Introduction
In the evaluation of a patient with a bone tumor, there are several areas where data can be gathered that bear on the differential diagnosis. These include the history of present illness, past medical history, family history, social history, physical examination, and review of imaging studies. Ultimately, it may be determined that histologic confirmation is required, at which time a careful evaluation of lesional tissue corroborated with the imaging study information will usually confirm a specific diagnosis.
The social history may elucidate behavioral or environmental factors which may impact the differential diagnosis. For example, exposure to Agent Orange has been associated in some studies with a possible increased incidence of soft tissue sarcoma. More commonly, a significant smoking history is associated with an increase in a variety of cancers, many of which metastasize to bone. The family history is typically noncontributory for any specific lesion. Rare exceptions include patients with a family history of inherited diseases such as Li-Fraumeni syndrome, neurofibromatosis, and multiple hereditary exostoses, which will usually be apparent in the family history.
The history associated with the presence of a musculoskeletal tumor in a specific patient defines the clinical context of the lesion. Age, sex, location of lesion within a bone, duration of symptoms, presence and quality of pain, history of trauma or weight loss, smoking history, and history of prior malignancy are all important historical factors. Critical to the early diagnosis of a skeletal tumor is an appreciation of the fact that early symptoms associated with skeletal neoplasms mimic all types of ordinary musculoskeletal disorders. Any pain that extends beyond the expected duration associated with a tentative diagnosis should raise the suspicion of an underlying tumor. Night pain is another red flag again leading to the suspicion of an occult lesion although many non-neoplastic conditions may also cause pain at night. Osteoid osteoma is the classic example of this phenomenon.
One of the most confusing parts of the history in a patient with an occult tumor is a history of trauma. Frequently, patients will experience some mild trauma to the affected area and then notice pain which would probably not have occurred or would have rapidly resolved in the absence of an underlying lesion. This is frequently not clear to the patient, however, who may directly attribute the local symptoms and findings to the traumatic event. The history related in this way may mislead a treating physician who then follows the local lesion until it becomes obvious that the true nature of the lesion goes well beyond a minor trauma. An example of this is a story related by a patient who was working as a waiter. This young man kicked a kitchen door to open it while carrying a heavy tray. The door was stuck and did not move, resulting in an apparent leg injury. When the patient’s pain did not resolve, compartment syndrome was suspected, and it was not until several months later that tissue was obtained which revealed an underlying lymphoma. Similar is the history of an elderly female on full-dose warfarin for a mechanical heart valve who bumped her thigh on a kitchen table and found out months later that the large anterior thigh mass was a soft tissue sarcoma and not a simple hematoma. In light of the fact that hematomas are rather frequently confused with soft tissue sarcomas, it has been the author’s experience to be especially suspicious of any possible hematoma that has not had an associated occurrence of localized ecchymosis. It is only through intellectual discipline and diligence that early diagnoses can be accomplished.
Once a comprehensive history is obtained, a careful and thorough physical examination is then essential in the evaluation of any patient with a suspected musculoskeletal tumor. Some tumors, such as multiple myeloma, may present with generalized symptoms without specific localized physical findings. When localized physical findings are present, a careful examination of the body part involved is essential. Physical findings such as warmth, redness of the overlying skin, presence of a mass, tenderness to palpation, pain with joint range of motion, localized lymphedema, and mechanical pain are frequent significant findings.
Another potential pitfall is confusion associated with referred pain. Deep pains associated with a tumor are typically pains that the individual has not experienced previously. It is not uncommon in such cases that the patient will perceive pain in a location not actually involved in the skeletal pathology. In nonorthopedic diseases, this can be seen in patients who experience left arm pain during a myocardial infarction. In these patients, there is no mechanical or structural problem with the left arm, but our brains frequently find it difficult to localize deep pain that we have never experienced before. A classic orthopedic example is a patient with a hip tumor complaining of knee pain. This is a common finding in non-neoplastic conditions such as slipped capital femoral epiphysis but occurs with bone tumors as well. For example, the author recalls seeing a patient who had previously undergone arthroscopic meniscectomy for complaints of knee pain which were actually associated with a lymphoma of the proximal femur. A key to understanding these cases is that while the patient may have pain referred to another location, the examination of those other locations would typically be normal. Considering this fairly frequent tendency for hip lesions to present with knee pain, it is extremely important that a physician examining a knee also examine the thigh and hip which may be the source of the knee symptoms.
