Keywords

Introduction

Positron emission tomography (PET), combined with CT (PET/CT) and MR (PET/MR), using fluorodeoxyglucose (FDG) as radiotracer, is often used for diagnosing, staging, assessing response to therapy, and follow-up for recurrence of a variety of malignancies. Other chapters in this book outline the utility of FDG-PET in melanoma and sarcoma. While both of these can spread throughout the body, they nonetheless have some important differences—melanoma (Fig. 4.1) is almost always very intense on FDG-PET and is notorious for spreading to unusual organs, whereas sarcomas (Fig. 4.2) are variably intense. A systematic approach is needed in each on these malignancies for proper evaluation of a patient’s PET-CT scan for minimizing the chances a potential metastasis is missed.

Fig. 4.1
figure 1

MIP (maximum intensity projection) shows multiple metastases throughout the body, with individual axial fused images showing intense metastases throughout the body

Fig. 4.2
figure 2

Heterogeneous osteosarcoma in distal femur of adolescent. Note the wide variation in FDG uptake in different parts of the tumor. SUV ranges from 0.4 (in the necrotic core) to 10.2 in the mass. At the same time, due to lack of IV contrast, the CT appearance of the mass lacks the heterogeneous appearance often seen in sarcoma on contrasted CT

Systematic Approach: Looking at the Whole Body

As with chest radiographs, a variety of systems may be used, and each reader must use the method they personally find most useful. Similarly, in FDG-PET, there is disagreement on the proper windowing, the use of fused images, and the utility of the standardized uptake value (SUV), to name a few key issues. We will describe the method we use at our institution, keeping in mind that other orders of operations and methodologies may prove just as effective.

Before even opening the study, it is worthwhile to investigate the purpose of the study and narrow down the question to be answered. Is the patient being staged initially, in which case one must look first at the initial site of disease and the expected location of the draining nodes? Are we assessing response to therapy, in which case it would be useful to know how conspicuous the prior disease was and where it was located? Or are we doing investigation of a suspected recurrence, in which case the results of other imaging and non-imaging studies would be crucial to focus on one area? Or are we doing surveillance scan in which case, the entire scanned area could conceivably hide a recurrence?

It is also good to know what prior surgeries the patient has had and even if there are any other malignancies. A mildly avid axillary node may become much more concerning if near the site of recurrence or less so if the patient has been recently operated on in this area. A history of radiation therapy is also a potentially relevant fact to be aware of, as post-radiation change and scarring may persist and even be mildly FDG avid. Finally, recent chemotherapy can cause activity in bone marrow and occasionally spleen, and this should not be confused with diffuse metastatic disease.

Once the history is known, one way we start is to look at the maximum intensity projection (MIP) images (example in Fig. 4.1). These are a series of projection images obtained at a range of angles around the patient in the axial plane, which when viewed as a movie give the effect of the patient’s scan spinning around; as a result this has also been referred to as the “spinning person” view.

Apart from the always intense brain and bladder activity and, frequently, intense and sometimes heterogeneous heart activity, metastases will stand out as intense foci. For the remainder of this discussion, we will assume a gray-scale MIP display where higher intensities appear “dark” and the principal locations of metastases can be assessed. Are they in the lung, liver, bones, muscles, or other locations? The overall metastatic burden (if any) can be assessed—if there were plenty of metastases throughout the body last time, are they more or less numerous and/or more or less metabolically active now?

A word on overall intensity is in order. PET images usually have a window, where a particular set of values in the source data set (usually an SUV) is mapped to a gray scale, ranging from black to white, or to a color scale. Numbers below the lower limit appear white on gray scale (e.g., MIP), and those above the upper limit appear black. Display protocols may reverse black and white or use color scales for fused PET/CT images, but the principle remains the same. The lower limit is set to 0, indicating no activity, and negative numbers are meaningless as there cannot be negative amounts of radioactivity. Note that this is different from CT, where 0 indicates pure water and tissues with negative numbers exist and include less dense materials like fat and air. The upper limit varies depending on user preference, but is usually set to 5–6 when beginning to read. Usually the intensity increases roughly linear to SUV, but this depends on the specific program and thus depends on the software developer.

It should be noted that while majority of malignant lesions independent of their intensity are usually visualized and detectable, using this SUV-derived default display of images, there are circumstances, where the intensity of images have to be increased or decreased in order to visualize malignant lesions. The two extreme ends of the spectrum are brain metastases on one hand and small soft tissue metastases on the other hand. In order to visualize brain metastases, the intensity of the image has to be “decreased” in order to decrease physiologic brain FDG uptake and give the metastases the opportunity to “come through.” On the other hand, in order to visualize the subtle soft tissue metastases, the intensity of the image has to be “increased,” in order to increase their target-to-background ratio and conspicuity (Fig. 4.3). With the SUV-derived display, the liver, which has an SUV of 2–3, is used as a background reference. However, this reference activity is empiric and other factors such as size of a lesion and its FDG avidity at cell level, and resulting count density has to be taken into account, when assessing a lesions, particularly smaller ones, for malignancy, as detailed in Chap. 1.

Fig. 4.3
figure 3

Brain metastasis needs to be windowed with a high upper limit to be visible (note relative low uptake of normal brain), whereas muscle metastasis needs to be windowed with a lower upper limit (note relative intensity of muscle around it)

While maximum standard uptake value (SUVmax) is generally not useful for assessing benign versus malignant tumors [1], the mean SUV (SUVmean) of the liver should be about 2–3, and if it is much higher or lower, an error in dose, weight, or uptake time may be present. The liver should appear as a light to intermediate shade of gray (with the upper SUVmax window around 5–6)—an excessively dark or bright liver may suggest a window that is too narrow, in which case noise will appear bright and produce many false positives and a grainy appearance making the image difficult to read, or too bright, in which case many smaller or less avid metastases may be missed (Fig. 4.4).

