• Open Access

Cardiac Troponin-I Concentration, Myocardial Arteriosclerosis, and Fibrosis in Dogs with Congestive Heart Failure because of Myxomatous Mitral Valve Disease

Authors


  • Preliminary results of this study were presented as a research abstract at the 2010 American College of Veterinary Internal Medicine (ACVIM) Forum, Anaheim, California
  • Corrections made after online publication April 3, 2013: references have been updated.

Corresponding author: Torkel Falk, Department of Clinical and Animal Sciences Veterinary Sciences; Faculty of Health and Medical Sciences, University of Copenhagen, 7 Grønnegårdsvej, 1870 Fredriksberg C, Denmark; e-mail: tof@life.ku.dk

Abstract

Background

Few previous studies have investigated the association between biomarkers and cardiac disease findings in dogs with naturally occurring myxomatous mitral valve disease (MMVD).

Aim

To investigate if histopathological changes at necropsy could be reflected by in vivo circulating concentrations of cTnI and aldosterone, and renin activity, in dogs with naturally occurring congestive heart failure because of MMVD.

Animals

Fifty privately owned dogs with MMVD and heart failure.

Methods

Longitudinal Study. Dogs were prospectively recruited and examined by clinical and echocardiographical examination twice yearly until time of death. Blood was stored for batched analysis of concentrations of cTnI and aldosterone, and renin activity. All dogs underwent a standardized necropsy protocol.

Results

cTnI were associated with echocardiographic left ventricular end-diastolic dimension (P < .0001) and proximal isovolumetric surface area radius (< .004). Furthermore, in vivo cTnI concentrations reflected postmortem findings of global myocardial fibrosis (P < .001), fibrosis in the papillary muscles (P < .001), and degree of arterial luminal narrowing (< .001) Aldosterone or renin activity did not reflect any of the cardiac disease variables investigated.

Conclusion and clinical importance

Cardiac fibrosis and arteriosclerosis in dogs with MMVD are reflected by circulating cTnI concentration, but not by aldosterone concentration or renin activity. Cardiac troponin I could be a valuable biomarker for myocardial fibrosis in dogs with chronic cardiac diseases.

Abbreviations
CHF

congestive heart failure

CKCS

Cavalier King Charles Spaniel

cTnI

cardiac troponin I

EDTA

ethylenediamine-tetraacetic acid

EF

ejection fraction

FS

fractional shortening

LA/Ao

left atrial-to-aortic root ratio

IQR

interquartile range

LVIDd

left ventricular internal diameter in diastole

LVIDd inc

increase of left ventricular internal diameter in diastole

LVID

left ventricular internal diameter in systole

MMVD

myxomatous mitral valve disease

MR

mitral regurgitation

PISA

proximal isovolumetric surface area method

VHS

vertebral heart scale

Myxomatous mitral valve disease (MMVD), which is the most common cardiac disorder in dogs,[1-4] is characterized by progressive degeneration of the mitral valve,[3-5] causing mitral regurgitation (MR) and subsequent chronic volume overload with left atrial dilation and left ventricular eccentric hypertrophy. The disease is typically chronic in nature and affected dogs compensate the MR for years; but chronic congestive heart failure (CHF) might eventually develop.

During progression of MMVD and development of CHF secondary to the disease, various cardiac remodeling events occur, which might adversely affect overall cardiac performance.[6-8] The process of cardiac dilatation in response to MR is characterized by dissolution of the interstitial collagen weave, a process allowing myocyte slippage. Cardiac dilatation inevitably leads to altered left ventricular geometry and, at some stage, to secondary MR. The process of cardiac dilatation has primarily been studied in experimental dogs, and represents fundamental mechanisms for cardiac remodeling. However, it is presumably accompanied by additional myocardial processes in dogs with MMVD, such as myocyte death and replacement fibrosis.[9, 10] Myocyte death can occur because of several processes, including necrosis and apoptosis.[11][12] Arteriosclerosis denotes a degenerative arterial change resulting in narrowing, hardening and loss of elasticity in the vessel wall.[13] Intramural coronary arteriosclerosis is more pronounced in dogs with MMVD compared with age-matched control dogs.[14] Arteriosclerosis can lead to a reduction in vasodilatory reserve and in combination with increased wall stress caused by volume overload induced by MMVD, a discrepancy between oxygen demand and supply might develop. Ultimately, these processes can lead to ischaemic injury and myocyte death. There is an association between arteriosclerosis and in experimental dog models of heart failure as well as in dogs with naturally occurring MMVD,[14, 15] indicating involvement of arteriosclerosis in the development of myocardial fibrosis in these dogs.[16]

