Pulmonary arteriovenous malformations: Clinical features and diagnostic evaluation in adults

INTRODUCTION — Pulmonary arteriovenous malformations (PAVMs) are abnormal communications between pulmonary arteries and veins [1]. Alternative names include pulmonary arteriovenous fistulae, pulmonary arteriovenous aneurysms, cavernous angiomas of the lung, and pulmonary telangiectases [2]. PAVMs are uncommon, but they are an important consideration in the differential diagnosis of common pulmonary problems, including dyspnea, hemoptysis, hypoxemia, and pulmonary nodules.

Clinical features, indications for diagnostic testing, and an approach to the diagnostic evaluation of suspected PAVMs are reviewed here. The epidemiology, etiology, pathology, and treatment of PAVMs are discussed separately. (See "Pulmonary arteriovenous malformations: Epidemiology, etiology, and pathology in adults" and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations".)

CLINICAL MANIFESTATIONS — Although the literature prior to 2000 [3] suggested that most PAVMs were associated with pulmonary symptoms, studies since 2000 describe pulmonary symptoms in only 20 to 65 percent (approximately 40 percent) and the remainder are asymptomatic, typically found incidentally on chest imaging or during screening of patients with known or suspected hereditary hemorrhagic telangiectasia (HHT) [4-8]. The most common pulmonary symptoms are dyspnea in 13 to 56 percent and hemoptysis in 7 to 30 percent (table 1) [4,5,8,9]. Patients with underlying HHT (which is the most common cause of PAVMs), often show symptoms attributable to this disorder including epistaxis and mucocutaneous telangiectases (table 2). In addition, PAVMs should always be suspected in patients who present with unexplained dyspnea or hypoxemia as well as in patients with nodules and a history of a stroke or brain abscess (stroke and brain abscess are common complications of PAVMs). (See 'Features attributable to PAVM' below and 'Features attributable to HHT' below and 'Features attributable to PAVM complications' below and "Pulmonary arteriovenous malformations: Epidemiology, etiology, and pathology in adults", section on 'Etiology'.)

Symptoms related to PAVMs typically begin during the fourth through sixth decades of life [10], whereas symptoms of HHT frequently develop before the age of 20 years (eg, epistaxis due to nasal telangiectases or appearance of telangiectases on the skin and lips (table 2)) [11,12].

The frequency and severity of symptoms tend to be greater among those who have larger PAVMs, multiple PAVMs, a higher shunt fraction, or HHT, although not all studies have found this [13-16]. Diffuse microvascular PAVMs are almost always symptomatic (image 1 and image 2) [17].

When compared with patients with HHT, patients with idiopathic PAVMs are expected to present with similar manifestations and complications (eg, dyspnea and stroke) except for a lesser incidence of epistaxis, telangiectases, and multiple PAVMs [18]. Two retrospective studies totaling 118 patients demonstrated that non-HHT PAVMs have a higher female predominance (60 to 73 percent), are more often solitary (83 to 90 percent), and are often symptomatic with a high rate of neurologic complications [19,20]. The causes of PAVMs are presented in the table (table 3). (See "Pulmonary arteriovenous malformations: Epidemiology, etiology, and pathology in adults".)

Features attributable to PAVM

Dyspnea — Dyspnea is the most common complaint directly attributable to PAVMs (table 1). Older retrospective series report about half of patients with PAVMs have dyspnea [4,18]. However, with the increased use of computed tomography (CT) particularly in patients with HHT, the detection of smaller PAVMs may have reduced the frequency of this symptom with case series published since 2006 reporting dyspnea in 13 to 56 percent [4,5,8,9,13,18,21]. In a study of 202 patients with HHT and PAVM, 44 percent had dyspnea [21]. However, HHT patients with or without PAVM frequently had other causes of dyspnea such as asthma or anemia [21]. Dyspnea is especially common among patients with clubbing and among patients whose PAVMs are large, multiple, bilateral, or diffuse [3]. It may occasionally be caused by high output heart failure due to hepatic AVM or severe anemia due to bleeding in patients with HHT (ie, extrapulmonary shunts) [22]. The mechanism of dyspnea is unclear but may relate to hypoxemia from shunt or airflow limitation [23]. (See "Measures of oxygenation and mechanisms of hypoxemia".)

Hemoptysis — Hemoptysis, presumably due to endobronchial or intraparenchymal PAVM rupture, occurs in 7 to 30 percent of patients with PAVM (table 1) [4,8,9]. Hemoptysis may be massive with the potential to be life-threatening especially during pregnancy, the details of which are discussed separately. Hemoptysis in PAVM patients should be taken seriously as it may sometimes be a sign of an impending and life threatening rupture. (See 'Hemothorax and hemoptysis' below and "Evaluation of nonlife-threatening hemoptysis in adults" and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Pregnancy' and "Hereditary hemorrhagic telangiectasia (HHT): Evaluation and therapy for specific vascular lesions", section on 'Pregnancy' and "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)", section on 'Pulmonary AVMs' and "Evaluation and management of life-threatening hemoptysis".)

Platypnea and orthodeoxia — Platypnea is dyspnea that is induced by the upright position and relieved by recumbency; it is believed to be the result of increased blood flow through PAVMs in the dependent portions of the lungs upon assuming the upright position. Platypnea is a classic but rare symptom in patients with PAVMs. As an example, in one eight-year prospective series of 258 patients with PAVMs, no patient reported platypnea [24]. In contrast, several studies have found orthodeoxia (ie, a decrease in the oxyhemoglobin saturation by 2 percent or more when rising from the supine to the upright position) was reported in at least 30 percent of patients [24-27]. Platypnea and orthodeoxia are not unique to PAVM; they may also occur with the hepatopulmonary syndrome, atrial septal defects (including patent foramen ovale), and other conditions [28].

Chest pain and cough — Other less common symptoms that may be seen among patients with PAVM include chest pain and cough, presumably related to pleural or endobronchial involvement [13,18].

Clubbing and cyanosis — Clubbing and cyanosis indicate severe and/or multiple/diffuse PAVMs (table 1). In older case series, they were often seen in more than half of patients [2,14,15,29], but they have been observed less often (<5 percent) in subsequent series [13,18].

Murmurs and bruits — Murmurs or bruits auscultated over the PAVMs are auscultated in less than 10 percent of cases of PAVMs (table 1). The murmurs are loudest during inspiration and more likely to be heard when the PAVM is in a dependent position. (See "Auscultation of cardiac murmurs in adults", section on 'Arteriovenous fistulas'.)

Features attributable to HHT — The most common clinical manifestations of hereditary hemorrhagic telangiectasia (HHT) are epistaxis, mucocutaneous telangiectases (picture 1), gastrointestinal (GI) bleeding from GI tract telangiectases, and blood loss anemia (table 2). About one-third of patients with HHT have dyspnea, although the proportion of this group that have PAVMs is unclear [21]. Detailed discussion of the clinical manifestations of HHT is provided separately. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)", section on 'Clinical features'.)

Telangiectases on the skin and mucous membranes are usually flat or slightly papular, ruby colored, 1 to 3 mm in diameter, and sharply demarcated from the surrounding skin. In addition, they tend to blanch easily with pressure and are most commonly located on the face, tongue, lips, and distal upper extremities (picture 2) [12]. These characteristics differentiate telangiectases on the skin from the spider nevi that are common in advanced liver disease, which are almost always flat, blanch readily, refill centrally first, and are mainly found on the face and anterior chest. Capillary microscopy of the finger nail bed is a simple and noninvasive technique that may occasionally reveal abnormal capillary loops in patients without characteristic telangiectases, although this test is rarely performed. (See "Clinical manifestations and diagnosis of Raynaud phenomenon", section on 'Nailfold capillary microscopy'.)

