Abstract
Background:
Identifying patients in whom ultrasound may be inadequate to exclude the presence of hepatocellular carcinoma (HCC) can inform interventions to improve screening effectiveness.
Aim:
Characterize correlates of suboptimal ultrasound quality and changes in ultrasound quality over time in patients with cirrhosis undergoing HCC screening.
Methods:
We performed a retrospective cohort study of patients with cirrhosis who underwent ultrasound examination at two large health systems between July 2016 and July 2019. Exam adequacy was graded by radiologists using the LI-RADS Visualization Score (A, B, C); we evaluated changes in visualization over time among patients with >1 ultrasound exam. We performed multivariable logistic regression to identify characteristics associated with limited ultrasound visualization (scores B or C).
Results:
Of 2,053 cirrhosis patients, 1,685 (82.1%) had ultrasounds with score A, 262 (12.8%) score B, and 106 (5.2%) score C. Limited visualization was associated with alcohol-related or NAFLD cirrhosis and presence of class II–III obesity. Among 1,546 patients with >1 ultrasound, 1,129 (73.0%) had the same visualization score on follow-up (1,046 score A, 60 score B, 23 score C). However, 255 (19.6%) of 1,301 with score A at baseline had limited visualization when repeated (230 score B, 25 score C), and 130 (53.1%) of 245 patients with baseline limited visualization had good visualization when repeated.
Conclusions:
Nearly one in five patients with cirrhosis had moderately-severely limited ultrasound visualization for HCC nodules, particularly those with obesity or alcohol-related or NAFLD cirrhosis. Ultrasound quality can change between exams, including improvement in many patients with limited visualization.
Keywords: Screening, cirrhosis, liver cancer, early detection, sensitivity
INTRODUCTION
Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer deaths worldwide and has rising mortality rates in the U.S., likely attributable to increasing numbers of patients with advanced hepatitis C infection and NAFLD.1,2 Early HCC detection remains one of the strongest drivers of prognosis, with several studies highlighting the association between HCC screening and improved survival in patients with cirrhosis.3,4 Across professional society guidelines, ultrasound with or without AFP remains the primary method of HCC screening in patients with cirrhosis.5,6
While ultrasound remains the most cost-effective modality for screening, there are limitations associated with its use. One of the largest criticisms of ultrasound in HCC screening is variability in sensitivity, ranging between 32% and 89% when used alone.7,8 A systematic review found ultrasound alone had a pooled sensitivity of only 47% for early-stage HCC detection, suggesting it misses over half of HCC at an early stage.8 A prior study found 20% of ultrasounds have moderate-severe limitations in visualization for exclusion of HCC nodules during screening, particularly in patients with steatosis, Child Pugh class B/C cirrhosis, and obesity.9 However, the optimal management of these patients is unknown. Although ultrasound visualization may remain stable over time, it is possible that it may vary between ultrasound exams given its operator dependent nature. Contrast-enhanced CT and MRI have been proposed as alternatives for HCC screening;10,11 however, they are associated with radiation exposure and contrast-related concerns, may have suboptimal accuracy in patients prone to compromised ultrasound quality, and have limited radiologic capacity. Further, these alternatives are costly, reinforcing the benefit of continuing ultrasound-based screening in patients when possible.
The aim of our study was to assess the quality of ultrasound, measured by the Liver Imaging, Reporting and Data System (LI-RADS) Ultrasound Visualization Score in patients with cirrhosis undergoing repeat HCC screening over time. These data are important to identify patients in whom ultrasound is inadequate and alternative screening strategies are needed, as well as those in whom it may be prudent to continue using ultrasound-based screening.
METHODS
Patient population
We conducted a retrospective cohort study among patients with cirrhosis at two health systems – Parkland Health and Hospital System (Parkland), the safety-net health system of Dallas County, and University of Texas Southwestern Medical Center (UT Southwestern), the academic tertiary-care referral health system in North Texas. Patients were identified using validated ICD-9/ICD-10 codes for cirrhosis.12 Cirrhosis diagnosis was confirmed by manual chart review, based on: 1) consistent histology; 2) non-invasive markers of fibrosis (e.g., transient elastography or FibroSure) demonstrating F4 fibrosis; or 3) imaging showing a cirrhotic-appearing liver with signs of portal hypertension including splenomegaly, varices, or thrombocytopenia. We included patients with at least one ultrasound examination between July 2016 and July 2019. We excluded patients with history of HCC or liver transplantation given lack of benefit of HCC screening in those subgroups. The study protocols were approved by the IRB at UT Southwestern Medical Center.
