Author + information
- Received December 8, 2008
- Revision received April 23, 2009
- Accepted May 5, 2009
- Published online November 17, 2009.
- Gaku Nakazawa, MD*,
- Aloke V. Finn, MD†,
- Marc Vorpahl, MD*,
- Elena Ladich, MD*,
- Robert Kutys, MS*,
- Isidora Balazs, BS*,
- Frank D. Kolodgie, PhD* and
- Renu Virmani, MD*,* ()
- ↵*Reprint requests and correspondence:
Dr. Renu Virmani, Medical Director, CVPath Institute, Inc., 19 Firstfield Road, Gaithersburg, Maryland 20878
Objectives The aim of this study was to perform pathologic assessment on stent fracture.
Background Clinically, stent fracture has been reported in 1% to 2% of patients after drug-eluting stent (DES) implantation.
Methods High-contrast film-based radiographs of 177 consecutive lesions from the CVPath DES autopsy registry were reviewed. Stent fracture was graded as I (single-strut fracture), II (≥2 struts), III (≥2 struts with deformation), IV (with transection without gap), and V (with transection causing gap in stent segment). The incidence of adverse pathologic findings (thrombosis and restenosis) was assessed histologically.
Results Stent fracture was documented in 51 lesions (29%; grade I = 10, II = 14, III = 12, IV = 6, and V = 9). Lesions with stent fracture had longer duration after implantation (172 days [interquartile range (IQR) 31 to 630 days] vs. 44 days [IQR 7 to 270 days], p = 0.004), a higher rate of Cypher (Cordis Corp., Miami Lakes, Florida) stent usage (63% vs. 36%, p = 0.001), longer stent length (30.0 mm [IQR 22.0 to 40.0 mm] vs. 20.0 mm [IQR 14.0 to 27.3 mm], p < 0.0001), and a higher rate of overlapping stents (45% vs. 22%, p = 0.003). Although fracture with grade I to IV did not have significant impact on the occurrence of adverse pathologic findings such as thrombosis and restenosis, 67% of the grade V fracture lesions were associated with adverse pathologic findings at fracture sites. Longer stent length, use of Cypher, and longer duration of implant were identified as independent risk factors of stent fracture by logistic regression analysis.
Conclusions The incidence of stent fracture was 29% lesions at autopsy, which is much higher than clinically reported. A high rate of adverse pathologic findings was observed in lesions with grade V stent fracture, whereas fracture with grade I to IV did not have a significant impact on the pathological outcome.
After pivotal clinical trials comparing bare-metal stents (BMS) to polymer-based sirolimus (Cypher, Cordis Corp., Miami Lakes, Florida) and paclitaxel (Taxus, Boston Scientific, Natick, Massachusetts) stents demonstrated a significant reduction in rate of restenosis, drug-eluting stents (DES) quickly became the standard of care for the percutaneous treatment of symptomatic coronary artery disease (1,2). Although late stent thrombosis has been raised as a safety concern in DES (3–5), the incidence of restenosis has dramatically decreased and is reported generally as 5%. However, stent fracture has emerged as a complication following DES implantation and is recognized as one of the contributors of in-stent restenosis (6–8) and possibly stent thrombosis (9,10). Clinically, the incidence of stent fracture is reported in 1% to 2% of patients at 8- to 10-month follow-up angiography (6,11). However, because of limited sensitivity of angiography to detect fracture, its true incidence is still unknown. The aim of this study was to assess the incidence of stent fracture at autopsy using high-contrast film-based radiography and to investigate the impact of stent fracture on the pathologic findings and clinical outcomes.