Importantly, some patients with HHT may present with PAVMs as their only manifestation or have limited findings of HHT [30].

Features attributable to PAVM complications — PAVMs can be associated with a variety of complications, some of which are life-threatening, including stroke, brain abscess, and hemoptysis (table 4). The true incidence of patients that initially present with a complication is unknown but probably about 15 to 30 percent in the era of CT evaluation of patients with HHT for PAVMs. The spectrum and rate of complications from retrospective case series, where screening was not commonly performed, are discussed in detail below.

Neurologic — The rate of neurologic complications in patients with PAVMs ranges from 9 to 41 percent in several large series published since 2004 [4-9]. The most common neurologic complications are stroke and brain abscess, but patients can also suffer from seizures, migraines, hemorrhage, headaches, dizziness, syncope, diplopia, and tinnitus [13,18]. Higher rates of stroke or brain abscess have been reported in those with diffuse PAVMs (up to 70 percent) [3,29,31]. (See "Initial assessment and management of acute stroke" and "Pathogenesis, clinical manifestations, and diagnosis of brain abscess".)

As examples:

●One series of 76 consecutive patients undergoing embolotherapy reported the following frequency of neurologic events [25]:

•CT evidence of stroke – 36 percent

•Symptomatic stroke – 18 percent

•Transient ischemic attack – 37 percent

•Brain abscess – 9 percent

•Migraine headache – 43 percent

•Seizure – 8 percent

•Cerebral arteriovenous malformation (CAVM) – 5 percent

●A case series of 219 patients with PAVM found evidence for neurologic events in 34 percent including clinical stroke in 14 percent and brain abscess in 13 percent [7].

Paradoxical embolization across a PAVM is the most likely mechanism for both strokes and brain abscesses, which is also supported by the observation that these complications are more frequent in patients with feeding arteries >2 to 3 mm in diameter. The mechanism for seizures, migraines, and other neurologic symptoms is less clear, but cerebral hypoxia due to shunting through PAVMs or symptoms due to concurrent CAVMs may contribute. Cerebral hemorrhage may be a complication of embolic stroke or due to a ruptured CAVM.

The risk of neurologic complications increases with the number of PAVMs and the patient’s age [31,32]. As an example, in one case series of 75 patients with PAVMs, radiologic evidence of cerebral infarction was found in 32 percent of patients with single PAVMs and 60 percent of patients with multiple PAVMs [31]. In another study of 71 patients with HHT and PAVM, embolic neurologic complications (ie, stroke, transient ischemic attack, or cerebral abscess) occurred in 10 percent of patients less than 30 years old and 45 percent of older patients [32].

Hemothorax and hemoptysis — Hemoptysis of any significant degree (more than blood streaking) in a patient with known PAVMs should be considered a medical emergency, prompting urgent hospitalization, evaluation, and institution of appropriate treatment. Hemothorax and hemoptysis occur in less than 10 to 20 percent of patients with PAVMs, but are potentially life-threatening [4,8,9,33]. Hemothorax results from rupture of a subpleural PAVM, while hemoptysis is likely due to rupture of either a parenchymal PAVM or an endobronchial telangiectasia. In a series of 143 patients with PAVMs who were referred for embolotherapy, 11 patients (8 percent) had a history of hemoptysis or hemothorax requiring hospitalization [34]. (See "Evaluation of nonlife-threatening hemoptysis in adults".)

Pulmonary hemorrhage may occur during pregnancy, especially during the third trimester and can be fatal. Pregnancy may exacerbate PAVMs through increased blood volume, increased cardiac output, and increased vascular distensibility, all of which promote increased blood flow through PAVMs. As a result, women with HHT should be evaluated for PAVMs and treated maximally before pregnancy, although treatment may be safely undertaken during pregnancy (preferably between 16 and 26 weeks of gestation), if required [35]. (See "Therapeutic approach to adult patients with pulmonary arteriovenous malformations" and "Hereditary hemorrhagic telangiectasia (HHT): Evaluation and therapy for specific vascular lesions", section on 'Pregnancy'.)

Polycythemia and anemia — Polycythemia and anemia were each seen in approximately 18 percent of patients with PAVMs in older series but polycythemia is now rare [13,14,17,36]. Polycythemia is likely due to chronic hypoxemia, and anemia is due to blood loss from epistaxis and/or associated gastrointestinal lesions related to underlying HHT. (See "Diagnostic approach to the patient with erythrocytosis/polycythemia" and "Diagnostic approach to anemia in adults".)

Pulmonary hypertension — Approximately 10 percent of patients with PAVMs associated with HHT have pulmonary hypertension [3,29,37-39]. As an example, one study of 143 patients undergoing PAVM embolization reported that 9 percent of patients had a mean pulmonary artery pressure >20 mmHg and 6 percent had a pressure >25 mmHg [37]. Another study of 651 HHT patients undergoing PAVM embolization reported that 13 percent of patients had a mean pulmonary artery pressure ≥25 mmHg and 2.8 percent had a pressure >35 mmHg [39]. Importantly, the elevated pulmonary vascular pressures and flow may predispose to enlargement of the underlying PAVM, leading to an increased risk of rupture [38] and possibly mortality [39].

The development of pulmonary hypertension is not directly due to the PAVM itself, but rather due to one of the following:

●Concurrent hepatic disease where chronic left-to-right shunting through hepatic AVMs causes high output left heart failure (ie, group 2 pulmonary hypertension) or portopulmonary hypertension (ie, group 1 pulmonary artery hypertension) [3,22]. (See "Causes and pathophysiology of high-output heart failure" and "Portopulmonary hypertension" and "Clinical features and diagnosis of pulmonary hypertension of unclear etiology in adults", section on 'Postdiagnostic testing and classification'.)

Pulmonary artery catheterization will distinguish group 1 from group 2 pulmonary hypertension. Patients with high output left heart failure (group 2) typically have an elevated pulmonary artery wedge pressure (PAWP), a markedly increased cardiac index (CI), and a normal pulmonary vascular resistance, while those with portopulmonary hypertension (group 1) have normal PAWP, a normal or low cardiac index, and a high pulmonary vascular resistance. In one study, 73 percent of patients with measurement of cardiac output were diagnosed as having postcapillary pulmonary hypertension [39], highlighting the importance of catheterization for an accurate diagnosis. (See "Clinical features and diagnosis of pulmonary hypertension of unclear etiology in adults".)

●Heritable pulmonary arterial hypertension (PAH) in patients with concomitant HHT, which is indistinguishable clinically from idiopathic pulmonary arterial hypertension (IPAH) and is usually related to defects in ALK-1 (ie, product of the HHT 2 gene) [40-42]. One study identified ten patients with pulmonary hypertension from four families with HHT, four of whom had histology-proven IPAH [41]. Genetic linkage analysis in those patients was consistent with alterations in ALK-1, which has a documented role in the pathogenesis of IPAH [42]. (See "Treatment and prognosis of pulmonary arterial hypertension in adults (group 1)".)

