Classic “do not touch” and benign osseous lesions are sometimes detected on ¹⁸-F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) studies. These lesions are often referred for biopsy because the physician interpreting the PET/CT may not be familiar with the spectrum of ¹⁸F-FDG uptake patterns that these lesions display.

To show that “do not touch” and benign osseous lesions can have increased ¹⁸F-FDG uptake above blood-pool on PET/CT; therefore, the CT appearance of these lesions should dictate management rather than the standardized uptake values (SUV).

This retrospective study evaluated 287 independent patients with 287 classic “do not touch” (benign cystic lesions, insufficiency fractures, bone islands, bone infarcts) or benign osseous lesions (hemangiomas, enchondromas, osteochondromas, fibrous dysplasia, Paget’s disease, osteomyelitis) who underwent ¹⁸F-FDG positron emission tomography/computed tomography (PET/CT) at a tertiary academic healthcare institution between 01/01/2006 and 12/1/2018. The maximum and mean SUV, and the ratio of the maximum SUV to mean blood pool were calculated. Pearson’s correlations between lesion size and maximum SUV were calculated.

The ranges of the maximum SUV were as follows: For hemangiomas (0.95-2.99), bone infarcts (0.37-3.44), bone islands (0.26-3.29), enchondromas (0.46-2.69), fibrous dysplasia (0.78-18.63), osteochondromas (1.11-2.56), Paget’s disease of bone (0.93-5.65), insufficiency fractures (1.06-12.97) and for osteomyelitis (2.57-12.64). The range of the maximum SUV was lowest for osteochondromas (maximum SUV 2.56) and was highest for fibrous dysplasia (maximum SUV of 18.63). There was at least one lesion that demonstrated greater ¹⁸F-FDG avidity than the blood pool amongst each lesion type, with the highest maximum SUV ranging from 9.34 times blood pool mean (osteomyelitis) to 1.42 times blood pool mean (hemangiomas). There was no correlation between the maximum SUV and the lesion size except for enchondromas. Larger enchondromas had higher maximum SUV (r = 0.36, P = 0.02).

The classic “do not touch” lesions and classic benign lesions can be ¹⁸F-FDG avid. The CT appearance of these lesions should dictate clinical management rather than the maximum SUV.

¹⁸F-Fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) is increasingly utilized for staging and surveillance of many common malignancies[1,2]. Approximately 1.9 million ¹⁸F-FDG PET/CT studies were performed in the United States in 2017, a 13% increase compared to 2015[3]. Cancer cells often switch from oxidative to glucose metabolism, even in the presence of oxygen, resulting in aerobic glycolysis[4], first described by Warburg[5,6]. This change in cancer cell metabolism is easily detected and measured in vivo using ¹⁸F-FDG PET/CT[3]. The rapid rise in the number of ¹⁸F-FDG PET/CT scans performed annually has the potential to increase the number of incidental findings[7-9]. Simu-ltaneously, there has been increased specialization of radiology with most physicians that interpret ¹⁸F-FDG PET/CT not being simultaneously fellowship-trained in musculoskeletal radiology[10-14].

As a result, incidentally detected ¹⁸F-FDG-avid osseous lesions are often subject to sometimes inappropriate clinical management. A previous meta-analysis showed that although a general approach to ¹⁸F-FDG-avid incidental lesions may be recommended for selected organ systems, osseous lesions remain less amenable to blanket re-commendations for or against biopsy, for incidentally noted ¹⁸F-FDG-avid osseous lesions[15]. Anecdotally, we have noted a concomitant increase in request for biopsies of incidentally noted ¹⁸F-FDG-avid osseous lesions in clinical practice, that, on review by a fellowship musculoskeletal trained radiologist, do not warrant biopsy because these lesions are classic “do not touch lesions” including non-ossifying fibromas, bone islands/enostoses, unicameral bone cysts, bone infarcts, and geodes/subchondral cysts)[16] or have a classic computed tomography (CT) imaging appearance including fibrous dysplasia, Paget’s disease and enchondromas. The “do not touch” osseous lesions are mostly benign osseous lesions; however, some may rarely undergo malignant degeneration. The term “do not touch” was coined by Clyde Helms, and refers to lesions that the radiographic/CT appearance is pathognomic, and additional diagnostic tests, biopsies and surgery involving these lesions may be misleading, potentially harming the patient[16]. Numerous case reports exist in the literature demonstrating that several of these benign lesions have resulted in ¹⁸F-FDG-avid osseous lesions and were confirmed histologically by subsequent biopsy[17-22].

