Introduction

Catheter-induced coronary artery dissection is an infrequent but potentially life-threatening complication of percutaneous procedures. We have recently studied its patterns in electrophysiological procedures; however, the incidence is much greater for coronary interventions [1, 2]. Involving proximal segments of major coronary vessels, it usually jeopardizes a large myocardial area and may propagate into the aorta more easily than guidewire-related dissections. It is also more dangerous than injuries induced by a balloon or stent because it may occur prior to coronary artery wiring or during diagnostic angiography.

Traditionally, studies of iatrogenic dissections focused on one dissected vessel or structure, be it the left main coronary artery (LMCA) or the sinus of Valsalva (SoV), thereby presenting only selective data [2–6]. Still others exclusively included dissections complicating specific procedures, such as chronic total occlusion interventions [7]. Registries of iatrogenic injuries of the aorta are also available, and although catheter-induced aortocoronary dissection is a distinct entity with its specific risks and treatment, coronary artery dissection and its retrograde extension into the aorta represent not only an anatomic but also a pathophysiologic continuum [8]. At the same time, various dissection-triggering devices (e.g. wires, stents) were usually taken into account, which precluded the analysis of mechanisms and factors specific to the catheter-related trauma. We sought to identify all catheter-induced injuries complicating coronary procedures to analyze the mechanisms of dissection, predisposing and precipitating factors, and propagation causes. Iatrogenic complications are avoidable by their nature; it is thus crucial to search for patterns and risk factors.

Methods

We performed a retrospective study of catheter-induced coronary artery and aortic dissections (CICAAD) at two high-volume centers. CICAAD cases (including coronary graft dissections) were retrieved from prospectively collected registries of procedural complications. They were double--checked by a query of electronic catheterization databases, utilizing keywords such as dissection, iatrogenic, spiral, or extraluminal. Recorded images of the right coronary artery (RCA) intervention with multiple stents as well as ad hoc angioplasty of LMCA were additionally reviewed to check for underreported cases. All CICAAD angiograms were assessed by two interventionists to confirm and classify a dissection, its mechanism, predisposing and precipitating factors, and propagation causes. If there was a discrepancy, a third reviewer was included in the adjudication.

Definitions

Dissections were graded according to the National Heart, Lung, and Blood Institute criteria, Eshtehardi and Dunning classification (Suppl. Table 1) [9, 10].

After the review of the first 60 cases, a consensus was reached to discern three main mechanisms of dissection. ‘Wedged contrast injection’ was defined as a primarily hydraulic injury caused by contrast agent delivery with damped outflow, i.e., catheter tip embedded in the artery wall. Initial vessel wall filling was usually recorded in statu nascendi with dye persistence afterwards (Suppl. Video 1). ‘Forceful catheter engagement’ (or vigorous pecking motion) described mechanical trauma caused by a catheter tip impact with a spot damage of a coronary artery or the aortic wall, typically inflicted at first contact with forceful intubation and immediately visible, or after vigorous pecking at a vessel wall (Suppl. Video 2). ‘Deep catheter insertion’ indicated a long-segment mechanical injury caused by deep intubation for support, or by inadvertent dragging of the catheter into a vessel during difficult device retrieval (Suppl. Video 3).

Multiple predisposing and precipitating factors suggestive of their role in the injury have been identified. They were assigned after cineangiography review. Since the project launch in 2017, operators provided internal reports detailing problems encountered during CICAAD procedures, with presumable triggers pointed out. Predisposing factors were defined as unmodifiable anatomic conditions, both normal (e.g., shepherd’s crook take-off) and anomalous or pathological (e.g., arteria lusoria, ostial atheroma), involving the arterial access route or a coronary artery origin, making it difficult to advance and/or manipulate a catheter. Precipitating factors included modifiable conditions that could have been reversed or corrected, such as catheter behavior (e.g., respiratory instability), vessel response (e.g., concertina effect), or procedural conditions (excessive blending) (Suppl. Video 4–6).

Hemodynamic collapse was defined as new signs and symptoms of organ hypoperfusion with systolic blood pressure drop < 90 mmHg, frequently requiring inotrope infusion and intra-aortic balloon pump insertion. Periprocedural infarction was defined according to the 2018 Fourth Universal Definition of Myocardial Infarction [11].

Low-, medium-, and high-volume operators were defined as per ACCF/AHA/SCAI 2013 [12] update of the competence statement on coronary interventions and the paper by Fanaroff et al. [13]. As complicated cases are usually taken over by senior interventionists, for the purpose of volume and experience estimation, the initial, culprit operator data was taken into account.

To account for each curve’s dissection potential, the catheter culpability index (CCI) was calculated with the following formula (aortic CCI was calculated in the same way): CCI = n/N × 1000, where n = number of dissections induced by a particular curve, and N = number of procedures in which the curve was used.