Study population and histologic preparation
The study comprised of 144 autopsy cases with 200 DES lesions. All hearts and/or coronary arteries had been fixed in 10% buffered formalin. A whole heart was X-rayed using high-contrast film-based radiography (Faxitron 43855A, Faxitron X-Ray, Lincolnshire, Illinois) in anterior-posterior view (70 to 75 kVp/15 s) when a whole heart was available. Epicardial coronary arteries were then dissected from the heart and re-X-rayed at a lower kilovolt potential setting (40 kVp/15 s). Stented arteries were submitted for plastic embedding and serially cross-sectioned at 2- to 3-mm intervals. All sections were stained with hematoxylin and eosin and Movat pentachrome as previously described (4,12).
Assessment of stent fracture and pathologic sections
All radiographs of stented vessels were assessed by light microscopy at 40× magnification. Stents with artifacts were excluded, including those that had been cut, distorted, or crushed during the removal of the vessel from the heart or were received in pieces. Stent length and extent of overlap was determined by examining the radiographs. Strut fracture was defined as complete transection of a strut. The presence of stent fracture was recorded and classified as grade I to V: I = involving a single-strut fracture; II = 2 or more strut fractures without deformation; III = 2 or more strut fractures with deformation; IV = multiple strut fractures with acquired transection but without gap; and V = multiple strut fractures with acquired transection with gap in the stent body (Fig. 1A).The severity of calcification in the stented lesion was also recorded as none, mild (calcification barely seen or focally localized; <25% stented segment), moderate (multiple sites of calcification), or severe (highly visible diffuse calcification; >70% stented segment) (Fig. 1B).
To correlate the pathologic findings and stent fracture, histologic sections were also reviewed. The presence of acute thrombus was defined as a platelet-rich thrombus that occupied >30% of the cross-sectional area of the lumen, and restenosis was defined as >75% cross-sectional area narrowing by neointimal formation within the stented segment. In all cases of stent fracture, an attempt was made to section the area of the fracture. In cases with thrombosis or restenosis, the relationship between these adverse pathologic findings and the site of fracture was recorded (i.e., histologic finding in sections at the fracture site or close to fracture site). In addition, histologic comparisons were performed between stents with fracture versus without. Furthermore, the same comparisons were made between various fracture grades. Neointimal thickness, fibrin deposition, and inflammatory reaction, including giant cell and eosinophil infiltration, were assessed as previously described (4,5).
Continuous variables with normal distribution were expressed as mean ± SD. Continuous variables with non-normal distribution were expressed as median and interquartile range (IQR). Comparisons between fracture and nonfracture lesion were tested by Student ttest for normally distributed continuous variables and chi-square test for categorical values. A Wilcoxon rank sum test was used for comparisons of non-normally distributed parameters or discrete variables. Normality of distribution was tested with the Wilk-Shapiro test. For multiple logistic regression analysis, a forward stepwise method was used to detect independent predictors of stent fracture. Variables in Table 1with p < 0.05 were entered into the analysis, and those with p > 0.10 were removed. A value of p < 0.05 was considered statistically significant.
