• Blood pressure;
  • Cardiomyopathy;
  • Diastolic function;
  • Obesity;
  • Overweight


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References


Cardiomyopathy of obesity occurs in humans, but the gross and cellular myocardial response to obesity in dogs is not well defined.


To characterize in vivo myocardial morphology and function in normotensive obese dogs, and quantitate collagen, triglyceride and myocyte cross-sectional area (CSA) in postmortem tissues from obese dogs.


Echocardiographic-Doppler measurements of normotensive obese dogs (n = 19) without historical or physical examination evidence of disease, and lean healthy dogs (n = 19) matched for age and ideal weight. Postmortem data were obtained from a separate population of 4 obese and 12 lean dogs without evidence of cardiac disease.


A prospective, observational study of myocardial morphology and function was conducted by echocardiographic-Doppler measurement. Left ventricular (LV) tissue was collected for quantitation of triglyceride, collagen, and myocyte CSA.


Compared with lean control dogs, obese dogs had increased systolic blood pressure (obese 153 ± 19 mm Hg; lean 133 ± 20 mm Hg; P = .003), and increased LV free wall thickness at end-diastole (obese 9.9 ± 1.8 mm, lean 8.7 ± 1.5 mm; P = .03) and end-systole (obese 15.2 ± 2.3 mm, lean 12.9 ± 2.3 mm; P = .004). Isovolumic relaxation time was prolonged in 7/19 (37%) of obese dogs, compared with normal ranges. Myocardial triglyceride and collagen content and myocyte CSA were similar between groups.

Conclusions and Clinical Importance

As in humans, LV hypertrophy and diastolic dysfunction can be an early myocardial change in some obese dogs.


body condition score


cross-sectional area


Colorado State University Veterinary Teaching Hospital


ratio of peak early to peak late left ventricular inflow velocities


fractional shortening




isovolumic relaxation time


interventricular septum thickness at end-diastole


interventricular septum thickness at end-systole


left atrial on aortic ratio


left ventricular free wall thickness at end-diastole


left ventricular free wall


left ventricular free wall thickness at end-systole


left ventricular hypertrophy


left ventricular internal dimension at end-diastole


left ventricular internal dimension at end-systole


left ventricular


systolic blood pressure



Obesity is a 30% gain in body weight over the ideal range[1] attributable to excess adipose accumulation. Obesity is the most common nutritional disease of dogs, affecting 34[2]–40%[3] of canids. Risk factors in this species include age, neutering, gender, breed, diet, owner characteristics, and geographic region.[2, 4-6] Obesity has deleterious effects on the canine musculoskeletal,[7, 8] endocrine[9] and urinary[10, 11] systems. Obese dogs demonstrate increased left atrial pressure with exercise and reduced activity tolerance[12]; yet, little is known about the myocardial effects of obesity in this species.

Human studies suggest that “healthy” obesity (ie, without concomitant hypertension, insulin resistance, or ischemic heart disease) is associated with altered myocardial morphology and function; however, the data are discrepant. Left ventricular hypertrophy (LVH) with[13, 14] and without[13, 15, 16] left ventricular (LV) chamber dilatation has been observed. Systolic function is impaired in some populations[14, 16, 17] but preserved in others[13, 15, 17, 18]; diastolic dysfunction is more consistently observed.[14-18] Preliminary studies suggest that overweight dogs without overt heart disease have systolic and diastolic dysfunction.1

Left ventricular hypertrophy, a commonly identified and early morphologic change in obese humans, is associated with increased risk of heart failure.[19-21] It is currently unknown whether obese dogs develop gross LVH. Furthermore, potential contributors to this morphologic change in humans have yet to be explored in dogs. These include lipid accumulation, excess collagen deposition, and myocyte hypertrophy.[22-25]

Given the contribution of obesity to cardiomyopathy in humans, the increasing prevalence of obese dogs, and evidence that LV dysfunction occurs in this population, characterizing the effect of obesity on the canine heart is critically important. We hypothesized that as in humans, concentric LVH would be present in normotensive obese dogs, and that this gross change would be associated with myocardial lipid accumulation, excess collagen, increased myocyte cross-sectional area (CSA), or a combination thereof. Furthermore, we hypothesized that as in healthy obese humans, diastolic dysfunction would be observed concomitantly with preserved systolic function.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

