When Does Metabolic Memory Start? Insights From the Association of Medical Diabetologists Annals Initiative on Stringent HbA_1c Targets

Poor glycemic control in patients with type 2 diabetes (T2D) is associated with increased risk of cardiovascular diseases (CVD) (1). On the other hand, an early, intensive glycemic control has been associated with a long-term benefit in the development of CVD, a phenomenon referred to as the legacy effect (2). This phenomenon was first described in the follow-up of the United Kingdom Prospective Diabetes Study (UKPDS) trial, showing that patients with a recent diagnosis of T2D benefited from an intensive glycemic control even after the intensive therapy was discontinued (3). While some subsequent studies in patients with more advanced stages of T2D did not confirm these results (4–6), three large observational studies and updated follow-up data of the same UKPDS cohort provided solid evidence that T2D patients with poor glycemic control in the years following diagnosis have an increased risk of late CVD and death, supporting the existence of long-lasting damage promoted by hyperglycemia (7–10). However, in most of the studies exploring the legacy effect, strict metabolic control was obtained using sulfonylureas or insulin, and the elevated risk of cardiovascular events and death associated with hypoglycemia could mitigate the positive effects of tight metabolic control (11). For this reason, existing guidelines generally recommend an HbA_1c target <7.0% in the vast majority of patients or suggest <6.5% in selected subgroups (12,13).

Thus, evidence on the potential benefits obtained by achieving more stringent glycated targets, as close as possible to a normal HbA_1c value, that is, HbA_1c <5.7%, is very limited to date, especially in real-world studies in patients with T2D. To explore this aspect, we used data from a large Italian clinical registry of people with T2D, the Association of Medical Diabetologists (AMD) Annals Initiative. Newly diagnosed patients free of CVD at baseline were stratified according to the average HbA_1c attained during the first 12, 24, and 36 months from diagnosis, and the incidence of CVD in the following years was assessed.

Data were derived from the registry of the Italian AMD Annals Initiative, which was established in 2004 to monitor quality of diabetes care in Italy (14). The database includes information on all patients with T2D receiving care at over 300 diabetes clinics in Italy from 1 January 2004 to 31 December 2022. All diabetes clinics adhering to AMD Annals Initiative, a third of those existing throughout the country, used a common electronic clinical record system for the everyday management of outpatients, and software was specifically developed to extract information from these clinical databases. Anonymized data from all participating clinics were collected and centrally analyzed. Available data included demographic, clinical, and biochemical information. The use of specific classes of drugs (glucose-lowering, lipid-lowering, and antihypertensive agents), based on Anatomical Therapeutic Chemical codes, was available. Information on the presence of diabetes complications was based on International Classification of Diseases, Ninth Revision, Clinical Modification, codes. The study design is summarized in Fig. 1. To explore the effects of various periods of early glycemic exposure, we adopted three definitions of early exposure periods (0–1, 0–2, and 0–3 years). The mean HbA_1c value was calculated for each early exposure period by using all HbA_1c measurements, except the value at diagnosis. The value at diagnosis was excluded since it reflects control before treatment was initiated, and the glycemic legacy effect has been demonstrated only in populations receiving diabetes treatment. To assess the effect of various degrees of glycemic control, the mean HbA_1c value for each of the three early exposure periods was categorized into either HbA_1c <5.7% (<39 mmol/mol), 5.7–6.4% (39–46 mmol/mol), 6.5–7.0% (47–53 mmol/mol), 7.1–8.0% (53–64 mmol/mol), or >8.0% (>64 mmol/mol).

The exposure period started at the date of diagnosis and ended after 12, 24, or 36 months from diagnosis. Accordingly, the follow-up period started after 12, 24, or 36 months from diagnosis (baseline/t0) and ended after the first occurrence of the outcome of interest or was censored at last visit. The outcome of interest was the composite of myocardial infarction, stroke, coronary or peripheral revascularization, and coronary or peripheral bypass. Patients with prevalent CVD at baseline were excluded. The risk factors used to adjust the analysis were measured at baseline. In case of missing data relative to covariates, a category of missing data was added for each covariate in the multivariate analysis.

The analyses are based on data relating to all newly diagnosed T2D patients seen by participating centers in the period 2010–2019.

Data on patient characteristics were summarized using means and SDs for continuous variables and counts and percentages for categorical variables, and stratified by the five classes of mean HbA_1c in each early exposure period. Characteristics were compared by the Kruskal-Wallis one-way ANOVA and χ² test for continuous and categorical variables, respectively.

