Fibrinogen Albumin Ratio and Ischemic Stroke During Venoarterial Extracorporeal Membrane Oxygenation

Abstract

Fibrinogen is a clotting factor and a major determinant of platelet aggregation. Albumin, on the other hand, inhibits platelet function and thrombus formation. Taken together, an elevated fibrinogen albumin ratio (FAR) has been described as a marker of disease severity during prothrombotic conditions. We evaluated the association of FAR and ischemic stroke during venoarterial extracorporeal membrane oxygenation (VA ECMO) support. A single center, retrospective study was performed including all adult patients placed on VA ECMO. FAR was calculated from fibrinogen and albumin measurements in the first 24 hours of VA-ECMO initiation. Patients were categorized into high (≥125) and low (<125) FAR groups and the risk of eventual ischemic stroke was determined. There were 201 patients who underwent VA ECMO placement and 157 had a FAR. They were 56 ± 14 years old and 66 (42%) had a high FAR. Patients with a high FAR had lower survival free from an ischemic stroke during VA ECMO (log rank p < 0.001; adjusted hazard ratio 5.51; 95% CI: 1.8–16.5). In tertile analysis, the level of FAR was associated with an incrementally higher likelihood of eventual ischemic stroke (log rank p = 0.004). Those with a high FAR had greater mean platelet volume (10.8[10.4–12] vs. 10.5[10.2–11.9] fl, p = 0.004). An elevated FAR during the first 24 hours of VA ECMO placement is associated with a greater risk of a subsequent ischemic stroke. Our findings suggest that assessment of FAR soon after VA ECMO placement may assist with early stratification of patients at risk for an ischemic stroke.

The use of venoarterial extracorporeal membrane oxygenation (VA ECMO) in patients with refractory cardiogenic shock has significantly increased in the last decade.¹ Despite significant technologic advances, associated morbidity remains high with only about 40% of patients surviving to discharge.² Neurologic complications may occur in more than one-third of patients and lead to longer length of stay and higher in-hospital mortality.^2–5 Although the rate of neurologic complications has decreased over the last decade, the incidence of stroke has remained elevated^4,6 and is reported in up to 33% of the patients.⁷ Moreover, the rate of stroke may be underreported as approximately 9% of strokes in patients on VA ECMO are clinically unrecognized.⁸ Several risk factors for neurologic injury including hypotension, metabolic acidosis, and duration of ECMO have been described.^9,10 However, predictive biomarkers of impending stroke are lacking.

Fibrinogen is a hepatically synthesized plasma glycoprotein (GP) and is the most abundant clotting factor in the body.¹¹ As an acute phase reactant, fibrinogen levels are elevated due to inflammation in severely ill patients with cardiogenic shock and commonly noted in patients requiring mechanical circulatory support.^12,13 An elevated fibrinogen level is associated with a higher rate of fibrin formation, thrombus fibrin content, fibrin network stability, and platelet aggregation.¹⁴ Albumin is a plasma protein which directly and indirectly inhibits platelets and thrombus formation.^15–21 Taken together, an elevated fibrinogen to albumin ratio (FAR) has been described as a marker of disease severity in various prothrombotic states such as ST elevation myocardial infarction, chronic venous insufficiency, and ischemic retinal vein occlusion.^22–24 However, the utility of the FAR in predicting ischemic stroke in patients on VA ECMO is unknown. In this investigation, we sought to evaluate the association between FAR and ischemic stroke during VA ECMO support.

Materials and Methods

All adult patients (age 18 and above) admitted to Montefiore medical center and placed on VA ECMO between January 2012 and September 2017 were included in this retrospective study. Patients were supported by the RotaFlow centrifugal pump. The Maquet Quadrox oxygenator was used during VA ECMO support. To exclude peri-procedural events, patients on VA ECMO for <24 hours were excluded. Patients without data on albumin or fibrinogen during the initial 24 hours of VA ECMO placement were excluded (Figure 1). Ischemic stroke was defined by i) the presence of new-onset neurologic symptoms that were confirmed by neurology consultation and ii) a corresponding acute cerebral infarction without the presence of hemorrhage noted on a head computed tomographic scan. The FAR, calculated from concurrent fibrinogen and albumin measurements, was recorded during the first 24 hours after VA ECMO initiation.

Figure 1.

