk. The study showed that compared with insulin glargine, tirzepatide did not result in excess cardiovascular risk for up to 104 weeks follow-up.¹⁸ There is another ongoing trial, SURPASS-CVOT which is expected to provide long-term (54 month) data on the cardiovascular safety of tirzepatide against dulaglutide among people with type 2 diabetes and established atherosclerotic cardiovascular disease. But this study is event-driven and expected to complete at the end of 2024. There have been no results reported from long-term Cardiovascular Outcomes Trials (CVOTs) evaluating tirzepatide to date.

Microsimulation models in diabetes are widely used tools to predict the long-term outcomes of treatments.^21,22 The latest developed Building, Relating, Assessing, and Validating Outcomes (BRAVO) diabetes model can provide accurate treatment assessments on long-term macrovascular and microvascular outcomes based on individuals’ dynamic characteristics and treatment effects data.^23,24 To bridge the knowledge gap in long-term outcomes of tirzepatide, this study compared the 5-year incidence rate of macrovascular and microvascular complications for tirzepatide or semaglutide versus insulin glargine – a standard glucose-centric approach to treating diabetes – in individuals with type 2 diabetes, using the BRAVO diabetes simulation model.

Methods

Study design and population

The target population included individuals with type 2 diabetes inadequately controlled by metformin. We used the baseline characteristics of the SURPASS-2 trial to populate the simulation sample.¹⁶ Details were summarized in the Appendix (eTable1). We then used the BRAVO diabetes simulation model to translate the treatment effects of tirzepatide, semaglutide, and insulin glargine on key biomarkers to the 5-year study outcomes (e.g., cardiovascular complications).²⁴

We simulated four intervention arms and one control arm: 1) three intervention arms received once-weekly subcutaneous injections of tirzepatide at dosages of 5 mg, 10 mg, and 15 mg; 2) one intervention arm received semaglutide 1 mg; and 3) the control arm received insulin glargine. The short-term treatment effects of tirzepatide (at a dose of 5 mg, 10 mg, and 15 mg) and semaglutide (1 mg) on key biomarkers were extracted from the SURPASS-2 trial and extrapolated to a five-year window under two alternative disease progression scenarios: optimistic and conservative scenarios (see below for details). To estimate the short-term treatment effects of insulin glargine compared to tirzepatide, we applied an indirect comparison methods using data from SUSTAIN-4 trial (insulin glargine vs semaglutide) and SURPASS-2 trial (semaglutide vs tirzepatide).^16,25

Model overview and study outcomes

The BRAVO diabetes model was a person-level discrete-time microsimulation model, utilizing target populations’ time-varying characteristics and treatment regiments to predict the progression of diabetes complications over a lifetime.²⁴ We have externally validated the BRAVO model against a large number of global diabetes trials and calibrated the model to a nationally representative sample of individuals with T2D in the US.^26,27 The BRAVO simulation model was previously used for several research purposes, including trial simulation, program evaluation, and cost-effectiveness analysis.^23,28,29 Detailed descriptions of the model were well described in a previous paper.²⁴

We used the BRAVO diabetes model to simulate the 5-year study outcomes for the five study arms, using the SURPASS-2 trial population as the target population. For each study arm, we entered the baseline characteristics of the SURPASS-2 trial population and the treatment efficacy data on key biomarkers (i.e., HbA1c, SBP, LDL, and BMI) into the BRAVO model. The baseline characteristics included patient’s sociodemographic characteristics (i.e., gender, race, age, region), current biomarkers (i.e., HbA1c, SBP, LDL, and BMI), and clinical characteristics (i.e., diabetes duration). The BRAVO model translated 5-year changes of key biomarkers to risk reductions in study outcomes. The biomarker driven incidence rate of study outcomes were simulated at a yearly cycle until 5 years or encountered death events. The incidence rate of each outcome event in the next year was predicted by patients’ parameter values from the current year. We ran the simulation 100 000 times for each study arm to achieve stable estimation. The primary predicted outcomes encompassed macrovascular complications, microvascular composite and all-cause mortality. Specifically, the macrovascular complications included myocardial infarction(MI), stroke, congestive heart failure (CHF), and major adverse cardiovascular events (MACE). MACE was a three-component composite endpoint of non-fatal stroke, non-fatal MI, and cardiovascular mortality. The microvascular composite encompassed end-stage renal disease (ESRD), blindness, and severe pressure sensation loss (SPSL). The simulation results were summarized to produce the predicted 5-year incidence rate of each complication and the predicted rate ratio (95% CI) for the intervention arm compared to the control arm, under optimistic and conservative scenarios. The risk reduction of diabetes complications between the intervention and control arm were driven by the treatment efficacy on key biomarkers.

