Introduction

The Fontan operation has transformed the lives of many children born with single-ventricle physiology, offering them the potential for survival and good quality of life well into adulthood [1]. The aim of the Fontan procedure is to redirect systemic venous return from the venae cavae directly to the pulmonary arteries. This permits the available functioning ventricle to perform solely as a systemic ventricle free from the burden of volume overload [2].

One of the most important late complications of this procedure is thrombosis, including arterial and venous thromboembolism, as well as intracardiac thrombosis. Thrombotic complications indicate worse mortality outcomes. These late thrombotic complications warrant pursuit of excellent thrombo-prophylaxis strategies [1].

Optimal prophylaxis, however, continues to be controversial. In a recent meta-analysis of ASA versus warfarin in Fontan patients, the pooled incidence rate of TE appears to be around 9% despite anticoagulant treatment [3]. However most studies do not report on time spent in the therapeutic range of warfarin or other markers of successful warfarin therapy. This makes accurate evaluation of the effectiveness of warfarin in preventing thrombotic and bleeding complications in this population difficult.

In the general non-congenital heart disease population, many studies have demonstrated that the efficacy and safety of warfarin is related to the actual level of anticoagulation represented as the international normalized ratio (INR) [5–7]. It is often difficult to maintain an optimal INR over time due to a narrow therapeutic index and many known interactions with warfarin therapy [7–10]. Therefore, monitoring anticoagulation intensity is a key component when managing patients receiving warfarin. Time in therapeutic range (TTR) has been used to monitor warfarin therapy and has been shown in multiple studies to correlate with incidence of thromboembolic events and bleeding [7]. In clinical practice, TTR is used as a quality indicator of outpatient oral anticoagulation management [8]. In the Fontan population, TTR was used to monitor warfarin treatment in only one prospective study and showed a significant increase in thromboembolism after the Fontan procedure in patients with low TTR compared to those within goal TTR [2]. It is not known whether a focus on TTR will yield greater insight into successful warfarin management. This retrospective longitudinal observational study aims at assessing whether TTR is a valid marker with which to calibrate successful therapy and outcomes in Fontan patients managed on warfarin.

Methods

Study population

All Fontan patients on warfarin at our institution who were enrolled in the cardiology-based pharmacist managed anticoagulation clinic (PMAC) and managed between the dates of November 1, 2013 and May 31, 2015 were eligible for inclusion. PMAC was started in 2013 to assist in anticoagulation management in patients receiving warfarin followed by pediatric cardiology. The clinic staff includes a clinical pharmacist with special interest and experience in anti-coagulation and two nurses. The staff follows patients during their cardiologist visits and in separate visits as needed. The team expanded their services to follow some patients when they are admitted to the hospital. Patients were excluded if lost to follow-up, presence of a recognized non-cardiac indication for long-term anticoagulation or on anticoagulation therapy with warfarin less than 1 month. Data compiled for each patient consisted of demographic data, anatomic diagnosis, hemodynamic information, INR values, warfarin dosing, concurrent medications as well as thrombotic and bleeding event history. The study was approved by the IRB at Cincinnati Children’s Hospital.

Study design

This study was a single-center, retrospective longitudinal observational study to assess whether TTR could be utilized as a practical target for monitoring warfarin treatment in Fontan patients, the factors affecting TTR and whether it correlates with thrombotic and bleeding events in this population. The proportion of time patients spent within, above and below their indicated therapeutic INR range (target INR ± 0.2 units) was calculated using the method developed by Rosendaal et al. [12]. This method assumes linear pharmacokinetics between INR assessments and calculates the percentage of time spent within the desired INR range. Percentage of time within therapeutic range is a validated measurement and is calculated as a percentage of days the INR was therapeutic out of the total duration of anticoagulation therapy during the study period [13].

Study outcomes

The primary outcome was TTR. Major bleeding was defined as a bleeding event which required emergency department care or inpatient admission. A sub-group analysis of the primary outcome including reasons for abnormal INRs was also performed.