The cardiac remodeling process can stimulate release of various biomarkers. Cardiac troponins are essential myofibrillar proteins that regulate the calcium-mediated action between actin and myosin filaments in cardiac muscle cells.[17] Myocardial injury causes a release of troponins into the circulation, and if the rate of release exceeds the rate of synthesis the contractile function of the myocardium becomes affected.[18] In humans, measurement of cardiac troponins can help quantify the degree of myocardial injury, monitor progression of disease, and offer prognostic information in heart failure patients.[19-23] In dogs with MMVD, however, concentration of cTnI increases with severity of disease, although diversity in concentration existed among dogs with similar disease severity.[23] The reason for this diversity is currently not known.

In dogs with advanced MMVD but with no or mild clinical signs of CHF, plasma concentrations of aldosterone and renin activity have been reported to be increased in one study[24] and decreased in another.[25] The aldosterone concentration and renin activity increase in CHF-dogs receiving furosemide.[25] Limited information is available in dogs regarding associations between circulating cTnI, aldosterone concentration, renin activity, and histopathological changes,[26-29] and these associations have, to our knowledge, not been systematically investigated in dogs with naturally occurring chronic cardiac diseases. Hence, the potential of these circulating substances as indirect markers of myocardial histopathological changes is currently not known.

Accordingly, the aim of our study was to investigate if histopathological changes at necropsy, particularly the degree of arteriosclerotic and fibrotic changes in the myocardium, could be reflected by in vivo circulating concentrations of cTnI and aldosterone, and renin activity, in dogs with naturally occurring CHF because of MMVD.

Material and Methods

Dogs

Fifty dogs of different breeds and gender with CHF because of MMVD were included in this study. Included dogs were part of another longitudinal study of naturally occurring MMVD in which dogs were followed during the progression of MMVD into CHF until they ultimately died or were euthanized.[30] The investigation conforms with the guide for the care and use of laboratory animals published by the US National Institutes of Health (publication 85-23, revised 1996).

Inclusion Criteria

Dogs were originally included in the longitudinal study provided that they presented a characteristic heart murmur of moderate-to-high intensity with maximal intensity over the mitral area, and showed echocardiographic evidence of advanced MMVD defined as characteristic valvular lesions of the mitral valve apparatus (leaflet thickening, valve prolapse), moderate-to-severe mitral regurgitation on color Doppler echocardiography, and echocardiographic evidence of moderate-to-severe left atrial and left ventricular dilatation, i.e, LA/Ao (left atrial-to-aortic root) ratio > 1.9 and cardiomegaly on thoracic radiographs (vertebral heart scale > 10.5).[31] To be included in the present study, dogs had to, during the course of the study, have developed radiographic evidence of pulmonary edema, and furthermore, shown clinical signs of decompensated congestive heart failure before death or euthanasia.

Exclusion Criteria

Dogs were excluded from this study if they had cardiac disease (congenital or acquired) other than MMVD, histological evidence of dilated cardiomyopathy (median myocardial atrophy score > 2, denoting an atrophy or attenuation of more than 20% of myofibrils, or evidence of fibrofatty degeneration), or presence of another significant systemic disease including evidence of significant organ dysfunction.

The dogs were repeatedly examined at 6 month intervals by a standardized protocol that included physical examination, thoracic radiography, echocardiography, and electrocardiography (ECG), until they died or were euthanized because of refractory CHF. Only clinical, echocardiographical, and hormonal data from the last examination before death are analyzed and reported in this paper. Thoracic radiographs were performed mainly to exclude presence of lung pathology and to ascertain presence of cardiomegaly and these data were not included in the statistical analyses.

Blood Sampling

Blood was collected from the cephalic vein by venipuncture into plain tubes and tubes containing ethylenediamine-tetraacetic acid (EDTA). The blood samples were centrifuged and plasma and serum were separated within 30 minutes and stored initially, because of lack in house facilities, at −20°C for a maximum of 14 days and then, after transport on dry ice, at −80°C until batched analysis of cTnI and aldosterone concentrations and renin activity.