Flowchart showing the study population.

Categorical variables were expressed as percentages and were evaluated using the χ² test. Continuous variables were expressed as mean (±standard deviation) or median (25–75% quartiles) and were evaluated using Student’s t-test or Mann–Whitney U test, respectively. Survival free from ischemic stroke during the observation period was calculated according to the Kaplan–Meier method. Log-rank testing was conducted to compare survival curves obtained from Kaplan–Meier estimates. Observation time was started at the initiation of VA ECMO and censored for hemorrhagic stroke, weaning off VA ECMO, device upgrade, heart transplantation, or death. Cox proportional hazard ratios (HRs) were calculated to estimate the risk of an ischemic stroke at a FAR cut-point of 125. This cut-point was derived by employing a receiver operating characteristic (ROC) curve. The adjusted HRs and corresponding 95% confidence intervals (CIs) were calculated after controlling for covariates with a univariable p < 0.2 including hypertension, baseline creatinine, mean platelet volume (MPV) atrial fibrillation, and hours on VA ECMO (model 1). A separate model was created (model 2) to calculate the adjusted HR after controlling for age, indication for ECMO, and total VA ECMO duration in hours. To limit over fitting, the indication for VA ECMO was grouped as postcardiotomy and non-postcardiotomy shock. All reported statistical tests were two-sided and p < 0.05 was considered significant. All analyses were performed using Stata 14.0 (StataCorp LLC, Texas). This study was approved by the institutional review board at Montefiore Medical Center.

Results

Patient Characteristics

There were 201 patients who underwent VA ECMO placement during the study period. As shown in Figure 1, 44 patients did not have fibrinogen or albumin measurements within 24 hours of VA ECMO initiation and were excluded. The remaining 157 patients, who comprised the study population, were 56.5 ± 14.0 years old; 97 (61%) had a history of hypertension and 55 (35%) had diabetes mellitus. The major indication for VA ECMO placement was postcardiotomy shock (31%), followed by acute myocardial infarction (27%). At device placement, lactic acid was 8.6 ± 5.6 mmol/L. Sixty-six (42.04%) patients survived to discharge.

Ischemic Stroke

Overall, 22 (14%) of the patients had an ischemic stroke in 4 ± 2 days on VA ECMO support. Patients who experienced an ischemic stroke were slightly younger (50 ± 16 vs. 58 ± 13 years, p = 0.02) but had no significant differences in comorbidities including hypertension and diabetes mellitus, and had similar implant indications (Table 1). There was a trend for longer VA ECMO support time in patients who had an ischemic stroke (129 [101–170] vs. 100 [60–157] hours, p = 0.08, Table 1).

Table 1.

Patient Characteristics of Patients with and Without an Ischemic Stroke

Characteristics	Patients with Ischemic Stroke (n = 22)	Patients Without Ischemic Stroke (n = 135)	p Value
Age in years	50.09 ± 13.44	57.48 ± 15.91	0.02
Gender n (%)
Male	17 (77.27)	92 (68.15)	0.39
Female	5 (22.73)	43 (31.85)
Comorbidities n (%)
Diabetes mellitus	9 (40.91)	46 (34.33)	0.55
Hypertension	11 (50)	86 (64.18)	0.20
Coronary artery disease	5 (22.73)	50 (37.04)	0.19
Peripheral vascular disease	0 (0)	5 (3.70)	0.36
Atrial fibrillation	4 (18.18)	26 (19.26)	0.91
COPD	1 (4.55)	9 (6.67)	0.71
Laboratory values (25–75% IQ)
Hemoglobin (g/dl)	10 (9.2–11)	9.7 (8.5–11.9)	0.60
Platelet (×1,000/μl)	141 (101–257)	142 (98–224)	0.72
Mean platelet volume (fl)	10.8 (10–11.4)	10.6 (9.8–11.6)	0.57
Baseline creatinine (mg/dl)	1.51 (1.1–2.3)	1.3 (0.9–2.1)	0.25
GFR (ml/min/1.73 m2)	38 (21–91)	45 (28–68)	0.77
Lactic acid (mmol/L)	7.6 (3–13)	7.5 (4.1–11.5)	0.90
24 hour PTT (seconds)	58.5 (44.8–67.8)	58 (46.6–66.6)	0.93
48 hour PTT (seconds)	55.5 (39.2–62.1)	53 (44.5–64.3)	0.63
Indication for VA ECMO n (%)			0.54
Graft failure after OHT	1 (4.55)	16(11.85)
Acute myocardial infarction	9 (40.91)	33 (24.44)
Postcardiotomy shock	8 (36.36)	41 (30.37)
Cardiomyopathy	2 (9.09)	27 (20)
Respiratory failure	1 (4.55)	13 (9.63)
Sepsis	1 (4.55)	4 (2.96)
Cardiac arrest	0	1 (0.74)
Intra-aortic balloon pump n (%)	9 (40.91)	52 (38.52)	0.83
Impella n (%)	1 (4.55)	9 (6.77)	0.69
Total VA ECMO duration in hours (25–75% IQ)	129 (101–170)	100 (60–157)	0.08
Intracardiac or aortic root thrombi during VA ECMO support n (%)	3 (13.64)	12 (9.45)	0.45
Survival at discharge n (%)	4 (18.18)	62 (46.62)	0.01

