Characterization of the hepatosplenic and portal venous findings in patients with Proteus syndrome

Abstract

Proteus syndrome (PS) is a rare disorder caused by a mosaic AKT1 variant that comprises patchy overgrowth of tissues derived from all three germinal layers affecting multiple viscera. We sought to delineate the extent of hepatoportal manifestations in patients with PS. We identified patients with PS who had abdominal imaging from 1989 to 2015 in a natural history study. Imaging was characterized for evidence of focal findings in the liver, spleen, and portal vasculature and for organomegaly. Relevant clinical and laboratory data were compared among those with or without organomegaly. Abdominal imaging was available on 38 patients including 20 who had serial studies. Nine patients had focal hepatic lesions including vascular malformations (VMs). Focal splenic abnormalities were noted in seven patients. Patients without cutaneous VMs did not have visceral VMs. Nine patients had splenomegaly, 12 had portal vein dilation, and 4 had hepatomegaly. There was a weak correlation of portal vein dilation to spleen height ratio (r² = 0.18, p < .05). On laboratory evaluation, hepatic function was normal but there was thrombocytopenia in those with splenomegaly; platelet counts were 179 ± 87K/μL compared to those with normal spleen size at 253 ± 57K/μL (p < .05). Overall, focal hepatosplenic abnormalities occurred in 11 of 38 (29%) patients with PS. Splenomegaly and portal venous dilation were both found in 8 of 38 (21%) patients; however, other than relative thrombocytopenia, there was no evidence of portal hypertension. Although the AKT1-E17K somatic variant is a suspected oncogene, there were no malignant lesions identified in this study.

1 |. INTRODUCTION

Proteus syndrome (PS) is a rare disorder characterized by patchy overgrowth of tissues derived from all three germinal layers, including skin, connective tissue, and vasculature (L. Biesecker, 2006; L. G. Biesecker et al., 1999). It is caused by a somatic c.49G>A p.(Glu17Lys) activating variant in the AKT1 gene (Lindhurst et al., 2011). The incidence is less than 1 per million, and only around 200 patients in total have been identified in the United States and Western Europe (L. G. Biesecker et al., 1998).

The differential diagnosis of PS includes the PIK3CA-related overgrowth spectrum (PROS), which encompasses the former designations of CLOVES syndrome, Klippel–Trenaunay syndrome (KTS), and isolated hemihyperplasia (Keppler-Noreuil et al., 2015), along with neurofibromatosis Type 1 (Williams et al., 2009) and PTEN hamartoma tumor syndrome (Pilarski et al., 2013). Because both PS and PROS are typically mosaic disorders, they are remarkably variable, which led to diagnostic confusion. To address this, stringent diagnostic criteria were developed to differentiate PS from other overgrowth disorders (L. G. Biesecker et al., 1999; Turner, Cohen Jr., & Biesecker, 2004).

The most common manifestations in PS include overgrowth of multiple tissues including skin, skeletal, adipose, and vascular tissue. Less common manifestations include the disproportionate growth of viscera including thymus, kidney, gonadal tissues (ovary, testis), liver, and spleen with tumors such as ovarian cystadenomas and parotid adenomas. In addition, there is an increased risk of deep vein thromboses and pulmonary emboli resulting in the major cause of morbidity and mortality in PS (Keppler-Noreuil, Lozier, Sapp, & Biesecker, 2017).

Only a few patients with PS having hepatoportal involvement have been reported (Turner et al., 2004). Portal vein thrombosis has been suspected in two patients (Slavotinek, Vacha, Peters, & Biesecker, 2000; Staub, Schmid, & Huber, 2006), but it remains unclear if the latter patient met the diagnostic criteria put forward. A large cohort of individuals with PS has been followed at the National Institutes of Health Clinical Center (NIHCC) as part of a natural history protocol under the guidance of National Human Genome Research Institute (NHGRI). Most of these research participants undergo routine screening evaluations including imaging studies. We present the clinical findings and evaluations of these individuals with PS for the purposes of better characterizing the prevalence, extent, and natural history of liver, spleen, and portal vascular abnormalities in this disorder.

