The aims of the study were to investigate the value of clinical evaluation, biomarkers, and diagnostic methods to detect and identify pathophysiologic progression of mitral regurgitation caused by degenerative myxomatous mitral valve disease. The specific aims were to:
# evaluate the influence of age and gender on N-terminal A-type natriuretic peptide (NT-proANP), nitric oxide (NO), measured as its degradation products nitrate and nitrate (NOx), and endothelin (ET-1) in dogs free of heart murmur (paper I)
# estimate the impact of progressive MR on PTT in dogs (paper II)
# evaluate the effect of severity of MR on right heart chamber size and shape (paper III)
# measure pulmonary blood volume (PBV) by means of pulmonary transit time (PTT) and stroke volume assessed by first pass radionuclide angiocardiography and Doppler echocardiography, respectively (paper IV)
# assess the predictive value of NT–proANP and NOx for congestive heart failure (CHF) in dogs caused by mitral regurgitation (MR) by means of appropriate longitudinal statistical methods (paper V)
# explore whether true longitudinal statistical methods can detect trends not detected by traditional methods so far used in veterinary cardiology (paper V)
V@*Q8D9HB8AC*8!R*Q9DPERC*
V@?*8-1'()+*(-.*+/#.1"+*
Studies were conducted at the Small Animal Hospital of University of Helsinki, Hämeentie 51, Finland; the current location of which is Viikki, Helsinki, and the Small Animal Hospitals of the Swedish University of Agricultural Sciences in Uppsala, and the Royal Veterinary and Agricultural University in Copenhagen, respectively.
Except for study III, dogs exclusively of the breed Cavalier King Charles spaniel (CKCS) were enrolled. All dogs were privately owned, either free of heart murmurs (healthy control dogs), or dogs with different degrees of mitral regurgitation (MR) due to degenerative myxomatous mitral valve disease (MMVD). Only dogs with no signs of other disease were recruited for the studies. All owners were informed of the studies to be conducted. The University Ethics Committee (Finland, Sweden and Denmark) approved the study protocols and owner consent was received before any procedures.
The recruited CKCS dogs (paper II- V) also participated in the Scandinavian Enalapril Prevention trial (SVEP) (Kvart, Häggström et al. 2002).
In study I healthy CKCS dogs were recruited via the local breed association. Dogs of different age were screened for absence of a heart murmur and signs of any concurrent disease. All dogs underwent clinical examination, thoracic radiography, echocardiography, blood and urine sampling (n=17). N-terminal proANP, BNP, ET-1 and NOx were analysed from blood samples and cGMP from urine samples.
Studies II and IV comprised CKCS with a regurgitant murmur originating from the mitral valves (n= 44 and 33, respectively). Some of the dogs in study II and IV also took part in the SVEP study (Kvart, Häggström et al. 2002). Blood pulmonary transit time was measured by first pass radionuclide angiocardiograms (FPRNAs) (Paper II) and in paper IV pulmonary blood volume was calculated. The author of this thesis performed examinations of dogs, analysed data and contributed equally in writing.
Study III was a retrospective study using archived FPRNAs made on healthy small breed dogs and dogs with varying degrees of MR. This study included six other small breed dogs in addition to CKCS dogs. The material included FPRNAs from study II in Finland, and Swedish dogs participating in the QUEST study (Häggström, Boswood et al. 2008) at the Swedish University of Agricultural Sciences. The author of this thesis took part in the examination of dogs, the analyses and the writing of the manuscript.
Study V utilized Finnish, Swedish, and Danish dogs, enrolled and examined at the aforementioned Universities, taking part in the prospective SVEP study between 1995 and 2000 (Kvart, Häggström et al. 2002). Dogs were allocated to receive either enalapril, an ACE inhibitor, or placebo in a double-blinded manner. All dogs included had a characteristic mitral murmur. Some had enlargement of the left atrium and ventricle but none had clinical or radiographic signs of cardiac failure. The dogs in the weight range of 5 but less than 10 kg (at entrance to trial) received 2.5 mg enalapril or
placebo and dogs in the range 10-15 kg received 5 mg enalapril or placebo. Animals were evaluated at entry and every 12 months there after until signs of congestive heart failure (CHF). Additional examinations were done whenever owners suspected signs of CHF (n=78). N-terminal proANP was measured from blood samples drawn at every visit. In addition, NOx was measured from Finnish dogs. The author of this thesis participated in all parts of the process of this study.