Model calibration

GLP-1 receptor agonists might have additional mechanisms to further reduce risks of cardiovascular complications unexplained by traditional risk factors (i.e., HbA1c, SBP, LDL, and BMI). We assumed that tirezepatide reduced the risk of cardiovascular complications under a similar mechanism as other GLP-1 receptor agonist classes. To predict cardiovascular outcomes of our study medications accurately, we have tested this hypothesis by validating and calibrating the BRAVO model against four CVOTs for GLP-1 receptor agonists classes (including ELIXA, LEADER, SUSTAIN 6, and EXSCEL).^11,30–32 We extracted baseline characteristics of the study subjects and clinical efficacies of the included treatments in reducing risk factors (i.e., HbA1c, SBP, LDL, and BMI) from each CVOT and used the BRAVO model to predict the corresponding risk-factor driven risk reductions in cardiovascular outcomes. We pooled the observed and simulated hazard ratio and the associated 95% CIs across four CVOTs using a random-effects meta-analysis. To quantify the additional cardiovascular benefits of GLP-1 receptor agonists, we calculated the ratio of simulated traditional-risk-factor-driven hazard ratio to the observed hazard ratio in cardiovascular outcomes as a measurement of additional relative risk reductions (ARRR) of GLP-1 receptor agonists not related to traditional biomarker control. The ARRR ≥ 10% was considered clinically meaningful, indicating the BRAVO model needed to be calibrated to account for additional mechanisms of risk-reduction for GLP-1 receptor agonists.

Optimistic vs. conservative scenarios

The tirzepatide (5 mg, 10 mg, or 15 mg) and semaglutide’s 40-week effects data on key biomarkers (HbA1c, SBP, LDL, and BMI) were extracted from the SURPASS-2 trial. We assumed the 40-week treatment effects could last till the end of the year (week 52). Given the lack of data on how long the treatment effects would last throughout a 5-year study period, we used the BRAVO model to simulate two scenarios under the following assumptions: 1) Under the optimistic scenario, the first-year treatment effects would persist for the 5-year study period (the difference in HbA1c, SBP, LDL, and BMI between tirzepatide and semaglutide vs. insulin glargine remained constant); 2) Under the conservative scenario, the first-year treatment effects diminished to none at a constant rate over the 5-year study period (the difference between tirzepatide and semaglutide vs. insulin glargine annually decreased 25%). The background progression rate of HbA1c was derived from the ORIGIN trial.³³

Decomposition analysis

To understand the contribution of each key biomarker to the overall effect of tirzepatide and semaglutide, a decomposition analysis was conducted focusing on the comparison between tirzepatide (15mg) vs. insulin glargine under the conservative scenario. We conducted the simulation with treatment impacting only one biomarker at a time to estimate biomarker-specific treatment effects on study outcomes. We then compared the effect size of all four biomarkers to measure the proportion of contributions attributable to each biomarker.

