Pregnancy-Associated Spontaneous Coronary Artery Dissection: Clinical Characteristics, Outcomes, and Risk During Subsequent Pregnancy

Objectives. Spontaneous coronary artery dissection (SCAD) is a common cause of pregnancy-associated myocardial infarction. Methods. This study compares the clinical course and longitudinal follow-up of 22 cases of pregnancy-associated SCAD (P-SCAD) with 285 cases of non-pregnancy SCAD (NP-SCAD) from Kaiser Permanente Northern California between September 2002 through June 2017. Results. Age in the P-SCAD group was significantly lower than in the NP-SCAD group (37.1 ± 5.7 years vs 50.9 ± 9.9 years, respectively; P<.001). Both cohorts were racially diverse, but the P-SCAD group had fewer whites (27.3% vs 50.7%; P=.03). The P-SCAD group had higher multigravidity (54.6% vs 31.4%; P=.03) and 68.2% were of advanced maternal age. The rates of ST-elevation myocardial infarction, ventricular tachycardia/fibrillation, and left main coronary dissection were similar. Proximal vessel dissection (31.8% vs 7.7%; P<.01), multiple vessel dissection (31.8% vs 9.5%; P<.01), and reduced ejection fraction at presentation (49.6 ± 10.5% vs 55.7 ± 10.4%; P=.01) were more common in the P-SCAD group vs the NP-SCAD group, respectively. More P-SCAD patients had cardiogenic shock and/or required intra-aortic balloon pump support (9.1% vs 1.1%; P=.04). Medical management was the principal coronary treatment strategy in both groups. P-SCAD patients experienced more major adverse cardiovascular events (50.0% vs 26.0%; P=.02), driven by persistent reduced ejection fraction ≤45% at follow-up (18.2% vs 5.3%; P=.04). Recurrent SCAD (18.2% vs 11.2%; P=.31) and cardiovascular death (0% vs 0.4%; P>.99) were similar in the P-SCAD group vs the NP-SCAD group, respectively. Seven patients had successful subsequent pregnancies without cardiac complications. Conclusion. P-SCAD has a higher-risk presentation, but similar long-term prognosis compared with NP-SCAD. In addition, subsequent pregnancy after SCAD may present acceptable risk.

J INVASIVE CARDIOL 2021;33(6):E457-E466. Epub 2021 May 14. doi:10.25270/jic/20.00529

Key words: MACE, outcomes, pregnancy, spontaneous coronary artery dissection

Spontaneous coronary artery dissection (SCAD) accounts for only 0.5% of hospitalizations for acute coronary syndrome, but it is a common cause of pregnancy-associated myocardial infarction, potentially accounting for 43% of acute coronary syndrome presentations in this population.^1,2 SCAD results from vaso vasorum rupture and intramural hematoma development with coronary dissection, a pathology distinct from atherosclerotic plaque rupture and thrombosis, which account for most cases of acute coronary syndrome. Hormonal influences are postulated to increase SCAD risk; hence, pregnancy, breastfeeding, multiparity, and exogenous hormones have been identified as risk factors.^3-5 Although pregnancy-associated SCAD (P-SCAD) accounts for a relatively low percentage of SCAD cases, it is associated with higher-risk presentations and worse outcomes, including more left main and multivessel involvement, cardiogenic shock, urgent percutaneous intervention (PCI), or coronary artery bypass surgery (CABG), and higher mortality.^6,7 Several case series of P-SCAD have been published, but the limited data available may be influenced by ascertainment and referral bias.^6,7 Therefore, P-SCAD was identified as a research priority in a recent scientific statement on SCAD from the American Heart Association.⁶ This study identified all SCAD patients in a large community-based practice and compared the clinical course and long-term outcomes of P-SCAD with non-pregnancy SCAD (NP-SCAD).