●New or increased pulmonary hypertension may occur following embolization or resection of large or multiple PAVMs, especially when the pretreatment shunt fraction is greater than 20 percent [43,44]. When PAVMs are obliterated during embolotherapy or resected, flow is redirected through residual pulmonary vasculature. A relative increase in flow through the residual pulmonary vascular bed may increase pulmonary vascular pressure, particularly in those with mild or subclinical pulmonary hypertension at baseline. The value of pulmonary artery catheterization in predicting post embolization pulmonary pressures is discussed separately. (See "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Procedural complications'.)

Other complications — Other complications of PAVM include the following:

●Non neurologic systemic embolization – There are anecdotal reports of patients with infectious endocarditis, and abscesses in the soft tissue, muscle, or other organs (eg, spleen), all presumably due to paradoxic embolization [4]. (See "Clinical manifestations and evaluation of adults with suspected left-sided native valve endocarditis".)

●Complications of HHT – Patients with underlying HHT can suffer from epistaxis, gastrointestinal hemorrhage, and skin bleeding (table 2) [45].

●Growth – Most PAVMs remain stable in size. However, approximately 25 percent of PAVMs will enlarge slowly, usually at a rate of 0.3 to 2 mm/year [11,14]. Macroscopic PAVMs do not spontaneously resolve. (See "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Natural history'.)

Imaging findings — Conventional chest radiography (image 3) and/or nonenhanced CT of the chest (image 4 and image 5 and image 6) are abnormal in most patients who have symptoms attributable to PAVMs [2,11,13,14,16,17,36,46-49]. While the typical appearances of PAVMs on chest radiography and CT are similar (ie, a smooth nodule with a feeding artery and draining vein), in the era of CT evaluation of HHT patients and frequent use of CT in the general population, many patients present with atypical findings on chest CT [50]. Pulmonary angiography is not routinely used for the diagnostic evaluation of suspected PAVMs except in the setting of concomitant embolotherapy. In most centers in the United States, magnetic resonance imaging (MRI) is rarely used for the routine evaluation of suspected PAVM. A concise review of the appropriateness of various modalities for imaging PAVM has been published by the American College of Radiology [51].

●Chest radiography – Typical chest radiographic findings include a smooth nodule(s) with linear parallel shadows representative of the feeding pulmonary artery and draining pulmonary vein.

●Computed tomography – In a study of 108 patients with definite HHT and PAVM who also had CT scans, 62 percent had multiple lesions and 11 percent had 10 or more lesions [4]. Typical CT imaging characteristics that together are considered diagnostic include the following (image 7 and image 8):

•Round or oval nodule (<3 cm) or mass (≥3 cm) of uniform density (ie, the "sac").

•Well demarcated smooth borders with occasional lobulation.

•Usually 0.5 to 5 cm in diameter, but occasionally exceeding 10 cm in diameter.

•Usually 2 to 8 lesions among the patients who have multiple PAVMs [4].

•Visible feeding vessels that are visualized on CT as tubular structures, often parallel, that seamlessly blend into/terminate at the sac; the artery radiates from the hilum and the vein deviates toward the left atrium. Vessels are also typically larger than other vessels in the vicinity (especially for peripheral lesions).

•Good contrast enhancement on CT if the sac is 1 cm or greater and if contrast is used. Contrast is not necessary for diagnosis.

Importantly, not all patients present with these classic findings. Atypical findings on CT include an irregular sac with poorly defined borders, only one tubular structure leading into the sac, a tubular structure that continues past the nodule/sac, and irregular or no contrast enhancement (image 9). In some cases, these findings may indicate other diagnoses such as tumors or other vascular malformations such as systemic artery to pulmonary artery malformations or pulmonary vein varices [50]. (See 'Computed tomography' below.)

Telangiectatic-type PAVMs are another atypical finding in which an area of ground-glass density, typically 0.5 to 3 cm, is seen on CT (image 10). These are often associated with small- to moderate-sized feeding and draining vessels and may be associated with a smaller central tangle of vessels that have not yet coalesced into a typical "sac." Investigations are ongoing to determine the risk profile of these telangiectatic-type PAVMs and whether they turn into typical PAVMs over time.

●Conventional pulmonary angiography – Pulmonary arterial angiography is the gold standard for defining the anatomy of PAVMs identified previously on CT that are considered potentially suitable for embolotherapy but is also occasionally performed for the sole purposes of definitive diagnosis (image 11 and image 12 and image 13 and image 14) [47]. Typical findings include the demonstration of a feeding artery leading to the abnormal arteriovenous communication, which is drained by a pulmonary vein (image 15). (See 'Pulmonary angiography' below and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations".)

●Magnetic resonance angiography – Although MRI may show similar findings to those seen on CT, it is not as sensitive or as specific diagnostically. It is also more expensive, and highly specialized expertise is required for accurate interpretation. (See 'Magnetic resonance imaging' below.)

Laboratory tests

●General – Patients with PAVMs may have anemia or polycythemia when they occur in association with HHT involvement of the nasal passages and gastrointestinal tract or with long standing hypoxemia, respectively. (See "Clinical manifestations and diagnosis of polycythemia vera" and "Diagnostic approach to the patient with erythrocytosis/polycythemia" and "Diagnostic approach to anemia in adults".)

●Arterial blood gases – The incidence of hypoxemia in PAVM is highly variable and depends on how patients were accrued. In one series in which PAVM were diagnosed by screening HHT patients, 80 percent of patients had a SaO₂ of 96 percent or higher. (See "Measures of oxygenation and mechanisms of hypoxemia".)

Patients with diffuse PAVMs are uniformly hypoxemic (image 1 and image 2), with a mean arterial oxygen tension (PaO₂) of 47 mmHg according to a series of 16 patients with PAVMs [29]. Other case series have found that 81 to 100 percent of patients with PAVMs have either a PaO₂ <80 mmHg or an SaO₂ <97 to 98 percent on room air [14,16,17].

●Orthodeoxia – Orthodeoxia is the laboratory correlate of platypnea. It is defined as a decrease in the oxyhemoglobin saturation by 2 percent or more when rising from the supine to the upright position, the details of which are discussed above. (See 'Platypnea and orthodeoxia' above.)

Cardiopulmonary exercise testing — Cardiopulmonary exercise testing is usually performed in patients with unexplained dyspnea and is not typically performed in patients with PAVM for diagnosis. Most patients with PAVM have reduced exercise tolerance; however, their exercise tolerance tends to be better maintained than patients who have an equivalent degree of hypoxemia from another cause [52-54]. (See "Approach to the patient with dyspnea", section on 'Cardiopulmonary exercise testing' and "Approach to the patient with dyspnea", section on 'Cardiopulmonary exercise testing with pulmonary artery catheterization'.)

As an example, a study of 15 patients with PAVM reported that 73 percent achieved >70 percent of their predicted maximum workload despite their SaO₂ decreasing from 86 percent at rest to 73 percent with peak exercise [53]. Patients with similar degrees of hypoxemia due to other cardiac and pulmonary diseases have considerably less tolerance for exercise due to an increase in baseline pulmonary vascular resistance that prevents an appropriate increase in cardiac output during exercise. Presumably, patients with PAVM are able to increase pulmonary blood flow appropriately during exercise due to a subnormal pulmonary vascular resistance [52].

