METHODS

Stents

The BiodivYsio stent used in this study is 15 mm long and laser cut from a tube of 316L electropolished stainless steel with an “interlocking arrowhead” design incorporating S shapes for flexibility. It is coated with a phosphorylcholine containing polymer and is premounted on a 3.5 mm balloon. The coating has been shown by us to have no effect on neointima formation in this model.¹¹

Stent implantation

The experiments were performed in accordance with the UK Animals (Scientific Procedures) Act, 1986. Yorkshire White pigs (15–20 kg) were used. General anaesthesia was induced with azaperone 12 mg/kg and maintained with inhaled enflurane 5% in oxygen using an endotracheal tube. Vascular access was gained through the carotid artery. Under fluoroscopic guidance, with ECG monitoring, a 6 French Amplatz right coronary guide catheter was used to intubate both coronary arteries. Quantitative angiography was performed in each of the right and left coronary arteries. Stent implantation sites were chosen so that the balloon to artery ratio was 1.25:1 (a stent mounted on a 3.5 mm balloon placed in a 2.8 mm segment of artery). Balloon inflation was at 8 atm for 30 seconds (to produce a moderate injury). Heparin (2500 U) was given intravenously before implantation, and oral aspirin 150 mg orally for five days, starting on the day of implantation. An angiogram was obtained at the end of the procedure to document arterial patency.

Processing of stented arteries

The animals were killed by an intravenous injection of sodium pentobarbitone (pentobarbital) at 28 days. The hearts were explanted and the stented segments of artery flushed with saline, removed, and immersion fixed in 4% buffered formalin for 24 hours. The stented arteries were cut, ground, and polished using methods published by our group, in which the metal remains in situ and the tissue is undamaged, enabling accurate assessment of tissue morphometry without tissue damage.¹² Multiple sections were cut from each stent. Up to three sections were selected from each stent—from the proximal, middle, and distal thirds. Other sections were rejected on the grounds of an incomplete ring of struts, a large branch, distortion, tissue loss, or proximity to the end of the stent. Two or three good quality representative sections were thereby obtained from each stent.

Measurement of deep injury

Two measures of deep injury were made: rupture of the internal elastic lamina, and rupture of the external elastic lamina (the latter implying rupture of the media).

Injury scores

Two previously described injury scores were measured: the “deep injury” Schwartz score,¹³ and the “media compression” score of Waale.¹⁴ Briefly, the Schwartz score assesses the presence or absence of rupture of the internal elastic lamina, the media, and the external elastic lamina, and the Waale score assesses the degree of angular deformation of the media by the struts. In addition, to include both deep injury and stretch, a novel score was constructed as follows: 0, no impression of metal upon media; 1, deformation of the internal elastic lamina by < 45°; 2, deformation of the internal elastic lamina by > 45°; 3, rupture of the internal elastic lamina; 4, rupture of the external elastic lamina (that is, complete medial rupture). The score for each section was the mean of the individual strut scores (fig 1).

Arterial morphometry

Sections were analysed by computed morphometry. The cross sectional areas of the lumen, neointima, media, adventitia, stent, and whole vessel were recorded. Neointimal thicknesses (defined as the minimum distance between the strut and the lumen) were also measured.

Statistical analysis

Descriptive statistics (vessel cross sectional areas and thicknesses) are expressed as mean (SEM). Comparisons between groups were made with unpaired t tests. The relation between each injury score and the measurements of the lumen, neointima, and arterial wall layers were investigated with product–moment correlation coefficients. A probability value of p < 0.05 was regarded as significant.

RESULTS

Stent implantation and processing

Thirty BiodivYsio stents were deployed in 15 pigs, 15 in the left anterior descending coronary artery and 15 in the right coronary artery. All stents were angiographically patent after implantation. The achieved balloon to artery ratio, determined by quantitative coronary angiography, was 1.23 (0.02) to 1. From the 30 stents harvested at 28 days, we obtained 77 sections (a mean of 2.6 per stent) which fulfilled the criteria described above.

Stent related injury

Sections displaying different types of stent related injury are shown in fig 2. Four examples of stretch and two of deep injury are shown. Sections displaying varying degrees of stent related injury are shown in fig 3.

Measures of stretch and deep injury

Twenty per cent of struts (23% of sections) displayed deep arterial injury. Rupture of the internal elastic lamina was present in 14% of sections, and of the external elastic lamina in 9%. Seventy eight per cent of struts (77% of sections) showed evidence of arterial stretch without deep injury; 2% of struts (0% of sections) showed no stretch.

Injury scores

The Schwartz score was 0.23 (0.06). The Waale score was 0.42 (0.02). The new score was 1.74 (0.09).

Arterial morphometry

The quantitative morphometry of the wall layers of all 77 sections is shown in fig 4.

Deep injury and neointimal area

In sections containing a rupture of the internal elastic lamina, neointimal cross sectional area was 1.7-fold greater than in those in which there was no deep injury (p = 0.0002). In sections containing a rupture of the external elastic lamina, neointimal cross sectional area was 1.5-fold greater than in those with internal elastic lamina rupture only (p = 0.02) (fig 5).

Deep injury and remodelling

The total vessel cross sectional area was 8.91 (0.19) mm² in arteries with stretch only, whereas arteries containing an internal elastic lamina rupture had a 7% smaller total cross sectional area (downsize remodelling, p = 0.19) and those with an external elastic lamina rupture an 11% larger total cross sectional area (upsize remodelling, p = 0.08) (fig 5).