Results

The temporal treatment effects for five arms in year one were presented in Table 1. Tirzepatide at a dose of 5 mg, 10 mg, and 15 mg was found to reduce HbA1c (−2.1 %, −2.4 %, and −2.5 %), SBP (−4.8mmHg, −5.3mmHg, −6.5mmHg), LDL (−6.7 mg/dl, −4.9mg/dl, −4.5mg/dl), BMI (−2.9 kg/m², −3.8 kg/m², −4.6 kg/m²) respectively. Semaglutide was found to reduce HbA1c (−1.9%), SBP (−3.6 mmHg), LDL (−5.6 mg/dl), and BMI (−2.3 kg/m²) in one year. Insulin Glargine was estimated to reduce HbA1c (−1.1%) and SBP (−0.1 mmHg), but increase LDL (0.8mg/dl) and BMI (0.03 kg/m²) in one year. Symptomatic hypoglycemia (blood glucose level <54 mg per deciliter) was reported by 0.6% patients, 0.2% patients, 1.7% patients who received tirzepatide at a dose of 5 mg, 10 mg, and 15 mg respectively, as compared with 0.4% patients with 1mg semaglutide and 11% patients with insulin glargine. Severe hypoglycemia (an event where the patient required assistance of another person to actively administer carbohydrates, glucagon, or undertake other corrective actions and was confirmed by plasma glucose level < 36 mg per deciliter) was reported by 0.2% patients, 0.0% patients, 0.2%, 0.0%, and 1% patients receiving 5 mg tirzepatide, 10mg tirzepatide, 15 mg tirzepatide, semaglutide, and insulin glargine, respectively.

Table 1.

Short-term treatment effects of Tirzepatide, Semaglutide, and Insulin Glargine in year 1.

Treatment Effects		Tirzepatide			Semaglutide	Insulin Glargine**
		5 mg	10 mg	15 mg	1 mg
HbA1c (%)	Start Value	8.3 (8.3,8.2)	8.3 (8.3,8.2)	8.3 (8.3,8.2)	8.3 (8.3,8.2)	8.3 (8.3,8.2)
	Change *	−2.1 (−2.2, −2.0)	−2.4 (−2.5, −2.3)	−2.5 (−2.6, −2.4)	−1.9 (−2.0, −1.8)	−1.1
	P-value	<0.001	<0.001	<0.001	<0.001
SBP (mmHg)	Start Value	130.6 (130.0,131.2)	130.6 (130.0,131.2)	130.6 (130.0,131.2)	130.6 (130.0,131.2)	130.6 (130.0,131.2)
	Change *	−4.8 (−5.8, −3.8)	−5.3 (−6.3, −4.3)	−6.5 (−7.6, −5.4)	−3.6 (−4.7, −2.5)	−0.1
	P-value	<0.001	<0.001	<0.001	<0.001
LDL (mg/dl)	Start Value	87.8 (86.1,89.5)	87.8 (86.1,89.5)	87.8 (86.1,89.5)	87.8 (86.1,89.5)	87.8 (86.1,89.5)
	Change *	−6.7 (−9.2, −4.2)	−4.9 (−7.5, −2.3)	−4.5 (−7.1, −1.9)	−5.6 (−8.1, −3.1)	0.8
	P-value	<0.001	<0.001	<0.001	<0.001
BMI (kg/m2)	Start Value	34.2 (33.9,34.5)	34.2 (33.9,34.5)	34.2 (33.9,34.5)	34.2 (33.9,34.5)	34.2 (33.9,34.5)
	Change *	−2.9 (−3.1, −2.6)	−3.8 (−4.0, −3.5)	−4.6 (−4.8, −4.3)	−2.3 (−2.5, −2.0)	0.03
	P-value	<0.001	<0.001	<0.001	<0.001
Symptomatic Hypoglycemia No.of patients(%)†		(3) 0.6	(1) 0.2	(8)1.7	(2) 0.4	(38) 11
Severe Hypoglycemia No.of patients(%)†		(1) 0.2	(0) 0	(1) 0.2	(0) 0	(5) 1

^*Treatment effects of tirzepatide and semaglutide were estimated using 40-week data and assumed the effects can last till the end of the year (week 52).

^**Reference Group. Confidence interval was not assigned for the reference group in the simulation.