INDICATIONS FOR DIAGNOSTIC EVALUATION — PAVMs should be suspected and a diagnostic evaluation initiated in individuals with one or more of the following characteristics:

●Clinical features suggestive of PAVMs (table 1):

•Pulmonary nodule(s) on chest imaging that is suspicious for PAVM. (See 'Imaging findings' above.)

•Stigmata of right-to-left shunting, such as dyspnea, hypoxemia, polycythemia, clubbing, cyanosis, cerebral embolism, or a brain abscess.

•Unexplained hemoptysis, hemothorax, hypoxemia, or dyspnea.

•Platypnea or orthodeoxia.

The clinical features of PAVMs are discussed in detail separately. (See 'Clinical manifestations' above.)

●Suspected or known diagnosis of hereditary hemorrhagic telangiectasia (HHT; Osler Weber Rendu syndrome). As examples, those with:

•Mucocutaneous telangiectases or other features suggestive of HHT (table 2).

•Personal history of HHT (including pregnant women).

•Family history of HHT.

Individuals with a personal or family history of HHT should be evaluated for possible PAVMs, since there is a high incidence of unsuspected PAVMs in this population [10,11,55-57]. This is particularly important if at least one member of the family with HHT has already been diagnosed with PAVM, as the incidence of PAVM is approximately 35 percent among relatives [57,58]. It is also recommended that all women with HHT undergo evaluation before becoming pregnant, since PAVMs are associated with a high incidence of hemoptysis or hemothorax during the last half of pregnancy [34]. Issues regarding evaluating patients with HHT for undetected PAVMs are discussed separately. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)" and "Pulmonary arteriovenous malformations: Epidemiology, etiology, and pathology in adults", section on 'Hereditary hemorrhagic telangiectasia (HHT)' and "Hereditary hemorrhagic telangiectasia (HHT): Routine care including screening for asymptomatic AVMs", section on 'Overview of screening strategy'.)

PAVMs are more common with HHT-1 than HHT-2 mutations but both populations are at risk and should be screened. Genetic evaluation of patients with HHT and evaluation for cerebral AVMs are discussed separately. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)", section on 'Cerebral vascular abnormalities' and "Hereditary hemorrhagic telangiectasia (HHT): Routine care including screening for asymptomatic AVMs", section on 'Genetic counseling and testing of family members'.)

DIAGNOSTIC APPROACH — For patients with suspected PAVMs, we and others agree that transthoracic contrast echocardiography (TTCE) is the initial test of choice to evaluate for the presence of a right-to-left shunt [59]. Subsequent diagnostic testing with nonenhanced CT of the chest depends upon the identification and severity (ie, grade) of shunt found on TTCE. Our preferred approach is outlined in the algorithm (algorithm 1) and the sections below.

Shunt fraction assessment — PAVMs, particularly large and multiple PAVMs, shunt blood directly from the pulmonary artery to the pulmonary vein such that shunted blood is poorly oxygenated. TTCE is the test of first choice for the evaluation of a right-to-left shunt due to suspected PAVMs. Importantly, TTCE is not diagnostic of PAVMs but does indicate the presence of a shunt. The 100 percent oxygen method and radionuclide perfusion scanning can be used as alternatives for the measurement of shunt when TTCE is not helpful, equivocal, or not available. Among these options, TTCE is the only test that can differentiate shunt at the level of the lung from heart and great vessels, thereby making it more useful for the diagnostic evaluation of pulmonary (and cardiac) AVMs.

Transthoracic contrast echocardiography — TTCE (also known as "bubble echocardiography") is the preferred initial test for patients who have an indication for diagnostic evaluation [56,57,59-65]. For most patients with suspected PAVMs in whom diagnostic testing is indicated, based upon its high sensitivity and low risk, TTCE is the initial test of choice to evaluate for the presence of a PAVM-associated right-to-left shunt. Subsequent diagnostic testing with nonenhanced CT of the chest depends upon the identification and severity (ie, grade) of shunt found on TTCE (algorithm 1). (See 'Indications for diagnostic evaluation' above and "Contrast echocardiography: Contrast agents, safety, and imaging technique" and "Contrast echocardiography: Clinical applications".)

TTCE is highly sensitive for the diagnosis of shunt from PAVMs [56,57,60,62-64]. TTCE identifies PAVM-associated shunt with a sensitivity, specificity, positive predictive value, and negative predictive value of 100, 49, 32, and 100 percent, respectively [63]. One cited potential limitation of TTCE is that it may be too sensitive, detecting clinically unimportant PAVMs [66-68]. This was illustrated by several studies that found that TTCE continued to detect intrapulmonary shunting in 80 percent of patients even after successful embolization of all visible PAVMs [68].

TTCE involves the injection of echocardiographic contrast, usually 10 to 20 mL of agitated saline (generates microbubbles), into a peripheral vein while simultaneously imaging the right and left ventricles with two-dimensional echocardiography. The contrast is normally visualized in the right ventricle soon after injection and should not be visualized in the left cardiac chambers at all (microbubbles are filtered out by the pulmonary capillary network). The appearance of contrast in the left ventricle (microbubbles) is consistent with right-to-left shunt. The Valsalva maneuver is not necessary and is not recommended when screening for PAVM as shunting from PAVMs is continuous and the Valsalva maneuver may artificially open a patent foramen ovale or other shunt, yielding a false positive result.

PAVMs are not directly visualized on echocardiography but shunts of cardiac or great vessel origin may be directly visualized which may help to distinguish PAVMs from other anatomic shunts. Visualizing an intracardiac shunt by TTCE should not preclude further evaluation of possible PAVMs if the suspicion for PAVMs is high. For example, in the presence of an atrial right-to-left shunt, contrast will appear in the left atrium immediately and therefore obscure any delayed shunt from a PAVM. In this case, TTCE or especially transesophageal echocardiogram (TEE) may be used to diagnose PAVM by looking for contrast in the pulmonary veins [69].

In the case of right–to-left shunt, contrast first appears in the right ventricle and then the left ventricle. The number of cardiac cycles that elapse between first appearance of contrast in the right ventricle and left ventricle can help to anatomically locate the shunt and, if visualized, microbubbles are counted so that the degree of shunting can be graded:

●Timing – When microbubbles are seen in the left ventricle in less than one cardiac cycle of their appearance in the right ventricle, this is typically associated with an intracardiac shunt; in greater than three but less than eight cycles is consistent with an intrapulmonary shunt; and between one to three cardiac cycles the location of the shunt is indeterminate (ie, the shunt could be located within the heart, great vessels, or lung) [60,61,65,70,71]. However, the location of shunt using this method has not been validated and is considered by many experts as a general guideline only because timing of contrast depends upon many factors including cardiac output, shunt size, and heart rate. For example, the shunt from large PAVM may occasionally be seen within one cardiac cycle.

