¹⁸F-NaF PET/CT imaging in fibrous dysplasia of bone

Abstract

Fibrous dysplasia (FD) is a mosaic skeletal disorder resulting in fractures, deformity, and functional impairment. Clinical evaluation has been limited by a lack of surrogate endpoints capable of quantitating disease activity. The purpose of this study was to investigate the utility of ¹⁸F-NaF PET/CT imaging in quantifying disease activity in patients with FD. 15 consecutively evaluated subjects underwent whole-body ¹⁸F-NaF PET/CT scans, and FD burden was assessed by quantifying FD-related ¹⁸F-NaF activity. ¹⁸F-NaF PET/CT parameters obtained included 1) SUVmax (standardized uptake value of the FD lesion with the highest uptake), 2) SUVmean (average SUV of all ¹⁸F-NaF positive FD lesions), 3) total volume of all ¹⁸F-NaF positive FD lesions (TV), and 4) total FD lesion activity determined as the product of TV multiplied by SUVmean (TA = TV × SUVmean) (TA). Skeletal outcomes, function al outcomes, and bone turnover markers were correlated with ¹⁸F-NaF PET/CT parameters. TV and TA of extracranial FD lesions correlated strongly with skeletal outcomes including fractures and surgeries (p-values ≤ 0.003). Subjects with impaired ambulation and scoliosis had significantly higher TV and TA values (p < 0.05), obtained from extracranial and spinal lesions, respectively. Craniofacial surgeries correlated with TV and TA of skull FD lesions (p < 0.001). Bone turnover markers, including alkaline phosphatase, N-telopeptides, and osteocalcin were strongly correlated with TV and TA (p < 0.05) extracted from FD lesions in the entire skeleton. No associations were identified with SUVmax or SUVmean. Bone pain and age did not correlate with ¹⁸F-NaF PET/CT parameters. FD burden evaluated by ¹⁸F-NaF-PET/CT facilitates accurate assessment of FD activity, and correlates quantitatively with clinically-relevant skeletal outcomes.

INTRODUCTION

Fibrous dysplasia (FD) is an uncommon disorder characterized by replacement of normal bone and marrow with fibro-osseous tissue¹. Somatic activating mutations in GNAS lead to discrete, expansile skeletal lesions prone to fracture and deformity, resulting in pain and physical impairment ^2,3. Patients present along a broad clinical spectrum with a phenotype that varies widely between individuals. FD may affect one bone (monostotic) or multiple (polyostotic) and may occur in isolation or in combination with extraskeletal manifestations, including skin macules and hyperfunctioning endocrinopathies. The combination of FD and one or more extraskeletal features is termed McCune-Albright syndrome (MAS)¹. Current clinical management in FD is inadequate. There are no medical therapies that have been shown to prevent FD lesion expansion or associated skeletal deformities ^4–6. Surgical treatment is the mainstay approach for fractures and deformities, however outcomes are frequently unsatisfactory due to post-operative FD regrowth in the craniofacial skeleton ^7,8, and implants that are incapable of providing adequate long-term support in weight-bearing bones, particularly in growing children ^9–11.

Because of the wide phenotypic variability between patients, accurate assessment of skeletal disease burden is a critical component of the evaluation, management, and study of FD¹. Radiographs and computed tomography (CT) scans can provide anatomical characterization of individual FD lesions (Fig 1A and B), however nuclear medicine studies are necessary to determine total skeletal involvement ¹². The most commonly used method is the Skeletal Burden Score, a 99m-technetium-methylene diphosphonate (^99mTc-MDP)-based technique which assesses the percentage of the skeleton involved with FD ¹³ (Fig 1C and D). This is a reliable, reproducible technique that correlates with mobility impairment ¹³. However, Skeletal Burden Score is based on visual interpretation of conventional bone scans and lacks the quantitative capability to evaluate and monitor activity within individual lesions. Because of this important limitation, Skeletal Burden Score is an inadequate surrogate endpoint for trials of bone-altering therapies ^4,14,15. Thus, there is a critical need to identify surrogate endpoints that are capable of accurately quantifying FD lesion activity, while correlating with clinically meaningful outcomes.

Figure 1. Representative images demonstrating typical radiographic features in fibrous dysplasia.

(A) X-ray of the proximal femur shows a discrete, expansile lesion with a homogeneous, “ground glass” appearance. Note the characteristic cortical thinning, and coxa vara deformity of the femoral neck shaft (red arrow). (B) Axial computed tomography of the skull shows an expansile lesion with a homogenous, “ground glass” appearance in the mandible (red arrow). A smaller expansile lesion is also seen in the spinous process (red arrowhead). (C) and (D) Whole-body ^99mTc-MDP bone scans from two individuals demonstrate patchy tracer uptake consistent with mosaic disease (red arrowheads). Panel (C) demonstrates mild disease affecting only a few regions with a Skeletal Burden Score of 10, and Panel (D) demonstrates severe disease affecting most of the skeleton with a Skeletal Burden Score of 68. The bone scan shown in panel (C) was obtained from a child with FD, as indicated by the increased tracer uptake at the growth plates.

