Heart and Circulatory Physiology

Effects of rosuvastatin on cardiovascular morphology and function in an ApoE-knockout mouse model of atherosclerosis

J. Grönros, J. Wikström, U. Brandt-Eliasson, G. B. Forsberg, M. Behrendt, G. I. Hansson, L. M. Gan


This study investigated the effects of rosuvastatin on plaque progression and in vivo coronary artery function in apolipoprotein E-knockout (ApoE-KO) mice, using noninvasive high-resolution ultrasound techniques. Eight-week-old male ApoE-KO mice (n = 20) were fed a high-fat diet with or without rosuvastatin (10 μmol·kg−1·day−1) for 16 wk. When compared with control, rosuvastatin reduced total cholesterol levels (P < 0.05) and caused significant retardation of lesion progression in the brachiocephalic artery, as visualized in vivo using an ultrasound biomicroscope (P < 0.05). Histological analysis confirmed the reduction of brachiocephalic atherosclerosis and also revealed an increase in collagen content in the statin-treated group (P < 0.05). Coronary volumetric flow was measured by simultaneous recording of Doppler velocity signals and left coronary artery morphology before and during adenosine infusion. The hyperemic flow in response to adenosine was significantly greater in left coronary artery following 16 wk of rosuvastatin treatment (P < 0.001), whereas the baseline flow was similar in both groups. In conclusion, rosuvastatin reduced brachiocephalic artery atherosclerotic plaques in ApoE-KO mice. Coronary artery function assessed using recently developed in vivo ultrasound-based protocols, also improved.

  • apolipoprotein E-knockout mouse
  • ultrasound imaging
  • atherosclerosis
  • rosuvastatin

the inhibition of 3-hydroxy-3-methyl-glutaryl coenzyme A (HMG-CoA) reductase using statins has become a standard treatment regimen in patients with atherosclerosis. In addition to their cholesterol-lowering effects (11, 20), statins have been shown to retard the progression of atherosclerosis (36), improve endothelial function (37, 45), reduce systemic inflammatory markers (32, 41), and reduce cardiovascular mortality and morbidity (1). Recently, the clinical regression of coronary artery atherosclerosis was reported with the use of a third-generation statin (35).

Despite the success of statins, cardiovascular mortality and morbidity still remain high in developed countries. When one explores new therapeutic strategies to treat atherosclerotic cardiovascular disease, it is increasingly important to demonstrate the beneficial antiatherosclerotic effects on top of the cholesterol-lowering effect of statin as well as potential pleiotrophic effects. In line with this approach, atherosclerotic animal models that can respond to currently available statin treatments are key for translational research.

The effects of various statins in atherosclerotic mice are inconclusive. Differences in genetic modifications, sex, diet, as well as differences in the statins themselves, have produced different results (55). Although first and second generation statins gave variable and sometimes contradictory data in mice (5, 50), several investigators demonstrated antiatherosclerotic effects with concomitant lipid-lowering effects using atorvastatin and rosuvastatin (3, 9).

Both acute and chronic statin treatment improve coronary flow velocity (CFR) reserve in humans, indicating enhanced coronary artery function (15, 17). Interestingly, CFR can be improved just 1 h following intravenous statin infusion, clearly demonstrating a lipid-independent mechanism of action (18).

Using a high-resolution ultrasound biomicroscope (UBM), we are now able to follow mouse atherosclerotic plaque growth over time in a noninvasive manner (16). In addition, mouse coronary artery function can be studied in vivo using color-Doppler echocardiography using a recently developed protocol (52).

In the present study, we studied the effects of rosuvastatin on lesion progression using in vivo imaging in an apolipoprotein E-knockout mouse model (ApoE-KO). The effect of treatment on coronary artery function was also examined in vivo.



Twenty 8-wk-old male ApoE-KO mice on a mixed background of 129SvJ and C57BL/6 (Taconic, Denmark) were fed a high-fat diet (21% pork lard, 0.15% cholesterol; SDS, Witham, England) with or without an additional dose of rosuvastatin mixed in the diet (10 μmol·kg−1·day−1, AstraZeneca R&D, Sweden, n = 10 in each group) for 16 wk.

