Clinical Features and Diagnosis.

Most individuals with HHT have nosebleeds, but are otherwise often minimally symptomatic. Telangiectasias increase with age: by age 16 years, 71% of individuals will have developed some sign of HHT, increasing to more than 90% by age 40 years.

To permit a high level of clinical suspicion without leading to overdiagnosis, international consensus diagnostic criteria were developed based on four findings: (1) spontaneous recurrent nosebleeds, (2) mucocutaneous telangiectasias, (3) visceral involvement, and (4) an affected first-degree relative ( Table 61-2 ). When three criteria are present, “definite HHT” can be diagnosed; with two criteria, most commonly family history and nosebleeds, “suspected HHT” is diagnosed. With only one criterion, typically spontaneous nosebleeds, in a patient with neither a family history nor a first-degree relative of an HHT patient, and with no signs of the disease, the diagnosis of HHT is “unlikely.” A crucial issue for families (and medical practitioners) is that no child of a patient with HHT can be informed that he or she does not have HHT, unless the child has had a molecular diagnosis that demonstrates that he or she has not inherited the HHT-causing gene mutation for that family.

Genetics and Pathogenesis.

HHT is a genetically heterogeneous condition. Three disease-associated genes have been identified. HHT1 is caused by mutations in ENG, encoding endoglin, and HHT type 2 (HHT2) by mutations in ACVRL1, encoding activin receptor–like kinase (ALK-1). More rarely, mutations in SMAD4 cause HHT, usually in association with juvenile polyposis/HHT. An HHT-like syndrome is caused by mutations in the gene coding for bone morphogenetic protein (BMP) 9. There are at least two further unidentified genes that can cause classic HHT, that is, HHT3 on chromosome 5q between D5S2011 and D5S2490 and HHT4 on chromosome 7p between D7S2252 and D7S510 .

A body of evidence indicates that HHT mutations result in a nonfunctional allele and haploinsufficiency, which leads to a lack of sufficient protein for normal function. The most obvious mutations that will fail to generate a protein include entire gene deletions, start codon mutations, and mutations with no detectable mutant RNA. In addition, the majority of mutations reported on the HHT mutation database lead to premature termination codons, indicating that the mutated RNA species will undergo nonsense-mediated decay. Although there have been suggestions that the telangiectasias/AVMs may develop at sites where there was a genetic “second hit,” it is believed that in most, if not all cases, HHT results from haploinsufficiency.

Phenotype-Genotype Correlations.

Pulmonary AVMs are more common in patients with either HHT1 (ENG) or juvenile polyposis/HHT (SMAD4) than in HHT2 ( ACVRL1 mutations). A characteristic finding is that different members of the same HHT family display different patterns of disease. Recent data have identified a modifier gene that makes PAVMs more likely. Although there was an initial suggestion that overall severity of disease is greater in HHT1 than HHT2, this study predated the recognition of pulmonary hypertension (see later) and, in a later series, there was no difference in 90-month mortality.

Cellular Basis.

The genes mutated in HHT encode proteins that mediate signaling by the transforming growth factor-β (TGF-β) superfamily. Superfamily ligands such as TGF-βs, activins, and BMPs affect cellular growth and differentiation through signal transduction cascades from transmembrane receptor complexes. Endoglin is a relatively endothelial-specific co-receptor for multiple receptor complexes of the TGF-β superfamily. ALK-1 represents an endothelial-specific type I receptor that structurally and mechanistically belongs to the BMP branch of type I receptors. ALK-1 can associate with at least two type II receptors, BMPR2 and TβRII. In turn, TβRII can associate with two different TGF-β type I receptors in endothelial cells (TβRI [also known as ALK-5] or ALK-1), activating different Smad pathways, and apparently resulting in opposing endothelial cell responses in terms of proliferation, migration, and proangiogenic or antiangiogenic gene expression. Recent HHT concepts include the “balance hypothesis,” whereby the HHT mutations modify the predominant endothelial TGF-β type I receptor, Smad pathway, and ultimately endothelial cell response ; and models incorporating BMP9 and BMP10, which are specific ALK-1 ligands that can also bind endoglin. The likelihood of BMP9 being the ligand contributing to HHT pathogenesis increased with the recent identification of an HHT-like syndrome caused by BMP9 mutations. How the disease gene mutations lead to the vascular pathology has proved difficult to unravel. Attention now focuses on aberrant vascu­lar responses to injury-induced angiogenic stimuli; in this setting, the mutated genes in HHT appear to result in the inability of a blood vessel to mature appropriately.

Cavopulmonary Shunts.