This is a retrospective study of all women diagnosed with SCAD at Kaiser Permanente Northern California (KPNC) from September 2002 through June 2017 with follow-up through June 2018. KPNC is an integrated healthcare system with comprehensive longitudinal follow-up for over 4 million members (approximately one-quarter of the Northern California population). This study received institutional review board approval from the Kaiser Foundation Research Institute with waiver of consent and funding from the KPNC Community Benefit Program. SCAD cases were identified through database searches for International Classification of Disease 9/10 and Current Procedural Terminology-4 codes (eg, dissection of coronary artery 414.12/I25.42, dissection of other artery 443.29/I77.9, and coronary aneurysm 414.11/I25.41), supplemented and confirmed by review of cardiac catheterization laboratory records and by referral from system cardiologists.

SCAD cases were confirmed by an independent review of angiograms and consensus diagnosis by 2 interventional cardiologists. A third cardiologist independently adjudicated discrepancies to confirm SCAD diagnoses. Angiographic criteria included contrast dye staining or multiple lumens consistent with dissection, abrupt diffuse vessel narrowing demonstrating intramural hematoma or stenosis (without evidence of atherosclerosis), and intravascular ultrasound confirmation per operator discretion.⁸ Patients with no coronary angiogram available for review were excluded, and patients with loss of KPNC membership for >6 months were considered lost to follow-up and censored.

Data collected included demographics, clinical presentation, risk factors, treatments, and long-term outcomes. P-SCAD was defined as during pregnancy or within 12 months postpartum, including during viable pregnancies, after miscarriage, or after elective termination of pregnancy. Multigravidity was defined as gravidity ≥4 and multiparity as parity ≥4.⁶Tobacco use was defined as current if tobacco was smoked within the preceding 12 months, and former for those who last smoked >12 months preceding the index SCAD. Active illicit drug use was defined as within 24 hours of the index SCAD and current drug use as within 12 months. Fibromuscular dysplasia (FMD) was defined by the pathognomonic “string-of-beads” appearance (ie, alternating narrowing and aneurysms) on femoral, carotid, and/or renal angiography, computerized tomography, or magnetic resonance angiography. Connective tissue disease included Marfan syndrome, Loeys-Dietz syndrome, and Ehlers-Danlos syndrome. Autoimmune disease included systemic inflammatory conditions (eg, systemic lupus erythematosus, rheumatoid arthritis, inflammatory bowel disease, and vasculitis).⁴Post menopause was defined as >12 consecutive months of amenorrhea.

Primary outcomes were the composite of major adverse cardiovascular events (MACE), including early propagation of the index SCAD, later recurrent de novo SCAD, persistent reduced left ventricular ejection fraction (LVEF) <45%, and cardiovascular mortality. Propagated SCAD was defined as dissection extension in the index SCAD vessel within 30 days of the index event; recurrent de novo SCAD was defined as dissection affecting a different vessel segment occurring >30 days after the index event.⁶ Propagated or recurrent SCAD was suspected if a new troponin rise occurred at least 1.5 times greater than the last available troponin level in the presence of angina pectoris. Angiographic confirmation of propagated or recurrent SCAD was reviewed when available. LVEF was measured by the most recent follow-up imaging study available (either left ventriculography in the cardiac catheterization laboratory or subsequent echocardiography).

Secondary outcomes included all-cause mortality, heart failure hospitalizations, the last measured LVEF, implantable cardioverter-defibrillator placement, additional PCI or subsequent CABG, and number of subsequent stress tests and emergency department (ED) visits or hospital admissions for cardiac issues after the index SCAD.

Statistical analysis. Statistical analysis was performed with SAS, version 9.4 (SAS Institute). Categorical variables were analyzed using the Chi-square or Fisher’s exact test, and continuous variables were analyzed with Student’s t-test. Rank-sum test was used to compare median follow-up. P-values <.05 were considered statistically significant. Cox proportional hazard models were used to evaluate predictors of outcome differences.