COPD, chronic obstructive pulmonary disease; GFR, glomerular filtration rate; PTT, partial thromboplastin time; VA ECMO, venoarterial extracorporeal membrane oxygenation; OHT, orthotopic heart transplant.

Fibrinogen to Albumin Ratio and Ischemic Stroke

Patients with an eventual ischemic stroke had a higher FAR 168 [125–226] in comparison to those who did not have an ischemic stroke (109 [77–148]; p = 0.0001, Figure 2). After adjustment for age, a high FAR was independently associated with an ischemic stroke (aHR: 4.5; 95% CI: 1.7–12.2, p = 0.004). ROC curve analysis showed that the area under the curve (AUC) for FAR and ischemic stroke was 0.75 (see Figure, Supplemental Digital Content 1, http://links.lww.com/ASAIO/A406).

Figure 2.

Box plot of the median FAR within 24 hours of venoarterial extracorporeal membrane oxygenation device placement in patients with and without ischemic stroke. FAR, fibrinogen albumin ratio.

Sixty-six (42%) patients had a high (≥125) FAR and 91 (58%) had a low FAR (<125) after VA ECMO placement. There was no significant difference in the distribution of age and sex among patients with a high or low FAR. The high FAR group had more patients with a history of hypertension (72% vs. 55%, p=0.03), whereas atrial fibrillation was more prevalent in patients with a low FAR (24% vs. 12%, p = 0.06). The baseline creatinine (1.6[1–2.4] vs. 1.3[0.9–1.85] mg/dl, p = 0.03) and median duration of VA ECMO (119[101–170] vs. 95[60–157] hours, p = 0.09) were higher in patients with a high FAR. Seventeen (26%, 0.049 events per patient day) patients with a high FAR had an ischemic stroke in comparison to only five (5.5%, 0.011 events per patient day) with a low FAR. As such, patients with high FAR had lower survival free from an ischemic stroke (log rank p < 0.001, Figure 3). Furthermore, the degree of FAR elevation was associated with an incrementally higher likelihood of an eventual ischemic stroke (log rank p = 0.004, Figure 4). Lastly, MPV, a surrogate of platelet activation and aggregation,²⁵ was greater in those with a high FAR (10.8[10.4–12] vs. 10.5[9.4–11.4] fl, p = 0.004) after VA ECMO placement

Figure 3.

Ischemic stroke free survival in patients with a high (≥125) and low (<125) FAR. Censored for death, explant, ventricular assist device placement, hemorrhagic stroke, and diffuse anoxic brain injury. FAR, fibrinogen albumin ratio.

Figure 4.

Ischemic stroke free survival in patients with a lower (>100), intermediate (100–200), and higher (<200) FAR. Censored for death, explant, ventricular assist device placement, hemorrhagic stroke, and diffuse anoxic brain injury. FAR, fibrinogen albumin ratio.

After adjusting for hypertension, creatinine, MPV, atrial fibrillation, and time on VA ECMO support, the aHR for an ischemic stroke in patients with a high FAR was 5.51, 95% CI: 1.8–16.5, p = 0.002, in comparison to a low FAR (Table 3). In a separate model with adjustment for age, indication, and duration of VA ECMO, the aHR remained elevated at 4.65, 95% CI: 1.69–12.79, p = 0.003 (Table 3). Survival to discharge for patients with a high FAR trended to be lower at 22 (33.9%) vs. 44 (48.9%), p = 0.06, Table 2.

