PAPILLON COURSE on THYROID ULTRASOUND

OPENING

Echogenicity is the most important characteristic of a thyroid nodule. The echogenicity itself can determine the next step of evaluation. As in most other ultrasound features, we lack a biological standard to which the echogenicity can be compared. Although we have clear rules, how to classify a nodule, the interobserver variation in the judgement of echogenicity is disturbingly high. This characteristic is very prone to technical issues and the influence of Hashimoto's thyroiditis.

In a similar way to other subchapters, we review the five most important guidelines and use their suggestions as the backbone of the discussion of the topic. The basic guidelines are that of the American Association of Clinical Endocrinologists/American College of Endocrinology/Associazione Medici Endocrinologi (AACE/ACE/AME) (1), the American College of Radiology (ACR) (2), the American Thyroid Association (ATA) (3), the European Thyroid Association (ETA) (4) and the Korean Society of Thyroid Radiology (KSTR) (5).

CATEGORIZATION OF NODULES ACCORDING TO THEIR ECHOGENICITY

We traditionally separate the lesions into five subgroups according to their echogenicities (see Table 1). These categories are the hyperechoic, the isoechoic, the minimally, the moderately and the deeply hypoechoic ones. The distinction between the different degrees of hypoechogenicity is based on the observation that deeper the hypoechogenicity, greater the risk of carcinoma. On the other hand, the hyperechoic and isoechoic nodules are treated in the same way in all protocols similarly to the minimally and moderately hypoechoic lesions. In the end, all protocols group the lesions into three categories: the iso/hyperechoic, the minimally/moderately hypoechoic and deeply hypoechoic ones. While all protocols use these three categories, two of them treat all hypoechoic lesions in the same way in their TIRADS, i.e. they use only two subgroups. (The details are

presented later in this subchapter.

On the one hand, it is clear that deeper the hypoechogenicity greater the likelihood of malignancy, and the risks are significantly different in minimally hypoechoic and deeply hypoechoic nodules. If we make distinction only between iso/hyperechogenic and hypoechogenic nodules, it would mean classifying nodules with significantly different malignancy risk in the same way. On the other hand, the interobserver agreement is only fair to moderate regarding the distinction between iso/hyperechoic and hypoechoic nodules and is even worse regarding the distinction between various degrees of hypoechogenicity.

ISSUE OF HETEROGENEOUS NODULES

The heterogeneity of the nodule always refers to the solid portion. A nodule is classified as heterogeneous if it contains both iso/hyperechoic and hypoechoic parts. (See Table 2.) A fully homogeneous nodule is the exception rather than the rule (6-7). Most nodules larger than 2 cm in diameter are composed of portions with different echogenicities.
Firstly, we need to determine the proportion of the minority pattern over which we classify a nodule as heterogeneous. Unfortunately, there is no suggestion in the basic guidelines for this. I suggest that the nodule be called heterogeneous if the proportion of the minority part exceeds 1 cm in maximal diameter. There is a consensus in the literature that the echogenicity of a nodule is determined by the dominant structure. At the same time, another problem of interpretation is that the guidelines treat those nodules differently that are predominantly echonormal but also have hypoechoic parts. The AACE/ACE-AME, the ACR and the ATA TIRADS do not mention this problem, while the Korean TIRADS handles this type of nodule as a homogeneous echonormal one. However, ETA TIRADS has a different opinion. It says that if a nodule has a hypoechoic part, it should be classified as an intermediate group lesion. ('In the case of heterogeneous echogenicity of the solid component, the presence of any hypoechoic tissue classifies the nodule as intermediate risk.') What does it mean to literally interpret what is described? Both a dominantly

hypoechoic and a dominantly echonormal should be grouped among intermediate risk category, i.e. the presence of a minority echonormal portion in a deeply hypoechoic lesion downgrades the nodule from high risk to intermediate, while the presence of a minority hypoechoic tissue in an echonormal nodule upgrades the lesion from low risk to intermediate risk. While the former sounds logical, the latter, i.e. the downgrading of deeply hypoechoic nodules, can be seen as not reflecting the real intent of the authors but an inaccurate wording. In our opinion and practice, downgrading of dominantly echonormal heterogeneous nodules results in a 5% to 10% increase in sensitivity means that every 10th to every 20th malignant nodules belong to those categories in which FNA becomes indicated on TIRADS score because of the upgrading of the nodule.