¹⁸F-NaF is a bone-seeking positron emitting radiopharmaceutical which is taken up through (¹⁸F⁻) ions, which exchange with hydroxyl ions (OH^‒) on the surface of hydroxyapatite-forming fluoroapatite ¹⁶. Skeletal sites with increased ¹⁸F-NaF activity reflect underlying bone processes, such as those in FD, which lead to increased bone crystal surface exposed to blood flow, resulting in higher availability of radiotracer binding sites ^17–19. The hybridization of PET imaging with multi-section CT allows for simultaneous anatomic evaluation of skeletal lesions, which is an attractive capability in deforming bone disorders such as FD. This technique has been applied widely to evaluate skeletal tumor burden and activity in metastatic diseases ^20–23.

The purpose of this study was to investigate the clinical utility of ¹⁸F-NaF PET/CT imaging in patients with FD. We hypothesized that FD burden determined by ¹⁸F-NaF PET/CT imaging would correlate with FD-related skeletal outcomes and biochemical markers of FD activity.

MATERIALS AND METHODS

Subjects

Subjects were evaluated as part of a long-standing NIH natural history study in FD/MAS (NCT00001727). This protocol involves collection of a standard set of data elements at the NIH Clinical Center, obtained through history and physical examination, physiatric, neuro-ophthalmologic and audiology evaluations, and biochemical testing. Subjects receive medical treatment for MAS-associated endocrinopathies and undergo imaging studies as clinically indicated. Nuclear medicine scans are clinically indicated to assess total skeletal disease burden at the time of diagnosis of FD/MAS, and at 3–5-year intervals for patients experiencing skeletal complications. Fifteen consecutively enrolled subjects who met clinical indications for nuclear medicine scanning underwent whole-body ¹⁸F-NaF PET/CT studies between 2012 and 2016. The protocol was approved by the NIDCR Institutional Review Board, and all subjects and/or their guardians gave informed consent/assent. The diagnosis of FD/MAS was established on clinical grounds with molecular testing as needed, according to previously published guidelines¹.

¹⁸F-NaF PET/CT studies

¹⁸F-NaF PET/CT scans were obtained on dedicated PET/CT scanners, after intravenous injection of an average ± standard deviation (SD) of 2.81 ± 0.83 mCi of ¹⁸F-sodium fluoride (range: 0.92 – 4.98). For adolescent and adult subjects, the injected dose ranged from 2.06 – 4.98 mCI; for one pediatric subject (age 6 years), a weight-based dose of 0.04 mCI/kg was administered. Scanners were SIEMENS Biograph 128_mCT models, with FlowMotion positioning and 256×256 matrix size. PET images from the vertex of the skull to the plantar surface of the feet were acquired, 62.36 ± 6.11 (mean ± SD) minutes (range: 58 – 76) post-tracer injection. Standardized uptake values (SUV) were measured on Time-of-Flight (TOF) PET images and were normalized to total body weight (as opposed to lean body mass or body surface area). A non-contrast, low dose CT scan was performed for attenuation correction and co-registration. Representative images are shown in Figure 2.

Figure 2. Representative ¹⁸F-NaF PET/CT images demonstrating features of fibrous dysplasia.

(A) ¹⁸F-NaF Maximum Intensity Projection (MIP) PET image of the skull and torso showing multiple areas of abnormally increased uptake corresponding to FD lesions involving the skull (light blue arrow), facial bones (purple arrow), left humerus (short deep blue arrows), multiple ribs (short green arrows), left radius (orange arrows), both pubic rami (black arrows) and both femurs (red arrows). Uptake in the left sacroiliac joint (red star) corresponds to site of FD-unrelated osteosclerosis seen on CT (panel C, open red arrows) and X-ray (panel D, closed red arrows). (B) Coronal ¹⁸F-NaF PET/CT image of the chest showing ¹⁸F-NaF avid FD lesions in several ribs (green arrows) and throughout the left humerus (short deep blue arrows). (C) Axial ¹⁸F-NaF PET/CT image of the pelvis showing focally increased uptake at the site of osteosclerosis in the left sacroiliac joint (open red arrow). (D) Anterior-posterior X-ray of the pelvis showing subarticular sclerosis on both sides of the ipsilateral left sacroiliac joint (closed red arrows), suggestive of likely sacroiliitis. FD lesions are apparent in both pubic rami (open red arrows) and proximal femoral bones with intramedullary rods noted on both femurs (open yellow arrows).

