Association of peripheral nerve conduction in diabetic neuropathy with subclinical left ventricular systolic dysfunction

Background Subclinical left ventricular (LV) longitudinal myocardial systolic dysfunction occurs in patients with diabetes mellitus (DM) and preserved LV ejection fraction (LVEF), and is closely related to DM-related complications. However, the association of diabetic neuropathy (DN) with subclinical LV systolic longitudinal dysfunction in such patients has not been fully clarified. Methods The subjects of this study were 112 consecutive DM patients with preserved LVEF (all ≥50%) without coronary artery disease and overt heart failure (aged 59 ± 14 years; 60 women, 52 men). Global longitudinal strain (GLS) was determined as the average peak strain of 18 segments from the three standard apical views, and was expressed as an absolute value. DN was diagnosed by experienced diabetologists. Median, ulnar, and sural nerves were subjected to motor and sensory nerve conduction studies. F-wave latency was defined as the minimum F-wave latency after a total of 16 stimulations of the tibial nerve. Results Forty-one (37%) patients were clinically diagnosed with DN. LV functions of DM patients with and without DN were similar except for GLS being significantly smaller in patients with than in patients without DN (18 ± 2% vs. 20 ± 2%, p < 0.001). It was noteworthy that, of the parameters for the nerve conduction study, only F-wave latency correlated with GLS (r = −0.34, p < 0.001), and also was identified as an independent determinative value of GLS in a multivariate linear regression model (β = −0.25, p = 0.001) even after adjustment for other closely related GLS factors. Conclusions Monitoring of F-wave latency may aid early detection of not only DN but also subclinical LV dysfunction. Joint planning of assessment by diabetologists and cardiologists is therefore advisable for better management of DM patients.


Introduction
Subclinical left ventricular (LV) myocardial dysfunction occurs in patients with diabetes mellitus (DM) and preserved LV ejection fraction (LVEF) but without coronary artery disease or hypertension [1][2][3]. This condition is considered a major contributor to the development of a type of heart failure (HF) known as diabetic cardiomyopathy. In addition, LV longitudinal myocardial systolic dysfunction may constitute the first marker of a preclinical form of diabetic cardiomyopathy in DM patients with preserved LVEF without overt HF [4]. In particular, global longitudinal strain (GLS) as assessed by two-dimensional speckle-tracking strain is reportedly effective for the assessment of subclinical LV myocardial dysfunction in DM patients with preserved LVEF [4][5][6][7]. Many DM-related pathophysiological mechanisms lead to harmful changes including free acid metabolism, increased apoptosis, activation of the reninangiotensin system or autonomic neuropathy, as well as an increase in oxidative stress or myocardial steatosis, have been found to play an important role in the development of diabetic cardiomyopathy [1,[8][9][10]. Moreover, it has been reported that DM-related complications or patient characteristics such as diabetic nephropathy, diabetic retinopathy, obesity, insulin resistance, or autonomic neuropathy are associated with subclinical LV myocardial dysfunction, because all of them are considered to be accumulative phenotypes of poor metabolic control [11][12][13][14][15][16][17]. Severe diabetic neuropathy (DN) also reflects poor blood sugar control, which leads to painful nerve damage as a result of increased deposition of sorbitol, fructose, and advanced glycation end products. Uncontrolled hyperglycemia generates microvascular ischemia, which results in peripheral nerve damage due to vasoconstriction and microvasculopathy [18]. Although this mechanism of DN is the same as that of diabetic cardiomyopathy, the relationship of diabetic peripheral neuropathy with impaired LV longitudinal myocardial systolic function has not yet been fully clarified. Accordingly, the purpose of the study presented here was to determine the relationship between DN and subclinical LV longitudinal myocardial systolic dysfunction in DM patients with preserved LVEF. We also assessed the relationship between the parameters of nerve conduction with LV longitudinal myocardial systolic function in such patients.