Fig. 4.4
figure 4

Windowed too high (upper limit 50), too low (upper limit 2), and adequately (upper limit 6). Notice that the lesion in the right leg is difficult to see in the first case and benign variations in the tissue appear significant in the second

Using more than one window may also be wise—it is worthwhile to use a window with a higher maximum, for example, to look for brain metastases (Fig. 4.2, above). PET is not sensitive for brain metastases due to the intense physiologic uptake in the brain, which uses a lot of glucose. However, particularly large and avid metastases (and melanoma is often very avid) may be visible on PET with proper windowing. Occasionally a tumor in the bladder may be seen as well.

Also visible on the MIP immediately are large extravasations and contaminations, appearing as large FDG-avid blotches. They will often be intensely avid (far more than any actual tumor, with SUVs in the hundreds) and will be recognizable as such, but should be identified to remove any confusion. In addition, extravasated FDG is eventually consumed by macrophages and brought to lymph nodes, possibly producing false positives in the nodal drainage pattern of the extravasation, so one should be aware of extravasations when they occur. The site of injection should also be noted—a mildly avid right axillary node becomes much more worrisome than a left axillary node if the original injection site was the left antecubital fossa.

If there are no obvious metastases, it then is useful to look at the usual drainage patterns of the tumor. Melanoma tends to spread to lymph nodes before going to other organs. In general, melanomas of the upper extremities drain to the axillae (Fig. 4.5) and melanomas of the upper extremities to the inguinal region (Fig. 4.6). Occasionally, epitrochlear nodes near the elbow or popliteal nodes behind the knee may also be seen to have metastases. Melanomas of the trunk may drain to either or both axillary or inguinal sites—they tend to drain to the closest one, but nothing is certain and all four sites should be examined.

Fig. 4.5
figure 5

Arm melanoma with axillary and subpectoral metastases

Fig. 4.6
figure 6

Leg melanomas with metastasis to inguinal lymph node (arrow)

Sarcoma, conversely, tends to invade locally or spread to the lung (Fig. 4.7), so particular attention should be paid to the lung, although lymph nodes should also be examined.

Fig. 4.7
figure 7

Lung metastasis from recurrent sarcoma

Systematic Approach: Part by Part

Once the whole-body images have been examined, the patient should be examined body segment by body segment. In general, we move head to toe, as with the physical exam, using the usual quasi-radiology division into head and neck, chest, abdomen and pelvis, and lower extremities and musculoskeletal system. Some readers prefer to view using fused images, while others prefer having separate CT and PET images and looking at both. One caution with the use of fused images is that registration may be imperfect, with the head and neck (usually the last part of the scan), lower chest (Fig. 4.8) and upper abdomen (due to diaphragmatic motion), and extremities being common trouble areas. It is usually fairly obvious when misregistration has occurred; if it has, separate viewing of CT and PET images and often a look at non-attenuation-corrected images is in order [1]. We will briefly discuss important CT findings to beware of at the end of each section.

Fig. 4.8
figure 8

Misregistered clear cell sarcoma lung metastasis. The lesion as depicted on PET is substantially superior to that on CT, likely due to the patient having taken a deeper inspiration on CT than on PET. On MIP images, streaks from motion (arrows) during acquisition are visible

It is also worth warning of “satisfaction of search.” A patient may have more than one disease, and discovery of one large, prominent finding may encourage a reader to stop looking and ignore other findings that may be just as important. Finding a pulmonary nodule, for example, does not mean a subtle periaortic node may not be present as well. The “big four” target sites of the brain, lungs, liver, and bone, as well as the relevant lymph nodes, should be specifically examined in each patient.

Head and Neck

As with the MIP, it is generally worth examining the brain. While sensitivity of PET for brain metastases is low [2] (and brain MRI is much more effective), if a brain MR is not ordered due to what is felt to be a low chance of metastatic disease, the PET gives at least some chance to detect one (Fig. 4.9). In general, the brain should be viewed using a typical brain window for CT and windowed to a higher upper limit for PET, in order to bring out brain lesions that would otherwise be obscured by the intense physiologic brain activity.

Fig. 4.9
figure 9

Choroid plexus metastasis is invisible with standard windowing with max SUV 6, but windowed to max SUV 20 a hot spot is visible (arrow). MRI clearly confirms metastasis at this location

Moving on to the head and neck, there are many sites of interest. Normal structures which may display intense avidity (generally not to be confused with melanoma metastases) include Waldeyer’s ring, including the adenoids, palatine tonsil, and lingual tonsil (Fig. 4.10), as well as the tongue and vocal folds (Fig. 4.11), which will light up if the patient has talked during the uptake period. (It is difficult to get some people not to speak for an hour.) In general, physiologic processes are symmetric—systemic activation of lymphatic tissue will tend to light up both sides at once—and relatively diffuse, such that diffuse uptake in a structure is less worrying than a focal “hot spot.”

