INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

ANGIOGENIC GROWTH FACTORS, such as vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF), have garnered significant attention for their potential therapeutic role in the protection of ischemic myocardium (10, 14, 24, 32, 33). Clinical trials are underway to define the therapeutic utility of vectors designed to increase the production of angiogenic proteins with the intent to improve blood flow to ischemic tissues, including the myocardium. The development of angiogenic interventions is limited by our incomplete understanding of the endogenous production of angiogenic factors. These initial clinical trials may be promising, but the role that VEGF or other angiogenic factors, like hepatocyte growth factor (HGF), exert endogenously in humans is not well established. Consequently, it is difficult to speculate on the effect that overproduction of angiogenic factors will have. This concern was substantiated by the recent finding that exogenous VEGF actually promotes atherosclerosis in animal models (4). Additionally, the optimal manner in which to deliver these agents and the clinical setting in which they would be most beneficial are unknown (30).

HGF is an angiogenic factor that has been investigated extensively and may participate in an endogenous cardioprotective response to coronary ischemia and infarction (22, 27). The concept that exogenous HGF will augment the endogenous protective mechanisms in response to coronary ischemia was recently confirmed in an animal model of ischemia and reperfusion (21). In these studies, endogenous HGF had an antiapoptotic effect and protected against infarct expansion, and recombinant HGF further enhanced this protective effect (21). Additionally, angiogenic activity has been demonstrated after transfection of the human HGF gene in both the noninfarcted and infarcted myocardium (1). These studies suggest that HGF may have a critically important role in cardioprotection from myocardial ischemia and infarction and that therapeutic delivery is feasible.

The development of collaterals in response to severe myocardial ischemia has long been observed in animal models and noted clinically in patients (5, 6, 26). In humans it appears that the major stimulus to develop collaterals is local ischemia and the pressure gradient (34). It is also fairly well established that those patients with the capacity to develop or augment collaterals are protected in some fashion from the deleterious effects of ischemia and infarction, presumably by maintaining sufficient perfusion to preserve viability in compromised myocardium (16). The difficulty, particularly in humans, is defining the cellular signal that leads to the anatomic changes. Hypoxia, in experimental models, is a potent stimulator of angiogenic peptide production and angiogenesis (29), but the role of hypoxia-driven collateral formation in humans has been called into question by virtue of the development of collateral vessels at a distance from the ischemic zone. Additionally, there is marked in vitro heterogeneity in the hypoxic regulation of VEGF production in human monocytes from patients with established collaterals (28).

In humans, there is a transient increase in venous serum levels of VEGF and HGF in response to a recent myocardial infarction (11, 13, 31). VEGF levels rise over several weeks after a myocardial infarction, whereas HGF levels appear to rise within days after the event (35). Although an upregulation of VEGF mRNA in human atherosclerotic plaques has been demonstrated (12), regulation by ischemia or other stimuli in humans in vivo is not well characterized. Furthermore, human data are unavailable to determine whether increased production of endogenous VEGF or HGF is correlated with increased collateralization or improved outcomes.

We sought to define the endogenous production of VEGF and HGF in patients with ischemic heart disease by examining VEGF and HGF levels in central vascular sites from patients with confirmed coronary artery disease (CAD) and ischemia compared with those without ischemia or CAD. Because of the lack of knowledge of endogenous angiogenic peptide response in human coronary ischemia and relative hypoxia, we investigated VEGF and HGF levels in patients undergoing angiography for the evaluation of ischemic heart disease. In an attempt to potentially detect local coronary production, samples were obtained from the coronary sinus (CS) in addition to the pulmonary artery (PA) and central aorta (Ao). These levels were then compared in patients without CAD (group 1) and those with CAD (group 2). Group 2 was then subdivided to include those with evidence of recent ischemia (group 2A) and those without recent ischemia (group 2B).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 74 patients (mean age 58 ± 10 yr; 31 women and 43 men) were enrolled in the study after informed consent was obtained. The study was approved by our institutional review board. The patients were considered for the study if they were referred for catheterization to evaluate for symptomatic ischemic heart disease. In these patients, left heart catheterization was performed in a standard fashion after CS and right heart catheterization were completed. With the use of the right internal jugular approach, a 6- or 7-Fr multipurpose catheter was inserted in the CS confirmed by fluoroscopy and oxygen saturation. (If uncertainty remained, placement was confirmed by a dye injection.) Samples from the PA and the Ao were obtained within 15 min of each other in duplicate, and oxygen saturation was immediately measured in all patients except for two, for whom data are unavailable. Supplemental oxygen was not given to any patients until after the saturations were obtained.

Two experienced angiographers independently performed coronary angiography in a standard fashion, and significant coronary disease was defined as 50% angiographic stenosis. Quantitative angiography was used in borderline lesions. The presence of visibly apparent angiographic collaterals was confirmed in a blinded fashion, defined by the Rentrop criteria (23), and graded as the following: 0, no collaterals seen; 1, some collaterals with incomplete, delayed filling of the occluded artery; 2, well-formed collaterals with delayed filling of occluded vessel; and 3, abundant collaterals that fill the occluded artery at the same rate as the artery being injected. The grading scale was applied to all angiograms by two experienced angiographers blinded to the clinical characteristics and serological testing.

