The obesity epidemic: The role of radiology

The pathogenesis of NAFLD is not fully understood, but is thought to be strongly associated with insulin resistance. In addition to hyperglycemia, insulin resistance alters the uptake, synthesis, degradation, and secretion of lipids in hepatocytes, resulting in an accumulation of triglycerides within the liver.⁷ A consequence of increased intracellular fatty acids is intensified oxidative stress within hepatocytes, which may be responsible for disease progression. The spectrum of NAFLD ranges from simple steatosis to the development of necro-inflammmatory change (non-alcoholic steatohepatitis [NASH]) to the formation of fibrosis and cirrhosis. Up to one third of NASH subjects demonstrate some degree of fibrosis, often at initial presentation.^1,8-12 Once cirrhosis occurs, there is an associated risk of developing hepatocellular carcinoma.^13,14

Cryptogenic cirrhosis, cirrhosis without a known underlying cause, is the third most common diagnosis in liver transplant recipients. Recent literature suggest a significant subset of these patients may actually have NASH, further highlighting the significant morbidity and mortality of this disease.^15-19

Hepatic steatosis may be associated with progressive liver injury not only in patients with NAFLD but also in patients with chronic hepatitis C infection and liver transplants. This has resulted in significant interest in the diagnosis and quantification of hepatic steatosis. The current gold standard for the diagnosis and quantification of steatosis is liver biopsy, although biopsy is invasive, painful and can result in complications and even death.²⁰ In addition, liver biopsy can be inaccurate due to sampling errors, since steatosis is frequently heterogeneous. Variation in observations in the qualitative assessment of histopathological fat grading also limits reproducibility.^21,22

Steatosis is defined sonographically by the manifestation of a hyperechoic liver parenchyma with tightly packed fine echoes and posterior beam attenuation. Severe steatosis is diagnosed by obscuration of hepatic vasculature by hepatic parenchymal echogenicity and acoustic edges from intracellular lipid accumulation (Figure 1). Severe fatty liver, containing >30% fat by weight, is detected by sonography with a sensitivity and specificity of 67% to 84% and 77% to 100%, respectively.^23,24 Sonography is poor in detecting smaller amounts of fat in liver and in general is limited by the subjective nature of the assessment.²⁵ This subjectivity also significantly hinders reliable assessment of interval change.^26-28

Computed tomography (CT) is used to quantify hepatic steatosis through the measurement of liver attenuation. Decreased liver attenuation is suggestive of increased liver fat. A low-dose vertebral scout to localize the T12 to L1 intervertebral space and a subsequent single image acquisition at T12 to L1 is a practical and reliable strategy for routine measurement of liver fat.²⁹ Unenhanced CT indicates hepatic steatosis when liver attenuation is at 10 Hounsfield units (HU) below that of the spleen (Figure 2). Steatosis may also be suggested by the presence of heterogeneous liver attenuation and by “pseudoenhancement” (relative high density of the hepatic vessels compared with surrounding parenchyma). The criteria for fatty liver on post-contrast CT includes a lower attenuation of the liver than the spleen by a difference of >20 HU, or liver attenuation that is lower than muscle. Like ultrasound, CT is not sensitive in detecting mild elevations of hepatic lipid content (5% to 30%).²⁶ Lall, et al., reported an overall sensitivity of 50% and specificity of 75% for estimation of liver fat.²⁵ Inherent pitfalls in assessing liver fat estimation with CT include the co-existence of conditions that may result in increased liver attenuation, such as therapy with amiodarone and hemochromatosis. Hyper-attenuating liver parenchyma in the range of 70 to 130 HU limits quantifying liver fat.³⁰

