Prognostic Implication of Admission HbA1c for All-cause Mortality in Ischemic Heart Failure Among a Chinese Population: A Prospective Cohort Study

Background: The impact of glycosylated hemoglobin A1c (HbA1c) on heart failure (HF) and ischemic heart disease (IHD) differs among studies, and IHD is a preponderant cause in HF. We investigated the link between admission HbA1c and all-cause mortality in a Chinese population with ischemic heart failure (IHF). Methods: Eligible patients with IHF at the Department of Cardiology, Guangdong Provincial People’s Hospital from December 2015 to June 2019 were enrolled to investigate the association between admission HbA1c and all-cause mortality of IHF with Kaplan-Meier survival analysis and Cox regression analysis. Results: Of 1413 participants, the median age was 63.2 ± 10.9 years, 85% were men and median admission HbA1c level was 6.82%. All-cause mortality was higher in HbA1c >7% group compared with HbA1c ≤ 7% group (hazard ratio (HR): 1.328, 95% condence interval (CI): 1.016-1.735, p =0.037), and sensitivity analysis appeared the consistent result. The association between HbA1c and all-cause mortality was also statistically signicant in the male and younger cohorts. Conclusions: Elevated admission HbA1c level (>7%) is an independent risk factor for all-cause mortality of IHF in the general population, and there is also a consistent trend among male and younger individuals. Further explorations are required to elucidate whether glycemic management plays a crucial role in the progression of IHF within female and elderly population.

demonstrated signi cant effect on peril of HF or all-cause mortality in spite of risk reduction in macrovascular or coronary incidents [11,12]. Moreover, it seems that intensive glucose lowering modality was associated with risk increase in congestive HF and non-signi cant impact on MI, and HbA1c ≤7% even signi cantly raised mortality of advanced systolic HF with diabetes [13,14]. These evidences remain contradictory regarding the role of HbA1c level or glycemic administration in HF and myocardial ischemia related disorders.
In our prospective ischemic heart failure (IHF) cohort study among a Chinese population [15], we intended to identify all-cause mortality of IHF in relation to admission HbA1c level, where sex-and age-related discrepancies were additionally explored, in consideration of sex differences in coronary heart disease (CHD) [16] and HF [17], and glucose targets varying from younger to elderly individuals recommended by current diabetes guidelines [18,19].

Study design and population
Design of our single-center, prospective cohort study targeting IHF patients has been published previously [15]. Patients were selected who were hospitalized in the Department of Cardiology, Guangdong Provincial People's Hospital from December 2015 to June 2019. Eligible individuals were included with: a) ≥18 years of age; b) left ventricular ejection fraction (LVEF) <45% assessed by echocardiogram during hospitalization; c) IHD con rmed by coronary angiography, antecedent MI, and/or antecedent revascularization; d) available HbA1c data at admission. Patients were excluded with LVEF <45% due to non-ischemic etiologies or without any follow-up since discharge. This cohort study was approved by the Clinical Research Ethics Committee of Guangdong Provincial People's Hospital (No. 2017128H), as well as performed in accordance with the Declaration of Helsinki. All the participants provided written informed consent before enrollment.

Data collection
Baseline data of interest, including demographics, reasons for referral, vital sign at admission, comorbidities and hypoglycemic drugs, were obtained from electronic medical records of Guangdong Provincial People's Hospital. As described already [15], venous blood at admission was drawn for evaluation of HbA1c, creatinine, high-sensitivity C-reactive protein (hs-CRP), high-sensitivity cardiac troponin-T (hs-cTnT) and N-terminal pro-brain natriuretic peptide (NT-proBNP) concentrations, and fasting venous blood was drawn for evaluation of fasting plasma glucose (FPG) and lipid parameters at the second day after admission. Morning urine was collected to examine urine albumin creatine ratio (UACR).
High-performance liquid chromatography (hemoglobin testing system D-10, Bio-Rad laboratories Inc, America) was utilized to test HbA1c concentration. Body mass index (BMI) was calculated as body weight in kilogram divided by height in square meter. Diabetes status was de ned according to diagnostic criteria of the American Diabetes Association (ADA) guideline [20]. Con rmed diabetes was de ned as having been diagnosed diabetes priorly or two abnormal test results while in hospital (FPG ≥7.0mmol/L and HbA1c ≥6.5%). Pre-diabetes was de ned as FPG levels from 5.6 to 6.9mmol/L or HbA1c levels from 5.7 to 6.4%. Excluded diabetes was de ned as the absence of prior diagnosis of diabetes and normal results of blood glucose and HbA1c during the current hospital visit. Unde ned status was de ned as not meeting the above conditions (for example, FPG ≥7.0mmol/L while HbA1c <5.7% or FPG <7mmol/L while HbA1c ≥6.5%).