Statistical analysis

Descriptive statistics were employed to denote the baseline characteristics utilizing mean ± SD or median and inter-quartile range (IQR; 25th and 75th percentile) for continuous variables as appropriate. Percentages and frequencies were used to present categorical variables. The difference in percentage of time within, above and below the therapeutic range was evaluated by Mann–Whitney U test (GraphPad Prism version 6.05, GraphPad Software, La Jolla, CA). Incidence of thrombotic events before and after enrollment into clinic was evaluated by Chi square test.

Results

Patient characteristics

Forty-seven patients were assessed for study eligibility. Two individuals were excluded due to being managed by the clinic for less than 1 month. A total of 45 Fontan patients on warfarin who were enrolled in the pharmacist-managed anticoagulation clinic met inclusion criteria. Of the 45 patients evaluated, 25 (55. 6%) were male and the median age at latest anticoagulation clinic follow-up was 19 (IQR 13, 29) years. Hypoplastic left heart and tricuspid atresia were the most common anatomic diagnoses in this study population with 12 (26.7%) and 9 (20%) patients, respectively. Regarding Fontan type, 20 (44.4%) patients had a lateral tunnel, 17 (37.8%) had an extra-cardiac conduit and 8 (17.8%) had atrio-pulmonary connections. Thirty-seven patients (82.2%) had a Fontan fenestration (Fig. 1). Thirty-five (77.8%) patients had a history of arrhythmias post-Fontan. History of a thrombotic event prior to clinic enrollment was present in 15 (33.3%) patients. INR testing method type was primarily lab/in-clinic INR testing for 35 (77.8%) patients and home INR monitor testing method for 10 (22.2%) patients. Patient demographics and antithrombotic therapy information are summarized in Table 1.

Table 1 Patient demographics and antithrombotic therapy
Full size table

Time in the therapeutic range

The mean and median TTR for patients was 84.1 and 90.3%, IQR (73.6–96.4%), respectively. Univariate determinants of TTR are summarized in Tables 2 and 3 (Fig.  2 and 3).

Table 2 Univariant analysis of % time in, above, & below therapeutic range
Full size table
Table 3 Logistic regression analysis of percentage time in therapeutic range
Full size table
Fig. 1
figure 1

Bar graph describing mean time (patient years) and percentage of time spent within, above, and below the patient specific therapeutic INR goal range during the study time frame. INR international normalized ratio; TTR time in therapeutic range

Full size image

The two most common reasons for supratherapeutic INR during this time period were diet changes (22%) and drug interactions (11%), while subtherapeutic INR were primarily due to medication non-compliance (23%) and diet changes (18%). Figures 2, 4 and 5  reviews all the explanations for abnormal INR obtained during the study time frame.

Fig. 2
figure 2

Bar graph illustrating the explanations documented at time of PMAC patient assessment for supratherapeutic and subtherapeutic INRs evaluated during the 19 month study time frame INR international normalized ratio

Full size image

Clinical events

Prior to being managed at our PMAC, 18 thrombotic events took place over 312.5 patient years (i.e., 1 event per 17.4 patient years). Throughout 52.9 years of total patient therapy during clinic management, only one thrombotic event took place. There were two major bleeds and 16 trivial to minor bleeding events in the PMAC. No data was available on bleeding events prior to PMAC. Despite these high levels of TTR in this overall cohort, there were no documented cerebral or gastrointestinal adverse events during the time of being managed in the PMAC. TTR for overall study time frame and monthly TTR values for the 3 months prior to the thrombotic and bleeding events. Thrombotic and bleeding event information is presented in Table 4.

Fig. 3
figure 3

A line graph of percent time in the therapeutic range 1, 2, and 3 months prior to thrombotic and bleeding events

Full size image
Table 4 Clinical events in the study population before and after clinic management
Full size table

The thrombotic event occurred in a 31 year old male with an atrio-pulmonary Fontan who prior to being managed at the warfarin clinic had an intra-atrial thrombus (Fig. 6). A follow up transesophageal echocardiogram (TEE), at the time of an atrial tachycardia, showed minor increase in the size of the intra-cardiac thrombus in this patient. No clinical symptoms were noted and he underwent successful Fontan conversion. His overall TTR during the study time frame was 69% and he had periods of low TTR prior to discovering the increase in the thrombus volume.