While image-guided percutaneous core needle biopsy of osseous lesions is generally regarded as a low-risk procedure, potential complications exist, including patient discomfort and anxiety, infection, bleeding, and the possibility of a non-diagnostic specimen. Non-diagnostic biopsies may occur in 5-29% of cases and may lead to repeat percutaneous or subsequent open biopsy[23,24]. Although the CT imaging characteristics of an osseous lesion may indicate its non-aggressive etiology, clinicians may persistently request biopsies of ¹⁸F-FDG-avid osseous lesions in patients that have a primary malignancy elsewhere due to concerns of under-staging or clinical concerns centered around missing osseous metastases.

To date, despite multiple case reports of ¹⁸F-FDG-avid benign osseous lesions in the literature, to the best of our knowledge a descriptive analysis of the ¹⁸F-FDG uptake of common benign osseous lesions in the musculoskeletal system has not been per-formed. There are no prior reports demonstrating the spectrum of ¹⁸F-FDG uptake patterns of several common benign skeletal osseous lesions with identifiable CT imaging characteristics. If this data existed, then it could be used as a guide for physicians that primarily interpret ¹⁸F-FDG PET/CTs and to eliminate referral of these benign lesions for biopsies. The aim of the study is to show that “do not touch” and benign osseous lesions can have increased ¹⁸F-FDG uptake above blood-pool, the-refore the CT appearance of these osseous lesions should dictate management rather than the PET/CT standardized uptake values (SUV).

There were 287 patients with either classic “do not touch” lesions or classic benign lesions. Patient characteristics are summarized in Table 1. Figure 1 shows the ¹⁸F-FDG avidity of each lesion type. Benign cystic lesions included subchondral cysts, bone cysts, herniation pits, and geodes. The maximum SUV detected was lowest for osteochondromas (maximum SUV 2.56), enchondromas (maximum SUV 2.69), and hemangiomas (maximum SUV of 2.88). The maximum SUV was highest for Paget’s disease of bone (maximum SUV of 5.65), benign cystic lesions (maximum SUV of 6.5), osteomyelitis (maximum SUV of 12.64), insufficiency fractures (maximum SUV of 12.97), and fibrous dysplasia (maximum SUV of 18.63) (Table 2). There was at least one lesion amongst each lesion type evaluated that demonstrated greater ¹⁸F-FDG avidity than the blood pool, with the highest maximum SUV ranging from 9.34 times blood pool mean (osteomyelitis) to 1.42 times blood pool mean (hemangioma) (Table 2, Figure 1A and 1B).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Plots of maximum standardized uptake value values, mean standardized uptake value/mean blood pool, and maximum standardized uptake value/blood pool by lesion size for each lesion type. SUV: Standardized uptake value; Max SUV: Maximum standardized uptake value; Mean SUV/Mean B: Mean standardized uptake value divided by the mean blood pool uptake; Max SUV/Mean B: Maximum standardized uptake value divided by the mean blood pool uptake.

There was no correlation between the maximum SUV and the size of benign cystic lesions (r = 0.08, P = 0.60), hemangiomas (r = 0.15, P = 0.44), bone infarcts (r = 0.12, P = 0.37), fibrous dysplasia (r = 0.13, P = 0.48), osteochondromas (r = 0.04, P = 0.95), Paget’s disease of bone (r = 0.16, P = 0.50), insufficiency fractures (r = 0.01, P = 0.96) and osteomyelitis (r = 0.15, P = 0.60). However, the size of enchondromas was associated with the maximum SUV (r = 0.36, P = 0.02).

The “do not touch” and benign lesions evaluated demonstrated a range of SUV values. All evaluated “do not touch” and benign lesions had at least one lesion with maximum SUV that was greater than and at least one lesion with maximum SUV that was less than the mean blood pool SUV. For this reason, the results suggest that the maximum SUV is not a reliable tool for the characterization of osseous lesions, pa-rticularly as benign lesions such as fibrous dysplasia can demonstrate a maximum SUV up to 8.87 times the average blood pool SUV.

These findings are consistent with multiple case reports in the literature that have described pathology-proven benign lesions with high ¹⁸F-FDG avidity prompting surgical excision on the basis of the ¹⁸F-FDG uptake[7,17-22]. Our results reaffirm that, when specific, the CT characteristics of such lesions need not be accompanied by an SUV at or below that of the blood pool to confidently diagnose a lesion as benign. In fact, SUV values may be misleading, and over-reliance on ¹⁸F-FDG uptake to characterize a lesion as benign or malignant may lead to over-diagnosis and un-necessary procedures such as biopsies, which although low-risk, are not risk-free.

