Cardiothoracic imaging findings of Proteus syndrome

In this work, we sought to delineate the prevalence of cardiothoracic imaging findings of Proteus syndrome in a large cohort at our institution. Of 53 individuals with a confirmed diagnosis of Proteus syndrome at our institution from 10/2001 to 10/2019, 38 individuals (men, n = 23; average age = 24 years) underwent cardiothoracic imaging (routine chest CT, CT pulmonary angiography and/or cardiac MRI). All studies were retrospectively and independently reviewed by two fellowship-trained cardiothoracic readers. Disagreements were resolved by consensus. Differences between variables were analyzed via parametric and nonparametric tests based on the normality of the distribution. The cardiothoracic findings of Proteus syndrome were diverse, but several were much more common and included: scoliosis from bony overgrowth (94%), pulmonary venous dilation (62%), band-like areas of lung scarring (56%), and hyperlucent lung parenchyma (50%). In addition, of 20 individuals who underwent cardiac MRI, 9/20 (45%) had intramyocardial fat, mostly involving the endocardial surface of the left ventricular septal wall. There was no statistically significant difference among the functional cardiac parameters between individuals with and without intramyocardial fat. Only one individual with intramyocardial fat had mildly decreased function (LVEF = 53%), while all others had normal ejection fraction.


Scientific Reports
| (2021) 11:6577 | https://doi.org/10.1038/s41598-021-86029-0 www.nature.com/scientificreports/ (NHGRI), Bethesda, MD, USA and was compliant with the Privacy Act (comparable to HIPAA). Individuals were included in this study if they met the following criteria: (1) had a clinical diagnosis of Proteus syndrome and (2) had cross-sectional imaging of the thorax (CT/CTA chest and/or cardiac MRI) available for review. Clinical diagnosis of Proteus syndrome was determined by both the presence of a mosaic AKT1 pathogenic variant as well as meeting clinical diagnostic criteria for Proteus syndrome detailed in Table 1. Individuals who lacked a mosaic AKT1 pathogenic variant or who only met clinical diagnostic criteria for AKT1-related overgrowth spectrum were excluded. Individuals who were clinically diagnosed with Proteus syndrome, but did not have cross-sectional imaging of the thorax were also excluded. Informed consent was obtained from all the participants or their parents/legal guardian for minors (under the age of 18). A subset of the current study population was previously reported by Hannoush et al. and Jamis-Dow et al. 17,18 . Participant imaging. Individuals referred to our institution for Proteus syndrome underwent baseline cardiothoracic imaging of the thorax to assess the degree of cardiothoracic involvement. Subsequent followup exams were also obtained if individuals became symptomatic (e.g. CTA for pulmonary embolism). CT and cardiac MRI scanner details are in Supplemental Table S1. The cardiac MRI protocol included cine images in the three long axes and a volumetric short axis cine stack for anatomic and functional analysis. Cardiac optimized, multi-echo Dixon fat-water separation images were also obtained at the same slices as the cine images as previously described 19,20 . This optimized fat-water sequence was used for easy detection of intramyocardial fat, even in thin structures such as the right ventricular (RV) free wall. Gadolinium was administered in only three of the cardiac MR studies and late gadolinium enhancement images were obtained in each of them.  11 . Proteus syndrome: a score of 10 or more points with a mosaic AKT1 variant or 15 or more points without an AKT1 mosaic variant. AKT1-related overgrowth spectrum: a score of 2-9 with an AKT1 mosaic variant.
Positive clinical criteria www.nature.com/scientificreports/ Image analysis. All chest CT studies were evaluated on a PACS workstation (CareStream, Version 12.1.6.1005, Carestream Health, NY) independently by two readers who were fellowship-trained in thoracic and body imaging and who were aware of the Proteus syndrome diagnosis but were blinded to all other clinical information. The readers (A.S., 10 years of experience and E.B.T., 13 years of experience) evaluated the lungs, heart, pleura, thoracic vasculature, chest wall and mediastinum for cardiothoracic imaging findings that were previously reported in the literature (Supplemental Table S2) 7,12,[14][15][16][17][18][21][22][23] . The entire thorax was assessed for the presence of nodules or masses suspicious for malignancy 8,17 and any other additional thoracic findings. Malignancy was determined via biopsy of suspicious lesions. Two fellowship-trained cardiac MRI readers (A.S., 10 years of experience and A.J.B., 3 years of experience) independently reviewed the cardiac MRI studies. Global and regional cardiac function were assessed using cardiac MR cine images. Functional cardiac MRI data (left ventricular ejection fraction (LVEF), left ventricular (LV) end-diastolic volume, LV end-diastolic volume index, LV end-systolic volume, LV end-systolic volume index, LV stroke volume, LV stroke volume index, cardiac output, cardiac index, anteroseptal wall thickness and posterolateral wall thickness) was analyzed using postprocessing software (Argus, Siemens Healthineers, Erlangen, Germany). The presence, location, and distribution of intramyocardial fat was evaluated on cardiac fat-water separation images using the 17 segment model of the American Heart Association 24 . Segmental findings were combined into the distribution of LV walls: anterior, septal, inferior and lateral. The presence of fat was described as epicardial (outer portion of the wall), endocardial (inner portion of the wall), midwall (middle of the wall), transmural (full thickness involvement of the wall), diffuse (entire wall involved), or focal (small, < 5 mm area of a wall). Consensus read was performed to resolve any disagreements between the readers for both chest CT and cardiac MRI studies.
Statistical analysis. Statistical analyses were performed using GraphPad Prism version 8.4.1 (676) for Windows (GraphPad Software, San Diego, California USA, http:// www. graph pad. com). The Shapiro-Wilk test was used to assess the normality of the data. Percentage was used as the descriptive index for the qualitative variables and mean ± standard deviation (SD) were used to describe the quantitative variables. Differences in cardiac functional parameters between subjects with and without intramyocardial fat were compared with the unpaired t-test and nonparametric tests, as appropriate. Fisher's exact test was used to compare the difference in the presence of fat location in LV and RV. A two-tailed P < 0.05 was considered to indicate a significant difference.

