• Ei tuloksia

This work has shown that plasma concentrations of NT–proANP and NOx increase with age in healthy dogs and that a plasma concentration of NT-proANP above 1000 pmol/l is a predictor of congestive heart failure (CHF) in dogs with mitral regurgitation (MR) caused by myxomatous mitral valve degeneration (MMVD). In addition, both a regurgitant murmur of grade 3 out of 6 and above, and heart rate above 130 beats/min, is associated with an increased hazard of CHF.

Dogs ending up with CHF had a lower plasma level of NOx than dogs not reaching CHF (censored) during the follow-up period and the average plasma concentration increased before CHF. This difference was not affected by treatment with enalapril.

Heart rate normalized pulmonary transit times (nPTT) increase with degree of MR as an indicator of decreased heart pump function and that pulmonary blood volume (PBV) increase as a consequence of increased nPTT, whereas there is only an insignificant decrease in forward cardiac stroke volume not related to PBV.

The right heart chamber enlarge only very late in the progression of MR. Apparent right heart enlargement on radiographs is due to septal wall deformity caused by a volume-overloaded LV and the displaced right heart chambers cause the contour of the RV to bulge cranially. Therefore, the contour of the right heart is not a useful signs of right-sided chamber dilatation on thoracic radiographs in MR.

X@?*!(/$1#$"/13*,",/1."+*(-.*"-.%/&")1()*$"+,%-+"*Y<(,"$+*B*(-.*IZ*

One aim of this work was to determine whether there is an association between plasma natriuretic peptides (NP) levels and age as demonstrated in humans (Redfield, Rodeheffer et al. 2002, Wang, Larson et al. 2002). We showed a significant increase in plasma NT–proANP and NOx with age, but not with BNP. Changes were independent of echocardiographic indices.

The effect of natriuretic peptides (NPs) and nitric oxide (NO) is mediated via the same secondary messenger cGMP. This indicates the increase could be a response to decreased local concentration of cGMP with age (Kawai, Hata et al. 2004). We measured cGMP as urinary “spill-over” and observed a significant correlation to plasma NT–proANP and BNP, but not to NOx. The latter may be due to a large variability in NOx concentrations. However, the increase in cGMP, if relevant on a cellular level, and blood NP concentrations, which we found, contradicts a report in aging humans (Kawai, Hata et al. 2004) and implies that the physiologic response may be impaired. Another explanation could be that NP signal transduction mechanisms are not limited to cGMP (Ruskoaho 1992).

X@?@?*!1/$13*%U1."*Y,(,"$*B*(-.*IZ*

Our finding of increased plasma NOx concentrations with age and the inverse impact on HR (0.86) for CHF in the multivariate Cox model is in contrast to a previous study (Pedersen, Schutt et al. 2003). However, that study included only nine healthy control dogs and lacked statistical power to determinate possible increase with age and decreased NOx concentrations with degree of MR may have obscured a true association to age in diseased animals. Another explanation is related to the complex relationship and feedback mechanisms between NOx and other mediators.

Endothelial dysfunction decreases local capability of cell NO production. On the other hand, with maintained endothelial function, NO together with the NPs, is an important beneficial mediator of beneficial vasodilation. Therefore, the decreased hazard with increased baseline NOx concentrations indicates that the dogs in our study (Paper V) initially had normal endothelial function. This explanation seems plausible, because mean age at enrolment was only 6.6± 2.1 years, whereas arteriosclerotic changes, as a marker of endothelial dysfunction, reported in necropsy material were from older animals (median age 11 years) (Falk, Jönsson et al. 2006).

Plasma NOx level is influenced by gastrointestinal sources (digested protein, bacterial source, etc.). Although blood samples, both in our and the previous study (Pedersen, Schutt et al. 2003), were taken from fasted animals, we cannot exclude the possibility that animals could have ingested food, this contributing to the high variability in plasma NOx concentrations as compared to NT-proANP. Nevertheless, based on our results, the high variability in plasma and urine NOx measurements limit their value in clinical risk evaluation of progression of endothelial function and heart disease.