●Grading – Many experts grade the degree of shunting seen on TTCE in order to select those suitable for diagnostic CT scanning [57,62,63,72-74]. Although there is little evidence to support the use of one grading system over another, we and some other experts prefer to use a quantitative grading system due to a preponderance of evidence to support its use [57,65]. Subsequent diagnostic testing with nonenhanced CT of the chest varies depending upon the grade:

•Grade 0 – Grade 0 refers to no bubbles reaching the left ventricle (ie, no right-to-left shunt). In general, patients whose TTCE finds no shunt require no further evaluation for PAVM unless there is a high clinical suspicion for PAVM, such as unexplained hypoxemia, suspicious nodule on imaging, or evidence to suggest paradoxical embolization (eg, stroke). In such patients, additional testing may include measurement of the shunt fraction by the 100 percent oxygen method, radionuclide perfusion scanning, or chest CT; chest CT is our preference. The preference for CT is based upon the higher diagnostic sensitivity compared with 100 percent oxygen and radionuclide perfusion scanning. For patients with no shunt and low clinical suspicion for PAVM (eg, a nodule with some features suggestive of cancer), additional investigations for the presenting complaint are indicated. In patients with hereditary hemorrhagic telangiectasia (HHT) and a grade 0 shunt, some experts will repeat TTCE in 5 to 10 years to follow for development of new PAVM, though this is clinical practice that is unproven. One study performed repeat TTCE five years later in 86 patients with HHT and grade 0 shunt and found that 26 percent developed a grade 1 shunt, though no one developed treatable PAVM or increased to grade 2 shunt during that time [75]. In another study of 172 HHT patients with initially negative TTCE or CT scan, 5.2 percent developed new PAVM over a median of seven years, including one with a feeding artery diameter (FAD) of 3 mm [76].

•Grade 1 – Grade 1 refers to 1 to 29 bubbles seen in the left ventricle on the still frame with the largest number of bubbles. Although historically many of these patients were subjected to further diagnostic testing, in practice most patients with a grade 1 shunt do not require further investigation for PAVM. This is because there appears to be no difference in the risk of paradoxical embolization or bleeding between the two groups (1.8 percent risk with grade 0 shunt, 0.4 percent risk with grade 1 shunt) and while a PAVM is occasionally discovered with CT following a grade 1 TTCE, such PAVMs are almost always too small to require embolization [33,74].

We generally advise the following:

-In most patients with grade 1 shunt, no further evaluation is performed and patients should undergo yearly observation clinically for the development of progressive symptoms which may prompt a repeat TTCE. Many experts consider it prudent in this population that a repeat TTCE be performed within five years so that small PAVMs that may have grown during that interval will be detected. Shorter intervals (weeks or months) are not required due to the slow growth of PAVMs. One study performed repeat TTCE five years later in 55 patients with HHT and grade 1 shunt and found that 17 percent developed a grade 2 shunt, and some of these developed treatable PAVM [75].

-In patients with a grade 1 shunt where the clinician has a high suspicion for clinically significant PAVM including unexplained hypoxemia, suspicious nodule on imaging, or evidence to suggest paradoxical embolization (eg, stroke), we prefer diagnostic testing with chest CT to confidently diagnose or exclude treatable PAVM. (See 'Computed tomography' below.)

•Grade 2 or 3 – Grade 2 shunt (30 to 100 bubbles in the left ventricle) or grade 3 shunt (>100 bubbles in the left ventricle) should undergo chest CT scanning next. (See 'Computed tomography' below.)

This approach is based upon our clinical experience and observational studies that report low likelihood of a treatable PAVM (<2 percent) in patients with grade 1 shunt [33,63] and that higher grade shunt is more likely to be associated with the detection of symptomatic PAVMs that require therapy [33,63,72,74]. As examples:

•TTCE and thoracic CT were performed in 95 patients with HHT using a 4 point grading system (as opposed to the 3 point system described above). No patient with grade 0 or grade 1 shunt on TTCE was found to have PAVMs by thoracic CT [73]. In contrast, 25, 80, and 100 percent of patients with grade 2, 3, and 4 shunts, respectively, were found to have PAVM by CT.

•When the results from four studies that used a grading scale were pooled, none of the 178 patients with a grade 1 shunt had treatable PAVMs on CT scan and only 6 percent had any visible PAVMs on CT [57,62,63,73].

•In another study of 1038 patients with HHT, pulmonary shunt grade 2 and 3 were both independent predictors for the prevalence of a cerebral ischemic event or brain abscess (odds ratio [OR] 4.78, 95% CI 1.14-20 and OR 10.4, 95% CI 2.4-45, respectively) [33]. In addition, patients with a grade 1 shunt were no more likely to suffer cerebral complications than those with a grade 0 shunt (OR 0.44) [33].

The optimal test for patients with an indeterminate TTCE (eg, appearance of contrast in the left atrium within 1 to 3 cardiac cycles or poor visualization of contrast or chambers due to poor technique) is unknown. We believe that the presence of shunt on TTCE in HHT patients should be assumed to be due to PAVM rather than an intracardiac shunt, even if seen before three cycles. If workup for PAVM is negative, then intracardiac shunts should be looked for with transesophageal echocardiography. In the case of poor visualization, we suggest that the clinical suspicion direct the course of investigation: observation may be appropriate in those with a low suspicion for PAVM; a repeat TTCE for those with poor technical views; or a CT chest for those with a high suspicion for PAVM.

The techniques, safety, and clinical applications of contrast echocardiography are discussed separately. (See "Contrast echocardiography: Contrast agents, safety, and imaging technique" and "Contrast echocardiography: Clinical applications".)

Other — Several additional methods exist to measure shunt. These include measuring arterial oxygen and carbon dioxide tension on 100 percent oxygen, oxygen saturation on room air, alveolar-arterial gradient, and radionuclide perfusion scanning. In general, these methods can be used to confirm the suspicion of shunt when TTCE is not available or is equivocal, but they are not as well validated as TTCE.

100 percent oxygen — The shunt fraction is the fraction of the cardiac output that bypasses the pulmonary capillaries. It can be determined by the 100 percent oxygen method, which involves measuring the arterial oxygen tension (PaO₂) and saturation (SaO₂) and the arterial carbon dioxide tension (PaCO₂) after breathing 100 percent oxygen through a mouthpiece or airtight mask for 15 to 20 minutes and then using those values to calculate the shunt fraction [3]. A shunt fraction of >5 percent is considered abnormal and warrants additional evaluation. (See "Measures of oxygenation and mechanisms of hypoxemia".)

The sensitivity of the shunt fraction at detecting PAVMs depends upon whether the PAVMs are clinically significant or asymptomatic. A shunt fraction >5 percent appears to detect approximately 98 percent of clinically significant PAVMs [3,13,16,67,77], but only 88 percent of asymptomatic PAVMs [55]. The magnitude of the shunt fraction varies among patients with PAVMs, with values ranging from 14 to 55 percent [3].

Oxygen saturation on room air — When the shunt fraction cannot be measured using the 100 percent oxygen method, an alternative approach is to measure the PaO₂ and arterial oxyhemoglobin saturation (SaO₂) while the patient is breathing room air. A PaO₂ >95 mmHg and an SaO₂ >96.5 percent effectively excludes a significant shunt, while a PaO₂ <85 mmHg or an SaO₂ <96 percent suggests a shunt fraction >5 percent [3,77]. However, this method is less specific than the 100 percent oxygen method because it does not differentiate a true shunt (either anatomic or physiologic) from ventilation-perfusion mismatch. The use of pulse oximetry alone is an insensitive screening method for PAVM and should not be used for this purpose, except perhaps in children where arterial puncture is generally avoided [78]. In the case of screening children for PAVM, we measure pulse oximetry in the supine position and then every 15 seconds for 2 minutes after assuming the seated position; a saturation of <96 percent in either position is considered abnormal.