Magnetic resonance imaging (MRI) is recognized as a sensitive technique for the quantification of liver fat. The normal hepatic fat fraction derived from MRI is reportedly <9%.³¹ MRI is able to measure liver fat based on the slightly different precessional frequencies of fat and water protons in a magnetic field. The most commonly used quantitative method is the acquisition of in- and out-of-phase imaging, or chemical shift imaging, in which different echo time (TE) parameters are used in gradient-echo images so that the signal from fat protons is added or subtracted, respectively, from the signal of water protons.²⁵ Reduced signal on out-of-phase T1-weighted (T1W) images is an accurate proxy of histologic hepatic fat content with a Pearson correlation coefficient of 0.86 to 0.9132-34 (Figure 3). An important pitfall to the use of in- and out-of-phase imaging in steatosis is the potential susceptibility effect of liver iron; particularly with longer TEs, iron overload can result in an underestimation of liver fat.^{32, 35-37} Recently developed chemical shift imaging techniques incorporate the use of multiple TEs to create a T2* map, thereby correcting for the susceptibility effects of liver iron. Post-processing techniques based on mathematical algorithms to include multiple fat frequencies have also been suggested to provide a more robust MRI quantification of liver fat.³⁸

Proton (1H) MR spectroscopy (MRS) (Figure 4) has been found to correlate with hepatic lipid content. Estimates of the correlation coefficient between MRS measurements and histologic fat measurements range from 0.91 to 0.98.^39-41 This technique is considered sensitive to small variation (0.5% change) in hepatic lipid content and has the potential to assess for small changes in steatosis due to treatment.⁴¹ However, a small sampling area can result in inaccurate assessment of the overall amount of fatty change due to the heterogeneous distribution of fat.

Liver steatosis has been shown to be clearly linked to obesity. Clinically, BMI has often been used as a surrogate anthropometric marker of obesity-related morbidity. This measurement does not account for body tissue composition or distribution of body fat.^42-44 The highly metabolically active visceral fat has been shown to correlate more closely with liver steatosis than BMI.^45-51 MRI-derived estimates of visceral fat have proven more accurate than the anthropometric indices, including the waist-to-hip ratio (Figure 5).³¹ A variety of MR techniques have been employed for visceral fat estimation, with most using select slices to extrapolate to the rest of the body.^47-50 The single slice method can be inaccurate due to the significant extrapolation required.⁴⁹ A study by Brennan, et al., attempted to reduce this inaccuracy by dividing the body into 3-dimensional voxels of 2.02 2 2.02 2 8.00 mm3, and subsequently employing computer-assisted diagnostic techniques to provide a definite distribution of fat deposition.⁵² Visceral and subcutanaeous adisposity may also be estimated using unenhanced CT with the measurement of tissue areas demonstrating fat attenuation (Figure 6).

Dual-energy X-ray (DEXA) was originally created for bone density analysis. However, it can also be used in body composition studies.^53,54 Output measurements include total weights of fat, lean tissue, and bone mineral content. Radiation exposure of <15 μSv and a gantry that does not induce claustrophobia make this technique ideal for the pediatric population.⁵⁵ This technique, however, cannot provide separate measurements of subcutaneous and visceral fat.⁵⁶

Staggering obesity rates, failure of traditional weight-loss methods, and evidence-based effectiveness have placed bariatric surgery in the forefront of morbid obesity treatment.^57-59 Radiology plays an important role in the evaluation of the post operative bariatric patient.

Roux-en-Y is the most common bariatric procedure. The surgical protocol entails the creation of a gastric pouch separated completely from the remainder of the stomach, which is called the remnant. The jejunum is separated 30 cm to 40 cm distal from the ligament of Treitz and mobilized from the mesentery to form a gastrojejunostomy with side-to-side anastomosis. The anastomosed jejunal loop, called the Roux limb, is placed retrocolic through an aperture formed in, or in front of, the transverse mesocolon. Finally, a jejunojejunostomy is created 100 cm to 150 cm distal from the original gastrojejunostomy.⁶⁰

The post operative day 1 assessment protocol includes an initial gastrograffin fluoroscopic imaging study to evaluate for anastomotic leak. If a nasogastric tube is present, often after an open Roux-en-Y gastric bypass, the tube must be pulled back as to not obscure leakage from the anastomosis.⁶¹ Rarely, leaked contrast entering the lumen of a surgical drain may be the only fluoroscopic evidence of leak and must not be overlooked. CT can be a useful adjunct modality, capable of demonstrating leaked contrast, as well as intra-abdominal collections and abscesses.