Table 3.

Hazard Ratio of an Ischemic Stroke in Patients on VA ECMO with a FAR ≥ 125 When Compared with Those with a FAR < 125

Risk of Ischemic Stroke	Hazard Ratio	95% CI	p Value
Unadjusted	4.74	1.75–12.88	0.002
Adjusted model 1*	5.51	1.84–16.49	0.002
Adjusted model 2†	4.65	1.69–12.79	0.003

Model 1

^*: Adjusted for history of hypertension, atrial fibrillation, mean platelet volume, baseline creatinine, and hours on VA ECMO. Model 2

^†: Adjusted for age, indication for ECMO (grouped as postcardiotomy and non-postcardiotomy) and total ECMO duration in hours.

ECMO, venoarterial extracorporeal membrane oxygenation; FAR, fibrinogen albumin ratio; OHT, orthotopic heart transplant.

Table 2.

Clinical Characteristics of Patients with a High (≥125) and Low (<125) FAR

Characteristics	High FAR (≥125) (n = 66)	Low FAR (<125) (n = 91)	p Value
Age in years	54.95 ± 13.89	57.53 ± 14.05	0.26
Gender n (%)			0.45
Male	48 (72.73)	61 (67.03)
Female	18 (27.27)	30 (32.97)
Comorbidities n (%)
Diabetes mellitus	24 (36.92)	31 (34.07)	0.71
Hypertension	47 (72.31)	50 (54.95)	0.03
Coronary artery disease	23 (34.85)	32 (35.16)	0.97
Peripheral vascular disease	1 (1.52)	4 (4.40)	0.31
Atrial fibrillation	8 (12.12)	22 (24.18)	0.06
COPD	2 (3.03)	8 (8.79)	0.15
Laboratory values (25–75% IQ)
Hemoglobin (g/dl)	9.9 (8.7–12.2)	9.8 (8.5–11.8)	0.99
Platelet (×1,000/μl)	149 (104–231)	131 (94–220)	0.18
Mean platelet volume (fl)	10.8 (10.4–12)	10.5 (9.4–11.4)	0.004
Baseline creatinine (mg/dl)	1.6 (1–2.4)	1.25 (0.9–1.85)	0.03
GFR (ml/min/1.73 m2)	39 (19–74)	46 (29–68)	0.23
Lactic acid (mmol/L)	7.6 (3.3–11.1)	7.5 (4.2–12.3)	0.53
Fibrinogen (mg/dl)	451 (373–563)	242 (197–303)	<0.001
Albumin (g/dl)	2.4 (2.1–2.6)	2.65 (2.3–3)	<0.001
24 hour PTT (seconds)	58.7 (51.5–67.8)	57.65 (44.4–65.4)	0.14
48 hour PTT (seconds)	56.25 (45.15–64.8)	52.45 (41.75–62.4)	0.12
Indication for VA ECMO n (%)			0.20
Graft failure after OHT	3 (4.55)	14 (15.38)
Acute myocardial infarction	17 (25.76)	25 (27.47)
Postcardiotomy shock	21(31.82)	28 (30.77)
Cardiomyopathy	16 (24.24)	13 (14.29)
Respiratory failure	5 (7.58)	9 (9.89)
Sepsis	3 (4.55)	2 (2.2)
Cardiac arrest	1 (1.52)	0
Intra-aortic balloon pump n (%)	24 (36.36)	37 (40.66)	0.59
Impella n (%)	6 (9.09)	4 (4.49)	0.25
Total VA ECMO duration in hours (25–75% IQ)	119 (81–167)	95 (57–152)	0.09
Intracardiac or aortic root thrombi during VA	4 (6.06)	11 (12.64)	0.24
ECMO support n (%)
Survival at discharge n (%)	22 (33.85)	44 (48.89)	0.06

FAR, fibrinogen albumin ratio; COPD =chronic obstructive pulmonary disease; PTT= partial thromboplastin time; VA ECMO = venoarterial extracorporeal membrane oxygenation; OHT =orthotopic heart transplant

