Diagnostic Categories

No, the categories reflect probabilities; they do not correspond to exact histological categories. For example, while all LR-1 observations are benign, not at all benign entities can be categorized LR-1. In particular, regenerative nodules and low-grade dysplastic nodules cannot be categorized LR-1 because imaging cannot definitively exclude malignant foci in such lesions. Similarly, all LR-5s are HCC, but not all HCCs can be categorized LR-5. In particular, HCCs < 10 mm, HCCs without APHE, and HCCs with atypical features (e.g., rim APHE) cannot be LR-5.

Did you find FAQ-023 helpful? Yes No

FAQ-024: What is the differential diagnosis of each diagnostic category?

Did you find FAQ-024 helpful? Yes No

FAQ-025: What is the percentage of HCC and malignancy associated with each LI-RADS category?

The percentage (with 95% confidence intervals) associated with LR-1, LR-2, LR-3, LR-4, LR-5, and LR-M is summarized below:

The above graph represents data from the literature using versions 2014 and 2017. Data using version 2018 are not yet available.

Reference List:

van der Pol CB et al.. Gastroenterology. 2019 Mar;156(4):976-986

Did you find FAQ-025 helpful? Yes No

FAQ-026: Why doesn’t LI-RADS have a category for path-proven HCC analogous to BI-RADS 6?

Biopsy plays a less important role in the workup and management of HCC in at-risk patients than it does in breast cancer. Nevertheless, such a category may be useful in some situations and a Path category is being considered for the next LI-RADS major update in 2021 or 2022.

Did you find FAQ-026 helpful? Yes No

FAQ-027: How do I determine if an observation is noncategorizable (LR-NC)?

An observation is considered noncategorizable if it cannot be categorized meaningfully because key phases were omitted or degraded, preventing assessment of one or more major features. As a direct result, reasonable categories range from those where cancer is unlikely (LR-1 or LR-2) to those where cancer is likely (LR-4, LR-5, LR-M).

Do not assign LR-NC if the range of categories can be narrowed to LR-1 vs LR-2 or to LR-4 vs LR-5 vs LR-M.

Do not assign LR-NC if categorization is challenged only by unusual imaging features or by inability to characterize ancillary features.

Did you find FAQ-027 helpful? Yes No

FAQ-028: What is the difference between LR-5 and LR-M?

LR-5 indicates 100% certainty of HCC. LR-M indicates high certainty of malignancy but the features are not specific for HCC.

Did you find FAQ-028 helpful? Yes No

FAQ-029: Why does LI-RADS have a LR-M category?

LR-M was created to categorize observations that are highly likely to be malignant but which lack imaging features specific for HCC. The differential diagnosis for such observations includes HCC with atypical features and non-HCC malignancies such as iCCA and cHCC-CCA. The distinction between these various malignancies can be important due to potential differences in prognosis and management. For example, liver transplantation is usually contraindicated in patients with iCCA or cHCC-CCA, whereas liver transplantation can be a curative treatment in patients with HCC. The presence of LR-M alerts the treating team and the patient to the possibility of a cancer other than HCC, information that may influence patient management.

Did you find FAQ-029 helpful? Yes No

FAQ-030: Does LR-M exclude HCC?

LR-M does not exclude HCC. The differential diagnosis for LR-M includes HCC with nonspecific imaging features in addition to iCCA and other malignancies. Based on current data, about 1/3 of LR-M observations are HCC, about 2/3 are non-HCC malignancies, and about 5% are benign.

Did you find FAQ-030 helpful? Yes No

FAQ-031: If the distinction between HCC, iCCA, and cHCC-CCA may be clinically important, why not just have a separate LI-RADS category for each tumor type?

There is not yet sufficient scientific evidence to develop imaging criteria and categories that differentiate these tumor types with 100% certainty. Hence, LI-RADS has a single category, M, to indicate a high likelihood of the presence of one of these types.

As mentioned elsewhere, however, there are times when imaging features allow the radiologist to narrow the list of likely tumor types. In those circumstances, communicating clearly the most likely tumor type(s) may help guide patient management even if there is less than 100% certainty.

Did you find FAQ-031 helpful? Yes No

FAQ-032: How do I categorize a mass with infiltrative appearance not meeting LR-TIV or LR-5 criteria?

LR-M. Infiltrative appearance suggests malignancy but does not provide 100% certainty for HCC. Non-HCC malignancies and some benign processes may have an infiltrative appearance. Moreover, imaging criteria to reliably differentiate infiltrative-appearing HCC from non-HCC malignancy have not been validated yet. Therefore, LR-M is the most appropriate category.

