Stent fractures are common in the field of peripheral vascular interventions and used to be unrecognized in coronary arteries [5]. Chowdhury and Ramos first described a fracture of a coronary BMS in a saphenous vein graft in 2002 [6]. BMS fracture is a rare finding [3, 7, 8], which might be explained by a stabilizing effect of greater neointimal proliferation [9, 10] but also a more difficult diagnosis of fracture due to lower radiopacity [11]. The incidence of SF became more considerable with the introduction of DES. Sianos et al. [12] were the first to report SF in DES in two cases involving sirolimus-eluting stents. In their meta-analysis of 8 studies assessing SF, Chakravarty et al. [13] report rates of SF ranging between 0.8% and 8.4% with a mean incidence of 4.0%. Notably, all but one SF in their analysis occurred with SES. Similar results were obtained by other studies with an incidence of 1.0% up to 18.6% [10, 11, 14, 15, 16, 17, 18, 19, 20]. An analysis with new generation DES by Schochlow et al. [21] showed incidental detection of SF in 8% immediately after implantation and almost 60% in the setting of device failure. The largest multicenter study has been carried out by Kan et al. [22] reporting an incidence of SF in 12.3% of the 6555 involved patients, 22.0% of stents and 17.2% of vessels. Limiting to clinical studies however is the varying definition of SF as well as an incomplete angiographic follow-up with the potential to miss out especially on late occurring SF. Accordingly, the highest incidence of SF has been reported in a post-mortem analysis, where Nakazawa et al. [23] found evidence of DES fracture in 29% of patients.

The mechanistic culprit behind SF is material fatigue. Biomechanical demands of coronary stents are high. Important factors include vascular biocompatibility, resistance to corrosion, high elasticity and plasticity for expansion, rigidity at body temperature for the maintenance of dilatation and resistance to elastic recoil as well as radiopacity to allow X-ray tracking [24]. While the most broadly used stent backbone material used to be stainless steel, modern 2nd generation DES are preferably fabricated with a cobalt-chromium (Co-Cr) alloy, typically L-605 and MP35N. Radial strength is supported by hoop elements, which are linked by connectors. The latter provide longitudinal stability and their design varies between stent types. In an experimental approach, Ormiston et al. [25] showed highest susceptibility to fracture in less flexible stents and stents with three connectors between hoops. Accordingly, most studies have found the relatively inflexible sirolimus-eluting stent (SES) to be the most susceptible to SF [5, 3, 11, 13, 20, 23]. In 10 million cycles of repetitive bending no fracture occurred in the more flexible Element (Boston Scientific, Natick, Massachusetts, MA, USA), Promus (Bostin Scientific, Natick, Massachusetts, MA, USA) and Integrity (Medtronic, Santa Rosa, CA, USA) stents [25]. From their analysis of the American Food and Drug Administration’s (FDA) Manufacturer and User Facility Experience Database (MAUDE) Omar et al. [26] also report the highest SF rate in Cypher stents, followed by Xience and Promus stents. But not only the infrastructure of the stent can make a difference. DES are coated with a polymer that controls the release of an antiproliferative drug. Especially in first generation DES, hypersensitivity reactions of the surrounding endothel induced by the polymer with consecutive inflammation has been described. The inflammatory process can then cause late stent malapposition with modification of the mechanical integrity of the device. The results are endothelial hinge points, excessive motion and torsion and finally stent thrombosis or SF [10, 27, 28]. The frequency of polymer-associated inflammation appears to be lower in 2nd generation DES with lowest reported SF rates in everolimus-eluting stents [29].

Further technical improvements have been made to increase flexibility and fracture resistance. Platinum-chromium alloys were recently introduced, allowing for thinner stent struts and higher radiopacity without decreasing radial strength [30, 31, 32, 33]. Kuramitsu et al. [34] observed SF in 1.7% of lesions and 2.2% of patients treated with platinum-chromium alloyed everolimus-eluting stents and found a numerically higher incidence of clinically-driven TLR compared to non-SF-lesions. To date, there is no data comparing SF rates in platinum-chromium stents with previous models. However, though not reporting specifically on SF, studies have shown comparable rates of stent thrombosis between cobalt-chromium and platinum-chromium alloy with a significantly lower rate of target lesion failure in platinum-alloyed stents [35]. No stent thrombosis was reported after treatment of small vessels and long lesions, both known as SF-susceptible [36].