V@J*Q"/&%.+*
V@J@?*5)1-13()*"U('1-(/1%-+*
All dogs underwent a thorough physical examination. In multicenter studies (Papers II and V), a special evaluation form was used to allow evaluation of dogs in a uniform manner. Murmurs of mitral origin were graded on the conventional scale from 1 to 6 (Freeman and Levine 1933).
The origin of murmur was determined by echocardiography. Thoracic radiographs in left lateral and ventrodorsal projections were made. All radiographs were evaluated for signs of cardiomegaly, interstitial or alveolar pulmonary edema. In addition, first pass radionuclide angiocardiograms (FPRNAs) were made in dogs participating in studies II and IV. For study I examinations were performed to rule out cardiac and any other disease.
Congestive heart failure (CHF) was defined as the time when dogs were presented with signs of heart failure. The diagnosis of CHF was only accepted when the data from the case history and physical examination were accompanied by cardiomegaly, including left atrial enlargement and interstitial or alveolar pulmonary edema on thoracic radiographs. Information of special interest from dog owners was dyspnoea, cough, nocturnal restlessness and exercise intolerance (Appendix I). In case of spontaneous death, CHF could also be considered to be reached if the diagnosis of heart failure, i.e. pulmonary edema and pulmonary congestion, was confirmed by post mortem examination.
V@J@J*G)%%.*(-.*#$1-"*(-()0+"+*
Blood and urine samples were taken according to defined protocols. Blood was drawn from each dog for measurement of plasma immunoreactivity of NT-proANP, BNP, and NOx. For analyses of NT-proANP, BNP and ET-1 we used pre-chilled tubes (in ice-bath) containing sodium-EDTA 0.15mM (final concentration) and aprotinin 50IU/ml plasma. Plasma was separated in a cold centrifuge and stored at -80 degrees Celsius until assayed.
NT-proANP was measured using a direct radioimmunoassay (RIA) as described (Vuolteenaho, Koistinen et al. 1992). Antibodies were raised in rabbits against human proANP- (79-98), which is identical to the corresponding sequence in the dog, except for amino acid number 95. NT-proANP was assayed directly from plasma as follows.
Standards [synthetic human proANP-(79-98) Peninsula] and samples (25µl) in duplicate were incubated with 200µl of radioiodinated human Tyr-(O)-proANP-(79-98) and 200µl of rabbit antiserum 135 (final dilution l/40 000) overnight at +4 Co. The bound and free fractions were separated by double antibody precipitation in the presence of polyethylene glycol. The sensitivity of the assay was 0.5 fmol/tube and the within and between assay coefficients of variation were <10% and <15%, respectively.
Gel filtration high-performance liquid chromotographic (HPLC) analyses of plasma samples consistently showed the presence of only one immunoreactive species corresponding in size to that of intact NT-proANP [proANP-(1-98)].
BNP was analysed using an antiserum raised in a rabbit against porcine BNP1-26, a synthetic peptide that was also used as the tracer and standard. Canine BNP peptide sequence differs from that of porcine BNP at 2 sites. The divergent amino acids reside at the ends of the porcine BNP sequence (positions 1 and 26). To form the immunogen, carbodiimide was used to couple porcine BNP1-26 from both ends to thyroglobulin.
Therefore the divergent amino acids should not interfere with the binding of canine BNP to the antiserum. The cross-reaction of the antiserum to human ANP and NT-proANP was <0.1%. Recovery ranged from 60 – 81% and the detection limit was 1.5 fmol/tube. Serially diluted dog plasma and porcine BNP standards were diluted in parallel thus validating quantification of dog BNP by the assay. However, as for NT-proANP, although the relative changes observed in the peptide concentrations were accurate, the absolute values may not have been, because assays are heterologous for dog samples.
Plasma immunoreactive ET-1 was measured by use of RIA as described earlier using synthetic human ET-1 and endothelin antiserum raised in rabbits (Fyhrquist, Saijonmaa et al. 1990). Sequences of ET-1 from dogs, rats, mice, humans, and pigs are identical (Masaki, Yanagisawa et al. 1992). Recovery of synthetic ET-1 added to plasma was 84%. Before radioimmunoassay plasma samples were purified using Bondelut C18-OH analytical columns prewashed with methanol, distilled water and 4% acetic acid. The samples were then lyophilized under vacuum. Dried samples were dissolved in a prepared RIA buffer. The sensitivity of the assay was 0.8 pg/tube. The RIA intra-assay and inter-assay CV were 10 and 11%, respectively
Blood and urine nitrate (NO3) was measured as nitrite (NO2) using a fluorometric method (Misko, Schilling et al. 1993). Briefly, red blood cells are removed by centrifugation in the presence of heparin or EDTA. Plasma is filtered through a 10 000 Dalton cut-off filter to remove albumin and haemoglobin resulting from cell lysis.
