Iron deficiency (ID) is believed to be prevalent among HIV-1-infected patients and is associated with a decreased physical capacity and increased mortality (1–3). It is also becoming recognized that iron plays an essential role in cardiovascular disease, especially in chronic heart failure (HF) (4, 5). Notably, ID is also known to drive mitochondrial abnormalities in cardiac muscle (6) by modulating cellular energy functions (7, 8) and regulating mitochondrial biogenesis (9). Thus, correcting ID in HIV-1 patients is necessary to improve anemias and reduce cardiovascular complications (10).

Circulating biomarkers, such as erythrocyte indices, serum iron, ferritin, and soluble transferrin receptor, have been widely used to identify ID in clinical settings because these tests are commonly available and easy to perform. Nevertheless, the accuracy of these tests is often negatively affected by inflammatory states and renal and liver conditions, which are more common in the context of HIV-1 infection (11). However, studies have shown that these circulating biomarkers are poor indicators of myocardial iron levels (12, 13).

With the advancement of magnetic resonance imaging (MRI), T2* cardiac MR (CMR) imaging has become widely used to assess myocardial iron overload in hematologic diseases, such as thalassemia (14). Studies have also shown that changes in iron within the myocardium to be highly correlated with intramyocardial hemorrhage in ischemic heart disease (15). However, only a limited number of studies have reported utilized T2* CMR for ID in non-ischemic cardiomyopathies (16, 17). Notably, the assessment of myocardial iron content in patients using CMR has not been reported in HIV-1-infected patients. We investigated whether myocardial iron content in HIV-1-infected patients is altered based on T2* CMR and whether those changes are associated with systolic dysfunction observed in HIV-1 patients.

The institutional ethics committee approved this prospective case-controlled study, and all participants gave written informed consent prior to CMR examinations. The data underlying this article will be shared on reasonable request to the corresponding author.

All statistical analyses were performed using SPSS (version 23.0, IBM statistics, Armonk, NY, USA) and GraphPad Prism software (Version 8.1, GraphPad Software Inc). Continuous variables were expressed as mean ± SD or median ± interquartile range depending on the normality of the data. Categorical variables were expressed as counts (percentages). The Shapiro-Wilk test was used to test if the data were distributed normally. Categorical variables were compared using the c² test (with a cell count >5) or the Fisher's exact test (with a cell count ≤5). Continuous variables were compared among the groups using a one-way analysis of variance or independent t-test analysis (for normal distributions) and Kruskal-Wallis tests (for non-normal distributions). Post-hoc analysis was performed for pairwise group comparisons. The Spearman rank correlation test was used to test correlations between the variables. Quantitative variables were transformed into categorical variables according to their normal range or cutoff values provided by receiver operating characteristic (ROC) curve analyses. A univariable and multivariable binary logistic regression analysis was applied to test the impact of clinical and CMR imaging variables on being able to predict mild LVSD in the 41 patients who underwent CMR examinations with Gd injection. After a forward selection of relevant covariates with a P < 0.1 for the univariable analysis, covariates were added to a multivariable model to assess the impact of the variables further. Reproducibility was assessed using a Bland-Altman analysis, linear regression, and intraclass correlation coefficients. The level of statistical significance was set at P < 0.05.

In a univariate model, our study revealed that the myocardial T2* was independently associated with mild LV dysfunction in HIV-1 infected patients. This association remained significant even after adjusting for other imaging co-variates in a multivariate model. Importantly, we found that the T2* of myocardium of HIV-1 infected patients to be significantly elevated compared to matched controls, suggesting myocardial iron deficiency is a hallmark in HIV-1 patients. Our findings here also lend early support for the hypothesis that restoration of iron could be a potential strategy to mitigate the functional impairment in HIV-1 infected patients.

Although systemic ID has been previously described, the present study is the first to demonstrate myocardial iron deficiency in HIV-1 subjects (23). Reduced myocardial iron level might result from systemic iron derangements due to decreased iron intake, poor bioavailability, and repeated HIV-1 infections (24). Also, higher level of hepcidin in both treated and untreated HIV patients can inhibit iron release by macrophages, decrease iron absorption through the internalization and impede ferroproteins degradation in enterocytes and macrophages (25). From our data, it is not possible to determine whether myocardial iron deficiency in HIV-1 infected patients were resulted from relative iron deficiency (decreased systemic iron availability despite overall normal total/body iron) or absolute iron deficiency (reduced iron stores). There was no significant difference among HIV+/LVEF+, HIV+/LVEF−, and healthy controls in hemoglobin and serum ferritin (Supplementary Table S1). Also, there was no correlation between T2* values and those indexes. Besides, none of the patients in this study had decreased SF despite a significant proportion of patients (9/47, 19.1%) were identified to have anemia (10). These facts may reflected a lack of agreement between traditional markers of ID and myocardial tissue iron, as shown in previous publications (26). Previous direct tissue analyses demonstrated that myocardial iron content might be independent of absolute iron regulation (27). In addition, as an acute reactant that rises with the chronic inflammation, ferritin is often elevated in HIV-infected individuals regardless of iron status. Thus, Iron stores likely overestimated in HIV-infected populations (28).

Our findings of cardiac dysfunction in the setting of compromised iron are supported by other studies as well. In an in vitro human cardiomyocytes experiment, when human embryonic stem cell-derived cardiomyocytes were deprived of iron, cardiomyocyte function was reduced as characterized by impaired mitochondrial respiration and reduced contractility and relaxation (29). In other studies, on the basis of T2* CMR (16, 17, 30–32), elevated myocardial T2* values (hence reduced iron content) has been shown to occur in non-ischemic heart failure (16, 32), hypertrophic cardiomyopathy (31), and hemodialysis-dependent end-stage renal disease (ESRD) (17, 30). Increased T2* has also been shown to be a an independent risk factor for LV dysfunction in ESRD patients (17) and a predictor of adverse cardiac events in non-ischemic heart failure (16).