Cardiac Troponin-I (cTnI) Concentration

Concentration of cTnI was measured in serum in all dogs by a high sensitivity enzyme-linked immunosorbent assay,1 as previously described.[23] The samples were assayed in duplicate and the mean of these results was used in the statistical analysis. Values are reported as ng/mL. The lower limit of detection for the assay used in this study reported here was 0.001 ng/mL, in accordance with earlier studies.[23]

Aldosterone Concentration and Renin Activity

Plasma aldosterone concentrations were measured in duplicate in 42 dogs by a radioimmunoassay kit previously described and validated for use in canine samples.2 Values are reported as pg/mL. The lower limit of detection for the assay used in the study reported here was 0.12 pg/mL, in accordance with earlier studies.[32]

Plasma renin activity levels were assayed in 33 dogs by radioimmunoassay as previously described and validated for use in canine samples.[33] For the purpose of statistical analysis, plasma renin activities above the detection limit of 192 ng/mL/h were assigned a value of 192 ng/mL/h. In accordance with previous studies,[34] the lower limit of detection for the assay used in the study reported here was 0.1 ng/mL/h.

Echocardiography

Echocardiography was performed in all dogs by the same experienced echocardiographer (TF). Standard echocardiographic equipment (Acuson Aspen, Siemens Medical) with 5–7 and 2–4 MHz sector transducers were used. Unsedated dogs were positioned in lateral recumbency and ECG electrodes were attached to the limbs. Standard echocardiographic views were obtained. A minimum of 3 digital short sequences (2–3 seconds) from each location was stored. From stored images the following data were obtained: left ventricular free wall thickness in diastole and systole left ventricular diameter in diastole (LVIDd) and systole (LVIDs), and interventricular septum thickness in diastole and systole (M-mode), left ventricular length (2D long axis view), left ventricular area in diastole (luminal and total) and systole (luminal) to compute ejection fraction (EF), LA/Ao (2D short axis view, left atrial/aortic root level).[35] The following variables were calculated: LA/Ao ratio, fractional shortening (FS), and EF by Bullet formula.[36] Left ventricular size measurements were indexed to body size as follows: [observed dimension – expected normal dimension]/expected normal dimension  × 100 (LVIDd inc). Expected normal dimensions were calculated as previously described.[37] From color-flow Doppler recordings the regurgitant jet size was measured in a left apical 4-chamber view with the baseline shifted to decrease the Nyquist limit to 25–60 cm/s in the flow direction, and then the Nyquist limit in flow direction was optimized to obtain a semicircular flow convergence pattern for measuring proximal isovelocity surface area (PISA) radius. Aortic flow and mitral inflow velocities (modified 5 and 4-chamber view) were determined by continuous wave Doppler.[38]

Postmortem Examination

All dogs underwent postmortem examination within 3 days of death (94% within 1 day). For postmortem analysis, the heart was removed and dissected. Both ventricles were opened from the apex up through the aorta and pulmonary artery and the heart was assessed macroscopically according to a predefined protocol: dilation or hypertrophy of the heart chambers, fibrosis, valvular changes according to the Whitney 0–4 grade scale,[39] presence of jet lesions and chordal ruptures. Tissue specimens of approximately 1 × 1 cm were collected for histopathological examination from the following sites: upper and lower parts of cranial and caudal papillary muscle; upper, middle, and lower parts of left ventricular wall, interventricular septum and right ventricular wall (Fig 1). All tissue blocks were wax embedded in cassettes3 and underwent routine sectioning and staining with hematoxylin-eosin at a commercial pathology laboratory.4 Four to 8 tissue sections (2–4 μm) from each localization of the myocardium were examined histologically. The examining histopathologist (TF) was blinded to clinical and echocardiographical status of the dogs by by coded histology slides. At least 6 intramyocardial arteries were digitally photographed from each myocardial localisation. Vessels were included if they were 100–400 μm in diameter and cut transversely, excluding nonround and obliquely cut vessels. Fibrosis was scored in the myocardium according to a 5 grade scale: 0 =  no evidence of fibrosis; 1 =  mild interstitial fibrosis; 2 =  moderate interstitial fibrosis; 3 =  areas of confluent (replacement-type) fibrosis; and 4 =  large areas of confluent fibrosis. Each section was examined and scored and median scores for each localization were calculated. An average of the fibrosis score based on all collection sites for each heart examined is referred to as global fibrosis score, whereas fibrosis scores from a specific site, e.g, cranial papillary muscle, are an average score of all the samples examined from that collection site for each heart.