Nitrate is converted to nitrite with nitrate reductase and finally 2,3-diaminonaphtalene is added to obtain fluorescence at 560nm.
Urine samples containing acetic acid were analysed for cyclic guanosine monophosphate (U-cGMP) content using a commercial radioimmunoassay (cGMP assay system RPA 125, Amersham, Buckinghamshire, UK). 5ml urine was put into vials containing 50µl acetic acid and 0.5ml of papaverine (10µmol/l). Samples were stored at -20 C0 until analysed according to manufacturer's instructions.
V@J@M*R1(2-%+/13*1'(21-2*
<4,%4-.5/%3.-$,6'
Standard right transthoracic two-dimensional echocardiography was done according to established guidelines (Thomas, Gaber et al. 1993). Left ventricular end diastolic (LVEDD) and systolic (LVESD) dimensions were measured from M-mode recordings (Figure 10). The arithmetic mean of at least three measurements was used.
Mean left ventricular dimensions were normalized for body size (nLVEDD and nLVESD, respectively) according to formulas published(Cornell, Kittleson et al.
2004): nLVEDD= LVEDD[cm]/(body weight [kg])0.294, and nLVESD=
LVESD[cm]/(body weight[kg])0.315, respectively.
The left atrial to aortic root ratio (LA/Ao) was measured from the right parasternal short axis view in mid diastole (Figure 5, page 29) after aortic ejection, where the Ao appeared as a symmetric three-leaf clover with closed aortic valves, similar to the technique later published (Hansson, Häggström et al. 2002), except they measured in early diastole.
=,%.-4/4'.-5/%3.-$,6'
Thoracic radiographs in left lateral and ventrodorsal projections were made for all dogs. All radiographs were evaluated for signs of cardiomegaly and interstitial or alveolar pulmonary edema (Figure 11). The radiographic examinations of dogs in the prospective study (V) were reviewed at annual researchers evaluation meetings to ensure uniform film readings. At the completion of the trial, a veterinary radiologist, blinded to drug assignment and dog identity, evaluated all radiographs according to the protocol. If not confirmed on radiographs by signs of cardiomegaly, including left
Figure 10. Transverse echocardiographic view of the heart (left panel) with M-mode cursor placed between the papillary muscles. LV, left ventricle;
RV, right ventricle. Corresponding M-mode view (right panel) with left ventricular end-diastolic (LVEDD) and left ventricular end-systolic dimensions (LVESD) measured.
atrial enlargement and presence of interstitial or alveolar pulmonary edema, the proposed clinical diagnosis and date for CHF were to be rejected (Kvart, Häggström et al. 2002). The vertebral heart score (VHS) was measured according to a slight modification (Hansson, Häggström et al. 2005) to the method described (Buchanan and Bucheler 1995, Hansson, Häggström et al. 2005) to reduce the variability in the selection of points to mark the short axis of the heart at the vena cava.
>/.#+'$-##'.-5/%&140/5"'-&3/%4-.5/%3.-$,6'
First pass radionuclide angiocardiography (FPRNA) or nuclear angiocardiography, was used to assess pulmonary transit times (PTT) (Papers II and IV) and pulmonary blood volume (PBV) (Paper IV). Technetium-99m diethylenetriaminepentaacetic acid (Technescan® DPTA, Mallinckrodt Medical B.B., Petten, Holland) was used as the tracer. It was placed as a 1ml bolus into the iv tubing inserted into a cephalic vein. To ensure a tight bolus the tubing was then flushed with saline. An ECG recording was started simultaneously to the gamma camera acquisition (10 frames/s).
To measure the PTT, regions of interest (ROI) were drawn around the right ventricle (RV), a representative area of the caudal lung excluding aorta near the LA, and the LA (Figure 12). As flow in the lungs overlies the LA and would create a false early rise in LA activity, pixel-normalized lung activity was subtracted from the LA pixel values. PTT was then measured as the time from the first appearance of the radioactive bolus in the pulmonary trunk to the first appearance in the LA (leading edge method) (Slutsky, Bhargava et al. 1983). The mean times between ventricular beats (mRR), i.e. R-waves, during the transit time were measured on the ECG strip.
