Benign lytic lesions

A benign, bubbly lytic lesion of bone is probably one of the most common skeletal findings a radiologist encounters. The differential diagnosis can be quite lengthy and is usually given on an “Aunt Minnie” basis (I know that’s Aunt Minnie because she looks like Aunt Minnie); in other words, the differential diagnosis is structured on how the lesion looks to the radiologist based on his or her experience. This method, called pattern identification, certainly has merit, but it can lead to many erroneous conclusions if not tempered with some logic. For instance, most radiologists would justifiably miss the diagnosis of a rare presentation of a primary malignant neoplasm that initially looks benign. Many of these radiologists would subsequently insist on including primary malignant neoplasms in their benign lytic differential even though the rare malignancy is “1 in a million.” If every differential is geared to cover even the long shots, there would be a lot of extremely long differentials and the clinicians wouldn’t get much useful information from us. We might as well give the clinician the index to a multivolume bone book as the differential to ensure we never miss anything.

Then again, you don’t want a differential diagnosis list that is wrong half the time. You could almost do better with a coin flip. I’m willing to accept a differential that is accurate (i.e., one that contains the correct diagnosis) 95% of the time. This is acceptable to me for most skeletal entities; however, I would be remiss if I were willing to accept a 1-out-of-20 miss rate for fractures and dislocations. Nevertheless, for most of the entities in this book I will accept 95% accuracy in differentials and would expect most radiologists to concur. If you want to be more accurate than that, you simply add more diagnoses to the list of differential possibilities.

The shorter the differential diagnosis list is, the handier it is to remember and apply. As the differential list gets longer, it generally gets more accurate but it can be difficult to remember and often falls into disuse. Mnemonics are helpful in recalling long lists of information, and I will pass on many that I use; many people, however, do not like to use mnemonics (for no good reason that I’ve been able to ascertain) and will have to use whatever method works for them to remember the differentials. The list of entities that cause benign lytic lesions is quite long; therefore a mnemonic is helpful in recalling them.

I was a flight surgeon in the Air Force before my radiology residency, and I would spend a half-day every week or so with the radiologist, trying to pick up some pearls. This radiologist was Ivan Barrett, and he did me a great favor: he taught me the mnemonic FEGNOMASHIC, which is made up from the first letter of each of the entities in the differential of benign lytic lesions of bone. For instance, the F stands for fibrous dysplasia; the E, for enchondroma; and so on. I diligently learned what each letter stood for, even though I had no idea what most of the processes were or looked like on an x-ray film. Before I could learn another mnemonic from Ivan (I was a slow learner), I moved away to begin residency.

Sure enough, the first week of residency, in a formal conference with 15 to 20 residents present, I was chosen as the sacrificial lamb among the first-year residents to take an unknown case. It happened to be a benign lytic lesion, which I proceeded to expound on with a list of 12 to 15 differential possibilities. The conference room got quiet, I was thanked cordially but a little frostily, and the conference was adjourned. One of the first-year residents whom I barely knew at the time asked how I knew so many of the possibilities on that case, because the staff man showing the case (who was a chest radiologist) didn’t even know that many. I explained, with a straight face, that those simply seemed like the logical things to mention. I was trying to be matter-of-fact and not come off as too much of a show-off, but I couldn’t help laughing. I then told the resident how I had learned a single mnemonic and how with a little luck it made it seem like I knew a lot more than I really did. He and the rest of my fellow residents were relieved that I really wasn’t any smarter or more advanced than them and quickly learned the mnemonic themselves. I became hopelessly addicted to mnemonics from that day on.

FEGNOMASHIC

FEGNOMASHIC is defined in Funk and Wagner’s unabridged dictionary, 13th edition, as “one who uses mnemonics.” It serves as a nice starting point for discussing possibilities that appear as benign lytic lesions in bone. That mnemonic has been in general use for many years, but I have never heard a claim as to who first coined it. The first mention of it that I saw in print was in 1972 in a radiology article by Gold, Ross, and Margulis.¹ By itself it is merely a long list—about 14 entities—and needs to be coupled with other criteria so that it can be shortened into a manageable form for each particular case. For instance, if the lesion is epiphyseal, only three to five entities need to be mentioned, depending on how accurate you care to be. If multiple lesions are present, only half a dozen entities need to be discussed. Ways of narrowing the differential are discussed later in this chapter.