Figure 1.

Normal intramural myocardial artery in a dog (hematoxylin-eosin, 40x).

Figure 2.

Arteriosclerotic intramural myocardial artery in a dog with arterial narrowing (hematoxylin-eosin, 40x).

Figure 3.

Fitted regression line and 95% confidence interval for the regression of the logarithm of the concentration of cardiac troponin I (log(cTnI)) on the average of mean cardiac fibrosis score in all localisations (global fibrosis).

Myocyte atrophy was defined as myofibers with a diameter of less than 6–7 μm. Grade of atrophy was scored in all sections according to a 5 grade scale: 0 =  no evidence of atrophy; 1 =  evidence of atrophy involving less than 20% of the myocardium and mainly subendocardially; 2 =  evidence of atrophy involving 20–50% of myocytes; 3 =  severe atrophy but involving ≤ 80% of the myocytes; 4 =  severe atrophy involving > 80% of myocytes. Measurements were made from the digital photographs by publicly available software.5 The previously described lumen area ratio (LAR),[14] defined as the luminal area of the vessel divided by the total vessel area, not including the adventitia, was used as a measure of vessel narrowing.

Statistical Analyses

All statistical calculations were performed by a commercially available statistical software program.6 Data are presented as medians and interquartile range (IQR). To what extent the histopathological findings were reflected by the selected blood variables and clinical findings was examined by use of least squares multiple regression analyses with the selected blood variables as response variables, histopathological and echocardiographic findings as predictors, and dog characteristics (including age, sex, breed (CKCS or other breed), body weight, storage time of the blood from sampling until batched analysis, and time from sampling to death) as covariates.

The postmortem variables fibrosis and LAR (global as well as in the papillary muscles), as well as the echocardiographical variables (LVIDd inc, LA/Ao, FS, EF, and PISA) were all entered in the model separately because of high degree of covariance between these variables.

For each multiple regression model, the residuals were tested for homogeneity and normality of variation by means of residual plots and Shapiro–Wilks test, respectively. Cardiac TnI, aldosterone concentration, and plasma renin activity were all log transformed to achieve normal distribution of skewed data. Because the variables global LAR, LAR pap, LA/Ao, PISA, FS, and EF were all proportions, these variables were logit-transformed, to enter them in the multiple regression analyzes as continuous variables. In the multiple regression models, analyses were performed in a backward stepwise manner, until all the remaining variables had a value of P < 0.01. All variables were assessed only as main effects, since no interactions were included in the model.

Results

Dogs

The 50 dogs (35 males and 15 females) included in this study comprised 19 different breeds, including 20 CKCS and 9 Dachshunds. Among the other breeds, Cocker Spaniel (3), Kelpie (2), and German Shepherd (2) were represented by more than 1 dog, and 14 breeds (Standard Poodle, Bichon Frise, Bichon Havanaise, Chihuahua, Miniature Schnauser, Jack Russel Terrier, Norfolk Terrier, Old English Sheepdog, Rottweiler, Shetland Sheepdog, Springer Spaniel) by only 1 dog. The dogs had a median age at inclusion of 10 years (IQR 8.8–11.6) and the median body weight was 10.4 kg (IQR 9.9–12.1).

Echocardiographic Variables

Included dogs were characterized by an average LVIDd inc of 55.2 (IQR 41.8–67.6)%, LA/Ao ratio of 2.1 (IQR 1.9–2.2), FS of 33.5% (IQR 28.0–37.0), and EF of 63.0% (IQR ± 52.0–68.0).

Concentrations of cTnI and Aldosterone, and Renin Activity

Median cTnI concentration was 0.08 ng/mL (IQR 0.05–0.15), but 1 dog, excluded from statistical analysis, had a cTnI value of 20.57 ng/mL, the highest value measured in this study, whereas the 95% quantile was 0.67 ng/mL. Median aldosterone concentration was 83.50 pg/mL (IQR 36.00–190.00) and median renin activity was 14.44 ng/mL/h (IQR 2.60–25.44).