A separate group of 18–26-wk-old ApoE-KO mice (n = 30) fed the same diet was used for the validation of ultrasound accuracy. All mice had free access to tap water and chow. The Regional Animal Ethics Committee, Göteborg University, approved the experimental protocol.

Blood samples and plasma analysis.

Blood samples were collected from the saphenous vein before the diet was started, and after 4, 8, 12, and 16 wk of treatment, samples were also obtained via heart puncture under 1.5% isoflurane anesthesia (Forene isoflurane, Abbott Scandinavia) at the study end. The samples were placed in lithium heparin-containing tubes and centrifuged at 4°C at 2,000 g for 10 min. The plasma was then stored at −80°C until analysis. Total cholesterol levels were analyzed continuously during the study using commercial reagents (Cat. No. 12016630, Roche Diagnostics, Mannheim, Germany) and Cobas Mira Plus (Roche Diagnostics) with enzymatic-spectrophotometric methods. The levels of the acute inflammatory parameter serum amyloid A (SAA) at death were analyzed using a Murine Serum Amyloid Enzyme Immunoassay kit (Cat. No. TP802-M, Tridelta Development).

Color-Doppler echocardiography: coronary artery function and cardiac function.

After 16 wk of treatment, a 15-MHz linear transducer (Entos CL15-7, Philips Medical Systems, Bothell, Washington) connected to an ultrasound system (ATL-HD15000, Philips Medical Systems) was used for echocardiographic measurements. A 6-MHz pulsed Doppler with a gate size of 0.5–1 mm was used for the coronary flow velocity measurements.

For the measurement of coronary artery hyperemic volumetric flow, a vasodilating substance, adenosine (ITEM Development, Stocksund, Sweden), was infused at the rate of 140 μg·kg−1·min−1 using a catheter (0.4, 27 gauge, Sa-TE-Z Set, Temse, Belgium) in the tail vein connected to a syringe and infusion pump (syringe pump 22, Harvard Apparatus, Holliston, MA). Coronary artery volumetric flow was calculated during baseline and hyperemia and has previously been described by our group (see appendix for formulas) (53). All measurements were made off-line using the Image Arena software analysis program (Image Arena 2.9.1, Tomtec Imaging Systems, Unterschleissheim, Germany). Cardiac measurements were made in accordance with guidelines from the American Society of Echocardiography (42). Stroke volume and the end-diastolic volume were calculated using the cubic formula, and ejection fraction, fractional shortening, and cardiac output were calculated from previously validated formulas (27). Left ventricular mass was calculated according to the area-length formula (7).

UBM protocol and analysis.

Before the diet was started and after 4, 8, 12, and 16 wk of treatment, a UBM (Vevo 770, version 2.2.2, Visualsonics, Toronto, Canada) with a transducer frequency of 40 MHz (providing a theoretical resolution of 40 μm at a frame rate of 32 Hz) was used for vascular imaging. The brachiocephalic artery and ascending aorta were visualized in two different projections, a long-axis view and a cross-sectional short-axis view in line with previously published protocol (16). According to the published data, for measurements of aortic intima media thickness, intra- and interobserver variability are 4.7% and 6.7%, respectively. For plaque area measurements, the corresponding values are 8.2% and 10%, respectively. To minimize the user dependency, all measurements were performed by one operator blinded to the animal identity and treatment. The left coronary artery (LCA) was visualized with the probe in a laterally tilted short-axis view of the aorta. Loops of at least 100 frames were stored in each projection. The frame with the largest plaque in each projection was chosen for off-line, blinded measurements.

Validation of UBM accuracy.

The additional group of ApoE mice, aged 18–26 wk, was used for the validation of UBM accuracy. The brachiocephalic artery lesion was measured within the proximal 200 μm of the vessel from its bifurcational site from the aortic arch. The histological lesion size was averaged from three consecutive cross-sectional slices starting from the first slice without visible aortic tissue.

Lipid extraction and fractionation.

Frozen liver tissues were homogenized in methanol, and the protein concentration was measured as previously described (2). Liver lipid extraction was performed with methanol and chloroform (13). A normal-phase HPLC system was used for lipid class separation according to Homan et al. (19) with slight modifications. Lipid classes were detected using an evaporative light-scattering detector (PL-ELS 1000, Polymer Laboratories, Amherst, MA). Calibration curves for detector response versus mass of lipid were obtained by injecting lipid standards. The results are presented as mass of lipid corrected for the amount of protein recovered.

