Author + information
- Received February 1, 1999
- Revision received January 17, 2000
- Accepted March 6, 2000
- Published online July 1, 2000.
- Satoshi Yasuda, MDa,* (, )
- Yoichi Goto, MDa,
- Takeshi Baba, MDa,
- Toru Satoh, MDa,
- Hitoshi Sumida, MDa,
- Shunichi Miyazaki, MD, FACCa and
- Hiroshi Nonogi, MDa
- ↵*Reprint requests and correspondence: Dr. Satoshi Yasuda, Division of Cardiology, Department of Internal Medicine, National Cardiovascular Center, 5-7-1 Fujishiro-dai, Suita, Osaka 565-8565, Japan
We hypothesized that the hepatocyte growth factor (HGF) may play a cardioprotective role in human myocardial infarction (MI).
The HGF is a novel, multifunctional growth factor implicated in wound healing, angiogenesis and promotion of cell survival. Recent animal studies have demonstrated the existence of an HGF system in the heart, where it is activated in response to myocardial ischemia and reperfusion.
We studied 40 patients with acute myocardial infarction (AMI), who underwent coronary reperfusion therapy upon admission. Approximately four weeks later, left ventricular (LV) catheterization was repeated to determine the LV ejection fraction (EF), end-diastolic volume index (EDVI) and pressure (EDP). The levels of HGF and brain natriuretic peptide (BNP) were measured by collecting blood samples from cardiac veins draining the infarcted region (MI region) and those draining the noninfarcted region (non-MI region). The ratio of the HGF level in the MI region to that in the non-MI region (= MI/non-MI ratio) was calculated in each patient as an index of the MI-related HGF secretion. The MI/non-MI ratio for BNP was also calculated.
The MI/non-MI ratio for HGF correlated inversely with LVEDP (r = −0.644, p < 0.0001) and LVEDVI (r = −0.843, p < 0.0001) and positively with LVEF (r = 0.763, p < 0.0001). These correlations were completely opposite in direction from those for BNP and LVEDP (r = 0.678, p < 0.0001), LVEDVI (r = 0.783, p < 0.0001) and LVEF (r = −0.805, p < 0.0001). These findings indicate that cardiac HGF acts in contrast to BNP, a biochemical marker for the development of LV remodeling.
Enhanced secretion of cardiac HGF from the MI region is associated with an attenuation of ventricular enlargement and an improvement in cardiac function. The HGF system may modulate the process of ventricular remodeling and thus have important clinical implications.
Hepatocyte growth factor (HGF) is a disulfide-linked heterodimeric molecule composed of a 69-kD kringle-containing alpha-chain and a 34-kD beta-chain (1). Hepatocyte growth factor is synthesized in large amounts in the liver, where it has been reported to induce liver regeneration and prevent hepatic destruction resulting in fibrosis/cirrhosis (2). In addition to the liver, HGF and its receptor c-Met have been identified in various tissues (2). There is a cardiac HGF system that is upregulated in an animal model of myocardial infarction (MI) and reperfusion (3). However, the role of cardiac HGF in human MI is not yet fully understood.
After MI, left ventricular (LV) enlargement is a frequent development. This process, termed postinfarction ventricular remodeling, is accompanied by an elevated level of brain natriuretic peptide (BNP) (4) and is associated with heart failure and increased mortality (5). Ventricular remodeling is known to be affected by several factors (6). One major factor is the infarct size, which is limited by opening the infarct-related artery (early reperfusion and its patency) or by the presence of collateral vessels. Another major factor is scar formation in the infarcted region, which is attenuated by angiotensin-converting enzyme inhibitors (ACE-I) (7).
We hypothesized that cardiac HGF plays a beneficial role in the process of ventricular remodeling after MI in humans. The HGF has potent mitogenic activity (angiogenesis and collateral formation) (8–10) and may contribute to the protection from or repair of endothelial dysfunction (11–13), both of which help in maintaining coronary perfusion. The HGF may also modulate the wound-healing process by attenuating tissue fibrosis (14,15).
In this clinical study, we found that an enhanced secretion of cardiac HGF from the infarcted region is associated with an attenuation of ventricular enlargement and an improvement in cardiac function after MI. The HGF system may play a cardioprotective role during ventricular remodeling and thereby have potential therapeutic implications (16).