Nevertheless, since the majority of mases with infiltrative appearance in at-risk patients are HCC, consider including in your report a statement that the most likely etiology is HCC.

For example: “LR-M with infiltrative appearance, likely represents HCC that does not meet LR-5 criteria.”

Did you find FAQ-032 helpful? Yes No

FAQ-033: On page 23, LI-RADS indicates that elevated levels of circulating biomarkers such as AFP and CA 19-9 can help narrow the differential diagnosis of LR-M. What defines elevated AFP and CA 19-9 levels and how do I interpret the elevations?

There are no simple cutoffs for interpreting AFP or CA 19-9 levels. The “normal” values for these biomarkers are ≤ 10 ng/mL and ≤ 37 Units/mL, respectively, but interpretation of values above these levels can be challenging. Values depend on the etiology of the underlying liver disease and may fluctuate with changes in liver disease activity. For these reasons, low-level elevations have a low positive predictive value for HCC or CCA.

Conversely, very high values, e.g. AFP >400 ng/mL, provide high specificity but low sensitivity. Often, the trend over time is more helpful than a one-time value. For example, a rise in the AFP level from 20 to 60 ng/mL is suggestive of HCC, whereas a one-time value of 60 ng/mL is not as informative. Another challenge is the limited knowledge of how well these biomarkers differentiate cHCC-CCA from HCC or iCCA. Interpretation of biomarker elevations can also be difficult if a patients has lesions with different imaging appearances, as it is possible to have synchronous cancers, e.g. HCC in an LR-5 lesion and iCCA in an LR-M lesion. Research is needed to better understand the interpretation of AFP, CA 19-9, and other emerging biomarkers.

Despite the caveats above, some guidance is helpful and provided below.

Below is a general guide for interpreting AFP elevation

Marked elevation of AFP (≥ 200 ng/mL) à high probability of HCC
Moderate elevation of AFP (≥ 100 ng/mL) àmoderate to high probability of HCC

Below is a guide for interpreting CA 19-9 elevation:

Marked elevation of CA 19-9 (≥ 200 Units/mL) à high probability of iCCA
Moderate elevation of CA 19-9 (≥ 100 Units/mL) à moderate to high probability of iCCA

Radiologists are encouraged to interpret biomarker values in the context of multidisciplinary discussion.

Did you find FAQ-033 helpful? Yes No

FAQ-034: How do I categorize a tumor in vein?

LR-TIV. Tumor in vein indicates malignancy. Although HCC is the most common cause, other tumors can grow into vessels such as iCCA, cHCC-CCA, and rarely metastases to the liver.

Scrutinize the images for a contiguous parenchymal mass. The LI-RADS category of a contiguous mass (if present) can help you suggest the most likely etiology of the tumor in vein. While this information does not alter the category, it may influence patient management and should be reported if possible.

Examples:

"LR-TIV contiguous with LR-4 parenchymal mass, probably due to HCC."
"LR-TIV contiguous with LR-5 parenchymal mass, definitely due to HCC."
"LR-TIV contiguous with LR-M parenchymal mass, may be due to non-HCC malignancy."
"LR-TIV contiguous with infiltrative-appearing parenchymal mass, probably due to HCC.”

Did you find FAQ-034 helpful? Yes No

FAQ-035: Why did LI-RADS change LR-5V to LR-TIV?

Since non-HCC malignancies (e.g., iCCA, cHCC-CCA) can cause tumor in vein, categorizing all observations with tumor in vein as LR-5V (i.e., definitely HCC with tumor in vein) is incorrect.

Did you find FAQ-035 helpful? Yes No

FAQ-036: The LI-RADS diagnostic table includes observations without APHE or any additional major feature. I am confused. How could such an observation be visible in the first place?

Observations without APHE or any additional major feature may be visible based on ancillary features favoring malignancy, ancillary features of benignity, or other imaging features (e.g., T1 hyperintensity, T2 hypointensity, DWI hypointensity, HBP hyperintensity).

Did you find FAQ-036 helpful? Yes No

FAQ-037: Can you explain how to categorize distinctive nodules < 20 mm without associated major features or LR-M features?

A distinctive nodule < 20 mm without associated major features or LR-M features usually is categorized LR-2. Examples include siderotic nodules, T1 hyperintense nodules, T2 hypointense nodules, DWI hypointense nodules, and HBP hyperintense nodules.