Cox proportional hazards models were used to examine associations between glycemic control and the risk of CVD. Cox models were adjusted for baseline potentially confounding variables: sex (male vs. female), age (by 5 years), total cholesterol (by 10 mg/dL), HDL cholesterol (by 10 mg/dL) and LDL cholesterol (by 10 mg/dL), triglycerides (by 10 mg/dL), BMI, systolic blood pressure (by 5 mmHg), smoking status (yes vs. no), eGFR (<60 vs. ≥60 mL/min/1.73 m²), microalbuminuria (yes vs. no), use of different classes of glucose-lowering drugs (yes vs. no for each one), use of statin (yes vs. no), use of antihypertensive medication (yes vs. no), HbA_1c, and number of HbA_1c measurements during the exposure period. A backward selection was introduced in the Cox models to exclude the confounders without a significant association with the outcome. For each model, patient follow-up was censored after the first occurrence of the outcome of interest or last visit. Results of Cox models are expressed as hazard ratios (HRs) with their 95% CI. Descriptive and multivariate analyses were performed in the three cohorts. A two-sided P < 0.05 was considered statistically significant for all analyses. Analyses were performed using SAS 9.4 statistical software (SAS Institute, Cary, NC).

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

The analysis involved a total of 251,339 individuals with newly diagnosed T2D and free of CVD at baseline, seen between 2010 and 2019. Overall, 5.7% of patients presented average HbA_1c values <5.7% during the first year after diagnosis, while 29.0% showed average values between 5.7 and 6.4%. Conversely, the average HbA_1c values during the first year exceeded 8.0% in 16.7% of case patients.

Clinical characteristics of the groups categorized according to the degree of glycemic control attained in the 12 months after diagnosis are reported in Table 1, while the characteristics of the groups categorized according to the degree of glycemic control attained in the 24 and 36 months after diagnosis are reported in Supplementary Tables 1 and 2.

Overall, individuals with HbA_1c <5.7% after diagnosis were younger and had lower BMI, a better lipid profile, and lower blood pressure levels as compared with the overall population. They also showed a lower prevalence of albuminuria and reduced eGFR. The groups showed statistically significant differences for all the characteristics assessed. Therefore, all these variables were added as covariates to adjust the subsequent multivariate analyses.

During a mean follow-up of 4.6 ± 2.9 years, 13,822 patients (5.5%) developed a major cardiovascular event.

Cox analysis shows that, compared with patients with a mean HbA_1c <5.7%, those above this threshold had an increased risk of CVD at follow-up for all the three early exposure periods and for all strata of glycemic control considered (Fig. 2). In detail, compared with mean HbA_1c <5.7% during the first year after diagnosis, patients with mean HbA_1c between 5.7 and 6.4% had a 24% increased risk of CVD (HR = 1.24; 95% CI 1.13–1.35), those with HbA_1c between 6.5 and 7.0% had a 42% increased risk (HR = 1.42; 95% CI 1.30–1.56), those with HbA_1c between 7.1 and 8.0% had a 49% increased risk (HR = 1.49; 95% CI 1.36–1.64), and patients with HbA_1c >8.0% had a 56% increased risk of CVD (HR = 1.56; 95% CI 1.42–1.72). The same trend was documented for the 0- to 2-year and the 0- to 3-year exposure periods (Fig. 2). Results of Cox models in the three different periods of exposure are reported in Supplementary Table 3.

The legacy effect is a well-recognized phenomenon clearly documented in cohort studies and in selected clinical trials, suggesting that poor glycemic control after T2D diagnosis promotes enduring damage on the vasculature (2,7–10). A recent analysis from the UKPDS study evaluated the impact of historical HbA_1c values on CVD risk, demonstrating that a 1% HbA_1c reduction obtained at diagnosis was able to reduce CVD risk by 27%, but the same reduction obtained 10 years later lost most of its beneficial potential in CVD risk (10).

It is well known that, in T2D patients, cardiovascular risk is a continuum that starts early in the clinical history of the disease, because of the contribution of several risk factors including hyperglycemia. Although lipidologists recommend achieving LDL cholesterol targets as stringent as possible according to the principle of “lower is better” and, after the introduction in the market of PCSK-9 inhibitors, “lowest is best,” this principle was hard to propose in the diabetology field because of the risks associated with hypoglycemia.

