Despite more than 140 years of clinical use, the long-term prognostic impact of nitrates remains uncertain (1). Nitrates continue to be among the most commonly prescribed medications for patients with coronary artery disease (CAD). According to current guidelines, long-acting nitrates may be used for angina prevention or symptom relief in patients who remain symptomatic despite treatment with aspirin, statins, β-blockers, or calcium channel blockers (2). In addition CAD, nitrates also have been recommended in the treatment of congestive heart failure (3). However, controversy surrounding nitrate use has persisted.

Several studies have suggested that long-term nitrate therapy may be associated with increased adverse clinical events in patients with CAD. Nevertheless, few studies have explicitly examined whether baseline left ventricular ejection fraction (LVEF) modifies this association (4–6). In some studies involving patients with reduced LVEF, nitrate treatment has demonstrated convincing benefits in terms of remodeling and exercise tolerance. In the case of African-American patients with LVEF ≤ 45%, nitrate treatment was even associated with reduced mortality (7, 8). However, approximately half of the patients enrolled in these studies did not have CAD. Therefore, evidence regarding the impact of nitrate therapy on patients with CAD across different levels of LVEF remains limited.

Discharge medications, as a critical component of the discharge summary, play a very important role in patient safety and continuity of care (9). The purpose of this study was therefore to analyze and compare if CAD patients discharged on nitrates experienced different prognoses across different levels of LVEF.

A total of 3,714 patients diagnosed with coronary heart disease by coronary angiography in the cardiology department at Guizhou Provincial People's Hospital between July 2012 and August 2016 were retrospectively selected. Of these, 3,367 patients (90.7%) completed follow-up, while 347 patients (9.3%) were lost to follow-up. The mean duration of follow-up was 27.2 ± 13.5 months. The follow-up flowchart is presented in Figure 1. Among the 3,367 patients, another 963 patients were excluded from this study because of in-hospital death, incomplete follow-up, absent echocardiograph records, or missing discharge records. Finally, 2,404 patients were eligible for this study. Of these, 1,153 patients (48%) were discharged on nitrates and 1,251 patients (52%) were not discharged on nitrates. Classifying these patients into three groups based upon LVEF, 299 patients had LVEF < 45%, 333 LVEF between 45% to 55%, and 1,772 LVEF > 55%.

The detailed baseline characteristics of the study population are shown in Table 1. Based on these sample sizes, a post-hoc power analysis indicated that the statistical power was sufficient (>80%) to detect clinically relevant hazard ratios in the LVEF < 45% and >55% subgroups, which comprised the majority of the study population. Compared with patients not discharged on nitrates, patients discharged on nitrates had similar distributions of sex, acute coronary syndrome, diabetes, body mass index, and LDL-C. However, patients with nitrates at discharge were older, appeared to have more hypertension, were less likely to have a history of myocardial infarction, and had fewer current smokers. Moreover, these patients seemed to have higher LVEF and serum creatinine, and fewer implanted stents during the index procedures. Importantly, patients discharged on nitrates seemed to take aspirin, statin, angiotensin-converting enzyme inhibitor (ACEI)/angiotensin receptor blocker (ARB), β-blockers, and calcium channel blocker (CCB) before admission. However, patients discharged on nitrates received comparable rates of aspirin, clopidogrel, and ACEI/ARB, but were prescribed fewer statin and more CCB and β-blockers.

Chronic nitrate therapy remains widely used for symptom control in CAD, yet its long-term prognostic impact has not been established in contemporary practice. Moreover, whether baseline left ventricular systolic function modifies any association between nitrates and clinical outcomes has been insufficiently characterized. In our angiography-confirmed CAD cohort, nearly half of the patients were discharged on long-acting nitrates. In the overall population, nitrate prescription at discharge was not associated with all-cause mortality or MACE. However, after stratification by LVEF, an adverse association was confined to patients with LVEF < 45%, in whom discharge on nitrates was independently associated with higher risks of all-cause death and MACE; no significant associations were observed in patients with LVEF 45%–55% or >55%.