2 |. METHODS

2.1 |. Study population

A database of all individuals with PS who underwent abdominal imaging including ultrasonography (US), computed tomography (CT), or magnetic resonance imaging (MRI) or had their imaging reviewed at NIHCC from 1989 to 2015 was obtained, and their clinical charts were reviewed. Only the individuals who met diagnostic clinical criteria (L. G. Biesecker et al., 1999) for PS summarized in Supporting Information Table S1 were included.

Informed consent was obtained for all participants who were enrolled in clinical research protocols that were approved by the NHGRI IRB (NCT00001403).

2.2 |. Data collection and radiologic evaluation

The medical charts were reviewed, and the relevant clinical data were abstracted including basic demographics, clinical data, and laboratory studies performed at the time when radiographic imaging was performed. An informal PS severity score was calculated by summing points for increasing severity of skeletal involvement and overgrowth of other organ systems. Up to five points were assigned for asymmetric, disproportionate overgrowth of the skeletal system. Up to six points were assigned for overgrowth in the central nervous system, eye, tonsils/adenoids, or testes. Lung cysts or lipoma and cutaneous and visceral vascular malformations (VMs) were assigned one point each.

The abdominal imaging studies were read and analyzed by a radiologist (A.L.). In the case of multiple studies during a single visit, all available imaging including 48 ultrasounds in 27 patients, 32 MRI scans in 16 patients, and 26 CT scans in 11 patients for each organ were used to identify any visceral findings (Supporting Information Table S2). Among these patients, 15 had US evaluation only.

The measured visceral sizes were dichotomized into normal or enlarged. Given the wide age range in the cohort, hepatomegaly was defined using established growth models (Konus et al., 1998). Spleen size was analyzed using two methods: spleen length meeting splenomegaly criteria per previously published norms (Megremis, Vlachonikolis, & Tsilimigaki, 2004) and by spleen length-height ratios (Gunay-Aygun et al., 2013). Lastly, a portal vein diameter of 13 mm or higher was determined to be enlarged per evaluation of literature (Patriquin et al., 1990). Evaluation of vascular abnormalities and mass lesions was made using multiple modalities, when available.

2.3 |. Statistical analysis

Baseline data are presented as frequencies and described with appropriate measures of central tendency. Comparisons between groups were performed using Χ² or Fisher’s exact test for categorical variables and the Mann–Whitney U test for continuous variables. Correlation between portal vein diameter, spleen size, and platelet count was analyzed using Spearman’s rank correlation coefficient. Significance was observed at p < .05. No correction for multiple testing was performed. Statistical analyses were performed using RStudio (v.0.99).

3 |. RESULTS

3.1 |. Clinical findings

Thirty-eight individuals with PS were evaluated. The median age of the patients was 12.9 (6.7–20.2) years. Overall, the study cohort had a male predominance at 22 (58%) and 71% of them were white (Table 1).

TABLE 1.

Demographics, clinical characteristics, and laboratory data of the study population at the last follow-up^a