In non-HHT patients, PAVMs commonly develop in those who have undergone surgical treatments of several forms of complex cyanotic congenital heart disease resulting in anastomoses between the superior vena cava (SVC) and inferior vena cava (IVC) and the pulmonary arteries. In the Glenn anastomosis, the SVC is redirected to provide the venous blood flow for the pulmonary arteries. Using angiography, PAVMs were detected in 31% of patients undergoing classic Glenn anastomoses after a mean follow-up of 6.8 years.

The incidence of PAVMs increases substantially if microscopic AVMs detectable by contrast echocardiography are included ; moreover, microscopic arteriovenous shunting may develop within 2 hours of the procedure. Some have suggested that, following superior bidirectional cavopulmonary anastomosis (Kawashima procedure, in which almost all venous blood except the hepatic venous blood flow is redirected to the pulmonary arteries), the development of functional intrapulmonary shunts may be universal. Contrast-enhanced magnetic resonance (MR) angiography is emerging as a useful assessment modality for these patients.

The key etiologic feature appears to be the route taken by hepatic venous flow, because PAVMs develop in the lung that receives no or minimal hepatic venous return, and regress if hepatic venous blood flow to the lung is restored. Serum levels of endostatin, a potent inhibitor of angiogenesis produced by the liver, drop after Glenn procedures (n = 17; 4.42 vs. 3.34 ng/mL; P < 0.001) (in which only SVC, not IVC, blood is directed to the lungs), but not after Fontan procedures (n = 13) (in which both SVC and IVC blood is directed to the lungs). Because a decrease in endostatin may promote angiogenesis, the authors propose that the potential role of endostatin in the pathogenesis of PAVMs warrants further study.

Right-to-Left Shunt.

PAVMs provide an anatomic R-L shunt, with the proportion of the cardiac output using the shunt pathways (shunt fraction) varying widely but reaching up to 60% in severe cases. In early series, mean R-L shunt fractions were generally higher (23% to 38%) than was found in later reports (8% to 13%), in which earlier diagnosis had taken place owing to more energetic screening procedures. Arterial P o ₂ and arterial oxygen saturation (S o ₂ ) are inversely related to the size of the R-L shunt fraction. The contribution of ventilation-perfusion mismatch to arterial hypoxemia in PAVMs is small, except in the occasional patient with significant coexisting lung disease.

Compensations for hypoxemia are usually highly successful. Chronic adaptations include secondary erythrocytosis which preserves the arterial oxygen content. Acute responses utilize increased cardiac output through heart rate (for acute falls in arterial oxygen content, e.g., on standing), and stroke volume. which can preserve the oxygen pulse (oxygen utilized/delivered per heart beat) at rest and on exercise.

Pulmonary Hemodynamics.

The absence of a microvascular network of capillary vessels in the PAVMs means that the pulmonary vascular resistance (PVR) of PAVMs is less than that of the surrounding normal lung. The effect on the overall PVR depends on the proportion of the cardiac output flowing through the shunt channels. In one study of eight patients with PAVM with large R-L shunt fractions (mean 31% ± 4% [standard error]), mean PVR was 0.33 ± 0.08 mm Hg/L/min (normal 0.5 to 1.3 mm Hg/L/min) and pulmonary artery pressure (P pa ) was 14 ± 0.6 mm Hg (normal 12 to 16 mm Hg), respectively. PVR was low, despite normal P pa , because total pulmonary blood flow was high (8 ± 0.8 L/min [160% of predicted]). In later studies with smaller mean R-L shunt fractions (8.5% to 11.5%), mean pulmonary systolic and diastolic pressures have been in the normal range, and recent studies have focused on the occasional presence of coexisting pulmonary hypertension (see section on HHT-related pulmonary hypertension).

Pulmonary Function.

Vital capacity is generally normal. There is no airflow obstruction unless a second pathology such as asthma or COPD is present. With large R-L shunts (>20%), the carbon monoxide diffusing capacity (D l _CO ) is often moderately reduced (71% to 78%), but in the majority of patients with less R-L shunting, D l _CO is equal to or greater than 90% of predicted (interquartile range, 76% to 100%). Patients with the lowest D l _CO values generally have widespread and small vascular malformations.

Posture.

PAVMs are more common in the lower lobes than in the upper lobes. Hence, for gravitational reasons, the shunt fraction tends to increase when patients stand up; in one study, R-L shunt increased from 28.7% to 39% in eight patients upon standing. Accordingly, arterial oxygen saturation falls. In 257 patients reported recently, 75 (29%) demonstrated orthodeoxia with an oxygen saturation fall of at least 2% on standing. This was accompanied by an orthostatic tachycardia, with an age-adjusted pulse rise of 0.79 min ^-1 per 1% arterial S o ₂ fall ( P < 0.001).

