Biomarkers of Endothelial Dysfunction in Cocaine Overdose and Overdose-Related Cardiovascular Events

Abstract

Overdose of stimulant drugs has been associated with increased risk of adverse cardiovascular events (ACVE), some of which may be ascribed to endothelial dysfunction. The aims of this study were to evaluate biomarkers of endothelial dysfunction in emergency department (ED) patients with acute cocaine overdose and to assess the association between in-hospital ACVE in ED patients with any acute drug overdose. This was a prospective consecutive cohort study over 9 months (2015–16) at two urban, tertiary-care hospital EDs. Consecutive adults (≥18 years) presenting with suspected acute drug overdose were eligible and separated into 3 groups: cocaine (n=47), other drugs (n=128), and controls (n=11). Data was obtained from medical records and linked to waste serum specimens, sent as part of routine clinical care, for biomarker analysis. Serum specimens were collected and analyzed using ELISA kit assays for three biomarkers of endothelial dysfunction: (a) endothelin-1 (ET-1), (b) regulated upon activation normal T-cell expressed and secreted (RANTES), and (c) soluble inter-cellular adhesion molecule-1 (siCAM-1). Mean siCAM was elevated for cocaine compared to controls and other drugs (p<0.01); however, mean RANTES and ET-1 levels were not significantly different for any drug exposure groups. Receiver operating characteristics curve analysis for prediction of in-hospital ACVE revealed excellent performance of siCAM-1 (area under curve 0.86, p<0.001), but lack of predictive utility for either RANTES or ET-1. These results suggest that serum siCAM-1 is a viable biomarker for acute cocaine overdose and that endothelial dysfunction may be an important surrogate for adverse cardiovascular events following any drug overdose.

Introduction

The current drug overdose epidemic has been declared a national emergency in the US.¹ Patients who present to the Emergency Department (ED) with acute drug overdose commonly experience adverse medical consequences, including an increased risk of adverse cardiovascular events (ACVE).² Abuse of cocaine in particular has been associated with the increased risk of developing adverse cardiovascular medical consequences, including accelerated atherosclerosis, myocardial infarction, sudden cardiac death, and ischemic stroke.³ Some cardiovascular events, such as myocardial infarction, shock, or cardiac arrest, may potentially be ascribed to endothelial dysfunction, thrombus formation, or vasoconstriction.^4–5

Previously, it has been shown that chronic cocaine use produces elevation in several serum markers of endothelial dysfunction.^6–8 Candidate biomarkers for endothelial dysfunction have previously included endothelin-1 (ET-1),⁶ Regulated upon Activation Normal T-cell Expressed and Secreted (RANTES),⁷ and Soluble Inter-Cellular Adhesion Molecule-1 (siCAM-1).⁸ Elevations of these biomarkers in the setting of acute drug overdose, however, has not been demonstrated before.

In this study, we evaluated Emergency Department (ED) patients for the association of endothelial dysfunction with cocaine drug overdose (primary aim) and with overdose-related ACVE (secondary aim). We intended to include polysubstance overdose in the present study to be generalizable to “real-world” ED populations, since poly-substance use tends to be the norm in both cocaine-use and opioid-use disorder populations. For the primary aim, we hypothesized that biomarkers of endothelial dysfunction would elevate following acute cocaine overdose; furthermore, we also hypothesized that these biomarkers may predict in-hospital occurrence of drug-overdose related ACVE. To test this hypothesis, we studied ED patients with acute drug overdose and compared mean serum RANTES, siCAM-1, and ET-1 levels among cocaine exposures, non-cocaine drug exposures, and non-drug control ED patients. For the secondary aim, we hypothesized that endothelial biomarkers would also predict ACVE outcomes for this entire patient population regardless of drug of exposure. To test this hypothesis, we calculated receiver operating characteristics (ROC) and test characteristics of the endothelial biomarkers for any in-hospital occurrence of ACVE.

Methods and Methods

Study Design and Population

This was a prospective cohort study at two urban, tertiary-care hospitals over a 9-month period (June 2015 – March 2016). Consecutive adult (≥18 years old) patients who presented to the Emergency Department (ED) with a chief complaint of acute drug overdose were eligible, and were included in the present study if the hospital clinical laboratory had any available waste serum specimens from the ED visit that were drawn as part of routine clinical care. Exclusion criteria were children (age <18), missing outcome data (left against medical advice or elopement), absence of waste serum specimens, prisoners, exclusively dermal exposures, do-not-resuscitate orders, or chronic poisoning (i.e., not acute). The study protocol was approved by the Program for Protection of Human Subjects with waiver of informed consent and authorization at the study institutions.