Results
Of 53 individuals referred to our institution, 53 were diagnosed with Proteus syndrome and no individuals were diagnosed with AKT1-related overgrowth spectrum. Of these, 15 of the 53 individuals with Proteus syndrome did not have any cross-sectional cardiothoracic imaging. Thus, the remaining 38 individuals with Proteus syndrome (mean age: 23; range: 9-61 years) who had cross-sectional cardiothoracic imaging available for review were included in the study. This group comprised 23 males (mean age: 24 years; range: 9-61 years) and 15 females (mean age: 23 years; range: 10-54 years).
Of the 38 individuals with cross-sectional cardiothoracic imaging, 18 underwent only chest CT imaging, 16 chest CT imaging and cardiac MRI, and four only cardiac MRI. Thus, of the 38 individuals with cardiothoracic imaging, 34 had chest CT imaging and 20 had cardiac MRI. Among the 34 individuals who underwent chest CT, ten had a routine chest CT without contrast, four had a routine chest CT with contrast and 20 had CT pulmonary angiography.
Cardiac involvement. Of the cohort of 38 patients who underwent imaging of the thorax, a subset of 20 individuals underwent dedicated cardiac MRI assessment for myocardial fat. Nine of the 20 individuals (45%) showed intramyocardial fat on the fat-water separation sequence ( Fig. 1). Within the LV, the intramyocardial fat was more commonly present in the septal wall (n = 9, 45%) compared to other walls of the LV (P = 0.0029). The RV was involved less often than the LV (n = 6 and n = 9, respectively). Within the RV, the RV free wall was the most common location for the intramyocardial fat (n = 6, 30%), however this was not significantly different from other RV locations (P > 0.05). Intramyocardial fat distribution patterns were variable and included endocardial, midwall, epicardial, transmural, diffuse, and focal ( Table 2). Some individuals had more than one pattern of intramyocardial fat within the same LV segment. In the LV, intramyocardial fat was most commonly located in an endocardial distribution. In the RV, intramyocardial fat most commonly was present diffusely within the RV free wall.
Chest CTs were obtained in 34 individuals and also evaluated for the presence of intramyocardial fat ( Fig. 1, Table 4). 16 of the 34 individuals, which included seven of the nine individuals with intramyocardial fat on cardiac MR, also had a chest CT. However, intramyocardial fat was missed on CT in one of these seven individuals by both readers who agreed this was most likely due to cardiac motion on the non-gated chest CT study. There were a total of 18 patients who only had chest CTs and no dedicated cardiac MRI to assess for intramyocardial fat. Evaluation of these 18 individuals who only had chest CTs for intramyocardial fat only detected one individual with intramyocardial fat. Thus, evaluation of the chest 34 chest CTs only detected a total of 7 (21%) cases of intramyocardial fat.
There was no statistically significant difference among any of the cardiac functional parameters in the individuals with intramyocardial fat compared to those without ( Table 3). The average LVEF of the individuals who underwent cardiac MR was normal (60 ± 5) with only one individual (female, age = 10 years-old) who had intramyocardial fat that was associated with a mildly decreased LVEF of 53% and mild diffuse hypokinesis without a focal regional wall motion abnormality. No wall motion abnormalities were found in any of the other individuals. Only three individuals received gadolinium and only one individual had a small patch of Airway and lung parenchymal involvement. Among those who underwent chest CT (n = 34), the most common findings in the lung parenchyma were hyperlucent lung parenchyma (n = 17, 50%), pulmonary nodules (n = 17, 50%), and bandlike areas of scarring (n = 19, 56%). Hyperlucent lung parenchyma usually involved < 50% of the lobe and was most common in the right upper lobe (n = 11, 32%). Lung cysts nearly always involved < 50% of a lobe and were most commonly involved the right lower lobe (n = 6, 18%). Both calcified and non-calcified nodules were present. Nine individuals had only non-calcified nodules, four individuals had only calcified nodules and four had both calcified and non-calcified nodules. All nodules were well-circumscribed, smoothly marginated, and round or oval in shape. The calcified nodules were completely calcified and ranged in size from 1 to 6.5 mm. Non-calcified nodules ranged from 1 to 17 mm in size and were completely soft tissue in attenuation. Large non-calcified nodules (10 to 17 mm) were observed in only one individual of our cohort. Interlobular septal thickening was present in four individuals (12%). Incidental, acute lung parenchymal findings in 4 individuals (12%) who had pneumonia at the time of scanning included consolidation and ground glass opacity. Minimal bronchiectasis was present seen in one individual (3%). Tracheal diverticulum and tracheal bronchus were incidental findings in two individuals (6%). One individual (38 year-old non-smoker) had a perihilar, left upper lobe mass (3%), which was biopsied and findings were compatible with primary lung adenocarcinoma. These findings are illustrated in Figs. 2 and 3 and summarized in Table 4.