We identified 307 patients, 22 with P-SCAD and 285 with NP-SCAD (Table 1). The average age of the P-SCAD group was significantly lower (37.1 ± 5.7 years vs 50.9 ± 9.9 years; P<.001), as expected among childbearing women. Both cohorts were racially diverse, but there were significantly fewer white patients in the P-SCAD group than in the NP-SCAD group (27.3% vs 50.5%; P=.03). The P-SCAD cohort included 27.3% Asian, 13.6% Black, and 27.3% Hispanic patients, with no significant difference by subgroup analysis by race. The rates of hypertension, hyperlipidemia, diabetes mellitus, smoking status, and illicit drug use (ie, cannabis, cocaine, and methamphetamines) were similar (Table 1). The rates of migraine, depression, and anxiety were comparable, as were identifiable physical or emotional triggers (Table 1). Approximately half of all patients underwent some form of FMD screening; of those screened for FMD, 18.2% were diagnosed with FMD in the P-SCAD group vs 27.7% in the NP-SCAD group (P=.50).

The P-SCAD group had a significantly higher rate of multigravidity (54.6% vs 31.4%; P=.03). The mean gravidity of the P-SCAD cohort was 4.1 ± 2.0, which is significantly higher than in the NP-SCAD group (2.6 ± 1.9; P<.01). The rates of oral contraceptive pill and hormone therapy use were similar. More patients underwent fertility treatments in the P-SCAD group than in the NP-SCAD group (9.1% vs. 1.4%, respectively), but the difference was not significant (P=.08). Fertility treatments included in vitro fertilization for 2 P-SCAD patients and use of clomiphene or intrauterine insemination 2-6 years before SCAD presentation for 4 patients in the NP-SCAD cohort. There were no newborn complications (eg, preterm delivery, low birth weight) or fetal mortalities.

The P-SCAD cohort included 4 patients presenting with SCAD during the first trimester (1 spontaneous abortion, 3 therapeutic abortions) and 18 patients who experienced SCAD after delivering viable infants (Figure 1). The average age was 37.7 ± 5.7 years at delivery; 90.9% were older than age 30 and 68.2% were older than age 35. Twelve patients had vaginal deliveries and 6 had caesarean sections (C-sections). Eight of the patients (36.4%) had high-risk features during pregnancy, including 2 patients with gestational diabetes, 3 with gestational hypertension, and 5 with pre-eclampsia (resulting in 1 induced delivery and 1 urgent C-section). Most P-SCAD events occurred in the early postpartum period, with 83% within the first month after delivery. Sixteen patients were known to be actively breastfeeding.

The rates of ST-segment elevation myocardial infarction and ventricular tachycardia or fibrillation (VT/VF) were similar (Table 2). More involvement of the left main and left anterior descending (LAD) coronary arteries was observed in the P-SCAD group, but this was not statistically significant. There was significantly more involvement of proximal vessels (31.8% vs 7.7%; P<.01) and multiple vessels (31.8% vs 9.5%; P<.01) in the P-SCAD group vs the NP-SCAD group, respectively. The LVEF was significantly lower in the P-SCAD group at initial presentation (49.6 ± 10.5% vs 55.7 ± 10.4%; P=.01). More P-SCAD patients presented with cardiogenic shock and/or required hemodynamic support with an intra-aortic balloon pump (9.1% vs 1.1%; P=.04).

The treatment strategy was similar in both groups, with the majority treated medically without revascularization (Table 2). At presentation, the patients in the P-SCAD cohort were on minimal medications (none were prescribed aspirin for pre-eclampsia prevention). At discharge, the majority (>70%) of both groups were prescribed aspirin, thienopyridines, beta-blockers, and statins (Table 3). The rates of balloon angioplasty, PCI, and CABG were similar (Table 2). Only 3 patients required urgent CABG (1 P-SCAD patient for inability to deploy a LAD stent for multivessel propagated SCAD, 1 NP-SCAD patient after attempted PCI of propagated LAD SCAD, and 1 NP-SCAD patient for left main SCAD).