X@?@J*!(/$1#$"/13*,",/1."+*(-.*(2"*Y<(,"$*BZ*

We found that plasma concentration of NT-proANP increases with age in healthy dogs. The increase in NT–proANP based on linear regression was 25 pmol/l per year).

Despite a good linear correlation (R2= 0.48, P= 0.002) the visual inspection of the scatter plot implied that there was little increase in plasma NT-proANP concentrations before 6 years of age (Figure 13). After using partition methods statistical software showed that a cut-off age of 5.9 years produced no overlap in NT–proANP levels (Figure 13). This difference was not explained by differences in creatinine or urea values and the observed increase in urinary cGMP suggests that total production of NPs is increased with age. The apparent difference in NT–proANP levels in young (<

5.9 years) vs. older dogs indicates different normal values should be used for different age groups. The polynomial curve fit indicated a progressive increase with age (up to 50 pmol/year).

We failed to demonstrate a similar increase in BNP with age. One explanation for this could be the use of BNP instead of the more sensitive and stable NT–proBNP.

Little is known about degradation of the active BNP1–32 molecule in vivo in dogs.

Another explanation may be the use of porcine BNP1–26 antibodies instead of canine

specific BNP1–32. Although differences in molecules are in places theoretically not interfering with the analysis (Paper I) we cannot exclude an impact of this heterology.

Because MR is a chronic disease, dogs in CHF will be older than dogs with asymptomatic MR. Therefore a considerable part of reported increase in plasma NPs may be related to age, because normal plasma NT-proANP concentrations for a young to middle aged dog (230 (190– 270 [IQR])pmol/l) may double during the lifespan (Figure 13, page 49). On the other hand, a moderate increase (of the order of 200 pmol/l) in a young dog may indicate activation of the natriuretic pathway. In consequence, statistically significant differences in papers using cross-sectional statistical methods can be confounded by physiological changes in plasma NPs and results erroneous. Further studies are warranted in a broader material including several breeds to establish age-related values for natriuretic peptides.

Few canine studies have detected an association between NPs and age. The studies may not have been adequately powered or designed to demonstrate such a difference (Häggström, Hansson et al. 2000, Borgarelli, Zini et al. 2004, Chetboul, Serres et al.

2009, Tarnow, Olsen et al. 2009, Moonarmart, Boswood et al. 2010, Moesgaard, Falk et al. 2011). In a study comprising only Doberman Pinchers (Wess, Butz et al. 2011), apparently healthy dogs over 8 years old had higher plasma levels of NT–proBNP than younger.

Sex, weight and M-mode measurements were not associated with NT–proBNP values. However, this breed is prone to dilated cardiomyopathy (DCM) and there is no reliable way to exclude the possibility that presence of subclinical DCM may have biased the data. The association between age and NT–proBNP was later verified in a large study (n=1134), which included dogs of several breeds and size (Ettinger, Farace et al. 2012). That and our study demonstrate that plasma levels of NPs increase with age in dogs as they do in humans.

!"-#%&#'(%.'4,-&3"'/&'&-+./1."+/4'$"$+/5"'0"8"0#'D/+,'-3"'

Decreased renal function may increase blood NP levels both in humans (deFilippi, Seliger et al. 2007) and dogs (Raffan, Loureiro et al. 2009, Schmidt, Reynolds et al.

2009). Application of 3 recommended cut-off values led to misclassification of dogs with azotemia as having heart disease (Raffan, Loureiro et al. 2009) and NT–proBNP levels were associated with both blood urea and creatinine concentrations (Ettinger, Farace et al. 2012). Whether this association is directly related to glomerular filtration on NP levels in old dogs remains unclear. However, creatinine and blood urea nitrogen (BUN) values did not change with age in our material.

Although decreased renal function was associated with increased levels of NPs in dogs (Raffan, Loureiro et al. 2009, Schmidt, Reynolds et al. 2009), NP concentrations would not rise in advance to increased plasma creatinine levels. We therefore consider that decreased renal function is an unlikely cause of increased NT–proANP levels in older dogs used in this study. However, serum ß2-microglobulin, an endogenous marker of renal filtration function (Schardijn and Statius van Eps 1987), correlates

strongly with increased NP concentrations in older humans (Kawai, Hata et al. 2004).