Injury score and neointimal area

The correlation between the three injury scores and neointimal cross sectional areas were: Schwartz, r = 0.64 (p < 0.0001); Waale, r = 0.43 (p = 0.001); and new score, r = 0.60 (p < 0.001) (fig 6).

Injury score and neointimal thickness

The correlation between the new injury score (for each strut) and neointimal thickness (for each strut) was r = 0.49 (p < 0.001) (fig 6).

DISCUSSION

In this study we showed that stretch of the media leads to neointima formation even in the absence of deep injury. With moderate stent oversize, stretch is more common than deep injury and is associated with neointimal growth. The greater the stretch, the larger the neointima. Deep injury, where it occurs, is a more powerful stimulus to neointima formation and is associated with a larger neointima than stretch alone. Rupture of the external elastic lamina (a deeper structure than the internal elastic lamina) is a yet more potent stimulus than rupture of the internal elastic lamina alone. A new injury score, devised to include all grades of stretch and deep injury, correlates well with neointima formation. Because of its universality, stretch might be considered to be the most important overall stimulus to in-stent restenosis where vascular injury is moderate.

We divided deep injury into rupture of the internal elastic lamina and rupture of both internal elastic lamina and external elastic lamina. This distinction appeared to be of biological importance (fig 5), rupture of the deep layer providing a further boost to neointima formation. Occasionally, deep injury could be associated with a single strut protruding from its neighbours (fig 2A), but it also occurred where there was overall oversizing (fig 2F). Stretch injury was synonymous with deformation of the circular artery by the polygonal stent. It occurred in almost every section examined, even in those with some localised deep injury.

Where there was no tension exerted upon the arterial wall, there was no neointimal growth (fig 3B). This is an important point. It implies (1) that a phosphorylcholine coated stent per se is completely biologically neutral and elicits no reaction from the artery; and (2) that stent geometry is only important in the presence of stretch.

In our model, with scant deep injury, the Schwartz score was low and did not differentiate between sections with mild and extreme strut impaction. There was thus a pronounced clustering of neointimal responses around a zero Schwartz score (fig 6A), although enough data points were left to provide a significant positive relation between deep injury and neointimal response. The Waale score provided a better data spread but showed a less intimate relation between injury—as assessed by media compression—and neointima formation than the Schwartz score. The new injury score included all grades of injury, and showed both good data spread and a close relation between injury and neointimal cross sectional areas (fig 6B). This score might therefore be more appropriate for the analysis of stented segments exposed to moderate injury. It would also allow correction for the degree of injury in the experimental setting.

The relation between the new injury score for the whole section and the neointimal cross sectional area was closer than the relation between the new injury score for each strut and the neointimal thickness at each strut (fig 6Bv 6C). This suggests that local stretch has effects not confined to the region immediately around the offending strut, but transmitted to the whole artery. This is an argument in favour of measuring arterial cross sectional area in stent studies rather than the maximum or average neointimal thickness. More importantly, cross sectional area reflects clinical restenosis better than a point measurement of intimal thickness.

There are two possible mechanisms whereby stretch might lead to neointima formation independent of deep injury. First, impaction of the stent causes compression of the vascular smooth muscle cells (VSMC) in the media beneath each strut. With major strut penetration, even with intact elastic laminae, this may cause atrophy of the media, although where cells abut the strut it is difficult to identify the presence or absence of nuclei. Cell death (necrosis or apoptosis) might then start a biological process leading to neointima formation. Second, stretch may transduce biological events directly. Stretch is transmitted cell to cell by adhesion molecules, and transduced through the cell by intracytoplasmic integrins, which may activate kinase signalling pathways leading to cell proliferation¹⁵ or matrix metalloproteinases leading to connective tissue production.¹⁶ The suggestion that neointima might occur in response to stent geometry, independent of injury,¹⁷ might now be explicable by the presence of stretch deformation.

There was a trend towards downsize remodelling of the whole vessel where there was a ruptured internal elastic lamina, whereas there was a trend towards upsize remodelling in the presence of a ruptured external elastic lamina. This may be explained by the constraining effect of an intact external elastic lamina allowing only inward expansion of neointima and loss of this constraint allowing inward and outward expansion of tissue. Stents have always been thought to eliminate remodelling, but our data suggest that, while this may be the case for deep injury where tissue growth can be outwards, it may not be where the injury is mild.

Limitations

There were some limitations of this study. First, the arteries were undiseased. On the other hand, undiseased arteries reveal the effect of subtle variations in stent geometry, independent of the wide variations inherent in disease models. Second, the vessels were porcine, not human. The pig is, however, the closest practical model with anatomical similarity to man. Third, arterial morphometry may change after 28 days. This time point was chosen as a practical compromise between completeness of neointima formation and the costs of animal husbandry.

Conclusions

Arterial stretch commonly follows from stent implantation, and is enough to stimulate neointima formation, even in the absence of deep injury, although deep injury is a more potent stimulus. A new injury score, including all injuries from mild stretch to deep penetration, works in this model. The effects of stretch are manifested more in the total (circumferential) amount of neointima than in its thickness at any one point. Without stretch, there is no neointima; thus stent geometry is only important in the presence of stretch. These findings may contribute to developing improved stent designs and stent related drug delivery systems.