Bone Tumor Radiology
In most cases, the first imaging study to be obtained in a patient with a suspected skeletal lesion is a plain radiograph. In order to properly interpret these radiographic findings, it is essential to first understand the various relationships that occur between a bone lesion and the bone itself as well as the mechanical properties of bone that limit the growth of the tumor and/or the response of the bone to the tumor.
Unlike soft tissues which can move in space in response to compression from a mass, bones are rigid structures which can only accommodate tumor growth by bone dissolution or permeation of the lesion through an otherwise mainly intact bone. Lesions which create bone marrow replacement without a calcified matrix appear as lytic lesions on radiograph if the tumor causes bone lysis in conjunction with marrow replacement. If bone marrow replacement is not accompanied by bone lysis, then the radiograph will be remarkably normal early in the disease process.
In cases where lysis does occur, as the tumor grows it stimulates osteoclasts to resorb bone. While tumors can cause resorption of both medullary bone and cortical bone, the interface between the tumor and the medullary bone gives a more refined indication of the behavior of the tumor since medullary bone is a fine bone lattice as opposed to the dense lamellar bone of the cortex.
In this setting of medullary bone destruction, there are 3 potential scenarios that define the relationship between the tumor and the surrounding medullary bone. These three appearances of medullary destruction adjacent to a tumor can be described as either geographic, moth-eaten, or permeative.
Geographic destruction yields a fine and easily traceable border between totally abnormal and totally normal bone. This is the same as the boundaries between countries on a map. If the lesion is not locally aggressive and it is recognized as foreign by the body, the bone will wall it off with a sclerotic rim. This will result in a geographic margin with sclerosis. Again, this indicates a nonaggressive lesion which has been walled off and constrained by the surrounding bone. This type of margin would be typical for a nonaggressive lesion such as a non-ossifying fibroma as seen in Figure 1.
Not every geographic margin, however, indicates an underlying benign lesion. Multiple myeloma typically causes punched-out lesions in bone which have sharp geographic margins. In this case, however, the lesion is not recognized as abnormal by the body. Plasma cells are normal inhabitants of the marrow and so they are just “guys in the neighborhood.” The bone might not be aware that this increased number of these inhabitants or “guys” represents a disease and so this appears as a geographic margin without a sclerotic rim.
Benign lesions in the bone marrow may also not be recognized as foreign by the body and so they may generate no sclerosis surrounding the lesion. Benign enchondromas typically have an irregular but sharp geographic margin and should not demonstrate convincing growth during adulthood. Despite this, sclerosis is virtually never seen around these lesions (Figure 2). Remember that all long bones are the result of endochondral bone formation such that cartilage is functionally the bone’s parent. Walling off your parents is probably never a good idea and so bones do not do this either.
The next pattern of adjacent medullary bone destruction is described as “moth-eaten.” In this scenario, the interface from completely abnormal marrow signal to completely normal marrow signal as seen on radiograph is a few millimeters wide (Figure 3). In this zone of interdigitation, the tumor is chewing on the bone, but still causing this lysis via osteoclastic bone resorption. This is typical of tumors that have an intermediate grade of local aggressiveness within the bone. This is found in lesions such as giant cell tumor and chondrosarcoma. Most of these lesions are growing too quickly for the body to respond with a sclerotic rim and so there usually is none seen even if the bone recognizes the process as foreign.