In Vivo Experiments


Adult obese dogs without a history of heart disease were prospectively recruited over a 3-month period by electronic communications to employees and students at the Colorado State University Veterinary Teaching Hospital (CSUVTH). Presence of obesity was determined by assigning a body condition score (BCS) of 7 or greater on a 9-point BCS scale. Each unit represents approximately 10% deviation from the ideal score of 5; scores of 7, 8, or 9 characterize progressively obese subjects.[26]

Dogs recruited for the in vivo study received a physical examination, and systolic blood pressure (SBP) was noninvasively measured (see “Blood Pressure” below). Dogs with SBP ≥180 mm Hg were considered hypertensive[27] and were excluded from the study. Dogs with a grade I/VI murmur were not excluded unless an associated abnormality was identified. Dogs with a grade II/VI murmur or greater, valve insufficiency determined to be greater than trivial, or suspected subclinical heart disease were excluded from the study. Specifically, dilated cardiomyopathy was minimally defined as increased LV internal dimension at end-systole (LVIDs) with concomitant reduced fractional shortening (FS); reduced FS in the face of normal LVIDs was not a basis for exclusion. Dogs receiving steroids were not included in the study.

Blood Pressure

Measurements of SBP were obtained using a Doppler flow detector2 applied to the right dorsal metatarsal artery, with the dogs in left lateral recumbency. Cuffs were fitted such that the width was approximately 40% of the circumference of the portion of the limb it was to encircle.[28] Three consecutive measurements were obtained by the same 2 observers, and the mean of these measurements was used for statistical analysis.

Echocardiographic-Doppler Measurements

Cardiac morphology and function were characterized using conventional transthoracic 2-dimensional echo-Doppler performed by a single experienced operator. Examinations were performed using a GE Vivid 7 ultrasound machine3 with a 5 MHz multifrequency phased array transducer using harmonics. For each measurement, mean values derived from 3 separate cardiac cycles were included in the dataset. Two-dimensional and M-mode views were obtained using a long axis inflow/outflow right parasternal view; M-mode measurements included LV free wall thickness at end-diastole (LVFWd) and end-systole (LVFWs), interventricular septal thickness at end-diastole (IVSd) and end-systole (IVSs), and LV internal dimension at end-diastole (LVIDd) and LVIDs. An index of LV systolic function, FS was calculated using the following formula: %FS = (LVIDd-LVIDs/LVIDd) X 100. Using a left apical 4-chamber view, pulsed-wave Doppler measurements of peak transmitral flow velocities were recorded with sample volume placement at the mitral leaflet tips. The ratio of peak early to peak late LV inflow velocities (E:A) was determined as a measure of diastolic function; data were excluded from analysis if the E and A waves were merged. A left apical 5-chamber view was obtained to quantitate the isovolumic relaxation time (IVRT) using continuous wave Doppler recordings, with the cursor located between the aortic and mitral valves. The IVRT is the time from aortic valve closure to onset of LV inflow and is used to approximate diastolic function. Left atrial and aortic dimensions were measured using the long axis inflow/outflow right parasternal view. The left atrial on aortic ratio (LA:Ao) and E:IVRT were calculated to estimate LV filling pressures.4,[29, 30] Previously published reference ranges were used to establish normal values4,[31, 32]; LVID was normalized for body surface area and LVFWd, LVFWs, IVSd, and IVSs were normalized for actual body weight. Simultaneous ECG recordings were obtained, and heart rate was determined using the mean R-R interval derived from 3 measurements.

Measurements in lean dogs were obtained from a CSUVTH database of 80 animals of various ages and breeds that received an examination in 2000. Data from 2 additional lean dogs were obtained from an 8-year retrospective search of CSUVTH patient records.

Obese dogs were matched by age group and ideal weight to lean controls, and were also matched to reference ranges based on their ideal weight, estimated by BCS and actual weight. The IVRT, FS, and E:A reference ranges used are not weight-dependent, so obese measurements were compared with established reference values. The E:A and IVRT were not measured in lean dogs, so the mean E:A, IVRT and E:IVRT derived from obese dogs were compared with reference ranges.