A different and promising approach was made with the development of bioresorbable scaffold stents. The latest models, which consist of poly-l-lactic acid (PLLA), degrade over time and were therefore thought to decrease rates of late stent failure. However, bioresorbable PLLA stents have proven inferior compared to non-degradable metallic stent platforms due to increased rates of target lesion restenosis and more frequent target vessel myocardial infarction [37]. Combining the mechanical advantages of non-resorbable metallic platforms and the biodegradability of resorbable stent material bioresorbable metallic stents are currently being advanced. The magnesium-alloyed Magmaris stent (Biotronik AG, Berlin, Germany) has shown safety and efficacy in early registries [38], however randomized controlled trials have not been carried out yet. No data is currently available regarding the use of scaffolds in SF predilection lesions. It is thinkable however, that while the scaffold technology may reduce the risk of fracture in angular and hinge point lesions their rather soft structure might make them prone to breakage or dismantling especially in heavily calcified lesions. Moreover, an essential issue with magnesium-based implants is their rapid degradation and corrosion in aqueous environments like body fluid. A more stable zink-silver (Zn-Ag) alloy has shown promising results in a porcine model with treatment of iliofemoral arteries [39].

Characteristics of the treated lesions are another important factor in the pathomechanism of SF. Numerous studies have shown, that stenting in the RCA is an independent predictor of SF [5, 16, 17, 40, 41, 42, 43, 44]. In their meta-analysis Chakravarty et al. [13] estimated 56.4% of SF to occur in the RCA followed by 30.4% in the left anterior descending artery (LAD). SF was least frequent in left main artery lesions [13]. Omar et al. [26] reported 47.7% of SF in the RCA, followed by the LAD. The most likely explanation for this clustering of SF events is the tortuosity and contortion of the RCA, leading to higher mechanical force and earlier material fatigue [3, 17, 20]. Ino et al. [20] showed a higher degree of hinge motion in the RCA or left circumflex artery (LCX) than in the LAD (31.0° $\pm$ 13.1° vs. 22.8° $\pm$ 4.9°).

Stenting across angular or hinge regions is a risk factor for SF regardless of the vessel. Park et al. [44] found a more than 6-fold higher risk of SF when stenting across an angle of $>$45°. Popma et al. [45], while finding no significant difference in SF incidence between RCA and LAD, reported cyclic angulation changes of 32.3° $\pm$ 15.2° in patients with SF. Shaikh et al. [11] demonstrated a 14-fold risk increase for SF when stenting across a bend of $>$75° (Odds Ratio 13.8). Kuramitsu et al. [15] reported an Odds Ratio for SF prediction of 14.6 by hinge motion as defined by an at least 16° difference between systole and diastole. And in a study by Park et al. [16] 79% of SFs occurred at hinge points, either adjacent to edges of overlapping stents or at angles $>$45°. But not only does coronary anatomy have an effect on stent mechanics. Vice versa the presence of a stent can significantly alter vascular geometry. Especially stents with low conformability, i.e., small amount of longitudinal flexibility after deployment, can create maldistribution of force during the cardiac cycle predisposing for SF [46].

Other lesional risk factors include ostial [15, 47] and bifurcational lesions [21, 48] as well as plaque calcification [15, 21, 49].

Finally, procedural details can also contribute towards SF. Stent length plays a significant mechanistic role [13]. In the study by Park et al. [16], SF occurred in stents with a mean length of 48.3 mm. Omar et al. [26] report from the MAUDE database that 65.2% of SF happened in stents longer than 30 mm. Overlapping stents are also a risk factor, since they can change vascular angulation and potentially create new hinge points. Omar et al. [26] discuss that half of the SF events in the RCA and LAD registered in the MAUDE database involved lesions with overlapping stents. Stent overlap was also significantly associated with SF in the large meta-analysis by Chakravarty et al. [13]. Finally, stent overexpansion can exceed material capacity and induce strut distortion causing early fatigue and SF [5, 12, 21, 50]. Omar et al. [26] report a mean postdilation pressure of 18 atm (IQR 16–20) in fractured stents. However, Lee et al. [51] showed the occurrence of stent thrombosis in only 0.2% of 1037 patients with intravascular ultrasound (IVUS) guided postdilation with a mean pressure of 18.7 $\pm$ 4.1 atm. SF risk from postdilation is likely to vary between different lesions and might be higher in calcified or tortuous lesions [5, 12]. Risk factors of SF are summarized in Fig. 3.

When a stent fractures, the new lesion predisposes for ischemic events. The altered stent geometry, mechanical shear stress and impaired contact of the antiproliferative drug to the endothelium can cause neoproliferation of endothelial tissue and smooth muscle cells, intimal hyperplasia and alteration of hemodynamic factors [13, 15, 52, 53]. Furthermore, complete SF is associated with the formation of coronary aneurysm [10, 45]. As a side note, occasional case reports have been published showing SF as a complication of post-stenting pyogenic infections and mycotic coronary pseudoaneurysm [54, 55]. Furthermore, an infected SF-induced coronary artery aneurysm has been described as a highly probable origin for formation of abscesses and sepsis with Staphylococcus aureus [56, 57].