Postmortem Findings

Dogs died or were euthanized at a median time of 70 days (IQR 29–107 days) from the time of the last clinical examination and collection of plasma. At necropsy, all dogs had pronounced MMVD changes, i.e, grade 2 (n = 2), 3 (n = 47), or 4 (n = 1) according to the Whitney classification. Median fibrosis score from all sections was 1.2 (IQR 0.8–1.6) and median fibrosis score from papillary muscle sections was 1.8 (IQR 1.3–2.3). Median LAR from all sections was 0.20 (IQR 0.16–0.27) and median LAR from papillary muscle sections was 0.15 (IQR 0.12–0.23).

Multiple Regression Analyses of cTnI and Signalment, Echocardiographic and Postmortem Variables

In the multiple regression analyses performed with serum cTnI concentration as a dependent variable and baseline dog characteristics, histopathological and echocardiographic findings as independent variables, cTnI concentration increased with increasing grade of both global fibrosis (P < .0001) and fibrosis in the papillary muscles (< .0001), and decreased with increasing LAR (P < .0001) (Fig 1, Table 1). All the echocardiographic variables were included in the statistical analyses, but except for the ones reported, none of them, including mitral inflow velocities, did reach statistical significance. Body weight was not significantly associated to any of the echocardiographic variables. Furthermore, cTnI increased with increasing LVIDd inc and increasing PISA (= .0007, and = .0044, respectively) (Table 1). The formerly mentioned outlier (serum concentration of cTnI 20.57 ng/mL) was excluded from all further statistical analyses, but multiple regression analyses were also repeated with inclusion of the same observation, and the results were consistent and robust regarding the covariates of interest.

Table 1. Summary of the entered predictors in each multiple linear regression model with cTnI as response variable, as well as P values of the slope of the regression coefficients and the adjusted R2 of the respective final models.[1]
ResponsePredictorP ValueFinal Model Adj R2
  1. The covariates body weight, age, sex, CKCS/not CKCS, time from blood sampling until death, and time from blood sampling until death were entered in each model but were, however, not significantly associated to cardiac troponin-I (cTnI) in any model. In order to achieve normality of the residuals, cTnI was log-transformed.

  2. NS, nonsignificant; Fibrosis, global fibrosis; Fibrosis pap, fibrosis in the papillary muscles; LAR, lumen area ratio; LAR pap, lumen area ratio in the papillary muscles; LA/Ao, left atrial-to-aortic root ratio; LVIDd inc, percentage increase in left ventricular internal diameter; PISA, proximal isovelocity surface area; EF, ejection fraction; FS, fractional shortening.

cTnIFibrosis<.00010.34
Fibrosis pap<.00010.31
LAR<.00010.29
LAR papNS
LA:AoNS
LVIDd inc.00070.21
PISA.00440.16
EFNS
FSNS

Multiple Regression Analyses of Plasma Aldosterone and Renin and Dog Characteristics, Echocardiographic and Postmortem Variables

None of the included dog characteristics, echocardiographic or histopathological variables were reflected by plasma aldosterone concentrations in the included dogs, neither were echocardiographic or histopathological findings reflected by plasma renin activity (data not shown).

Discussion

This study demonstrates that serum cTnI concentrations reflect the presence of intramyocardial fibrosis and arteriosclerosis in dogs with MMVD. This could imply a novel approach to the use of cTnI as a predictor for fibrosis in naturally occurring chronic cardiac disease.

The increased cTnI concentrations in our population of dogs are most likely because of an ongoing release of cTnI caused by a continuous remodeling process, rather than an acute one. Most cTnI is structurally bound within the myocyte, and it is released into the circulation only after cell injury.[28] In the multiple regression analysis, the variables fibrosis and arterial narrowing explained most of the observed variation in cTnI concentration. Serum cTnI concentrations seem to reflect the presence of intramyocardial fibrosis and arteriosclerosis, and this lends support to the hypothesis that arteriosclerosis might cause myocardial damage and play a role in the pathogenesis of heart failure in MMVD in dogs, as proposed previously by some authors.[16, 40] We interpret the association of increased heart size and PISA-radius to increasing cTnI concentrations in this study as a consequence of advancing heart failure or advancing heart disease in itself. The fact that aldosterone concentrations and plasma renin activity were not associated to histological changes in the myocardium might be because a marked influence of treatment regimens on these hormones masks associations with disease processes.