The final and most aggressive pattern of local bone destruction is described as showing a permeative margin with adjacent medullary bone. In these cases, the tumor is rapidly advancing through the marrow cavity at such a rate that the tumor does not have time to destroy all of the medullary bone as it marches forward. As such, the width of the margin from totally abnormal to totally normal as seen on radiograph may be a distance of 2 to 5 cm. This is typical of lesions which are very locally aggressive such as osteosarcoma, Ewing sarcoma, infection, and some metastatic bone tumors (Figure 4). In these cases, there is no time at all for the bone to wall off these lesions and so typically no sclerotic rim is seen.
As a specific intramedullary bone tumor is destroying bone and growing within the bone, it may deposit matrix tissues which can be either purely radiolucent (lytic) or may possess calcified or ossified portions within them. Tumors that create a calcified matrix give some additional information as to the type of tissue involved in the tumor. Chondroid calcification, as seen in the enchondroma in Figure 2, has a typical punctate appearance described as “stippled calcification,” “rings and arcs,” or “popcorn calcification.” Ossification, on the other hand, indicates an underlying lesion that is forming bone histologically.
At some point in its growth, a tumor in the medullary canal will contact adjacent cortical bone. Again, the type of destruction that this interaction will create depends on the level of aggressiveness of the tumor involved. A slow-growing tumor which cannot permeate lamellar bone will cause lysis of the bone from the endosteal surface going out in a radial direction. The shape of this expansion mirrors the shape of the tumor itself. If the tumor is one large mass that is uniform in size within the medullary canal, the bone resorption will be one large erosion. If the tumor tissue itself is lobulated, as is typical for cartilage tumors, this cortical lysis will appear as multiple scallops into and through the endosteum into the cortex (Figure 5). As the tumor progressively thins the cortical bone adjacent to it, if the destruction is slow enough, the bone will respond by expanding around the lesion to maintain mechanical integrity of the bone. The net result of this process is a cortex that appears thinned and expanded. In cases of more aggressive destruction of cortical bone, the tumor is moving too quickly for the bone to respond and so gross cortical destruction is the result.
Once the tumor makes its way through the cortex, the next tissue it encounters is the periosteum. The periosteum is a sheet on the surface of bones and it only has one capability, namely making bone. In fact, it can do nothing else. Depending upon the aggressiveness or lack of aggressiveness of the incident or lesion stimulating the periosteum, an assortment of clinical scenarios may result.
Since periosteum is a sheet, it is never elevated in a punctate fashion but rather as a mound, like a small pillow put under a sheet on a bed. Again, as the periosteum is elevated it will begin to do the only thing it knows how to do, which is to make bone. The result will be a periosteal reaction which is either benign or aggressive.
If a sheet of periosteum is elevated but not otherwise traumatized or invaded, it will produce bone between itself and the underlying cortex and create a single-layer ossified periosteal reaction. This is a benign periosteal reaction and can be seen following nonprogressive interventions such as stress fracture or localized inflammation such as adjacent to an osteoid osteoma (Figure 6).
More locally aggressive lesions such as infection and malignant tumors are too difficult for the periosteum to completely contain. As such, the underlying lesion is like an army breaking through a defensive line. The army breaks through and then the defending soldiers try to regroup but cannot do so effectively. The army breaks through again and the same process happens repeatedly. This results in a very irregular periosteal reaction which can be described as “onion-skin” if it has layers that appear lamellated. This is a classic finding associated with Ewing sarcoma (Figure 7). Alternatively, the periosteal reaction may appear as thin wisps of bone at a roughly right angle to the long axis of the bone. This creates a picture described as “sunburst” or “hair-on-end.” This is typical for osteosarcoma (Figure 8).
It is not uncommon for some lesions to reach the periosteum but have no associated periosteal reaction. This is for one of two possible reasons. Either the tumor itself cannot get through the periosteum and so does not percolate through the periosteal tissues. An example of this is giant cell tumor which may produce significant cortical destruction but usually no periosteal reaction unless there is an associated fracture. In these cases, the tumor is not permeating through the periosteum and so the periosteum is intact on the surface and is not stimulated to make new bone.