These findings are clinically relevant because long-acting nitrates continue to be prescribed frequently in coronary artery disease, despite limited evidence for sustained prognostic benefit. Current guidelines recommend long-acting nitrates primarily for angina prevention and symptom relief, particularly when symptoms persist despite first-line antianginal therapy (13). In contrast, several observational studies have raised concerns about chronic nitrate exposure in ischemic heart disease. Using databases from two large postinfarction studies, Nakamura and colleagues reported an association between long-term nitrate use and higher mortality (17, 18). Similarly, Ishikawa et al. found that long-term oral nitrate therapy was associated with more cardiac events in a cohort of patients with healed myocardial infarction (4), and Kanamasa et al. reported higher event rates during follow-up among patients treated continuously with oral and transdermal nitrates (6). More contemporary registry data have extended these observations to other clinical contexts. For example, among higher-risk stable coronary disease populations—including patients with diabetes undergoing elective percutaneous coronary intervention—chronic oral nitrate therapy has been linked to increased major adverse cardiovascular events in observational analyses (19). In addition, in cohorts of vasospastic angina, long-acting nitrates have not demonstrated prognostic benefit, and certain nitrate-based regimens have been associated with higher rates of adverse events (20–22). Taken together, although these data are subject to confounding by indication, the consistency of signals across diverse cohorts supports the possibility that chronic nitrate therapy may be prognostically non-neutral in selected subgroups, which provides clinical context for our finding of an adverse association primarily in patients with reduced LVEF.

However, most prior studies have not evaluated whether baseline LVEF modifies the association between chronic nitrate therapy and clinical outcomes in coronary artery disease. To address this gap, we assessed discharge prescriptions for long-acting nitrates across prespecified LVEF strata in an angiographically confirmed CAD cohort. The primary finding was that an adverse association with both all-cause mortality and major adverse cardiovascular events was observed among patients with LVEF < 45%, whereas no significant associations were detected among those with LVEF ≥ 45%. Although the observational design precludes causal inference and residual confounding remains possible, these data suggest that the clinical risk profile associated with chronic nitrate therapy may differ according to baseline systolic function.

Although these results may appear to conflict with heart failure trials that evaluated nitrate-based regimens, several distinctions are important. First, the trials most often cited in support of nitrate therapy in HFrEF evaluated hydralazine plus isosorbide dinitrate rather than nitrate monotherapy, and they were conducted in therapeutic eras that differed substantially from contemporary guideline-directed medical therapy (8, 23). In our cohort, most patients received background neurohormonal therapies, including an ACEI or ARB and a beta blocker, which may influence both symptoms and prognosis and may also modify the net effect of vasodilator therapy. In A HeFT, which enrolled African American patients with heart failure, hydralazine plus isosorbide dinitrate was associated with reduced mortality; however, the restriction to a single racial group and early trial termination limit generalizability, particularly to non-Black populations (8). Consistently, a large observational cohort study including 76,828 patients with heart failure did not show an apparent benefit of hydralazine plus nitrate therapy overall or within major racial or ethnic subgroups (24). Furthermore, contemporary guidance positions hydralazine plus isosorbide dinitrate as a therapy for selected patients with HFrEF rather than a broadly applicable nitrate strategy to improve prognosis (25). Finally, more recent randomized data in HFpEF have not supported benefit from nitrate-based approaches. Extended release isosorbide mononitrate reduced daily activity without improving exercise capacity or quality of life in NEAT HFpEF (26), and inhaled inorganic nitrite did not improve peak oxygen consumption in INDIE HFpEF (27). Taken together, the broader contemporary literature suggests that nitrate-related strategies are not uniformly beneficial across heart failure phenotypes, which is compatible with our observation that adverse associations were confined to CAD patients with reduced LVEF.