Histology and immunohistochemistry.

At the end of the study, the heart, aorta, and brachiocephalic arteries were dissected and fixed in formaldehyde before being embedded in paraffin. Serial cross-sections of 5 μm were made using a microtome (RM2255, Leica) at one site on the aorta and at three different sites on the brachiocephalic artery (proximal and midbrachiocephalic and more distally at the bifurcation with the subclavian artery) and placed on TechMate slides (ChemMate Capillary Gap Microscope Slide S2024, DakoCytomation). Sections were stained with Picro Sirius red (Histocenter, Gothenburg, Sweden) for cross-sectional plaque area measurements and collagen content quantification. For macrophage accumulation, the sections were stained with a Mac-2 antibody at a dilution of 1:15,000 (CL8942AP clone M3/38, Cedarlane). Quantitative measurements of collagen and macrophage content as a percentage of plaque area were made using Image-Pro Plus 5.1.2 software (MediaCybernetics, Bethesda, MD).

Statistical analysis.

Data are presented as means ± SE. Differences in total cholesterol levels and UBM-assessed lesion size between treated and control groups over time were tested using repeated-measurement two-way ANOVA. Comparisons between each time point were performed using the Student's t-tests only when the two-way ANOVA tests showed significance. A P value < 0.05 at the 95% confidence interval was considered to be statistically significant.

For statistical analysis of the basal and hyperemic flow measurements at the end of the study, i.e., for the Student's t-test with an adjustment for two comparisons, a P value < 0.025 (0.05 of 2), was considered significant. Bland-Altman analysis was used for the validation of UBM and histology-assessed measurements of plaque area in the brachiocephalic artery. Linear regression analysis was used for correlations between two parameters. All statistical evaluations were made using Pris 4 software (GraphPad, San Diego, CA).


UBM visualization of atherosclerotic lesions in the brachiocephalic artery.

The validation study revealed good correlation between cross-sectional plaque areas in the proximal part of the brachiocephalic artery as determined by UBM and histology (P < 0.001, r = 0.59, Fig. 1). Bland-Altman analysis demonstrated a slight underestimation of lesion area by UBM; the discrepancies were uniform throughout the range of lesion size (Fig. 2).

Fig. 1.

Correlation between brachiocephalic artery atherosclerotic lesions measured with an ultrasound biomicroscope (UBM) compared with histology measurements (P < 0.001, r = 0.59).

Fig. 2.

Bland-Altman plot comparison (difference vs. average) of UBM and histology measurements of the brachiocephalic artery plaque area.

Effects of rosuvastatin on plasma total cholesterol, hepatic cholesterol ester, and triglyceride levels.

Total cholesterol levels were not different between the rosuvastatin and control groups of ApoE-KO mice before the onset of the high-fat diet [11.6 ± 0.5 mM and 11.8 ± 0.9 mM, respectively; P = not significant (NS)]. After the commencement of the high-fat diet, the plasma cholesterol levels were markedly increased in control animals. After 4 wk of treatment and until euthanasia at 16 wk of treatment, the total cholesterol levels were reduced in the rosuvastatin-treated group (P < 0.05, Fig. 3). Over time, the treated group showed a reduction of 22% at 4 wk (P < 0.001), 18% at 8 wk (P = NS), 27% at 12 wk (P < 0.001), and 23% at 16 wk of treatment (P < 0.0001) compared with the control group. At the study end, the amount of hepatic cholesterol ester was significantly decreased in rosuvastatin-treated mice (45.3 ± 6.6 μg/mg protein) compared with controls (84.3 ± 12.2 μg/mg protein; P < 0.05), together with lowered triglyceride content (68.6 ± 9.8 μg/mg protein in the treated group compared with 192 ± 42.6 μg/mg protein in controls; P < 0.01).

Fig. 3.

Total cholesterol levels over time in control and rosuvastatin-treated apolipoprotein E-knockout mice. Pre, previous to diet onset; w, week; NS, not significant. ***P < 0.001.