Follow-up and clinical endpoint
Methods of follow-up and clinical endpoint were presented previously [15]. Of note, all-cause mortality was served as clinical endpoint in the current study because of the unavailability of judgement for exact reasons of death.

Statistical analysis
Patients enrolled in the study were rstly divided into two groups based on HbA1c concentrations (HbA1c >7% and HbA1c ≤7%). In addition, we analyzed the sex and age (age ≥65 years old and age <65 years old) subgroups. Missing values of BMI, hs-CRP, hs-cTnT, NT-proBNP and UACR were lled by regression imputation. Continuous variables consistent with normal distribution were expressed as mean value ± standard deviation (SD), while the data consistent with skewed distribution expressed as median (interquartile range) and categorical variables were expressed as number and proportion. A student's ttest or Mann-Whitney U-test was used to evaluate continuous variables properly, and chi-square test or Fisher's exact test was used to compare categorical variables. Differences in baseline characteristics were examined between HbA1c >7% vs. HbA1c ≤7%.
Univariate and multivariate logistic regression analyses were then performed to discover factors connected with HbA1c >7%. Brie y, factors with P-value <0.1 in univariate analysis were further included in multivariate analysis with stepwise adjustment. Odds ratio (OR) and 95% con dence interval (CI) were recorded.
To determine the relationship between admission HbA1c level and all-cause mortality in IHF patients as well as in the sex and age subgroups, survival analysis with Log-Rank test and multivariate Cox proportional hazards analysis with stepwise adjustment were performed for covariates including baseline characteristics, clinical presentations, antecedent medical histories, hypoglycemic agents and medications prescribed at discharge. Due to Kaplan-Meier survival curves of HbA1c >7% and HbA1c ≤7% categories intersecting in the elderly subgroup, which means non-proportional risk model, time-dependent Cox regression analysis was then applied to identify the association between admission HbA1c level and all-cause mortality of IHF in the elderly subgroup.
Sensitivity analysis was lastly employed to verify robustness of our results. There were 567 participants with missing values of BMI in our study owing to their serious illnesses, making them need to rest in bed and unable to measure height or weight. The remainder with BMI data were subsequently selected to explore the relationship between HbA1c and all-cause mortality by survival analysis and multivariate Cox proportional hazards analysis. SPSS version 22.0 of Windows was utilized for all analyses, and two-sided P-value <0.05 was considered statistically signi cant.

Association between HbA1c and all-cause mortality in all patients
After a median follow-up time of 2.9 years, 233 patients died with an overall mortality rate of 16.5% in the study cohort. Figure 2 shows higher hazard of death among those with admission HbA1c >7%, compared to ones with HbA1c ≤7% (Log-Rank p =0.037). Cox proportional hazards regression analysis showed that HbA1c >7% was still associated with increased risk of all-cause mortality (HR: 1.330, 95%CI: 1.012-1.749, p =0.041) after stepwise adjustment for baseline characteristics, clinical presentations, antecedent medical histories and medications prescribed at discharge. Other independent predictors of all-cause mortality included age, BMI, TG, LVEF, ST-segment elevation myocardial infarction (STEMI), prior stroke and tumor, aspirin therapy, beta-blockers (BB) therapy and diuretics therapy ( Figure 3).
Association between HbA1c and all-cause mortality in sex and age subgroups Figure 4a exhibits that male patients with admission HbA1c >7% had higher hazard of all-cause mortality compared to those with HbA1c ≤7% (Log-Rank p =0.016). In the female population, linkage between HbA1c and all-cause mortality didn't reach statistical difference (Figure 4b). In multivariate Cox proportional hazards regression, male participants with HbA1c >7% predicted greater risk of all-cause mortality (HR: 1.419, 95% CI: 1.047-1.922, p =0.024) in comparison to ones with HbA1c ≤7% (Figure 5a). Figure 4c shows that all-cause mortality was signi cantly higher in the younger patients (age <65years old) with HbA1c >7% compare to those with HbA1c ≤7% (Log-Rank p =0.002). In multivariate Cox proportional hazards regression, there was higher hazard of all-cause mortality in younger sufferers with HbA1c >7% (HR: 1.531, 95% CI: 1.014-2.312, p =0.043) in contrast to ones with HbA1c ≤7% (Figure 5b). In the elderly cohort (age ≥65years old), Kaplan-Meier survival curves of HbA1c >7% and HbA1c ≤7% intersected, meaning proportional hazards rate is violated. Hence, we implemented time-dependent Cox regression analysis instead of Cox proportional hazards model. Table 3 illustrates that in the timedependent Cox analysis, no signi cant effect of HbA1c on all-cause mortality was observed in the elderly individuals.