Incidence of stent fracture
Twenty-three lesions were excluded because of artifacts that were observed when we received the samples. In the remaining 177 lesions, stent fracture was documented in 51 lesions (29%) (grade I = 9, II = 14, III = 11, IV = 6, and V = 9). Most of the fractures in the Cypher stents were located in flexible N-shaped, undulating longitudinal intersinusoidal-ring linker segments, whereas in Taxus Express stents, fractures were observed in the straight longitudinal intercrown linker or the modular ring portion (13). Furthermore, in single-stented lesions, the majority of stent fractures were localized in the middle portion of the stent body, except for stents >25-mm length where the fracture sites were slightly shifted toward the proximal portion (Fig. 2A).On the other hand, in overlapping stents, most fractures were observed within 5-mm distance from the overlapping zone, with similar frequency in proximal and distal regions (Fig. 2B). Mean age and sex of patients were not different between lesions with versus without fracture (Table 1). There was no significant difference in cause of death between patients with fracture and those without. Lesions with stent fracture had longer duration of stent implants (fracture; 172 days [IQR 31 to 630 days] vs. nonfracture; 44 days [IQR 7 to 270 days], p = 0.004). There was no statistical difference in stent implant duration between each stent fracture of grade (grade I: 31 days [IQR 5 to 616 days], II: 105 days [IQR 27 to 1,095 days], III: 376 days [IQR 72 to 570 days], IV: 331 days [IQR 31 to 833 days], and V: 172 days [IQR 44 to 450 days], p = 0.70). Furthermore, lesions with stent fracture showed a higher rate of Cypher stent usage (63% vs. 36%, p = 0.001), longer stent length (30.0 mm [IQR 22.0 to 40.0 mm] vs. 20.0 mm [IQR 14.0 to 27.3 mm], p < 0.0001), greater number of stents (1.7 ± 0.9 vs. 1.3 ± 0.6, p = 0.008), and a higher rate of overlapping stents (45% vs. 22%, p = 0.003). However, there were no statistical differences in severity of vessel calcification between the 2 groups. Stent fractures were more commonly found in the right coronary artery (RCA) and in bypass grafts, but this did not reach statistical significance. Longer stent length (p < 0.0001, odds ratio [OR]: 1.07, 95% confidence interval [CI]: 1.036 to 1.100), use of Cypher stent (p = 0.002, OR: 3.40, 95% CI: 1.57 to 7.33), and longer duration (p = 0.002, OR: 1.002, 95% CI: 1.001 to 1.003) were identified as independent risk factors of stent fractures by a forward stepwise logistic regression analysis.
Histologic comparison between fracture and nonfracture lesions
Neointimal thickness was similar between stents with fracture (0.11 mm [IQR 0.06 to 0.19 mm]) and without (0.11 mm [IQR 0.03 to 0.19 mm], p = 0.62). There was no significant difference in fibrin deposition (fibrin score: fracture (+); 1.0 [IQR 0.1 to 1.5] vs. fracture (−); 1.4 [IQR 0.4 to 2.0]) and inflammation (inflammatory score: fracture (+); 1.0 [IQR 0.5 to 1.6] vs. fracture (−); 1.4 [IQR 0.4 to 2.0]) including a similar degree of giant cell and eosinophil infiltration. Furthermore, no differences in these parameters among various fracture grades (i.e., grades I to V, data not shown).
Correlation to adverse pathologic findings
Six adverse pathologic findings (5 thrombosis and 1 restenosis) were associated with grade V fracture (67%), whereas there were no fracture site-related adverse pathologic findings in stents with grades I to IV except for 1 stented lesion with grade IV (p < 0.0001 vs. grade V) (Table 2).Total rates of adverse pathologic findings irrespective of fracture severity were similar between lesions with fracture and those without (35% vs. 37%, p > 0.99). However, stents with grade V fracture showed a significantly higher rate of adverse pathologic findings as compared with those without fracture (78% vs. 37%, p = 0.03) (Table 2).