The P-SCAD cohort experienced a significantly higher rate of the primary composite MACE outcomes (50.0% vs 26.0% in the NP-SCAD cohort; P=.02) (Table 4), which was driven by a significantly higher proportion of patients with persistent reduction of LVEF at follow-up (18.2% vs 5.3% in the NP-SCAD cohort; P=.04). Early propagation of the index SCAD episode was comparable (27.3% in the P-SCAD cohort vs 13.3% in the NP-SCAD cohort; P=.10), and most early propagations were confirmed by angiography. The rate of recurrent SCAD in both groups was comparable (18.2% in the P-SCAD cohort vs 11.2% in the NP-SCAD cohort; P=.31). Use of beta-blocker medications at the time of propagated or recurrent SCAD was similar in both cohorts. Seven patients, all in the NP-SCAD cohort, had more than 1 episode of recurrent SCAD.

Cardiovascular death (0% in the P-SCAD cohort vs 0.4% in the NP-SCAD cohort; P>.99) and all-cause mortality rates were low (0% in the P-SCAD cohort vs 1.4% in the NP-SCAD cohort; P>.99), and all deaths occurred in the NP-SCAD cohort (Table 4). Only 1 patient had a SCAD-related cardiovascular death due to a third SCAD recurrence complicated by VT/VF arrest and anoxic brain injury. The other 3 deaths were related to sepsis, metastatic lung cancer, and end-stage chronic obstructive pulmonary disease. Although MACE outcomes among P-SCAD patients were driven by an LVEF <45% at follow-up, on average the ejection fraction improved to normal (LVEF 54.6 ± 7.7% at an average of 3 years after the index SCAD), and subsequent heart failure admissions was comparable to the NP-SCAD cohort (Table 4).

The P-SCAD patients required more subsequent PCI after the index SCAD event (22.7% vs 8.1% in the NP-SCAD cohort; P=.04), mostly driven by in-stent restenosis (Table 4). Patients who received PCI (including plain old balloon angioplasty or stents) during their index presentation often needed additional stents for propagated SCAD, subsequent stents for in-stent restenosis, or even subsequent CABG; however, the long-term event rates including recurrent SCAD and death were low (Table 5). The rates of subsequent CABG were low; 2 NP-SCAD patients underwent subsequent elective CABG for refractory chest pain and multiple episodes of in-stent restenosis. The 3 implantable cardioverter-defibrillator placements were all in the NP-SCAD cohort — 1 for persistent severe systolic dysfunction and 2 following cardiac arrest at index SCAD presentation. Resource utilization is high after a SCAD diagnosis; about half of the patients in both cohorts received subsequent stress tests, repeat ED visits, or hospital admissions, with no difference in utilization between cohorts (Table 4). For our entire cohort, the average number of ED visits or hospitalizations was 3.2 ± 4.1, and 19.2% of patients had ≥3 post-SCAD visits. In addition, 12.1% of patients received myocardial perfusion studies, which substantially increases radiation exposure risk in this relatively young population.

Bivariate analysis of patients with and without recurrent SCAD showed that patients with multiple vessel involvement at index SCAD presentation were more likely to have recurrent SCAD (22.2% vs 9.6%; P=.04). Sixteen patients with <6 months of in-plan drug coverage were excluded from the prescription medication multivariable analysis. Patients with recurrent SCAD were less frequently prescribed thienopyridines (53.1% with recurrent SCAD vs 62.9% with no recurrent SCAD; P=.04). Cox multivariable modeling of our data did not identify any other variables associated with recurrent SCAD rates.

Image

This study also describes 7 patients who had uncomplicated pregnancies after an index SCAD event, including 6 vaginal births and 1 C-section delivery (Figure 2). The average age of these patients was advanced (40.8 ± 5.2 years), and many had high-risk features requiring management by cardiologists and maternal-fetal medicine specialists. Two patients with prior P-SCAD had subsequent pregnancies, including patient #1 with postpartum hypertension and patient #6 who received in vitro fertilization at age 48 years with a course complicated by recurrent SCAD of a left circumflex coronary branch that was medically managed. Patient #2 had high-risk features including previous multivessel and propagated SCAD, multivessel FMD, advanced maternal age, and prior gestational hypertension. Her management included a prepartum normal surveillance coronary angiogram, peripartum aspirin and labetalol, and planned induction resulting in an uneventful delivery. High-risk features in Patient #3 included left main stenting for prior SCAD related to methamphetamine use, FMD, multigravidity, and a post-SCAD diagnosis of systemic lupus erythematous. She had a positive methamphetamine toxicology test during pregnancy and was not treated with beta-blockers, but successfully delivered twins. Patient #4 had an uneventful post-SCAD pregnancy managed on aspirin and metoprolol, but 7.9 years later required an implantable cardioverter-defibrillator placement for VT attributed to restenosis of a stent placed during her index SCAD event. Six of these patients chose to breastfeed, including 1 who breastfed for 3 years. Of note, patient #8 in our study elected for therapeutic abortion after counseling with her physician since her index SCAD event occurred the year prior.