Ultrasound examinations
Parkland and UT Southwestern perform >7500 inpatient and outpatient ultrasound examinations per month. At both institutions, ultrasounds are performed by one of 90 registered sonographers using a standardized imaging protocol, and captured images are interpreted by one of 52 fellowship-trained radiologists. The liver-dedicated ultrasound protocol involves sequential longitudinal and transverse imaging through the left and right hepatic lobes; high-resolution imaging of the hepatic capsule and underlying parenchyma; color and spectral Doppler interrogation of the main portal vein; evaluation of the gallbladder and biliary system; measurement of spleen length and volume; and assessment for ascites. In contrast, a right upper quadrant ultrasound evaluates liver components without assessment of the spleen or ascites. All examinations were performed on an iU22, EPIQ 5 or EPIQ 7 ultrasound system (Philips Healthcare, Bothell, WA, USA) using a C5-1 or C9-2 curvilinear transducer for deep imaging and an L12-5 or L12-3 linear transducer for superficial imaging, or ACUSON Sequoia (Siemens Healthineers, Issaquah, WA, USA) using a 5C1 curvilinear and 10L4 or 14L5 transducers, respectively.
Data Collection
Patient characteristics including demographics, clinical history, and imaging results were obtained from electronic medical records. As previously described, liver disease etiology was classified as hepatitis C, hepatitis B, alcohol-related liver disease, NAFLD, and other.13 Body mass index (BMI) was obtained at each ultrasound exam (+/− 3 months) and categorized according to World Health Organization classification: normal (BMI < 25), overweight (BMI 25–29.99), class I obesity (BMI 30 – 34.99), class II obesity (BMI 35 – 39.99), and class III obesity (BMI ≥40). Child Pugh class was also derived at time of each ultrasound. Ascites and hepatic encephalopathy were classified as none, mild or controlled, and severe or uncontrolled. Laboratory indices at each ultrasound (+/− 3 months) included platelet count, creatinine, AST, ALT, bilirubin, albumin, and INR. For patients with multiple laboratory results, results closest to the ultrasound exam were recorded.
For each ultrasound exam, we recorded patient location (inpatient, outpatient, or emergency department) and exam intent (screening, diagnostic, other, or unknown). Indications such as “screening”, “rule out HCC”, and “cirrhosis” were classified as screening. Ultrasounds performed for indications such as abdominal pain, swelling, or elevated liver enzymes were recorded as diagnostic exams. Ultrasounds performed to monitor small liver lesions were classified as “other”. Ultrasound LI-RADS reporting recommendations and structured radiology reports were used during the study period at both institutions, permitting collection of visualization scores and imaging findings from the original radiologist’s clinical interpretation.14 The LI-RADS ultrasound visualization score is a standardized and validated score that is recommended by the American College of Radiology (ACR). It is a qualitative evaluation that is categorized by the interpreting radiologist as: A – no or minimal limitations, B – moderate limitations (e.g., moderate parenchymal heterogeneity or beam attenuation), or C – severe limitations (e.g., severe heterogeneity or beam attenuation, >50% liver not visualized). We recorded reasons for visualization limitations when provided in the radiologist’s report. Finally, we recorded the presence and size of any suspicious liver masses. For ultrasound examinations with a liver lesion, diagnostic CT or MRI imaging was reviewed, if available, and observations were categorized using the CT/MRI LI-RADS algorithm, ranging from LR-1 (definitely benign) to LR-5 (definitely HCC).
Statistical Analysis
Our primary outcome was ultrasound visualization score on baseline ultrasound exam, as adequate visualization is necessary to exclude HCC during screening. We used univariable and multivariable logistic regression analyses to identify patient-level factors associated with limited ultrasound visualization. Multivariable analysis included variables of a priori importance (e.g., Child Pugh and obesity) and those with p<0.05 in univariable analyses. Our primary analysis evaluated visualization as a dichotomous outcome: good visualization (score A) vs. limited visualization (B or C). We performed subgroup analyses evaluating correlates of ultrasound visualization among outpatients, including those ultrasounds performed using a liver-dedicated imaging protocol and those done for screening purposes. As some providers believe a visualization score of B may be adequate for HCC screening, we also performed a sensitivity analysis to identify correlates of patients with severely limited visualization (score C). In secondary analyses, we performed univariable and multivariable logistic regression to identify correlates of change in ultrasound visualization over time, considering visualization as a categorical outcome (better vs. no change vs. worse visualization). Statistical significance was defined as p<0.05 for multivariable analysis as well. All statistical analyses were performed using Stata 14.0 (StataCorp, College Station TX).
RESULTS
Patient Characteristics
Characteristics of 2,053 patients with cirrhosis with at least one ultrasound exam are detailed in Table 1. Median age of the cohort was 59.5 (IQR 53.6 – 65.0) years, and 58.1% were male. The cohort was racially/ethnically diverse, with 31.0% being non-Hispanic White, 39.7% Hispanic White, and 27.6% Black. The most common liver disease etiologies were hepatitis C (48.4%), alcohol-related (22.2%), and NAFLD (18.0%). Most patients had compensated cirrhosis (71.9% Child Pugh class A), with less than one-fifth having ascites (19.5%) or hepatic encephalopathy (15.5%). Median BMI was 29.3 (IQR 25.6 – 34.4), with 32.7% of patients classified as overweight and 23.3%, 11.9%, and 10.8% having class I, II, or III obesity, respectively. Nearly three-fourths (74.4%) of ultrasound exams were performed in the outpatient setting, 79.2% were performed using a liver-dedicated protocol, and nearly two-thirds (65.2%) had screening as the primary indication.