Nitrate tolerance, the potential role of intermittent nitrate therapy, neurohumoral effects, aggravation of myocardial ischemia, and hypervascular responses in coronary arteries may contribute to the adverse outcomes associated with chronic nitrate therapy (28, 29). In the setting of prior myocardial infarction, the ischemic myocardial microenvironment is often characterized by persistent oxidative stress, inflammation, and microvascular dysfunction, which may further impair myocardial and vascular homeostasis (30). In addition, the thrombotic microenvironment in postmyocardial infarction patients can exacerbate endothelial dysfunction, further compromising vascular integrity and increasing the risk of adverse cardiovascular events (31, 32). These factors may contribute to the impaired endothelial function observed in patients with reduced LVEF, which could amplify their vulnerability to additional endothelial or hemodynamic stress. As a result, this adverse ischemic milieu may potentially contribute to the unfavorable associations observed with chronic nitrate therapy in these patients. In addition, both experimental and human studies have suggested that sustained exposure to organic nitrates may impair endothelial function through oxidative stress-related pathways (33, 34). In a well-characterized animal model, short-term treatment with isosorbide 5 mononitrate was associated with endothelial dysfunction and increased oxidative stress even in the absence of clear hemodynamic tolerance, which supports the possibility that vascular dysfunction may occur independently of classic tolerance (33). Consistently, once-daily isosorbide 5 mononitrate therapy impaired endothelial-dependent vasomotor function in humans, and this effect was reversed by antioxidant supplementation, further implicating reactive oxygen species in this process (34). Because hydralazine has been reported to exert antioxidant-related effects, it is possible that concomitant hydralazine could attenuate nitrate-associated oxidative stress, thereby influencing the net clinical effects of combination therapy; however, this remains a mechanistic hypothesis rather than a definitive explanation for trial outcomes (35). Endothelial dysfunction is an established marker of worse long-term prognosis in patients with coronary artery disease and heart failure (36–38). Moreover, patients with reduced systolic function may have less physiologic reserve and could be more vulnerable to additional vascular dysfunction, which may help explain why adverse associations with nitrates were observed in patients with LVEF < 45% but not in those with preserved systolic function (37–39).

In addition to the primary exposure of interest, several established clinical and treatment-related factors were independently associated with outcomes, supporting the overall clinical plausibility of our models. In the Cox regression analyses, preadmission beta blocker use, higher LVEF, discharge aspirin prescription, and stent implantation were associated with lower risks of all-cause mortality and major adverse cardiovascular events, whereas older age, higher serum creatinine and white blood cell count, left main coronary artery disease, and current smoking were associated with higher risks of these outcomes. These associations are consistent with prior reports in patients with coronary artery disease and heart failure (40–44). By contrast, the direction of the association between discharge statin therapy and outcomes in the subgroup with LVEF greater than 55% did not align with the broader literature supporting statins for secondary prevention (13, 45). This finding should be interpreted cautiously because only a small proportion of patients were not discharged on statins, which may have limited statistical power and increased susceptibility to imbalance and model instability. In addition, residual confounding and treatment selection related to disease severity or comorbidity could have influenced this estimate. Accordingly, while these covariate associations provide clinical context for our main findings, the statin signal in this subgroup should be viewed as exploratory and is further addressed in the limitations.

This study has several limitations. First, its retrospective, single-center observational design precludes causal inference and leaves the possibility of residual confounding despite multivariable adjustment. Second, nitrate exposure was defined according to discharge prescription, and detailed information on postdischarge adherence, treatment duration, or dose changes was not available, which may have resulted in exposure misclassification. Third, coronary artery disease was defined using angiographic criteria without routine physiological assessment, and anatomical stenosis severity may not always reflect functional ischemic significance. Fourth, echocardiographic data were not available for all patients undergoing coronary angiography, potentially introducing selection bias. Finally, as the cohort was derived from a single center, the generalizability of these findings to other populations requires external validation.

Our study suggests that nearly half of patients with coronary artery disease are discharged on nitrates. The impact of prescribed nitrates at discharge on prognosis is diverse across LVEF. Being discharged on nitrates was not a predictor of MACE or all-cause death in patients with normal or mildly reduced LVEF, but it was a strong predictor among those with moderately or severely reduced LVEF.