Here we have reported cMRI-derived measurements of the effects of trientine treatment on cardiac function in rats with established heart failure caused by diabetes. Effects of diabetes on cardiac function were characterized, inter alia, by substantive impairment of functional indexes including LVEF, CO/LVM (a measure of myocardial efficiency [30, 33]), LVEDV/LVM, LVESV/LVM, stroke volume/LVM and elevated LVM/BW (the latter a measure of relative LV hypertrophy). These findings are consistent with prior reports and provide substantive validation of the model used in the current studies .
Eight weeks trientine treatment significantly improved cardiac function without modifying hyperglycemia, thus effectively breaking the link between defective glucose homeostasis and organ damage, and implicating defective copper regulation in the mechanism by which diabetes causes heart failure. Cardiac function improved in every drug-treated rat, consistent with marked trientine-evoked improvement in cardiac performance in rats with established heart failure caused by diabetic cardiomyopathy. Furthermore, the beneficial effect of trientine treatment on CO in these studies was consistent between the cross-sectional and time-dependent in vivo cMRI analyses, and the subsequent cross-sectional ex vivo endpoint measurements in the same individuals.
Thus, trientine treatment substantively improved cardiac function in this widely-used model of diabetes-evoked heart failure. Previous studies have shown that trientine treatment can improve LV hypertrophy, ex vivo indexes of diastolic and systolic function, and numerous cardiac biomarkers in animals with diabetes-induced heart failure [6, 34–36]. To date, however, there has been no direct in vivo evidence that trientine can significantly improve cardiac function in animals with established heart failure. The current results are significant because they provide robust support for ongoing development of trientine treatment for the experimental treatment of diabetic patients with heart failure, and validate the use of cMRI to measure the responses of trientine treatment in diabetic patients with impaired CO and LVEF.
Different values for LVEF have been reported in the literature by different groups for normal rats or those with heart failure [33, 37–39]. Between-group differences in reported values may be related to differences in rat strains (for example Sprague–Dawley vs. Wistar), methodological differences, or differences in the severity of diabetes or the time that measurements were made after the onset of disease. The values obtained in our study (Tables 1, 2 and 3) are consistent with data reported by others in the literature, as measured by cMRI , PET scanning  or echocardiography . For example, LVEF values comparable to ours in normal adult rats have been reported that were determined by cMRI (79 ± 4%) , cardiac PET scanning (83.2 ± 8.0%)  and echocardiography (81.6 ± 6.0%) . The latter group also reported values of LVEF in heart failure, of 54.6 ± 15.9% by PET scanning and 54.2 ± 13.3% by echocardiography that are also comparable to our measurements (Tables 2, 3, 4). By contrast, others have reported moderately lower estimates of LVEF in control (~64-67%) and diabetic (~42-45%) rats [33, 39]. Thus there is considerable variability in published LVEF measurements in both normal rats and those with heart failure, and our values lie in the mid-range of reported values. In studies such as ours of the reversal of disease effects, it is probably more important that measurements in control, diabetic and drug-treated diabetic animals were made by using a consistent analytical approach, as was done here. Furthermore, our experiment employed significantly larger numbers of animals per group than were used by some of the others  so the power of our study was considerably greater.
Treatment with the long-acting calcium channel blocker, azelnidipine in STZ-diabetic rats has also been reported to cause significant improvements in myocyte contractile function, oxidative stress and myocardial apoptosis, which were attributed to improved myocellular calcium homeostasis . However, there is compelling evidence that impaired myocellular calcium handling does not explain the impaired contractility of diabetic cardiomyopathy in STZ-diabetic rats, which is actually attributable to LV remodeling and consequential impairments in calcium responsiveness . Therefore, the actual mechanism of the reported effects of azelnidipine  is by no means clear.
The current results extend our previous work in nonclinical models of diabetic organ damage [5, 7, 8, 36, 42, 43] and in clinical trials [10, 13, 44, 45], where we have shown that trientine is safe in both nondiabetic volunteers and diabetic patients, and that it significantly improves antioxidant defences and improves LVM in diabetic patients with LVH [6, 12]. Consistent with our prior results, trientine treatment also improved diabetes-mediated cardiac damage in Zucker diabetic rats, an animal model of T2DM, whose metabolic disturbances were less severe than those in the animals we studied here . Taken together with our nonclinical and clinical results, these latter independent findings are consistent with our current conclusions and provide further support for the idea that trientine treatment can improve cardiac function in diabetic individuals with different degrees of metabolic perturbation.