One of the major bleeds was epistaxis in a 32 year old male that required an emergency department visit resulting in packing and eventually cauterization. This patient’s overall TTR during the study time was 72.4% and 3 months prior to this bleeding event the time above goal range was 31.4%. The second major bleeding event occurred in a 16 year old female with history of trisomy 21 who presented with hemoptysis and was found to have pulmonary hemorrhage from the left lower lobe bronchus. A cardiac catheterization showed large aorto-pulmonary collaterals from the celiac trunk to the left lower lobe and the patient underwent a successful particle embolization. This patient’s overall TTR during the study time was 95.7% and 3 months prior to this bleeding event the time above goal range was 14.8%. Of the 16 minor bleeding events, 11 (68.8%) were epistaxis, 4 (25%) were due to gingival bleeding, and one (6.1%) was from rectal bleeding.

Fig. 4
figure 4

Scatterplot of univariant analysis for time within, above and below goal INR range based on designated INR target range INR international normalized ratio

Full size image

Discussion

This study demonstrates for the first time, that it is possible to obtain very high proportions of time in therapeutic range for Fontan patients. We achieved this through a pharmacist managed anticoagulation clinic. These superior therapeutic times were associated with excellent clinical outcomes. This brings into question the previously demonstrated incidences of thrombotic complications in those Fontan patients managed on warfarin, as TTR was rarely mentioned in these studies, and in the single investigation where it was mentioned the TTR was only 59% [2, 3].

Time in the therapeutic range

In the general patient population, the national benchmark for time in the therapeutic range is 65% [14]. In general, warfarin is considered ineffective when TTR is below 60% [15]. Our patient population, who represent some of the most complex of congenital heart defects, frequently with coexisting end organ disease such as Fontan associated liver disease, had a median TTR of 84.4%, comparing very favorably with published general adult literature [16, 17]. Most of the adult studies were done in elderly patients with atrial fibrillation and valvular heart disease. In pediatric and adult congenital heart disease cohorts, very few studies looked at TTR as a measure of successful warfarin management, or as a marker for good outcomes. At commencement of our study, the only randomized controlled trial to report TTR in a solely Fontan group, Monagle et al., reported a 5.9 times increased incidence of thromboembolic events in patients on warfarin who were below goal TTR. In that particular study, a high proportion of patients had TTRs less than 30% whilst only two-thirds of the population fulfilled the study definition of controlled warfarin being goal TTR > 30% [2]. Despite these suboptimal goal TTR ranges, the investigators found warfarin to be equivalent to ASA in terms of thrombotic and bleeding outcomes. However the study’s relatively less aggressive level of warfarin/INR control may have contributed to the apparently equivalent results observed between warfarin and ASA. In the secondary analysis of the Monagle et al. study, McCrindle and colleagues demonstrate that suboptimally controlled warfarin resulted in 3.5 times increased incidence of thromboembolic events and was associated with worse outcomes than ASA, and presumably worse than no anticoagulation at all. Additionally, these investigators reported well controlled warfarin was superior to all.

In the current series, we achieved very high TTR, and this was associated with only one thrombosis event in 52 patient years, significantly lower than the 18 events during 576 patient years prior to being managed in our pharmacist-managed anticoagulation clinic. We were not able to further stratify TTR to thrombotic propensity, given the single event in the relatively small series.

Subtherapeutic INRs were associated with medication non-compliance, diet changes, drug interactions, intercurrent illness and a relatively large proportion of undefined contributory factors in the current study. These observations reflect many of the known limitations of warfarin therapy. Probably the most important single factor is that of the influence of dietary content and appetite changes particularly in pediatric patients, including vitamin K containing food. Diet needs to be evaluated thoroughly when managing warfarin in this population and patient and staff education needs to be meticulously adhered to in achieving good TTR [18]. Drug interactions are another key contributing factor. The S-enantiomer of Warfarin is the most active isomer. S-warfarin is responsible for 70% of the drug’s activity because it has up to five-fold higher potency than R-warfarin. S-warfarin is metabolized primarily by CYP2C9. Therefore, it is readily susceptible to drugs that inhibit or induce CYP2C9 metabolism such as certain antibiotic, antifungal, antidepressant and antiarrhythmic drug classes commonly used in Fontan patient population and result in out of range INRs. Additionally, patients with a CYP2C9 genetic polymorphism can be at added risk for abnormal INRs. In our evaluation, we observed that patients who are on three or more interacting medications need to be monitored more closely due to their increased time spent out of range. The same is true for patients that are older than 30 years as it is clearly a very important contributory factor to poor TTR and associated adverse outcomes.