The study group included 40 patients (all men; age 61 ± 9 [mean ± SD] years) with acute MI (AMI), who were admitted to the National Cardiovascular Center between June 1997 and May 1998. Diagnosis of AMI was based on the presence of any two of the following criteria: typical chest pain >30 min duration, ST segment elevation >0.1 mV in two or more leads of ECG with subsequent evolution of a typical infarct pattern, or an increased serum creatine phosphokinase level. Patients with the following disorders were excluded from the study: prior MI; uncontrolled diabetes mellitus; and liver (elevated activity in aminotransferases), kidney (elevated levels of creatinine or urea), or lung dysfunction (restrictive or obstructive pattern in spirometry). Thirty-five patients underwent percutaneous transluminal coronary angioplasty of the infarct-related artery (with adjunctive stenting in 10 patients), and the remaining 5 patients received intravenous administration of tissue-type plasminogen activator in the acute phase. In all patients, coronary angiography immediately after the treatment proved Thrombolysis in Myocardial Infarction (TIMI) III grade flow in the infarct-related artery. Following the reperfusion therapy, left ventriculography was performed in 30 patients. The baseline clinical characteristics of the patients who underwent reperfusion therapy are summarized in Table 1.
In a separate protocol, we also studied 10 patients (all men; age 62 ± 11 [mean ± SD years]) with AMI who did not undergo reperfusion therapy in the acute period and were referred to us later for the further examination.
The use of ACE-I or beta-adrenergic blockers was determined on the basis of the individual clinical conditions (mainly for the purpose of controlling blood pressure in patients without contraindications).
All patients gave their written informed consent prior to participation in the study. The institutional committee on human research approved the study protocol. The patients in the present study were not consecutive but were not selected with any intention or prejudice.
A 5F multipurpose catheter (model ISH 5.0F-110C; Cathex Co.; Tochigi, Japan) was introduced into the coronary sinus (CS) through the left subclavian vein under fluoroscopic guidance. The catheter was then advanced into the anterior interventricular vein (AIV) using a guidewire (17). The AIV ascends parallel to the left anterior descending artery, predominantly receives the blood from the area supplied by this artery, and becomes the great cardiac vein (18,19). The position of the catheter tip was confirmed by injection of contrast media. The blood samples were collected from the AIV and the CS before intravenous administration of heparin because heparin is known to affect HGF production (20,21). Following collection of further blood samples from the right brachial artery (as peripheral blood) through a 6F sheath, heparin was administered, and coronary angiography and left ventriculography were performed according to the conventional Judkins technique. Left ventricular pressure was measured with a 2F high-fidelity micromanometer catheter (model: SPC-320; Miller Instruments; Houston, Texas) advanced into the left ventricle via the lumen of a 6F pigtail catheter. Restenosis of the treated artery was defined as a >50% narrowing of the artery as determined by coronary angiography.
Analysis of LV volume and pressure
Angiographic evaluation of the LV volume was performed by a cardiologist who was blind to the results of the HGF and BNP assays. Ventricular silhouettes in a 30° right anterior oblique projection were digitized with a ventriculography analysis system (ANCHOR; Siemens-Elema; Solna, Sweden). By the area-length method, the LV end-systolic and end-diastolic volume index (LVEDVI) and ejection fraction (LVEF) were calculated. Changes in LVEDVI (ΔLVEDVI) and LVEF (ΔLVEF) were obtained by subtracting the respective values in the acute phase from the corresponding values in the chronic phase in 30 patients.
The time constant of LV pressure decay (tau) was determined from analysis of the LV micromanometer pressure and defined as the interval from the time of the peak negative first derivative of pressure to the time when LV pressure decreased to 5 mm Hg above the end-diastolic pressure (22).
HGF and BNP assay
Blood samples were centrifuged, and the serum was stored at −80°C until assay. The HGF levels were measured with specific enzyme-linked immunosorbent assay kits (Otsuka Assay Laboratories; Tokushima, Japan). Microtiter plates coated with anti-HGF murine monoclonal antibody were incubated with standard HGF or serum samples before anti-HGF rabbit polyclonal antibody was added. After the addition first of anti-rabbit goat immunoglobulin G-peroxidase conjugate and then o-phenylene diamine, the absorbance was measured at 492 nm using a plate reader (23). The sensitivity of the HGF kit is 0.1 ng/ml. Both the interassay and intra-assay coefficients of variation were <5%. This assay system detects only the bioactive, heterodimeric (mature) form of HGF, proteolytically cleaved from its single-chain precursor (pro-HGF), in the blood samples (24,25). Previous studies have demonstrated a strong (r = 0.986) positive correlation between HGF levels measured by this system and those measured by bioassay (23,24).