Vascular involvement.
More than half of the individuals with chest CTs had pulmonary venous dilation (n = 21, 62%), as shown in Fig. 3. Affected pulmonary veins were diffusely dilated. Right and left pulmonary veins were involved similarly (n = 17 and n = 16, respectively). Ten individuals also had abnormal enlargement of systemic veins: internal jugular vein, subclavian vein, brachiocephalic vein, superior vena cava and the azygos vein. Pulmonary embolism was detected in two symptomatic individuals (6%) on CT angiography (Fig. 3). Mild aneurysmal dilation (diameter 42-43 mm) of the ascending aorta was present in two individuals (6%). The vascular findings are summarized in Table 4.

Pleural involvement.
Pleural effusion and thickening were detected in three (9%) and five (15%) individuals, respectively (Table 4). Pleural effusion was found incidentally in individuals acutely ill with pneumonia and likely unrelated to Proteus syndrome. No individuals had pleural nodules.
Chest wall involvement. Skeletal findings included scoliosis, asymmetric vertebral body growth, overgrowth of posterior elements of the spine, ribs, and scapulae. Scoliosis secondary to asymmetric overgrowth of various portions of the spine was the most common chest wall manifestation (n = 32, 94%) in this study. Skeletal findings of Proteus syndrome are depicted in Fig. 5. Overgrowth of chest wall fat was seen in ten individuals (29%). One individual (3%) had a well-circumscribed area of focal fat overgrowth in the right axilla suspicious for lipoma and three individuals (9%) had asymmetric breast size and/or asymmetric breast tissue/fat composition related to fat overgrowth. In individuals who had both deformed skeletal anatomy (e.g. scoliosis) as well as  www.nature.com/scientificreports/ soft tissue asymmetry, the soft tissue asymmetry was not associated with the areas of skeletal deformity. Chest wall adipose overgrowth findings are depicted in Fig. 6 and summarized in Table 4.