^†Proportional meta-analysis of the incidences of symptomatic hypoglycemia and severe hypoglycemia among 4 intervention arms revealed an overall symptomatic hypoglycemia incidence 0.7% (95% CI: 0.3%−1.5%), overall severe hypoglycemia incidence 0.1% (95% CI: 0.03%−0.4%), both significantly lower than the incidence in the insulin glargine arm (Appendix, eFigure2).

95% confidence intervals are presented in the brackets.

HbA1c: glycated hemoglobin; SBP: Systolic blood pressure; LDL: low-density lipoprotein; BMI: body mass index

Source: Frías JP, Davies MJ, Rosenstock J, et al. Tirzepatide versus Semaglutide Once Weekly in Patients with Type 2 Diabetes. N Engl J Med. 2021;385(6):503–515.

Aroda VR, Bain SC, Cariou B, et al. Efficacy and safety of once-weekly semaglutide versus once-daily insulin glargine as add-on to metformin (with or without sulfonylureas) in insulin-naive patients with type 2 diabetes (SUSTAIN 4): a randomised, open-label, parallel-group, multicentre, multinational, phase 3a trial. Lancet Diabetes Endocrinol. 2017;5(5):355–366.

The 5-year trajectories of HbA1c, SBP, LDL, and BMI for each of the five treatment arms under optimistic and conservative scenarios were presented in Figure 1. We also summarized the model calibration results in the Appendix (eTable2). We found that the cardiovascular benefit of GLP-1 receptor agonists could be mostly explained by reductions in traditional risk factors (i.e., HbA1c, SBP, LDL, and BMI), thus model calibration to capture additional treatment effect of this drug class was not required.

Figure 1.

5-Year trajectories of HbA1c, SBP, LDL, and BMI for the Tirzepatide arms, Semaglutide arm, and Insulin Glargine arm.

Note: Optimistic scenario: treatment effects lasted for five years. Conservative scenario: treatment effects diminished to zero by year five.

HbA1c: glycated hemoglobin; SBP: Systolic blood pressure; LDL: low-density lipoprotein; BMI: body mass index.

Figure 2 presented the rate ratio (RR) of diabetes-related complications for tirzepatide and semaglutide, when compared with insulin glargine. Both tirzepatide and semaglutide were estimated to be superior to insulin glargine in reducing the risk of diabetes complications over a 5-year time horizon. Tirzepatide at a higher dose (i.e., 15 mg) was associated with a higher 5-year risk reduction in all-cause mortality (RR 0.90, 95% CI 0.90–0.91) [Explanation for this narrow confidence interval were provided in discussion], MI (RR 0.75, 95% CI 0.73–0.77), stroke (RR 0.51, 95% CI 0.47–0.56), CHF (RR 0.60, 95% CI 0.57–0.63), MACE (RR 0.64, 95% CI 0.61–0.67) and microvascular composite (RR 0.67, 95% CI 0.64–0.70) [Further breakdown of microvascular events types were provided in the appendix(eTable 3)] under the optimistic scenario, as compared with insulin glargine. Under conservative scenario, the 5-year risk reductions in observed diabetes-related complications in Tirzepatide 15 mg arm were 17%−33% less than that in the optimistic scenario. Tirzepatide at a dose of 5 mg, 10 mg were associated with a relatively lower but still significant 5-year risk reduction in observed diabetes-related complications. Combining results from the optimistic and conservative scenarios, semaglutide 1mg was associated with a 5-year risk reduction in all-cause mortality (RR 0.90–0.94), MI (RR 0.82–0.93), stroke (RR 0.67–0.89), CHF (RR 0.77–0.96), MACE (RR 0.75–0.91) and microvascular composite (RR 0.79–0.94) when compared with insulin glargine. More details regarding the simulation results can be found in the Appendix (eTable 3,eFigure1).

Figure 2.

5-Year relative risk of diabetes-related complications.

Note: 95% confidence intervals were presented as the bar widths.

MI: myocardial infarction; CHF: congestive heart failure; MACE*: major adverse cardiovascular events, including non-fatal stroke, non-fatal myocardial infarction, and cardiovascular mortality. Microvascular Composite^† includes end-stage renal disease, blindness, and severe pressure sensation loss.