Pregnancy-associated SCAD accounts for only 2%-8% of all SCAD cases, but is associated with a worse prognosis.^6,7,9 Higher-risk features of P-SCAD are thought to be related to hormonal effects on arterial vasculature and hemodynamic stress associated with pregnancy. Elevated progesterone can weaken the collagen and elastin strength of the coronary artery media, and estrogen can compromise the arterial vaso vasorum.^10,11 In this study, fertility treatment rates in the P-SCAD cohort were high for a childbearing population. The impact of therapeutic hormone exposure with in vitro fertilization on the incidence of P-SCAD compared with intrauterine insemination or clomiphene treatments received in the NP-SCAD group is speculative and cannot be determined by our limited data. Multigravidity was higher in the P-SCAD cohort, and multiple pregnancies can cause progressive vessel wall degeneration over time.⁹ Hemodynamic changes in pregnancy, including an increase in plasma volume by 30%-50% and cardiac output to 10 L/min during labor, can further increase risk.¹² Although the majority of P-SCAD has been reported in the third trimester or post partum, there is a risk of P-SCAD early in pregnancy. One case report describes P-SCAD at 10 days of gestation.¹³ Our study identified cases of very early antepartum SCAD within the first 21 days of gestation: for 2 of these patients, the urine human chorionic gonadotropin test was negative at the index SCAD presentation but resulted positive when repeated shortly thereafter. In addition, higher rates of a history of gestational diabetes, gestational hypertension, and pre-eclampsia were seen in the P-SCAD cohort (40.9% of P-SCAD vs 1.4% of NP-SCAD patients with parity ≥1), and these conditions may further increase cardiovascular risk during pregnancy.^14,15 Pre-eclampsia, P-SCAD, and peripartum cardiomyopathy have been hypothesized to share similar pathophysiology.^9,16 As for other predictors of P-SCAD, one study identified more hypertension, hyperlipidemia, and smoking history in P-SCAD patients compared with a baseline pregnant population.¹⁷ A previous case-control study (including this population) also identified significant hyperlipidemia in SCAD patients.¹⁸ Maternal age >30 years has been reported as a significant P-SCAD risk factor, consistent with our findings.¹⁷

Our data confirm that P-SCAD patients have higher-risk presentations than NP-SCAD patients, with more proximal and multiple vessel involvement and cardiogenic shock. However, the rates of ventricular arrhythmia, cardiogenic shock, mechanical circulatory support, and urgent CABG were substantially lower by at least 2 fold than reported previously (4.6% vs 9%-16%, 9.1% vs 20%-24%, 9.1% vs 28%, and 4.5 vs 26%-37%, respectively).^{7,17, 19} This difference may be explained by lower rates of ST-segment elevation myocardial infarction and left main coronary involvement in our P-SCAD cohort (36% vs 64%-69% and 5.0% vs 36%, respectively).^7,17 Another possibility is ascertainment and publication bias of more complicated cases reported from tertiary-care centers compared with our community cohort.²⁰ In addition, the increase in published SCAD research over the past decade may have led to earlier recognition and improved care, thereby decreasing adverse outcomes.²¹