Methods
A total of 118 consecutive DM patients who were admitted to Kobe University Hospital between July 2013 and January 2015 were recruited, and data sets of nerve conduction and echocardiographic studies were prospectively collected. The diagnosis of DM was based on the World Health Organization criteria [19]. Preliminary exclusion criteria for this study were as follows: (1) ischemic heart disease; (2) LVEF < 50%; (3) a previous history of open-heart surgery and congenital heart disease; (4) serious renal dysfunction defined as glomerular filtration rate (GFR) < 30 mL/min/1.73 m 2 ; (5) uncontrolled hypertension >180/100 mmHg; (6) significant valvular heart disease; (7) atrial fibrillation; (8) left bundle branch block. Each patient underwent exercise stress testing such as treadmill exercise or stress myocardial perfusion scintigraphy within 2 weeks after admission. The study protocol was approved by the ethics committee of our institution and all patients gave informed consent before participating in this study.

Echocardiography
All echocardiographic data were obtained by means of a commercially available echocardiographic system (Vivid E9; GE-Vingmed, Horten, Norway) within 2 weeks after admission. Digital routine grayscale two-dimensional cine loops from three consecutive heartbeats were obtained at end-expiratory apnea from the standard parasternal long-axis and three apical views with mean frame rates of 66 ± 7 frames/sec. Sector width was optimized to allow for complete myocardial visualization with maximization of the frame rate. Digital data were transferred to dedicated software (EchoPAC version 113; GE Vingmed) for subsequent offline analysis. Standard LV measurements were obtained in accordance with the current guidelines of the American Society of Echocardiography/European Association of Cardiovascular Imaging [20]. LV volumes and LVEF were calculated using the modified biplane Simpson's method. Left atrial volume was also calculated with the modified biplane Simpson's method using apical 2-and 4-chamber views at the ventricular end-systole, and then normalized to body surface area. LV stroke volume was determined in terms of the velocity-time integral assessed by means of pulsed-wave Doppler positioned at the LV outflow tract. The early diastolic (E) and atrial wave velocities and the E-wave deceleration time were measured by using the pulsed-wave Doppler recording from the apical fourchamber view. Spectral pulsed-wave Doppler-derived early diastolic velocity (E') was obtained from the septal mitral annulus.

Assessment of GLS
GLS was assessed by means of two-dimensional speckletracking longitudinal strain from the three standard apical views with the aid of a single dedicated software (EchoPAC version 113; GE Vingmed) as previously described in detail. Briefly, a region of interest was traced on the endocardium at end-systole with a point-andclick approach for each of the three apical views. A second larger region of interest was then generated and manually adjusted near the epicardium. Each apical image was divided into six standard segments, followed by the generation of six corresponding time-strain curves ( Figure 1). GLS was determined as the averaged peak strain of 18 segments from the three standard apical views [20], and was expressed as an absolute value. The physicians who evaluated GLS were blinded to patient data including nerve conduction parameters.

Nerve conduction study
Sensory and motor nerve conduction velocities for the median, ulnar and sural nerves were obtained from surface recordings and determined by means of conventional techniques [21]. The F-wave is considered to be an electrophysiological artifact bypassing the spinal cord after antidromic activation of motor neurons following distal electrical stimulation of motor nerve fibers. Fwave latency reflects the combined conduction time which consists of the times for antidromic conduction from the site of the stimulation of the peripheral nerve to the spinal cord, and for orthodromic conduction for reactivation of the motor neurons from the spinal cord to the peripheral motor nerve ( Figure 2). In this study, F-waves were obtained from the tibial nerve with surface electrodes placed on the on lower leg muscle, and Fwave latency was defined as the minimal F-wave latency in a series of 16 tibial nerve stimulations [22,23]. F-wave occurrence was expressed as a percentage of the F-wave appearance rate for 16 stimulations.