Benign cystic lesions, including bone cysts, geodes, and herniation pits, have in common a well circumscribed, lytic appearance [27-30]. All may have a sclerotic rim, although such a rim is only occasionally present in simple bone cysts [27]. The terms geode and herniation pit both describe lesions arising in the setting of reactive changes at the surface of bone, although herniation pits have classically been described in the femoral neck, remote from the articular surface, whereas geodes are subarticular in location and regarded as synonymous with subchondral cysts (Figure 2), although these lesions lack an epithelial lining and are not true cysts[31]. Subchondral cysts/geodes are often accompanied by other signs of joint degeneration including subchondral sclerosis and eburnation as well as marginal osteophyte formation. Subchondral cysts are more common in the elderly. These lesions are lytic on CT and may show increased ¹⁸F-FDG uptake on PET.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Axial computed tomography and axial fused positron emission tomography/computed tomography of the pelvis in an 82-year-old male with history of colon and parotid gland cancer with subchondral cyst in the right femoral head (arrows) (maximum standardized uptake value 5. 2).

Osseous hemangiomas are most commonly seen in the vertebral bodies, but may occur in the calvarium, calcaneus, and long bones [28]. Vertebral body hemangiomas have a characteristic striated appearance on radiographs and a corresponding “polka dot” appearance on axial CT images (Figure 3A)[28]. On MRI, hemangiomas typically have high signal on both T1- and T2-weighted images[32] and commonly show loss of signal on in- and out-of-phase gradient sequences[33]. The PET/CT appearance is similar to the CT appearance, with some lesions showing increased ¹⁸F-FDG uptake on PET.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Computed tomography and fused positron emission tomography/computed tomography of two patients. A: Sagittal computed tomography (CT) and sagittal fused positron emission tomography (PET)/CT of the spine in a 73-year-old male with history of lymphoma and a hemangioma in the T5 vertebral body (arrows) [maximum standardized uptake value (SUV) 2.86]; B: Coronal CT and coronal fused PET/CT of the distal tibia in a 54-year-old female with history of a neuroendocrine lung tumor demonstrating a bone infarct (arrows) in the distal tibia (maximum SUV 0.37).

The term “bone infarct” is commonly used to refer to osteonecrosis seen in the metaphyses and diaphyses of bones [34], whereas osteonecrosis occurring in the epiphysis is commonly referred to as avascular necrosis. On CT, these lesions may demonstrate a serpentine rim of sclerosis (Figure 3B), although this may not be seen early in the disease course [32]. Bone infarcts have a similar appearance on PET/CT to CT, but may show variable amounts of ¹⁸F-FDG uptake on PET. On MRI, these lesions demonstrate a characteristic serpentine rim of low signal with variable internal signal [34]. Bone infarcts rarely undergo malignant degeneration, most commonly into osteosarcoma, pleomorphic sarcoma and fibrosarcoma, but here malignant degeneration is characterized by development of a soft tissue mass, cortical destru-ction and in some cases development of a pathological fracture [35].

Bone islands, or enostoses, are benign osteoblastic lesions consisting of dense intramedullary lamellar bone that appear on CT as hyperdense oval lesions with spiculated margins and are classically periarticular (Figure 4A)[28]. The data presented show that enostoses appear similar on PET/CT to CT and may show ¹⁸F-FDG uptake on PET above background. Although osteoblastic metastatic lesions may have a similar appearance, bone islands can be identified with high sensitivity and specificity by their higher attenuation value on CT[36]. On MRI, osteoblastic metastases may also show a halo of surrounding bone marrow edema, which may help distinguish these lesions from bone islands[28].

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Axial computed tomography and axial fused positron emission tomography/computed tomography of two patients. A: Axial computed tomography (CT) and axial fused positron emission tomography (PET)/CT of the pelvis in a 76-year-old male with a history of a solitary pulmonary nodule and a bone island in his left ilium (arrows); B: Axial CT and axial fused PET/CT of the chest in a 52-year-old female with a history of melanoma and an enchondroma in the left proximal humerus (arrows) (maximum standardized uptake value 2.33).