However, today, many drugs do not cause hypoglycemia, allowing us to rethink proper glucose targets in many patients. In fact, recommended glucose-lowering drugs with proven cardiovascular benefit, for example, sodium–glucose transport protein 2 inhibitors (SGLT2i) and GLP1 receptor agonists (GLP1-RA), are not burdened by the risk of hypoglycemia, also allowing the safe attainment of stringent glycemic targets. Furthermore, the SURPASS study program (15) showed that treatment with the new incretin system receptor agonist tirzepatide was associated with a high percentage of patients safely achieving the HbA_1c target of <6.5%. At maximum dosage, one in two patients reached a “normal” glycated hemoglobin value, that is, a value <5.7%, suggesting the possibility of potential new therapeutic approaches and new metabolic targets (13), which are still hard to achieve in the clinical practice (16).

In this regard, the strategy to readily address glucose control with a combination therapy at T2D diagnosis has been tested in Vildagliptin Efficacy in Combination with Metformin for Early Treatment of Type 2 Diabetes (VERIFY), a large randomized, double-blind, parallel-group study of newly diagnosed patients with T2D conducted across 34 countries. This study demonstrated the long-lasting beneficial effects of an early intensive combination intervention on glucose control, compared with the standard approach over the 5-year study duration. Moreover, this strategy was not associated with unexpected or safety issues, including hypoglycemia (17).

However, the benefit of achieving such stringent HbA_1c targets in a real-world setting has not been proven so far.

Moreover, it has been increasingly acknowledged that using only HbA_1c measurement does not allow evaluation of other important aspects of glycemic disturbance, such as glycemic variability, which has been previously demonstrated to predict CVD complications in this same cohort of individuals with T2D, irrespective of the achievement of targets (18). In the current analysis, we looked at another unsolved issue, which is whether the HbA_1c goals that we pursue in accordance with current guidelines would really reflect the CVD risk of our patients.

Accordingly, the current analysis from the AMD Annals data set aimed to answer two important clinical questions. First, is it feasible to achieve glucose targets in the “normal range,” that is, HbA_1c values <5.7% in our T2D outpatients? Second, is it worthy from a CVD risk perspective?

To address these questions, we evaluated a large cohort of newly diagnosed T2D patients with no CVD at baseline and followed up with them for >4 years to assess the incidence of CVD events (composite end point) according to different HbA_1c cut-offs achieved during 1, 2, and 3 years of their initial clinical history.

Here we show, in a sample of over 250,000 newly diagnosed T2D patients, the benefits of reaching and maintaining ambitious targets immediately after diagnosis, in terms of reducing cardiovascular risk. Comparing patients who maintained HbA_1c levels <5.7% during the first year after diagnosis to those who had HbA_1c between 5.7 and 6.4%, our data show, for the latter, an increase in CVD risk of 24%, rising to 27% if the target HbA_1c <5.7% was maintained for 2 or 3 years. The excess CVD risk reached 42% in patients who showed HbA_1c levels between 6.5 and 7.0%, the target generally recommended by existing guidelines for the majority of patients, excluding elderly, frail individuals and those with severe comorbidities. The CVD risk further increased in poorly controlled individuals, reaching 56%, 66%, and 77% among those with average HbA_1c levels >8.0% during 1, 2, and 3 years after diagnosis, respectively.

Notably, the risk associated with HbA_1c levels just above the “diagnostic” cut-off values was independent from major risk factors, including sex, age, lipid profile, BMI, systolic blood pressure, smoking status, renal function, use of different classes of glucose-lowering drugs, statins, and antihypertensive medication, and the number of HbA_1c measurements during the exposure period.

In this regard, a meta-analysis recently demonstrated that the intensive antihyperglycemic approach significantly reduces the incidence of cardiovascular outcomes (major adverse cardiovascular events) compared with conventional treatment when all available studies are considered (odds ratio [OR] = 0.86; 95% CI 0.77–0.96; P = 0.007), with a more consistent effect in the case of randomized controlled trials (RCTs) that enrolled patients with diabetes lasting <10 years (OR = 0.73; 95% CI 0.56–0.94; P = 0.01), and an even more pronounced protection when analyzing only RCTs that enrolled patients without previous cardiovascular events at baseline (OR = 0.64; 95% CI 0.48–0.86; P = 0.003). These results support the recommendation to intensify antihyperglycemic treatment with a view to cardiovascular prevention, in patients with short duration of diabetes, without previous cardiovascular disease and with a long life expectancy (6).