Summary.
•  A nodule is suggested to be classified as heterogeneous if the minor portion exceeds 1 cm.
•  A dominantly hypoechoic nodule should be denominated, grouped, and handled similarly to a homogeneous hypoechoic nodule.
•  A dominantly iso/hyperechoic/echonormal nodule is best to be described on the ultrasound report as it is and is proposed to be handled according to the hypoechoic part.

ISSUE OF DEFINITION OF NODULES ON THEIR ECHOGENICITY

Apparently, there is no problem in determining the echogenicity of a nodule. We have two tissues of reference, the normal thyroid and the strap muscle.

All protocols and TIRADS use the differentiation based on the echogenicity of the normal thyroid. There are basically two subtypes of nodules. Compared to the normal thyroid a nodule is either iso/hyperechoic (similarly bright or brighter than the normal thyroid) or hypoechoic (darker than the normal thyroid).
The lack of unequivocal definition of 'normal thyroid' is one of the greatest concerns in thyroid ultrasound. Does a 'normal thyroid' mean a healthy thyroid or the non-nodular tissue of the actual patient? In the former case, we should compare the echogenicity of the nodule to an imagined degree of greyness, and we used the term 'echonormal' for nodules with an echogenicity similar to the normal healthy thyroid. In the latter case, when we compare the echogenicity of the nodule to the non-nodular part of the specific patient, indeed we compare the nodule' echogenicity to a diseased parenchyma. In this case we use the term 'isoechoic' if the nodule's echogenicity is similar to the non-nodular part. There are two common situations in which the echogenicity of the 'non-nodular' thyroid and the healthy thyroid differs: the echogenicity of the thyroid is decreased in the event of autoimmune thyroiditis and it is also decreasing with age (8-10). The decreasing echogenicity with aging is in part caused by the former. This very common problem is not taken into account in the TIRADS of three professional societies. The Korean Society of Thyroid Radiologist recognizes this problem but suggests: "When the thyroid parenchyma shows abnormal hypoechogenicity in the case of thyroiditis, the nodule echogenicity should still be described relative to the same reference structures and the abnormal thyroid echogenicity should be described" (5). Only the ETA goes on (4). Unfortunately, the ETA system handles this situation inconsistently. Although ETA realizes the problem caused by the decreased echogenicity of the non-nodular part of the thyroid in Hashimoto's thyroiditis, it suggests a solution that is difficult to interpret and impossible to interpret uniformly: both the normal salivary gland and the thyroid with reduced echogenicity can serve as a reference.

(We only mention here that the echogenicity of the salivary gland is not the same in all persons.)

In addition, the coexistence of Sjögren syndrome with Hashimoto's thyroiditis is quite a common finding and the former decreases the echogenicity of the salivary gland (11). Both the negligence and the inconsistent handling of this issue have a huge impact in the event of Hashimoto's thyroiditis. As an example, let's take a look at this nodule that occurred in a Hashimoto's thyroiditis. The echogenicity of the nodule exceeds that of the non-nodular parenchyma but is almost as hypoechoic as the strap muscle fiber. If we use non-nodular part as reference than this nodule should be considered as hyperechogenic, if we use healthy thyroid as reference than this nodule is (moderately) hypoechoic.

