Laboratory characteristics of ischemic stroke patients with atrial fibrillation on or off therapeutic warfarin

Abstract

Background

In patients with atrial fibrillation (AF), despite adequate anticoagulation, ischemic stroke (IS) is an uncommon yet concerning occurrence.

Hypothesis

Specific laboratory parameters may affect the efficacy of warfarin despite therapeutic international normalized ratio (INR) in patient with AF who present with IS.

Methods

We used the database from a multicenter clinical trial to identify AF patients who presented with IS. We trichotomized the cohort into patients with therapeutic INR on warfarin, subtherapeutic INR on warfarin, and on no anticoagulants. We then compared baseline laboratory characteristics and other baseline features among the groups.

Results

Patients with therapeutic INR presented with higher serum creatinine (P = 0.01) and blood urea nitrogen (P = 0.02) and lower glomerular filtration rates (P = 0.001) compared with other groups. Other laboratory parameters were not different among the 3 groups. Patients with therapeutic INR also presented with milder stroke symptoms (P = 0.01). Medical history of the 3 groups was not different, except for history of valvular heart disease, which was more prevalent in patients with therapeutic INR (P = 0.004). In‐hospital mortality rates and 90‐day disability were not different among the 3 groups.

Conclusions

AF patients who presented with IS on therapeutic warfarin had higher average serum creatinine and blood urea nitrogen, and lower glomerular filtration rates, compared with others. Impaired renal function may be a factor contributing to occurrence of IS in AF patients despite adequate anticoagulation. Larger, targeted studies are needed to confirm these findings.

1. INTRODUCTION

Atrial fibrillation (AF) affects >2.7 million Americans and is responsible for 10% to 12% of ischemic strokes (IS) in the United States each year.1 The annual risk of IS among individuals with AF can be as high as 18.2%, depending on the coexistence of various risk factors.2 Anticoagulation with adjusted‐dose warfarin has been shown to lower this risk substantially—up to 68%, according to some reports.3

The optimal strength of warfarin anticoagulation for prevention of IS in patients with AF is an international normalized ratio (INR) of 2.0 to 3.0.4 Close monitoring and timely dose adjustments are necessary to maintain the intensity of anticoagulation within this range. In many instances, however, AF patients who take warfarin fall below therapeutic anticoagulation indices (INR <2.0) and present with IS.5 Conversely, sometimes those with INR within the therapeutic range of 2.0 to 3.0 present with IS. Although such occurrences are probably rare, they are especially concerning because they preclude treatment with intravenous thrombolytics.

Various studies have identified a range of patient characteristics that can lead to warfarin failure in patients with AF, including prior ischemic events, hypertension, carotid atherosclerosis, diabetes mellitus, hyperlipidemia, smoking, high CHA₂DS₂‐VASc scores, and elevated plasma homocysteine levels.6, 7, 8, 9 In this study, we used the data from a large multicenter clinical trial to explore differences in laboratory features and other baseline characteristics of AF patients who presented with acute IS on INR ≥2.0 on warfarin and those with INR <2.0 on and off warfarin.

2. METHODS

2.1. Data acquisition

Study data were obtained from the publicly available Field Administration of Stroke Therapy–Magnesium (FAST‐MAG) trial dataset and therefore, exempt from ethics review. Details of this trial have been published elsewhere.10 Briefly, FAST‐MAG was a multicenter, randomized, double‐blind, placebo‐controlled, clinical trial wherein degree of disability at 90 days was compared in IS patients who received intravenous magnesium sulfate within 2 hours of symptom onset vs placebo.10 For our study, patients of interest were those with a final diagnosis of IS and history of AF. As such, patients who were diagnosed with transient ischemic attack (TIA), intracerebral hemorrhage, and stroke mimics were excluded. We also excluded patients on the newer generation of oral anticoagulants.