	Entire cohort (n = 38)	Organomegaly (n = 21)	No organomegaly (n = 17)
Age	12.9 (6.7–20.2)	13.6 (6.6–18.1)	12.1 (6.3–23.2)
Male gender	22 (58%)	14 (66%)	8 (47%)
Ethnicity
White	27 (71%)	16 (76%)	11 (65%)
Hispanic	5 (13%)	2 (10%)	3 (18%)
Black	1 (3%)	1 (5%)	–
Others	5 (13%)	2 (10%)	3 (18%)
Subjects >18 years	10 (26%)	5 (24%)	5 (29%)
BMI (kg/m2)	17.9 (15.3–23.3)	17.7 (15.0–23.5)	18.0 (15.6–23.6)
<18 years	16.3 (14.8–18.8)	15.9 (14.8–20.0)	16.7 (14.8–18.5)
≥18 years	24.6 (19.6–28.7)	20.1 (18.1–32.4)	26.3 (22.3–29.1)
PS scale without L/S (n = 35)c	7.1 ± 2.3	7.2 ± 2.1	6.9 ± 2.6
Superficial VMs (n = 35)	26 (74%)	13 (62%)	13 (76%)
History of VTE	8 (21%)	5 (24%)	3 (18%)
Anticoagulation use	4 (11%)	3 (14%)	1 (6%)
Pulmonary emboli	6 (16%)	3 (14%)	3 (18%)
Laboratory data
WBC (K/μL) (n = 32)c	6.3 ± 2.0	6.0 ± 2.4	6.6 ± 1.5
Hgb (g/dL) (n = 32)c	12.6 ± 2.0	12.4 ± 2.3	12.8 ± 1.6
Platelets (K/μL) (n = 32)c	250 ± 105	215 ± 70	249 ± 72
PT (s) (n = 29)c	15.1 ± 3.4	15.9 ± 4.2	13.9 ± 1.2
Creatinine (mg/dL) (n = 36)c	0.5 ± 0.2	0.5 ± 0.2	0.5 ± 0.2
BUN (mg/dL) (n = 31)c	12.5 ± 4.4	13.3 ± 3.9	11.5 ± 5.0
Albumin (g/dL) (n = 23)c	3.9 ± 0.4	3.9 ± 0.4	4.0 ± 0.5
ALP (IU/dL) (n = 29)c	210 ± 119	221 ± 127	198 ± 112
<18 years (n = 21)	257 ± 105	265 ± 119	248 ± 92
≥18 years (n = 8)	87 ± 27	100 ± 35	74 ± 7
ALT (IU/dL) (n = 26)c	29 ± 18	25 ± 13	32 ± 22
AST (IU/dL) (n = 29)c	24 ± 17	21 ± 7	28 ± 23
Total bilirubin (mg/dL) (n = 25)c	0.5 ± 0.2	0.6 ± 0.2	0.5 ± 0.2
Direct bilirubin (mg/dL) (n = 25)c	0.14 ± 0.06	0.14 ± 0.05	0.15 ± 0.07

PS = Proteus syndrome; VTE = venous thromboembolism; L/S = liver/spleen; VMs = vascular malformations; ALP = alkaline phosphatase; AST = aspartate aminotransferase; ALT = alanine aminotransferase; PT = prothrombin time. – indicates absence of data.

^aNo significant differences between groups.

^bMedian, interquartile range.

^cMean, standard deviation.

The mean PS severity score was 7.1 ± 2.3 and a majority of the cohort (26 individuals, 68%) had evidence of cutaneous VMs. Patients younger than 18 years of age had significantly lower body mass index (BMI) than older patients (16.3 [14.8–18.8] kg/m² to 24.6 [19.6–28.7] kg/m², p < .001). Eight patients had a history of venous thromboembolism; six of them had documented pulmonary emboli, three of which were fatal. Four were undergoing long-term anticoagulation with warfarin. When comparing patients affected with visceral enlargement to those without, it was found that there were no significant differences in any of the above clinical parameters.

3.2 |. Radiologic findings

A total of 20 individuals (53%) had multiple radiographic evaluations over a median follow-up time of 4.7 (1.8–9.4) years. Fourteen individuals (37%) had abnormalities detected on initial imaging prompting further characterization by multiple advanced imaging studies on the same visit.

Focal visceral lesions and abnormalities were found throughout the hepatoportal system (Table 2). Given that cutaneous VMs are common in individuals with PS, we separated vascular and nonvascular lesions. Visceral VMs were found in 7 of 19 patients who had cutaneous VMs and in none of 19 patients who did not have cutaneous VMs.

TABLE 2.