The essence of the problem can be summarized by using the concept of echogenicity in a relative or absolute sense. When used echogenicity in a relative sense, as in most protocols, we claim that the inflammatory process, or even age, reduces the echogenicity of the nodular tissue in the same way as the non-nodular parenchyma. Do we have any evidence of this? No, we do not have any. Is it at all realistic to assume this than the opposite?
Until we did not get deeper impact into this problem, it seems better to use the echogenicity in absolute sense. It means that in hypoechoic thyroids, the nodule' echogenicity should be compared not to the diseased thyroid parenchyma of the actual patient but to the parenchyma of a healthy thyroid. Naturally, it is more difficult to compare the echogenicity of a nodule to an imagined structure than to a visualized one, but notice that this is exactly what the recommendations expect of us when we need to realize that a thyroid is hypoechoic. In other words, we have to be able to compare the echogenicity seen in that patient to a healthy thyroid parenchyma imagined from our previous experience.

As has been mentioned earlier, all protocols differentiate minimally/moderately hypoechoic nodules from deeply hypoechoic ones, and three TIRADS systems handle differently these two types of hypoechoic lesions. The reference tissue for this distinction is the strap muscle. Here's another issue which also has practical relevance. The muscle tissue as a whole is composed of three different parts with significantly different echogenicities: the connective tissue is hyperechoic, the adipose tissue is echonormal, while the muscle fiber is hypoechoic.

None of the papers have a clear statement whether the strap muscle as a whole or only the muscle fiber of the strap muscle with low or average or high adipose tissue content should be regarded as reference tissue. Although not defined in the papers, it seems very likely that the authors suggest the strap muscle fiber as reference. There is another problem which is caused by the fact that adipose tissue is within, mixed with the muscle fibers and the former has a deep impact on the echogenicity of the latter. It means that the echogenicity of muscle fiber is more echoic in an obese patient.  

Before we move on, we take a little detour. Although similarly to most ultrasound characteristics, we have no biological standard which the echogenicity can be compared to, we can measure the echogenicity of various tissues. A histogram is a graphic representation of the distribution of statistical data, and it has been recently researched as a quantitative analysis method in the imaging of various organs, including the thyroid with reliability and reproducibility (12-17).

The images of the thyroid nodules are stored as 8-bit bitmap images in the picture archiving and communication system. Most commercially available image analyzing software (e.g. DICOM Viewers, Windows Paint, Corel, Adobe Photoshop) are able to measure pixel intensity. At first, the region of interest has to be drawn manually, thereafter by applying histogram analysis, the program automatically measures the pixel intensity values (PIV) in the region of interest as 0 to 255 (0, black, 255, white), and calculates various parameters. The latter include among others the mean value and standard deviation of PIV. The mean is the average PIV within a region of interest classified by a 256-point scale.

The goal of histogram analysis is to gain exact and reproducible parameters on nodule' echogenicity instead of the subjective visual judgement of the researcher. The mean PIV of the nodule can be compared to the mean PIV of the reference tissues (normal or extranodular thyroid and the strap muscle).
Clearly, this is not a panacea and cannot replace the medical judgement. The results highly depend on the actual settings of the ultrasound equipment. Therefore, there is no sense to compare the absolute values gained by different machineries or by the same equipment but with different settings.

However, if we use the same settings in our patients, we will get a deeper inside and by measuring grey-scale value in hundreds of patients, we can get an internal standard. Essentially, we will be able to determine the exact grey-scale value of the reference tissues which a nodule echogenicity can be compared to.

As we have mentioned earlier the strap muscle is not a homogeneous tissue and is composed of three different components: the muscle fiber, the adipose tissue and connective tissue. Indeed, the first two components are mixed, which means that connective tissue fibers divide the muscle tissue into smaller parts, which differ in the ratio of adipose component to muscle fiber. In Table 3 we present the PIV of different components of a strap muscle.

The question is what is meant by 'strap muscle'? The entire muscle or all muscle fibers or a muscle fiber with low adipose tissue component?

These different tissues differ significantly in their echogenicity. See Table 3.

We propose to use the strap muscle fiber with low adipose tissue content as reference point. Otherwise, the lack or presence of obesity would influence greatly the echogenicity of this reference tissue. We mention here, that except for very rare patients, we can find strap muscle with low adipose content even in very obese patients.