The second scenario in which there may be no periosteal reaction is associated with those tumors that the periosteum does not recognize as foreign. It is common, for example, for lymphomas of bone to permeate the cortex and grow into the soft tissues without the presence of a significant periosteal reaction. As in the case of the myeloma discussed previously, it may well be that the periosteum is not equipped to recognize lymphocytes in a lymphoma as unusual. Again, in anthropomorphic terms, the periosteum just sees “folks in the neighborhood” and does not appreciate that this large number of “folks” represents something it should react to.
The final form of periosteal reaction seen is a mix of the first two. In this case, a lesion irritates the periosteum which again elevates as a sheet and there is an aggressive periosteal reaction in the center of the sheet. However, if the lesion does not reach the corners of the elevated periosteal sheet then that area of periosteum at one or both corners generates a single-layer benign-appearing periosteal reaction. This periosteal new bone which has been laid down in a single layer has a triangular appearance and is called a “Codman’s triangle” (Figure 9). This is simply a benign periosteal reaction at one or both ends of an aggressive periosteal reaction where the periosteum is beginning to get elevated by the soft tissue mass.
Describing a Radiograph
To communicate with one another, healthcare providers need to be able to describe the appearance of radiographic lesions they may encounter completely yet concisely. The following aspects of a bone lesion require comment:
- Name of the specific bone and laterality
- Size of the lesion in cm
- Location within the bone
a. Epiphyseal
b. Metaphyseal
c. Diaphyseal - Appearance of the lesion
a. Lytic
b. Blastic
c. Mixed lytic and blastic - Description of calcified matrix if present
a. Bone
b. Chondroid calcification
c. Uncertain - Position within the bone
a. Central
b. Eccentric
c. Expansile - Zone of transition with medullary bone
a. Geographic
b. Moth-eaten
c. Permeative - Presence or absence of surrounding sclerosis
- Cortical thinning or destruction
- Presence of a periosteal reaction
a. Benign
b. Aggressive
c. Codman’s triangle
A few examples follow.
Going back to Figure 3: This is a radiograph of the right/left distal femur of a skeletally mature individual showing a purely lytic lesion in the lateral/medial femoral condyle which is 7 cm in maximum width. The lesion has a moth-eaten zone of transition with adjacent medullary bone and significant cortical destruction but no periosteal reaction. This describes a typical giant cell tumor of bone (juxta-articular lytic lesion with a moth-eaten margin, cortical disruption, but no periosteal reaction).
Going back to Figure 8: This is a radiograph of a left distal femur in a skeletally mature individual showing an aggressive-appearing bone lesion that has a mixed lytic and blastic appearance. The proximal margin is permeative, as I cannot tell exactly where the tumor starts or stops. There is a soft tissue mass extending medially and laterally with Codman’s triangle periosteal reaction proximally and sunburst periosteal reaction laterally.
An alternative method of describing a radiograph is referred to as “Enneking’s four questions for bony lesions.” This is a reference to Dr. William Enneking who was a world-renowned pioneer in orthopedic oncology. Dr. Enneking used a system to describe bone lesions using the following four questions:
- Where is the lesion located?
- What is the lesion doing to the bone?
- What is the bone doing to the lesion?
- Are there any clues on the radiograph that would give information about the type of tissue in the lesion (e.g. calcified, blastic, or purely lytic)?
Whether you utilize the 10 steps outlined above or the Enneking method is a matter of personal preference.
Differential Diagnosis of a Bone Lesion
Differential diagnosis simply means that among the many lesions that are potentially encountered in bone, which are likely possibilities when considering a specific radiographic lesion. There are many systems that have been advocated to help a clinician develop a reasonably complete differential diagnosis. Personally, I have found many of them confusing and not user-friendly. Considering this, I am going to present my own attempt to simplify this for my own use as well as being a relatively simple method of teaching radiographic differential diagnosis to students and residents.