Postmortem Experiments


Myocardial tissue was collected from obese and lean dogs over a 2-year period. Dogs were received by the CSUVTH necropsy service after euthanasia for reasons unrelated to the present study, at which time the principle investigator was informed of a potential subject. Only subjects without evidence of ante- or postmortem cardiovascular disease were considered for inclusion. Tissues were processed within 3 hours of euthanasia. The presence of gross myocardial lesions was an exclusion criterion.

Tissue Processing

Using established landmarks, similar sections of the LV free wall (LVFW) were divided for formalin fixation and snap freezing in liquid nitrogen. Tissues designated for oil red O staining were embedded in optimal cutting temperature compound5 before freezing. Myocardial triglyceride (TG) content was quantitated in frozen LVFW tissue by colorimetry.6 Oil red O staining was applied to LVFW sections, and slides were examined in a random and blinded approach by 2 observers using light microscopy. Staining patterns were graded according to the following scale: 1 = intracellular staining of myocytes in perivascular space; 2 = light intracellular staining of myocytes outside of perivascular space; 3 = dark intracellular staining of myocytes outside of perivascular space. The total number of events were counted and expressed as a percentage of total number of fields observed, and graded as follows: <5% = 1; 5% to <10% = 2; 10% to <50% = 3; >50% = 4.

Hydroxyproline (HP), a principal amino acid in collagen, was quantitated in frozen LVFW tissue spectrophotometrically using described methods.[33] Masson's trichrome stain was applied to paraffin-embedded LVFW sections for detection of collagen, and quantitated using Image J7 software. Four images from each dog were examined; final values represent the mean area percent of positive staining.

Using hematoxylin and eosin-stained tissue sections, the CSA of 100 transversely sectioned cardiac myocytes with central nuclei was measured in LVFW tissue from each dog using Image J8 software. Sections were also examined for the presence of inflammatory or neoplastic infiltrates by 2 observers blinded to experimental group.

Statistical Analysis

Statistical analyses were conducted using JMP Pro 9.0.28 and Prism 5.0 for Macintosh9. Data are expressed as mean ± SD; statistical significance was set at P < 0.05. Homogeneity of variance was assessed by inspecting plots of residuals versus predicted values. Normality of the residuals was evaluated using the Shapiro–Wilk test. Except as noted, the unpaired t-test was used to compare means from lean and obese groups with normally distributed data (LVFWd, LVFWs, IVSd, IVSs, LVIDs). The Mann–Whitney nonparametric test was used to analyze nonnormal data (IVDd, FS, LA:Ao). The IVRT and E:IVRT data were analyzed using descriptive statistics; E:A data were analyzed using the 1-sample t-test.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

In Vivo Experiments


Nineteen of the 28 obese dogs recruited for the in vivo study met the selection criteria. Seven dogs were excluded for the following reasons: use of steroids (1), hypertension (1), valvular disease (3), dilated cardiomyopathy (1), and scheduling conflicts (1). Two additional dogs were excluded when age-matched controls could not be identified. Table 1 indicates signalment, body weight, and BCS of obese and lean subjects meeting all selection criteria.

Table 1. Signalment, body weight, and BCS of obese and lean subjects examined as part of in vivo studies
Obese Subjects
 Age (years)Weight (lb)BCSIdeal Weight (lb)BreedSex
Range3–1210.6–141.77–98.2–101.213 purebreed (8 breeds), 6 mixed breed10 FS; 1 FI; 8 MC; 0 MI
Mean ± SD7 ± 272 ± 368 ± 155 ± 26  
Lean Subjects
 Age (years)Weight (lb)BCSBreedSex
  1. BCS, body condition score; FS, female spayed; FI, female intact; MC, male castrated; MI, male intact; SD, standard deviation.

Range1–108.2–102.0512 purebred (9 breeds); 7 mixed8 FS; 1 FI; 10 MC; 0 MI
Mean ± SD6 ± 253 ± 265 ± 0  
Systolic Blood Pressure and Heart Rate

Mean SBP was greater in obese subjects compared with lean subjects (Table 2). Heart rate data were not available in 5 lean dogs; values were not different between lean and obese groups (Table 2).