Collection of baseline clinical data
HemoglobinA1c, glycoalbumin, estimated glomerular filtration rate and lipid profile were obtained the day after admission. Albuminuria was quantitated after 24-hour urine collection. All subjects were interviewed in detail and underwent a thorough neurologic examination which included sensory tests for light touch, pin-prick, and vibration and tendon reflexes. Experienced diabetologists methodically and comprehensively assessed DN according to the current guidelines which combine a patient's symptoms, neurological findings and results of a nerve conduction study [24]. In addition, diabetic retinopathy was defined as a condition more serious than simple retinopathy or previously treated by laser photocoagulation was based on the evaluation of retinal photographs by experienced ophthalmologists.

Statistical analysis
Continuous variables were expressed as mean values and their standard deviation for normally distributed data and the median and interquartile range for non-normally distributed data. Categorical variables were expressed as frequencies and percentages. The parameters of the two subgroups were compared by using the Student's t-test or the Mann-Whitney U test as appropriate. Proportional differences were evaluated with Fisher's exact test. Relationships between two variables were analyzed by using linear regression and are expressed as Pearson correlation coefficients. A multiple logistic regression analysis was performed to detect the most sensitive marker of nerve conduction for the prediction of DN using stepwise selection with the p levels for entry from the model set at <0.10.
Since many parameters were associated with GLS, baseline clinical data were first divided into the following four categories to clarify the independency: (1)Demographics and comorbidities: age, gender, body mass index, DM duration, DM type, retinopathy, and dyslipidemia (2)Blood exam and urinary test: HbA1c, glycoalbumin, low-density lipoprotein, triglyceride, estimated glomerular filtration rate and log-transformed albuminuria (3)Cardiac signs: systolic blood pressure, heart rate and rate pressure product (4)All parameters in nerve conduction study: motor and sensory nerve conduction velocities obtained for the median, ulner, and sural nerves, F-wave latency and F-wave occurrence Second, we chose determinative values for GLS by means of stepwise multivariate regression analysis for each group, with a cut-off p value of 0.15 selected so as not to miss potentially influential parameters. We then constructed the final multivariate regression model consisting of the variables selected from four models. The intraclass correlation coefficient was used to determine inter-and intra-observer reproducibilities for GLS from 15 randomly selected patients using identical cine-loops. Statistical significance was set at p < 0.05. All analyses were performed with SPSS version 16.0 (SPSS, Inc., Chicago, IL).

Comparison of echocardiographic parameters for patients with and without DN
Results of a comparison of echocardiographic parameters for patients with and without DN are summarized in Table 3. The standard echocardiographic parameters for the two groups were similar except for GLS for patients with DN being significantly smaller than for patients without DN (18 ± 2% vs. 20 ± 2%, p < 0.001).

Relationship between F-wave latency and GLS
We constructed a two-step multivariate analysis for evaluation of independent associations with GLS. Based on the results of the first multivariate linear regression analysis of four pre-defined models, ten individual parameters comprising gender, DM duration, type 2 DM, retinopathy, body mass index, rate pressure product, triglyceride, glycoalbumin, albuminuria and F-wave latency were chosen for the final multivariate linear regression analysis (Table 4). A significant finding was that F-wave latency was one of the independent determinative factors of GLS (β = −0.25, p = 0.001).

Discussion
The study presented here provides the first evidence that LV longitudinal myocardial systolic function in asymptomatic DM patients with preserved LVEF is significantly impaired in DM patients with DN compared to those without DN in spite of similar conventional Values are mean ± SD. DN = diabetic neuropathy; OR = odds ration; 95% CI = 95% confidential interval. Values are mean ± SD for normally distributed data. DN = diabetic neuropathy; LV = left ventricular; E = peak early diastolic mitral flow velocity; A = peak late diastolic mitral flow velocity; E' = Spectral pulsed-wave Doppler-derived early diastolic velocity from the septal mitral annulus. echocardiographic parameters. Furthermore, only Fwave latency, which is considered to be the most sensitive marker of diabetic peripheral neuropathy, was found to be independently associated with subclinical LV longitudinal myocardial systolic function in a nerve conduction study.

