Thyroid Nodules: TI-RADS and Bethesda Answer Different Questions

A thyroid nodule can acquire two numbers that look deceptively similar. Your ultrasound report may call it ACR TI-RADS 4. A later cytology report may call it Bethesda II. The first number describes sonographic features and a management threshold. The second describes cells collected by fine-needle aspiration. They are not competing scores, and they do not run on one continuous scale.

Understanding the sequence spares you unnecessary alarm. TI-RADS helps decide which nodules merit sampling or surveillance. Bethesda helps interpret the sample. Clinical history, thyroid function, lymph nodes, symptoms, size, prior imaging, patient priorities, and local resources complete the decision.

First ask why the nodule was found#

Many nodules are discovered incidentally on imaging performed for another reason. Others are felt during examination or noticed because of neck fullness, swallowing difficulty, voice change, or thyroid dysfunction, and the route to discovery affects pretest probability and the value of further testing.

An initial evaluation reviews prior childhood head or neck radiation treatment, family history of thyroid cancer or related syndromes, rapid growth, compressive symptoms, voice change, and abnormal lymph nodes. Serum thyroid-stimulating hormone, or TSH, helps determine whether the nodule may be functioning.

When TSH is low, radionuclide imaging may identify an autonomously functioning nodule. Hyperfunctioning nodules are seldom malignant and follow a different pathway. When ultrasound is indicated, it characterizes the thyroid and cervical lymph nodes rather than simply confirming that a lump exists. Population screening of asymptomatic adults is a different question from evaluating a known nodule or symptom; the distinction matters because indiscriminate ultrasound detects many small findings that would never cause harm.

ACR TI-RADS converts five feature groups into points#

The American College of Radiology Thyroid Imaging Reporting and Data System uses a standardized lexicon. For each nodule, the reader scores five domains.

Composition distinguishes cystic or spongiform nodules from mixed and solid nodules. Echogenicity compares the nodule with surrounding thyroid and neck muscles. Shape asks whether the nodule is taller than wide on a transverse image. Margins may be smooth, ill-defined, lobulated, irregular, or show extension beyond the thyroid. Echogenic foci include comet-tail artifacts, macrocalcifications, rim calcifications, and punctate foci.

Points are summed into TR1 through TR5. TR1 is benign, TR2 not suspicious, TR3 mildly suspicious, TR4 moderately suspicious, and TR5 highly suspicious. The labels are risk categories, not pathology. The system deliberately does not score every clinical clue. Abnormal lymph nodes, symptoms, thyroid function, age, prior treatment, and family history can override a routine nodule pathway.

Size modifies the recommendation#

ACR TI-RADS uses larger size thresholds for lower-suspicion categories and smaller thresholds for higher-suspicion categories, and under the original white paper, TR3 nodules generally reach fine-needle aspiration at 2.5 cm and follow-up at 1.5 cm. TR4 reaches sampling at 1.5 cm and follow-up at 1.0 cm. TR5 reaches sampling at 1.0 cm and follow-up at 0.5 cm. TR1 and TR2 generally need neither on TI-RADS grounds.

Those thresholds are not claims that a smaller nodule is harmless. They reflect the balance between detecting clinically important cancer and avoiding biopsy and treatment of indolent disease; ACR's 2026 FAQ states that the recommendations were informed by recognition that many thyroid cancers are unlikely to cause harm during a person's lifetime.

Other systems, including American Thyroid Association sonographic patterns, use different categories and thresholds. A report should say which system it applied. If you mix the score from one system with the threshold from another, you can end up with the wrong recommendation.

Growth needs a consistent measurement rule#

Ultrasound measurements vary with probe angle, borders, cystic change, and reader technique. A small millimeter difference is not automatically biological growth. ACR TI-RADS follows established definitions of meaningful enlargement rather than treating any change as progression.

When a nodule grows enough to cross a size threshold, management can change. Growth by itself does not diagnose malignancy, and some benign nodules enlarge. New suspicious features, abnormal lymph nodes, or symptoms may matter more than volume change alone. Prior images should be compared directly when possible. A report that relies only on old text can miss differences in which nodule was measured.

Fine-needle aspiration samples cells, not the whole capsule#

Ultrasound-guided fine-needle aspiration places a thin needle into selected parts of a nodule and collects cells for microscopic review. Multiple passes can improve adequacy. The procedure does not remove the whole nodule or show every architectural relationship.

This sampling explains two limits. A focal cancer can be missed if the sampled region is not representative, and follicular adenoma and follicular carcinoma can have similar cytology because carcinoma is defined by capsular or vascular invasion, which requires examination of the resected capsule rather than loose cells. Rapid adequacy assessment may reduce nondiagnostic results where available. Blood, cyst fluid, calcification, and difficult location can still limit the sample.