Study Protocol

ED patients with a chief complaint relating to drug overdose were screened for eligibility by research assistants. After applying inclusion/exclusion criteria, prospective data collection was gathered from the ED visit (including pre-hospital/ambulance information) up until the time of ED/hospital discharge, in-hospital death, or medical clearance (i.e., for transfer to the psychiatry service). Hospital medical record data collection was performed by research assistants trained in medical abstraction and recorded using standardized data collection forms. After the time of ED/hospital discharge or medical clearance, patients had no further follow-up.

Drug Exposure Groups

The “cocaine group” consisted of ED patients with suspected acute cocaine overdose with exposure confirmed by urine toxicology testing. The “other drug group” consisted of ED patients with suspected acute drug overdose with exposures confirmed by either serum or urine routine toxicology testing including the absence of cocaine exposure. The “control group” consisted of ED patients initially suspected to have drug overdose but who had negative serum/urine routine toxicology results and were later determined to have alternative diagnoses according to the discharge diagnosis upon medical records review. Multi-drug use with confirmed cocaine exposure were classified into the “cocaine group”. All patients had urine cocaine testing as part of routine clinical care. For other drugs not present in routine serum/urine toxicology testing, patient history was used to classify patients into the “other drugs” group.

Data Collection

Data collection from medical records was prospectively performed by trained research assistants who were blinded to the study hypothesis. These data included demographics (gender, age, race), toxin identification (details from the medical record, serum drug concentrations if available), and toxicology screens (urine ELISA panel and serum concentration, if any). Blood and urine toxicology screen results sent as routine part of clinical care (i.e., no confirmatory GC/MS) were recorded in order to confirm exposure, and cocaine exposure was confirmed with positive urine toxicology screens in all “cocaine-group” patients. Data collection from the medical chart occurred in accordance with accepted guidelines for valid medical chart abstraction,⁹ including training of abstractors and 95% agreement of a random sampling of ten test charts prior to mass data abstraction.

Endothelial Biomarkers

Waste serum specimens were initially drawn at the bedside on ED presentation as part of routine clinical care. This was typically performed by antecubital puncture with a 20-gauge needle performed by trained and credentialed nursing staff. After clinical specimens were analyzed based on tests ordered by the clinical team, but before specimens were routinely discarded, waste samples were collected by research staff and de-identified with a study ID by laboratory staff familiar with the protocol in the clinical laboratory. Samples were then centrifuged, and serum was aliquoted to cryotubes and stored frozen at −80C for later use.

Soluble intercellular cell adhesion molecule (sICAM) was measured by enzyme-linked immunosorbent assay (ELISA), incubating platelet-poor serum in a microwell plate coated with monoclonal antibodies to human sICAM-1a. Horseradish peroxidase-conjugated anti-human sICAM-1 monoclonal antibody was used to reveal the reaction, and serum levels were calculated by reading the absorbances from a standard curve generated with purified sICAM (Bender Med Systems, CA).

Regulated on activation normal T cells expressed and secreted (RANTES) was measured using solid phase sandwich ELISA using monoclonal antibodies and recombinant RANTES as the standard. The sensitivity of this assay is 6.6 pg/ml (R&D Systems, MN).

Endothelin-1 (ET-1) was measured following the specified procedures for ELISA kit of ET-1 (R&D Systems, Inc., Minneapolis, MN). Standards of different concentration, control, and the samples extracted were assayed simultaneously in duplicate. The within-run precision correlation of variation was 5%.