Fig. 4.10
figure 10

Uptake in Waldeyer’s ring. Given its symmetry, this is most likely benign physiologic uptake; more asymmetric uptake might suggest a focal malignancy (possibly a second primary) or metastasis, although local inflammation might also be involved

Fig. 4.11
figure 11

Symmetric vocal cord uptake is usually physiologic. If the left side is less avid, however, it can reflect a tumor in the AP window in the chest or elsewhere along the course of the left recurrent laryngeal nerve

The head is also a common site of melanoma due to its relatively high degree of sun exposure. The head and neck contains many lymph nodes in close proximity to each other. While the seven-level otolaryngology classification can be useful in describing the location of lymph nodes, it is designed for squamous cell carcinomas of the aerodigestive tract and does not include many nodes, such as mastoidal and occipital, which may be important sites of drainage for a scalp melanoma, for example. If one is not certain of the precise nomenclature of a node, describing local landmarks may be useful.

As inflammation lights up on PET just as cancer does, deciding whether a node is hot enough to call malignant is a difficult process. Some considerations include history (has the patient recently had an infection or surgery in the area, which may cause inflammation and hence a false positive?), the original site of disease (the same mildly avid jugular node is much more likely to be a malignant metastasis if a primary melanoma is on the cheek than if it is on the leg), symmetry (bilaterally symmetric metastases exist but are relatively rare), and CT appearance of the node (round, larger nodes with effacement of the hilum and in particular cystic or necrotic components are more likely to be malignant than oval, smaller nodes with a preserved hilum and without such components) [1]. Even so, judging a node benign or malignant is often a difficult judgment call.

It is also important to look at the subcutaneous tissues for nodules that may not be in typical nodal beds, but nonetheless represent sites of metastasis (Fig. 4.12). Uptake within fat with absolutely no CT correlate, if diffuse, symmetric and seen in a younger person, more likely represents brown fat, a normal variant commonly seen in colder weather. However, focal uptake with no CT correlate presents something of a conundrum. In this case we would advise to look carefully for a small nodule that may be misregistered; if none is found, directing careful attention on follow-up may be the best strategy. The thyroid may have metabolically active nodules, and these have a reasonable risk of malignancy ranging from 15% to 36% depending on the meta-analysis [3, 4]; if a hot spot is found in the thyroid, this should be brought to the clinician’s attention in the impression and a thyroid ultrasound recommended if this has not already been done.

Fig. 4.12
figure 12

Subcutaneous neck nodule. This is roughly in the location of a level 5 cervical node, but melanoma metastases can also be found in locations not corresponding to standard lymph node locations

Small muscles may light up with muscular activity and become confusing. If purely due to anxiety, they will usually be symmetric, but in the case of asymmetric activation due to muscle strain, often after surgery, they can be hard to tell from small metastases, particularly if the patient moved their head during the scan. In this case a look at the maximum intensity projection images can be useful [2].

A look at the calvarium and later face and spine is also worthwhile; many tumors send metastases to the bone, and melanoma and sarcoma are definitely no exception.

Important CT findings to be on the lookout for in the head and neck include hemorrhagic or ischemic strokes (Fig. 4.13) or large brain tumors. If an acute stroke or hemorrhage is detected, the referring physician should be called immediately and the patient held and directed to the emergency room if they are still present.

Fig. 4.13
figure 13

This is an old stroke, visible in the right frontal lobe by the lack of uptake. Unfused CT shows the low attenuation characteristic of an old stroke—new strokes will show a more subtle blurring of the gray-white matter differentiation. Note that this is much more visible on the brain window images (center) than on the general soft = tissue window images (right). In most cases, strokes are old and well-known, but correlation with old imaging is recommended

Chest

As one moves into the chest, it is normal to look first at the lungs, which occupy most of the cross-sectional area and, receiving close to 100% of the body’s blood supply, are common sites of metastasis from both melanoma and sarcoma (as well as many, many other tumors). There are different schools of thought on this, some holding that the lungs should be left for last to avoid ignoring other thoracic structures, others holding that as a high-yield structure they should be examined first to quickly identify patients who are not candidates for surgical resection.

Whatever one’s order of operation, the lungs are a major target of metastasis (Figs. 4.14 and 4.15) as well as one of the more common sources of primary cancers, and one should pay close attention to them. Generally, one scrolls up and down through the lungs, looking for nodules or masses. The normal branching of the bronchial and arterial trees produces vessels that are round in cross-section, and this is one of the most technically challenging parts of the body to read. However, normal structures have a tree-like fractal branching pattern that moves from larger to smaller as one moves peripherally, and focal swellings are suspicious. Metastases, of course, may not connect to the bronchial or arterial trees at all, at least at CT resolution.

Fig. 4.14
figure 14

Typical metastatic lung lesion—round and intense. Note that some primary lung lesions can have this appearance as well. Metastases to lungs are quite common

Fig. 4.15
figure 15

Clear cell sarcoma with multiple lung metastases. Despite the different histology, these metastases have a roughly spherical appearance, similar to melanoma metastases in the previous image

Telling whether a nodule or mass (a mass is over 3 cm) is primary or metastatic can be difficult. Metastases tend to be round in shape, whereas primary cancers are more likely spiculated (Fig. 4.16), but these rules are not absolute, and some primary tumors such as carcinoids are often round in shape. A biopsy may be required. Whatever the case, a “hot” or even “warm” nodule or mass is suspicious for malignancy. Exactly how “hot” is “hot” enough to be called malignant again requires judgment. Previous work on telling whether nodules were benign or malignant focused on SUVmax [5], but the many variations due to size, motion artifact, etc., as well as the problem of “hot” granulomas and “cold” adenocarcinomas and carcinoids, have made SUV an unreliable guide. In particular, due to issues involving volume recovery and partial volume effects [6], uptake will appear higher in larger tumors up to about 3 cm, so a lesion that grows larger without any changes in metabolism per cubic centimeter will appear to grow brighter as well.