Fig. 5
figure 5

Scatterplot of univariant analysis for time within, above and below target INR range based on number of interacting medications with warfarin. INR international normalized ratio

Full size image

Serious bleeding complications occurred rarely in the present series and we did not observe any cerebral or gastrointestinal hemorrhage. The two patients who had serious bleeding events, had prolonged and heavy epistaxis and the other hemoptysis. These two patients, in the months prior to the bleeding event spent a great proportion of time above the therapeutic range. In a review of 4883 pediatric patients on warfarin therapy, 2% were readmitted within 30 days of hospital discharge for bleeding, of which Fontan patients represented only 5.2% [19]. This study observed similar bleeding event types as our study with epistaxis (22.7%) being the most common type of bleeding observed while hemoptysis was observed as well (7.2%). Risk factors identified in this study included higher target INR ranges. Recent studies comparing warfarin to new oral anticoagulant agents found an annualized rate of major bleeding for warfarin to be 3.1–3.4% when TTR is 55–64.9% [20–23]. These reassuring observations suggest that warfarin when well managed can achieve truly superior outcomes. Interrogation of the time period leading up to major bleeding events in this study demonstrated that patients may spend less time in the therapeutic range. In fact, this trend should prompt concern for bleeding and thrombotic risk.

Fig. 6
figure 6

Bar graph illustrating incidence of thrombotic events prior to clinic enrollment and post clinic enrollment

Full size image

The excellent TTR achieved in this current cohort can be attributed to a number of factors that were achieved in a PMAC. Components of the program include individualized management approaches, excellent patient engagement and compliance with a personalized approach. This approach involves a centralized cardiology anticoagulation team for consistency and continuity including patient INR reminders, use of patient-preferred communication method and INR testing method. High quality patient and staff education was also utilized to anticipate periods where INR control is likely to be at risk and taking appropriate steps to minimize large swings in INR. Kauffman and colleagues affirmed these anticoagulation management techniques in their evaluation of factors that influence warfarin compliance including assigned anticoagulation providers to work with the same patients consistently, providing formal INR reminders, avoiding lecturing patients following missed INR tests, reinforcing the clinical and psychological utility of INR results and facilitating access to INR testing [24]. The EMPoWarMENT and EMPoWARed studies also clearly illustrated incorporating and engaging patients in their warfarin management improves adherence, time within therapeutic range and outcomes. The EMPoWarMENT and EMPoWARed pediatric studies revealed that TTR was improved after home INR monitoring was implemented (77.7% pre vs. 83.0% post) [25, 26]. However, only 22.2% of patients in the present cohort had access to home INR monitoring. This important finding clarifies that high TTR can still be achieved in patients utilizing lab/in-clinic testing given that limited access to home INR monitoring is common.

Limitations of the study

The most significant finding of this study was our remarkably low thrombotic event rate in this closely monitored cohort. Our findings may be confounded by the relatively small numbers and relatively short follow-up years. We were not able to further stratify TTR to thrombotic propensity, given the single event in the relatively small series. Due to the observational nature of the study it is prone to unmeasured possible confounders. Additionally, lack of data available on bleeding events prior to PMAC limited the ability to evaluate bleeding event rate change with PMAC enrollment and correlation with TTR.

Conclusions

In this study, exemplary TTRs can be achieved in Fontan patients managed with warfarin in an outpatient cardiology PMAC and this is associated with superior clinical outcomes. These preliminary observations suggest that well managed warfarin therapy may in fact offer excellent prevention from thrombotic and bleeding events in this complex population. TTR is an important factor to consider when determining the most effective warfarin management strategies and when deciding which Fontan patients will require the closest monitoring. We believe TTR maintained at a high level >80% will help facilitate such desirable outcomes .