Clinical Outcomes

In-hospital occurrence of ACVE following acute drug overdose was the clinical outcome and focus of the secondary aim. ACVE has been validated previously² as the occurrence of at least one of the following: (A) myocardial injury (troponin elevation at any time during hospitalization), (B) shock (hypotension requiring vasopressors), (C) ventricular dysrhythmia (ventricular tachycardia or VT, ventricular fibrillation or VF, Torsade de Pointes or TdP), and (D) cardiac arrest (loss of pulses requiring chest compressions). Myocardial injury was defined as any elevation of the traditional cardiac biomarker cardiac troponin I, which was measured using Bayer reagents (Bayer Healthcare, Cambridge, MA) on the Bayer Advia Centaur analyzer, and the standard cutoff concentration was used (0–0.09 ng/ml, detection limit 0.01 ng/ml, 10% imprecision, 99th percentile cutoff >0.1 ng/ml). Patients without a laboratory evaluation of troponin I were recorded as “missing” in the database to avoid misclassification bias, but for the purposes of ACVE coding the missing value was assumed to be negative. Inpatient rhythm strips in the medical record of inpatients receiving telemetry monitoring were reviewed by a blinded cardiologist to evaluate the alarmed segments for occurrence of ventricular dysrhythmia. ACVE was abstracted based on information from electronic medical records (laboratory results, discharge summary, discharge diagnosis), pharmacy records (dispensed and administered medications), alarmed rhythm strip segments (printed daily as part of routine clinical care if patient admitted to telemetry units, interpreted by a blinded cardiologist), and electronic billing records (diagnosis codes, current procedural terminology codes). These records were reviewed for all enrolled patients and recorded using standardized data collection forms.

Statistical Analysis

Sample size was calculated a priori for the primary analysis (cocaine-biomarker associations); assuming baseline prevalence of 20% cocaine exposure in the overdose population, we calculated the need to analyze 155 patients to have 80% power to detect an effect size of 25% mean difference between groups. The t-test was calculated with 5% alpha (2-tailed) for differences between mean biomarker levels across all drug exposure groups as follows: (a) cocaine compared to the other drugs group; (b) cocaine compared to the control group; and (c) cocaine compared to the entire cohort (i.e. other drugs plus controls). Optimal biomarker cutpoints for endothelial biomarkers were determined using the Youden index of the area under the curve (AUC) from receiver operating characteristics (ROC). ROC analysis was performed by pooling the “cocaine group” with the “other drugs group” into one large cohort, and controls were excluded for ROC analysis (drug overdose was the intended study population for which the chosen study outcomes were relevant). We calculated 95% confidence intervals around AUC values and odds ratios (OR) using the estimated standard error method. Statistical analysis was performed with STATA v.16 computer software (StataCorp, College Station, TX).

Results

Patient Enrollment

During the study period, 492 patients were screened, of whom 101 were excluded (prisoner 34, missing data 29, dermal 19, children 14, do-not-resuscitate 3, chronic 2), and 205 had no waste specimens, thus resulting in 186 patients available for biomarker analysis with the following demographics: mean age 42.1 years, 57.5% male, 45.2% Hispanic, 27% White, 12.4% Black, 7.0% Asian, 8.6% other/unknown.

Drug Exposure Groups

Confirmed drug overdoses among study patients included 47 cocaine exposures (3 cocaine only), 128 other drug exposures, and 11 controls. The “other drug” exposure group included 59 opioids (6 opioids only), 26 non-cocaine stimulants, and 22 cannabinoids (4 cannabinoids only). Table 1 demonstrates demographic breakdowns of the 3 main exposure groups (cocaine, other, control); there were no substantial demographic differences between the three groups. Multi-drug exposures were reported in the majority (74%) of study patients including 93.6% of the cocaine group. For multi-dug exposures in the cocaine group, additional drugs were the following: 24 opioids, 11 non-cocaine stimulants, 10 cannabinoids, and 28 other.

Table 1.

Baseline Characteristics of the Drug Exposure Groups

Characteristic	Cocaine	Other Drugs	Control
Demographic
Age (years)	41.6 (14.7)	42.1 (16.4)	41.8 (14.8)
Female	15 (32)	59 (46)	5 (45)
Race/Ethnicity
White	6 (13)	41 (32)	3 (27)
Black	8 (17)	13 (10)	2 (18)
Asian	2 (4)	9 (7)	2 (18)
Hispanic	26 (55)	55 (43)	3 (27)
Other/Unknown	5 (11)	10 (8)	1 (9)
TOTAL N	47	128	11

Each cell lists the raw N (%) for categorical characteristics or means (SD) for the continuous characteristic.

Abbreviations: N=number; SD=standard deviation.

Outcomes

Overall, ACVE occurred in 15 patients (12 myocardial injury, 4 shock, 3 ventricular dysrhythmia, 2 cardiac arrest) and there were 2 deaths (both included as ACVE). Categorizing ACVE by drug exposure group, 12.8% of cocaine group patients had ACVE (6 myocardial injury, 1 shock, 0 ventricular dysrhythmia, 0 cardiac arrest), 5.4% of other drug group patients had ACVE (4 myocardial injury, 2 shock, 3 ventricular dysrhythmia, 2 cardiac arrest), and 18.2% of control group patients had ACVE (2 myocardial injury, 1 shock, 0 ventricular dysrhythmia, 0 cardiac arrest).