The next step after learning the names of all the lesions is getting some idea of what each one looks like. This is where experience becomes a factor. For the medical student or first-year resident it is difficult to go beyond saying that they all look lytic or bubbly and benign. However, the fourth-year resident should have no trouble differentiating between a unicameral bone cyst and a giant cell tumor (GCT) because he or she has seen examples of each many times before and knows what each looks like. The fourth-year resident may have a difficult time verbalizing the differences but should be able to tell them apart.

A novice can quickly gain experience by looking at the examples of each of these lesions in a major skeletal radiology text. In fact, I highly recommend that you compare my description and differential points on each lesion with multiple examples in other books. Some of these lesions can be diagnosed radiologically only on a pattern identification or Aunt Minnie basis. In other words, there are no hard-and-fast criteria to differentiate some of the lesions from the others.

After getting a feel for what each lesion looks like radiographically and overcoming the frustration that builds when you realize that many of them look alike, you should try to learn ways to differentiate each lesion from the others. I have developed a number of keys that I call discriminators that will help you differentiate each lesion. These discriminators are 90% to 95% useful (I will mention when they are more or less accurate based on my experience) and are by no means meant to be absolutes or dogma. They are guidelines but have a high confidence rate.

Textbooks rarely tell you that a finding always or never occurs. Authors temper their descriptions with virtually always, invariably, usually, or characteristically. I have tried to pick out findings that come as close to always as I can, realizing that I will often be only 95% accurate. That’s good enough for me. If it is not good enough for you, you will have to get your own differential criteria or discriminators. Try these and see if they work for you. If they don’t, modify them as necessary. Whatever you do, develop reasons for including things in your differential. Have concrete criteria of some kind for including or excluding each entity.

I will give a brief description of each entity, because complete descriptions are readily available in any skeletal radiology text. What I will dwell on, however, are the points that are unique for each entity, thereby enabling differentiation from the others. Table 2-1 is a synopsis of these discriminators.

TABLE 2-1

Differential Criteria for Benign Lytic Bone Lesions

Fibrous dysplasia

It is unfortunate that this differential starts with fibrous dysplasia because fibrous dysplasia can look like almost anything. It can be wild looking, a discrete lucency, patchy, sclerotic, expansile, multiple, and a host of other descriptions. It is therefore difficult to look at a bubbly lytic lesion and unequivocally say it is or is not fibrous dysplasia. When assessing such a lesion, radiology residents usually say, “I suppose it could be fibrous dysplasia, but I’m not sure.” The resident is feeling insecure and becomes immediately defensive, setting the tone for the entire differential diagnosis. It would be better if the differential started on a positive note, say, with GCT or chondroblastoma, where there are some hard and definite criteria. This would help the resident set a tone of self-assurance and decisiveness rather than appearing wishy-washy. Then, when the resident mentions fibrous dysplasia and uses the same words, “I suppose it could be fibrous dysplasia, but I’m not sure,” it is looked upon as thoughtful deliberation rather than insecurity or ignorance. It is pure gamesmanship, but it makes a difference in how the resident is perceived.

How do you know whether to include or exclude fibrous dysplasia if it can look like almost anything? Experience is the best guideline. In other words, look in a few texts and find as many different examples as you can; get a feeling for what fibrous dysplasia looks like. A few examples are shown in Figures 2-1 to 2-6, but pouring over another text for 10 to 15 minutes will be time well spent.

FIGURE 2-1 Fibrous dysplasia. A predominantly lytic lesion with some sclerosis and expansion is seen in the distal half of the radius in a child. A long lesion in a long bone typifies fibrous dysplasia. Although parts of this lesion indeed have a ground-glass appearance, most of it does not. Expansion and bone deformity like this is commonly seen in fibrous dysplasia.

FIGURE 2-2 Fibrous dysplasia. The ribs are often involved with fibrous dysplasia, as in this example. When the posterior ribs are involved, the process is often a lytic expansile lesion, whereas when the anterior ribs are involved, it is commonly a sclerotic process.

FIGURE 2-3 Fibrous dysplasia. An expansile mixed lytic/sclerotic process in the proximal femur is a common pattern for fibrous dysplasia. Note that the supraacetabular region is also involved. The ipsilateral proximal femur is invariably affected when the pelvis is involved with fibrous dysplasia.

FIGURE 2-4 Fibrous dysplasia. The entire pelvis and proximal femurs are diffusely involved with polyostotic fibrous dysplasia. The pelvis is severely deformed with predominantly lytic lesions. The proximal femurs are involved with both lytic and sclerotic lesions.