The current study complements and extends our previous studies in patients with T2DM and LV hypertrophy, in whom 12-months trientine treatment significantly improved LV mass indexed to body-surface area (LVMbsa) without adverse remodeling consequences . In that study, patients had LVH and abnormal diastolic filling as demonstrated by mitral inflow Doppler with preload reduction and LV ejection fraction ≥ 45% by echocardiography, with evidence of diastolic dysfunction, and it remains to be determined whether trientine treatment can improve indexes of cardiac function in patients with established heart failure.
Previous mechanistic studies have indicated that trientine decreases LVH by a combination of beneficial effects through which it restores the integrity, inter alia, of ventricular composition, cardiac myocellular actomyosin filaments, mitochondria, and the ECM [5, 7, 43]. At the molecular signaling level, these responses are mediated, at least in part, by normalization of the myocardial and arterial TGF-β/SMAD signaling pathways  with concurrent bolstering of antioxidant defense mechanisms . These beneficial effects on tissue and organ structure/function are probably mediated through trientine-evoked normalization of copper regulation [6–8] with resulting lowering of oxidative stress, and accompanying improvements in ROS metabolism and mitochondrial function [36, 43].
Of potential relevance to the current findings are results from treatment with another copper chelator, tetrathiomolybdate (TTM) that can inhibit acute inflammatory responses in vivo . In a recent study, TTM reportedly inhibited vascular inflammation and atherosclerotic lesion development in apolipoprotein-E deficient mice , raising the possibility that similar effects could also contribute to the cardiovascular responses to trientine. However, in this study, 10-weeks treatment with TTM induced significant lowering of circulating ceruloplasmin levels, raising the possibility that the reported findings could have been confounded by concomitant copper deficiency, as signaled by marked lowering of serum ceruloplasmin levels. By contrast, under the conditions used in this study, trientine does not cause copper deficiency [5, 12].
What information is currently available about the dosage and tolerability of trientine in diabetes? In previous nonclinical studies [5, 7, 34, 35], trientine dosages have been equivalent on a mg-per-kg basis in rats to those employed in the current study and in our recently reported 12-month clinical trial . Trientine has been well tolerated by diabetic and non-diabetic rats alike in previous studies, in which it has typically been administered orally in the drinking water for periods of between 6 and 8 weeks. It has also been well-tolerated by diabetic patients in recent Phase 2 studies in one of which it was administered at 1200 mg/d for 12 months [6, 12]. Furthermore, the trientine-treated patients in these studies did not develop copper deficiency, as shown by levels of informative biomarkers [10, 12], namely serum copper and ceruloplasmin . Equivalent and higher doses were also well tolerated by the non-diabetic volunteers in our two recent Phase 1 pharmacokinetic-pharmacodynamic studies [10, 13]. In addition, comparable trientine dosages have been used successfully for many years in the chronic treatment of patients with Wilson’s disease [17, 18]. An optimal trientine dosage for the treatment of heart failure in diabetic patients is likely to lie in the region of 1200–2400 mg/day (600–1200 mg b.i.d.) [6, 10, 12, 13].
The current study has limitations. Trientine treatment significantly improved cardiac function in diabetic rats with established heart failure. Nevertheless, although substantively improved, their cardiac function was still markedly impaired after eight weeks drug treatment compared with non-diabetic controls. In comparison, T2DM patients with LVH displayed on average a ~50% lowering of elevated LVMbsa values after 12-months trientine treatment . Moreover, although CO was apparently fully restored in this ex vivo study, comparison with the cMRI-based measurements shows that the level of in vivo recovery was less, at about ~30%. This difference may well reflect greater sensitivity of the in vivo cMRI-based measurements. Further studies will be required to determine the maximal restoration of CO, CO/LVM, LVEF and stroke volume/LVM values achievable by longer periods of trientine treatment, and what length of treatment might be required to achieve an optimal response. In addition, the long-term sustainability of trientine-evoked improvements in cardiac function also remains to be established. Finally, it remains to be determined whether improvements in functional parameters might herald lowered rates of clinically-relevant cardiac endpoints and increased survival in patients with diabetes-associated heart failure.