There are basically two ways to interpret the normal thyroid parenchyma; this can be the non-nodular part of the case in question or can be the normal healthy thyroid. As described earlier, the distinction has great relevance in autoimmune thyroids and in elder patients. Both conditions decrease the echogenicity of the thyroid.

Apart from the above, there is another problem. The non-nodular thyroid

presents in most cases with a heterogeneous pattern. Which part of the non-nodular part should be considered? The entire non-nodular part or only the dominant part?

It seems to be evident that thick connective tissue fibers, cystic areas, coarse calcifications, and the acoustic shadow caused by a coarse calcification should be excluded from the calculation of PIV and from the visual judgement.

Until there is no consensus in the literature, we suggest the following:

•  The strap muscle as a reference tissue

It seems to be logical and reproducible to consider muscle fiber with low adipose content as a reference point.

•  The normal thyroid as reference point

Most guidelines suggest a controversial solution. They compare the diseased, hypoechoic thyroid to an imagined healthy echonormal thyroid while the nodule occurring in hypoechoic thyroids are suggested to compare to the diseased, hypoechoic thyroid. If we follow the proposal of these guideline, it is advisable to describe it as suggested by the KSTR: the nodule is hyperechoic compared to the hypoechoic non-nodular parenchyma (5). In this case, the FNA of the nodule should be considered irrespectively of the nodule' echogenicity.

The second solution which I prefer is, that in hypoechoic thyroid, not only the extranodular tissue, but also the nodule should be compared to the echogenicity of the healthy thyroid.

THE ROLE OF ECHOGENICITY IN VARIOUS TIRADS

The echogenicity is the most important sign in characterizing of thyroid nodules.

The AACE/ACE-AME, the ATA, the European and the Korean TIRADS

ACR TIRADS

THE ECHOGENICITY OF THE MOST FREQUENT THYROID CANCERS

THE INTEROBSERVER VARIATION IN THE JUDGEMENT OF THE ECHOGENICITY OF THE NODULES

TECHNICAL ISSUES

Similar to other organs, the echogenicity of the thyroid tissue shows a relatively large variability in the population. This is an underinvestigated field in thyroid studies, while for example in the case of the pancreas or liver, there are quite large number of publications that focus with the influence of body weight, age and gender on the echogenicity (57-58).
Firstly, it is worth noting that the echogenicity of a solid nodule is very stable even over decades. In most if not all of those cases in which the echogenicity of a nodule seems to have changed, the issue is the inconsistent judgement. Among tens of thousands of cases in which we performed ultrasound at different times, we have found only two completely solid nodules, where echogenicity indeed has been changed over time.
Secondly, the anatomical situation has a deep impact on the performance of thyroid US. We have to fit the ultrasound probe all along to the neck of the patient. If it is impossible to fit adequately the probe, the echogenicity will decrease. This occurs in short necked patients or in those who have anatomical deformity, and in nodules which protrude ventrally. In such

cases we cannot visualize the entire nodule or lobe properly, so the nodule must be analyzed in more details.

Another issue might arise in mixed, cystic nodules. It is a well-known phenomenon that the echogenicity gets amplified dorsal to the fluid. This has impact on the judgement of nodule' echogenicity in two cases. The first is in mixed nodules where the solid part is located dorsal to the fluid, and also in multinodular goiter if a nodule is found dorsal to a cystic lesion. The echogenicity of the nodule can change after the aspiration of cystic fluid which not infrequently changes the categorization of the nodule in the TIRADS system. The guideline does not mention the issue caused by the amplification of the fluid. It seems to be reasonable to classify the echogenicity of the nodule on that portion of the solid part which is not located dorsally to cystic fluid. It is also conceivable that the definition of the solid part is more realistic after than before the aspiration of the fluid. (For examples see the introduction and courses.)  

THE ISSUE OF AUTOIMMUNE THYROIDITIS