Data of interest included initial laboratory results, demographics, medical history, initial stroke severity based on the National Institutes of Health Stroke Scale (NIHSS) score, hospital mortality, intervention, and 90‐day outcomes based on the modified Rankin Scale. Glomerular filtration rate (GFR) was calculated using the Modification of Diet in Renal Disease equation.

2.2. Statistical analysis

The study cohort was trichotomized into patients on warfarin with INR ≥2.0, those on warfarin with INR <2.0, and those not on warfarin. Categorical data were presented as proportions, continuous data as mean ± SD, and ordinal data as median with IQR. The 1‐way analysis of variance was used to compare mean values. The Marascuilo procedure was used to compare proportions between groups. Ordinal variables were compared using the Kruskal‐Wallis H test. The α value was set at ≤0.05.

3. RESULTS

Out of a total of 1700 patients included in FAST‐MAG, 290 IS patients had a history of AF. From this number, 5 were excluded due to unavailable initial INR, and 2 were excluded for being on dabigatran. Our study cohort consisted of 283 AF patients, of which 21 (7.4%) presented with INR ≥2 (mean INR, 2.7 ± 0.9), 71 (25.1%) on warfarin with INR <2 (mean INR, 1.35 ± 0.28), and 191 (67.5%) patients not taking warfarin.

Table 1 lists and compares the baseline characteristics of the 3 groups. Valvular heart disease (VHD) was more prevalent in patients with INR ≥2, compared with those with INR <2 on or off warfarin (respectively, 42.9%, 22.5%, 11.5%; P = 0.004). Patients not on warfarin were more frequently on antiplatelet agents (P < 0.0001) than those on warfarin, regardless of INR.

Table 1.

Baseline characteristics of AF patients presenting with ischemic stroke with INR ≥2 on warfarin, INR <2 on warfarin, and no warfarin

	Warfarin INR ≥2, n = 21	Warfarin INR <2, n = 71	No Warfarin, n = 191	P Value
Age, y	74.4 ± 11.4	74.7 ± 11.2	76.4 ± 10.7	0.4
Female sex	10 (47.6)	39 (54.9)	94 (49.2)	0.7
White race	15 (79.4)	42 (59.2)	133 (69.6)	0.3
Weight, kg	78.2 ± 19.1	78.2 ± 17.3	76.7 ± 18.9	0.8
HTN	20 (95.2)	63 (88.7)	167 (87.4)	0.6
DM	3 (14.2)	25 (35.2)	53 (27.8)	0.2
Hyperlipidemia	14 (66.7)	42 (59.2)	87 (45.6)	0.06
CAD	8 (38.1)	24 (33.8)	60 (31.4)	0.8
VHD	9 (42.9)	16 (22.5)	22 (11.5)	0.004
Prior stroke	4 (19.1)	7 (9.9)	16 (8.4)	0.3
Smoking	1 (4.8)	4 (5.6)	19 (10)	0.4
Alcohol usea	2 (9.5)	6 (8.4)	9 (4.7)	0.4
APA use	2 (9.5)	11 (15.4)	98 (51.3)	<0.0001

Abbreviations: AF, atrial fibrillation; APA, antiplatelet agent; CAD, coronary artery disease; DM, diabetes mellitus; HTN, hypertension; INR, international normalized ratio; SD, standard deviation; VHD, valvular heart disease.

Data are presented as n (%) or mean ± SD.

^aMore than 2 drinks per day.

Patients with INR ≥2 presented with higher average serum creatinine (sCr), blood urea nitrogen (BUN), and lower mean GFR than those with INR <2, on or off warfarin (Table 2). Leukocyte count, hematocrit, platelet count, serum glucose, and the frequency of AF on the initial electrocardiogram were not different among the 3 groups. Patients with INR ≥2 presented with lower median NIHSS score than the other 2 groups (respectively, 6, 14, and 11, P = 0.01; Table 2).

Table 2.