Optimistic scenario: the short-term drug effects of tirzepatide and semaglutide lasted for 5 years.

Conservative scenario: the short-term effects of tirzepatide and semaglutide diminished wthin 5 years.

Figure 3 presented the proportion of the additional clinical benefits for tirzepatide 15 mg when compared with insulin glargine under the conservative scenario explained by each biomarker. Tirzepatide demonstrated a significant reduction in glucose levels and body weight. Most of the clinical benefit of tirzepatide was achieved through HbA1C reduction: all-cause mortality (9%), MI (80%), stroke (64%), CHF (25%), MACE (65%), microvascular composite (78%). The benefit of body weight reduction from tirzepatide contributed substantially in reducing the risk of Mortality (82%) and CHF (55%). SBP reduction explained a moderate proportion of the total clinical benefit: all-cause mortality (9%), MI (2%), stroke (27%), CHF (20%), MACE (16%), microvascular composite (20%).

Figure 3.

Contribution of key biomarker reduction to the observed benefit of Tirzepatide.

Note: Blue, orange, gray, and green bars denote risk reduction in trial outcomes due to glycated hemoglobin, Systolic blood pressure, low-density lipoprotein, and body mass index control, respectively.

HbA1c: glycated hemoglobin; SBP: Systolic blood pressure; LDL: low-density lipoprotein; BMI: body mass index. MI: myocardial infarction; CHF: congestive heart failure; MACE*: major adverse cardiovascular events, including non-fatal stroke, non-fatal myocardial infarction, and cardiovascular mortality. Microvascular Composite includes end-stage renal disease, blindness, and severe pressure sensation loss.

Discussion

In the 5- year study period, tirzepatide at a dose of 5 mg, 10 mg, and 15 mg and semaglutide at a dose of 1 mg were projected to be superior to insulin glargine, with respect to the reduction in the risk of macrovascular and microvascular complications when administered in people with type 2 diabetes. The 5-year risk reduction in diabetes-related complication events and mortality for the 15 mg tirzepatide compared to insulin glargine ranged from 49% to 10% under an optimistic scenario, which was reduced by 17%−33% when a conservative scenario was assumed. We also observed dose-dependent risk reductions of macrovascular and microvascular complications using tirzepatide. The clinical benefits of tirzepatide at a dose of 5 and 10 mg were slightly lower than tirzepatide administered at 15 mg, but the additional benefits were still statistically significant compared to insulin glargine. The high dose of tirzepatide (15 mg) may potentially outperform semaglutide (1 mg) in this aspect. Stroke and CHF manifested pronounced long-term benefit from tirzepatide, while MI and microvascular composite received relatively modest but still clinically significant benefit.

The benefit of tirzepatide in preventing all-cause mortality was lower than its effects in preventing other diabetes-related complications, and the 95% CI of the all-cause mortality risk reduction was also narrow. Because the BRAVO model was developed using data from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, which found a higher mortality rate in the intensive glycemic control group (HbA1c ≤ 6%) compared with the standard glycemic control group (HbA1c 7.0–7.9%).³⁴ A post-hoc analysis of the ACCORD data found the optimal glycemic control level was 7.1%, and no additional mortality benefits were observed when HbA1c was further reduced below 7%. In our study, the HbA1c level after receiving insulin glargine (i.e., the reference group) was already 7.2%, which explained the limited clinical benefit in all-cause mortality when further reducing HbA1c through tirzepatide.