Sensitivity analysis
In order to validate reliability of our results, we carried out sensitivity analysis in participants with available BMI data (n =846). Supplementary Table 1 presents baseline characteristics between patients with BMI data and without BMI data. Those without BMI data seemed to experience more serious illnesses during hospitalization and need to rest in bed, leading to inability to measure height and weight. The results of sensitivity analysis were similar to those from the imputed data (Supplementary Figure 1 and Supplementary Figure 2). In patients with BMI data, HbA1c >7% group suffered from greater risk of all-cause mortality compared with HbA1c ≤7% group (Log-Rank p =0.031). Multivariate Cox proportional hazards regression illustrated that HbA1c >7% remained predicting higher hazard of all-cause mortality than HbA1c ≤7% (HR: 1.552, 95% CI: 1.086-2.218, p =0.016).

Discussion
This prospective cohort study amongst a Chinese population with IHF denoted that higher level of admission HbA1c (HbA1c >7%) was signi cantly associated with increased all-cause mortality in general IHF sufferers, which was likewise exhibited within male and younger individuals, with the absence of such relevance in female and elderly ones on the contrary.
The present study primarily indicated that stringent glycemic administration (HbA1c ≤7%) facilitated reduction in risk of death and improved prognosis of general patients with IHF, conforming to glucose goals for a majority of DM subjects recommended by contemporary diabetes guidelines [18,19]. Diabetes contributes to structural cardiac alterations and HF partly via myocardial ischemia or infarction, where hyperglycemia and hyperinsulinemia escalate atherosclerosis [21]. HbA1c magnitude is positively correlated with coronary and peripheral atherosclerosis in non-diabetic population [22,23], along with severity of coronary artery stenosis in acute myocardial infarction (AMI) with DM and non-DM [24]. Meanwhile, HbA1c level is considered as one of the strongest predictors with regard to the peril of death, HF hospitalization and AMI amongst T2DM sufferers. Decreased HbA1c concentrations are linked with lower risks of these outcomes, more so of atherothrombotic incidents, even though HbA1c ≤7% marginally shrank risk of HF hospitalization [25]. Furthermore, some of AMI patients without known diabetes history have glucose metabolic disorders, and HbA1c ≥39 mmol/mol demonstrates a signi cant increase in risk of HF amongst those AMI subjects [26]. These ndings, combined with our current study, inform detrimental effects of elevated HbA1c level or hyperglycemia on HF and myocardial ischemia related alterations, and rigorous glycemic administration to achieve low HbA1c targets predisposes patients with IHF to better outcomes.
As is known, hypoglycemia (sulfonylureas and insulin), uid retention (thiazolidinediones), weight gain (sulfonylureas, insulin and thiazolidinediones) potentially prompt worsening in HF with certain hypoglycemic medications [7]. Intensively treated patients with poor prognosis of HF in the past studies [13,14] are likely to encounter more adverse reactions on account of a greater amount of conventional antihyperglycemic drugs to reach lower HbA1c goals. Moreover, in the study where elevated HbA1c level ameliorated survival of advanced systolic HF with diabetes, only 123 participants were involved, and ones with HbA1c >7% had higher ejection fraction (EF) [14], signaling the requirement of involving more subjects and comparable baseline data between HbA1c groups. The exact connection between HbA1c level and peril of IHF should be assessed by more research including randomized clinical trials.
With respect to sex-related disparities in linkage between death risk of IHF and HbA1c, our research implicated that HbA1c showed positive relationship with all-cause mortality of IHF among male sufferers, while there was no signi cant difference for female ones. IHF subjects involved in our investigation were recognized with LVEF <45%, which was de ned as HF with reduced ejection fraction (HFrEF). Previous study found higher lifetime risk of HFrEF in men than women, and it might be attributed to men encountering heavier load and prior episode of CHD compared with women [27]. In addition, it's thought that HFrEF is predominantly a disease of men, where macrovascular lesions (e.g. CAD and MI) have been postulated to play a critical role, and conversely, women preponderate in HF with preserved ejection fraction (HFpEF), attributable to their susceptibility to coronary microvascular dysfunction and endothelial in ammation [17], although other research has demonstrated that lifetime risk of HFpEF was almost on a par between both sexes [27]. It can therefore be assumed that male individuals with HFrEF bene t more from strict glucose conduct in the setting of ischemic cardiomyopathy, whilst female ones fail to display the connection due to their predisposition to HFpEF and a paucity of adequate female participants (25% in our study).