Table 1.
Patient characteristics, stratified by LI-RADS visualization score on baseline exam
| LI-RADS Vis A | LI-RADS Vis B | LI-RADS Vis C | |
|---|---|---|---|
| (n=1685) | (n=262) | (n=106) | |
| Age | 59.7 (53.8 – 65.1) | 58.2 (52.0 – 65.0) | 58.1 (49.9 – 63.8) |
| Sex (% male) | 983 (58.3%) | 151 (57.6%) | 59 (55.7) |
| Race | |||
| White | 500 (29.7) | 97 (37.0) | 38 (35.9) |
| Black | 513 (30.5) | 43 (16.4) | 9 (8.5) |
| Hispanic | 642 (38.1) | 116 (44.3) | 56 (52.8) |
| Other | 28 (1.7) | 6 (2.3) | 3 (2.8) |
| Exam location | |||
| Outpatient | 1303 (77.3) | 157 (59.9) | 68 (64.2) |
| Inpatient | 322 (19.1) | 97 (37.0) | 36 (34.0) |
| Emergency department | 60 (3.6) | 8 (3.1) | 2 (1.9) |
| Type of exam protocol | |||
| Liver-dedicated | 1335 (79.3) | 207 (79.0) | 83 (78.3) |
| Right upper quadrant | 238 (14.1) | 39 (14.9) | 13 (12.3) |
| Complete | 111 (6.6) | 16 (6.1) | 10 (9.4) |
| Exam indication | |||
| Screening | 1127 (66.9) | 148 (56.5) | 64 (60.4) |
| Diagnostic | 558 (33.1) | 114 (43.5) | 42 (39.6) |
| Cirrhosis etiology | |||
| Hepatitis C virus | 876 (52.0) | 93 (35.5) | 25 (23.6) |
| Hepatitis B virus | 58 (3.4) | 2 (0.8) | 3 (2.8) |
| Alcohol-related | 351 (20.8) | 71 (27.1) | 34 (32.1) |
| NAFLD | 262 (15.6) | 72 (27.5) | 35 (33.0) |
| Other | 138 (8.2) | 24 (9.2) | 9 (8.5) |
| Child Pugh Class | |||
| A | 1255 (74.5) | 159 (60.7) | 62 (58.5) |
| B | 379 (22.5) | 83 (31.7) | 35 (33.0) |
| C | 51 (3.0) | 20 (7.6) | 9 (8.5) |
| Ascites | |||
| None | 1372 (81.4) | 195 (74.4) | 85 (81.7) |
| Mild | 272 (16.1) | 60 (22.9) | 14 (13.5) |
| Moderate-severe | 41 (2.4) | 7 (2.7) | 5 (4.8) |
| Encephalopathy | |||
| None | 1450 (86.5) | 197 (76.1) | 78 (73.6) |
| Mild | 223 (13.3) | 58 (22.4) | 27 (25.5) |
| Moderate-severe | 4 (0.2) | 4 (1.5) | 1 (0.9) |
| Body mass index | 28.9 (25.3 – 33.4) | 31.5 (26.4 – 37.5) | 37.0 (30.7 – 43.2) |
| AST | 47 (31 – 80) | 53 (36 – 85) | 45 (34 – 87) |
| ALT | 37 (24 – 64) | 37 (25 – 58) | 35 (22 – 50) |
| Albumin | 3.8 (3.3 – 4.2) | 3.6 (3.0 – 4.0) | 3.6 (3.0 – 4.0) |
| Bilirubin | 0.8 (0.5 – 1.4) | 0.9 (0.6 – 1.8) | 0.9 (0.6 – 1.8) |
| INR | 1.2 (1.1 – 1.3) | 1.2 (1.1 – 1.4) | 1.3 (1.1 – 1.5) |
| Platelet count | 134 (88 – 190) | 128 (75 – 172) | 120 (87 – 167) |
| Creatinine | 0.9 (0.7 – 1.1) | 0.8 (0.6 – 1.0) | 0.8 (0.6 – 1.1) |
Continuous data reported as median (IQR) and categorical data as n(%)
Correlates of Ultrasound Quality on Baseline Examination
Among 2053 patients with at least one ultrasound, 368 (17.9%) had limited visualization at baseline, with 262 having moderate limitations (score B) and 106 having severe limitations (score C) (Table 1, Figure 1). The most common reasons for limited visualization were heterogeneous echotexture and body habitus. In multivariable analysis, limited visualization was associated with inpatient location (OR 2.03, 95%CI 1.43 – 2.91), alcohol-related (OR 1.74, 1.25 – 2.43) or NAFLD (2.13, 95%CI 1.51 – 3.00) cirrhosis, Child Pugh C cirrhosis (2.11, 95%CI 1.21 – 3.67), and obesity class II (OR 1.69, 95%CI 1.06 – 2.67) or class III (OR 4.35, 95%CI 2.82 – 6.71) (Table 2). Whereas limited visualization was observed in less than 15% of patients with normal weight or overweight status, it was noted in 17.2%, 21.3%, and 40.6% of patients with obesity classes I, II, and III, respectively. Similarly, limited visualization was only observed in 11.9% of patients with hepatitis C-related cirrhosis, compared to 23.0% of those with alcohol-related cirrhosis and 29.0% with NAFLD cirrhosis.