Most SF are reported to occur several months after initial stenting [10, 13]. These events are likely due to the described pathomechanisms and risk factors of material fatigue. However, some SF have been reported only days after stent deployment [5, 50, 58] and are likely to have occurred in high-risk lesions (calcification, angle, overexpansion).

All the above risk factors and mechanisms show, that SF can be seen as a form of “patient-prosthesis mismatch” that requires further improvements in engineering and precision medicine [59].

Most studies reporting on SF have based their diagnosis on fluoroscopy [5, 15, 22, 60]. Ino et al. [20] discussed, that contrast injection might mask SF lesions and implanted stents should be assessed with and without contrast agents. However, the low spatial resolution of 300 $\mu$m limits the diagnostic abilities of fluoroscopy, especially in newer stent models with thin struts. Therefore, additional diagnostic modalities have been introduced, which significantly improved imaging. To enhance visibility of stent struts, high-resolution cine-angiography technologies such as StentBoost (Philips Healthcare, Best, Netherlands) can be helpful [61, 62] and even diagnosed SF in one case, where evidence by intravascular ultrasound (IVUS) was lacking [63]. In a large prospective study by Biscaglia et al. [64] enhanced stent visualization techniques proved highly effective and safe in detection of SF during the index PCI. Use of digitally enhanced fluoroscopic imaging can lower the incidence of TLR and MACE [65]. Davlouros et al. [66] suggested a staged approach for high-risk patients with routine flat panel digital detector cinefluoroscopy and invasive assessment in case of pathologic initial findings.

IVUS has been used successfully to visualize SF due to its higher spatial resolution (150–200 $\mu$m) [5, 10, 13, 15, 19, 67]. Yamada et al. [68] reported superiority of IVUS compared to angiography to detect SF.

Optical coherence tomography (OCT) with a spatial resolution of 10–15 $\mu$m has seen a rise in use for the differential diagnosis of stent failure [50, 69, 70, 71, 72]. OCT can reproduce the complex spatial stent configuration with high precision and reproducibility [73]. Most recently, Schochlow et al. [21] have demonstrated an unexpectedly high prevalence in OCT-diagnosed SF in the setting of elective control examinations.

Since many SF events remain asymptomatic non-invasive diagnostic options are useful. Computer tomography angiography (CTA) has been used as a reliable tool. Hecht et al. [48] have reported SF in 28% of patients with ISR as assessed by CTA. They defined evidence of SF with the following criteria: partial or complete (circumferential) gap or “crush” pattern and reduction of Hounsfield units 300 proving the absence of metal. CTA can be superior to fluoroscopic angiography for the detection of SF [74, 75, 76]. However, due to formation of severe artifacts some patients and stents are more suitable for follow-up by CTA than others: Carbone et al. [77] suggest patient selection according to stent diameter, stent material and type as well as heart rate and rhythm.

Depending on the diagnostic tool that is used, numerous classifications of SF have been suggested. The first categorizations of SF based on fluoroscopy were done with self-expanding nitinol stents which were implanted in the superficial femoral artery. Scheinert et al. [78] described minor SF as single-strut fracture, moderate SF as fracture of more than one strut and severe SF as complete separation of segments. This classification was later adapted for coronary stents by Lee et al. [5], Shaikh et al. [11] and Kim et al. [43]. Allie et al. [79] suggested four types of SF with type I being single strut fracture, type II being multiple strut fractures at different sites, type III being complete transverse linear fracture without displacement and type IV being complete stent displacement. Jaff et al. [80] introduced an additional type V with formation of a gap between stent fragments. A similar classification was used for coronary stents by Nakazawa et al. [23] in their pathological study.

Using IVUS, Doi et al. [10] differentiate between partial SF (absence of stent strut across at least one third of the stent) and complete SF (evidence of at least two fragments separated by an image slice with no visible struts).

Schochlow et al. [21] reported four OCT-patterns of SF, with pattern 1 being a single stacked strut and pattern 4 being stent transection with or without gap formation.

Finally, Hecht et al. [48] distinguished between partial and complete SF using CTA.

Stent-related adverse events are emerging as a significant issue for the interventional cardiologist. In a recent individual patient pooled study analysis, Madhavan et al. [60] showed an incidence of very late stent-related events of 2% per year with all stent types, without an evident plateau over time. SF is one proposed mechanism of stent failure and is known as a major risk factor for ISR, stent thrombosis, TLR and MACE [11, 13, 15, 40, 42, 81, 82, 83]. From their meta-analysis, Chakravarty et al. [13] report a significantly higher risk for ISR (38% vs. 8.2%, p 0.001) and TLR (17% vs. 5.6%, p 0.001) in lesions with fractured stents. In the large study by Kan et al. [22], SF increased the incidence of ISR, TLR and stent thrombosis more than three-fold. Kashiwagi et al. [83] showed a 5-fold higher rate of ISR in SF lesions compared to non-SF lesions. And Ohya et al. [82] described a significantly increased risk for clinically driven and all-cause TLR.