Endothelial Biomarkers

Mean RANTES (pg/ml) was 6536 for controls, 8043 for cocaine, and 8311 for other drugs (all p=NS). Mean ET-1 (pg/ml) was 24.8 for controls, 20.4 for cocaine, and 21.6 for other drugs (all p=NS). Mean siCAM (ng/ml) was 209.1 for controls, 1328.2 for cocaine (p<0.01 vs. other drugs, p=0.09 vs. controls, p<0.01 vs. entire cohort), and 229.8 for other drugs (p=NS compared to controls). Comparison of mean levels for all biomarkers is illustrated in Figure 1.

Figure 1. Mean biomarker levels in each drug exposure group.

Bars show mean levels (with lines showing standard errors) for all three biomarkers comparing patients with cocaine (middle columns), and other drugs (right columns) compared to controls (left columns).

Abbreviations: pg=picogram; ng=nanogram; ml=milliliter

Receiver Operating Characteristics (ROC)

In order to determine the extent to which biomarkers of endothelial dysfunction were viable predictors of in-hospital ACVE occurrence irrespective of drug exposure, we plotted results on ROC curves for the overdose cohort (cocaine group and other drugs group, excluding controls) with respect to study outcomes. The ROC curves (Figure 2) reveal excellent AUC for prediction of ACVE using the siCAM-1 (c=0.86, CI 0.76–0.97, p<0.01) but not ET-1 (c=0.44, CI 0.23–0.65) or RANTES (c=0.57, CI 0.39–0.75). The optimal siCAM-1 cutpoint (Youden index) for prediction of ACVE was 311.4 ng/ml. This cutpoint was 75% sensitive and 86.1% specific for ACVE, and conferred 14-fold increased odds of occurrence of ACVE (OR 14.0, CI 4.1–48.1).

Figure 2. ROC curves to evaluate whether endothelial biomarkers can predict ACVE.

Dotted lines represent ROC curves for each biomarker to predict ACVE outcomes for siCAM (red line, top), RANTES (green line, middle), and ET-1 (blue line, bottom). * Represents the point on the ROC curve that is the optimal cutpoint of siCAM for prediction of ACVE; above 311.4 ng/ml, siCAM was 75% sensitive and 86.1% specific.

Abbreviations: ACVE=adverse cardiovascular events; AUC = area under the curve; ROC = receiver operating characteristics.

Discussion

This is the first study to demonstrate endothelial dysfunction in ED patients with acute cocaine overdose, specifically detected by elevations in serum siCAM-1. In addition, this is the first study to demonstrate that endothelial dysfunction, detected by elevations in serum siCAM-1, was also strongly associated with occurrence of in-hospital ACVE in ED patients with acute drug overdose, regardless of drug class. Importantly, siCAM-1 was the only endothelial biomarker that demonstrated either a specific elevation in the cocaine overdose group or the ability to predict occurrence of in-hospital ACVE.

The finding that siCAM-1 was elevated following cocaine overdose in this study is consistent with, and builds upon, prior data regarding endothelial dysfunction due to cocaine use.^3–5,7–8 Intercellular adhesion molecule 1 (ICAM-1), also known as CD54, binds the leukocyte integrins LFA-1 and Mac-1.¹⁰ ICAM-1 expression is weak on leukocytes, epithelial and resting endothelial cells, as well as some other cell types, but expression can be stimulated by IFN-gamma, TNF-alpha, IL-1 beta and LPS.¹⁰ Soluble ICAM-1 is found in a biologically active form in serum, probably as a result of proteolytic cleavage from the cell surface.¹⁰ siCAM-1 has been associated with altered vasoreactivity, blunted vasodilation, and chronic cocaine use.^8,11 The association of siCAM with ACVE in the present study is a novel finding that requires further study and replication in larger validation studies. If confirmed, the use of siCAM as a predictor of ACVE in drug overdose patients could add substantially to the current clinical toolkit for acute care clinicians who take care of drug overdose patients.