Enchondromas are benign lesions that typically occur within long bones. Enchondromas histologically represent rests of hyaline cartilage within medullary bone. Radiographically, enchondromas are expansile lucent lesions with a narrow zone of transition and varying degrees of ring-and-arc mineralization (Figure 4B)[28], although mineralization may be absent when these lesions are seen in the hands and feet. Enchondromas are almost never seen involving the flat bones (ribs, pelvis, scapula) or spine. Distinguishing between enchondromas and chondrosarcomas can be challenging, both by imaging features and by histology[37]. Malignant degeneration may be seen and is more common (20%-50%) in patients with Ollier disease or Maffucci syndrome[28]. Malignant degeneration is characterized by increase in size of lesion after skeletal maturity and pain[37,38]. A prior report suggested that a maximum SUV > 4.4 was 99% specific for grade 2/3 chondrosarcoma[39]. Biopsy of low-grade chondroid matrix lesions, such as enchondromas, is of limited utility because enchondromas may be histologically mistaken for low-grade chondrosarcomas and vice versa. Enchondromas have a similar appearance on PET/CT to CT and ra-diographs, but may show variable amounts of ¹⁸F-FDG uptake on PET. The data suggest that the size of the enchondroma is associated with higher maximum SUV; however, it is unclear whether this confers a potentially higher risk for chon-drosarcoma. Development of a soft tissue mass with cortical destruction is highly suspicious for malignant degeneration.

Fibrous dysplasia is due to a post-natal sporadic mutation in the G-nucleotide binding protein alpha sub unit (GNAS) and results in development of fibrous tissue with ground-glass matrix on radiographs and CT studies[40]. Fibrous dysplasia may have areas of cystic change on CT. The PET/CT appearance of fibrous dysplasia is identical to its appearance on CT; however, fibrous dysplasia may be metabolically active and show increased ¹⁸F-FDG uptake. Fibrous dysplasia may be monostotic or polyostotic. McCune-Albright syndrome is associated with polyostotic fibrous dysplasia, café au lait macules and hyperfunctioning endocrinopathies, which may include precocious puberty, hyperthyroidism or Cushing’s syndrome. Mazabraud syndrome is characterized by fibrous dysplasia with intramuscular myxomas. Fibrous dysplasia typically affects ribs and long bones (femur, tibia, and humerus), and may result in a Shepherd’s crook deformity of the femurs. Fibrous dysplasia may involve the bones of the face/jaw and result in disfigurement, or involve the spine and result in scoliosis. Approximately 1% of patients will have a lesion that undergoes malignant degeneration, most commonly into osteosarcoma, fibrosarcoma or undifferentiated pleomorphic sarcoma[40,41]. Malignant degeneration is characterized by development of an osteolytic component with cortical destruction[41].

Osteochondromas are tumors of the bone that may be sporadic or inherited. These are bony exostoses that show contiguity with the intramedullary canal on CT studies (Figure 5A). These exostoses have a cartilage cap that may be along various stages of ossification. Osteochondromas should not grow after skeletal maturity. Osteo-chondromas may be sessile or pedunculated, and if pedunculated point away from the nearest joint. Osteochondromas are benign tumors, but may rarely undergo malignant degeneration into a chondrosarcoma. Malignant degeneration is more common in patients with multiple hereditary exostoses (MHE). A thickened cartilage cap on CT, greater than 2.0 cm, has been reported as 100% sensitive and 95% specific for degeneration into chondrosarcoma[42]. The communication of the bony excrescence with the intramedullary space on PET/CT is pathognomic; however, osteo-chondromas should not be confused with tubercles/muscle insertion sites. Osteochondromas may show increased ¹⁸F-FDG uptake on PET/CT.

Open in New Tab   Full Size Figure   Download Figure

Figure 5 Axial computed tomography and axial fused positron emission tomography/computed tomography of two patients. A: Axial computed tomography (CT) and axial fused positron emission tomography (PET)/CT of the pelvis in a 51-year-old male with diffuse large B-cell lymphoma and a sessile osteochondroma arising from the proximal left femur (arrows) [maximum standardized uptake value (SUV) 1.64]; B: Axial CT and axial fused PET/CT of the pelvis in a 61-year-old male with a history of a follicular lymphoma and Paget’s disease of the bone involving the right ilium (arrows) (maximum SUV 3.92).

Paget’s disease of the bone is a disorder of normal bone homeostasis with abnormal osteoblastic and osteoclastic activity. The exact etiology of Paget’s disease is unknown; however, Paget’s disease may present in a mixture of three phases: the osteolytic phase, the intermediate phase, and the quiescent phase. Pagetic bone in osteolytic phase may show osteolytic lesions, like osteoporosis circumscripta cranii; however, Pagetic bone in the intermediate and quiescent phase shows cortical and trabecular thickening on CT (Figure 5B). The PET/CT appearance of Paget’s disease is identical to that noted on CT; however, Paget’s disease may show variable ¹⁸F-FDG uptake, likely depending on the phase of Paget’s disease. A proximal femur shepherd’s crook deformity may be present. Paget’s disease tends to be localized affecting adjacent bones. The cotton wool spots/patchy areas of sclerosis of Paget’s disease are often mistaken for osteoblastic metastases, in particular metastases related to prostate or breast cancer. While Paget’s disease is benign, Pagetic bone may rarely undergo malignant degeneration into undifferentiated pleomorphic sarcoma, fibrosarcoma, or osteosarcoma amongst other sarcomas[43]. Development of a soft tissue mass with cortical destruction should be considered highly suspicious for malignant degeneration of Paget’s disease.