Alveolar-arterial oxygen tension gradient — The alveolar-arterial PaO₂ gradient is not routinely used to measure shunt. However, some clinicians have used it instead of the shunt fraction to identify PAVMs. One study of 105 patients with hereditary HHT who were being evaluated for treatable PAVMs found that an alveolar-arterial gradient >24.5 kPa (184 torr) measured while breathing 100 percent oxygen in the upright position detected PAVMs with a sensitivity of only 59 percent [56]. This threshold was chosen because it corresponded with a shunt fraction >5 percent in the investigators’ laboratory. However, the threshold corresponds to a shunt fraction of >10 percent if conventional equations are used to determine the shunt fraction, which may explain why the sensitivity was so poor (ie, higher gradients are likely to cause a lower sensitivity and a higher specificity).

Radionuclide perfusion scanning — Radionuclide perfusion lung scanning does not typically anatomically define a PAVM but can confirm or identify the presence of a right-to-left shunt. This test is useful if contrast-enhanced echocardiography or the 100 percent oxygen method for the evaluation of shunt is not available. The test involves the peripheral, intravenous injection of macroaggregated albumin labeled with technetium-99m (^99mTc). In healthy persons, these particles are filtered by pulmonary capillaries. However, in the presence of true right-to-left shunts, radiolabeled particles pass through the lungs (or intracardiac defects) and are trapped in the brain and kidneys. The shunt fraction may be calculated by quantifying the renal uptake as a percentage of the total dose given.

One study reported in 19 PAVM patients that the mean shunt fraction was 23 percent compared with 3 percent in normal subjects [79]. A second study of 66 patients with PAVM showed that a shunt of >3.5 percent by the radionuclide method was 87 percent sensitive and 61 percent specific for detection of residual PAVM after embolotherapy [80].

The radionuclide method of shunt detection and calculation has several potential advantages over the 100 percent oxygen method:

●Arterial blood sampling is not needed

●The 100 percent oxygen method may overestimate the shunt fraction if it is not performed correctly

●The radionuclide method allows shunt measurement during exercise

●Radionuclide scanning can sometimes show the location of a PAVM [3,79,81]

However, radionuclide perfusion scanning can be expensive and is not routinely available at many hospitals. In addition, an abnormal radionuclide perfusion scan does not typically differentiate between intrapulmonary and intracardiac shunts.

Computed tomography — CT is the test of choice for the diagnosis of PAVMs, particularly in patients with the following (algorithm 1):

●Evidence of grade 2 to 3 shunt by TTCE (or shunt >5 percent by other modalities)

●A high suspicion for PAVMs together with a low grade (grade 1) shunt, equivocal evidence for shunt, or indeterminate shunt

Further testing is dependent upon the results of the CT scan:

●If the CT scan shows one or more PAVM with a FAD ≥2 to 3 mm diameter, the patient should be referred for pulmonary angiography and potential embolotherapy (image 8). There is some controversy about the exact cutoff to use. In a survey study of HHT/PAVM experts, 80 percent used 2 mm as the cutoff, though some went even lower [82]. (See 'Pulmonary angiography' below and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations" and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Embolotherapy'.)

●If the CT scan shows PAVMs with a FAD <2 mm, in most patients pulmonary angiography may be deferred unless patients have clinical features suggestive of symptomatic PAVM (eg, prior stroke with no other identifiable cause, hemoptysis, or significant hypoxemia). Notably, the presence of a cerebral abscess does not imply that a PAVM is treatable because an abscess can occur in a PAVM with a FAD of any size.

For those in whom angiography is not performed, patients should undergo yearly observation clinically for the development of progressive symptoms, which may prompt a repeat TTCE and CT. Repeat CT can be performed after an appropriate interval to detect PAVMs that may grow to a size that is amenable to embolotherapy. Although this interval was historically recommended to be three to five years, longer intervals may be considered due to the slow growth of small PAVMs. A study of 88 patients with untreated PAVMs with a FAD <2 mm who underwent long-term CT monitoring showed no statistically significant change in dimension over mean duration of 8.4 years [83]. In another study, PAVM enlargement was seen in 2 of 18 patients with small PAVMs but only after long follow-up times (7.9 and 10.7 years) and only one patient required treatment [84].

●If the CT scan is negative for PAVM and shunt is present on TTCE, microscopic PAVMs may be responsible for the shunt and follow-up as described above is prudent (ie, yearly clinical evaluation and at appropriate intervals if symptomatic). The established recommendation is to repeat assessment at three- to five-year intervals, but development of clinically significant PAVMs in this scenario is rare. One retrospective study of patients with microscopic PAVMs (positive TTCE but negative CT) found that only 2 of 48 patients developed visible PAVMs on CT after 6.8 years follow up, which were asymptomatic and too small to require embolization [84]. The optimal repeat imaging technique is unknown at this time [82]. At our center, if the degree of shunt on TTCE is grade 1, we usually revert back to contrast echocardiography and only repeat the CT scan if the severity of shunt worsens to grade 2 or 3. This approach is based on the low incidence of complications in patients with grade 1 shunt [33]. However, some centers continue to follow CT scans [82]. Pulmonary angiography is rarely performed in this population for patients with, for example, idiopathic stroke; however, in our experience, the likelihood of demonstrating a significant PAVM suitable for embolotherapy in this population is low.

Alternatively, shunt may be due to hepatopulmonary syndrome (HPS) such that evaluation for HPS may be indicated. (See "Hepatopulmonary syndrome in adults: Prevalence, causes, clinical manifestations, and diagnosis".)

●If the CT scan is indeterminate (eg, atypical features suggestive but not diagnostic of PAVM (image 9) (see 'Imaging findings' above)), further evaluation should be individualized. Some experts occasionally perform contrast pulmonary angiography in patients with high grade shunt and indeterminate CT scans in whom the clinical suspicion is high. However, observation with follow-up TTCE and/or CT is also appropriate (eg, those with minimal symptoms and/or a low grade shunt).

We prefer multidetector, thin cut (ie, 1 to 2 mm collimation) CT, which readily detects PAVMs; three-dimensional reconstruction facilitates delineation of the vascular structures (image 6) [85]. Contrast material is not required, except for patients with atypical nodules on nonenhanced CT in whom the suspicion remains high (eg, large shunt, clinical symptoms of paradoxical embolization such as stroke). Diagnostic imaging findings on CT are discussed above. (See 'Imaging findings' above.)

This approach is based upon the high sensitivity of non-contrast-enhanced CT for the diagnosis of PAVMs. As an example, one study of 33 consecutive patients with 37 PAVMs compared non-contrast three-dimensional multidetector CT scanning with contrast-enhanced pulmonary angiography, which was the original gold standard at the time of the study [49]. CT provided reliable analysis of more PAVMs than unilateral pulmonary angiography (76 versus 32 percent), but fewer PAVMs than hyper-selective pulmonary angiography (76 versus 100 percent; ie, imaging the suspected PAVM only). (See 'Pulmonary angiography' below.)

CT is more sensitive than conventional chest radiography for detecting PAVMs. This is because on chest radiography, small PAVMs can be easily missed and discrete PAVMs are occasionally obscured by the diaphragms, heart shadow, or concomitant parenchymal lung disease. CT identifies nearly all PAVMs, even those that are poorly appreciated on conventional chest radiographs such that CT is the preferred imaging modality for diagnosis [48,49]. Thus, while a completely normal chest radiograph makes the diagnosis of symptomatic PAVM less likely, it should not preclude further evaluation with CT or angiography, particularly in a symptomatic patient with other features consistent with PAVM or HHT. As an example, in a symptomatic patient, CT of the chest may demonstrate multiple small PAVMs that are undetectable or present as a vague increase in pulmonary vascular markings on chest radiography [14].