In this study, we extracted the tirzepatide and semaglutide’s 40-week effects data on key biomarkers (HbA1c, SBP, LDL, and BMI) from SURPASS-2 trial and assumed the 40-week treatment effects would persist for the 5-year study period under optimistc senarios. This extrapolation was inspired by the SURPASS-4 trial which demonstrated the sustainability of one-year treatment effects of tirzepatide on key biomarkers into the second year. Upon further investigation, we noted that the trajectories for these biomarkers from the 40th week to the end of the second year demonstrated a relatively stable trend, corroborating our assumption of persistence. Besides, our model projects a higher 5-year risk reduction in MACE (RR 0.64, 95% CI 0.61–0.67) for the 15 mg tirzepatide, consistent with the 50% reduced risk (HR: 0.50, 95% CI: 0.26–0.95) in MACE-4 (cardiovascular death, myocardial infarction, stroke, hospitalization for unstable angina) reported in the SURPASS-4 trial.¹⁸ If the trajectories for the biomarker continued to decrease sharply after 1 year, it could lead to an underestimation of our simulation results. However, we believed the actual 5-year effectiveness of tirzepatide would be more likely to lean toward the lower bound of our estimation (i.e., the optimistic scenario).

Previous trials of tirzepatide have demonstrated its favorable effects on critical risk factors for diabetes-related macrovascular and microvascular complications, including glucose control, weight reduction, blood pressure, and lipid profile. In our simulation experiment, we used the BRAVO model to translate 5-year treatment efficacy on key biomarkers to risk reductions of diabetes complications. For instance, the benefit of tirzepatide in preventing CHF can be attributed primarily to the significant weight loss (55%),HbA1c reduction(25%), SBP reduction(20%). The decompostion analysis showed that HbA1c reduction and body weight from tirzepatide contributed to most of the long-term clinical benefits. This was consistent with a previous study that found the reduction in HbA1c explaining 94.1% of the hazard reduction in MACE for newer glucose-lowering drugs.³⁵ Compared with insulin glargine and semaglutide, which reduced HbA1c by −1.1% and −1.9% the first year after initiation, tirzepatide at a dose of 5 mg, 10 mg, and 15 mg showed more robust HbA1c reductions in HbA1c by −2.1%,−2.4%,−2.5%, respectively. In addition, tirzepatide demonstrated a significant dose-dependent effect on weight reduction, with an average BMI change of −2.9 kg/m², −3.8 kg/m², and −4.6 kg/m² at a dose of 5 mg, 10 mg, and 15 mg, respectively. Such large drops in BMI contributed to 55% of the total risk reduction in CHF and, most importantly, 82% of risk reductions in all-cause mortality. According to our previous analysis, a constant 5 kg/m² BMI reduction could potentially increase one’s life expectancy by over a year.³⁶ Thus, the tirzepatide’s significant effect on body weight reduction was likely to result in a substantial amount of health outputs.

Results from the model calibration indicated no additional cardiovascular protective effect beyond the control of traditional risk factors existed for GLP-1 receptor agonists. According to our analysis, reductions in traditional risk factors explained more than 90% of the cardiovascular benefit observed in the included CVOTs. This finding was consistent with a previous study conducted by Giugliano D et al., where blood glucose reduction was proven to be highly related to risk reduction for MACE by a meta-analysis of 12 CVOTs for newer glucose-lowering drugs.³⁵ Thus, no GLP-1 receptor agonist-specific model calibrator was needed for this simulation experiment. In addition, being able to model the risk reductions of cardiovascular outcomes through traditional risk factors demonstrated the strong predictive validity of the BRAVO diabetes simulation model. General concerns existed about whether simulation models can be used to predict treatment effects, because prediction models, in general, captured associations instead of causal relationships. For example, instead of capturing the causal relationship between A1c and risk of MACE, a prediction model was more likely to only captures the association. However, the uniqueness of the BRAVO simulation model was that it was derived from the ACCORD trial, a two-by-two factorial trial that randomly assigned A1c and SBP goals and lipids treatment to the study subjects.³⁷ The model development process took advantage of the trial’s randomization process to ensure that the model was more capable of capturing causal relationships between modifiable predictors (i.e., A1c, SBP, and LDL) and outcomes than traditional predictive models. The model calibration of this study provided compelling evidence to support this argument.