Myocardial fibrosis can be caused by a wide variety of insults to the myocardium, which can be of toxic, inflammatory, metabolic and idiopathic origins. The type of fibrosis seen in this study, especially pronounced in the most severely affected dogs, was of replacement type, where normal myocardial tissue is replaced by fibrous tissue in larger or smaller areas. This type of fibrosis is almost invariably caused by ischemia, and pathologists often refer to it as infarcts.[16] Replacement fibrosis is mainly oriented close to vessels, and through serial sectioning an association between arteriosclerosis and replacement fibrosis/infarcts has previously been demonstrated in dogs with MMVD.[16] Both associations between arteriosclerosis and fibrosis, as well as associations between arteriosclerosis/fibrosis and MMVD, have previously been reported.[1, 14] In previous studies we have demonstrated an association between intramyocardial arteriosclerosis and MMVD.[14] In this study it was also demonstrated that mild intramyocardial fibrosis could be an age-related change whereas more severe forms of fibrotic changes (replacement type) were not associated with age. Intramyocardial arteriosclerosis might affect the vascular supply to the myocardium ultimately promoting regional hypoxia and myocyte death.[42] Interestingly, in microembolization models in dogs where smaller myocardial vessels (~100 μm) were occluded by intracoronary injections of latex spheres, the same type of fibrotic changes as seen in this study was provoked, supporting the hypothesis that intramural arteriosclerosis might be involved in the pathogenesis of myocardial fibrosis.[15] Such an arteriosclerosis-induced myocyte death might be a causative factor in development of myocardial fibrosis in MMVD in dogs. Yet, a reversed scenario is possible: an increasing amount of myocardial fibrosis caused by remodeling of the extracellular matrix might reduce the capillary density in the myocardium and thereby the oxygen diffusion distance resulting in damaged myocyte integrity.

There are limitations to this study. The blood samples were stored for a variable and sometimes considerable time before batched analysis. The blood samples were collected at a variable time point before death or euthanasia because of the study design. This did, however, not seem to influence the concentrations of cTnI in the statistical model. The reason for this could be that the arterial changes and fibrosis might be progressive chronic disease processes during which changes in cTnI concentrations are slow and fairly constant over a long period of time.

The dogs were not on standardized treatment protocols, which might have influenced the results, especially the aldosterone concentrations and renin activities.

Lastly, no control group was included in this study. However, all dogs were in an advanced disease stage, which in itself decreases the usefulness of controls, since we compare histological changes in dogs with advanced disease and their different serum cTnI concentrations. Furthermore, a previous study, utilizing the same analyzing method in the same laboratory, showed that dogs of comparable breeds and ages without MMVD or with less advanced disease had lower circulating cTnI concentrations compared with cTnI concentrations in the severely affected dogs in the current study.[23] Some large breed dogs were included, which might be a weakness concerning the homogeneity of the study population. However, these dogs all had severe mitral valve changes, no histological signs of cardiomyopathy, and body weight was not significantly associated with any of the echocardiographic variables. Another study limitation is that some additional echocardiographic measurements, not included in the prospective protocol, could have been more appropriate for estimation of systolic and diastolic function.

In conclusion, circulating cTnI concentrations assayed by a high sensitivity assay might reflect the presence of intramyocardial fibrosis and arteriosclerosis in dogs with naturally occurring MMVD, and might therefore be of value as a predictor of myocardial fibrosis in this disease. Circulating aldosterone concentration and renin activity do not seem useful as biomarkers for myocardial fibrosis. The potential role of circulating cTnI as a biomarker for myocardial fibrosis in dogs with chronic cardiac diseases needs to be further investigated in future studies.

Acknowledgments

The study was supported by the Danish Council of Medical Sciences, F. Thure and Karin Forsberg's Foundation and Agria animal insurance company research foundation, Sweden.

Conflict of Interest Declaration: Authors disclose no conflict of interest.

Footnotes

  1. 1

    Access AccuTnI, Beckman Coulter, Fullerton, CA

  2. 2

    Coat-A-Count; Diagnostic Products Corporation, Los Angeles, CA

  3. 3

    Tissue-Tek system, Electron Microscopy Services, Liverpool, UK

  4. 4

    BioVet pathology lab, Stockholm, Sweden

  5. 5

    Image J, public domain JAVA-progam

  6. 6

    SAS statistical software, version 9.1, SAS Institute

Ancillary