Contrast-enhanced ultrafast (ie, electron beam) CT uses an electron beam rather than a conventional x-ray beam to obtain images. The electron beam obtains images more quickly and, therefore, motion artifact is minimized. In a study of 40 patients with suspected PAVM, contrast-enhanced ultrafast CT was more sensitive than selective pulmonary angiography at detecting PAVM (98 versus 60 percent), but selective pulmonary angiography was better at determining the angioarchitecture of individual PAVMs [48]. The disadvantage of contrast-enhanced electron beam CT compared with multidetector CT is that contrast is required, although they have not been compared with one another. (See "Principles of computed tomography of the chest".)

The major advantages of CT scanning over pulmonary angiography are that it is noninvasive, is readily accessible, and does not require contrast material. Its disadvantages include the need for breath-holding and occasional mischaracterization of vascular tumors as PAVMs.

Low-dose radiation protocols are used by many centers with expertise for the evaluation of suspected PAVM.

Pulmonary angiography — Catheter pulmonary angiography is the gold standard test for defining the vascular anatomy of PAVMs that are identified on prior CT as suitable for embolotherapy (ie, those with a feeding artery ≥2 to 3 mm in diameter) (image 11 and image 13 and image 14). It will additionally detect the presence of unsuspected PAVMs in an individual patient, some of which may also be suitable for embolotherapy. Thus, it is generally performed with the primary purpose of diagnostic confirmation and therapeutic management (ie, embolotherapy) of PAVMs (see "Therapeutic approach to adult patients with pulmonary arteriovenous malformations"). Although it is not routinely performed in patients for purely diagnostic reasons (eg, small asymptomatic lesions found on CT), pulmonary angiography can be performed in those with symptomatic lesions where uncertainty persists after other tests (eg, atypical lesion on CT with high grade shunt of unidentified etiology) (algorithm 1). Diagnostic imaging findings on pulmonary angiography are discussed above. (See 'Imaging findings' above.)

During angiographic testing, contrast should be directly injected into the feeding artery or a distal pulmonary artery (ie, hyper-selective angiography) to accurately define the angioarchitecture of individual lesions (image 12) [48,49,86,87]. Contrast should also be injected into the right and left main pulmonary arteries (ie, routine/unselective angiography) to detect additional PAVMs that may also be amenable to embolotherapy and to identify intra- or extra-thoracic vascular communications.

Other diagnostic tests

Magnetic resonance imaging — Magnetic resonance (MR) angiography is rarely used for the diagnostic evaluation of patients with suspected PAVMs in the United States, although some HHT centers in Europe with expertise in MR angiography include it as part of their evaluation. Although MR angiography may show similar findings to those seen on CT, it is not as sensitive or specific diagnostically, and is more expensive; additionally, MR requires specialized expertise in the acquisition and interpretation of the images. However, MR angiography has the advantage of avoiding ionizing radiation. Diagnostic imaging findings on MR are discussed above. (See 'Imaging findings' above.)

Conventional spin-echo MR imaging has shown low sensitivity and specificity in the evaluation of PAVMs due to poor sensitivity to blood flow [3]. Despite the use of techniques to improve MR sensitivity to blood flow, including gradient-refocused echo MR imaging, phase contrast cine sequences, and contrast-enhanced MR angiography with multiplanar reconstruction, the sensitivity and specificity have generally remained lower than CT and pulmonary angiography [88-92]. However, a blinded study comparing contrast-enhanced MR imaging showed excellent sensitivity of 92 percent in the detection of PAVM with FAD of ≥2 mm when compared with CT; specificity depended on the experience of the reader (97 percent for an experienced reader versus 62 percent for a less experienced reader) [93]. (See "Magnetic resonance imaging of the thorax".)

Routine laboratory tests — Complete blood count and chemistries should be performed to look for anemia and polycythemia as well as for the detection of underlying chronic liver and chronic renal disease as potential etiologies for the presenting complaint. If not already performed, oxygen saturation (or occasionally arterial blood gas analysis) in the recumbent and standing positions should also be obtained to uncover underlying hypoxemia and orthodeoxia, respectively. (See 'Platypnea and orthodeoxia' above.)

Pulmonary function test — Pulmonary function tests are often done for the investigation of the presenting complaint but are not specific or sensitive for the diagnosis of PAVMs. Standard spirometric indices are usually within normal limits in patients with PAVM. However, the diffusing capacity may be mildly elevated and resting minute ventilation is usually increased [27]. (See "Overview of pulmonary function testing in adults".)

Pulmonary artery catheterization — Pulmonary artery catheterization (PAC) may be performed in patients with PAVMs in whom pulmonary hypertension is suspected clinically (eg, loud pulmonic second heart sound, elevated jugular venous pressure, or leg edema). Details regarding the incidence of pulmonary hypertension and the value of PAC in predicting post embolization pulmonary artery pressures are discussed separately. (See 'Features attributable to PAVM complications' above and "Clinical features and diagnosis of pulmonary hypertension of unclear etiology in adults", section on 'Initial diagnostic evaluation (noninvasive testing)' and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Procedural complications'.)

Evaluation for HHT — In patients with suspected PAVMs without a known diagnosis of hereditary hemorrhagic telangiectasia (HHT), it is prudent to perform a thorough history and examination to look for clinical evidence of HHT (eg, personal or family history of epistaxis or arteriovenous malformations, and presence of mucocutaneous telangiectases). However, a lack of findings for HHT does not rule out the possibility for HHT [30]. The clinical manifestations and indications for genetic testing in patients with suspected HHT and their relatives are discussed separately. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)".)

DIAGNOSIS — PAVMs are typically diagnosed by chest CT with the presence of shunt supported by contrast-enhanced pulmonary angiography. Typical findings include a well demarcated nodule with a well-defined feeding pulmonary artery and draining pulmonary vein. These and other classic radiographic features are discussed above. (See 'Imaging findings' above.)

A PAVM should always be suspected in a patient with a nodule and a history of cerebral infarct or abscess as well as in those with unexplained dyspnea, hypoxemia, hemoptysis, or hemothorax.

DIFFERENTIAL DIAGNOSIS — The differential diagnosis of PAVMs is wide and depends upon the presenting complaint. Although PAVMs are unusual causes of pulmonary nodules, dyspnea, hemoptysis, stroke, and cerebral abscess, they should be suspected when more common etiologies for these symptoms and signs are absent.

●For those who present with dyspnea, distinguishing features include platypnea and orthodeoxia, although these can also be found in hepatopulmonary syndrome, atrial septal defects (including patent foramen ovale), and other causes of shunt. Dyspnea from underlying cardiopulmonary disease should be evident on chest imaging and pulmonary function testing. (See "Approach to the patient with dyspnea".)

●For those who present with hemoptysis, the cause may be related to nasal telangiectases, endobronchial telangiectases, or PAVM in the setting of hereditary hemorrhagic telangiectasia. Importantly, if PAVM is suspected, these lesions should not be randomly biopsied as significant bleeding can occur. (See "Etiology of hemoptysis in adults" and "Evaluation of nonlife-threatening hemoptysis in adults".)