This finding indicates preclinical renal failure may still have an impact on NP levels.

Decreased endothelial function may increase blood pressure thereby inducing atrial stretch, which release natriuretic peptides (Daniels and Maisel 2007, Lynch, Claas et al. 2009). The change in LA/Ao ratio with age was, however, minimal in our healthy CKCS dogs (Paper I). We did not measure blood pressure but clinically relevant hypertension is most unlikely to affect all dogs in our study group. Moreover, hypertension is not a common cause of increased NP concentrations in elderly humans (Wallen, Landahl et al. 1993). In an experimental model secretion of ANP was decreased in old rats in response to atrial stretch in comparison with young rats (Pollack, Skvorak et al. 1997). Together, these findings indicate that increased LA pressure and stretch are unlikely causes of increased concentrations of NT–proANP in old dogs. Other possible explanations include altered production, secretion, or metabolism of NPs.

NPR-C clearance receptor activity may be reduced with aging (Giannessi, Andreassi et al. 2001). This could be one reason for the increased NT-proANP with age in our healthy dogs (Paper I). However, human patients with chronic heart failure had an increase in platelet NPR-C receptor density (Andreassi, Del Ry et al. 2001), which potentially may decrease NT-proANP concentrations. This divergent trend in NPR-C receptor density could potentially attenuate differences between healthy dogs and dogs with MR and increase the overlap of NP concentrations between dogs with MR and healthy older ones.

Contrary to that, ß-adrenoreceptor agonists, with upregulated release in chronic heart failure, significantly decreased both NPR-C receptor densities and mRNA levels (Yoshimoto, Naruse et al. 1998), potentially decreasing degradation of NPs. Since our dogs were healthy without signs of heart disease, ß-activation is unlikely to be the cause of increased NP concentrations with age. However, with increased sympathomimetic tone in chronic MR plasma levels of NPs may increase due to decreased degradation. Little is known about the activation time-course and net balance of aforementioned mechanisms but the may be a reason to the low power of NPs in cross-sectional studies to determine severity of MR in the period before CHF.

Taken together, reports on the cause of elevated NP concentrations with age are far from unanimous but are most likely related to decreased degradation, renal clearance and possibly increased production (Raizada, Thakore et al. 2001) in response to attenuation of local action due to modulations in receptors and second messengers (Kawai, Hata et al. 2004, Daniels and Maisel 2007).

E+,".'8-./-?0"#'-##%4/-+"5'D/+,'&-+./1."+/4'$"$+/5"#'

Blood NP concentrations in humans correlate with female gender and inversely with increased body mass index (BMI) (Wang, Larson et al. 2002, Daniels and Maisel 2007). The reason for higher levels in women is unknown, but oestrogen may play a role. Our material (Paper I) did not have the power to detect a possible gender difference (n=17), but we observed an inverse association between weight and urinary

cGMP (P= 0.01) and plasma NT–proANP (P= 0.035). Both associations were significant when controlling for age. Although we did not estimate BMI, a higher weight within the same breed should correlate to BMI. Our finding is therefore in agreement with previous reports in human, though the reason for this inverse relationship is not fully understood (Daniels, Clopton et al. 2006, Daniels and Maisel 2007).

X@?@M*!(/$1#$"/13*,",/1."+*1-*,$%2$"++16"*'1/$()*$"2#$21/(/1%-*Y<(,"$*IZ*

Increased blood NT–proANP levels in CKCS dogs were related to a significant increase in risk of CHF with naturally progressing MR due to MMVD. The hazard ratio (HR) was 1.21 for every 100 pmol/l increase in NT–proANP (P< 0.001).

Moreover, the increase in hazard (i.e. hazard rate) was not linear, but increased steeply before 1000 pmol/l, whereafter the curve stabilized. The time to CHF was shorter for dogs with plasma NT-ANP higher than 1000 pmol/l (median 11 months) than for dogs with levels below 1000 pmol/l (median 54 months) (Paper V, Figure 1).

Previous reports have showed diagnostic value of NT–proANP (and BNP or NT–

proBNP) in dogs with MR in differentiating dogs in or not yet in failure (Häggström, Hansson et al. 1997, Häggström, Hansson et al. 2000, MacDonald, Kittleson et al.