Admission laboratory and clinical features of AF patients presenting with INR ≥2 on warfarin, INR <2 on warfarin, and no warfarin

	Warfarin INR ≥2, n = 21	Warfarin INR <2, n = 71	No Warfarin, n = 191	P Value
NIHSS, median (IQR)	6 (3–14)	14 (7–20)	11 (4–18)	0.01
AF on ECG, n (%)	15 (71.4)	53 (74.7)	115 (60.2)	0.08
WBC, 103/μL	8.2 ± 2.7	8.1 ± 2.6	8.1 ± 3.2	1.0
Hematocrit, %	40.6 ± 4.9	39.7 ± 4.8	39.4 ± 4.4	0.5
Platelets, 103/μL	217.2 ± 69.6	215.4 ± 73.1	209.1 ± 61.2	0.7
Serum glucose, mg/dL	144 ± 53.9	140.3 ± 57.3	134.5 ± 55.1	0.6
BUN, mg/dL	23.7 ± 8.5	19.5 ± 6.4	22.6 ± 9	0.02
sCr, mg/dL	1.22 ± 0.29	0.96 ± 0.27	1.14 ± 0.54	0.01
GFR, mL/min	54.8 ± 14.8	75.6 ± 22.2	67.5 ± 25.3	0.001

Abbreviations: AF, atrial fibrillation; BUN, blood urea nitrogen; ECG, electrocardiogram; GFR, glomerular filtration rate; INR, international normalized ratio; IQR, interquartile range; NIHSS, National Institutes of Health Stroke Scale; sCr, serum creatinine; SD, standard deviation; WBC, white blood cells.

Data are presented as mean ± SD unless otherwise noted.

Two patients in the INR ≥2 group, 12 patients in the INR <2 warfarin group, and 11 patients in the no‐warfarin group underwent intra‐arterial intervention (P = 0.02; Table 3). All‐cause hospital mortality rates and degree of disability at 90 days were not different among the 3 groups.

Table 3.

Intervention rates, hospital mortality rates, and 90‐day disability of AF patients presenting with INR ≥2 on warfarin, INR <2 on warfarin, and no warfarin

	Warfarin INR ≥2, n = 21	Warfarin INR <2, n = 71	No Warfarin, n = 191	P Value
Hospital death	3 (14.3)	15 (21.1)	30 (15.7)	0.60
IA intervention	2 (9.5)	12 (16.9)	11 (5.8)	0.02
mRS90	1 (0–4)a	2 (0–4)b	2 (1–4)c	0.40
mRS90 d	2 (0.5–4.5)	3 (1–5)	3 (1–5)	0.50

Abbreviations: AF, atrial fibrillation; IA, intra‐arterial; INR, international normalized ratio; IQR, interquartile range; mRS₉₀, modified Rankin Scale at 90 days.

Data are presented as n (%) or median (IQR).

^aN = 18.

^bN = 55.

^cN = 157.

^dIncluding death.

4. DISCUSSION

This exploratory study resulted in the novel finding that AF patients with INR ≥2 who presented with IS had higher average BUN and sCr levels and lower GFR than the other groups. There is some evidence suggesting that warfarin metabolism is influenced by kidney function.11 It has been reported that patients with reduced kidney function have worse anticoagulation control—that is, lower proportion of INRs in target range.11 In most cases, subtherapeutic INRs are seemingly more common than supratherapeutic INRs in patients with chronic kidney disease.11 It is possible that fluctuating INR in patients with kidney dysfunction led to formation of emboli in the days leading up to their IS, notwithstanding their therapeutic INR on presentation. Overall, however, there is no evidence that kidney function—evidenced by elevated BUN and sCr and reduced GFR—interferes with the effectiveness of warfarin when INR is within the therapeutic range. In fact, it has been shown that adjusted‐dose warfarin has similar efficacy for prevention of IS in patients with chronic kidney disease and in those with higher GFR.12

A higher average BUN and sCr does not always indicate renal dysfunction. Dehydration can also lead to elevation of these laboratory parameters due to renal stress. Approximately 1 in 3 patients who present with stroke are dehydrated during their first blood test performed on the day of admission or the day after.13 Some reports indicate that dehydration may be a significant risk factor for ischemic stroke in patients with AF.14 Dehydration changes the viscosity of blood by enhancing deformability of erythrocytes, subsequently increasing the propensity for thrombosis.15 It is not clear whether warfarin therapy reduces blood viscosity resulting from changes in red‐cell morphology. Important to note here is that hematocrit proportions among the 3 groups were not different.