Figure 1 –

Changes in LI-RADS visualization score between baseline and repeat ultrasound exam
Of 1546 patients with more than one ultrasound during the study period, 27% had better or worse visualization on subsequent exam. Of those with good visualization (score A) at baseline, 19.6% had worse visualization on subsequent exam. Among patients with visualization score B at baseline, 58% had improved visualization and 6.5% had worse visualization on the subsequent exam. Among those with severely limited visualization (score C) at baseline, 69.7% had improved visualization on repeat exam.
Table 2.
Correlates of limited visualization (LI-RADS score B/C) on baseline ultrasound exam
| Univariable | Multivariable | |
|---|---|---|
| (OR, 95% CI) | (OR, 95% CI) | |
| Sex (% male) | 0.95 (0.76 – 1.19) | |
| Race | ||
| White | Reference | |
| Black | 0.38 (0.27 – 0.53) | |
| Hispanic | 0.99 (0.77 – 1.28) | |
| Other | 1.19 (0.55 – 2.58) | |
| Exam Location | ||
| Outpatient | Reference | Reference |
| Inpatient | 2.40 (1.87 – 3.06) | 2.03 (1.43 – 2.91) |
| Emergency department | 0.97 (0.49 – 1.91) | 0.91 (0.43 – 1.90) |
| Type of exam protocol | ||
| Liver-dedicated | Reference | |
| Right upper quadrant | 1.08 (0.69 – 1.68) | |
| Complete | 1.01 (0.73 – 1.39) | |
| Diagnostic indication | 1.49 (1.18 – 1.87) | 0.94 (0.69 – 1.29) |
| Cirrhosis etiology | ||
| Hepatitis C virus | Reference | Reference |
| Hepatitis B virus | 0.64 (0.25 – 1.63) | 0.73 (0.28 – 1.89) |
| Alcohol-related | 2.22 (1.66 – 2.97) | 1.74 (1.25 – 2.43) |
| NAFLD | 3.03 (2.26 – 4.08) | 2.13 (1.51 – 3.00) |
| Other | 1.78 (1.16 – 2.72) | 1.85 (1.15 – 2.96) |
| Child Pugh Class | ||
| A | Reference | Reference |
| B | 1.77 (1.38 – 2.27) | 1.29 (0.95 – 1.76) |
| C | 3.23 (2.00 – 5.21) | 2.11 (1.21 – 3.67) |
| Presence of ascites | 1.34 (1.03 – 1.77) | Collinear |
| Presence of encephalopathy | 2.09 (1.59 – 2.76) | Collinear |
| Body mass index | ||
| Normal | Reference | Reference |
| Overweight | 0.84 (0.59 – 1.21) | 0.84 (0.57 – 1.24) |
| Obesity class I | 1.31 (0.91 – 1.88) | 1.35 (0.91 – 2.00) |
| Obesity class II | 1.70 (1.12 – 2.57) | 1.69 (1.06 – 2.67) |
| Obesity class III | 4.30 (2.93 – 6.33) | 4.35 (2.82 – 6.71) |
| Platelet count <150 × 109/L | 1.36 (1.06 – 1.75) | 1.18 (0.91 – 1.55) |
Results were consistent in subgroup analyses among outpatients, including those who underwent ultrasound using a liver-dedicated protocol and those who had an ultrasound for screening indication. In brief, limited visualization was observed in patients with NAFLD cirrhosis and those with obesity classes I–III (Supplemental Table 1). A sensitivity analysis examining severe visualization limitations continued to identify liver disease etiology and obesity as the most important correlates. Specifically, severely limited visualization was associated with alcohol-related (OR 2.69, 95%CI 1.50 – 4.83) or NAFLD (2.54, 95%CI 1.44 – 4.49) cirrhosis and obesity class II (OR 4.57, 95%CI 2.06 – 10.1) or class III (OR 9.00, 95%CI 4.21 – 19.2) status in multivariable analysis (Supplemental Table 2).
Ultrasound Findings
A liver lesion measuring ≥1 cm in diameter on baseline ultrasound was observed in 95 patients, of whom 64 had a diagnostic CT or MRI within 6 months. Upon diagnostic evaluation, 3 patients had LR-5 observations (definite HCC), 27 LR-3 or LR-4 observations (intermediate probability and probable HCC, respectively), and 16 had LR-1 or LR-2 observations (definitely or likely benign, respectively), whereas 18 did not have any lesion (i.e., false positive). Patients with good ultrasound visualization (score A) were more likely to have a liver lesion (any LR score) observed on CT or MRI (2.3% vs. 1.9%, p=0.63) and less likely to have a false positive result (25.0% vs. 41.7%, p=0.25), although neither reached statistical significance given limited sample size.