Exercise.

In the healthy lung during exercise, PVR falls to half its value at rest, attributed to dilation and recruitment of vessels in the pulmonary capillary bed. The effects of exercise on pulmonary hemodynamics in a PAVM-affected patient depend on the change in vascular resistance through the shunt channels in relation to the change in the resistance of the normal channels. In one group of eight patients with PAVM with severe arterial hypoxemia on exercise (arterial oxygen saturation 74% ± 3%), there was an excessive increase in total pulmonary blood flow (142% of predicted) in relation to the observed oxygen consumption ( ), resulting, in turn, in higher than predicted tissue oxygen delivery on exercise.

The change in arterial oxygen saturation from rest to exercise depends both on the change in the shunt fraction and the fall in mixed venous oxygen saturation on exercise. For patients with PAVM, overall, the fall in arterial oxygen saturation (rest to exercise) averaged 6% for mean shunts greater than 30%, 3% for mean shunts of 20% to 25%, and 1% to 2% for mean shunts less than 12%.

Overall, gas-exchange efficiency in terms of the ventilatory equivalent ( ) is abnormally high on exercise; observed values are directly related to the exercise R-L shunt and inversely related to exercise arterial oxygen saturation.

Work capacity is well preserved in PAVM patients, even when arterial oxygen saturation on exercise is less than 80%. The adaptive responses are lost following correction of hypoxemia. In a recent study, despite higher arterial S o ₂ , treated patients achieved similar work rates and similar peak oxygen consumption. It is noteworthy that treated patients reset to virtually identical peak oxygen pulses and, in many cases, to the same point on the peak oxygen pulse/work-rate plot.

Dyspnea.

Dyspnea is the respiratory symptom most commonly reported by PAVM patients, but this is present in less than 50% of all cases (see Table 61-1 ) and may not be appreciated until after the condition has been treated. In one series of 219 consecutive patients, symptomatic dyspnea was generally only present when resting arterial oxygen saturations were less than 80%. Three subsequent studies showed no relationship between arterial S o ₂ and dyspnea as self-reported (N = 165, arterial S o ₂ 78.5%-99%); by Borg scales during cardiopulmonary exercise testing (N = 21, arterial S o ₂ 80%-96%); or as self-reported during flight (N = 99; arterial S o ₂ 85%-99%). None of 75 consecutive patients demonstrating orthodeoxia reported platypnea (dyspnea on standing).

Hemoptysis.

The fragile vessels in HHT should be more prone to hemorrhage than normal pulmonary vessels but, surprisingly, hemoptysis and hemothorax are relatively rare features for PAVMs (see Table 61-1 ), with two important exceptions: (1) with a spontaneous or postembolization systemic arterial blood supply to PAVM sacs, and (2) with pregnancy-associated changes. Both conditions place patients at higher risk of hemorrhage from PAVMs, which may be massive and life-threatening.

Chest Pain.

Pleuritic chest pain of uncertain etiology may be described in up to 10% of PAVM patients (see Table 61-1 ). PAVM associations are likely to be overestimated, however, in series that do not correct for ascertainment bias of incidental PAVM detection following protocol-driven CT scans for suspected pulmonary embolism.

Contrast Echocardiography.

The method of contrast echocardiography (CE; synonym “echo-bubble”) demonstrates intrapulmonary shunts by tracing the circulatory transit of microbubbles generated by intravenously injected echocontrast material. Microbubbles seen in the left heart should therefore be the result of an R-L shunt, either cardiac (most commonly a patent foramen ovale) or pulmonary. Typically, with an intrapulmonary shunt such as a PAVM, the number of bubbles in the left heart increases after a matter of seconds ( Fig. 61-5 andVideo 61-1 ). The entry of bubbles is not affected by the cardiac cycle or respiration, and though influenced slightly by a Valsalva maneuver, this is in a more subtle manner to the changes observed with a patent foramen ovale.

Contrast echocardiography showing different grades of left ventricular opacification.

The grades shown are minimal (A) , moderate (B) , extensive (C), and extensive (D) with endocardial definition. The likelihood of detecting a pulmonary arteriovenous malformation on chest CT increases with higher grades of abnormality.

(From Parra JA, Bueno J, Zarauza J, et al: Graded contrast echocardiography in pulmonary arteriovenous malformations. Eur Respir J 35:1281–1285, 2010. Reproduced with permission of the European Respiratory Society.)

CE consistently detects more intrapulmonary R-L shunting in HHT patients than any other PAVM screening modality. False-negative results can result, but it was early reports demonstrating high sensitivity with no attendant radiation burden that led to current guidelines recommendations to use CE as the first-line method for PAVM screening.