Measurement of VEGF and HGF. Blood samples were collected in heparinized tubes, immediately placed on ice, and centrifuged for 10 min at 2,000 rpm. Plasma was collected in aliquots and frozen at 80°C until further analysis. The samples were cataloged and analyzed in an anonymous fashion. The concentrations of VEGF and HGF were measured in the plasma by sandwich ELISA by using commercially available assays (R&D Systems; Minneapolis, MN) that are sensitive in detecting these cytokines in concentrations <10 pg/ml for VEGF and 40 pg/ml for HGF. These ELISAs specifically recognize the respective human proteins.

Data analysis. The patients were subdivided based on the presence or absence of significant CAD into the normal control (group 1) and those with significant CAD (group 2), and the group with CAD was then further subdivided into those with recent ischemia (group 2A) and those without recent ischemia (group 2B). The patients with recent ischemia, group 2A, were those patients with CAD and angina at rest or minimal activity within the past 72 h requiring admission to the hospital. This group also included two patients with abnormal MB isozyme of creatine kinase (CK-MB) levels and documented regional wall motion abnormalities on ventriculography or echocardiography. The group without recent ischemia, group 2B, consisted of those patients with significant CAD that had a gradual progression of symptoms (but none in the preceding 72 h) and were undergoing an elective catheterization. The important variables generated in each group, including VEGF, HGF, and collateral grade are expressed as means ± SD and compared by ANOVA. A P  0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 74 patients consented for the study protocol, but the CS was not adequately cannulated in 3 patients. Therefore, 71 patients had complete sampling. There were no complications as a result of the study protocol. The demographics of the study patients are summarized in Table 1. No study patients exhibited severe lung disease and all had serum creatinine values <2 mg/dl. No patient in the study protocol had active liver disease nor did any have hepatocellular enzyme values above two times the normal value. There were no significant differences in known cardiovascular risk factors or other clinical characteristics among the groups; the levels of either HGF or VEGF were not significantly affected by risk factors such as smoking, hypercholesterolemia, hypertension, or diabetes (Table 2).

Angiogenic factor levels, vascular location, and CAD. The VEGF and HGF levels from the three separate vascular sites (PA, Ao, and CS) were not significantly different in the study groups (35 ± 18, 32 ± 13, and 33 ± 19 pg/ml for VEGF and 1,495 ± 904, 1,656 ± 1,011, and 1,445 ± 946 pg/ml for HGF, respectively). The VEGF levels in the selected vascular sites were similar in normal patients or those with CAD and the subgroups of those patients with CAD (Table 3). In examining the HGF levels, there were also no detectable differences in comparisons of the normal group to the entire group with CAD, although small differences in these parameters may be obscured by the large patient-to-patient variability. To examine whether recent ischemia had an influence on growth factor levels, comparisons were made between normal patients and the subgroups of patients with CAD (with recent ischemia and without recent ischemia). There was a marginal difference in CS HGF between the group with recent ischemia compared with the normal group (P = 0.05). There were no significant differences exhibited in the group without recent ischemia.

Effect of hypoxia on angiogenic factor production. To investigate the consequences of relative hypoxia in humans on growth factor levels, we subdivided the entire study population based on the mean value of Ao saturation and CS saturation. The entire study group had a mean Ao oxygen saturation of 93 ± 3% and a mean CS oxygen saturation of 31 ± 7%. Those patients with relative hypoxia were defined as having an Ao saturation of 92% or less (mean 89.64 ± 2.81% compared with 95.89 ± 1.65% for the nonhypoxic group). Similarly, those patients with relative increased coronary extraction were defined as having a CS saturation of 30% or less (mean 27.68 ± 7.50% compared with 32.14 ± 6.66% for the nonhypoxic group). Table 4 summarizes these results. There was no apparent difference in either VEGF or HGF levels in this population based on relative hypoxia or increased coronary oxygen extraction.