Discussion

We investigated the association of FAR after VA ECMO placement with subsequent ischemic stroke. Our major findings are as follows: First, a FAR > 125 within 24 hours of VA ECMO placement was associated with an over five times higher risk of an ischemic stroke compared with a FAR < 125. Beyond cut-point analysis, the level of FAR elevation was related to a higher likelihood of an eventual ischemic stroke. Second, MPV, an indicator of platelet activation and aggregation,²⁵ was higher in patients with a high FAR. And lastly, survival to discharge was lower in patients with a high FAR. These observed associations were independent of covariates that showed possible univariable associations with FAR levels. Such associations are also unlikely to be confounded by variations in anticoagulation parameters because patients in both groups were on intravenous unfractionated heparin and had similar partial thromboplastin time levels after device placement. In addition, patients with a high or low FAR had similar demographics, indication, and cannulation sites for VA ECMO and severity of baseline perfusion, as evidenced by similar lactic acid levels. These findings suggest that FAR may be a readily available hematologic biomarker for early stratification of patients at risk of an ischemic stroke during VA ECMO support.

Ischemic stroke remains a major source of morbidity and mortality in patients supported by VA ECMO; however, predictive biomarkers of this potentially catastrophic complication remain elusive. Fibrinogen binds to the GP IIb-IIIa receptor on activated platelets and leads to platelet aggregation by cross-linking adjacent activated platelets.²⁶ Although elevated fibrinogen may just be a consequence of the same risk factors that predispose to stroke, it is also directly related to fibrin formation and clot fibrin content.¹⁴ Since thrombus formation during ECMO is fibrin driven, it is conceivable that a high fibrinogen level plays a direct role in thrombus formation in patients on ECMO.²⁷

Albumin inhibits promoters of platelet aggregation such as thromboxane A2 (TXA2), platelet activating factor, and non-esterified fatty acids. It binds arachidonic acid, depriving the substrate needed for TXA2 formation.¹⁹ In addition, albumin also inactivates TXA2 via direct binding and is a negative acute phase reactant.¹⁷ Importantly, albumin interacts with fibrinogen, resulting in impaired fibrinogen activity²⁸ and higher albumin levels are associated with deceleration of fibrin build-up.²⁹ Due to such synergy, the FAR is implicated in prothrombotic conditions and has been shown to be a predictor of disease severity and poor prognosis in patients with acute coronary syndromes,²² chronic venous insufficiency,²³ and stroke.³⁰ Although the ROC for FAR yielded a slighter greater AUC for ischemic stroke than either fibrinogen or albumin, most of the predictive capacity was with fibrinogen alone (see Figure, Supplemental Digital Content 2, http://links.lww.com/ASAIO/A407). Our findings suggest that monitoring of FAR within the first 24 hours of VA ECMO placement may assist clinicians with awareness of patients at higher risk of an ischemic stroke. It remains to be seen whether alterations in management such as higher anticoagulation targets, corticosteroid therapy, which reduces fibrinogen level during ventricular assist device support,³¹ and minimizing VA ECMO exposure based on a high FAR would lead to favorable outcomes.

Our study has several limitations. First, this study is limited by a retrospective design that may lead to information and selection bias. To ensure accuracy, we verified all ischemic stroke outcomes documented in the medical record with concurrent radiological studies. Because of the relatively low number of ischemic strokes, the multivariable analysis could not be adjusted for all variables that may impact this outcome. Although, the study groups with a high or low FAR were quite similar in levels of covariates, there may be other, unmeasured or incompletely measured differences besides FAR that may increase the susceptibility of VA ECMO patients for an ischemic stroke. It remains plausible that strokes may have occurred in such critically ill patients even before VA ECMO placement or during cardiopulmonary bypass as it was not feasible to determine baseline neurologic state; however, patients with a high or low FAR were all exposed to such potential confounders. In addition, ischemic strokes were only adjudicated to cases with new onset neurologic symptoms that corresponded to simultaneous brain imaging. Since follow up ended at the time of the first detectable ischemic stroke, the impact of FAR on recurrent strokes was not assessed.

Conclusion

In conclusion, this study shows that a high FAR (>125) within the first 24 hours of VA ECMO placement is associated with higher risk of subsequent ischemic stroke. Further studies are warranted to confirm these findings and whether changes in management in this subset of patients would lead to a reduction in stroke burden.