Fig. 4.16
figure 16

Primary lung cancer, spiculated in appearance. It can sometimes be difficult to tell a second primary from a pulmonary metastasis, although primary lung cancers will tend to have a more irregular appearance

Presently, we consider any activity in a nodule under 1 cm or any activity above mediastinal blood pool in a nodule over 1 cm suspicious. In general, PET is not sensitive for detecting lesions below about 8 mm, so lesions with visible avidity that are much smaller than this are likely to be very avid indeed and hence suspicious. Similarly, larger nodules with no avidity are felt more likely to be benign, or at least a low-grade tumor [7].

It is also worth mentioning that, since infection is a common source of false positives, and the lungs are a common source of infection, pneumonia as well as inflammatory diseases like pulmonary fibrosis (Fig. 4.17) will be “hot” on PET and may be confused with malignancy if one is not careful. They will tend to have different appearances on CT, however. Scarring from radiation may be avid and has a linear appearance not seen in natural processes (Fig. 4.18). CT images for PET-CT are taken in shallow inspiration, and the appearance may be somewhat different from stand-alone chest CTs taken in deep inspiration, particularly near the bases.

Fig. 4.17
figure 17

Pulmonary fibrosis. Inflammation, like cancer, is hot on PET, and the area of uptake can often be seen to correspond to the area of fibrosis where inflammation is occurring. It can occasionally be confused with cancer

Fig. 4.18
figure 18

Radiation fibrosis from radiation to breast (note mastectomy) with impact on surrounding lung—notice the linear contour

While there is an entire chapter on artifacts and pitfalls, one worth mentioning here is the breathing artifact at the bases [2]. The diaphragm is one of the largest muscles and one in near-constant motion throughout normal physiologic functioning, and while we attempt to coach patients on shallow breathing to allow it to be in a similar position on the CT and PET, this is often difficult, especially with sick patients who may have poor breath control. As a result, misregistration is often greatest near the diaphragm, with the position of lung nodules on CT often quite different from their position on PET. This can lead to false negatives where the nodule appears cold because its PET position is registered a few slices up or down from its CT position. One countermeasure is to view the CT and PET separately in this region. Non-attenuation-corrected images may also be useful, as attenuation correction will be thrown off by misregistration.

While examining the lungs, it is also worthwhile to examine the pleura. Pleural metastases are much more common from primary lung cancers than from melanoma or sarcoma, but the well-known tendency of melanoma to send metastases to odd places means the pleura are at least worth investigating. Unfortunately, one generally cannot tell by avidity whether a pleural effusion is benign or malignant, as plenty of effusions with inflammatory causes may be “hot.” A rind of soft tissue, generally with an irregular appearance, is much more suspicious for spread to the pleura (Fig. 4.19).

Fig. 4.19
figure 19

Metabolically active pleural disease in this patient with melanoma

Once the lungs have been evaluated, one can turn one’s attention to the mediastinum, usually defined simply enough as everything between the lungs (or to be exact the visceral pleural surfaces). The mediastinum contains the heart, great vessels and pulmonary vessels, esophagus, trachea, and many lymph nodes. Among these, the heart, great vessels, and pulmonary vessels are rare sites of metastasis radiologically (though some pathologic studies suggest cardiac metastases may be more common than often thought), and the heart in particular has variable and sometimes very intense uptake, so PET findings here are relatively rare [2]. The esophagus and trachea are not commonly targets of metastasis either, although esophageal inflammation from chemotherapy in patients who have already been treated should at least be commented on to avoid clinicians reading the study on their own coming to the wrong conclusion. Inflammation from gastroesophageal reflux can also be seen.

Often a metastatic target, however, are lymph nodes. Mediastinal and hilar nodes are more commonly targets of metastasis from primary lung lesions, but can sometimes be targets of melanoma spreading down lymph node chains. One possible false positive to beware of is that low-level mediastinal and hilar uptake is fairly common and easily confused with spread of disease (Fig. 4.20). The distinction here is that it is bilaterally symmetric or close to it, and mediastinal and hilar uptake should be similar in level [2]. In cases where there is a granulomatous disease such as sarcoid, it can be quite florid (Fig. 4.21). Metastases, conversely, tend to be sharply focal in a few areas. While diffuse mediastinal and hilar metastasis may exist, it is usually accompanied by widespread disease elsewhere in the body (in melanoma and sarcoma).

Fig. 4.20
figure 20

Low-grade mediastinal and hilar uptake, bilaterally symmetric and possibly due to a small right lower lobe bronchopneumonia. The mild uptake and relative symmetry suggest a benign etiology

Fig. 4.21
figure 21

Intense, symmetric mediastinal and bilateral hilar uptake characteristic of sarcoid

When looking at the chest, one should not forget the superficial soft tissues. Axillary (and adjacent subpectoral) nodes are the preferred site of drainage for upper extremity malignancies and frequent sites of drainage for trunk melanomas. As a result, one should always make sure to check the axillary nodes on both sides. Similar to the case with the mediastinum and hila, low-grade bilaterally symmetric axillary nodes can be seen in viral infections. HIV is a particularly notorious cause of this, and one should check the medical record for a systemic disease that could cause diffuse inflammation; occasionally, this can even be the cause for diagnosing a systemic illness even if the scan is negative for melanoma!

The breasts can also have additional primary malignancies, especially but not exclusively in women (Fig. 4.22). Additionally, melanoma in particular can send metastases to subcutaneous tissues and muscles, so one should look carefully for hot lesions here as well. Cutaneous “hot” spots may indeed be additional melanomas or may be local infections, so this is a place to draw attention to the site and refer the clinician to correlate with physical examination.