Effect of established collaterals and left ventricular dysfunction on growth factor levels. When we examined the effect of established collaterals on growth factor levels, CS HGF, but not CS VEGF, was significantly higher in patients with established collaterals compared with the normal group (1,746 ± 1,321 pg/ml, n = 24 vs. 1,232 ± 533 pg/ml, n = 26, for HGF, P < 0.05; Fig. 1). The corresponding data for VEGF is shown in Fig. 2 [38 ± 27 vs. 32 ± 14, P = not significant (NS)]. There was not a statistically significant difference in either CS growth factor levels between patients with CAD but no collaterals and those with established coronary collaterals (1,366 ± 771 vs. 1,746 ± 1,321 pg/ml for HGF and 30 ± 11 vs. 38 ± 27 pg/ml for VEGF, respectively). Interestingly, patients with left ventricular dysfunction, defined as those with an ejection fraction (EF) <40% by echocardiography or ventriculography, had a highly significant elevation of CS HGF (Fig. 3) compared with those with normal or nearly normal left ventricular function (1,795 ± 1,328 pg/ml in those with EF < 40%, n = 21 vs. 1,298 ± 695 pg/ml, n = 50 in those with EF40% for HGF, P < 0.01). A similar effect was not seen with CS VEGF (36 ± 27 pg/ml with EF < 40% vs. 32 ± 13 pg/ml with EF  40% for VEGF, P = NS).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Coronary sinus hepatocyte growth factor (HGF) levels in patients with collaterals (CAD/Coll) compared with normals (NL) and patients with coronary disease without collaterals (CAD/no Coll). All data are means ± SD; n, number of patients. * P < 0.05 vs. normal.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Coronary sinus vascular endothelial growth factor (VEGF) levels in patients CAD/Coll compared with NL and patients with CAD/no Coll. All data are means ± SD; n, number of patients. No significant P values for all comparisons.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Coronary sinus HGF levels in patients with reduced left ventricular function [ejection fraction (EF) < 40%] compared with patients with normal or nearly normal left ventricular function (EF > 40%). All data are means ± SD; n, number of patients. * P = 0.01 vs. EF > 40%.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Investigators have demonstrated that VEGF is upregulated in animal models of ischemia (2, 3); however, clinical correlation in humans is lacking, particularly regarding the endogenous response of VEGF in different cardiac diseases. It is known that venous VEGF levels increase gradually over days after a myocardial infarction (11, 13, 31), but little is known about ischemia without infarction. One report in human ventricular biopsy samples indicated that hypoxia-inducible factor 1 (HIF-1) was an early response to severe ischemia or infarction but that VEGF mRNA expression may respond later (17). Additionally, HIF-1 was expressed in myocardial and endothelial cells, whereas VEGF was only expressed in endothelial cells. This finding may have importance if myocardial cells are the intended targets for angiogenic factors. Despite the encouraging preliminary evidence that growth factor administration with VEGF (10, 24, 33), FGF (14, 32), or even HGF (18, 21) may be beneficial in coronary and peripheral vascular ischemia, there are little data demonstrating how humans respond endogenously. In fact, there are indications from animal studies that VEGF may actually promote atherosclerosis (4) and may be responsible for enhanced neovasularization that can contribute to atherosclerosis and restenotic lesions (15, 25).

The principal findings of this study indicate that there is no dramatically increased production of VEGF that is detected in the circulation in the presence of significant CAD compared with those patients without CAD. There is no indication of increased production of VEGF in patients with significant CAD, and there is no associated increase production in those with risk factors for atherosclerosis; therefore, it is unlikely that an important difference in VEGF levels would be seen in patients with ischemia without infarction even if they were compared with true "normals." The HGF levels in patients with CAD were also not significantly elevated, but when the subgroup of CAD patients that had recent ischemia was examined, there was an increased production of coronary sinus HGF, which was of borderline significance (P = 0.05). The limitation in sample number in these (and other similar) invasive human studies raises the possibility of statistical error; in particular, that minor differences between patient populations were missed in our analyses. Nevertheless, the possibility that HGF may be an important angiogenic factor responding to recent ischemia in humans warrants further study in this setting.

The observation here that HGF was significantly elevated in patients with established collaterals compared with normals is evidence that endogenous growth factors may be upregulated in response to severe ischemia to increase collateralization. It is interesting to note that this observation was not documented with VEGF. It has been established that VEGF is upregulated during recovery from infarction, but it is not known how VEGF production responds to ischemia without infarction. This is obviously a clinically relevant question because administering exogenous VEGF may therefore be unlikely to affect patients with ischemia but no infarction, especially if the cellular machinery is not prepared to respond to VEGF stimulation. Other growth factors, like FGF and HGF, may be responsible for the endogenous protective mechanisms that result in increased collateralization in patients with unstable angina (8) or those with complications of hypertension (19) and diabetes (20). Perhaps the combination of growth factors is necessary to actually promote collateralization as suggested by a recent physiological study of human collateralization (7). The exact mechanisms that stimulate production of growth factors in humans and subsequently promotes coronary collateralization is currently unknown, although it is likely that many mechanisms are involved. Possible variables that may be important include local tissue hypoxia, the presence of local cytokines induced following ischemia, inflammatory cell recruitment, and potentially mechanical factors such as stretch that may differentially regulate angiogenic factors.

We found that patients with left ventricular dysfunction had higher HGF levels than those with normal or nearly normal left ventricular function, which may further substantiate the complexity of a protective endogenous angiogenic factor response system. It is unknown how much an effect deleterious remodeling or other chamber characteristics may have on growth factor stimulation. The fact that patients with systolic dysfunction have higher angiogenic peptide levels may represent an endogenous attempt for cardioprotection. It has been demonstrated that LV dysfunction can improve after administration of angiogenic growth factor in an animal model of heart failure (9), suggesting growth factors may be protective of myocardial function due to angiogenesis. It is possible then that growth factors, in a manner similar to brain natriuretic peptide in decompensated heart failure, could be augmented to provide benefit. Clearly, a more detailed investigation into these questions should provide important insight and perhaps indicate an endogenous protective system that could be augmented for important clinical benefits in patients with LV dysfunction.