As tirzepatide demonstrated superior reductions in the risk of macrovascular and microvascular complications, this novel GIP and GLP-1 receptor agonist can potentially be an effective second-line therapy to treat type 2 diabetes.^5,38 The US Food and Drug Administration approved tirzepatide injection to improve blood sugar control in adults with type 2 diabetes in May 2020. However, the economic value of tirzepatide needs to be assessed before including it in the formulary. Dongzhe and colleagues systematically reviewed the cost-effectiveness studies on newer glucose-lowering medications and found GLP-1 receptor agonists were cost-effective when compared with insulin from 24 studies.³⁹ However, the cost-effectiveness of GIP and GLP-1 receptor agonist remains unknown. We are evaluating the cost-effectiveness of tirzepatide as second-line therapy for US adults with type 2 diabetes and estimating the price threshold for this new agent to be cost-effective.

Our study has several strengths. First, this study produced timely insight on information that often took years through a CVOT to generate. Through an innovative experimental design, we extrapolated tirzepatide’s short-term treatment effects on key risk factors to long-term macrovascular and microvascular outcomes without a CVOT. Second, the use of optimistic and conservative scenarios helped us gain insight into the impact of uncertainties driven by the effect persistency on the benefit of tirzepatide. Last, we have validated the BRAVO diabetes simulation model comprehensively before conducting the simulation experiment to minimize prediction errors. The ideal prediction accuracy during the validation and after the calibration increased our confidence in the accuracy of our prediction.

Our study also has several limitations. First, the one-year treatment effects of insulin glargine on SURPASS-2 population were estimated by subtracting the difference in treatment effects between insulin glargine and semaglutide in the SUSTAIN-4 trial from the short-term treatment effects of semaglutide in the SURPASS-2 trial. Given the different populations in the SUSTAIN-4 trial and SURPASS-2 trial, this estimation may need further examination, even though the one-year differences in treatment effects on key biomarkers for tirzepatide (5 mg, 10 mg, and 15 mg) versus insulin glargine observed in SURPASS-4 trial was similar to our estimation in this study.¹⁸ Second, our simulation was conducted under the assumption that tirezepatide reduced the risk of diabetes complications under a similar mechanism as other GLP-1 receptor agonists classes. This assumption may not hold if tirzepatide, the dual GIP and GLP-1 receptor agonist, can further reduce the risk of cardiovascular and microvascular complications under mechanisms that are beyond traditional biomarker control. Third, the simulation population from the SURPASS-2 studies primarily includes patients in the early stages of type 2 diabetes transitioning from metformin. The BRAVO model developed from the ACCORD study, characterized by older participant with high risk for cardiovascular disease events. Several modeling assumptions derived from CVOTs may not entirely align with the demographic profile and risk factor levels of simulation population, while we’ve incorporated key predictors such as age, diabetes duration, and complication history into all BRAVO risk equations. Forth, potential long-term safety concerns associated with tirzepatide warrant attention. For instance, tirzepatide showed significant effect on body weight reduction, although beneficial in the management of diabetes, may potentially lead to excessive weight loss or contribute to frailty over the long-term. Our model, though predicting the five-year positive benefits of tirzepatide, didn’t fully account for possible compensating factors or the long-term safety issues. In addition, this study didn’t capture the potential benefits of higher doses of semaglutide as comparators. Higher dosages of semaglutide (2mg) could confer greater therapeutic benefits. While our model endeavors to provide a comprehensive and predictive framework, it remains a simplification of complex biological, clinical, and sociological interactions. These fingding should be interpreted judiciously, always considering the broader context and the inherent uncertainties associated with modeling.

With the use of BRAVO diabetes model, we predicted the potential long-term benefits of tirzepatide and semaglutide in reducing the risk of macrovascular and microvascular complications among individuals with type 2 diabetes, compared to that of insulin glargine. Based on the current modeling assumptions, tirzepatide (15 mg) may potentially outperform semaglutide (1 mg) in this respect. However, It’s crucial to highlight that these findings hinge upon our model’s assumptions and higher doses of semaglutide were not assessed in this study. Furthermore, due to the inherent limitations of our modelling approach, the long-term cardiovascular benefit of tirzepatide should be further validated in a prospective clinical trial.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.