Gastric staple-line disruption is readily identified with fluoroscopy.⁶² Leaked contrast will either enter the peritoneal cavity if the 2 portions of the stomach were surgically separated or will enter into the remnant if just a dividing staple line without surgical separation was used (Figure 7). Once contrast enters the gastric remnant, it may then be seen moving through the antrum and proximal duodenum (a path normally blocked by the Roux procedure). Contrast passing normally through the gastrojejunostomy may opacify the short blind-ending portion of the jejunum adjacent to the anastomotic site, mimicking leakage into the gastric remnant. This is distinguished from true leakage by a lack of progression of contrast beyond this site and the appearance of a jejuna mucosal pattern.⁶⁰

In 5% of patients, obstruction occurs as a complication at the gastrojejunostomy site, mesocolic window, and behind the Roux limb (Peterson's space).^61-63 Anastomotic stricture at the gastrojejunal anastomosis is seen on fluoroscopy as narrowing at the gastrojejunal anastomosis, with expansion of the gastric pouch and delayed transit of contrast into the Roux limb.⁶⁰ Mesocolic window stenosis is a potential cause of obstruction in patients with retrocolic placement of the Roux limb and is seen fluoroscopically as distention of the Roux limb proximal to the location of the mesocolic window.

Laparoscopic adjustable gastric banding involves an adjustable silicon band lined with an inflatable balloon that is placed around the superior stomach to create a small gastric pouch and an adjustable stoma to the rest of the stomach. The surrounding anterior gastric wall is usually sutured over the band towards the gastric pouch to prevent band slippage. The balloon is connected to a subcutaneous port in the abdomen that alters the diameter of the gastric stoma through injecting and aspirating saline into the port.^60,64,65

The first fluoroscopic evaluation of postbanding patients is done postoperatively on day 1 to assess the size of the gastric pouch, assess for leakage indicating iatrogenic injury, and to assess for normal passage of contrast through the stoma. The gastric pouch should be 3 to 4 cm maximally and symmetric when distended with contrast material. The stoma should measure 3 to 4 mm in diameter and should allow passage of all contrast material within 15 to 29 minutes postadministration.⁶⁶ The phi angle assesses slippage and is formed by the intersection of the lines parallel to the spinal column and antero-posterior plane of the gastric band. An angle that is 4° to 58° and 4 cm to 5 cm below the left hemidiaphragm is considered normal. Anterior or posterior herniation of the distal stomach through the band creates a phi angle < and > normal, respectively.^67,68 Band system leakage may occur at the level of the band, connector, tubing, or access port. Suspicion for this complication should arise when there is insufficient deflation volume and loss of eating restriction. The leak can be detected when adjusting the band diameter under fluoroscopy.⁶⁶ A 99mTc-albumin study can prove useful in locating the leak.⁶⁸ The lap-band procedure shares the majority of the common complications seen after Roux-en-Y and can be evaluated through similar radiologic studies. Chandler et al. provide a thorough, in-depth evaluation of rarer complications.⁶⁰

The size of obese and bariatric patients imposes certain limitations for imaging. CT and fluoroscopy table weight limits, often 450 to 500 lbs, and a gantry opening of 70 cm can preclude imaging certain patients. Dropping the CT table can potentially accommodate larger patients. The larger body habitus of bariatric patients can also degrade image quality. The use of conventional, rather than spiral scanning, techniques and increased mAs can improve quality imaging through increases in radiation.

Upper airway obstruction is an inherent risk of positioning obese patients in the supine position. The imaging team should be prepared for this possible complication and may consider having the lift team readily available.

Increasing incidence of obesity and NAFLD underscores the necessity of diagnosing fatty liver before progressing to advanced disease states. Imaging is an efficient and safe alternative to liver biopsy for estimating liver fat. Of the imaging options, MRI is considered the safest and most sensitive technique in detecting liver steatosis.

MRI and CT may also be used to estimate visceral fat, which demonstrates a better correlation with liver steatosis than anthropometric measures. Imaging also plays a pivotal role in the postoperative evaluation of bariatric patients.