Visceral abnormalities

	N	%
Visceral enlargement
Hepatomegaly (n = 34)	4	12
Splenomegaly (n = 30)	7	20
Portal vein ≥13 mm (n = 27)	12	44
Hepatic abnormalities (9 of 34 patients)a
Vascular malformations	2	6
Cysts	4	12
Bile duct hamartomas	1	3
Focal intrahepatic biliary dilation	1	3
Large middle hepatic vein variant	1	3
Splenic abnormalities (7 of 35 patients)
Vascular malformations	3	9
Granulomas	1	3
Noncystic lesion	1	3
Echogenic lesion	1	3
Infarct	1	3
Vascular abnormalities (5 of 38 patients)b
Chest wall collaterals	1	3
Abdominal wall collaterals	2	5
Large left testicular vein	1	3
SMV junction varix	1	3

Note: We use here the ISSVA terminology of vascular anomaly instead of the obsolete term of hemangioma.

^aOne patient had both hepatic cysts and vascular malformations.

^bOne patient had both chest and abdominal wall collaterals.

The cohort was assessed for visceral enlargement (Figure 1). The most common abnormalities of size in the study population were found in the spleen and portal vein (Table 2). Using radiographic assessment, 7 of 30 (23%) patients had splenomegaly on cross-sectional imaging available. Evaluating by spleen length-height ratio, the findings were comparable to radiographic assessment with 9 of 30 (30%) patients meeting criteria for splenomegaly (Figure 1d).

FIGURE 1.

Visceral enlargement in Proteus syndrome. (a) A 15-year-old individual with hepatomegaly at 21.5 cm on MRI abdomen, (b) a 16-year-old individual with severe scoliosis with marked splenomegaly at 21.5 cm with multiple sonographically echogenic splenic vascular malformations (arrowheads) on abdominal CT, (c) a 31-year-old individual with portal vein at 26 mm on abdominal MRI with contrast (dotted line), (d) scatterplot for spleen height ratio adjusted for age with points above grey line indicating splenomegaly, (e) association of portal vein diameter to spleen height ratios, and (f ) platelet counts in patients with high or normal spleen height ratios

Splenic imaging showed focal abnormalities in seven patients. Splenic VMs were identified in three patients (Supporting information Figure S1). Furthermore, one had granulomas and two others had lesions that were not differentiable. Another patient with splenomegaly had splenic infarcts. This individual eventually had a splenectomy and also severe reactive postsplenectomy thrombocytosis, which was treated with clopidogrel. He later died from a pulmonary embolism.

Portal vein dilation, defined as diameter ≥13 mm, was present in 12 of the 27 (44%) available patients. The mean portal vein diameter was 19.6 ± 5.3 mm compared to 9.5 ± 0.2 mm in the patients without dilation. Hepatomegaly was less common; it is present in only 4 of 34 (12%) patients in whom liver imaging was available.

There was also a weak, yet significant correlation of a larger portal vein diameter to a larger spleen height ratio (r² = 0.18, p < .05) (Figure 1e). In addition, although spleen height ratio was significantly different, liver span was not different in the two groups at 15 ± 4 cm and 16 ± 3 cm. However, there was no correlation with PS disease severity score or visceral or portal vein enlargement (Supporting Information Tables S3–S5).

Liver was the predominant location for focal lesions with several identified in eight individuals (Figure 2). Hepatic cysts were noted in four patients with one having multiple cysts measuring the largest 8.7 cm in size and another having septated cysts. Focal biliary dilation and bile duct hamartomas were the other nonvascular abnormalities. Hepatic vascular abnormalities were seen in three patients including one with a large, thrombosed hepatic VM (Supporting Information Figure S2) and another participant with a large middle hepatic vein variant.

FIGURE 2.

