Skeletal Radiology

, Volume 34, Issue 7, pp 367–374

18FDG-PET applications for cartilage neoplasms

Authors

    • Department of RadiologyNew York Presbyterian Medical Center
  • Ronald Van Heertum
    • Department of RadiologyNew York Presbyterian Medical Center
  • Chitra Saxena
    • Department of RadiologyNew York Presbyterian Medical Center
  • May Parisien
    • Department of Surgical PathologyNew York Presbyterian Medical Center
Scientific Article

DOI: 10.1007/s00256-005-0894-y

Cite this article as:
Feldman, F., Heertum, R.V., Saxena, C. et al. Skeletal Radiol (2005) 34: 367. doi:10.1007/s00256-005-0894-y

Abstract

Objective

To assess the value of [18F]fluoro-2-deoxy-d-glucose positron emission tomography (18FDG-PET) in defining aggressive cartilage neoplasms, particularly those with problematic or borderline histologic, imaging and clinical characteristics.

Design and patients

From 2000 to 2003, 29 cartilage lesions were studied with whole-body 18FDG-PET scans (Siemens Ecat Exact, Knoxville, Tenn.). Analyses of data in 20 females and nine males, 11–85 years old, were based on maximum standard uptake values (SUVs) in regions of interest (ROIs) on axial 3.37 mm thick, 3×3 pixel images. A statistically significant maximum SUV cutoff of 2.0 was used to distinguish benign from malignant cartilage neoplasms and correlated with the postoperative histopathologic findings.

Results

In 26 operated cases the overall sensitivity of whole-body 18FDG-PET in separating benign and malignant lesions was 90.9% (10/11), specificity 100% (18/18) and accuracy 96.6%.

Conclusions

Whole-body 18FDG-PET is a valuable adjunct in identifying primary, recurrent and metastatic cartilage malignancies. It supplements classic histology and morphologic imaging with functional data which may facilitate management in individual cases.

Keywords

CartilageNeoplasmsPETBone tumors

Introduction

Cartilage lesions are difficult to compartmentalize histologically since their spectrum includes an often problematic, gray middle zone in which distinction between benign and malignant members may be blurred [1, 2, 3, 4]. Imaging modalities are likewise limited, but positron emission tomography (PET), with currently simplified clinical use, is providing a new functional parameter for evaluating varied histologically unrelated neoplasms [5, 6, 7]. PET, by detecting hypermetabolic foci, could thereby improve the conspicuity of residual, recurrent and metastatic tumors in view of the difficulties found with morphologically oriented imaging modalities in identifying abnormality in distorted post-treatment anatomy and large heterogeneous lesions. However, its application to musculoskeletal oncology in general and cartilage tumors in particular has not been extensively explored [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19].

The purpose of this study was to evaluate 18FDG-PET as a problem-solver specifically in cartilage lesions by utilizing its functional data to distinguish benign lesions from their malignant counterparts in view of their often problematic clinical, imaging and histologic findings.

Materials and methods

Patients and 18FDG preparation for PET scans

Patients were NPO for 6 h with fasting blood glucose levels <160 mg/dl. Intravenous access was obtained 15–30 min before injecting 0.14 mCi/kg of [18F]fluoro-2-deoxy-d-glucose (18FDG). Its preparation utilized the nucleophilic reaction of 18F-fluoride with a protected mannose molecule [19]. Feet-first scans starting 60 min post-injection had a 60 min transmission corrected acquisition time for serial images.

Acquisition processing and display

PET studies were performed on a Siemens ECAT EXACT 921 dedicated PET scanner (Knoxville, Tenn.) with FWH=6 mm. A total of five or six bed examination (6 min emission, 4 min transmission) images used OSEM iterative reconstruction with segmented transmission attenuation corrected with Gaussian filters (cutoff 6–7 mm). Coronal, axial and sagittal 1.3 cm thick reconstructions were displayed with inverted gray-scale maps and successive 6 mm multiplanar slices of the body examined.

Image interpretation and analysis

Visual and region of interest (ROI) 3×3 pixel image analyses were based on axial whole-body 3.37 mm images. ROI data generated mean and maximum standard uptake values (SUVs), where SUV is defined as the radioactive tissue concentration in becquerels per gram divided by the body weight in grams. The maximum SUV cutoff of 2.0 between benign and aggressive lesions was based on our prior experience with various neoplasms studied with current equipment. Institutional Review Board approval was obtained for the study.