The prevalence of VHD was higher in AF patients who presented with acute IS on warfarin with INR ≥2, compared with those with INR <2 on or off warfarin. The distinction between “valvular” and “nonvalvular” AF remains a matter of debate.16 Whether AF patients with any evidence of valvular abnormality can be categorized as having valvular AF is also unknown. In a registry of patients with AF across 9 European countries, valvular abnormalities coexisted in 63.5% of AF patients.17 AF frequently complicates VHD, especially left‐sided valvular lesions.18 It can occur in 30% to 40% of patients with rheumatic mitral valve disease (stenosis) and up to 75% of patients age ≥ 65 years with mitral regurgitation.18

Except for a trend toward a higher rate of hyperlipidemia in patients in the INR ≥2 group, we did not find a higher prevalence of various vascular risk factors in patients who presented with IS on therapeutic INR. This is unlike prior studies, in which a variety of risk factors were reported to be associated with IS on therapeutic warfarin.6, 7, 8, 9 Each of the 4 studies that investigated this reported a different set of risk factors.6, 7, 8, 9 Previous ischemic events and smoking were the only risk factors shared in 2 studies.

Finally, we found fewer severe neurological deficits based on the NIHSS score on admission in patients with INR ≥2.0, compared with subtherapeutic INR groups. This finding confirms prior reports. Large registry‐based studies have shown that in AF patients who present with acute IS, therapeutic warfarin is associated with reduced IS severity at presentation.19, 20

4.1. Study limitations

Some of the limitations our study need reflection, including its relatively small size. This may have resulted from the exclusion of patients with TIA. We decided to exclude this population because of the subjectivity associated with the diagnosis of TIA. Another reason to explain the small numbers is that presentation with acute IS vis‐à‐vis therapeutic INR is a relatively uncommon occurrence. Our study was based on post hoc analysis of prospectively collected data, and it is also possible that there were factors contributing to warfarin failure that were unknown to us and subsequently not included in our analysis. For example, we could not eliminate the potential diluting effect of concomitant medications on warfarin. We did not compare medications among the 3 groups because there were numerous classes of drugs and compliance could not be confirmed. Stroke subtypes were not defined in the dataset. Therefore, there could have been noncardioembolic IS mechanisms among patients with INR ≥2, such as small‐vessel disease or IS due to high‐grade extracranial large‐artery stenosis. Information on congestive heart failure history or left ventricular function was also not available, which was, in our opinion, an important variable in cardioembolic IS. We also could not delineate the nature of VHD, because this information was not available in the FAST‐MAG database. However, those patients with mechanical valves—who may require higher intensity of anticoagulation—were clearly identified in the database. None of the patients in the INR ≥2 group had mechanical valves.

5. CONCLUSION

The reasons for IS on therapeutic warfarin remain elusive. Dehydration or compromised kidney function may be contributing factors. Presence of VHD of all types may also be a risk factor. However, the appropriate treatment for valvular AF—except for those with prosthetic valves—is inadequately defined. Large prospective studies are needed to specifically assess the effect of renal function (or dehydration) on the efficacy of warfarin in prevention of IS in patients with AF.

Aachi RV, Birnbaum LA, Topel CH, Seifi A, Hafeez S, Behrouz R. Laboratory characteristics of ischemic stroke patients with atrial fibrillation on or off therapeutic warfarin. Clin Cardiol. 2017;40:1347–1351. 10.1002/clc.22838