Discussion
In this study, we were able to further define the prevalence of cardiothoracic imaging findings in Proteus syndrome. Intramyocardial fat was common (45%) on cardiac MRI and its severity had no association with age. Although present in many places in both ventricles, the intramyocardial fat was always present in the septal wall. This intramyocardial septal wall fat is on the mild end of a spectrum of abnormal fat growth that includes the myocardial septal lipoma described within the clinical diagnostic criteria for Proteus syndrome (Table 1) at the more severe end of the spectrum. Interestingly, intramyocardial fat was also detected on the routine, non-cardiac gated chest CTs about half of the time (21%). This illustrates the importance of looking at the heart on routine chest CT exams, but also confirms that chest CT is an imperfect technique for intramyocardial fat detection as it missed cases detected by cardiac MRI. The intramyocardial fat does not appear to correlate with wall motion abnormalities or significantly impact LV function. Although a previous study reported arrhythmia (right bundle branch block) associated with intramyocardial fat in Proteus syndrome 25 , we observed no arrythmia associated with the presence of intramyocardial fat or any other adverse cardiac complications. As Proteus syndrome is an overgrowth syndrome, we hypothesize that this intramyocardial fat may represent overgrowth of areas of physiologic intramyocardial fat that has been well described in healthy individuals. In healthy individuals, intramyocardial fat is commonly seen in the RV up to 85% of the time, but is only present in the LV or in both ventricles < 20% of the time 26,27 . Thus, Proteus syndrome differs by commonly having intramyocardial fat in both ventricles, with LV predominance over the RV. However, intramyocardial fat in Proteus syndrome is associated with no wall motion abnormality, a feature it shares with the intramyocardial fat present in healthy individuals. Unlike Proteus syndrome, pathologic conditions with intramyocardial fat tend to have associated ventricular chamber enlargement, dysfunction and wall motion abnormalities. These include arrhythmogenic RV cardiomyopathy (ARVC), healed myocardial infarction, dilated cardiomyopathy, and Duchenne's muscular dystrophy. It should be noted that myocardial fat is no longer part of the diagnostic criteria for ARVC 28 . Further follow-up of the individuals with intramyocardial fat will be necessary to characterize its behavior over time and determine if it behaves similarly to a hamartomatous anomaly or more similarly to a fatty neoplasia.
Pulmonary nodules, areas of hyperlucent lung, and bandlike areas of scarring were present in at least 50% of individuals, which is much greater than the 8-13% described in prior reports 8,12,17,29 . Although we did not have pathology available for the pulmonary nodules in this study, they are suspected to be hamartomas based on prior published pathology 12 . In regards to the hyperlucent lung parenchyma, previous studies with pathology found that the areas of hyperlucent lung parenchyma corresponded to panlobular emphysema 12,15 . However, additional pathological evaluation is needed to further validate this correlation.
Pulmonary venous dilation was present in over half of our cohort. However, systemic venous dilation was less common. Venous dilation in Proteus syndrome is secondary to vascular overgrowth, but it remains uncertain as to why the pulmonary veins are affected more commonly than systemic veins. Vascular anomalies (tumors, malformations) have been previously reported 6,8,30 , however we did not observe any.
Chest wall abnormalities related to dysregulated skeletal and fat overgrowth are extremely common in Proteus syndrome 1,5,7,17 , and were also commonly observed in the current study. Scoliosis secondary to overgrowth of   www.nature.com/scientificreports/ various vertebral elements is almost universally present and tends to be progressive in Proteus syndrome, leading to significant deformity of the chest wall and eventual respiratory compromise that may necessitate orthopedic intervention. Additionally, we observed significant asymmetric breast size and/or composition related to an abnormal ratio of fat and parenchymal tissue in the affected breast. To our knowledge, this finding has not been previously reported.
The main thoracic complications of Proteus syndrome include pulmonary thromboembolism, respiratory failure, pneumonia, and malignancy. Pulmonary thromboembolism in Proteus syndrome, which was observed in our cohort, is elevated relative to the general population and is thought to be due to decreased anticoagulation proteins, venous stasis in vascular malformations, as well as the effects of the pathogenic AKT1 variant on endothelial cells [1][2][3]5,8,10 . Proteus syndrome is associated with an increased risk of developing malignancy of any kind which is already well documented in the literature, with meningioma and ovarian cystadenoma being among the more common 1,2,6-8 . In our population of individuals with Proteus syndrome, we found one left upper lobe mass that was found to be primary lung adenocarcinoma in an individual with no other risk factors other than Proteus syndrome.

Limitations
Our study has several limitations. The main limitation is that it is retrospective in nature. Although we have a large cohort of individuals for such a rare disease, not all individuals had both cardiac MRI and chest CT imaging performed. Second, the majority of the cardiac MRI studies and several of the chest CTs were performed without contrast. Lack of contrast limited our ability to assess for myocardial fibrosis on cardiac MRI and to assess for pulmonary embolism on the non-contrast chest CTs. Third, although many individuals in our cohort had lung parenchymal findings, pulmonary function testing was not available for correlation. This is an important clinical question and will be looked at in a future study. Finally, pathology was not available for all of the imaging findings we observed.

Conclusion
Proteus syndrome is a rare disorder with many cardiothoracic imaging findings that are much more common than previously described in the literature. The most common areas of involvement include the chest wall (skeletal and adipose tissue), lung parenchyma, thoracic vasculature, and myocardium. Additionally, it is already known that this syndrome has an increased risk of complications that include pulmonary thromboembolism, respiratory failure, pneumonia, and malignancy-findings which we also observed in our cohort of individuals with Proteus syndrome.