Presence of a positive shunt by echocardiography does not imply that a macroscopic PAVM will be found. Appreciable proportions of the general population exhibit intrapulmonary shunting; in one early study, intrapulmonary shunts were found in 5 of 19 healthy subjects in whom contrast was injected directly into the pulmonary artery, thereby bypassing any intracardiac shunt. More recent studies using CE documented shunting in 6% to 7% of control subjects at rest, rising to as high as 90% on exercise. Previous studies demonstrated a high frequency of false-positive results in which PAVMs were not identified by subsequent angiography, considered to be the “gold standard” for detection of PAVMs. For patients with endoglin mutations, a positive CE study only provided a positive predictive value for a treatable PAVM of 36.3%.

Reassuringly, in the HHT-based research studies, no treatable PAVMs were found in individuals with negative CE studies, although these accounted on average for only 39% of screened individuals. In one study of patients stratified by HHT mutations, 85% of those with endoglin mutations and 35% of those with ACVRL1 mutations had a positive shunt by CE.

Specialist groups have incorporated grading systems to examine whether clinically significant PAVMs could be excluded by low-grade positive CE studies (see Fig. 61-5 ). The broad grades of CE shunt severity range from grade 1 (found in at least 7% to 8% of the general population ), to grades 3 and 4, which are more frequently associated with visible PAVMs seen on CT (see Fig. 61-3 ) and neurologic complications. In a recent two-center study of 1038 patients undergoing PAVM screening, 530 (51%) had a positive CE shunt, but there was no enhanced stroke risk in patients with grade 1 shunts (<30 microbubbles per frame). For patients with grade 2 (30 to 100 microbubbles per frame) or grade 3 (>100 microbubbles per frame) shunts, the odds ratios for cerebral ischemic event or brain abscess were 4.8 ( P = 0 .03) and 10.4 ( P = 0.002), respectively. In general, with increasing grade of abnormality on CE, the likelihood of finding a PAVM on CT and the risk of embolic events increase.

Long-Term Outcome.

Anatomic long-term results of transcatheter embolization have been evaluated by several centers. After embolization, residual shunting through untreatable (<2 to 3 mm diameter) arterial feeding vessels is common (see eFig. 61-4H ). Additionally, treated PAVMs may recanalize and/or reperfuse. Factors associated with reperfusion include a low number of coils, oversized coils, proximal placement of coils, and altered pulmonary hemodynamics (development of pulmonary hypertension or presence of hepatic AVMs). In one series of 192 PAVM patients in whom feeding arteries less than 3 mm in diameter were embolized, 70% had residual disease, a finding supported by other studies ( eTable 61-1 ). It has been recommended to use CT scans for 6-month and 1-to-3-years follow-up, but the recently documented radiation burden calls this into question: In one series of 246 PAVM patients with HHT, CT scans accounted for 46% of the mean cumulative effective dose (CED).

Physiologic Outcome.

Substantial improvement in oxygen saturation is the rule for patients with preembolization hypoxemia (see eTable 61-1 ), with little effect on other pulmonary function measurements. Compensatory mechanisms reset; falls in hemoglobin restore arterial oxygen content to pre-embolization levels. Stroke volume and cardiac output also fall after treatment of PAVMs, and oxygen consumption at peak exercise is unchanged.

Clinical Outcome.

Unsurprisingly exercise capacity, which is often relatively unimpaired before embolization, only improves in a subgroup of patients. Concurrent cardiopulmonary disease was a predictor of improvement in one series of 98 treated patients. Several studies have now demonstrated the clinical efficacy of embolization in improving stroke/abscess risk and reducing the prevalence of migraine.

Radiation Exposure.

New data demonstrate that currently employed protocols can result in levels of radiation exposure that would be classified as harmful. In a single center study of 246 PAVM patients (53 years mean age), the mean CED over an 11-year period was 51.7 mSv, and CED exceeded 100 mSv in 26 patients (11%). Interventional procedures accounted for 51% of the CED.

Development of Systemic Arterial Supply.