Fig. 4.22
figure 22

Intense breast lesion, incidentally discovered, not visible on CT

Finally, a glance at the ribs is wise, as the bone is a common target for metastasis from many tumors. As there are many ribs (usually twelve), it is difficult to follow each one individually, but a look at the ribcage as a whole can allow one to see many metastases. The sternum, scapulae, and clavicles should be examined too, as should the proximal humeri.

One important CT finding is the presence of a pneumothorax (Fig. 4.23). This may be a medical emergency depending on size and pressure and should prompt an immediate call to the referring clinician on detection. Other important findings to report include pleural effusions (Fig. 4.24), pericardial effusions (Fig. 4.25), pulmonary consolidations such as pneumonia, and coronary calcifications (Fig. 4.26), which suggest coronary artery disease. Signs of a median sternotomy and coronary artery bypass grafts may be seen as well. Occasionally, one may see an aneurysm here.

Fig. 4.23
figure 23

Small pneumothorax (red arrow) incidentally found on chest CT

Fig. 4.24
figure 24

This large non-avid pleural effusion in a patient with a non-avid desmoplastic round cell tumor may be benign or malignant; the lack of avidity suggests a benign etiology, but this is not certain

Fig. 4.25
figure 25

This pericardial effusion is metabolically active, as could result from either benign inflammation or tumoral infiltration

Fig. 4.26
figure 26

Coronary artery calcification. This can suggest coronary artery disease, though specificity is rather low. It is usually worth calling attention to in the findings section at least

Abdomen and Pelvis

The wide variety of organs here makes interpretation of the abdomen and pelvis relatively complex. While, as in most cancers, the liver and lymph nodes are common sites of metastasis, the viscera may become targets as well in melanoma.

The liver is a common target for metastasis in melanoma, with its dual blood supply meaning cancers have two methods of ingress. The liver has moderate FDG uptake, enough that it is used for benchmarking uptake in various applications. However, the somewhat grainy uptake means that metastases can be more difficult to detect, and while PET may find some metastases CT or MR do not, the reverse is also true. As a result it is often useful to look at the CT separately, possibly using a liver window to see liver lesions better (Fig. 4.27). If there is continued clinical concern for liver metastasis despite a negative PET/CT, MRI can be useful in the liver [2].

Fig. 4.27
figure 27

Liver metastasis (arrows) is better seen on liver window (middle) than on soft tissue window (left) on CT. On PET (right), lesion clearly is quite avid and malignant

Lymph nodes are also common sites of metastasis. While axillary nodes drain relatively quickly into subclavian trunks and reach the mediastinum, disease entering the inguinal nodes draining the lower extremities passes to external iliac and common iliac nodes along lumbar trunks and finally periaortic nodes before entering the thorax. Similarly, lymph leaving the bowel (and malignancies with it) drains into mesenteric nodes. As a result, there is a wider range of areas that must be checked for metastasis in the abdomen.

Inguinal nodes, draining the legs, often show mild, symmetric uptake and indeed have the largest acceptable dimensions for short axis apart from the subcarinal and lower aortic nodes [1]. Inguinal nodes are seen as suspicious if uptake is particularly intense (much more than liver) or asymmetric, particularly if there is a lower extremity melanoma ipsilateral to the node or the node has a malignant appearance by CT (round, lacking fatty hilum, large) (Fig. 4.28).

Fig. 4.28
figure 28

Round, large, asymmetric inguinal node lacking fatty hilum, known melanoma. The heterogeneous uptake suggests necrosis

Mesenteric nodes may also be sites of metastasis, particularly if the disease has spread to the bowel. It is worthwhile in the case of a melanoma (assuming it is of the usual cutaneous source) to exclude a visceral source of inflammation which might be another cause.

Retroperitoneal nodes (Fig. 4.29), along the major vessels, are common sites of spread once disease has passed the inguinal nodes; in sarcoma cases they may be earlier sites of spread. In general, metastases will pass to the external iliac, then common iliac, and then periaortic/aortocaval nodes, but few malignancies are this regular in their behavior. All of these areas should be examined for enlarged or avid nodes.

Fig. 4.29
figure 29

This large desmoplastic round cell tumor (the same as previous patient with this tumor) has periaortic, right common iliac, and bilateral external iliac metastases

The viscera are less common sites for metastasis, but melanoma in particular is notorious for sending metastases to uncommon locations. The bowel is a reasonably common target (Fig. 4.30), but evaluation of the bowel is complicated by the variable and sometimes intense uptake that can be seen in the bowel and particularly the colon [2]. Hot spots in the colon, if solitary, may reflect primary colon cancers or polyps with malignant potential, and a solitary area of focally increased uptake should prompt a colonoscopy (Fig. 4.31). An exception would be if the patient already has widespread metastatic disease and detecting a colon cancer early would have little effect on quality or duration of life. Metformin, a common oral hypoglycemic, causes the entire bowel to become avid and may complicate diagnosis of bowel disease.