Case material

From 2000 to 2003, 29 cartilage lesions in nine males and 20 females, 11―85 years old, were referred for radiographs, CT or MRI and whole-body PET scans by various clinicians. Of 11 enchondromas, seven osteochondromas and 11 chondrosarcomas, 26 were surgically excised and histologically evaluated. Three nonoperated cases included two presumed enchondromas (cases 8 and 9) and one chondrosarcoma (case 28) with an incidentally noted breast mass preferentially referred for treatment. PET analyses based on a statistically significant (P=0.001) maximum SUV cutoff of 2.0 separated benign from malignant lesions.

Results

Benign lesions

In 11 benign enchondromas SUV ranged from 0.8 to 1.8 (Figs. 1, 2). In seven benign exostoses SUV ranged from 0.9 to 1.4 (Table 1).
Fig. 1A―H

A 62-year-old woman with left distal femoral pain. A Radiograph and B coronal T1-weighted, C sagittal T1-weighted, D axial T1-weighted, E coronal T2-weighted fat-suppressed FSE, F sagittal and G axial T2-weighted fat-suppressed FSE MR images show a large, inhomogeneously calcified diametaphyseal femoral lesion with endosteal scalloping with an SUV of 1.4 on 1 h coronal PET. Histology: Enchondroma

Fig. 2A, B

An 80-year-old woman. A Radiograph shows a large, calcified lesion involving most of the right humeral medullary cavity with lateral cortical thinning. B Coronal PET shows symmetrically hypometabolic humeri (SUV 1.3) and an unrelated left inflammatory axillary node. Histology: Enchondroma

Table 1

The efficacy of 18FDG-PET in cartilage neoplasms (H humerus, F femur, IL ilium, T tibia, P proximal, D distal)

Site

SUV

Type

Grade

Case no.

Age (years)/Sex

Primarya site metastases

Pathology

Enchondroma

1

42/F

L PH

1.7

All solitary

2

53/F

R PH

1.6

3

39/F

R DF

1.8

4

82/F

L PH

1.3

5

62/F

R DF

1.4

6

23/M

L DF

0.9

7

45/F

L DF

0.8

8

54/F

L PH

1.5b

9

77/F

R PH

1.6b

10

37/F

R DF

0.9

11

81/F

R PH

1.2

Exostoses

12

33/M

L PF

0.9

All multiple

13

12/M

L PT

1.3

14

14/F

R SC

1.2

15

20/M

R PF

0.9

16

34/F

L IL

1.4

17

64F

R Hand

1.2c

18

23/M

L DF

1.1c

Chondrosarcoma site of origin

Peripheral intramedullary

19

49/M

R PH

6.0

Muscle, axilla 8

Dedifferentiated to fibrosarcomaa

3

20

85/F

L DF

12.4

Dedifferentiated to fibrosarcomaa

3

21

80/F

L DF

1.4d

1

Axial intramedullary

22

57/F

R IL

3.0

Ilium, spine, rib, vertebra, lung 3–20

Mesenchymal

3

23

46/M

Soft tissue

4.0

Lung, mediastinum, pelvis, femora 5

Myxoid

2

24

51/F

Retroperitoneum

2.4

Psoas, ilium 3

Myxoid

2

25

62/F

Retroperitoneum

2.4

Axial exostosis

26

32/M

Pubis

3.4

Solitary

1

27

42/M

Sacrum

3.0

Multiple

1

28

64/F

R pubis

4.6

Multiple

***

Axial enchondroma

29

29/F

L acetabulum

3.0

Multiple

1

aRecurrence dedifferentiated to fibrosarcoma

bNO OR: case 28 referred for left breast carcinoma (SUV 3.3), case 20 referred for right breast carcinoma (SUV 2.4) following femoral resection

cMild focal atypia

dSUV <2.0; histological diagnosis: low-grade chondrosarcoma?, “bizarre cells”