The normalized pulmonary transit time (nPTT) and pulmonary blood volume (PBV) were calculated by established formulas [Formula 2, page 23](Milnor, Jose et al. 1960, Giuntini, Lewis et al. 1963, Slutsky, Shabetai et al. 1983).
Figure 11. A lateral thoracic radiography of a dog with only minimal enlargement of the heart (Panel A). Panel B presents the same dog years later in congestive heart failure with severely increased global heart size, dilated pulmonary veins (arrow) indicating congestion, and increased pulmonary radiodensity caudal to the tracheal bifurcation consistent with pulmonary edema.
Static images of the right heart (RH) and left heart (LH) chambers were made from both of the FPRNAs (Paper III). Ten to eighteen 0.1 second frames of the dynamic study were summed so that each new frame was 1.0–1.8 seconds. The images were scrolled in time to show the bolus filling and outlining the right atrium and ventricle. The number of summed frames and the position of the composite frame in the RH chambers were adjusted to minimize overlap of the pulmonary artery and caudal vena cava with the endocardial margins, and maximize the filling of the chambers. This image was saved as a static image. The images were then scrolled in time to one in which the bolus of radioactivity was filling and outlining the left atrium and ventricle, and this static image was saved. A red-blue color table was used to give a sharp definition between red and blue for location of ROIs (see Paper III for details).
The degree of distortion of the RH chambers was measured as an index of circularity, i.e. the ratio short axis/long axis. An index of 1.0 indicates a circular figure, whereas flattening decreases the value. The shape of the septal border of the RH nuclear angiocardiogram was assessed subjectively as convex or flattened.
V@M*C/(/1+/13+*
Routine statistics were done with commercial statistical software (JMP v. 7 – 9, SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513, USA) using
Figure 12. An angiocardiograms after a bolus of a radioisotope into a peripheral vein in the front leg showing high activity in the right ventricle (RV) and early activity in the lungs. LA, left atrium; LV, left ventricle.
appropriate parametric or nonparametric standard tests, including adjustments for several groups. Normal time to event methods (survival analysis) including Cox regression analyses (proportional hazards model) and the Kaplan-Meier curve estimates were used to estimate hazard ratios (HR) and survival times, respectively; in this case (survival) time to congestive heart failure (CHF).
Dog owners were instructed to request an immediate clinical examination whenever they had reasons to suspect signs of CHF according to given instructions. In addition to the scheduled annual visits, the prospective longitudinal study (Paper V) therefore included several unevenly randomly distributed time points performed according to the full protocol. Standard longitudinal models rely on defined time points equal for all individuals. To be able to utilize all unevenly distributed time points, the free statistical software R (R Core Team, R Foundation for Statistical Computing, Vienna, 2012) was used to construct an enhanced Cox model (Allignol and Latouche 2012) using the "survival"- package (Therneau 2012). In the Cox model the fundamental quantity used to assess the risk of event occurrence in each discrete time period is known as hazard. The mathematical formula of the ordinary Cox function is:
[4] h(tij) = h0(tj)e[ß1X 1ij + ß
2X
2ij +…+ ß nX
nij] ,
where i = individual, j = time point, and ß = effect on n predictors (X).
Thus, the Cox hazard function summarizes the probability (h(tij)) that individual i experiences the event in time period j, given no previous event occurrence. In other words, in the time period when individual i experiences the event, the individual contributes h(tij) to the likelihood function (i.e. how well the variables describe the real situation). The hazard ratio (HR) in turn expresses the difference (calculated as ratio) between two hazard functions. That in turn is the relative risk for individuals with a certain level or type of predictor compared to another type. By definition then, the standard Cox model can deal only with discrete ("pre-fixed") time periods.