Our measurements of myocardial fat indicate that the model of severe STZ-evoked diabetes we employed here is unlikely to yield useful information about copper/trientine-myocardial lipid interactions. We have employed the STZ-mediated model of diabetes-induced heart failure because it is reproducible and consistent in terms of time of onset, progression, and severity, and its cardiac structure/function responses are clearly consistent with those in diabetic patients [5, 12]. However, the metabolic disturbance in this model is more akin to that in patients with severe T1DM rather than T2DM, where the metabolic disturbance is usually less severe. Trientine treatment did not modify plasma lipid levels in rats with hypercholesterolemia and hypertriglyceridemia caused by diabetes . We conclude that effects of trientine on myocardial metabolism will need to be studied in other nonclinical models of hyperlipidemic diabetes, such as the Zucker diabetic rat, in which myocardial lipid levels are elevated in untreated diabetic animals, in order to better understand its effects on lipid regulation.
Our measurements of simultaneous copper and triethylenetetramine concentrations in diabetic animals and humans undergoing trientine treatment confirm that the molar trientine free-base to Cu(II) ratio is consistently > > 1 in the urine [14, 44]. These findings, taken together with the strong, selective binding of triethylenetetramine to Cu(II) [9, 11], are consistent with protection by the drug against any copper toxicity that could otherwise be caused in the body during trientine-mediated extraction of copper into the urine. Trientine can thus effectively suppress Cu(II)-mediated cytotoxicity. Biological effects that may explain trientine’s therapeutic efficacy include its ability to elicit structural and functional improvements in the cardiac myocytes , the extracellular matrix  and cardiac mitochondria , and to suppress oxidative stress, including that in cardiovascular and renal tissues [8, 13, 36, 49]. Current data strongly support the view that trientine’s efficacy is underpinned in substantive part by its ability to suppress Cu(II)-mediated oxidative stress and mitochondrial dysfunction [36, 43].
Administration of trientine and measurement of the resulting urinary copper excretion unmasks the extent of the copper overload in diabetes [5, 6, 12, 14, 45]. For example, an oral challenge with trientine in diabetic patients revealed a significant increase in chelated Cu(II) . The presence of elevated systemic copper following trientine administration is an important sign because such copper is catalytically active and therefore (presumably) capable of catalyzing the formation of highly-destructive hydroxyl radicals from substrates such as superoxide anion or hydrogen peroxide [9, 22]. Probing of diabetic animals and patients with various applicable types of spectroscopy, including atomic absorption, atomic emission, ICP- MS, EPR and particle-induced X-ray emission (PIXE), has demonstrated the presence of copper excess in the whole body [5, 6, 12], urine, coronary arteries , and kidneys . Thus, defective copper regulation could well make a major contribution to the causation of oxidative stress in diabetes, which can thus be suppressed by treatment using trientine with the outcome that organ damage is thus substantively ameliorated.
How might particular organs be selectively targeted by copper-mediated oxidative stress? Elevated pathogenetic tissue binding of copper occurs in diabetes [5, 6]. This phenomenon could well be caused by copper-catalyzed ‘glycoxidation’ that mediates formation of advanced glycation end-products (AGEs) and consequential AGE-modification of susceptible amino-acid residues, particularly lysine, arginine, histidine and cysteine, in long-lived fibrous proteins such as connective-tissue collagens [5, 50–54]. Such AGE-modified amino-acid residues are thought to act as endogenous chelators [9, 21, 55] that can increase the copper content of organs such as the coronary arteries and kidneys by binding increased amounts of catalytically-active Cu(II), thereby focusing the related oxidative stress into susceptible tissues. The pathogenetic significance of these phenomena was not apparent before we demonstrated suppression of diabetes-mediated organ damage by Cu(II)-selective chelation [5–8, 12, 42, 43].
What biological properties of copper underpin its pathogenetic behaviour in diabetes? Most intracellular copper is tightly bound and regulated by copper-binding proteins , and intracellular free copper is essentially undetectable . Cu(II), which is present in urine from drug-treated diabetic rats, is the most effective divalent ion for binding to organic molecules and the main extracellular copper ion, whereas Cu(I) predominates inside cells . Trientine binds Cu(II) less strongly than most physiological copper-binding proteins . These observations, taken together with our demonstration of the prompt increase in Cu(II) excretion after oral trientine administration in diabetic patients [5, 6], and after injection into the coronary arteries in ex vivo cardiac preparations from diabetic rats , indicate that this increased Cu(II) is unlikely to be released acutely from an intracellular pool. More likely, the Cu(II) is bound to extracellular matrix (ECM) components, such as collagen; because it is readily extracted by trientine, the increased Cu(II) must be loosely bound and is therefore the probable cause of the observed increase in oxidative stress suppressible by trientine . Importantly, consistent with our recent X-ray crystallographic studies, triethylenetetramine-binding to Cu(II) can suppresses its catalytic activity, thus protecting the renal tubules from the toxicity that could otherwise ensue.