Cases with grade V fracture are listed in Table 3.A 68-year-old-female (Case 1), who died of late stent thrombosis in the left circumflex coronary artery (LCx), had 2 stents in the LCx/left obtuse marginal branch (LOM) bifurcation. Both LCx (Taxus) and LOM (Cypher) stents showed grade V fracture. Taxus stent fracture in LCx was located in the area close to the bifurcation site and was near the thrombus, which extended into the bifurcation site, whereas the Cypher stent in LOM had restenosis at the site of fracture (Fig. 3).The Taxus stent was distorted at the fracture site, and some struts protruded into the lumen where thrombus was observed. Cypher stent, on the other hand, showed moderate chronic inflammation, including macrophages and lymphocytes at the site of fracture. A 60-year-old male (Case 2) with grade V Cypher stent fracture had chest pain 60 days following stent placement and underwent angiography. Stent thrombosis was detected by angiography, which was treated by balloon angioplasty followed by bypass surgery; however, the patient died from surgical complications. The histologic sections at the fracture site showed fragment of thrombus with moderate fibrin deposition and inflammatory cell infiltration, including macrophages, giant cells, and occasional neutrophils. A 58-year-old male (Case 3) who died of subacute stent thrombosis 11 days following stent implantation showed complete separation of the Cypher stent distal to the site of overlap with thrombotic occlusion. At the site fracture, the stent was underexpanded, and thrombus was observed in the lumen (Fig. 4).However, there was no significant difference in vascular reaction, such as fibrin deposition and inflammation, between fracture and nonfracture sites. Another 83-year-old male (Case 4) who died 27 days following Taxus stent implantation, had a stent fracture that showed septic thrombotic occlusion in the distal portion of 4 overlapping stents. The thrombotic occlusion was observed, not only at the fracture site, but extended into the proximal stents, and was thought to have lodged at the site of fracture related to the infection. A 57-year-old male (Case 5) who died of diffuse coronary artery disease had a Cypher stent placed in the distal RCA 1.800 days ante mortem. The stent showed a transection with a gap, and histology revealed chronic total occlusion with organized thrombus at the site of fracture. A 71-year-old male (Case 6) who died with late stent thrombosis had a long Taxus stent implanted in the RCA, and a thrombus was observed only in the proximal portion, whereas grade V stent fracture was located in the middle portion of the stent. A Cypher stent in a 68-year-old male (Case 7) had grade V fracture; however, the vessel was widely patent at 570 days, and the patient died of severe coronary artery disease in nonstented segments (Fig. 5).A 62-year-old male (Case 8) received 2 Cypher stents 130 days ante mortem and died of diffuse coronary artery disease. The Cypher stent showed an acquired transection distal to the overlapping site with a gap. Histologic section showed widely patent lumen at the fracture site and mild fibrin deposition with minimal inflammation, including macrophages and lymphocytes.
Main findings in the present study are: 1) a higher rate of stent fracture in DES at autopsy (29%) than what has been clinically reported; 2) a higher event rate was observed in lesions with grade V stent fracture, whereas grade I to IV fracture showed no significant impact on the clinical outcome; and 3) longer stent length, Cypher stent usage, and longer stent duration of implant were identified as independent predictors of stent fracture.
The incidence of stent fracture has been reported to be 1% to 2% in clinical settings (6,7). In the current study, stent fracture was observed in 29% of lesions, which is far more frequent compared with what was reported clinically. The reason for the higher detection rate of fracture is due to the use of a high-contrast film-based radiography that allowed us to detect even minor stent fracture in contrast to poor visibility by angiography or intravascular ultrasound (IVUS) because of insufficient resolution. The resolution of a high-contrast film-based radiography is reported as ≅80 μm, which is greater than that of angiography (300 μm) or IVUS (200 μm). Our findings suggested that there are a fair number of patients who have “subclinical,” low-grade stent fracture that remains clinically silent. However, 9 lesions (5.1%) were associated with grade V fractures (i.e., complete separation) in the present study. Given that only severe fractures with gap can be detected by angiography, the incidence of fracture with grade V in the current study is within expectations and consistent with previous clinical reports.
Although the stent fracture was more frequently documented in our population, the incidence of adverse pathologic findings was similar between stents with and without fracture when all grades of fractures were included. This could be explained by the study population since ours is an autopsy study and therefore might have a higher incidence of DES failures in both groups. However, if only grade V fractures are included, stent fracture is associated with a significantly higher rate of adverse pathologic findings as compared with nonfracture group as well as fracture with grade I to IV. We observed 6 adverse pathologic findings at the fracture sites (5 stent thrombosis and 1 restenosis) among 9 grade V fracture lesions (67%) in the present study. Although it is not fully understood why stent fractures cause adverse events, the lack of stent integrity, such as distortion or acquired underexpansion, may also play an important role in occurrence of adverse events as shown in some of our cases. In additional analysis, no significant pathological differences (i.e., inflammation, giant cell reaction, and fibrin deposition) were observed between stents with and without fracture, as well as among the various fracture grades. Although stent fracture clinically has been reported to predominantly be associated with restenosis, we had a higher incidence of acute thrombus in our study. Nevertheless, some case reports have described cases of stent thrombosis with stent fracture (9,10).