FIGURE 2-5 Monostotic fibrous dysplasia. The proximal femur is a common place for monostotic fibrous dysplasia. This presentation is typical and should not be confused with an infection. Some speckled calcification is noted within the lesion that should not be misconstrued as chondroid calcification in an enchondroma.

FIGURE 2-6 Monostotic fibrous dysplasia. Another example of proximal femur involvement that shows a lytic lesion with a thick sclerotic border reminiscent of a chronic infection. This is characteristic of fibrous dysplasia in the hip.

Fibrous dysplasia will not have periostitis associated with it; therefore, if periostitis is present, you can safely exclude fibrous dysplasia. It would be possible to have a pathologic fracture through an area of fibrous dysplasia, which then had periostitis, but I have never seen this occur. Fibrous dysplasia virtually never undergoes malignant degeneration and should not be a painful lesion in the long bones unless there is a fracture.

Fibrous dysplasia can be either monostotic (most commonly) or polyostotic and has a predilection for the pelvis, proximal femur, ribs, and skull. When it is present in the pelvis, it is invariably present in the ipsilateral proximal femur (Figures 2-3 and 2-4). I have seen only one case in which the pelvis was involved with fibrous dysplasia and the proximal femur was spared. The proximal femur, however, may be affected alone, without involvement in the pelvis (Figures 2-5 and 2-6).

The classic description of fibrous dysplasia is that it has a ground-glass or smoky-appearing matrix. This description confuses people rather than helps them, and I do not recommend using ground-glass appearance as a buzzword for fibrous dysplasia. Fibrous dysplasia is often purely lytic and becomes hazy or takes on a ground-glass look as the matrix calcifies. It can go on to calcify quite a bit, and then it presents as a sclerotic lesion. Also, I often see other lytic lesions that have a distinct ground-glass appearance; therefore the ground-glass quality can be misleading.

When fibrous dysplasia is in the differential diagnosis for a lesion in the tibia, an adamantinoma should also be mentioned (Figure 2-7). An adamantinoma is a malignant tumor that radiographically and histologically resembles fibrous dysplasia. It occurs almost exclusively in the tibia (for unknown reasons) and is rare. Because it is rare, you may choose not to include it in your memory bank—you won’t miss more than one or two in your life, even if you are a busy radiologist.

FIGURE 2-7 Adamantinoma. A wild-looking mixed lytic/sclerotic lesion in the tibia that resembles fibrous dysplasia is classic for an adamantinoma. This lesion occurs almost exclusively in the tibia and has some malignant potential. An adamantinoma should always be considered when a lesion resembling fibrous dysplasia is seen in the tibia.

Polyostotic fibrous dysplasia occasionally occurs in association with café-au-lait spots on the skin (dark-pigmented, frecklelike lesions) and precocious puberty. This complex is called McCune-Albright syndrome. The bony lesions in this syndrome, and even in the simple polyostotic form, often occur unilaterally (i.e., in one half of the body). This does not happen often enough to be of any diagnostic use in differentiating fibrous dysplasia from other lesions. The presence of multiple lesions in the jaw has been termed cherubism, which relates to the physical appearance of the affected child. Such children have puffed out cheeks, producing an angelic look. The jaw lesions in cherubism regress in adulthood.

Fibrous dysplasia often has areas of chondroid matrix, which, when a biopsy is performed, can resemble chondrosarcoma (as can any chondral tissue). I have seen sampling errors result in inappropriate radical surgery when the diagnosis should have been made based on the plain films with no biopsy performed.

Discriminator: No periosteal reaction.

Enchondroma and eosinophilic granuloma

Enchondroma

The most common benign cystic lesion of the phalanges is an enchondroma (Figure 2-8). Enchondromas occur in any bone formed from cartilage and may be central, eccentric, expansile, or nonexpansile. They invariably contain calcified chondroid matrix (Figure 2-9, A) except when in the phalanges. If a cystic lesion is present without calcified chondroid matrix anywhere except in the phalanges, I will not include enchondroma in my differential.

FIGURE 2-8 Enchondroma. A benign lytic lesion in the hand is an enchondroma until proved otherwise. This is a common presentation of an enchondroma. Enchondromas in any other part of the body should contain some calcified chondroid matrix before they are included in the differential. However, calcified chondroid matrix is unusual in the phalanges.