Although the range of T2* values in the healthy controls of our study was lower than that reported by Meloniet al. (33), our findings are close to Xu a (17). Besides, the T2* values in HIV-1+/LVEF− were similar to those with non-ischemic cardiomyopathy (16). Even the extent of myocardial iron deficiency we observed in HIV-1+/LVEF+ did not reach the levels reported in patients with ESRD (17, 30) as it was considerably higher than the HIV-1+/LVEF− patients and normal control subjects of our study. Due to a large fraction of asymptomatic patients (63.8%, 30 of 47) enrolled in our study, the iron metabolism disorder might have been relatively mild compared with the patients in end stage of renal disease. Additional studies with larger sample sizes are needed to explore whether the myocardial iron deficiency in patients the late stages of HIV-1 disease also experience more severe impairment in LV function.

Furthermore, we found that CMR markers of myocardial inflammation were elevated in HIV-1-infected patients, indicating subclinical myocardial inflammation. Previous CMR studies revealed that HIV-1-related cardiovascular disease is associated with chronic inflammation, frequent pericardial effusion, and probable myocardial edema (33–35). Increases in T2 and native T1 relaxation times, as a quantitative CMR measure of water content, although nonspecific, can be seen in myocardial infarction, myocarditis, amyloidosis, and focal and diffuse fibrosis (36–38). It is worth noting that we observed higher ECV in HIV-1+/LVEF+ than HIV-1+/LVEF− and normal controls, which has also been designated a risk factor for mild LV dysfunction. Increased ECV values were found to be associated with the severity of diffuse myocardial fibrosis as identified on histopathology. ECV values were also a major determinant of altered diastolic filling and LV systolic pumping function (39). Studies have also shown a correlation between ECV and LV function in patients with end-stage HF secondary to mitral regurgitation (MR) (40). We note that the range of ECV values in the healthy controls of our study were higher than previously reported by Julian et al. (33), they were close to a Chinese cohort in a study by Lin et al. (41).

It is possible that T2* values may also be influenced by edema, but it is far less insensitive compared with T2-based methods (15, 42). On 1.5 T MRI, in patients with acute non-hemorrhagic myocardial infarction experienced an increase in T2* of 6%, but 78% increase in T2-weighted signal intensity with respect to remote myocardium (42). Similar findings have been found in large animals' models as well with more than a 3-fold greater difference in the sensitivity to edema in T2-based vs. T2*-based approaches (15). These observations along with our observations in T2 and T2* in HIV-1 patients here suggest that the increase in T2* in HIV-1 patients we observed is likely a consequence of ID than from edema. Besides, our previous study (21) had been reported a higher rate of subclinical myocardial edema (represented by T2 values) and fibrosis (represented by LGE) in HIV-infected patients, but we further found that it was fibrosis but not edema that changed with the severity of the disease. In this study, we also found that HIV patients had higher T2 values than healthy controls, meanwhile, there was no differences in T2 values between patients with normal LVEF and patients with abnormal LVEF. Contrarily, T2* values showed a significantly negative ecoefficiency with LVEF in HIV-infected patients. Thus, we suspected that with the extensively use of antiretroviral therapy, it was chronic inflammation rather than acute inflammation that existed in HIV patients, and edema alone cannot explain the increasing in T2* values between patients with and without normal LVEF.

There are several limitations in this study. First, myocardial T2* measurement in clinical practice is typically performed at 1.5 T, and our studies are performed at 3.0 T, where B0 and B1inhomogeneities are greater (43). Despite this we found our data to be reproducible. Further, the use of 3.0 T for CMR exams is growing rapidly given the well-known SNR benefits that can be exchanged for imaging speeds and higher-spatial resolution. Next, none of the HIV-1 patients in our study had severe LV dysfunction. Thus, additional studies in symptomatic patients are needed to evaluate whether iron deficiency and greater impairment in LV dysfunction are linked. On the other hand, iron overload has been shown to be associated with increased susceptibility to certain infections (44). Therefore, prophylactic iron supplementation without repeat assessment for myocardial iron can facilitate the development of opportunistic infections. Hence, additional studies are required to establish the benefits of using T2* CMR for guiding appropriate iron treatment strategies. Further, our findings are determined from a cross-sectional and exploratory study with a scant number of patients, so it was impossible to assign causal links between myocardial iron deficiency and altered LV function seen in the HIV-1 group. Longitudinal studies with a larger cohort are warranted to confirm our findings. ART is a potential confounder in the present study. However, with a push toward early HIV-1 treatment, it is unlikely that a large cohort of ART-naive patients would be available for comparative studies to help determine the influence of ART on myocardial iron deficiency. Furthermore, given the diverse combinations of antiretroviral drugs used on patients in the present study, we could not identify drug- or drug class-specific effects. This would necessitate a suitably powered prospective study.

In the present study, we demonstrated a previously underappreciated high burden of iron deficiency in HIV-1 subjects using T2* CMR. We found that myocardial iron deficiency is an independent risk factor and predictor of LV dysfunction. Further studies are required to determine whether changes in iron metabolism plays a causal role in the increased cardiac morbidity and mortality of HIV-1-infected patients and whether iron supplementation could improve the management of HIV patients.