Of note, the positive results of our study were obtained despite the fact that only a minority of patients were treated with an SGLT2i or a GLP1-RA. These glucose-lowering agents were consistently proven to provide cardiorenal protection while showing a very low risk of hypoglycemia (19). Furthermore, a recent, large study on data derived from the AMD Annals database suggested that the early introduction of SGLT2i might be able to ameliorate or even suppress the noxious long-term consequence of early, poor glycemic control on the vasculature (9).

Our study has important strengths, particularly the very large sample size, representative of routine clinical practice, and the long follow-up.

The study also has limitations. Despite our effort to adjust for all known risk factors, residual unmeasured confounders are inherently linked to all registry-based studies. Moreover, differences in baseline disease severity are likely intertwined with the inability of reaching HbA_1c targets early in the course of the disease. It is important to note that, as the reference group used less medications (for instance, insulin therapy was reported in 16% vs. 42% of the <5.7% vs. >8% HbA_1c groups, respectively), the possibility that our results may reflect an earlier stage of the disease cannot be ruled out, because of the observational nature of our study, despite multivariate adjustments for all major CVD risk factors and diabetes-related variables, and the large study sample. Finally, no information on hypoglycemic episodes was available, since it is not usually reported in a standard classification in medical records, precluding the possibility of investigating this important aspect. However, our results showing that the lower the HbA_1c level, the lower the risk of CV events suggest that the availability of modern glucose-lowering drugs associated with a very low hypoglycemic risk should allow the safe attainment of a more ambitious HbA_1c target.

Despite these limitations, the study showed a consistent trend of CVD risk with increasing average HbA_1c levels, and the same trend was confirmed when analyzing data relative to 1, 2, or 3 years of early exposure, thus testifying to the robustness of our findings.

Achieving more ambitious glycemic targets, immediately after diagnosis, may prevent cardiovascular risk in patients with T2D, and the findings of the current analysis support the importance of not neglecting the treat-to-target approach in a treat-to-benefit era, provided that an optimal metabolic control is achieved with the most appropriate drugs.

In conclusion, our results suggest that, in a large population of T2D outpatients, in routine care, the early achievement of more stringent targets of HbA_1c in the normal range (i.e., HbA_1c <5.7%) is feasible and worthy in terms of CVD prevention. Further specifically designed RCTs are urgently needed to confirm our observational data.

Funding. The study was promoted by AMD. The list of participating centers can be found at the website https://aemmedi.it/wp-content/uploads/2020/10/Annali-nuova-versione-2020_1-ok.pdf. The analysis was partly supported by Eli Lilly SpA.

Duality of Interest. G.T.R. is on the advisory board and does consultancy and lectures for Novo Nordisk, AstraZeneca, Sanofi, Boehringer, Lilly, Mundipharma, and Sanchio. A.N. has received honoraria from AstraZeneca, Eli Lilly, and Novo Nordisk, and research support from Alfasigma, Novo Nordisk, Sanofi, Shionogi, and Sobi. M.C.R. has received research support from Alfasigma, Novo Nordisk, Sanofi, Shionogi, and Sobi. A.C. is on the advisory board and does consultancy and lectures for AstraZeneca, Berlin-Chemie, Eli Lilly, Novo Nordisk, Mitsubishi, Roche Diagnostics, and Theras Lifetech. F.P. is a lecturer for Berlin-Chemie. V.M. has received lecture fees from MSD, AstraZeneca, and Eli Lilly. S.D.C. received honoraria for lectures from Eli Lilly, Boehringer, AstraZeneca, Mundipharma, MSD, Sanofi, Novo Nordisk, Daiichi Sankyo, and Bayer. R.C. has received consultancy fees from Boehringer Ingelheim, Eli-Lilly, Novo Nordisk, AstraZeneca, Sanofi-Aventis, Roche Diabetes Care; speaking fees from AstraZeneca, Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Sanofi, Mundipharma Pharmaceutical, Abbott, MSD, Neopharmed Gentili, Menarini, Essex Italia, and Ascensia Diabetes. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. G.T.R. and A.N. conceived the idea and wrote the manuscript. A.N., G.L., and M.C.R. contributed to study design, made the statistical analysis, and discussed the manuscript. A.C., F.P., V.M., A.R., P.D.B., S.D.C., G.D.C., and R.C. reviewed the manuscript for intellectual content. All authors collected data and approved the final version of the manuscript. R.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.