Body weight differed significantly between the controls and the rosuvastatin-treated group at the study end (30.9 ± 1.2 and 26.9 ± 0.8 g, respectively, P < 0.05), despite no significant difference in daily chow intake.

Effects of rosuvastatin on serum inflammatory biomarkers.

At the end of the 16-wk treatment period, the levels of SAA were significantly reduced (P < 0.001) in the rosuvastatin-treated group (32 ± 4 μg/ml plasma) compared with controls (125 ± 18 μg/ml plasma).

Effects of rosuvastatin on coronary artery function.

Calculated coronary volumetric flow in the LCA was increased by rosuvastatin during hyperemia but not at baseline (P < 0.001, P = NS, respectively, Figs. 4 and 5). The proximal mean flow velocity in the LCA was significantly increased during hyperemia in the rosuvastatin-treated group (76.1 ± 7.2 m/s) compared with controls (48.6 ± 3.6 m/s; P < 0.01). The change in the calculated average LCA diameter between the two groups at baseline (controls, 240 ± 6 μm; and treated, 260 ± 12 μm) and during hyperemia (controls, 270 ± 7 μm; and treated, 270 ± 13 μm) was nonsignificant after 16 wk of treatment.

Fig. 4.

Flow per minute (in μl/min) in the proximal left coronary artery (LCA) calculated from UBM-measured diameter and color-Doppler-guided flow velocity. **P < 0.01.

Fig. 5.

Illustration of the LCA images for the flow measurements assessed with UBM during baseline (A) and hyperemia (B). C and D: the flow velocity profile in baseline (C) and hyperemia (D) measured with color-Doppler echocardiography. Ao, aorta. Scale bar = 1 mm.

There was no significant difference in cardiac parameters between the two groups (Table 1).

View this table:
Table 1.

Cardiac parameters at 16 wk for control mice compared with the rosuvastatin-treated apolipoprotein E-knockout mice

Effects of rosuvastatin on plaque progression in the brachiocephalic artery.

Both the long-axis and short-axis view projections from the UBM revealed a significant reduction in plaque progression in the rosuvastatin-treated mice over time (P < 0.001 and P < 0.05, respectively, Figs. 6, A and B, and 7). The effect of rosuvastatin treatment on lesion size in the brachiocephalic artery was seen after 8 wk and was sustained until the 16-wk end point. Histology confirmed these results at the end point, indicating that lesion size was reduced in the treated group close to the bifurcation of the brachiocephalic and subclavian artery (P < 0.05). The aortic plaque lesion size measured with UBM was significantly reduced at the end of the study after 16 wk of treatment with rosuvastatin (0.27 ± 0.02 mm2) compared with controls (0.36 ± 0.04 mm2, P < 0.05).

Fig. 6.

A and B: progression curve of atherosclerosis in the brachiocephalic (Br) artery measured in a long-axis view (LAX; A) and a short-axis view (SAX; B) using UBM before the start of a high-fat diet (Pre) and after the indicated number of weeks on the diet. *P < 0.05; ***P < 0.001.

Fig. 7.

Typical recordings of LAX (A and B) and SAX (C and D) images of the ascending aorta and brachiocephalic artery captured with UBM with the atherosclerotic plaques delineated. Scale bar = 1 mm.

Total cholesterol levels correlate with brachiocephalic plaque area.

Instead of correlating cholesterol values at a certain time point to plaque burden, the exposure over time was correlated to plaque burden, since the model is highly cholesterol driven. When all values were pooled in each animal, the total area under the curve for cholesterol correlated significantly with the plaque area measured from brachiocephalic histology sections (P < 0.05, r = 0.57).

Effect of rosuvastatin on brachiocephalic artery lesion composition.

Histology results from the end point of the study showed an increase in collagen content (percentage of plaque area) in sections from the brachiocephalic artery in the statin-treated group compared with the controls (55.7 ± 2.5% vs. 45.8 ± 3.1%; P < 0.05). The macrophage content (percentage of plaque area) in the brachiocephalic artery plaque was similar between the groups (P = NS).