Statistical analysis

Categorical variables are presented as numbers and percentages, and compared by the Fisher exact test. Continuous variables were checked for normal distribution and expressed as means ± standard deviation, with comparisons made using Student’s t-test. Due to unequal sample sizes of the CICAAD vs. non-CICAAD cohorts, the results were verified and confirmed with Mann-Whitney U test. P-values were two-sided with a significance level of 0.05. Analyses were performed using SPSS version 24.0 (IBM Corp., Armonk, NY, USA).

Results

CICAAD incidence

Overall, 76,104 diagnostic and therapeutic coronary procedures were performed between June 2000 and September 2020, and 96 cases of coronary artery (including 2 saphenous grafts) and aortic dissection were identified. The total incidence of CICAAD was 0.126%, with aortic involvement in 0.021% (16 cases). During the 21-year timespan, dissection occurred annually in 0–3 cases per 1000 procedures (Fig. 1). No dissection caused by sheathless and guide extension catheters was found.

Figure 1. Annual number of catheter-induced coronary artery and aortic dissections per 1000 procedures.

CICAAD vs. non-CICAAD cohort

Compared to the non-CICAAD population, patients with dissection were significantly more often female (48% vs. 34%, p = 0.004), with a higher prevalence of comorbidities such as hypertension (56% vs. 25%, p < 0.001), chronic kidney disease (10% vs. 4%, p = 0.002), and prior stroke (8.3% vs. 2.8%, p = 0.006) (Table 1).

They were more likely to present with acute myocardial infarction (72% vs. 43%, p < 0.001), and undergo percutaneous coronary intervention (PCI) (85% vs. 39%, p < 0.001), with transradial access also more frequently utilized in this group (77% vs. 65%, p = 0.011). Six French catheters were significantly more often used in the procedures complicated by dissection (81% vs. 44%, p = 0.012), as opposed to 5 French, which were more commonly utilized in the non-CICAAD group (55% vs. 16%, p < 0.001).

Dissection and procedure characteristics

The right (RCA) and left (LCA) coronary artery were almost equally affected, in 47% vs. 46%, respectively (Table 2).

The right SoV was dissected in twice as many cases as the left (8 vs. 4, respectively). The ascending aorta dissection extending above SoV occurred in 8 (8%) cases, in 6 involving the right cusp.

Reassuringly, all dissections manifested acutely during index procedure, and no cases of urgent reintervention due to delayed occlusion caused by unrecognized or conservatively treated dissection were found. In 30%, dissection occurred during an off-hours procedure (Table 2). Automatic injection of contrast was used in 35% of all CICAAD, and in 46% of aortic dissections. Dissections were managed conservatively in 28%, with stent in 71%, and surgically in 2%.

Hemodynamic collapse and mortality

The in-hospital mortality rate for all dissections was 4.2%, and 6.25% for dissections involving the aorta. Four patients with CICAAD died during index hospitalization, and in 3 of them, death could be attributed to the consequences of dissection or its management.

In 15.6% of cases, hemodynamic compromise was observed (Table 2). Five patients were either admitted in shock or manifested circulatory instability due to slow flow during primary angioplasty, while in the remaining 10, the collapse could be attributed to the dissection. Of the 10 dissection--related hemodynamic collapses (6 spiral RCA occlusions, 2 zipper and 1 spiral LMCA dissection, and 1 left anterior descending artery injury), in 8 cases vessel occlusion occurred (with cardiac arrest in 5). Eight of these patients had been admitted with acute coronary syndrome.

Predisposing and precipitating factors

In 52% of CICAAD, at least one predisposing factor was identified (Table 3). Cases of unfavorable origin of a coronary artery were more often found than unfavorable arterial access route (49% vs. 7%, respectively). Among the former, small proxi­mal diameter and/or ostial atheroma were the most commonly identified (35%), whereas the most fre­quent access route obstacle was mediastinal arteries tortuosity.

The most commonly observed phenomena precipitating dissection were catheter deep seating or deep insertion during device delivery/removal (51%), catheter malalignment (34%), vessel prodding (32%), catheter systolic-diastolic mobility (27%), and respiratory instability (14%).

Mechanisms of dissection and its propagation

‘Wedged contrast injection’ proved to be the most common mechanism of CICAAD (46%) (Table 3).

It was also responsible for all cases of primary aortic or aortocoronary dissection (11 pa- tients). ‘Forceful catheter engagement’ and ‘deep catheter insertion for device delivery/retrieval’ were implicated in 28% and 26% of cases, respectively.