Clinical presentation seems to be associated with the extent of SF. SF was seen in 20% of asymptomatic control group devices in a study by Schochlow et al. [21], suggesting SF as a common phenomenon often without clinical implications. Lee et al. [19] showed that only patients with SF grade III and IV were admitted with acute myocardial infarction. No cardiac deaths occurred in their study. In their post mortem analysis Nakazawa et al. [23] found adverse pathologic findings such as thrombosis and restenosis at the SF site in 67% of type V fractures. No significant impact on pathologic findings was seen in type I to type IV SF [23]. While not reporting on the respective grade of SF, Park et al. [17] did not find a significant difference in severity of angina pectoris or incidence of acute coronary syndrome in patients with SF compared to a matched control group. And in their large multicenter study, Kan et al. [22] did not find a difference in mortality between SF and non-SF patients.

On the other hand, reports of SF leading to STEMI have been published with BMS [7] as well as 1st generation, cobalt-chromium alloyed DES [84, 85]. In a study with 2nd generation DES STEMI occurred in 15.8% of patients with SF [21]. Kuramitsu et al. [15] described that the risk for myocardial infarction was more than 12-fold higher in SF compared to non-SF patients. Chhatriwalla et al. [86] reported that 12% of patients with SF presented with STEMI or stent thrombosis and 19% with unstable angina or NSTEMI. Omar et al. [26] found that one third of patients with SF presented with acute coronary syndrome. Ohya et al. [82] report a significantly higher risk for myocardial infarction and very late stent thrombosis in SF patients.

Heterogeneity of the available data might again be due to the varying definition and different diagnostic modalities in SF studies. Undoubtedly, however, SF has a hazardous potential and interventional cardiologists should be familiar with therapeutic strategies to prevent adverse outcome.

Despite the clinical experience with risk factors of SF, management of SF remains challenging and poorly researched. To date, no randomized controlled trials have been carried out to suggest an optimal treatment. Options for management of SF patients are drug therapy, re-stenting, balloon angioplasty and, in some cases, coronary bypass grafting [87]. Omar et al. [26] reported from the MAUDE database that half of patients re-admitted with STEMI or stent thrombosis due to SF were treated with DES, 23% with medical therapy alone, 13% with ballon angioplasty and 8% with surgery. While treatment of myocardial infarction due to SF is mainly interventional, there is an ongoing debate how asymptomatic SF lesions should be managed. Since the clinical course of SF, especially minor types with single strut fractures, are often benign and asymptomatic and pose a low risk for adverse cardiac events the cost-benefit ratio of re-intervention is often doubted. Therefore, Adlakha et al. [88] proposed to leave asymptomatic SF-related restenosis without treatment and reserve intervention for symptomatic patients. Lee et al. [19] advocated treatment of SF with continuation of dual antiplatelet therapy irrespective of symptoms and suggested re-intervention in symptomatic or asymptomatic ISR with $>$70% stenosis or symptomatic ISR with 50–70% stenosis, which shows positive results in physiological stress test. None of the conservatively treated patients in their study had significantly aggravated restenosis during follow-up nor did cardiac death occur [19]. Ino et al. [20] reported similar results with no adverse outcomes in SF patients without significant restenosis who were treated with dual antiplatelet therapy. Park et al. [17] only performed re-intervention in SF patients with ISR $>$70% and achieved excellent results without adverse events. SF Patients without ISR or with acceptable fractional flow reserve were treated conservatively with dual antiplatelet therapy and no patient required TLR during a follow-up period of median 30.5 months [17].

Different strategies are needed for patients presenting with SF and myocardial infarction. Most reports of STEMI due to SF have been treated by re-intervention and stent deployment [7, 84]. However, stenting in mal-apposed stent struts results in a double layer of metal, leading to an increased risk of thrombogenicity, ISR and recurring SF. Case reports have been published using “plain old balloon” [85] or drug-coated balloon (DCB) angioplasty only in a setting of STEMI achieving good short-term results [87]. A balloon only approach to prevent SF in de-novo lesions or stent failure including re-SF could be feasible. The BASKET-SMALL-2-trial has shown non-inferiority of drug-coated balloon (DCB) application compared to DES implantation in small de-novo coronary lesions [89]. DCB was also shown to be non-inferior to 2nd generation DES for treatment of DES ISR [90, 91]. However, specifically in SF DCB application alone could not achieve lower rates of re-ISR and TLR compared to DES implantation [92]. The authors proposed maintained mechanical stress at an SF site as the mechanism of re-re-ISR regardless of the used device [92].

Coronary artery bypass grafting can be seen as the “last resort” in recurrent SF or SF with heavy ISR or stent thrombosis that makes the vessel inaccessible for intervention wires.