The long-term prognosis for P-SCAD patients after index hospitalization seems favorable. Three of our P-SCAD patients had scheduled surveillance angiograms performed in the absence of symptoms, all of which showed healed vessels or minimal residual dissection. Other studies have shown lower rates of healed vessels (47.8%-65.2%) on surveillance angiograms for P-SCAD.^7,22 It is unclear whether selection bias or timing differences after index SCAD accounts for these differences. The risk of recurrent de novo SCAD was similar in our P-SCAD and NP-SCAD cohorts, consistent with other reports.⁹ The recurrence rate for our entire cohort was 11.7%, similar to published rates of recurrent de novo SCAD, ranging from 11%-17%.⁶ Beta-blockers decreased the risk of recurrent SCAD (hazard ratio, 0.36) in a prospective study, and 67% of our patients with recurrent SCAD were on a beta-blocker.²³ For SCAD patients presenting with ventricular arrhythmias, the prognosis may be favorable, with low risk for recurrent arrhythmia.²⁴

Counseling women with SCAD regarding the risk of subsequent pregnancy is challenging given the extremely limited data available.²⁵ The CARPREG risk index and mWHO classification do not specifically address cardiac risk for SCAD patients.^26,27 In the only published case series of patients with subsequent pregnancy after SCAD, 1 of 8 patients experienced recurrent SCAD post partum and required CABG.²⁸ Our series includes 7 successful post-SCAD pregnancies, with 1 patient experiencing recurrent SCAD medically managed in the setting of in vitro fertilization and advanced maternal age, which elevated her risk. Our post-SCAD pregnant patients had less severe index SCAD presentations (no previous VT/VF or reduced LVEF), and 43% were on aspirin and beta-blocker medications during the post-SCAD pregnancy, which may have mitigated risk. Although the number of patients is small, our data suggest that future pregnancy may not be absolutely contraindicated in SCAD patients. More data are needed to allow valid risk stratification and knowledgeable counseling for SCAD patients contemplating pregnancy. A multidisciplinary approach with cardiology and maternal-fetal medicine specialists to plan medical therapy and mode of delivery may be beneficial.^29,30 Lastly, the impact of breastfeeding on SCAD risk is unclear, but SCAD during lactation has been described.⁹ However, we had high rates of breastfeeding in our P-SCAD cohort (16 out of 17) and in our patients with pregnancies after SCAD (6 out of 7) without recurrent SCAD during the breastfeeding period.

Strengths and limitations. The strengths of this study include the large integrated system of KPNC, wherein the membership accounts for at least a quarter of the population in Northern California. The integrated electronic medical record at Kaiser Permanente allows for systematic assessment of longitudinal outcomes and comprehensive follow-up data. The membership retention rate is high; over a third of our cases had at least 5 years of follow-up, and 10% had more than a decade of follow-up. Because this was a community-based population study, it may have less referral bias than large single-center referral cohorts, and a thorough database query was performed to identify every patient with SCAD in KPNC. This is also the most racially and ethnically diverse SCAD cohort described in the literature.

The limitations include the retrospective study design. SCAD patients who received emergency care outside of KP facilities may not have been found by queries to the in-system electronic medical records, and late events related to SCAD may have occurred after loss of health plan membership. Furthermore, some cases of recurrent de novo SCAD were not confirmed with angiography, so it is possible that these cases were misdiagnosed. If coronary thrombosis was misclassified as recurrent dissection, the result suggesting that thienopyridine use is protective against recurrent SCAD may be erroneous. The prevalence of FMD may be underestimated, given that only 46% of patients were screened, and screening protocols were inconsistent. We were unable to obtain detailed pregnancy histories (eg, gravidity, parity, history of pre-eclampsia) for 4.5% of our cohort, but these patients were all >40 years old and in the NP-SCAD cohort. Fertility treatment data may be incomplete because they were ascertained from obstetrics chart documentation and not confirmed by patient contact.

P-SCAD patients have higher-risk initial presentations than NP-SCAD patients. However, the long-term prognosis for P-SCAD patients is favorable and comparable to that of NP-SCAD patients. Further data are needed to accurately determine risk for SCAD patients who undergo subsequent pregnancy. Currently available data suggest that post-SCAD pregnancy may not be absolutely contraindicated.

Acknowledgment. Naomi L. Ruff of RuffDraft Communications edited the manuscript.