There are several strong endogenous copper chelators including spermine, spermidine and carnosine, which are localized mainly within cells. There are no available data known to us concerning their ability to extract systemic copper into the urine in diabetes, or to treat diabetic cardiomyopathy. Spermine was reportedly highly toxic when administered to chicks  and long-term studies of its tolerability on administration to diabetic rats are currently unavailable. Based on their structures and related thermodynamic considerations, putrescine (a diamine) and spermidine (a triamine) cannot bind copper as strongly as triethylenetetramine (a tetraamine). Much research would be required before the putative utility of any of these endogenous polyamines in the treatment of diabetic heart disease could be ascertained.
Some studies with clinically significant drugs used to treat diabetic patients, such as metformin, pioglitazone and some of the ARBs, have suggested that they are chelators that could exert their therapeutic effects through copper chelation [59, 60]. The strength of their copper-binding ability appears to be relatively weak compared with that of triethylenetetramine based on currently available evidence. Moreover, there is no available in vivo evidence to indicate that any of them acts by correcting the dysregulated copper homeostasis that occurs in diabetes .
Mechanisms of action of trientine and metformin are now briefly compared with respect to roles of copper. Metformin is an anti-hyperglycemic biguanide drug that is used extensively in the treatment of type 2 diabetes . It acts via AMP-activated protein kinase (AMPK), whose activation is required for metformin’s inhibitory effect on hepatic glucose production via inhibition of the diabetes-mediated activation of the gluconeogenic genes phosphoenolpyruvate carboxykinase and glucose-6-phosphatase [62–64]. The idea has recently been advanced that metformin might function through binding to copper in locations such as mitochondria . As has been discussed above, there is a substantive body of evidence linking defective copper homeostasis to the pathobiology of diabetes and its experimental treatment with trientine [9, 21] which points to major differences between the mechanisms of action of the two drugs. For example, trientine does not lower blood glucose in diabetic rats or patients, whereas glucose lowering is metformin’s major therapeutic effect [5, 12]. Moreover, in a study comparing patients with type-2 diabetes and LV hypertrophy, who were taking standard therapeutic doses of metformin with or without trientine, LV hypertrophy was unchanged in those taking metformin only, but was substantively improved in the trientine-treated patients . Furthermore, trientine acts in vivo as a chelator that removes excess Cu(II) from the body, whereas patients treated with metformin have rates of urinary copper excretion similar to basal (that is, untreated) values in diabetic patients [6, 12]. Therefore, any role played by copper binding in metformin’s mode of action must be very different from that it plays for trientine.
In conclusion, eight weeks trientine treatment caused significant improvement in numerous indexes of cardiac function as determined by cMRI in a widely employed nonclinical model of diabetes-induced heart failure, the STZ-diabetic rat. These findings support and extend previous studies in diabetic rats, and are consistent with results from a recent Phase 2 trial of trientine treatment in diabetic patients with LVH. The Cu(II)-selective chelator trientine is the first in a new class of orally-active molecules with application in the experimental pharmacotherapy of the diabetic complications. The evidence presented here supports the further clinical investigation of trientine treatment in patients with diabetes and heart failure.
aAbbreviations: AGEs, advanced glycation endproducts; ANOVA, analysis of variance; ARB, angiotensin II receptor blocker; BW, body-weight; cMRI, cardiac magnetic-resonance imaging; CI, 95% confidence interval of the mean; CO, cardiac output; Cu(I), univalent copper; Cu(II), divalent copper; ECM, extracellular matrix; EPR, electron paramagnetic resonance spectroscopy; LME, linear mixed-effects; LVEDV, LV end-diastolic volume; LVEF, left-ventricular ejection fraction; LVESV, LV end-systolic volume; LVM, LV mass; PET, positron-emission tomography; PIXE, particle-induced X-ray emission spectroscopy; RBS, Rutherford backscattering spectroscopy; STZ, streptozotocin; T1DM, type-1 diabetes; T2DM, type-2 diabetes; TE, echo time; TR, repetition time; TTM, tetrathiomolybdate.