**7. Conclusions**

PAVMs are important to detect and challenging to treat. As the vast majority of PAVMs occur in the setting of HHT, particular attention should be given to screening and surveillance of PAVMs in this patient population. Detailed review of the current literature suggests that contemporary practices often deviate from previously published guidelines on PAVM management.

Grade 1 shunts on initial screening TTCE are not associated with cerebral complications and do not predict presence of treatable PAVMs, and therefore chest CT can be withheld. The same is true of shunts which are seen only with Valsalva maneuver. In these patients, TTCE screening can be repeated every 5 years.

Chest CT should be performed in patients with grade 2 shunts or higher on screening TTCE. CT can be performed su fficiently with or without contrast.

Serious complications including stroke and brain abscess often occur in PAVMs with feeding artery diameter ≥3 mm. These complications can also occur with smaller feeding artery diameters, though it is less common. It is recommended to treat any PAVM with feeding artery 2 mm or greater, and any symptomatic PAVM. Furthermore, embolization of all angiographically visible PAVMs at the time of initial procedure appears to significantly reduce the likelihood of reintervention and risk of ischemic stroke.

Both standard and conservative approaches are acceptable screening methods for PAVM in pediatric patients. Standard screening with TTCE, followed by CT for grade 2 or higher shunts, is safe and e ffective at detecting PAVM, but risks exposing a larger percentage of pediatric patients to radiation. Conservative screening with physical exam, pulse oximetry, and chest radiograph every 5 years is also safe, e ffective, and may reduce radiation exposure. However, with the conservative algorithm, there is a slightly higher risk of occasionally missing a treatable AVM which is not detected until adulthood.

PAVMs are expected to grow during childhood, potentially increasing associated risks and symptoms. Pediatric patients should be rescreened for PAVMs at 5-year intervals.

We recommend treatment of symptomatic AVMs and asymptomatic PAVMs ≥ 3 mm in pediatric patients, and monitoring of smaller asymptomatic PAVMs until age 18 to reduce childhood radiation exposure.

For untreated PAVM in pregnancy, both the morbidity and mortality exceed 1%, while the risk of PAVM embolization is much lower. Thus, pregnan<sup>t</sup> patients with untreated PAVMs should be treated, prior to pregnancy if possible. If PAVMs are discovered during pregnancy, they should be treated. Symptomatic patients should be treated as needed, regardless of gestational age. Otherwise, early in the second trimester may be the ideal time to treat pregnan<sup>t</sup> mothers with PAVM, as this stage of fetal development has the least susceptibility to radiation and is generally a stable period with low risk of preterm labor. Unscreened pregnan<sup>t</sup> patients with HHT should undergo screening.

Persistent PAVMs are di fficult to treat, with high rates of reperfusion or recanalization following repeat embolization. Embolization distal to the existing embolic results in a better rate of durable occlusion, especially when treating the recanalization pattern of persistence.

Patients with di ffuse PAVMs are very high risk for adverse events. In patients with di ffuse PAVM affecting only one segmen<sup>t</sup> or a few segments, the optimal treatment strategy combines embolization of focal PAVMs ≥ 3 mm and peripheral-to-central occlusion of the most severely a ffected segment(s) to achieve peripheral blood flow redistribution and improve hypoxemia. In patients with truly di ffuse bilateral PAVM a ffecting all lung segments, attempts at blood flow redistribution are unlikely to achieve meaningful improvements in hypoxemia, though embolization of focal PAVMs ≥ 3 mm should be pursued to reduce risk of paradoxical embolization. Even after treatment, patients with di ffuse PAVM remain high risk for complications.

Patients with coexisting PH and PAVM should be considered for therapy on a case-by-case basis. Treatment is typically indicated in patients with mild-to-moderate PH. Severe baseline PH should prompt careful consideration of risks versus benefits, due to the possibility of worsening PH after closure.

For patients with initial negative screening CT or small untreatable PAVM on CT, there is conflicting evidence regarding what constitutes an appropriate screening interval. The most recent studies demonstrate slow growth of small untreated PAVMs, challenging the recommendation to obtain follow-up CTs every 3–5 years, and suggesting that a longer screening interval of 5–10 years may be warranted for these patients.

For follow-up of treated PAVMs, we recommend initial follow-up with CT within 6 months of embolization. This should be followed by a repeat CT every 3–5 years based on perceived likelihood of persistence, favoring the 3 year follow-up interval for more complex PAVMs, and 5 years for all others.

Emerging data suggests that MR may have similar diagnostic accuracy as CT for follow-up of treated PAVMs, in particular for PAVMs with feeding artery diameter >2 mm. Preliminary data also suggests that serial graded TTCE has potential to be used as a post-embolotherapy follow-up tool, with high predictive accuracy for presence of treatable PAVMs on CT. These methods provide avenues to avoid repeated radiation exposure and nephrotoxic contrast.

**Author Contributions:** Conceptualization, S.M. and J.P.M.; writing—original draft preparation, S.M.; writing—review and editing, S.M. and J.P.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** S.M. declares no conflict of interest. J.P.M. has received a speaker honorarium from Penumbra, Inc.