2003, Moesgaard, Pedersen et al. 2005, Boswood, Dukes-McEwan et al. 2008, Chetboul, Serres et al. 2009, Oyama, Rush et al. 2009, Serres, Pouchelon et al. 2009, Tarnow, Olsen et al. 2009, Moonarmart, Boswood et al. 2010, Hori, Yamano et al.

2011, Moesgaard, Falk et al. 2011, Wess, Butz et al. 2011, Ettinger, Farace et al. 2012, Reynolds, Brown et al. 2012, Wolf, Gerlach et al. 2012). However, reports on the usefulness of NPs to predict future failure are inconsistent. Some studies have suggested that the natriuretic peptides have potential for the prediction of outcome, with elevated concentrations of ANP (Greco, Biller et al. 2003), BNP (MacDonald, Kittleson et al. 2003) and NT-proBNP (Serres, Pouchelon et al. 2009) shown to be predictive of a worse out- come, though these studies have some limitations. They only included univariable analyses with either relatively limited numbers of dogs (Greco, Biller et al. 2003, MacDonald, Kittleson et al. 2003) or relatively short periods of follow-up (Serres, Pouchelon et al. 2009).

In one study NT–proANP could differentiate dogs between modified NYHA classes I-IV, whereas BNP could differentiate only dogs with or without failure (Häggström, Hansson et al. 2000). In another study comparing MR jet size to NP levels, NPs could predict further progression in MR. In the receiver operating characteristic (ROC) analyses, NT-proBNP slightly out-performed NT-proANP (Tarnow, Olsen et al. 2009). However, specificity was low for both peptides and the authors suggested NPs can be used primarily as a rule-in diagnosis primarily in dogs with MR.

NT-proBNP could predict Doppler derived peak tricuspid regurgitation, as a sign of precapillary PH, whereas NT-proANP could not (Kellihan, Mackie et al. 2011).

However, their conclusion appears unjustified because the number of dogs in different

groups was only three to six, making statistical comparisons underpowered.

Nevertheless, there was graphical evidence that both NT-proANP and NT-proBNP concentrations increase with degree of PH.

In a study assessing an increase in plasma BNP increased with mortality rate over 4 months time by approximately 44% for every 10-pg/ml increase in plasma BNP (MacDonald, Kittleson et al. 2003). That study comprised only 25 dogs with MMVD, either 9 or 15 of them reaching CHF depending on which of the two inconsistent classification criteria were used. One cross-sectional study did report significant differences in plasma NT–proBNP levels between grades of cardiac disease (Ettinger, Farace et al. 2012). However, the methods section of this paper state that statistical tests between groups were performed by analysis of variance, whereas in the results significant differences between two groups were determined using t-tests. Moreover, since the authors used a different grading of cardiac disease than is commonly used (see Table 1), these results cannot be compared to other reports.

One study has reported predictive value of survival of lowered NT–proBNP, but not proANP31-67 (a section of the NT-proANP1-98 molecule), in response to treatment of MR dogs (P= 0.04 and 0.3, respectively; n= 11) (Wolf, Gerlach et al. 2012). We could not detect an effect by ACE inhibition (enalapril) on the time to CHF in MR dogs (Paper V). However, the designs of the two studies were different. Dogs in our study received prophylactic treatment, whereas dogs in the other study received effective treatment for signs of CHF caused by MR. The sample size in that study was also too small to provide evidence that NT–proBNP was superior to proANP31-67 for predicting survival. Furthermore, they used an antibody segment against proANP, which was different from ours (proANP79-98). Molecular heterogeneity of the circulating forms has been reported to cause a serious risk of preanalytical errors in assays for NT-proBNP and, to a lesser extent, NT- proANP. The most robust and reliable assays use antibodies directed at the central portions of proANP or NT-proBNP (Ala-Kopsala, Magga et al. 2004).

Several other studies have evaluated the usefulness of NT–proBNP as predictor of heart failure. Serres and others (2009), using a multivariate analysis, showed that NT-proBNP was an independent predictor of death within 6 months in patients with symptomatic MMVD. Moonarmart and others (2010) reported that NT-proBNP was a marker of increased all cause mortality in dogs with asymptomatic and symptomatic MMVD. However, none of these studies applied longitudinal statistical methods.