Changes in Ultrasound Quality on Repeat Exams
We identified 1,546 patients with more than one ultrasound during the study period; visualization remained the same for 1,129 (73%), with 1,046 exams consistently graded as score A, 60 as score B, and 23 as score C. Among the 1301 patients with good visualization (score A) at baseline, 17.7% (n=230) had the subsequent exam scored as B, and 1.9% (n=25) as C (Figure 1). Among the 169 patients with visualization score B at baseline, 58% (n=98) had improved visualization on subsequent ultrasound, while 6.5% (n=11) had worse visualization. Among the 76 patients with severely limited visualization (score C) at baseline, most showed improvement on subsequent exams, with 42.1% (n=32) improving to score A and 27.6% (n=21) to score B.
Results were consistent in the subgroup of 1,189 outpatients with more than one ultrasound, with visualization remaining the same in 75.8% of patients. Similarly, visualization worsened on repeat exam in 18.7% of those with good visualization at baseline and improved in 57.1% of those with impaired baseline visualization.
In multivariable analysis, worse ultrasound visualization on repeat exam was associated with inpatient status (OR 1.56, 95%CI 1.09 – 2.25), alcohol-related (OR 1.88, 95%CI 1.38 – 2.55) and NAFLD (OR 1.65, 95%CI 1.65 – 2.28) cirrhosis, and obesity class I (OR 1.47, 95%CI 1.02 – 2.13), class II (OR 1.87, 95%CI 1.23 – 2.86), and class III (OR 2.91, 95%CI 1.88 – 4.50) (Table 3). Correlates of worse ultrasound visualization on repeat exam were consistent among the subgroup of patients with good baseline visualization (Supplemental Table 3).
Table 3.
Predictors for worse LI-RADS visualization score between baseline and repeat ultrasound exams*
| Univariable | Multivariable | |
|---|---|---|
| (OR, 95% CI) | (OR, 95% CI) | |
| Exam location | ||
| Outpatient | Reference | Reference |
| Inpatient | 2.02 (1.55 – 2.63) | 1.56 (1.09 – 2.25) |
| Emergency department | 0.67 (0.32 – 1.39) | 0.58 (0.26 – 1.27) |
| Type of exam protocol | ||
| Liver-dedicated | Reference | |
| Right upper quad | 1.48 (0.96 – 2.29) | |
| Complete | 1.25 (0.91 – 1.71) | |
| Diagnostic indication | 1.39 (1.10 – 1.76) | 1.10 (0.82 – 1.48) |
| Cirrhosis etiology | ||
| Hepatitis C virus | Reference | Reference |
| Hepatitis B virus | 0.86 (0.42 – 1.75) | 0.98 (0.47 – 2.02) |
| Alcohol-related | 2.19 (1.65 – 2.90) | 1.88 (1.38 – 2.55) |
| NAFLD | 2.13 (1.57 – 2.88) | 1.65 (1.19 – 2.28) |
| Other | 1.08 (0.68 – 1.73) | 1.11 (0.69 – 1.81) |
| Child Pugh Class | ||
| A | Reference | Reference |
| B | 1.69 (1.31 – 2.18) | 1.30 (0.97 – 1.76) |
| C | 2.30 (1.33 – 4.01) | 1.48 (0.80 – 2.75) |
| Body mass index | ||
| Normal | Reference | Reference |
| Overweight | 0.93 (0.66 – 1.31) | 0.90 (0.63 – 1.29) |
| Obesity class I | 1.47 (1.03 – 2.09) | 1.47 (1.02 – 2.13) |
| Obesity class II | 1.88 (1.26 – 2.81) | 1.87 (1.23 – 2.86) |
| Obesity class III | 3.13 (2.07 – 4.73) | 2.91 (1.88 – 4.50) |
| Platelet count <150 × 109/L | 1.17 (0.92 – 1.50) |
Multivariable logistic regression comparing any better vs. no change vs. any worse; predictors are for worse visualization on repeat ultrasound imaging
Changes in ultrasound visualization over four rounds of screening is illustrated in Supplemental Table 4. Overall, changes in visualization in subsequent rounds was consistent with changes observed between the first two screening rounds.
DISCUSSION
We found one in five ultrasound exams performed in patients with cirrhosis have limited visualization and potential inadequate quality for HCC screening. Limited ultrasound visualization was more likely in patients with obesity, Child Pugh class B or C cirrhosis, and NAFLD or alcohol-related cirrhosis – populations growing in prevalence.2 Notably, ultrasound visualization can change over time, with 20% of patients with good baseline visualization having moderate to severe limitations on repeat exam and half of those with limited baseline visualization showing improvement on repeat exams.