The risk of massive hemoptysis from PAVM sacs that persist after embolization and that acquire a systemic arterial collateral blood supply was first highlighted in a small series published in 1998. In view of the known importance of pulmonary-bronchial communications (see Systemic-to-Pulmonary Vascular Communications), systemic collaterals might be expected to develop to supply an area of the pulmonary capillary bed that had lost its pulmonary arterial supply as a result of embolization. Systemic supply is only of consequence if the fragile PAVM sac persists. None of the patients in Brillet and coworkers’ series experienced hemoptysis, in contrast to the majority of those in a smaller series. To reduce the risk for development of systemic arterial collateral supply to any persistent sac, standard practice is to place embolization devices as close as possible to the neck of the malformation (see eFigs. 61-4C-H and 61-5 ). This is particularly difficult in patients with diffuse PAVMs, in whom sac persistence is inevitable. Expert institutions differ in the degree to which such PAVMs are embolized. At our institution, more limited embolization is undertaken and in the presence of ongoing neurologic symptoms or hemoptysis, surgical resection is considered. Other approaches include dense packing of the most severely involved segmental arteries.

Development of Pulmonary Hypertension.

PAVM embolization may be expected to elevate P pa , because PAVMs provide low-resistance pathways for pulmonary blood flow. Individual cases in which P pa increased both after embolization and after surgical resection are reported. In one series of 35 patients, there was no evidence of a sustained or acute change in P pa in the majority of patients and, in half, embolization led to a fall in P pa .

Should embolization of PAVMs be performed in patients with severe preexisting pulmonary hypertension? This question was specifically addressed in light of lower stroke risk and data indicating that test balloon occlusion did not predict subsequent rise in P pa following definitive embolization. The main indications for PAVM embolization are to reduce the risk of paradoxical embolic stroke/brain abscess and, for individuals with hypoxemia, to improve dyspnea and exercise tolerance. The authors concluded that, for patients with preexisting severe pulmonary hypertension, the risks of PAVM embolization generally outweigh potential benefits. It was recognized that the most difficult judgments relate to individuals with severe pulmonary hypertension and major hemoptysis. Additionally, it is now recognized that higher P pa is one of the predictors for symptomatic improvement post embolization.

Surgery.

Surgery remained the treatment of choice until the 1980s but was never the ideal solution for the multiple PAVMs of HHT; more recently, however, surgery has been a useful adjunctive therapy for selected cases. There may be times when small PAVMs are single or sufficiently localized for thoracoscopic resection when embolization is not feasible. At our institution, and elsewhere, elective surgical treatments are reserved for patients demonstrating ongoing ischemic strokes or transient ischemic attacks following maximal embolization. In emergency situations, particularly associated with massive hemoptysis, lobectomy or pneumonectomy may be appropriate.

Lung transplantation has been undertaken in a few patients with severe hypoxemia secondary to diffuse disease. The long-term complications of untreated PAVMs, however, are likely in most cases to be less than transplantation-associated morbidity and mortality. The three patients with PAVM in our clinic (one male, two female) who elected not to proceed with transplantation—after discussion of the risks at two different transplantation centers—have since remained stable over 20, 22, and 25 years, and one patient has had three successful pregnancies. In a retrospective series of 36 patients with diffuse PAVMs for whom follow-up data were available for a mean of 8.5 years (range, 0.12 to 26 years), 24 of the 27 survivors were working or studying full time; one of the deaths was transplantation associated.

Dental Issues.

For patients with PAVMs and HHT, owing to the strong link between oral bacteria and cerebral abscess, antibiotic prophylaxis before dental and surgical procedures was recommended based on the endocarditis paradigm. The evidence for an association between oral microorganisms and brain abscess has since been strengthened. American Heart Association and British National Institute for Health and Care Excellence guidelines indicate that antibiotic prophylaxis is no longer required for most patients with structural heart disease at risk for infective endocarditis. This has led to confusion among dentists and medical practitioners caring for patients with PAVM. A recent article exploring why PAVM/HHT patients do not fall into the groups considered by the guidelines provided recommendations to reduce the risk of dental bacteremias, including the use of antibiotic prophylaxis before dental procedures.

Pregnancy.

In view of the risks of PAVM growth and rupture during pregnancy, it is recommended that female patients be advised to defer pregnancy pending formal PAVM assessment and treatment. Pregnancies should be managed with close liaison between obstetricians, pulmonologists, and interventional radiologists, using appropriate “high-risk” obstetric management strategies. Patients and their medical practitioners should be alerted to the possibility of hemoptysis or sudden severe dyspnea that requires urgent admission and management: embolization in the second and third trimesters is feasible. The question of whether PAVM embolization should be offered to asymptomatic pregnant women differs between countries; at our institution it is not performed.

Thromboembolic Risks.

The American Stroke Association recommends antiplatelet agents for secondary prevention of ischemic stroke in PAVM patients. Antiplatelet therapy can be considered on a case-by-case basis for patients, even if there is underlying HHT. Venous thromboemboli are common, associated with conventional venous thromboembolism risk factors and iron deficiency, and both prophylaxis and treatment with anticoagulants may be required, even when HHT is present.