A previous animal study has demonstrated upregulation of the HGF system in the infarcted region of the myocardium (3). As a relative index of the infarct-related HGF secretion, we calculated the ratio of the HGF level in the infarcted region to that in a noninfarcted region (= the MI/non-MI ratio) in each patient. In a patient with anterior infarction, the serum level in the CS was considered to be the level in the noninfarct region, whereas, in a patient with inferior or lateral infarction, the serum level in the AIV was considered to be the level in the noninfarct region.
The BNP levels were measured with specific immunoradiometric assay kits (Shionogi Co.; Osaka, Japan), as previously reported (17). The sensitivity of this BNP kit is 2 pg/ml. Brain natriuretic peptide has been considered as a biochemical marker of ventricular remodeling after MI (4,25). The MI/non-MI ratio for BNP was also calculated in a manner similar to that for the HGF.
Comparisons between two groups were made by Student t test. Comparisons among three or more groups were carried out by analysis of variance. When a significant difference among groups was indicated by the initial analysis, individual paired comparisons were made using the Student-Newman-Keuls method. Linear regression analysis was used to assess the relationship between the changes in LV functional parameters and the secretion of HGF or BNP. In all cases, differences were considered as being significant at p < 0.05. Data are presented as mean ± SD.
In 40 patients with reperfusion therapy upon admission, the chronic phase cardiac catheterization was repeated 26 ± 6 days after the infarction. In 10 patients, restenosis was found in the follow-up coronary angiography: 75%, 90%, and 100% stenosis in 4, 3, and 3 patients, respectively. In the remaining 30 patients, the treated sites of the infarct-related artery remained patent. Twenty-two patients (55%) were treated with ACE-I, and seven (17.5%) were treated with beta-blockers after the infarction. Three patients (7.5%) were treated with both ACE-I and beta-blockers.
Cardiac HGF production
Table 2 shows the comparisons of the HGF levels among the samples from the AIV, CS and peripheral artery. In patients with inferior/lateral infarction, the HGF levels were significantly elevated in the CS (infarcted region), but not in the AIV (noninfarcted region), compared with those in the peripheral artery. By contrast, in patients with anterior infarction, the HGF levels in the AIV (infarcted region) were significantly higher than those in the peripheral artery. Thus, increased HGF levels were observed in the cardiac vein draining the infarcted region, suggesting infarct-related enhancement of cardiac HGF production. The HGF levels in the CS in patients with anterior infarction were also elevated because the CS drains blood from both the anterior and inferior wall regions.
Because angiotensin II is known to be an important modulator of the HGF system (12), we investigated whether ACE-I treatment may affect HGF production. Cardiac HGF levels (in the infarcted region) were different between patients treated and those not treated with ACE-I (with ACE-I: 0.27 ± 0.09 vs. without ACE-I: 0.22 ± 0.07 ng/ml, p < 0.05). However, neither beta-blocker therapy (with beta-blockers: 0.25 ± 0.09 vs. without beta-blockers: 0.22 ± 0.05 ng/ml, NS) nor the progression to restenosis (with restenosis: 0.23 ± 0.06 vs. without restenosis: 0.25 ± 0.09, NS) affected the HGF levels.
The differential roles of HGF and BNP on LV function
The MI/non-MI ratio for HGF, an index of the infarct-related HGF secretion, correlated inversely with LVEDVI and positively with LVEF, as shown in Figure 1. Moreover, there were significant inverse correlations between the MI/non-MI ratio for HGF and LVEDP (r = −0.644, p < 0.0001) and tau (r = −0.691, p < 0.0001). Similar results in relation to HGF were found in the subanalysis limited to the anterior infarction group (n = 18: r = 0.775, p < 0.0001 in LVEF, r = −0.842, p < 0.0001 in LVEDVI) and the inferior/lateral infarction group (n = 22: r = 0.755, p < 0.0001 in LVEF, r = −0.856, p < 0.0001 in LVEDVI). These findings indicate that the enhanced secretion of HGF from the infarcted region was related to attenuated progression of LV dysfunction.