This study showed that ApoE-KO mice treated with rosuvastatin had reduced plasma total cholesterol levels and reduced brachiocephalic artery atherosclerotic lesion progression. Plaque reduction was evident after 8 wk of treatment and until the study end at 16 wk. It was accompanied by a total cholesterol-lowering effect that was detectable after 4 wk of treatment. In addition, we observed improved coronary volumetric hyperemic flow in the LCA, as well as an anti-inflammatory effect due to a decrease in SAA in mice treated with rosuvastatin.

Contradictory results have been reported regarding the response to statin treatment in various atherosclerosis-prone mouse strains. One explanation for the inconsistent results might be related to the physical properties of the different generations of statins, like tissue permeability and the ability to bind to and inhibit the enzyme HMG-CoA reductase (26, 43). Other possible causes may depend on the sex, the genetic background of the knockout mice, the lipid type, the cholesterol content of their diet, and the vascular site of investigation (40, 47). Zadelaar et al. (55) recently published a review describing the most commonly used mouse models in atherosclerotic research and the response to different pharmaceutical treatments. The review states that there is no consensus about which mouse model responds to statins. Results similar to ours have been observed in ApoE-KO mice on a high-cholesterol diet treated with rosuvastatin. These studies showed both a plasma lipid-lowering effect and reduced lesion formation after 12 wk of treatment (3, 4). Simvastatin, on the other hand, showed no positive effect in ApoE-KO mice after 3 mo of treatment but lowered the total cholesterol levels and reduced the atherosclerotic lesions in low-density lipoprotein receptor-knockout mice (49, 50). Atorvastatin showed the positive effects on atherosclerotic lesion progression and cholesterol levels in female ApoE3*Leiden mice after 16 and 28 wk of treatment (9, 47a). On the other hand, there are also data demonstrating an atherosclerotic plaque reduction and an anti-inflammatory effect of rosuvastatin, independent of its lipid-lowering effect (23, 33). In our case, liver lipid analysis showed clearly that the rosuvastatin treatment decreased cholesterol ester and triglyceride content, which are the major components of very low-density lipoproteins (VLDLs). This is most likely the result of the reduced cholesterol synthesis through the inhibition of HMG-CoA reductase through rosuvastatin. In addition, Delsing et al. (10) proposed that rosuvastatin acts through the inhibition of liver VLDL production and enhancing hepatobiliary lipid excretion (10). Interestingly, there are indeed indications of hepatic VLDL production reduction following statin treatment in humans (38). Parini et al. (38) further suggest that statin treatment inhibits hepatic acetyl-coenzyme A acetyltransferase 2 (ACAT2) activity, which may explain the decreased liver cholesterol ester content in our mice. Statin-treatment can additionally also inhibit systemic ACAT2 activity, especially in intestines, which would decrease both the uptake from the intestines and the remnant particle formation. These suggested mechanisms add speculative explanations to our findings regarding the lowered plasma total cholesterol together with a decrease in the hepatic lipid content in our rosuvastatin-treated mice. The mechanism behind the reduced body weight of the rosuvastatin-treated mice, despite no difference in daily chow intake, is still unclear. However, one may consider that a statin-induced change in intestinal lipoprotein and fatty acids secretion and/or synthesis pattern might have an effect on satiety and lipid oxidation. This intestinal lipid sensing hypothesis has recently been suggested by others (28, 29, 48). In the present study, we show that the total cholesterol expression over time correlates significantly with the plaque burden in the brachiocephalic artery as measured by histology. This provides evidence that the plaque regression seen in our study is dependent on the cholesterol-lowering effect from rosuvastatin treatment rather than being a result of another effect of the medication, although additional effects that are independent of cholesterol lowering cannot be ruled out.