In the PREDICT study (Reynolds, Brown et al. 2012) a regression formula including LA/Ao and NT-proBNP was proposed for predicting risk of developing CHF in MR dogs. However, both these studies used cross-sectional statistical methods and logistic regression (which cannot account for time) to assess risk, therefore unfortunately limiting the value of the report. Moreover, given that we detected a significant and escalating increase in NT-proANP with age (unrelated to heart disease), and that none of the aforementioned studies accounted for the effect of age on NPs, the reports on both the discriminative and the predictive value of NPs in dogs with MR are biased or at least incomplete.

To our knowledge, we are the first to apply extended longitudinal statistical methods in companion animal cardiac research. Extended models can include unevenly distributed time points, as well as defined common time points to all individuals, and account for change of a variable by time for the same individual. In contrast, simple or ordinary models (e.g. ordinary Cox regression models) do not handle time-dependent (changing with time) covariates (or are confounded by them) and all individuals must share time points. With an extended Cox model we could detect the predictive value of NT-proANP (HR= 1.7 !100 pmol/l, P= 0.0002) for CHF in MR dogs. Because the model corrects for inter-individual change it is robust for confounding factors, including age. Furthermore, NT-proANP predicted time to CHF (median 11 months when NT-proANP< 1000 pmol/l) when assessed by methods utilizing the fitted extended Cox model.

X@?@N*D"3&-13()*(+,"3/+*%4*-(/$1#$"/13*,",/1."*'"(+#$"'"-/+*

At the initiation of this work commercial RIA kits for measurement of canine BNP were not available. We therefore used previously validated methods with porcine BNP antibodies raised in rabbits. Although divergent end nucleic acids sequences in canine and porcine BNP theoretically should not have interfered with the analysis, results were not in conformance with NT-proANP, as we had expected. The variability in the results may relate to nonspecific cleavage of BNP by plasma proteases (Clerico, Fontana et al. 2007). Moreover, although blood samples were drawn on ice and deep frozen, nonsiliconized glass tubes were used for some samples. Blood kallikrein is reported to become activated by glass and rapidly degrade BNP but not NT-proBNP (Ordonez-Llanos, Collinson et al. 2008).

Measurement of the proANP36-67 in EDTA plasma gave significantly higher peptide concentrations than corresponding serum samples from dogs, whereas there was no difference in NT-proBNP levels between tubes (Boswood, Dukes-McEwan et al. 2008). This may be due to the nonproteolytic properties of EDTA, but authors did not report whether they used plastic or glass tubes and whether glass tubes (if used) were siliconized. Although not proven, we cannot exclude that part of the variance in our year to year intra-individual levels of BNP may have been due to differences in in vitro degradation.

X@J* <#)'%-($0* /$(-+1/* /1'"* (-.* S)%%.* 6%)#'"* 1-* '1/$()* 1-+#44131"-30*

Y,(,"$+*BB*(-.*BIZ*

X@J@?*8++"++'"-/*%4*,#)'%-($0*S)%%.*/$(-+1/*/1'"*(-.*6%)#'"*

Pulmonary transit time normalized to heart rate (nPTT) increased with severity of MR (Paper II). Normalized PTT is a measure of the pulmonary blood volume (PBV) to stroke volume ratio (PBV/SV)[Formula 2] (Milnor, Jose et al. 1960, Hannan, Vojacek et al. 1981). Heart rate normalized PTT is one of few measures not affected by e.g.

size, heart rate, or even exercise (Lewis, Gnoj et al. 1970, Kuikka, Timisjarvi et al.

1979, Okada, Pohost et al. 1979, van der Walt, Van Rooyen et al. 1981, Thorvaldson, Ilebekk et al. 1984, van Rooyen and van der Walt 1989).

Theoretically, nPTT should be associated with increases in pulmonary vascular pressures, which in turn precedes development of CHF. Consequently, an increase in

Theoretically, nPTT should be associated with increases in pulmonary vascular pressures, which in turn precedes development of CHF. Consequently, an increase in