Malignant lesions

Ten of 11 chondrosarcomas had SUVs >2.0 (Figs. 3, 4, 5, 6). SUVs of grade 1 chondromas ranged from 1.4 to 3.4, grade 2 from 2.4 to 4.6, and grade 3 from 6 to 20. Their metastases generally had higher SUVs compared with the primary loci, particularly those with fibrosarcomatous dedifferentiation in primary and metastatic sites(cases 19 and 20; Figs. 7, 8; Table 1). One intramedullary chondrosarcoma with SUV <2 (case 21) was histologically classified as low grade due to “degree of cellularity, cytological atypia and frequent binucleation.” Two cases (cases 17 and 18; SUV 1.2, 1.1) with “mild histologic atypia” are additional examples of “borderline” lesions which some pathologists have graded 0 to 1/2 [20, 21]. In 1992, 79 of 800 bone chondrosarcomas reviewed at the Mayo Clinic as having “clinical and radiographic features suggesting malignancy, with histologic features intermediate between benign enchondroma and frank chondrosarcoma, were histologically graded as 1/2 or borderline. After curetting and bone grafting, none had any evidence of disease after long-term follow-up” [20]. Our cases were similarly treated and followed, as was case 21 (SUV<2), histologically classified as low-grade chondrosarcoma. None recurred after 4 year follow-up.
Fig. 3A, B

A 29-year-old woman with multiple enchondromatosis and left hip pain. Two right hand enchondromas had been excised 5 years previously. Radiographs, CT and MRI revealed a left acetabular lesion. A Coronal and B axial PET slices show a left acetabular hypermetabolic focus (SUV 3.0). Histology: Grade 1 chondrosarcoma

Fig. 4A, B

A 56-year-old woman with right hip pain who had had a maxillary chondrosarcoma resected 5 years previously. Radiographs, CT and MRI showed a right iliac fracture extending to the acetabulum. Coronal PET image showed hypermetabolic metastases to the sacrum, right ilium contiguous with the pelvic soft tissues (right lower arrows, SUV 3) and right rib (right upper arrow), lung (left upper arrow) and spine lesions (SUVs 3–20). Histology: Grade 2 mesenchymal chondrosarcoma

Fig. 5A, B

A 46-year-old man 2 years after surgery, chemotherapy and radiotherapy to neck, shoulder and buttock lesions elsewhere, admitted with left groin and buttock pain. A Coronal and B sagittal PET show femoral prosthesis (right arrows), hypermetabolic foci in the left thigh and buttock (arrows; SUV 4) and left femoral, lung (arrow) and mediastinal metastases (SUVs 3–5). Histology: Metastatic myxoid chondrosarcoma

Fig. 6

A 52-year-old woman who had undergone resection of a right retroperitoneal chondrosarcoma, radiotherapy and chemotherapy (1988), resection of lung metastases (2002) and was admitted (2003) for right flank hernia. Coronal PET shows recurrent tumor in the hernia abutting the liver (right middle arrow), iliopsoas, bladder, pelvic (right lower arrow) and mediastinal metastases (right upper arrow; SUV 3). Histology: Chondrosarcoma

Fig. 7

A 53-year-old man who had undergone excision of an initially diagnosed right proximal humeral enchondroma (1996), a humeral prosthesis for “recurrent” chondrosarcoma (1997) and chemotherapy for lung metastases (1999). Coronal PET (2000) shows a defect due to the humeral prosthesis (right upper arrow), and muscle and axillary tumor extension (lower arrow). Its “donut” configuration reflects peripheral hypermetabolic viable tumor (SUVs 6–8) with central hypometabolism due to hypovascularity, necrosis and mineralization. Histology: Chondrosarcoma with fibrosarcomatous dedifferentiation

Fig. 8A―D

An 85-year-old woman had undergone right breast cancer lumpectomy (2001; no nodes); uterine cancer radiated in 2002 (no breast recurrence noted). Radiographs (1997–1998) were read as unchanged infarct or cartilage lesion. CT and MRI (2003) showed a left femoral medullary, calcified lesion with extraosseous extension, progressive since December 20, 2002. PET was carried out in February 2000. A Coronal and B axial PET show distal left femoral hypermetabolism (SUV 12.4; arrows). C Axial and D sagittal thoracic slices show a right breast lesion (SUV 2.4; arrows). Femoral histology: Chondrosarcoma with fibrosarcomatous dedifferentiation. Referred for mammogram

Outcomes

Outcomes were monitored by follow-up appointments with referring physicians.

Benign lesions

Among 16 operated benign lesions, six of nine enchondromas did not recur after 1―4 years (1 year in 1 case, 2 years in 2 cases, 3 years in 2 cases, 4 years in 1 case). Three remaining patients with enchondroma who were asymptomatic postoperatively did not return.