In the same patient groups, the correlations of the MI/non-MI ratio with LV functional parameters were also determined for BNP. As shown in Figure 2, the MI/non-MI ratio for BNP correlated positively with LVEDVI and inversely with LVEF. Moreover, there were significant positive correlations between the MI/non-MI ratio for BNP and LVEDP (r = 0.678, p < 0.0001) and tau (r = 0.700, p < 0.0001. These data confirm the previous findings that BNP secretion from the infarcted region was increased in proportion to the severity of LV dysfunction (17).
There also existed an inverse relationship (r = −0.817, p < 0.0001) between the MI/non-MI ratio for HGF and that for BNP. Taken together with the results shown in Figures 1 and 2, this finding suggests that HGF may have actions opposite to, or may even counteract the actions of, BNP in the MI.
The role of HGF in postinfarction LV remodeling
In 30 patients who also underwent left ventriculography upon admission, ΔLVEDVI and ΔLVEF were calculated as indicators of the magnitude of LV remodeling four weeks after the infarction. Figure 3 shows that the MI/non-MI ratio for the HGF correlated inversely with ΔLVEDVI and positively with ΔLVEF. These findings suggest that the enhanced secretion of cardiac HGF from the infarcted region is associated with attenuated ventricular enlargement and improved cardiac function.
HGF production in patients who did not undergo reperfusion therapy
Additionally, we studied 10 patients with AMI (anterior: n = 5, inferior/lateral: n = 5) who did not undergo reperfusion therapy. In these patients, cardiac catheterization (performed 30 ± 5 days after the infarction) revealed a lower LVEF (with reperfusion: 44 ± 9 vs. without reperfusion: 36 ± 7%, p < 0.05) and larger LVEDVI (with reperfusion: 68 ± 14 vs. without reperfusion: 82 ± 22 ml/m2, p < 0.05) compared with the 40 patients who underwent reperfusion therapy. As shown in Figure 4, between the patient groups with and without reperfusion therapy, there were significant differences in the MI/non-MI ratio for the HGF (with reperfusion: 1.04 ± 0.34 vs. without reperfusion: 0.74 ± 0.19, p < 0.05) and BNP (with reperfusion: 2.38 ± 0.99 vs. without reperfusion: 3.26 ± 1.02, p < 0.05).
The major finding of this study is that after MI, an enhanced HGF secretion from the infarct region correlated inversely with the LVEDVI, LVEDP and tau, and positively with LVEF, which were completely opposite to the correlations between cardiac BNP secretion and the aforementioned variables of LV function.
Ventricular remodeling and BNP
Myocardial loss as a consequence of infarction initiates a vicious cycle of contractile dysfunction and progressive LV dilation, referred to as postinfarction ventricular remodeling. Much attention has been focused on the pathogenesis and prevention of LV enlargement, because this is an important determinant of the patient’s prognosis and the incidence of heart failure after MI (5). Recently, BNP has been implicated as a useful biochemical marker of postinfarction LV dysfunction/remodeling (4,25). Brain natriuretic peptide is secreted primarily in response to chronically increased ventricular wall tension or stretch, and thus sensitively reflects the driving force of ventricular remodeling. We found that BNP secretion from the infarcted region was increased in proportion to the severity of LV dysfunction (Fig. 2), which is consistent with the results of previous studies (17).
The potential role of cardiac HGF in MI
In the present study, we focused on the HGF, a novel, multifunctional growth factor (2). The HGF has recently been characterized as being one of the most potent angiogenic mitogens specific for endothelium among various growth factors, including vascular endothelial growth factor (VEGF) (9). Several lines of evidence suggest that the HGF may play a role in recovery from MI. In patients with AMI, circulating HGF levels are markedly elevated within 3 h after infarction to levels comparable with or even higher than those in fulminant hepatic failure (23,26). In the infarct animal model, the HGF level remained increased for >24 h (3). There is a local HGF system (HGF and its receptor c-Met) in the heart (3). The cardiac HGF system is upregulated for >120 h only in the ischemic region (3), where the secretion of HGF is presumed to be enhanced. However, the role of cardiac HGF following AMI in humans, especially in the chronic phase (weeks after infarction), has not been well studied.
We have found in the present study that cardiac HGF secretion remained enhanced up to approximately four weeks after infarction (Table 2). The enhanced secretion of cardiac HGF was associated with attenuated progression of LV dysfunction (Figs. 1 and 3). Hepatocyte growth factor production appeared to act in contrast to BNP (Fig. 2) and to be a marker for improved LV function.