Hepatic lesions. (a) a 55-year-old individual with hepatic cyst on T1-weighted abdominal MRI, (b) a 29-year-old individual with septated cyst on liver ultrasound (arrowhead), and (c) a 18-year-old individual with multiple bile duct hamartomas on abdominal CT

Besides the VMs described above, several other vascular abnormalities were discovered in this cohort (Figure 3). There were abdominal and chest wall collateral vessels in two patients. A large left testicular vein and a diffusely ectatic splenic vein were noted in another patient. One patient with splenomegaly and large portal vein diameter was found to have a superior mesenteric vein varix at the junction between the portal vein and splenic vein (Supporting Information Figure S3). We did not find evidence of portal vein thromboses in our study population.

FIGURE 3.

Vascular abnormalities in patients with Proteus syndrome. (a) a 28-year-old individual with multiple abnormalities including splenomegaly at 15 cm on the left and (b) MR angiography on the right shows portal vein enlargement (a′), an ectatic diffusely enlarged splenic vein (b′), enlarged left testicular vein (c′), and enlarged anomalous vessels associated with right colon (d′)

3.3 |. Laboratory findings

Laboratory data were available for 36 patients (Table 1). The mean hematologic parameters were normal for the study population including white blood cell count at 6.3 ± 2.0K/μL, hemoglobin at 12.6 ± 2.0 g/dL, and platelets at 250 ± 105K/μL.

Liver enzymes were normal with mean alanine aminotransferase (ALT) at 29 ± 18 U/L and aspartate aminotransferase at 24 ± 17 U/L. The hepatic excretory function was normal with total serum bilirubin at 0.5 ± 0.2 mg/dL and direct bilirubin at 0.14 ± 0.06 mg/dL along with normal synthetic function with serum albumin at 3.9 ± 0.4 g/dL and prothrombin time at 15.1 ± 3.4 s. Age-appropriate mean alkaline phosphatase was in normal range for those aged 18 years and older at 87 ± 27 U/L and those younger than 18 years at 257 ± 105 U/L.

Patients with hepatomegaly, splenomegaly, and portal vein enlargement had consistently normal mean liver enzymes and function without significant differences compared to the normal-sized counterparts (Supporting information Tables S3–S5). There was a trend towards relative thrombocytopenia at 192 ± 42K/μL compared to 242 ± 77K/μL in the hepatomegaly group (p = .09); however, given the small number of individuals, this was not significant. Similarly, patients with splenomegaly had relative thrombocytopenia at 179 ± 87K/μL compared to 253 ± 57K/μL in normal-sized spleen group (p < .05) (Figure 1f ). Lastly, one patient with hepatic steatosis on US had an ALT of 57 U/L, which may reflect a diagnosis of nonalcoholic steatohepatitis.

3.4 |. Histopathology correlation

Autopsy was available for review on two patients. Both died from pulmonary emboli. One patient was the subject of a recent case report (Doucet, Bloomhardt, Moroz, Lindhurst, & Biesecker, 2016) and was found to have numerous dark brown to black nodular lesions in her spleen that microscopically were identified as cavernous VMs.

Another patient discussed earlier (Figure 4) with postsplenectomy thrombocytosis was found to have pale nodular lesions, sized from 0.5 to 1.2 cm, in the subcapsular area of the liver. Grossly, the nodules were lobulated with cystic spaces. Microscopically, a 0.2-cm nodule was found to show aggregation of irregular-shaped small bile ducts or varying size, embedded in collagenous connective tissue, consistent with atypical biliary hamartoma. The larger nodules consisted of cystic spaces surrounded by thick collagenous connective tissue wall without definite evidence of edema or inflammation and lined by bland-looking cuboidal to low-columnar biliary-type epithelial cells. The remaining liver parenchyma was remarkable for mild steatosis and sinusoidal congestion. Unfortunately, the portal vasculature was not specifically examined at autopsy.

FIGURE 4.