Among six of seven benign exostoses, none recurred after 1―4 years of follow-up (1 year in 4 cases, 2 years in 1 case, 3 years in 1 case). One patient did not return for follow-up.

Malignant lesions

Among 10 patients who were operated on for chondrosarcoma, three died of metastases and three were alive with local extension, metastases and periodic tumor debulking (9 years after surgery in 1 case, 1 year in 2 cases). Two cases did not recur after 3 and 2 years follow-up, respectively. The latter patient with a wound infection left the country. One had no recurrence 1 year after amputation. One case with “atypical cells” listed as grade 1 chondrosarcoma (SUV<2.0) has not returned since 2000. Based on a statistically significant (P=0.001) maximum SUV cutoff of 2.0 in ROIs, overall sensitivity for differentiating malignant and benign lesions was 90.9% (10/11) in 26 operated cases, with an overall specificity of 100% (18/18) and an accuracy of 96.6% (Fig. 9).
Fig. 9

18FDG-PET SUV values of benign and malignant cartilage neoplasms

Discussion

PET differs from other single-photon radionuclide scans in its ability to correct for tissue attenuation signal loss and its relatively uniform spatial resolution. Regional tracer measurement in absolute units, difficult in planar and SPECT imaging, allow PET data conversion into functional images, since each pixel in an anatomic area has a numerical value measuring a physiologic parameter. While most imaging modalities depict morphology, the initial rationale for PET’s estimation of aggressiveness was the avidity of malignant cells for glucose [5]. 18FDG ([18F]fluoro-2-deoxy-d-glucose) is a favored metabolic tracer due to its intracellular transport mechanism, indistinguishable from that of glucose, its increased uptake in cells with high metabolic rates and similarly increased utilization in malignancies. Low membrane permeability also limits FDG back-diffusion during PET, with the FDG trapped in tumor cells determining glucose utilization.

Some early studies [8] failed to correlate glucose metabolism and aggressiveness. Others held that 18FDG distinguished benign and malignant lesions [9, 10, 11, 12, 13]. Differences may be due to older PET technology or inclusion of bone or soft tissue tumors of diverse histology and biologic behavior. This study was therefore confined to cartilage lesions.

Clinical, imaging and histologic guidelines have inconsistently distinguished between benign and malignant cartilage lesions, with chondrosarcomas classically being held to be more often painful, palpable, proximal, larger and commoner in older patients than their benign counterparts. Large, asymptomatic, incidentally encountered lesions still pose diagnostic and management problems. Pain is also unreliable as an indicator of malignancy. Benign lesions may be symptomatic due to impingement on local anatomy while chondrosarcomas, despite age or size, may be painless and incidentally noted. Imaging criteria for malignant cartilage lesions, i.e., endosteal scalloping, remodeling, post-contrast enhancement on MRI and enlarged cartilage cap exostoses, are suggestive rather than conclusive evidence of malignancy, with variations in age and overlap continuing to be noted [1, 2, 3, 4, 20, 21].

Radiographs and CT best define mineralization and subtle periosteal changes. MRI best depicts morphology, anatomic extent and enhancement patterns which favor chondrosarcoma but are inconsistent. T1-weighted signal in cartilage, often lower than in muscle, usually increases in viable tumor on T2-weighted images. However, the converse can occur, especially in mature or heavily calcified cartilage lesions, so that inconstant MRI patterns may be less helpful than radiographs in distinguishing malignancy.

Pathologists have also had well-known difficulties in differentiating benign from low-grade cartilage tumors [1, 2, 3, 4, 20, 21], with tissue, depending on its origin and patient age, variously interpreted. Tissue from peripheral tubular bones that is considered benign may be problematic when originating in long bone intramedullary or cortical sites in multiple enchondromatoses or exostoses and in large symptomatic solitary lesions, particularly in the young with their often increased cellularity [20, 21]. 18FDG-PET, by reflecting tumor metabolism in such cases, aids in distinguishing aggressive from indolent behavior. It supplements static morphologic evaluation by helping resolve inconsistent clinical data, i.e., no pain, incidentally detected lesions, both small and large, those without extraosseous extension on MRI and “atypical or borderline histology” as in our cases 17, 18 and 21 (Table 1).