If you consider all the lesions that can present in bone, it becomes apparent that they actually represent a fairly small number of entities. This list of possible lesions becomes more manageable if it is divided into three shorter lists:
- Bone-forming tumors
a. Osteoid osteoma
b. Osteoblastoma
c. Osteochondroma
d. Osteosarcoma
e. Blastic metastasis
f. Paget’s disease - Cartilage-forming tumors
a. Enchondroma
b. Chondroblastoma
c. Chondromyxoid fibroma
d. Chondrosarcoma
e. Osteochondroma - “Third list”
a. Infection
b. Metastasis
c. Round cell tumors
i. Children: Ewing sarcoma, neuroblastoma
ii. Adults: myeloma, lymphoma, small round cell carcinoma
d. Fibrous dysplasia
i. Cortical osteofibrous dysplasia
e. Non-ossifying fibroma
f. Simple (unicameral) bone cyst
g. Aneurysmal bone cyst
h. Langerhans cell histiocytosis
i. Giant cell tumor of bone
j. Stress fracture
k. Metabolic conditions (e.g. hyperparathyroidism)
l. Very rare lesions
i. Adamantinoma
ii. Chordoma
In fact, these three lists include nearly every lesion you are likely to encounter throughout your orthopedic career. You can either memorize these lists and/or carry them with you. Once you are presented with a specific radiographic lesion, you can simply go through these three lists and either rule in or rule out each of these as being relevant or not relevant to the differential diagnosis of the lesion in question. In this way, you can easily assemble a very complete differential diagnosis for any bone lesion.
This obviously assumes that you are familiar with the classic radiographic findings associated with each lesion plus some normal variation. In fact, you need to know what each of these lesions looks like on radiograph as well as understanding their appearance on other imaging modalities such as nuclear medicine bone scan, CT scan, and MRI scan as well as the clinical context in which they typically appear. A comprehensive approach to the imaging of each entity will be handled in subsequent chapters.
Several radiographic scenarios can generate shorter differential diagnostic lists based on the specific radiographic appearance of the lesion along with the typical clinical context of the patient. These short differentials are listed below:
- Aggressive bone lesion in a young person: the major differential here includes osteosarcoma, Ewing sarcoma, and infection, which frequently mimics sarcoma, along with aneurysmal bone cyst and, once the growth plates are closed, giant cell tumor of bone. Repeat after me: osteosarcoma, Ewing sarcoma, infection, aneurysmal bone cyst, and giant cell tumor. Repeat it again until it is memorized!
- Aggressive bone lesions in adults: From the first list, osteosarcoma. From the second list, chondrosarcoma. From the third list, metastasis, lymphoma, myeloma, and giant cell tumor. A shortened way of listing the same tumors would be sarcoma, metastasis, adult round cell tumor, and giant cell tumor. I prefer to remember it as osteosarcoma, chondrosarcoma, metastasis, adult round cell tumor, and giant cell tumor. Repeat after me: osteosarcoma, chondrosarcoma, metastasis, adult round cell tumor, and giant cell tumor. Repeat it again until it is memorized!
- Lytic lesion in the epiphysis of a child: chondroblastoma versus infection. Pretty easy to memorize a list of two!
- Lesion in the anterior cortex of the shaft of the tibia: adamantinoma versus cortical osteofibrous dysplasia.
- Fibrous dysplasia: this is listed in the differential diagnosis of every benign-appearing lesion in bone as it has a very variable appearance ranging from a small lesion resembling a non-ossifying fibroma to a very large long lesion in a long bone to extensive polyostotic involvement.
- Langerhans cell histiocytosis: this is listed in the differential diagnosis of every intramedullary lytic lesion in a young person.
- Osteoblastoma: The first ten elements of the differential diagnosis of any lytic lesion of the posterior elements of the spine are all the same: osteoblastoma.
Hopefully, these shortcuts will help you quickly assemble a reasonable differential diagnosis but remember that you can always go back to the three lists