Despite ultrasound-based HCC screening being recommended by international guidelines, most patients with HCC are diagnosed beyond an early stage. Late stage HCC detection can be partly attributed to screening underuse,15 although studies have also highlighted the low sensitivity of ultrasound for early-stage HCC detection.8 Our study results highlight that one in five patients with cirrhosis may be at higher risk of failure of ultrasound-based screening given suboptimal visualization. This finding is consistent with prior studies that found ultrasound has suboptimal quality in approximately 20% of patients at risk of HCC.9,16 These data are concerning in light of poor visualization being associated with higher risk of false negative results and non-diagnostic ultrasound guided biopsies.17,18 In a recent modeling study, increasing screening sensitivity had the largest effect in increasing overall survival.19 Further, poor ultrasound visualization may be associated with higher risk of indeterminate or false positive exams, increasing the risk of screening-related harms.20–22
Patients at risk for HCC represent a heterogeneous population in which ultrasound-based screening can have varying effectiveness. Tailored screening strategies that consider individual factors affecting the accuracy of screening tools and HCC risk are likely to be superior to a “one-size-fits-all” approach.23 In our study, poor ultrasound visualization was associated with obesity class II or III and NAFLD or alcohol-related cirrhosis. Prior studies have also suggested these factors affect ultrasound quality and increase risk of screening failure.9,16,24 Alcohol-related liver disease remains one of the leading causes of cirrhosis and its incidence is increasing in the U.S. Similarly, NAFLD has become the fastest growing cause of cirrhosis, and NAFLD-related HCC is expected to increase in nearly every country over the next decade.2,25,26 The increasing disease burden of populations at risk for failure of ultrasound-based screening underscores the importance of identifying better screening strategies.
Similar issues have been observed in other cancer screening programs such as breast cancer. Increased breast density decreases sensitivity for early cancer diagnosis, is associated with increased false positive results, and is associated with higher mortality.27,28 In this population, abbreviated MRI has higher cancer detection rate, and addition of molecular breast imaging can increase sensitivity versus mammography alone.29,30 Similar to this, patients with poor quality ultrasound might benefit from alternative screening strategies. Abbreviated MRI and novel biomarkers (e.g., methylated DNA and GALAD) have shown promising results for early-stage HCC detection.31–34 However, alternative modalities have their own disadvantages including cost, limited availability, and radiation or contrast exposure. Additionally, some factors limiting ultrasound quality (i.e., hepatic steatosis and inflammation) can also impact the quality of other cross-sectional imaging techniques such as MRI and CT and the accuracy of serum biomarkers.35,36
As screening occurs at repeated time intervals, considering changes in ultrasound quality and visualization is important when evaluating alternative strategies. To the best of our knowledge, our study is one of the first to demonstrate how LI-RADS visualization scores change over time. Our findings suggest it may be reasonable to simply repeat short-interval ultrasound-based screening in some patients despite an initial study with poor visualization score, as half of low-scoring ultrasound examinations had optimal quality on repeated exams. This may be particularly true for patients whose index ultrasound was performed as an inpatient. However, patients with obesity, alcohol-related liver disease, or NAFLD cirrhosis have increased odds of continued poor visualization and CT or MR imaging may be considered in those subgroups. On the other hand, approximately 20% of those with good visualization at baseline can have worse visualization when repeated – with higher risk of worsening ultrasound quality in patients with obesity and non-viral cirrhosis. These patients may similarly require consideration of alternative screening strategies.
Our study has limitations. This was a dual center study performed in tertiary centers, and our results may not generalize to all clinical settings. This is particularly true for health systems outside the United States, in which ultrasounds are often performed by physicians instead of radiology technologists. Second, ultrasound visualization scores were obtained retrospectively from clinical radiology reports, and images were not independently reviewed for purposes of this study. However, we believe our findings reflect reporting of ultrasound quality in a “real world” setting and helps inform anticipated results in clinical practice. Third, although we used ultrasound LI-RADS visualization scores as a standardized measure of ultrasound visualization, interrater reliability for LI-RADS visualization scores is unknown.37 However, we found a similar proportion with suboptimal visualization in a prior study where images were independently evaluated by radiologists, prior to implementation of LI-RADS scores. Fourth, there are limitations inherent to the retrospective nature of this study, including the presence of unmeasured cofounders. Finally, while poor ultrasound visualization scores have been associated with screening failure, the correlation is imperfect.
In summary, we found that one in five patients with cirrhosis undergoing ultrasound exams have suboptimal liver visualization, with higher risk among those with alcohol- or NAFLD- cirrhosis and those with class II or III obesity. However, exam quality was noted to change upon repeated ultrasound-based screening exams, with improvement in one-half of patients and worsening in one-fifth of patients, among patients in whom ultrasound-based screening was repeated. These data help identify patients at risk for ultrasound-based screening failure and the need for highly accurate alternative screening strategies.
Supplementary Material
What You Need to Know.
Background:
While ultrasound is the most cost-effective modality for screening, it misses over one-third of early-stage hepatocellular carcinoma (HCC). Prior studies have suggested this may be due to inadequate visualization in some patients.
Findings:
One in five patients with cirrhosis have suboptimal ultrasound visualization, with higher risk among those with alcohol- or NAFLD- cirrhosis and those with class II or III obesity. Notably, ultrasound exam quality changed in many patients if repeated.
Implications:
These data help identify patients in whom ultrasound is likely of sufficient quality for HCC screening as well as those at risk for failure of ultrasound-based screening who may benefit from alternative screening strategies.