FIGURE 2-9 A, Enchondroma. A lesion in the distal femur is seen with irregular speckled calcification typical of chondroid matrix. This is virtually pathognomonic of an enchondroma. A chondrosarcoma could have an identical appearance but should be painful for consideration in the differential diagnosis. A bone infarct can be similar in appearance to an enchondroma. B, Bone infarcts. Bilateral lytic lesions in the femurs are noted with a densely calcified serpiginous border characteristic of bone infarcts. Compare these lesions with (A), which does not have a well-defined serpiginous border. Often the differentiation between bone infarcts as in this example and an enchondroma (A) is not so clear-cut. (Case courtesy of Dr. Hideyo Minagi.)

Often it is difficult to differentiate between an enchondroma and a bone infarct. Although some of the following criteria are helpful in separating an infarct from an enchondroma, they are not foolproof. An infarct usually has a well-defined, densely sclerotic, serpiginous border, whereas an enchondroma does not (Figure 2-9, B). An enchondroma often causes endosteal scalloping, whereas a bone infarct will not.

It is difficult, if not impossible, to differentiate an enchondroma from a chondrosarcoma. Clinical findings (primarily pain) serve as a better indicator than radiographic findings, and indeed pain in an apparent enchondroma should warrant surgical investigation. Periostitis should not be seen in an enchondroma either. Trying to differentiate an enchondroma from a chondrosarcoma histologically is also difficult, if not impossible at times. Therefore biopsy of an apparent enchondroma should not be performed routinely for histologic differentiation.² Magnetic resonance imaging (MRI) criteria for benign versus malignant includes lack of a soft tissue mass and no surrounding T2 high-signal edema in benign enchondromas.

Because an enchondroma can histologically mimic a chondrosarcoma, biopsies on painless chondroid lesions of all types should not be routinely done. Most tumor surgeons prefer to watch them with serial imaging (every 6 months to a year) and close clinical supervision. Radiologists should refrain from adding “cannot exclude chondrosarcoma” to a formal report when looking at a benign-appearing chondroid lesion, even though that is technically correct. The problem is there are few surgeons who will not biopsy a lesion if the dictation mentions chondrosarcoma as a possibility. I have seen multiple examples of radical surgery for benign cartilaginous lesions being undertaken because a biopsy was incorrectly interpreted as chondrosarcoma (including a forequarter shoulder amputation). The correct dictation for an enchondroma should simply say “benign-appearing chondroid lesion with no aggressive features noted”—no differential diagnosis should be given.

Multiple enchondromas occur on occasion, and this condition has been termed Ollier’s disease (Figure 2-10, A). It is not hereditary and does not have an increased rate of malignant degeneration. Older books say that Ollier’s disease has a high rate of malignant degeneration; this is because any chondroid lesion can look malignant when a biopsy is performed and needs to be correlated radiographically and clinically. The presence of multiple enchondromas associated with soft tissue hemangiomas is known as Maffucci’s syndrome (Figure 2-10, B). This syndrome also is not hereditary; however, it is characterized by an increased incidence of malignant degeneration of the enchondromas.

FIGURE 2-10 A, Ollier’s disease. Multiple lytic lesions in the hand—multiple enchondromas—are seen in this patient. This is known as Ollier’s disease. B, Maffucci’s syndrome. Multiple enchondromas associated with soft tissue hemangiomas are seen in the hand in this patient. This is Maffucci’s syndrome. Note the multiple rounded calcifications in the soft tissues, which are phleboliths in the hemangiomas.

Discriminators:

1. Must have calcification (except in phalanges).

2. Must be no periostitis.

Eosinophilic granuloma

Eosinophilic granuloma (EG) is a form of histiocytosis X, the other forms being Letterer-Siwe disease and Hand-Schüller-Christian disease. Although these forms may be merely different phases of the same disease, most investigators categorize them separately. The bony manifestations of all three disorders are similar and are discussed in this text simply as eosinophilic granuloma, or EG.

EG, unfortunately for radiologists, has many appearances. It can be lytic or blastic, may be well defined or ill defined (Figures 2-11 and 2-12), may or may not have a sclerotic border, and may or may not elicit a periosteal response.³ The periostitis, when present, is typically benign in appearance (thick, uniform, wavy) but can be lamellated or amorphous. EG can mimic Ewing’s sarcoma and present as a permeative (multiple small holes) lesion.