Another notable indication in our results was that elevated level of HbA1c was linked with poor survival of IHF in the younger category (age <65 years old) rather than in the elderly category (age ≥65 years old). It has previously been observed that risks of HF hospitalization, AMI and death related with diabetes illustrated a stepwise decline from younger to older age groups in the context of HbA1c beyond target range, and also of note, hazard of hospitalization for HF was not expected to shrink with HbA1c not outside target range [25], which is not congruous with our outcome. What's more, preceding evidence suggests that younger sufferers aged <55 years with T2DM are susceptible to excessive risk of HF, of which excess risk declines with advancing age [28]. Intensive glucose lowering therapy signi cantly raises risk of hypoglycemia [11,13], and diverse studies appear a consistently U shaped relationship between HbA1c and risk of death amongst HF subjects, with moderate HbA1c level (7% to 8%) conferring the lowest mortality [21]. Accordingly, up-to-date diabetes guidelines have recommended that more lenient HbA1c goals may be suitable for individuals with limited life expectancy and severe comorbidities [18,19]. A plausible explanation for the results and evidence might be that there are greater potential bene ts from more aggressive treatment on blood glucose for younger patients with IHF, whereas probable gains from lower admission HbA1c level (≤7%) across elderly ones are offset by deleterious impact of hypoglycemia.
Our observational investigation has several strengths and also some noteworthy limitations. Patients with HFrEF caused by ischemic cardiomyopathy or IHD in China were included. The positive association between admission HbA1c magnitude and peril of IHF in the majority of population was substantiated.
Nevertheless, this is a single center based and observational cohort study. More female participants are supposed to be incorporated to disclose the power of glucose control over prognosis of IHF in women. There was a comparatively small portion of application in novel glucose-lowering drugs in our participants, containing glucagon-like peptide-1 receptor agonists (GLP-1 RA), dipeptidyl peptidase-4 (DPP-4) inhibitors, and sodium-glucose cotransporter-2 (SGLT-2) inhibitors, signaling the need for more employment of these new therapies to recognize how drop in HbA1c contributed by them affects hazard of IHF.
In conclusion, elevated admission HbA1c level (>7%) is an important indicator of increased all-cause mortality of IHF among general patients, and so for male and younger individuals. Further explorations are required to elucidate whether glycemic management plays a crucial role in the progression of IHF within female and elderly population.

Declarations
Ethics approval and consent to participate The Clinical Research Ethics Committee of Guangdong Provincial People's Hospital approved the study protocol (No. 2017128H) in accordance with the Declaration of Helsinki, and written informed consent was obtained from all participants.

Consent for publication
The manuscript was approved by all authors for publication.

Availability of data and materials
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
All authors declare that they have no competing interests.

Funding
This study was supported by the Guangzhou Science and Technology Program (No. 202002030088).
Authors' contributions QWD contributed to the data collection, analysis and drafted the manuscript. CSC drafted the manuscript.
CAP critically revised the manuscript. XXJ contributed to the data collection. GZP, LJJ and LLW contributed to the conception and design of the work. All authors read and approved the nal manuscript.