Financial Source:
This study was conducted with support from NIH U01 CA230694, R01 CA222900, and R01 CA212008. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding agencies had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation of the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest:
Amit Singal has served as a consultant or on advisory boards for Bayer, Wako Diagnostics, Exact Sciences, Roche, Glycotest, and GRAIL.
Jorge Marrero has served as a consultant for Glycotest.
Neehar Parikh has served as a consultant or on advisory boards for Bayer, Wako Diagnostics, Exact Sciences, Glycotest, and Freenome.
None of the other authors have any relevant conflicts of interest.
REFERENCES
- 1.Altekruse SF, McGlynn KA, Reichman ME. Hepatocellular carcinoma incidence, mortality, and survival trends in the United States from 1975 to 2005. J Clin Oncol. 2009;27(9):1485–1491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Moon AM, Singal AG, Tapper EB. Contemporary Epidemiology of Chronic Liver Disease and Cirrhosis. Clin Gastroenterol Hepatol. 2020;18(12):2650–2666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Singal AG, Pillai A, Tiro J. Early Detection, Curative Treatment, and Survival Rates for Hepatocellular Carcinoma Surveillance in Patients with Cirrhosis: A Meta-analysis. PLoS Med. 2014;11(4):1–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Choi DT, Kum HC, Park S, et al. Hepatocellular Carcinoma Screening Is Associated With Increased Survival of Patients With Cirrhosis. Clin Gastroenterol Hepatol. 2019;17(5):976–987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Marrero JA, Kulik LM, Sirlin CB, et al. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology. 2018;68(2):723–750. [DOI] [PubMed] [Google Scholar]
- 6.Galle PR, Forner A, Llovet JM, et al. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J Hepatol. 2018;69(1):182–236. [DOI] [PubMed] [Google Scholar]
- 7.Singal AG, Nehra M, Adams-Huet B, Yopp AC, Tiro JA, Marrero JA. Detection of Hepatocellular Carcinoma at Advanced Stages Among Patients in the HALT-C Trial: Where Did Surveillance Fail? Am J Gastroenterol. 2013;108(3):425–432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tzartzeva K, Obi J, Rich NE, et al. Surveillance Imaging and Alpha Fetoprotein for Early Detection of Hepatocellular Carcinoma in Patients With Cirrhosis: A Meta-analysis. Gastroenterology. 2018;154(6):1706–1718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Simmons O, Fetzer DT, Yokoo T, et al. Predictors of adequate ultrasound quality for hepatocellular carcinoma surveillance in patients with cirrhosis. Aliment Pharmacol Ther. 2017;45(1):169–177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Joshi K, Mendler M, Gish R, et al. Hepatocellular Carcinoma Surveillance: A National Survey of Current Practices in the USA. Dig Dis Sci. 2014;59(12):3073–3077. [DOI] [PubMed] [Google Scholar]
- 11.Kanwal F, Singal AG. Surveillance for Hepatocellular Carcinoma: Current Best Practice and Future Direction. Gastroenterology. 2019;157(1):54–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Nehra MS, Ma Y, Clark C, Amarasingham R, Rockey DC, Singal AG. Use of administrative claims data for identifying patients with cirrhosis. J Clin Gastroenterol. 2013;47(5):e50–e54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Singal AG, Tiro JA, Murphy CC, et al. Patient-Reported Barriers are Associated With Receipt of Hepatocellular Carcinoma Surveillance in a Multicenter Cohort of Patients With Cirrhosis. Clin Gastroenterol Hepatol. 2020;(August):1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tang A, Singal AG, Mitchell DG, et al. Introduction to the Liver Imaging Reporting and Data System for Hepatocellular Carcinoma. Clin Gastroenterol Hepatol. 2019;17(7):1228–1238. [DOI] [PubMed] [Google Scholar]
- 15.Wolf E, Rich NE, Marrero JA, Parikh N, Singal AG. Utilization of hepatocellular carcinoma surveillance in patients with cirrhosis: A systematic review and meta-analysis. Hepatology. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Del Poggio P, Olmi S, Ciccarese F, et al. Factors That Affect Efficacy of Ultrasound Surveillance for Early Stage Hepatocellular Carcinoma in Patients With Cirrhosis. Clin Gastroenterol Hepatol. 2014;12(11):1927–1933. [DOI] [PubMed] [Google Scholar]
- 17.Son JH, Choi SH, Kim SY, et al. Validation of US liver imaging reporting and data system version 2017 in patients at high risk for hepatocellular carcinoma. Radiology. 2019;292(2):390–397. [DOI] [PubMed] [Google Scholar]