Impact of GLS on diabetic neuropathy
Painful symptoms of diabetic peripheral neuropathy are thought to be triggered by increased depositions of sorbitol, fructose, advanced glycation end products, and free oxygen radicals which reflect uncontrolled hyperglycemia and its sustained duration. In addition, hyperglycemia conveys microvascular ischemia to the peripheral nerves via vasoconstriction and thickening of the basement membrane in blood vessels, resulting in painful damage to the peripheral nerves [18]. Since these pathological mechanisms of DN indicate the degree of diabetic seriousness and the exact duration for DM patients exposed to uncontrolled hyperglycemia [25], the presence of DN can reflect the extent to which LV myocardium has been impaired. Nonetheless, it is not easy to evaluate correctly such often subtle LV myocardial degeneration. Marangoni et al. used streptozotocin-induced rat model of diabetes to prove increased LV volume and impaired systolic and diastolic function just between 4 and 12 weeks after DM onset in parallel with development of insensate neuropathy [26]. Sacre et al. provided evidence of the impaired subclinical LV myocardial systolic and diastolic functions in type 2-DM patients with cardiac autonomic neuropathy [27]. In their study population, 16 (14%) subjects had cardiac autonomic neuropathy, and their peak systolic and early diastolic velocities determined with color tissue Doppler were significantly lower than those of patients without cardiac autonomic neuropathy. This finding suggests that LV longitudinal systolic and diastolic functions are impaired despite preserved LVEF in patients with cardiac autonomic neuropathy.

F-wave latency and GLS
Nerve conduction studies are generally sensitive, reliable, noninvasive, and objective means of examining diabetic neuropathy. Notably, minimal F-wave latency is a wellestablished component of nerve conduction studies and is considered to be the most sensitive and reliable of all relevant parameters [28]. In our subjects, a multiple logistic regression model identified F-wave latency as the most powerful predictor for DN. In addition, F-wave latency showed an independent correlation with GLS, suggesting the first evidence verifying the presence of subclinical LV myocardial systolic dysfunction. Previous investigators have demonstrated that activation of dipeptidyl peptidase-4, retinopathy, albuminuria, obesity and hypertriglyceridemia are associated with subclinical LV myocardial dysfunction [11,14,16,29,30]. In fact, almost of these clinical features were evident in the final multiple linear regression model used for the detection of GLS in our population. However, F-wave latency still proved to be a powerful independent determinative factor in addition to albuminuria. Our finding suggests that, among the many potential confounders, subclinical LV myocardial systolic dysfunction is independently associated with DN.

Clinical implications
The LV wall is not homogenous and is composed of three layers of fibers [31]. LV longitudinal myocardial function is governed by the subendocardial/subepicardial myocardial fibers which are aligned longitudinally so that selective impairment of longitudinal myocardial function may be related to an increase in subendocardial/subepicardial stress. Previous investigators have observed early manifestations of LV abnormalities caused by various diseases including DM in LV longitudinal myocardial function in spite of preserved LVEF [32][33][34]. Among the various parameters for the assessment of LV longitudinal myocardial function, GLS is thought to be the most reliable marker and to have good reproducibility [35]. Since speckle-tracking strain analysis requires experience to achieve reproducible results and its effective use requires training, it could be difficult to use routinely for general physicians such as diabetologists or even cardiologists who were unfamiliar with speckle tracking. Because LV longitudinal myocardial dysfunction can eventually lead to impaired global LV performance, and the presence of DM may cause changes in LV performance over time, watchful observation of DM patients with preserved LVEF but with impaired F-wave latency by both diabetologists and cardiologists may well prove to be of vital importance. Joint planning of assessment by diabetologists and cardiologists is therefore advisable for better management of DM patients.

Study limitations
This study covered a relatively small number of patients in a single center study, so that future studies of larger patient populations are needed to verify our findings. Moreover, our study population did not include normal healthy subjects so that the comparison with GLS in DM patients without DN and normal healthy subjects was not part of this study. Finally, 92 patients (82%) underwent treadmill exercise, and the remaining 20 patients (18%) did stress myocardial perfusion scintigraphy to exclude coronary disease in this study. The group who underwent treadmill exercise may potentially include patients with undiagnosed coronary artery disease.