The overall propagation rate was as high as 30.2%. In 13.5% of CICAAD patients, the propagation was minor without flow impairment, while in the remaining 16.7%, it was serious, leading to spiral dissection, vessel occlusion, or retrograde extension into the aorta. The main causes of dissection expansion (multiple factors were often found in 1 patient) were repeat injections (62%), unchanged catheter (48%), and not immediately stenting a dissection (35%). Four out of 8 aortic dissections extending above the SoV occurred due to the progression of the original coronary artery injury. The main aortic propagation triggers (found in all such cases) were repeat injections and unchanged catheter. In 10 out of 16 aortic dissection cases, the retrograde propagation to the aorta was the consequence of downstream expansion having been blocked by a stent, chronic total occlusion, or a tight and/or calcified lesion.

Individual and institutional expertise

Most dissections (94%) were caused by high--volume interventionists (> 100 PCI/year), whose invasive experience in 84% of cases was at least 5 years (Fig. 2A, B).

Figure 2. Operator’s percutaneous coronary intervention (PCI) volume (A) and experience 12 months prior to dissection (B); C. Operator’s PCI volume 12 months prior to dissection induced by a diagnostic vs. guiding catheter.

Only in 6 cases, an operator had performed < 100 PCI in the preceding 12 months and had < 3 years of experience. Dissections during angioplasty were caused by high-volume operators in 98% of cases (Fig. 2C).

Catheter culpability index

As per CCI, the most offending curves were HS III, ART4.5, IL3.5, and HS II (for catheter names see Table 4); however, being overall rarely used, they all caused only 6% of all dissections (Fig. 3; Table 4). The lowest CCI values (< 10) were calculated for EBU4.0, BLK4.0, and Judkins Left and Right curves, with Judkins curves being the most commonly used. The highest aortic CCI was calculated for HS II, AL1, and TIG4.0 curves, with AL1 implicated in 50% of aortic injuries.

Discussion

Catheter-induced coronary and aortic dissections have not been systematically studied as far as mechanisms of injury and its propagation, risk factors, and operator’s competence are concerned. There is only anecdotal evidence on triggers and predictors of catheter-related trauma, and even the seminal papers by Dunning and Eshtehardi were based on small series of 9 and 28 patients, respectively [2, 9, 14].

Incidence

The largest study of catheter-induced coronary dissections so far reported 0.09% incidence and 5.9% mortality; findings similar to our population, with 80 isolated coronary injuries (0.105%) and 3.75% mortality [10]. A retrospective analysis of complications of diagnostic cardiac catheterization found coronary dissection in 0.003% of patients, corresponding to 0.004% in our sample (3 diagnostic angiography-related dissections) [15]. In the analyses of iatrogenic aortic dissections, triggers other than a catheter were also included, with the calculated catheter-induced injuries rate ranging from 0.017% to 0.067%, as compared to 0.021% in our material [4, 5]. The cited mortality rate for the aortic dissections was 2.7%, and 6.25% in our cohort. Dunning et al. [9] reported a 0.02% rate of iatrogenic aortocoronary injuries, which in our population amounted to 0.014% — 11 cases. In keeping with its reported vulnerability, the right SoV was dissected twice as often as the left, and the dissection propagated above the aortic root three times more often during RCA cannulation [5, 16]. A novel finding of the study is that in 62.5% of aortic dissections, the retrograde extension into the aorta occurred when the downstream progression was blocked by a stent, chronic occlusion, or a tight, calcified lesion. In the presence of such antegrade obstacles, catheter wedging is particularly dangerous because intramural contrast injection is redirected into the aortic root wall.

Figure 3. A. Catheter curves involved in dissections (%); B. Catheter Culpability Index as calculated for major catheter curves.

CICAAD vs. non-CICAAD cohort

We did not find any significant differences between CICAAD and non-CICAAD populations regarding age and body mass index. Not unexpectedly for an adverse event cohort, CICAAD patients had more comorbidities and more often presented with acute myocardial infarction. As in previous reports, we also observed a much higher rate of dissections induced by guiding as compared to diagnostic catheters (85% vs. 15%, respectively, p < 0.001), with a diagnostic catheter being responsible for only 1 out of the 16 cases of aortic involvement [2, 4, 5, 7, 10]. The relation between the access route and catheter-induced injury has not been evaluated so far, although 2 studies did observe a higher prevalence of transradial approach in procedures complicated by dissection [10, 17].

Predisposing and precipitating factors

Several types of ominous situations and dangerous catheter behavior portending dissection have been identified in the CICAAD cohort, suggesting that catheter-induced injuries rarely come unheralded. Many of them have been previously mentioned in case reports/series [2, 3, 14, 18]. Despite all the disadvantages of a retrospective analysis, only in 48% of patients were no predisposing factors found, and in just 14.6%, no precipitating factors were present, while in the rest, multiple triggers typically contributed to a dissection.