Fig. 4.30
figure 30

Bowel metastasis from melanoma. Bowel metastases are often difficult to see on CT, even with oral contrast; if solitary bowel metastasis is suspected, colonoscopy may be in order to investigate it

Fig. 4.31
figure 31

Two focal areas of increased uptake in the colon, corresponding to masses on CT (poorly visualized without oral contrast); one proved malignant and the other a premalignant polyp on colonoscopy. Bowel metastases, primary colon cancer, and polyps are hard to tell apart, so focal areas of uptake in the colon should prompt a colonoscopy, unless the patient’s chance of survival are otherwise poor and this is unlikely to be of any benefit

Other viscera are usually less important as far as evaluation goes, though it is wise to check them systematically, as melanoma in particular sends metastases to odd places. The gallbladder is rarely a target but can be easily checked along with the liver. The spleen does accumulate metastases and should be checked, although solitary splenic metastases are rare. The adrenal glands are much more commonly the site of spread from lung cancers, although adrenal metastases are well-known. Adrenal adenomas are also quite common, and generally adrenal nodules more avid than liver are considered suspicious (Fig. 4.32; as benign adenomas and bilateral adrenal hyperplasia can produce abnormal uptake, an MRI or contrast-enhanced CT using an adrenal protocol may be useful if uncertainty persists [2]. Evaluation of the kidneys and ureters is difficult given the normal physiologic excretion of tracer into the collecting system, but metastases are sometimes visible, and primary renal cell carcinomas are occasionally found (and may not be avid). A focus that localizes to the ureter is much more likely to be urine in the ureter than an actual metastasis to this organ (Fig. 4.33). Renal cysts, if large enough, are usually visibly non-avid. The bladder is generally difficult to evaluate given that large amounts of radioactive urine collect in the bladder, although windowing may allow the detection of masses—note that these may be seen as cold spots.

Fig. 4.32
figure 32

Bilateral adrenal metastases. The liver is often used as a cutoff for determining “how hot is too hot” for adrenal nodules. Avid adrenal nodules can be further investigated by contrast-enhanced CT or MRI using special adrenal nodule protocols

Fig. 4.33
figure 33

Uptake localizing to ureter (arrows). Uptake has a linear, vertical, superolateral-inferomedial orientation and localizes to the ureter on CT. It can be easily confused with a retroperitoneal node

Reproductive organs such as the prostate (in men) and uterus and ovaries (in women) are uncommon targets of spread of disease. Avid foci in the ovaries in a woman of childbearing age may be entirely physiologic. Foci in the prostate along the course of the urethra may be similarly physiologic, but correlation with PSA is at least useful given the high background rate of prostate cancer in older men.

Soft tissues and bones are important in the abdomen and pelvis. Injection granulomas from subcutaneous injections of insulin or other drugs may be avid and appear as false-positive sites; these are usually sorted out with a check of the patient’s history and, if necessary, a note to the clinician about a physical examination. As in the head, neck, and chest, melanoma may send metastases to the subcutaneous tissues and muscles of the abdominal wall (Fig. 4.34).

Fig. 4.34
figure 34

Subcutaneous and muscular melanoma metastases (arrows). Note that muscular metastases in particular may be unimpressive and even invisible on CT without PET; the relatively high contrast of subcutaneous fat makes the metastases easier to see

Checking the skeleton in the abdomen and pelvis is important as much of the skeletal mass is here. The pelvis should generally be checked in multiple planes, both axial and coronal at the least, whereas a single sagittal view can see most of the spine unless the patient is very scoliotic. In the skeleton, it is also important to look at the scan CT only using bone windows, as subtle sclerotic metastases may be present.

The abdomen and pelvis are notorious for the variety of CT findings that may exist, and of course CT of the abdomen and pelvis is the preferred method of diagnosing abdominal pain in many areas due to the wide variety of pathologies that may exist. Fortunately, most of these produce inflammation and are FDG-avid. Diverticulitis (Fig. 4.35), appendicitis, cholecystitis, and pancreatitis are all FDG-avid stranding, so the inflamed organ will be FDG-avid in addition to the CT findings such as edema and fat stranding. Pyelonephritis may be somewhat more difficult to diagnose as the kidney may have increased or decreased uptake. Bowel obstruction may not produce increased uptake, but the dilated loops of bowel are easy enough to recognize. Kidney stones may not produce visible FDG changes as they occur in a system with high background uptake, but are easy enough to detect on CT, and can be seen with the fused images turned off.

Fig. 4.35
figure 35

Focus of uptake localized to a diverticulum in the colon (arrow), with mild soft tissue stranding surrounding the diverticulum, is suspicious for diverticulitis. Free air from perforation may also be seen

Musculoskeletal System and Extremities

The musculoskeletal system, practically speaking the bones and muscles for the purpose of searching for malignancy, is present throughout each part of the body and hence can be searched when previously searching the head and neck, chest, and abdomen and pelvis. An additional search at the end is nonetheless prudent, as one is frequently focused on the viscera on an initial run-through.

If the scan is a whole-body scan (head to toe), as is commonly done for melanoma, a final look at the legs is at least prudent. One can scroll down using the soft tissue window and then scroll back up using the bone window and then go up and look at the skeleton as one moves upward toward the head. Non-axial plane reformats are often useful for the skeleton, given its craniocaudal extent relative to other organs and the importance of a whole-skeleton evaluation given the problems with diffuse versus patchy disease, as described below. A sagittal reconstruction allows one to view the entire spine at once unless significant scoliosis is present, and coronal reconstructions allow one to see much of the pelvis in one view. A major complication in interpretation of the skeleton is the low-grade bone marrow activity [8] (Fig. 4.36) which is normal and increases after the growth factors frequently given with chemotherapy (Fig. 4.37) [2]. Metastases generally produce a lumpy, multifocal appearance (Fig. 4.38). The MIP and sagittal and coronal sections can be useful in telling these apart. It is worth mentioning that metastases may be hidden by the intense bone marrow uptake.