Table 2. Systolic blood pressure and heart rate in lean and obese dogs
 MinimumMaximumMean ± SD P
  1. SD, standard deviation.

  2. a

    Heart rate data unavailable in 5 lean subjects.

Systolic blood pressure
Lean105170133 ± 20.003
Obese100173153 ± 19
Heart rate
Leana7712499 ± 15.069
Obese72167114 ± 30
Echocardiographic-Doppler Examination

The LVFWd and LVFWs were greater in obese dogs compared with lean dogs using both 19 age-matched controls (Table 3) and reference ranges based on actual weight (Table 4). Lean and obese dogs had similar FS. Prolonged IVRT was measured in 37% (7/19) of obese dogs, compared with reference ranges. There was no difference in the E:A of obese dogs compared with a theoretical mean of 1. The E:IVRT was increased in 47% (9/19) of obese dogs compared with normal values. The LA:Ao was increased in lean compared with obese dogs, with larger aortic diameters measured in obese dogs.

Table 3. Echocardiographic findings of 19 obese and 19 lean age-matched controls
Mean ± SDMinMaxMean ± SDMinMax P
  1. Ao, aortic; FS, fractional shortening; IVSd/ IVSs, interventricular septum thickness at end-diastole/end-systole; LA:Ao, left atrial on aortic ratio; LVFWd/LVFWs, left ventricular free wall thickness at end-diastole/end-systole; LVIDd/LVIDs, left ventricular internal dimension at end-diastole/end-systole; SD, standard deviation.

LVFWd (mm)8.7 ± 1.55.711.39.9 ± 1.86.613.1.03
LVFWs (mm)12.9 ± 2.38.817.415.2 ± 2.311.118.9.004
IVSd (mm)11.2 ± 2.36.816.510.3 ± 1.47.512.7.17
IVSs (mm)14.7 ± 2.710.221.714.6 ± 2.011.418.4.82
LVIDd (mm)33.4 ± 6.215.540.735.2 ± 7.420.743.9.17
LVIDs (mm)22.3 ± 5.76.531.522.2 ±
FS (%)35.8 ± 6.529.558.038.2 ± 9.225.455.7.34
LA:Ao1.32 ± ± 0.200.541.40.003
Ao diameter (mm)19.6 ± ±
Table 4. Echocardiographic findings of 19 obese dogs and reference ranges for estimated ideal weight
 Reference rangesObese 
Mean ± SDMinMaxMean ± SDMinMax P
  1. E:A, ratio of peak early to peak late left ventricular inflow velocities; FS, fractional shortening; IVRT, isovolumic relaxation time; IVSd/IVSs, interventricular septum thickness at end-diastole/end-systole; LVFWd/LVFWs, left ventricular free wall thickness at end-diastole/end-systole; LVIDd/LVIDs, left ventricular internal dimension at end-diastole/end-systole; SD, standard deviation.

LVFWd (mm)8.0 ± 1.35.310.29.9 ± 1.86.613.1.001
LVFWs (mm)12.8 ± 2.18.616.115.2 ± 2.311.118.9.002
IVSd (mm)9.9 ± 1.66.612.510.3 ± 1.47.512.7.49
IVSs (mm)14.8 ± 2.59.718.914.6 ± 2.011.418.4.71
LVIDd (mm)34.3 ± 6.418.041.735.2 ± 7.420.743.9.33
LVIDs (mm)21.5 ± 4.410.226.622.2 ±
FS (%)39.5 38.1 ± 9.324.255.7.34
IVRT31–7368.5 ± 14.45099 
E:A11.08 ±
E:IVRT1.45 ± 0.391.1 ±

Postmortem Experiments


Myocardial tissue was collected from 4 obese and 12 control subjects representing 10 breeds and ranging in age from 6 months to 15 years. Reasons for euthanasia included cancer (5), advanced age (1), and poor adoptability per the local humane society (8).