If the nodule has one or more ancillary features (AFs) favoring malignancy and no ancillary features favoring benignity, it should be upgraded to LR-3. Examples include steatotic nodules, T2 hyperintense nodules, diffusion restricting nodules, and HBP hypointense nodules.

The algorithm below explains how to categorize such nodules, based on presence of AFs.

Did you find FAQ-037 helpful? Yes No

FAQ-038: Can you explain how to categorize distinctive nodules ≥ 20 mm without associated major features or LR-M features?

It depends on whether such nodules have ancillary features (AFs).

If they have no AFs or if the radiologist elects not to apply AFs, then these nodules are categorized LR-3 since they land in the second cell from left on the top row of the CT/MRI Diagnostic table:

These include T1 hyperintense nodules, T2 hypointense nodules, DWI hypointense nodules, and HBP hyperintense nodules.

If, on the other hand, the nodule has one or more AFs favoring malignancy, the nodule has no AFs favoring benignity, and the radiologist elects to apply the AF(s), then it is categorized LR-4. Examples include steatotic nodules, T2 hyperintense nodules, diffusion restricting nodules, and HBP hypointense nodules.

Did you find FAQ-038 helpful? Yes No

FAQ-039: With respect to distinctive nodules lacking major features and LR-M features, why are those measuring < 20 mm categorized LR-2 or LR-3 (if there are AFs favoring malignancy) whereas those measuring ≥ 20 mm are categorized LR-3 or LR-4 (if there are AFs favoring malignancy)?

Due to the difference in size. Nodules ≥ 20 mm are more likely to be malignant than nodules < 20 mm and so should be assigned a higher category.

Did you find FAQ-039 helpful? Yes No

FAQ-040: Why can’t ancillary features be used to upgrade to LR-5?

Ancillary features increase diagnostic confidence and modify the probability of malignancy. The features do not have sufficient specificity for HCC to allow upgrading to LR-5, however.

Did you find FAQ-040 helpful? Yes No

FAQ-041: Why did LI-RADS make ancillary features optional?

This was done to encourage more radiologists to adopt LI-RADS by reducing its complexity. As new users become familiar with LI-RADS, they can use ancillary features to further improve their practice.

Did you find FAQ-041 helpful? Yes No

FAQ-042: What should I do if some ancillary features favor malignancy and others favor benignity?

Do not change category.

Did you find FAQ-042 helpful? Yes No

FAQ-043: What if there many more ancillary features favoring malignancy and than favoring benignity?

Do not change category.

Did you find FAQ-043 helpful? Yes No

FAQ-044: Why does the tiebreaking rule choose the category reflecting lower certainty?

This maintains 100% certainty for LR-5 and LR-1. For example, if there is doubt about whether an observation is definitely or probably benign, then it cannot be considered definitely benign. The rule also helps achieve 100% positive predictive value of LR-5 for HCC. If there is uncertainty between LR-5 and LR-4 or between LR-5 and LR-M, the tiebreaking rule selects LR-4 or LR-M, respectively.

Did you find FAQ-044 helpful? Yes No

FAQ-045: Does a category of LR-3 or LR-4 exclude non-HCC malignancy?

No. LR-3 and LR-4 criteria are not specific for hepatocellular origin and so do not exclude non-HCC malignancy. Thus, a small minority of LR-3 or LR-4 observations may be non-HCC malignancies.

Did you find FAQ-045 helpful? Yes No

FAQ-046: Step 4 of the LI-RADS diagnostic algorithm is to assess if the assigned category seems reasonable and appropriate and to reevaluate if not. Can you provide an example?

One example would be a observation with rim APHE that has unequivocally decreased in size over serial exams without treatment and in which size reduction is not attributable to resorption of blood products. Although rim APHE usually prompts a categorization of LR-M, the spontaneous size reduction would be unusual for a malignant neoplasm and suggests an alternative diagnosis such as a sclerosing hemangioma. In this case, the best category would be LR-1, LR-2, or LR-3 depending on the confidence level.

Did you find FAQ-046 helpful? Yes No

FAQ-047: How do I categorize observations that land in the “diagonal” LR-4 or LR-5 cell?

Observations in this cell should be categorized LR-4 OR LR-5 as follows

Did you find FAQ-047 helpful? Yes No

FAQ-048: Regarding 10-19 mm observations with nonrim APHE and exactly one additional major feature, how do LI-RADS v2018, AASLD 2018, and OPTN categories compare?

How can we help you today?