●Patients with platypnea and orthodeoxia from hepatopulmonary syndrome should have evidence of underlying chronic liver disease, and if from atrial septal defect should have echocardiographic evidence of early right-to-left shunt. (See "Hepatopulmonary syndrome in adults: Prevalence, causes, clinical manifestations, and diagnosis" and "Clinical manifestations and diagnosis of atrial septal defects in adults".)

●CT imaging may reveal feeding arteries and veins that distinguish a PAVM from a primary or metastatic lesion of the lung and/or brain. (See "Diagnostic evaluation of the incidental pulmonary nodule".)

●MRI of the brain will distinguish a cerebral infarct from an abscess, and cerebral AVM. (See "Overview of the evaluation of stroke", section on 'Confirming the diagnosis' and "Pathogenesis, clinical manifestations, and diagnosis of brain abscess".)

●The differential diagnosis of anemia and polycythemia are discussed separately. (See "Diagnostic approach to anemia in adults" and "Diagnostic approach to the patient with erythrocytosis/polycythemia" and "Clinical manifestations and diagnosis of polycythemia vera".)

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Hemoptysis" and "Society guideline links: Hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)".)

SUMMARY AND RECOMMENDATIONS

●Definition – Pulmonary arteriovenous malformations (PAVMs) are abnormal communications between pulmonary arteries and veins. They are uncommon, but they are an important consideration in the differential diagnosis of common pulmonary problems, including hypoxemia, pulmonary nodules, and hemoptysis. (See 'Introduction' above and "Pulmonary arteriovenous malformations: Epidemiology, etiology, and pathology in adults".)

●Clinical manifestations – Approximately 40 percent of PAVMs are symptomatic and the remainder are found incidentally on chest imaging or during screening of patients with known or suspected hereditary hemorrhagic telangiectasia (HHT).

•The most common presenting features are those attributable to the PAVM itself, usually dyspnea and hemoptysis (table 1), as well as those attributable to underlying HHT, including epistaxis and mucocutaneous telangiectases (table 2). In addition, PAVMs should always be suspected in patients who present with nodules and a history of a stroke or brain abscess, which are common complications of PAVMs. (See 'Clinical manifestations' above.)

•Conventional chest radiography (image 3) and/or nonenhanced CT of the chest (image 6) are abnormal in most patients who have symptoms due to PAVMs. While the typical appearances of PAVMs on chest radiography and CT are similar (ie, a nodule with a feeding artery and draining vein), in the era of CT evaluation of HHT patients and frequent use of CT in the general population, many patients present with atypical findings on chest CT (image 9). Pulmonary angiography is not routinely used for the diagnostic evaluation of suspected PAVMs unless CT-identified lesions are potentially suitable for embolotherapy (image 11 and image 13 and image 14). (See 'Imaging findings' above.)

●Diagnostic evaluation – PAVMs should be suspected and a diagnostic evaluation initiated in individuals with one or more clinical features suggestive of PAVMs (eg, typical nodules on chest imaging), stigmata of right-to-left shunting (dyspnea, hypoxemia, cyanosis, cerebral embolism, brain abscess), unexplained hemoptysis, or hemothorax, as well as in patients with platypnea or orthodeoxia, and those with suspected, known, or a family history of HHT. (See 'Indications for diagnostic evaluation' above.)

•Shunt assessment – For most patients with suspected PAVMs in whom diagnostic testing is indicated, we suggest transthoracic contrast echocardiography (TTCE) as the initial test of choice to evaluate for the presence of a right-to-left shunt. Subsequent diagnostic testing with nonenhanced CT of the chest depends upon the identification and severity (ie, grade) of shunt found on TTCE (algorithm 1) (see 'Diagnostic approach' above and 'Transthoracic contrast echocardiography' above):

-For patients whose TTCE finds no right-to-left shunt (grade 0), we suggest no further evaluation for PAVMs, unless there are confounding factors that raise the clinical suspicion for PAVM such as unexplained hypoxemia, suspicious nodule on imaging, or evidence to suggest paradoxical embolization (eg, stroke).

-For most patients whose TTCE shows a grade 1 shunt, CT is not required due to the low risk of stroke and low likelihood of a treatable PAVM; yearly clinical observation and a repeat TTCE within five years is appropriate in this population so that small PAVMs that become symptomatic or have grown during the interval period will be detected. However, in patients with grade 1 shunt and a high clinical suspicion for treatable PAVM (eg, prior stroke), we prefer diagnostic testing with nonenhanced CT of the chest. (See 'Computed tomography' above.)

-For patients whose TTCE shows a grade 2 or 3 shunt, we suggest nonenhanced CT of the chest. (See 'Computed tomography' above.)

•Nonenhanced CT of the chest – For patients with suspected PAVMs in whom CT scanning is indicated, we perform non-contrast-enhanced, multidetector, thin cut (ie, 1 to 2 mm collimation) CT of the chest; further testing is dependent upon the results of the scan (algorithm 1) (see 'Computed tomography' above):

-If the CT scan shows one or more PAVMs with a feeding artery diameter (FAD) of ≥2 to 3 mm diameter, the patient should be referred for pulmonary angiography and potential embolotherapy. (See 'Pulmonary angiography' above and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Embolotherapy'.)

-If the CT scan shows PAVMs with a FAD <2 mm, in most patients pulmonary angiography can be deferred and patients should undergo yearly clinical evaluation and CT in five years, though longer intervals can be considered. Pulmonary angiography may be performed in patients with symptoms suggestive of a treatable PAVM (eg, stroke).

-If the CT scan is negative and shunt is present on TTCE, microscopic PAVMs may be responsible and yearly clinical evaluation and CT in five years is prudent, though longer intervals can be considered. While the yield is low, pulmonary angiography is rarely performed in patients with symptoms suggestive of a possible treatable PAVM (eg, stroke).

-If the CT scan is indeterminate (eg, atypical features), further evaluation should be individualized. Some experts occasionally perform contrast pulmonary angiography in patients with high grade shunt and indeterminate CT scans in whom the clinical suspicion is high (eg, stroke). However, observation with follow-up TTCE and/or CT is also appropriate (eg, those with minimal symptoms and/or a low grade shunt).

•Pulmonary angiography – For patients with suspected PAVMs on CT, we perform contrast-enhanced pulmonary angiography to define the vascular anatomy of PAVMs that are identified on prior CT as potentially suitable for embolotherapy (ie, those with a feeding artery ≥2 to 3 mm in diameter). Although it is not routinely performed in patients for purely diagnostic reasons (eg, small asymptomatic lesions found on CT), pulmonary angiography can be performed in those with symptomatic lesions where uncertainty persists after other tests (eg, atypical lesions on CT with high-grade shunt of unidentified etiology). (See 'Pulmonary angiography' above and "Therapeutic approach to adult patients with pulmonary arteriovenous malformations", section on 'Embolotherapy'.)

●Diagnosis and differential – PAVMs are typically diagnosed radiographically either by computed chest radiography or by contrast-enhanced pulmonary angiography. Typical findings include a well-demarcated nodule with a well-defined feeding pulmonary artery and draining pulmonary vein. The differential diagnosis of PAVMs is wide and depends upon the presenting complaint. Although PAVMs are unusual causes of pulmonary nodules, dyspnea, hemoptysis, stroke, and cerebral abscess, they should be suspected when more common etiologies for these symptoms and signs are absent. (See 'Diagnosis' above and 'Differential diagnosis' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges James R Gossage, MD, who contributed to earlier versions of this topic review.