Myocardial Outcomes

No gross myocardial lesions were observed in candidate subjects. Data relevant to myocardial lipid and collagen, as well as myocyte CSA, are summarized in Table 5. Myocardial TG concentrations were similar in lean and obese dogs. The majority of lean (8/12) and obese (3/4) subjects had no visible oil red O staining or exhibited primarily perivascular staining of myocytes that constituted <10% of total number of fields observed (Fig 1A). Positive staining was observed in >50% of fields in sections from 1 lean dog and 1 obese dog (Figs 1B and C). This obese subject, with a predominant grade 3 staining pattern, had the highest myocardial TG content (195 nmol/mg wet weight). In general, however, there was no correlation between TG content and oil red O staining score (r = 0.43; P = .09). Variable numbers of intercellular adipocytes were present in all sections, predominantly perivascularly.

Table 5. Postmortem quantification of myocardial lipid, hydroxyproline, % collagen, and myocyte cross-sectional area in obese and lean dogs
Mean ± SDMinMaxMean ± SDMinMax P
  1. CSA, cross-sectional area; HP, hydroxyproline; SD, standard deviation; TG, triglyceride.

TG (nmol/mg wet weight)48.4 ± 43.98.2165.282.0 ± 78.915.1195.2.30
HP (μg/mg wet wt)0.84 ± 0.6601.961.51 ± 1.710.334.02.51
Collagen (%)0.077 ± 0.0140.0500.1000.071 ± 0.0220.0430.095.67
CSA (μm2)420 ± 139163700864 ± 6033151700.24

Figure 1. Sections of left ventricle stained with oil red O. Panel A: light perivascular staining within myocytes in an obese subject (arrowheads), representing staining pattern “1”. Arrows denote examples of artifactual trapping of stain that occurred during processing; such signals were not included in designation of staining patterns. Panel B: LV tissue from a lean subject displays overall light intracellular staining of myocytes (pink coloration; staining pattern “2”) with focal areas of dark intracellular staining (arrowheads) representing staining pattern “3”. Panel C: the same lean subject represented in Panel B displays heavy intracellular staining of myocytes (arrowheads), characterized as staining pattern “3”. Arrows denote artifactual trapping of stain. 40X oil; scale bar = 20 μm.

Download figure to PowerPoint

There were no differences in HP or collagen content, or in CSA, between lean and obese groups. No inflammatory or neoplastic infiltrates were observed.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Obese dogs examined in vivo in this study exhibited gross hypertrophy of the LVFW without concurrent septal thickening. The LVID was unchanged; thus, this remodeling could be described as asymmetric concentric hypertrophy without chamber dilatation. This morphologic change is similar to that which occurs in some obese humans without conventional comorbidities.[13, 15, 16] In humans, a “typical” obesity cardiomyopathy in normotensive individuals has yet to be identified; it is likely that factors such as severity[13] and duration of obesity, as well as diet,[34-36] contribute to the array of phenotypy.

Although obese dogs had higher SBP than lean animals, the mean SBP of this group fell within normal canine ranges.[37, 38] The American College of Veterinary Internal Medicine, however, established ranges of SBP values associated with risk for target organ damage.[39] Interpreted in light of these parameters, the mean SBP of 153 mm Hg in obese dogs places the obese group at “mild” risk. Applying this classification to individual animals, fewer dogs in the obese group fell within the “minimal” risk category (lean 14/19; obese 6/19), while a greater number fell within the “moderate” category (lean 2/19; obese 9/19). It is evident that myocardial outcomes associated with even mild-to-moderate increases in blood pressure warrant further study.

In addition to hypertension, conditions associated with altered myocardial morphology in obese humans have been identified in fat-fed or obese dogs, including insulin resistance and dyslipidemia[40, 41] as well as aberrant adipokine expression.[42] Furthermore, exercise[43] and hereditary conditions[44] may alter cardiac morphology in the absence of obesity. Importantly, diet composition, highly variable in canids, also modulates the hypertrophic response.[34-36] Although care was taken to select healthy subjects, future studies of nonpredisposed individuals should incorporate serum metabolic indices to elucidate the contribution of the aforementioned factors. Activity and diet warrant serious study as modifiable regulators of cardiac morphology.

