Hepatic involvement of HHT is a complex process involving shunts between hepatic arterial, hepatic venous or portal venous systems (9, 12). Hepatic AVMs may be present in up to 74% of HHT patients and several modalities have been utilized in their detection and monitoring (5). Hepatic angiography is considered the gold standard to detect subtle abnormalities indicative of hepatic AVMs, however it is limited by the invasive nature of the procedure and has largely been replaced by non-invasive modalities such as ultrasonography and CTA (21). While it is known that CTA can readily distinguish between different hepatic shunts and is often performed to characterize hepatic AVMs, no previous studies have quantified the association between CTA findings and the presence of heart failure symptoms in HHT patients (1, 5, 22).
CTA may be superior to ultrasonography in detecting and characterizing hepatic AVMs (11, 12). Although hepatic AVM screening is not typically performed in HHT patients (10, 21), CTA including the hepatic vasculature is often available in these patients because of the frequent use of chest CTA to screen or diagnose pulmonary AVMs. A routine chest CTA protocol often includes much or all of the hepatic parenchyma and vasculature, and thus CTA chest can serve a dual purpose of screening for both lung and liver AVMs. Identifying these hepatic AVMs and subsequently assessing their risk may lead to better outcomes even in initially asymptomatic HHT patients due to the early detection of hepatic AVM related complications (10, 21).
The results of our multivariate analysis suggest that the degree of CHA dilation is an independent predictor of heart failure symptoms. Mean CHA diameter was significantly higher in symptomatic patients (11.1 versus 8.4 mm), and our data showed 2.6 times greater odds of having heart failure symptoms with each 1 mm increase in CHA diameter. These results are compatible with an ultrasonography study which demonstrated that high output heart failure only occurred in higher grade hepatic vascular malformations (10). This ultrasound grading system is reproducible by trained ultrasonographers and incorporates factors such as dilation of the hepatic vasculature and vascular flow abnormalities (23). Ultrasonography benefits by being low cost with a high sensitivity and specificity for liver AVM detection (10, 23–25). However, it requires capable and experienced ultrasonographers, and the subjectivity of the grading system may lead to discrepancies between operators. On the other hand, measurement of the CHA diameter using CTA is unambiguous and is not technologist dependent.
Symptomatic anemia overlaps with heart failure symptoms, and hemoglobin less than 7 g/dL may directly lead to heart failure in a mechanism that expands extracellular plasma volumes from increased sympathetic and renin-angiotensin activity (26, 27). Symptoms are temporary as correction of anemia demonstrates rapid and complete regression of anemia-related high output heart failure (26). We excluded patients who experienced symptoms correlating to their anemia because of the possible overlap with heart failure symptoms. None of the patients in our cohort had hemoglobin levels less than 7 g/dL at the time of testing. Nevertheless, mean hemoglobin levels were significantly lower in symptomatic patients (10.7 vs 12.6 g/dL). Our multiple logistic regression analysis determined hemoglobin to be an independent predictor of heart failure symptoms with an OR 0.489 for each 1 g/dL increase of hemoglobin.
In our stepwise logistic regression analysis the combination of these two independent variables, CHA diameter and hemoglobin level, allows for excellent discrimination between patients who do and do not have heart failure symptoms with an AUC of 0.906. Employing the ORs respective to both CHA diameter and hemoglobin may help clinicians quantify risks to their patients.
Most patients in our cohort exhibited normal ejection fractions and a non-dilated left ventricle. In the remaining cardiac chambers, dilation was more frequently seen in the symptomatic group, though these differences were not statistically significant. Left atrial enlargement, for example, occurred in 72.2% of symptomatic patients, versus 47.1% of asymptomatic patients. The relatively small sample size likely contributed to this difference not reaching statistical significance and left atrial enlargement may be prognostic in a larger sample size. Additionally, the degree of regurgitation was higher in symptomatic patients but this also did not reach statistical significance. Although further research is needed, echocardiograms should continue to be utilized to monitor cardiac manifestations of HHT (28, 29).
Initial treatment of high-output heart failure secondary to liver AVMs is supportive and includes diuretics, salt and fluid restriction, beta-blockers and maintaining adequate hemoglobin levels (1, 20). Bevacizumab, an anti-angiogenic agent, has been successfully used to reduce shunting and mitigate symptoms of high-output heart failure, but may expose patients to adverse events including hypertension and arterial thromboembolism (30). Endovascular embolization is infrequently performed because of high complication risk from ischemic cholangitis, ischemic cholecystitis or hepatic necrosis (21). Orthotopic liver transplant has been proposed as the only definitive curative treatment, but is typically reserved for patients with severe complications related to hepatic AVMs (8, 21, 31). The current study shows only association, not causation, but measurement of CHA diameter and hemoglobin may help identify patients who are at higher risk of development of heart failure symptoms. This group may benefit from more aggressive correction of anemia, more frequent monitoring of symptoms and regular echocardiography. Further research is needed to explore if early or prophylactic treatment of this higher-risk group could improve outcomes associated with hepatic AVMs.
Our study had limitations inherent to studies with a small sample size performed at a single academic center and a retrospective design. Patients at our institution are not universally screened for hepatic AVMs and abdominal CTAs were often performed in patients with suspected hepatic AVMs due to symptoms. This likely led to the high proportion of symptomatic patients in our cohort. In our study, 44.4% of patients with hepatic AVMs were classified as having symptoms attributed to hepatic AVMs compared to the 8%-15% of patients reported in the literature (5, 21, 23). Although chest CTAs were typically performed to screen for or evaluate PAVMs, they were included in this study because our chest CTA protocol includes the liver parenchyma and the hepatic vasculature. It is possible that hepatic AVMs in the inferior liver could have been excluded from the scan plane in some patients, causing mis-classification. These two factors could have led to exclusion of some patients with small, asymptomatic hepatic AVMs which would have only been detected by more extensive screening.