Rosuvastatin-treated mice had greater hyperemic flow in the LCA following adenosine infusion compared with control mice. We previously reported that adenosine primarily effects the distal microvessels of the heart; this effect in turn lowers the resistance of the myocardium and increases the flow as measured with ultrasound (52). Similar changes in coronary artery diameter in treated and control mice during hyperemia indicate a true increase in flow rather than just an alteration in flow velocity. This further indicates improved microcirculatory function after rosuvastatin treatment in these mice. Our data partially concur with previous studies showing improved ex vivo artery function in the hearts and aortas of ApoE-KO mice treated with rosuvastatin, despite unchanged cholesterol levels in these mice (39). Similar observations have been made in human trials in which both short- and moderate long-term statin therapy improved coronary flow and coronary flow velocity reserve (CFR) (1a, 15, 18). The mechanism underlying the improved vessel function seems to be an enhanced nitric oxide release from endothelial cells following statin treatment, which has been shown in both cell-based and animal studies (25, 44). Hematological parameters such as hematocrit and viscosity also play a role in coronary flow reserve in healthy humans (46), and patients with biopsy-proven inflammatory infiltrates in the myocardium had decreased coronary flow velocity reserve (24). Interestingly, improved coronary function is observed in humans after only 1 h of statin infusion (18). This demonstrates that statins have an effect on the microvasculature that is independent of a cholesterol-lowering effect, which could not occur during that time frame (1 h). Thus both coronary flow and CFR could be useful as integrated physiological markers for coronary artery disease since they reflect several of the major pathogenetic risk factors for atherosclerosis (inflammation, endothelial function, microvascular function, and blood viscosity). Indeed, a recent study by Cortigiani et al. (8) showed that CFR is a powerful independent indicator of cardiovascular events in both diabetic and nondiabetic patients.

The present study revealed a reduction in atherosclerotic lesion size in the brachiocephalic artery, but not in the aortic arch, following treatment with rosuvastatin. The atherosclerotic process is normally initiated at sites where there is a disturbed flow pattern, such as at bifurcations, where curves in the vessels are more exposed. The disturbed flow patterns enhance the gene expression of adhesion receptors on the endothelial surface, increasing the recruitment of inflammatory cells from the circulation to the arterial wall (47). VanderLaan et al. (47) also identify the coronary arteries as the most important vascular bed to study from a clinical point of view. However, because the small size of the mouse heart makes this site difficult to study, the brachiocephalic artery is also mentioned as an important site to study in mice (54). The lesions in the brachiocephalic artery of mice mimic many of the important features of lesions in the human carotid artery, which have been shown to be important predictors of coronary artery disease (22, 30). In our study, we observed an increase in the collagen content of the histological sections from the brachiocephalic artery in the treated group. This indicates that the rosuvastatin-treated mice express a more stable phenotype at this artery site. These findings are consistent with a previous study showing reduced ruptures in brachiocephalic artery plaques after pravastatin treatment of ApoE-KO mice (21).

Ultrasound and magnetic resonance imaging are two rapidly developing techniques, both for in vivo and ex vivo cardiovascular imaging in mice (12, 14, 31, 34, 56). For fast and reproducible measurements, in vivo high-resolution ultrasound provides an accurate tool not only for morphological measurements but also for investigating the effects of treatment over time. To the best of our knowledge, this is the first study validating brachiocephalic artery lesion imaging using UBM.

In conclusion, we observed retarded lesion progression in the brachiocephalic artery of ApoE-KO mice following treatment with rosuvastatin. Coronary artery function improved in vivo after 16 wk of rosuvastatin treatment with a concomitant decrease in plasma total cholesterol and SAA. The male ApoE-KO mouse treated with rosuvastatin thereby provides us with a robust animal model with a statin-responding phenotype, which may facilitate future translational drug intervention studies in mice.


For flow calculations the following formulas were used: LCA diameter = LCA segment area/LCA segment length, and LCA area = π(LCA diameter/2)2.

Flow (μl/min) was calculated both in baseline and in hyperemia as: Flow = (velocity time integral × LCA area × heart rate).


This work was supported by grants from The Swedish Medical Research Council, the Swedish Heart-Lung Foundation, AstraZeneca R&D Sweden.


J. Wikström, U. Brandt-Eliasson, G. B. Forsberg, M. Behrendt, G. I. Hansson, and L. M. Gan are employed by AstraZeneca R&D (Mölndal, Sweden).


We give special thanks to Mia Umaerus, Kim Ekroos, Marcus Ståhlman, and Anders Elmgren at AstraZeneca Research and Development (R&D) (Mölndal, Sweden) for helpful discussions and analysis.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


  1. 1.
  2. 1a.
  3. 2.
  4. 3.
  5. 4.
  6. 5.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 47a.
  49. 48.
  50. 49.
  51. 50.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
View Abstract