- 18.Kang JH, Choi SH, Kim SY, et al. Ultrasound LI-RADS visualization score: diagnostic outcome of ultrasound-guided focal hepatic lesion biopsy in patients at risk of hepatocellular carcinoma. Ultrasonography. 2020:1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mourad A, Deuffic-Burban S, Ganne-Carrié N, et al. Hepatocellular carcinoma screening in patients with compensated hepatitis C virus (HCV)-related cirrhosis aware of their HCV status improves survival: A modeling approach. Hepatology. 2014;59(4):1471–1481. [DOI] [PubMed] [Google Scholar]
- 20.Atiq O, Tiro J, Yopp AC, et al. An assessment of benefits and harms of hepatocellular carcinoma surveillance in patients with cirrhosis. Hepatology. 2017;65(4):1196–1205. doi: 10.1002/hep.28895 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Parikh ND, Singal AG, Hutton DW, Tapper EB. Cost-Effectiveness of Hepatocellular Carcinoma Surveillance: An Assessment of Benefits and Harms. Am J Gastroenterol. 2020;115(10):1642–1649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Konerman MA, Verma A, Zhao B, Singal AG, Lok AS, Parikh ND. Frequency and Outcomes of Abnormal Imaging in Patients With Cirrhosis Enrolled in a Hepatocellular Carcinoma Surveillance Program. Liver Transplant. 2019;25(3):369–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Goossens N, Singal AG, King LY, et al. Cost-Effectiveness of Risk Score–Stratified Hepatocellular Carcinoma Screening in Patients with Cirrhosis. Clin Transl Gastroenterol. 2017;8(6):e101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Samoylova ML, Mehta N, Roberts JP, Yao FY. Predictors of Ultrasound Failure to Detect Hepatocellular Carcinoma. Liver Transplant. 2018;24(9):1171–1177. [DOI] [PubMed] [Google Scholar]
- 25.Estes C, Razavi H, Loomba R, Younossi Z, Sanyal AJ. Modeling the epidemic of nonalcoholic fatty liver disease demonstrates an exponential increase in burden of disease. Hepatology. 2018;67(1):123–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Julien J, Ayer T, Bethea ED, Tapper EB, Chhatwal J. Projected prevalence and mortality associated with alcohol-related liver disease in the USA, 2019–40: a modelling study. Lancet Public Heal. 2020;5(6):e316–e323. [DOI] [PubMed] [Google Scholar]
- 27.Kerlikowske K, Zhu W, Hubbard RA, et al. Outcomes of screening mammography by frequency, breast density, and postmenopausal hormone therapy. JAMA Intern Med. 2013;173(9):807–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Mandelson MT, Oestreicher N, Porter PL, et al. Breast density as a predictor of breast cancer risk. J Natl Cancer Inst. 2000;92(13):1081–1087. [DOI] [PubMed] [Google Scholar]
- 29.Hruska CB, Conners AL, Jones KN, et al. Diagnostic Workup and Costs of a Single Supplemental Molecular Breast Imaging Screen of Mammographically Dense Breasts. Am J Roentgenol. 2015;204(6):134501353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Comstock CE, Gatsonis Co, Newstead G, et al. Comparison of Abbreviated Breast MRI vs Digital Breast Tomosynthesis for Breast Cancer Detection Among Women with Dense Breasts Undergoing Screening. JAMA. 2020;323(8):746–756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Besa C, Lewis S, Pandharipande PV., et al. Hepatocellular carcinoma detection: diagnostic performance of a simulated abbreviated MRI protocol combining diffusion-weighted and T1-weighted imaging at the delayed phase post gadoxetic acid. Abdom Radiol. 2017;42(1):179–190. [DOI] [PubMed] [Google Scholar]
- 32.Khatri G, Pedrosa I, Ananthakrishnan L, et al. Abbreviated-protocol screening MRI vs. complete-protocol diagnostic MRI for detection of hepatocellular carcinoma in patients with cirrhosis: An equivalence study using LI-RADS v2018. J Magn Reson Imaging. 2020;51(2):415–425. [DOI] [PubMed] [Google Scholar]
- 33.Berhane S, Toyoda H, Tada T, et al. Role of the GALAD and BALAD-2 Serologic Models in Diagnosis of Hepatocellular Carcinoma and Prediction of Survival in Patients. Clin Gastroenterol Hepatol. 2016;14(6):875–886. [DOI] [PubMed] [Google Scholar]
- 34.Parikh ND, Mehta AS, Singal AG, Block T, Marrero JA, Lok AS. Biomarkers for the Early Detection of Hepatocellular Carcinoma. Cancer Epidemiol Biomarkers Prev. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Yacoub JH, Elsayes KM, Fowler KJ, et al. Pitfalls in liver MRI: Technical approach to avoiding misdiagnosis and improving image quality. J Magn Reson Imaging. 2019;49(1):41–58. [DOI] [PubMed] [Google Scholar]
- 36.Gopal P, Yopp AC, Waljee AK, et al. Factors that Affect Accuracy of α-Fetoprotein Test in Detection of Hepatocellular Carcinoma in Patients with Cirrhosis. Clin Gastroenterol Hepatol. 2014;12(5):870–877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Yokoo T, Singal AG, Diaz de Leon A, et al. Prevalence and clinical significance of discordant LI-RADS® observations on multiphase contrast-enhanced MRI in patients with cirrhosis. Abdom Radiol. 2020;45(1):177–187. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