Fig. 4.36
figure 36

Relatively normal bone marrow uptake. There is uptake in the axial and proximal appendicular skeleton, but it is not immediately obvious on the MIP unless rotated posteriorly

Fig. 4.37
figure 37

Relatively homogeneous bone marrow and splenic uptake from growth factors. Note the uniform uptake throughout the spine (the sagittal view is usually good for this unless there is substantial scoliosis) and the slightly increased intensity of the spleen vis-à-vis the liver

Fig. 4.38
figure 38

Heterogeneous skeletal uptake from metastases—note the lumpy and heterogeneous uptake pattern, particularly in the spine, as opposed to the smoother uptake in the axial and proximal appendicular skeleton usually seen with bone marrow activation. Liver and bowel metastases are also present

Muscles can be targets of metastasis in melanoma (something much less common in other cancers). As such focal areas of uptake within muscle are much more concerning. This must be differentiated from physiologic activity due to the patient using the muscle during the uptake period [8] (Fig. 4.39). Usually this is simple enough as physiologic muscle uptake lights up the whole muscle in linear or triangular, muscle-shaped patterns, but smaller, short muscles in the neck may occasionally cause confusion.

Fig. 4.39
figure 39

Muscle uptake reflecting motion during the uptake period

Issues Specific to Melanoma

Surgical Field

Important in evaluation of melanoma is knowledge of prior surgical fields. Melanomas may recur multiple times in the same area, and it is important to know where the patient has been resected in order to pay special attention to those areas (Fig. 4.40). Postsurgical changes may be seen as well, but are usually more low grade and diffuse [2]. In addition, the duration of time from the last surgery should be known, as the area may continue to display uptake for a few weeks or even months after surgery. The reader should also be aware of any nodal dissections performed to stage the melanoma, not only as a possible site of metastasis, but because removal of a nodal basin may mean that recurrence occurs in other nodes nearby instead. Surgical clips can help show the site of axillary or inguinal (or epitrochlear, popliteal, or jugular) nodal dissection.

Fig. 4.40
figure 40

Recurrence in surgical bed. Arm melanomas frequently drain to the axilla, and in this case, after an axillary nodal dissection, this arm melanoma presented with a recurrence in the axilla. PET-CT clearly demonstrates the large, confluent axillary mass

Whole-Body Imaging

Most cancers are imaged from skull base to mid-thigh, the idea being that the low sensitivity of PET for brain metastases makes including the brain useless, and metastases to the lower extremities are rare in the absence of numerous other metastases. Melanoma is imaged from vertex to toes (literally whole-body imaging), the idea being that given its tendency to spread all over the body, scanning the scalp and legs becomes useful. This is somewhat controversial, as isolated leg metastases are relatively rare even in melanoma [9, 10]. Melanoma is somewhat unique in this regard (the only other tumor we scan this way at our institution is multiple myeloma), and there is an argument for scanning skull base to mid-thigh for melanoma as well. The exception, of course, is melanomas of the lower extremities, where the lower extremities should be scanned in their entirety to detect the primary and any in-transit disease, and melanomas of the scalp, in which the whole skull should be included for the same reason.

NAC Image Use

Scans for melanoma often make use of non-attenuation-corrected (NAC) images, which have greatly decreased uptake in the inside of the body (Fig. 4.41), but increased uptake in the skin and lungs, which have counts suppressed by attenuation correction algorithms as part of their normal function. This remains somewhat controversial, but in the case of melanoma, may be relevant due to the possibility of skin lesions undetected on attenuation corrected images [2]. As with boosting the gain on any imaging system, noise rises as well as signal, but given the ease of correlating with physical examination on the skin, this may be worth doing.

Fig. 4.41
figure 41

NAC IMAGES, MIP, and FUSED (with AC image for comparison). Notice the higher uptake in the skin and significantly decreased uptake in the viscera

Unusual Metastasis Locations (but Not in Melanoma)

The most common sites are lung, liver, bone, and brain, but significant numbers of metastases are also seen to bowel, adrenal, pancreas, heart, kidney, and spleen, as well as all the other viscera [11]. As a result, every organ should be checked, as described in the section-by-section lists above, and even relatively poorly perfused sites such as the subcutaneous tissues may become targets.

Immunotherapy

Immunotherapy has been a powerful new addition to the armamentarium of medical oncologists, with many previously untreatable cases of metastatic melanoma showing response to therapy. However, new side effects related to immunotherapy are being discovered, many of which are visible on PET/CT. In particular, the immune system may attack normal organs, which in addition to the obvious organ damage and consequent pathology appears as inflammation—and hence a false positive—on FDG-PET.

Current immunotherapies for melanoma include treatments targeted at mutations of the BRAF gene which result in constitutive activity of the MAPK or MEK cell proliferation and growth pathways and immunotherapies aimed at activating cytotoxic T cells to attack the melanoma. BRAF-targeted medications include oral BRAF inhibitors (vemurafenib, dabrafenib, or encorafenib) and MEK inhibitors (cobimetinib, trametinib, or binimetinib) [12].

With BRAF inhibitors, there may be increase in keratoacanthomas or squamous cell carcinomas, which may appear as metabolically active skin lesions and be seen on PET; as these are difficult to distinguish from cutaneous metastases, the reader should simply call attention to them, as being cutaneous lesions they will be easy to detect by physical exam. Lesions may fail to decrease in size despite therapy being effective, but FDG uptake usually occurs more rapidly (the terminal gene of the MAPK pathway, ERK, decreases glycolysis when downregulated). These inhibitors may also cause activation of the immune system, and one may see activation of lymph nodes in the lymphatic drainage basin of metastatic sites and increased splenic uptake, reflecting an immune flare, particularly if tumor shrinkage also occurs [12].