Previous clinical studies have reported that the main risk factors for stent fracture are longer stent length, RCA or saphenous vein graft lesion location, lesion with high motion, overlapping stent, and Cypher stent use (6,8,14). Although actual stent angulations or motion on the heart could not be assessed in the present study, the risk factors identified in the present study are similar to those reported clinically. Cypher stent was associated with higher incidence of stent fracture; the flexible N-shaped, undulating longitudinal intersinusoidal-ring linker segment was the most frequent location of the fractures, which are smaller in width than the sinusoidal-ring portion. RCA and bypass graft lesions were also more common in the fracture group than in the nonfracture group but did not reach statistical significance because of the small number of cases in the present study. On the other hand, the time course of stent fracture has not been investigated in clinical studies, since first follow-up angiography is usually planned at 8 to 12 months. In the present study, longer implant duration was identified as an independent risk factor for stent fracture, thus suggesting that stent fracture may result from continuous stress over time to the implant, which leads to greater metal fatigue with eventual fracture. However, it should also be noted that stent fracture was seen even in the patients who died shortly after stent implantation, which is probably procedure-related (high pressure and/or oversized balloon, overlapping stent, and so on).
Because this is an autopsy study, the results may not be representative of all patients who receive DES. Our population is biased toward patients dying from DES complications. Stent thrombosis and restenosis rates in our registry are considerably higher than those reported in clinical settings. Furthermore, vessel motion, tortuosity, and angulations on the beating heart, which also play an important role in stent fracture (15), cannot be assessed. The mechanisms of adverse pathologic findings in grade V fracture could not be elucidated by histologic assessment in the present study, probably because of the small sample size and time variation at autopsy. Unlike clinical or preclinical studies, the timing for the observation in autopsy studies cannot be controlled. In fact, the stent implant durations in grade V fracture range from 11 to 1,800 days. However, to the best of our knowledge, this is the first report of stent fracture in a large series assessed at autopsy with a highly sensitive method, and therefore, the findings in the present study remain valid and represent a more accurate assessment of fracture incidence than either angiography or IVUS.
The incidence of DES fracture is 29% of the stented lesions at autopsy, which is much higher than clinically reported. A high rate of adverse pathologic findings was observed in lesions with grade V fracture, whereas grade I to IV fracture did not have significant impact on the clinical outcome. Longer stent length, Cypher stent usage, and longer duration of stent implant were identified as independent predictors of stent fracture.
The authors thank Mr. John Newell for his valuable assistance in statistical analysis.
Dr. Finn has received research funding from Boston Scientific. Dr. Virmani receives research support from Medtronic AVE, Abbott Vascular, Conor Medsystems, OrbusNeich Medical, Terumo Corporation, Cordis Corporation, BioSensors International, Prescient Medical, Biotronik, and Alchimedics; and is a consultant for Medtronic AVE, Abbott Vascular, Prescient Medical, and Biotronik.
- Abbreviations and Acronyms
- bare-metal stent(s)
- drug-eluting stent(s)
- intravascular ultrasound
- left circumflex coronary artery
- right coronary artery
- Received December 8, 2008.
- Revision received April 23, 2009.
- Accepted May 5, 2009.
- American College of Cardiology Foundation
- Finn A.V.,
- Joner M.,
- Nakazawa G.,
- et al.
- Joner M.,
- Finn A.V.,
- Farb A.,
- et al.
- Shite J.,
- Matsumoto D.,
- Yokoyama M.
- Farb A.,
- Weber D.K.,
- Kolodgie F.D.,
- Burke A.P.,
- Virmani R.
- Butany J.,
- Carmichael K.,
- Leong S.W.,
- Collins M.J.