Insufficiency fractures can occur in patients after radiation, after chemotherapy, or after hormonal-altering therapy for some cancers including breast and prostate cancer, or be related to diminished bone mineral density related to senescent changes. Insufficiency fractures are more common in women. Acute sacral insufficiency fractures appear as unilateral or bilateral para-median linear lucencies with adjacent sclerosis oriented anterior-posterior along the sacral ala on CT (Figure 6A); however, a transverse component may be present, resulting in a “Honda” or “H” sign on Tc99m-MDP bone scans and PET/CT. Insufficiency fractures may also be noted involving the iliac wing, medial ilium, supraacetabular region, parasymphyseal region, and involving the pubic rami[44]. Proximal femoral insufficiency fractures are more common along the compression than tension side of the hip; however, insufficiency fractures related to bisphosphonate therapy are more common along the tension side of the sub-trochanteric proximal femur.

Open in New Tab   Full Size Figure   Download Figure

Figure 6 Axial computed tomography and axial fused positron emission tomography/computed tomography of two patients. A: Axial computed tomography (CT) and axial fused positron emission tomography (PET)/CT of the pelvis in an 84-year-old female with a history of non-Hodgkin’s lymphoma and an insufficiency fracture of the right sacral ala (arrows) [maximum standardized uptake value (SUV) 1.99]; B: Axial CT and axial fused PET/CT of the pelvis in an 82-year-old female with a sacral decubitus ulcer and osteomyelitis (arrows).

Osteomyelitis is an infection of the intramedullary canal and cortex of the bone. Signs of osteomyelitis may include elevated white blood cell counts (WBC), erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). Clinical history and physical examination are often required because the yield for bone biopsies for osteomyelitis is generally less than 50%. CT findings of osteomyelitis include periosteal reaction, sclerosis, osseous erosion/destruction, surrounding soft tissue inflammatory changes and reactive lymphadenopathy (Figure 6B). These areas of destruction may show increased ¹⁸F-FDG uptake depending on chronicity on PET/CT. Longstanding chronic ulcers (Marjolin’s ulcers) related to chronic infection, burns, injuries, or venous stasis and chronic osteomyelitis may undergo malignant degeneration into squamous cell or basal cell carcinoma [44,45]. Signs of malignant degeneration include development of bone destruction, soft tissue mass and per-iosteal reaction[44,45].

These findings are clinically important for radiologists and clinicians, because we have described the key CT imaging features of these “do not touch” and benign lesions, which can be gleaned from the CT component of the PET/CT. We anticipate that this will decrease the number of times the PET/CT interpreting physician and the bone biopsy proceduralist have differing opinions and reduce the number of un-necessary bone biopsies.

This study has a few limitations. The data are obtained from a single center tertiary care academic healthcare center, and therefore subject to ascertainment bias since our institution primarily treats adult patients. Lesions that are more common in the pediatric population including cortical desmoids and avulsion fractures were not described. SUV measurements, while standardized by weight, acquisition time, and injected dose, may be affected by body mass index, scanner calibration, and imaging artifacts, among other factors[26]. To address this limitation, blood pool mean SUV was included as an internal control to better characterize the FDG uptake of the target lesions. Several lesions were not histologically confirmed – this was because these lesions had the classic imaging appearance of a benign lesion or a “do not touch” lesion[16], and biopsies of these lesions are generally thought to be unnecessary since characterization by a trained musculoskeletal radiologist is pathognomic. The sample size was limited, but these are rare lesions, and prior publications referencing these lesions are almost all are restricted to case-reports. Nonetheless, this work clearly shows the spectrum of ¹⁸F-FDG uptake in several known “do not touch” and benign osseous lesions, which can be used to help in clinical decision making.

In conclusion, among benign lesions for which CT imaging findings are diagnostic, SUV values may vary, and these lesions may demonstrate ¹⁸F-FDG uptake above that of the blood pool. For this reason, ¹⁸F-FDG uptake above background should not be misconstrued to indicate the presence of malignancy for all osseous lesions.