Cartilage, like other individual cell types, should be considered separately. Unique properties shared by its benign and malignant counterparts include relative hypovascularity, gelatinous extracellular matrix and predominant anaerobic glycolysis [14). Vascularity and metabolism, even of aggressive cartilage lesions, seldom reach the SUVs of osteosarcoma, fibrosarcoma or Ewing tumor [12, 13, 14], excepting mesenchymal variants (case 22; Fig. 4), dedifferentiated lesions (cases 19 and 20; Fig. 7) and their metastases. Well-differentiated grade 1 or 2 chondrosarcomas and large, mature, heavily calcified or focally necrotic lesions may have relatively lower SUVs reflecting hypometabolism despite their size [12, 15, 16, 17]. Necrotic foci may also be ringed by “donut-like” viable peripheral tissue (Fig. 7). FDG, by “mapping” haphazardly distributed hypermetabolism, particularly in large heterogeneous lesions, can more reliably assess tumor grade than tissue obtained from undirected sampling [12, 15, 16, 17].

Recent PET analyses of known chondrosarcomas indicated that maximum SUV and histopathology, when combined, are useful parameters for tumor grading and outcome prediction for local relapse and metastases [12]. As in clinical and conventional imaging observations, tumor diameter “was not a significant discriminator of outcome” [12]. The current study also noted that FDG uptake increased with tumor grade. PET changed management in six current patients (cases 8, 9, 17, 18, 20 and 28).Biopsy or surgery was eliminated due to hypometabolism in two enchondromas (cases 8 and 9). Excision, rather than allograft placement occurred in two others (cases 17 and 18). Two serendipitous PET-defined breast lesions (cases 20 and 28), referred for preferential care, influenced management.

Single-site CT, MRI and ultrasound (excepting Doppler) chiefly define static morphology.

While low-grade chondrosarcoma is particularly difficult for conventional radionuclide scanning [18], PET can simultaneously detect primary tumors, local residua, recurrence and metastases earlier via metabolic activity both before and after treatment, regardless of post-treatment anatomic distortion.

Timelier treatment adjustments may then result [22, 23, 24]. PET also documents diverse physiologic dynamics of primary and metastatic lesions (cases 19, 22 and 23). Focal recurrences or metastases may change their aggressiveness or dedifferentiate with resultant higher SUVs than primary foci (cases 19 and 20) (Table 1).

Limitations of PET

Variables affecting SUV include small lesions, volume averaging, scan initiation time and duration post-injection, tumor type, patient weight and blood glucose, with varying results in the same patients on different days. Values for separating benign from malignant lesions also vary in different institutions (from a maximum SUV of 1.3 [17] to 4 [12] due to diverse equipment, protocols and histology of lesions, so that SUV criteria cannot be strictly compared [25]. In the current study a maximum SUV cutoff of 2.0 was considered statistically significant (P=0.001).

While glucose metabolism generally increases in malignancies [5], tumor SUVs may overlap with those of inflammatory lesions (osteomyelitis, eosinophilic granuloma), lesions with a high giant cell content (giant cell tumor, aneurysmal bone cyst, osteoblastoma, Paget disease) and fibrous lesions (fibrous dysplasia). However, these entities may be excluded by radiographs, history and clinical data [15, 26, 27]. Experience with lesions of shared histology in general and cartilage lesions in particular also aid in optimizing PET evaluation.

18FDG-PET is a valuable adjunct in identifying aggressive cartilage lesions: directly in primary and locally recurrent sites or indirectly on whole-body scans by detecting unsuspected metastases. PET may eliminate biopsy in incidentally noted hypometabolic lesions(case 8 and 9) or guide it in histologically problematic or heterogeneous cartilage lesions by defining hypermetabolic foci for selective sampling which may be further facilitated by PET/CT. PET can be advantageous in elderly or high-risk cases, especially in detecting small or scattered foci in deep pelvic or retroperitoneal soft tissues (cases 24, 25 and 27). Interval whole-body scans, by characterizing new lesions metabolically, may aid in decreasing the repeated interventional or imaging studies often required for initial staging or follow-up.

Conclusion

18FDG-PET served as an objective and quantitative modality in distinguishing benign and malignant cartilage neoplasms. It supplied a functional criterion with which to further assess or supplement morphologic imaging and histopathology. PET may be useful as a predictor of biologic behavior of cartilage lesions as a group or of its individual members. Inter-institutional coordination of SUVs from larger cohorts is needed to provide more uniform criteria for the evaluation of SUVs and to allow optimal management.

Copyright information

© ISS 2005