(a) Hepatomegaly with bile duct hamartomas seen on abdominal CT, (b) H&E stain showing typical biliary hamartomas, (c) dilated bile ducts with thick collagenous wall (inset: magnification of lining epithelium), and (d) hepatic steatosis with sinusoidal congestions

4 |. DISCUSSION

This study describes the hepatoportal system abnormalities in a well-characterized cohort of individuals with PS. The pattern of findings reflects distinct visceral enlargement in the portosplenic circuit and a myriad of vascular and nonvascular lesions in the liver and spleen. Patients with PS had multiple focal visceral abnormalities and often had unexplained visceral enlargement.

VMs in PS can be capillary, venous, or lymphatic, and these are components of the diagnostic criteria. VMs were most common in the liver and spleen. Hepatic VMs have a prevalence of only 1%–4% in general population (Gandolfi et al., 1991). These benign VMs are grossly well-circumscribed red-brown lesions with the presence of cavernous vascular channels seen histologically, and they are considered to be congenital hamartomas that enlarge by ectasia. Multiple hepatic VMs have been noted in several genetic diseases including hereditary hemorrhagic telangiectasia (Buonamico et al., 2008). There is also an association of hepatic VMs with cutaneous VMs, especially in children (Hughes et al., 2004). This finding holds true in this cohort with PS, as those with hepatic and splenic VMs all had cutaneous VMs. Conversely, those without cutaneous VMs did not have visceral VMs. Splenic VMs are proposed to have a similar etiology as hepatic VMs but are less common in the general population based on US studies with the prevalence of 0.2% (Bachmann & Gorg, 2005). In patients with PS, the incidence of splenic VMs was 8.6%.

Other focal hepatic abnormalities included biliary hamartomas and hepatic cysts. Interestingly, multiple foci of disorganized bile duct-like structures were also identified in one patient who also had biliary hamartomas and are not compatible with any category of fibrocystic liver disease. Hepatic cysts were seen in 4 of 34 (11.8%) patients with PS. Normally, hepatic cysts are seen in 1% of autopsied adults in the general population with most common being simple hepatic cysts. As in individuals with PS, these cysts are rarely symptomatic and have a benign course unless they are large or are complicated by infection, rupture, or cause biliary obstruction (Hanazaki et al., 1997; Salemis, Georgoulis, Gourgiotis, & Tsohataridis, 2007).

The vascular and biliary abnormalities in PS may be explained by the underlying AKT1 mutation. The mutated AKT1 protein kinase is overexpressed in affected tissue with increased signaling through PI3K-AKT (Bader, Kang, Zhao, & Vogt, 2005), which is integral to multiple intracellular pathways. Phosphorylated AKT1 is an upstream modulator of the mammalian target of rapamycin (mTOR)-signaling pathway (Porta, Paglino, & Mosca, 2014), which leads to evasion of cell apoptosis and induces proliferation through p21/p27 activation (Rossig et al., 2001). It also leads to sustained angiogenesis through a positive feedback loop with vascular endothelial growth factor (VEGF) and also leads to increased cell survival (Karar & Maity, 2011). These mechanisms can explain the scattered overgrowth noted in PS given the mosaic pattern of the AKT1 mutation in all three embryonic germ layers. In addition, there is direct evidence with hepatic VMs incidentally reducing in size with VEGF inhibition therapy and blockade of AKT/mTOR pathway, which give credence to this proposed mechanism (Mahajan, Miller, Hirose, McCullough, & Yerian, 2008).

The other major visceral abnormality was visceral enlargement including splenomegaly, which has been described previously (L. G. Biesecker et al., 1998). Splenomegaly was accompanied by relative thrombocytopenia and an enlarged portal vein diameter. Normally, splenic sequestration accounts for one-third of the total platelet mass. As spleen size increases, this ratio increases, leading to thrombocytopenia (Aster, 1966) and may partially explain our findings. This was most well demonstrated in one individual with PS who underwent splenectomy in this cohort. The accompanying portal vein enlargement in this individual is of interest because there was no evidence of functional hepatic compromise suggesting that it may be related to vascular ectasia. Furthermore, hepatomegaly is not common in PS, and when it is present, it is not associated with biochemical evidence of hepatobiliary dysfunction. In addition, the hepatic steatosis in this cohort is more related to elevated BMI rather than PS directly.