In humans, diastolic dysfunction often accompanies LVH.[14-18, 45] Impaired relaxation, estimated by IVRT in this study, existed in 37% of obese dogs. The E:IVRT, an indicator of LV filling pressure, was increased in 47% of the obese dogs in the present study; with one exception, these subjects were distinct from those with prolonged IVRT. These data suggest that there may be 2 types of diastolic dysfunction present in obese dogs: impaired relaxation and increased LV stiffness. The LA:Ao, a less sensitive indicator of LV filling pressure, was increased in lean dogs, a finding inconsistent with E:IVRT data. It is possible that inclusion of perivascular adipose in aortic measurements may have contributed to this outcome, given that aortic diameters were greater in obese dogs.

In contrast to prolonged IVRT, the E:A was unchanged in lean versus obese dogs. A similar discrepancy between these 2 measurements has been documented by others.[15] The IVRT may be less variable than E:A, especially when hemodynamics are normal.[46] Although IVRT and E:A are commonly employed indices, these parameters can be influenced by age, heart rate, and body weight,[32, 47] as well as loading conditions.[48] Tissue Doppler imaging is a less load-dependent and more sensitive measure of diastolic dysfunction in humans,[49, 50] although attempts to define normal parameters in a diverse canine population revealed an effect of age, breed, weight, and heart rate.[51]

The array of canine sizes and breeds poses a unique challenge in interpretation of echocardiographic findings in this species. Transforming the M-mode data to normalize for body weight[52] led to similar results; that is, lean and obese mean LVFW measurements were statistically different, but within reference ranges (data not shown). Lacking a canine-specific standardized approach to comparing lean and obese individuals, we adopted an experimental design commonly employed in human and rodent studies.

No clear contributors to gross hypertrophy emerged from the postmortem studies. Because postmortem samples represented only a select region of the myocardium, it is possible that lesions were underrepresented in the locations chosen for study. The need for representative sampling was particularly evident in relation to myocardial lipid. Intercellular adipocytes were variably present, and probably contribute to the lack of correlation between observed intracellular staining gradation and measured TG content. Additionally, in vivo and postmortem studies were conducted using subjects from distinct populations, and it was not confirmed that postmortem subjects had gross LVH.

Myocardial TG accumulation has not been reported in obese dogs. In obese humans, studies of TG accumulation in individuals reveal conflicting results, ranging from increased[53] to unchanged[23] TG content and combinations thereof.[24] Regarding CSA, dogs with induced heart failure have increased CSA of LV myocytes that is reduced with treatment.[54, 55] Human studies reveal a clear correlation between obesity and myocyte hypertrophy.[25]

Given the risk factors associated with LVH in humans, the finding of LVH in normotensive obese dogs may have important implications regarding health and longevity in this population. Longitudinal studies are needed to investigate the development of cardiomyopathy over time and better understand the possible contribution of metabolic comorbidities. Investigation of the role of exercise and dietary composition will provide insights into modifiable factors and prevention.


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

This study was supported by a Colorado State University Professional Veterinary Medicine Grant. The authors acknowledge Dr Carla Lacerda, Dr Kathy Cooney, and Mr Dennis Madden for identification of subjects for postmortem studies, Ms Carol Moeller for quantitation of myocardial lipid staining, and Mr Jay Oakes for assistance with figure production.

Conflict of Interest: Authors disclose no conflict of interest.

  1. 1

    Schwartz DS, de Oliveira VMC, Melo PRR, et al. Pulsed tissue Doppler identifies preclinical left ventricular myocardial dysfunction in obese dogs. Proceedings of the 27th American College of Veterinary Internal Medicine Annual Conference; 2009 June 3–6; Montreal.

  2. 2

    Parks Medical Electronics Inc, Aloha, OR

  3. 3

    GE Healthcare, UK

  4. 4

    Schober KE. Echocardiographic estimation of left ventricular filling pressure: fact or fiction? Proceedings of the 26th American College of Veterinary Internal Medicine Annual Conference; 2008 June 4–7; San Antonio, TX

  5. 5

    Sakura Finetek, Torrance, CA

  6. 6

    Biovision Triglyceride Quantification Kit, Mountain View, CA

  7. 7

    Image J, National Institute of Health, Bethesda, MD

  8. 8

    SAS Institute Inc, Cary, NC

  9. 9

    Graphpad Software, Inc, San Diego, CA


  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References