The inability of the present study to show any association between ET-1 and cocaine overdose was somewhat surprising, given that prior animal and human studies have shown an association with chronic cocaine use.^4,6 Previously, it has been shown that chronic cocaine consumption alters several functions of the endothelium towards a pro-thrombotic condition and that some of those functions remain abnormal even after short-term drug withdrawal.⁴ Moreover, in vitro⁵ and in vivo⁶ animal studies demonstrated a role for ET-1 as a correlate of endothelial dysfunction following chronic cocaine exposure. Additionally, an association between cocaine abstinence and lowered ET-1 levels suggested ET-1 could be used as a marker for reduction in cocaine use.⁷ Mechanistically, this peptide is a potent vasoconstrictor and is produced by vascular endothelial cells. It is possible therefore, that chronic use, rather than the setting of an acute drug overdose, is required to generate an appreciable elevation in ET-1. An alternative explanation is that the present study was simply underpowered to show a difference. Similarly, while mean ET-1 levels were lower for the cocaine group compared to controls, this finding was underpowered to show a significant difference. Future study, therefore, is warranted to examine a distinct sample across multiple time points to potentially detect delayed ET-1 elevation.

Though there was not a significant elevation in RANTES following cocaine overdose, it was lowest in controls and highest in other drug overdoses. Also known as CCL5 and SISd, RANTES is a chemokine that belongs to the CC chemokine subfamily. It is a secreted polypeptide produced by a wide variety of epithelial and mesodermal cell types and is involved in both cellular and humoral immunity. Prior studies had shown associations between RANTES and cocaine dependence.⁸ Again, it is possible that the present study missed a delayed elevation in this biomarker following drug-induced endothelial dysfunction; this should be addressed in subsequent studies to monitor multiple time points.

The present study has potential implications when developing future medications to treat cocaine addiction. Because treatment of cocaine addiction requires the ability to track exposure as well as abstinence, and cocaine urine toxicology results are relatively fleeting (undetectable after 36 hours from intermittent use),¹² biomarkers of endothelial dysfunction are attractive targets. Future study should also establish that these endothelial biomarkers are selective to overdose-related and/or cardiovascular events rather than chronic drug use or other chronic disease conditions. Overall, the present study provides the foundation for further exploration of siCAM-1 as a potential biomarker for detection of endothelial dysfunction following acute cocaine use.

The present study findings of the ability of siCAM-1 to predict ACVE for all drug overdose is novel. Clinical risk factors for prediction of ACVE in this population have been previously described,¹³ but no biomarkers for ACVE have been validated to date. However, improvement of risk prediction is still required and early risk stratification for cocaine (or other drug) overdose may improve care for thousands of patients per year who present to healthcare settings with need for emergency clinical intervention. Additionally, early assessment of endothelial dysfunction in ED patients with overdose may improve care by identifying patients at highest risk for ACVE.

The findings in this study should be interpreted with several important considerations. This was an observational study with notable limitations, including limited ability to consider chronic exposure or polysubstance effects. Polysubstance overdose was very common in the cohort, and >90% of the cocaine group had multiple drug exposures; thus, other drugs may have confounded the biomarker associations (i.e., siCAM-1). Multiple time points were not evaluated for estimation of rise and fall of biomarkers, which was a limitation of the study protocol; thus, we may have missed “late” elevation of ET-1 or RANTES. Therefore, further study of multiple time points is needed to confirm these findings. We did not control for diabetes or prior cardiac disease (coronary artery disease or congestive heart failure), which may confound the interpretation of endothelial dysfunction; future studies should therefore account for this at baseline. Due to low numbers of outcomes for each type of ACVE, we were not able to differentiate whether biomarkers were predictive for individual components of the composite outcome. A sizeable proportion of patients (205 of 391, or 52% of potentially eligible patients) were excluded due to absence of waste specimens, which may have biased the sample; while this likely biased the study towards the null (as less severe overdose would not have clinical laboratory specimens), whether the addition of these patients could have changed the overall analysis remains a matter of conjecture. Additionally, since the study was performed at the affiliate hospitals of only one medical school in an urban catchment area, these results should be externally validated in other settings with different patient populations. And finally, these results do not apply to excluded populations such as children. Follow-up work is required to determine the relative contribution of endothelial biomarkers in this population.

In conclusion, we observed that for adult ED patients with acute drug overdose, the initial serum siCAM-1 was elevated in the setting of cocaine overdose, and was a predictive biomarker for the in-hospital occurrence of ACVE. Endothelial dysfunction may be an important surrogate for both acute cocaine overdose and drug overdose severity. Future research is warranted to clarify the role of endothelial biomarkers as part of the clinical evaluation for ED patients with acute drug overdose. Given the present study’s limited ability to consider chronic exposure or polysubstance effects, follow-up work is required to determine the relative contribution of endothelial biomarkers in this population.