Data on the incidence of coronary artery origin variants in the general population are not available in the literature; however, with 14% and 13% rates, respectively, both the superior and inferior take-offs are likely to be overrepresented in the CICAAD group. Although not considered as predisposing to dissection per se, they may impede coaxial catheter alignment and its stable positioning. Similarly, Shepherd’s crook type RCA origin (4.3%) and coronary artery ectopy rate (3.2%), with cited incidence for the latter of 0.05–0.1%, were relatively high in our cohort.

Mechanisms of dissection and its propagation

Wedged contrast injection appears to be the commonest and most dangerous dissection mechanism. It was implicated in 46% of coronary dissections, 75% of spiral dissections, and was exclusively responsible for both primary and secondary (i.e. caused by propagation) aortic injuries. Although not always the principal mechanism of injury, intentional or inadvertent deep catheter insertion was observed in nearly half of CICAAD, suggesting that deep penetration of the vessel may facilitate subsequent dissection by its wall disruption. The mechanisms identified in our analysis, although never systematically studied before, agree with the observations of previous case series [2, 3, 14, 18].

The iatrogenic propagation rate of 30% is a disturbing finding, and although local expansion cases were also included, repeated injections or sticking to the culprit catheter led to the doubling of occlusive dissections. Moreover, out of 8 aortic dissections extending above SoV, 4 could have been avoided because they were caused by a retrograde progression. In a 2010 study of iatrogenic LMCA dissections, the reported propagation rate was also alarmingly high, reaching 32% [2].

Operator’s and center’s expertise

Both individual and institutional learning curves have been described for coronary invasive procedures, and an inverse relationship between operator’s angioplasty volume and in-hospital mortality was reported [13]. However, in most of the study cases, dissections were inflicted by high-volume operators with extensive experience. Despite equipment sophistication, the annual rate of dissections did not decrease with growing individual and institutional experience, showing an upward trend (Fig. 1). This fact, however disappointing, may at least partially be explained by more complex procedures having been performed over time, with older and sicker patients qualified to intervention.

Catheter culpability index

The most harmful catheter curves are reported in our study not just in absolute numbers but in relation to their overall use. CCI was calculated to compensate for the fact that the most frequently utilized catheters typically cause more dissections (Fig. 3). Non-conventional guides (e.g., HS or ART) are more challenging to handle and thus less commonly used, which accounts for an explosive combination of operator’s low familiarity and catheter’s high aggressiveness. Even with their infrequent usage, their CCI value was disproportionately high. Uncommonly employed curves (< 1% usage) accounted for 12% of dissections, although they comprised only 2.2% of all catheters used. Regular extra support guides like AL1 and EBU (both accounting for 14% of used catheters), despite their reported notoriety, had relatively low CCI, even though they were responsible for almost half of dissections (49%) [4, 9, 10]. Amplatz Left, however, had the second-largest aortic CCI and was implicated in nearly half of aortic dissections (7/16), as also suggested by previous studies [4, 8]. This correlation may be due to Amplatz’s frequent usage in RCA interventions and RCA’s susceptibility to retrograde dissection (Table 4) [16]. In fact, of all 18 dissections induced by Amplatz, 17 involved RCA and/or the right SoV. All culprit Amplatz Left catheters were guiding ones, and all of them were 6 French except one 5 French. Equally surprising was a good track record of bilateral catheters, infamous for popping into coronary ostia and dissection potential [3, 18]. On the other hand, despite being implicated in 37.5% of dissections, Judkins catheters had the lowest CCI thanks to their frequent utilization (73% of all used catheters).

Limitations of the study

Evaluation of uncommon complications is necessarily based on retrospective analyses. This is the largest series presented so far; however, 96 is too small a number of patients to draw definite conclusions. Despite the relative uniformity of techniques, our study may not reflect practices and equipment used elsewhere. Over the 21-year period, catheters from several manufacturers have been used, differing in material, and thus in stiffness of the shaft and tip, with slight differences also in their shape despite the same curve name, all of which may have additionally affected the dissection potential of a given catheter type. To limit the observer bias, all cases were reviewed by 2 operators, with a third interventionist called in whenever a discrepancy arose. Risk factors such as prohibitive chest vasculature, or radial artery spasm, were probably more prevalent than could be deduced a posteriori. As in any post hoc analysis, the causative role of the identified precipitating factors is speculative; however, they were repeatedly observed in the studied cohort and regularly preceded the injury.

Conclusions

It was the complexity and urgency of a procedure that decided the fate of a dissection rather than the operator’s inexperience. The significant number of serious propagations underscores how many severe complications could potentially be avoided.

Deeper understanding of the mechanisms and risk factors of catheter-induced dissections may help their prevention and enable prompt recognition and management.

Acknowledgments: The authors would like to thank following persons for their assistance in documentation and data retrieval: Grażyna Porowska, Bożena Machut, Katarzyna Bagińska, Dorota Jarocka, Mateusz Ligęza.

Conflict of interest: None declared.