Although, from the current data, a causal relationship between the HGF and LV function cannot be clearly determined, the HGF can potentially modify several important processes contributing to ventricular remodeling after infarction (6). The HGF can induce angiogenesis (8,9) or may restore ischemia-induced vascular dysfunction to maintain endothelium-dependent coronary flow, like VEGF (27–30). These effects may lead to the obvious benefit of salvaging the myocardium and reducing the infarct size by providing a patent infarct-related artery and developing collateral vessels. The HGF also exerts cytoprotective effects against ischemic injury (31–33), contributing to the viability of the myocardium in the acute phase of MI. Moreover, the HGF can modulate wound healing processes and prevent scar formation (14,15).
Therapies such as thrombolysis and angiotensin II inhibitors attenuate the progression of ventricular dysfunction after MI (6), which may be mediated in part through HGF. In a recent preliminary study on cardiomyopathic hamsters, angiotensin II inhibition prevented myocardial injury and fibrosis, accompanied by a significant increase in cardiac HGF levels (34). Interaction between angiotensin II and HGF (9,12,35) may be important in the present study, where we found that ACE-I treatment affected the cardiac HGF levels. Moreover, the comparison data shown in Figure 4 suggest the potentially beneficial role of early reperfusion on HGF production, which may be related to the recent finding that the HGF gene is downregulated by hypoxia (33).
There are several potential limitations of this study. First, we did not determine the absolute HGF release rate by measuring the coronary blood flow. Instead, the MI/non-MI ratio was employed as a relative index of infarct-related HGF production. Second, the HGF levels in the CS might have been underestimated (diluted by the AIV) in patients with inferior/lateral infarction, whereas they might have been overestimated (contaminated by the AIV) in patients with anterior infarction, because venous blood from the AIV joins the CS. However, this is not likely to compromise the present findings greatly, according to the subanalysis limited to each patient group that showed consistent correlation between HGF release and LV function, regardless of the site of infarction. Third, the present study was a relatively short-term study, whereas ventricular remodeling is known to progress over months or years. A longer follow-up would be desirable.
We have found that enhanced HGF secretion is associated with both attenuation of ventricular remodeling and improvement in cardiac function in patients after MI. Taking the recently reported biological actions of HGF into account (2,16), these findings suggest that HGF could prove to be a novel therapeutic target in MI in humans.
☆ This study was done with support from the Uehara Memorial Foundation, the Osaka Heart Club and the Japan Cardiovascular Research Foundation (Dr. Yasuda).
- angiotensin-converting enzyme inhibitors
- anterior interventricular vein
- acute myocardial infarction
- brain natriuretic peptide
- coronary sinus
- end-diastolic pressure
- end-diastolic volume index
- ejection fraction
- hepatocyte growth factor
- left ventricular
- myocardial infarction
- vascular endothelial growth factor
- Received February 1, 1999.
- Revision received January 17, 2000.
- Accepted March 6, 2000.
- American College of Cardiology
- Matsumoto K.,
- Nakamura T.
- Ono K.,
- Matsumori A.,
- Shioi T.,
- Furukawa Y.,
- Sasayama S.
- White H.D.,
- Norris R.M.,
- Brown M.A.,
- Brandt P.W.T.,
- Whitlock R.M.L.,
- Wild C.J.
- Braunwald E.
- Weber K.T.
- Grant D.S.,
- Kleinmann H.K.,
- Goldberg I.D.,
- et al.
- Nakamura Y.,
- Morishita R.,
- Higaki J.,
- et al.
- Belle E.V.,
- Witzenbichler B.,
- Chen D.,
- et al.
- Dunsmore S.E.,
- Rubin J.S.,
- Kovacs S.O.,
- Chedid M.,
- Parks W.C.,
- Welgus H.G.
- Schaper W.,
- Kubin T.
- Sumida H.,
- Yasue H.,
- Yoshimura M.,
- et al.
- Clemente C.D.
- Raff G.L.,
- Glanz S.A.
- Tsutamoto T.,
- Wada A.,
- Maeda K.,
- et al.
- Ku D.D.
- Quillen J.E.,
- Sellke F.W.,
- Brooks L.A.,
- Harrison D.G.
- Bauters C.,
- Asahara T.,
- Zheng L.P.,
- et al.
- Harada K.,
- Friedman M.,
- Lopez J.J.,
- et al.
- Miller S.B.,
- Martin D.R.,
- Kissane J.,
- Hammermann M.R.
- Taniyama Y.,
- Morishita R.,
- Matsumoto K.,
- Ogihara T.