In comparison with generalized overgrowth syndromes, similar clinical reports of abnormal abdominal visceral findings were found. In patients with Beckwith–Wiedemann syndrome (BWS), a disease characterized by hemihypertrophy along with other congenital features, hepatic and splenic VMs have been documented (Francisco, Goncalves, Borges, & Neto, 2013), and there is an increased risk of hepatoblastoma (Mussa & Ferrero, 2015). Overlapping clinical findings are found in PS and PROS and may be explained in part by their respective somatic mutations in AKT1 and PIK3CA, which are in the same PI3K-AKT signaling pathway. PROS describes phenotypes previously described as CLOVES syndrome and KTS because these conditions have been found to have somatic PIK3CA variants. In PROS, ectatic portomesenteric veins can occur and lead to thrombosis and portal hypertension (Alomari et al., 2011). Splenomegaly with multiple VMs has been documented in KTS (Misawa et al., 2014), which is characterized by a triad of port-wine stain, varicose veins, and bony and soft tissue overgrowth involving an extremity.

In summary, there were numerous focal lesions identified in the hepatoportal system of patients with PS. Although several patients had portal vein enlargement, splenomegaly, or thrombocytopenia, there was no evidence of hepatic dysfunction or portal hypertension. In addition, compared to the malignant potential in BWS, 20 individuals with PS had long-term follow-up ranging from 1 to 16 years without evidence of malignant transformation of focal lesions of the liver or spleen.

Based on similar findings of portal venous dilatation in non-cirrhotic portal hypertension to that seen in these patients with PS, we suggest consideration of hepatic vein catheterization, especially if patients have progressive thrombocytopenia with splenomegaly. Furthermore, it may be advisable to avoid surgical resection of the spleen in patients with PS as the resultant reactive thrombocytosis may increase the potential for thrombotic events in this already predisposed population.

Given our experience, we recommend a baseline ultrasound of the abdomen on all patients with PS. For cystic lesions, we recommend any septated or large cysts should be reimaged, preferably with MRI, for stability as they can carry malignant potential including rare risk of mucinous cystadenoma and cystadenocarcinomas (Devaney, Goodman, & Ishak, 1994). We do not believe that follow-up MRIs for visceral VMs are necessary unless patients have symptoms. One exception would be progressive thrombocytopenia with large hepatic VMs given the risk of Kasabach–Merritt syndrome (Concejero, Chen, Chen, Eng, & Kuo, 2009). Furthermore, we do recommend repeat abdominal ultrasounds with flow Doppler of portal vein for patients with thrombocytopenia and either splenomegaly or portal vein enlargement to determine whether patients are progressing to overt portal hypertension or developing a portal vein thrombosis. Finally, given the rare case reports of portal vein thrombi, we recommend that the portal vasculature and mesentery should be evaluated in deceased patients who undergo autopsies to look for evidence of ectasia and thrombi.

Abbreviations:

ALP

alkaline phosphatase

ALT

alanine aminotransferase

AST

aspartate aminotransferase

BMI

body mass index

BWS

Beckwith-Wiedemann syndrome

CT

computed tomography

Hgb

Hemoglobin

IQR

interquartile range

KTS

Klippel–Trenaunay syndrome

L/S

liver/spleen

MRI

magnetic resonance imaging

NASH

nonalcoholic steatohepatitis

NHGRI

National Human Genomic Research Institute

NIHCC

National Institutes of Health Clinical Center

PHTS

PTEN hamartoma tumor syndrome

PROS

PIK3CA-related overgrowth spectrum

PS

Proteus syndrome

PT

prothrombin time

SD

standard deviation

US

ultrasonography

VM

vascular malformation

VTE

venous thromboembolism

WBC

white blood cell

X²

chi-squared