Immunotherapies aimed at stimulating the immune system currently consist of antibodies that inhibit physiological inhibition of the immune system (thus making the immune system more active). These include antibodies to the CTLA4 receptor (ipilimumab) and to the PD1 receptor on activated T cells (pembrolizumab, nivolumab). Anti-CTLA4 agents cause infiltration of lymphocytes into sites of disease and germinal center activity in draining lymph nodes. Anti-PD1 agents increase cytotoxic killing by T cells. These can also produce a variety of toxicities (usually presenting at about 6–8 weeks) in various organs, most commonly the thyroid but also the pituitary, adrenals, lungs, liver, pancreas, and bowel; often the inflammation can be seen on PET as increased activity. As with BRAF inhibitors, one may see reactive nodal uptake in the drainage basin of metastases and diffuse splenic uptake, as well as a sarcoid-like symmetric hilar and mediastinal nodal pattern, more common with pulmonary metastases [12].

Accompanying CT findings include colonic wall thickening, pericolonic fat stranding, and possible perforation and free fluid for colitis, mild hepatomegaly, periportal lymphadenopathy, and periportal edema for hepatitis, and diffuse groundglass and reticular opacities, consolidation, and traction bronchiectasis for pneumonitis [12].

CTLA-4 agents in particular have been known to produce false-positives for progression due to the infiltration of lesions (which can cause an apparent increase in size) or the appearance of new ones due to immune infiltrate (which are not actually new cancers). While baseline lesions may shrink or durable stable disease may occur as with other agents, other patterns including response after initial “worsening” (pseudoprogression) and simultaneous response in baseline lesions and presence of new lesions which also result in partial or complete response have also been described [13].

Due to the possible problems with new areas of uptake representing either new tumor (progression) or areas of immune response (response), separate immunotherapy guidelines have been produced for assessing immune response to therapy (immune-related response criteria, irRC). Generally, new measurable lesions on CT are incorporated into total tumor burden (rather than representing progressive disease and hence treatment failure), and progressive disease is defined as more than a 25% increase in total tumor burden compared with nadir, in two consecutive observations at least 4 weeks apart. Formal criteria for PET do not yet exist, but new lesions in common sites of toxicity such as the adrenals, bowel, or pituitary should not necessarily be assumed to be new sites of metastasis [13]. A follow-up examination in 1–2 months with new lesions that are felt to be possibly due to immune reaction (whether due to new lymph nodes in a relatively diffuse distribution more typical of inflammation than tumor, presence of increased splenic uptake, or localization in a typical target organ for adverse reactions such as bowel or pituitary) may be useful [14].

Issues Specific to Sarcoma

Heterogeneity of Uptake and Grade

Sarcomas are often heterogeneous in degree of aggressiveness, and the grade of the sarcoma may be underestimated if the wrong area within it is selected for biopsy. As dedifferentiated sarcomas tend to be both more aggressive and more FDG-avid, areas of high SUV more likely represent areas of high histologic grade, which can then be targeted for biopsy (Fig. 4.42) [15]. Attempts to grade tumors solely by SUV have been made, but generally find a high degree of overlap between benign and malignant tumors [15].

Fig. 4.42
figure 42

Largely poorly avid fibrosarcoma with smaller area of more intense uptake. This area was targeted for biopsy

Metastases

Unlike the extremely wide range of metastases produced by melanoma, bone and soft tissue sarcomas often metastasize to the lungs. Retroperitoneal sarcomas often spread to the liver, and myxoid liposarcomas often spread to the retroperitoneum, spine, and paraspinous soft tissues. Synovial sarcoma, rhabdomyosarcoma, angiosarcoma, clear cell sarcoma, and epithelioid sarcoma often spread to locoregional lymph nodes, like other cancers [15] (Fig. 4.43).

Fig. 4.43
figure 43

This thigh rhabdomyosarcoma spread to local nodes, as well as a variety of more proximal and distant locations. Note that retroperitoneal nodes, while not that avid overall, are nonetheless visibly avid

Sarcomas are generally not as avid as melanomas (Fig. 4.44), and while not in the category of tumors unable to be staged by FDG-PET (like carcinoid or prostate cancer), careful attention to both PET and CT is useful.

Fig. 4.44
figure 44

Metastases to the bone. This retroperitoneal sarcoma has diffuse but relatively poorly avid bony metastases, which are even more subtle on CT

CT, which is most sensitive for small lung lesions, is thus necessary for full staging of a sarcoma. As most modern scanners now include a CT, this is generally not an issue, but older PET-only scanners may require addition of a separate CT scan. The reader should also inspect the lungs carefully for smaller lung metastases [15].

Response to Therapy

Lacking the extensive use of immunotherapy, a decrease in uptake after chemotherapy usually predicts tumor response. GIST tumors in particular often display a decrease in response after treatment with imatinib before decreasing in size [15].

Joint Replacements

One problem with extremity osteosarcoma in particular is that limb-salvage therapy with endoprosthetic replacement produces metallic artifacts from prosthetics on both CT and MR. PET is generally less affected by these artifacts intrinsically, but errors in attenuation correction may also hamper PET interpretation. A look at the non-attenuation-corrected images can be helpful, particularly as attenuation is less important in most extremity lesions. Also, chronic periprosthetic inflammation will take up FDG and may produce false positives. The reader should be aware of this and look specifically for focal areas of increased uptake [15].