Bethesda uses six cytology categories#

The 2023 third edition gives each category one preferred name:

1. Bethesda I, nondiagnostic.
2. Bethesda II, benign.
3. Bethesda III, atypia of undetermined significance.
4. Bethesda IV, follicular neoplasm.
5. Bethesda V, suspicious for malignancy.
6. Bethesda VI, malignant.

Each category carries an estimated risk of malignancy and typical management options. The third edition updated average risks and ranges using newer data and changes in tumor classification; risk varies by institution, case selection, how noninvasive follicular thyroid neoplasm with papillary-like nuclear features is counted, and which nodules proceed to surgery. Surgical series can overestimate risk because the most concerning indeterminate nodules are removed while lower-risk nodules remain under observation. Local audit is valuable.

Nondiagnostic does not mean benign#

Bethesda I means the sample did not meet adequacy criteria or could not be interpreted. It is not a reassuring negative test. Repeat ultrasound-guided sampling is common, with timing and technique adapted to the nodule.

Persistently nondiagnostic nodules require integration with ultrasound suspicion, size, symptoms, and clinical risk. A cyst that repeatedly yields fluid differs from a solid TR5 nodule with inadequate cells. Core-needle biopsy or surgery may be considered in selected circumstances.

The report should explain why the sample was nondiagnostic when possible. “No cancer seen” is not equivalent to “enough representative cells show a benign pattern.”

Bethesda III and IV are probability categories#

Bethesda III captures atypia that is more than clearly benign but insufficient for a more specific category, and the 2023 system simplifies subcategorization around nuclear versus other atypia because their risk profiles can differ.

Options may include repeat fine-needle aspiration, molecular testing, surveillance, or diagnostic surgery, so the choice depends on ultrasound pattern, nodule size, local malignancy rate, molecular test performance, symptoms, other nodules, your own preference, and whether the result would change what happens next.

Bethesda IV suggests a follicular-patterned neoplasm. Cytology cannot determine capsular or vascular invasion. Molecular testing can alter probability, but no test should be interpreted without its validation population and local pretest risk.

Molecular tests update probability rather than issue a verdict#

“Rule-out” tests seek a low residual risk when negative. “Rule-in” tests seek a high probability when positive. Sensitivity, specificity, and predictive values depend on the panel, variant, disease prevalence, and tumor classification.

A positive oncogenic alteration can carry different implications depending on the gene and variant, and a negative panel cannot exclude every cancer because not every tumor has a covered alteration and sampling remains imperfect.

Before you order a test, define the action it would trigger. Would a negative result support surveillance? Would a positive result change the extent or timing of surgery? Testing that cannot change the plan adds cost and ambiguity.

Discordance is information#

A TR5 nodule with Bethesda II cytology creates imaging-cytology discordance. Possibilities include benign suspicious morphology, sampling error, interpretive variability, or a lesion whose risk is not captured by one modality. Repeat review, repeat sampling, or another plan may be appropriate.

A TR3 nodule with Bethesda VI cytology should not be dismissed because ultrasound looked mild. Cytology and imaging have different sensitivities and targets. Discordance should trigger a review of image quality, target identity, sample adequacy, cytology, lymph nodes, and clinical context. Multidisciplinary review can be useful for high-impact decisions.

The report is a decision map#

Read your thyroid report in sequence: nodule location and size, ultrasound system, feature points, category, growth, lymph nodes, and recommendation. If it was sampled, add adequacy, Bethesda category, subtype or qualifiers, and molecular results.

Then ask which decision is still open: repeat imaging, repeat sampling, surveillance, surgery, treatment of thyroid dysfunction, or no action. The thyroid cancer overdiagnosis guide explains why less testing can sometimes protect health.

TI-RADS and Bethesda work well when each is allowed to answer its own question. The first organizes what ultrasound sees. The second organizes what sampled cells show. Clinical judgment connects them without pretending either number is certainty.

Before you act on a copied result, confirm that the score belongs to the same nodule; multinodular glands are easy to mislabel across reports, especially when locations are described differently or dimensions change order. The biopsy note, ultrasound images, and cytology requisition should identify the target consistently, because a correct Bethesda category attached to the wrong nodule is a process error that no probability table can fix. Clear laterality, pole, size, and image labeling are therefore safety information.

References#

1. ACR TI-RADS white paper
2. ACR TI-RADS FAQ
3. 2023 Bethesda system
4. American Thyroid Association thyroid nodule guideline
5. ACR Appropriateness Criteria for thyroid disease
6. NIDDK thyroid tests

For your own health, talk with your clinician.*