To be able to analyse our data (paper V) with continuous time and random event points we needed to extend the Cox model as follows:
[5] h(tij) = h0(tj)e[ß1X1D 1ij + ß2X2D ij +…+ ßnXnD nij],
In this model we divide continuous time into epochs ("time periods", coded with
“start, stop” time), each represented by one of n time-indicators, D1 – Dn. All epochs need not be the same length, but taken together, they must cover all observed time events (Singer and Willett 2003). ß1 is the difference in hazard associated with a one-unit difference in X (the predictor or covariate) during the first epoch (D1<D2), ß2 is the difference in hazard associated with a one unit difference in X. In addition, we had to take into account that consecutive measurements for dogs were not independent, meaning the same dog contributed Dn times to the data set. Therefore, the “cluster”-term (dog) was introduced in the model formula indicating there were multiple
observations (clusters) from the same subject, requesting robust standard errors be produced for the coefficient estimates. Robust standard errors are designed to account for the non-independence of observations from the same subject (Kleinbaum and Klein 2012). These concepts were further developed and coded for calculations in the
"survival"-package (Therneau 2012) available for the R-software.
To be able to utilize these types of analyses, data must be converted to a "person-period" form, i.e. for every individual there is a row for every time point. Time points were calculated in days. The survival package was instructed to calculate the time difference in each time epoch and time of event for each individual (start, stop).
The main differences to the ordinary Cox model are, that the extended model can:
# account for change in predictors by time, whereas the ordinary model only can handle baseline values (fulfilling the proportionality assumption)
# deal with time as a continuous variable, not only predefined discrete time points, i.e. measurements may be taken at random times for each individual
# control for inter-individual change
# evaluate hazards for intervals of covariates
W@*H9CLADC*
W@?*!(/$1#$"/13*,",/1."*(-.*"-.%/&")1()*$"+,%-+"*Y<(,"$+*B*(-.*IZ*
W@?@?*944"3/*%4*(2"*%-*-"#$%&#'%$()*$"+,%-+"+*1-*&"()/&0*.%2+*Y<(,"$*BZ*
We measured an increase in plasma levels of both NT-proANP and NOx with age (R2= 0.48, P= 0.002 and R2= 0.37, P= 0.01, respectively), whereas there was no detected association with age for ET-1, BNP and urinary cyclic GMP. However, ET-1 was significantly associated to heart rate (R2= 0.37, P= 0.0086). We also found a correlation between NT-proANP and NOx(R2= 0.38, P= 0.0002). Using simple linear regression analysis the mean estimated increase in plasma NT–proANP was 25pmol/year. The linear regression curve can be described as: NT–proANP [pmol/l]=
185+0.02576 x Age[Yrs.], adjusted R2= 0.45 (P= 0.0020). An even better fit was received by a second degree polynomial fit (NT–proANP [pmol/l]= 165 +0.02089 x Age+ 0.00466x [Age–4.31]2), adjusted R2= 0.52 (P= 0.0022). However, the small numbers of dogs (n= 17) makes mathematical curve fitting unrewarding, but when data was partitioned into ages < 5.9 and $ 5.9, there was no overlap in NT-proANP values (Figure 13). Median NT–proANP for dogs younger than 5.9 years was 230 (190– 270 [IQR]) pmol/l and 405 (307– 530) pmol/l for dogs of age 5.9 years and older, respectively.
Although there was a significant positive association between plasma NOx and age, the inter-individual variation in plasma NOx compared with NT–proANP was high. The optimal partitioning threshold in this material was 5.8 years with corresponding mean NOx values of 12 (± 3.2 [SD], 10.1– 16[CI]) and 17.9 (±6.0, 12.3– 22.1)µmol/l. The values for age groups were overlapping and the difference was not significant (P= 0.14).
Figure 13. Panel A. Linear (solid) and polynomial (dotted) fit of N-terminal proatrial natriuretic peptide (NT-proANP) vs age in healthy Cavalier King Charles spaniels. Panel B. Box-plot (median and quartiles) of plasma NT-proANP categorized into age groups.
W@?@J*!D;,$%8!<*(-.*!E*1-*/&"*,$%2$"++1%-*%4*'1/$()*$"2#$21/(/1%-*Y<(,"$*IZ*
We followed 78 Cavalier King Charles spaniels (CKCS) for a maximum of 4.5 years during the progression from an innocent heart murmur to congestive heart failure (CHF) due to mitral regurgitation (MR). During this time span plasma levels of NT-proANP increased from 465± 49 pmol/l to 1530± 93 pmol/l for dogs reaching CHF.
For dogs that did not progress to signs of CHF ("censored") the plasma level of NT–
proANP at the last visit was 860± 81 pmol/l (P< 0.0001).
The hazard ratio (HR) for CHF calculated in the extended Cox proportional
The hazard ratio (HR) for CHF calculated in the extended Cox proportional