Determination of FD burden

Whole-body skeletal FD burden was assessed by quantifying ¹⁸F-NaF activity using the MIM Vista workstation (version 6.5.9) (Fig 3A). A volume of interest (VOI) encompassing the entire skeleton was drawn (Fig 3B), and subsequently a standardized uptake value (SUV)-threshold based approach customized per subject was applied to include all disease-related bone uptake. The software enables automatic generation of separate VOIs encircling all areas above the SUV-threshold set by the user (Fig 3C). Automatic lesion demarcation generated by the software was compared with the lesions’ anatomic cross-sectional images, to achieve maximal overlap with <5% qualitative difference between functional and anatomic images. This approach was implemented to customize the applied SUV-threshold per subject to the entire skeleton, avoiding incomplete segmentation of pathologic activity or inclusion of background physiologic uptake. Subsequently, an experienced nuclear medicine physician (GZP), blinded to subject clinical data, manually excluded areas with physiologic or non-FD-related ¹⁸F-NaF activity (Fig 3D). Non-FD related ¹⁸F-NaF activity was defined by uptake in the urinary tract (kidneys, ureters, urinary bladder, etc) and any abnormal skeletal uptake that did not correspond to confirmed areas of FD involvement on corresponding CT imaging. Non-FD related skeletal activity was identified and excluded in 3 subjects: 1) an area of focal increased uptake in a sacroiliac joint, consistent with sacroiliitis (Fig 2 C), 2) an isolated area of focal increased uptake in a posterior mandible, subsequently found to be associated with a dental infection, and 3) growth plate activity in a child. Finally, the following parameters were automatically obtained: 1) standardized uptake value (SUV) of the FD lesion with the highest uptake (SUV_max), 2) average SUV of all ¹⁸F-NaF positive FD lesions, determined as the mean of all 18F-NaF-positive lesions’ means (SUV_mean), 3) whole skeleton disease indices of total volume by summation of the volumes of all ¹⁸F-NaF avid FD lesions (TV), and 4) total lesion activity, determined as the product of SUV_mean multiplied by total volume of all ¹⁸F-NaF avid FD lesions seen on PET/CT (TA = SUV_mean × TV) (TA). PET/CT parameters were separately assessed for FD lesions in the skull, spine, and extracranial skeleton to investigate correlations with clinical and functional outcomes related to these specific regions. The quantification analysis for all ¹⁸F-NaF parameters was performed twice blindly by the same reader. By employing the same quantification approach, the obtained parameters (SUVmax, SUVmean, TV, TA) from each scan exhibited an intra-observer variation of <2% [(Value obtained from the 2nd analysis-Value obtained from the 1st analysis)/Value obtained from the 1st analysis) x100].

Figure 3. Representative images demonstrating semi-automated approach for quantification of FD-related ¹⁸F-NaF activity.

(A) Whole-body Maximum Intensity Projection ¹⁸F-NaF PET image of a patient with FD, showing abnormal radiotracer distribution. (B) Semi-automated drawn volume of interest (VOI) encompassing the entire skeleton. (C) Automatically generated VOI post SUV-threshold application, encircling all areas with uptake above the threshold. (D) Final VOI after manual extraction of areas with physiologic (e.g. kidneys, urinary bladder) and FD-unrelated activity (for example, the osteosclerotic lesion demonstrated in Figure 2). Green arrows in panels (C) & (D) demonstrate ¹⁸F-NaF activity in the kidneys, which was excluded from the final VOI.

Clinical Outcome Measures

The following skeletal-related clinical outcomes measures were evaluated: 1) fractures, 2) orthopedic surgeries, 3) impaired ambulation, 4) scoliosis, 5) craniofacial surgeries, 6) optic neuropathy, 7) hearing loss, and 8) bone pain. Outcomes measures were determined by retrospective review of clinical data elements collected on the natural history study. Fracture, surgery, and pain data were obtained by review of history and physical examinations. Ambulation impairment was defined as regular use of one or more assistive ambulation devices and was obtained by physiatric evaluation. Spinal radiographs were performed in patients with known spinal FD involvement and/or evidence of spinal curvature on exam, and scoliosis was defined as a Cobb angle of >10 degrees on radiographs ⁵. Optic neuropathy was determined by neuro-ophthalmologic evaluation, and hearing loss was determined by audiologic evaluation. These clinical outcomes measures were correlated with ¹⁸F-NaF-PET/CT parameters (SUV_max, SUV_mean, TV, and TA). ¹⁸F-NaF-PET/CT parameters were also correlated with subject age and biochemical bone turnover markers alkaline phosphatase, osteocalcin, and N-telopeptides (NTX).

Statistical Analysis

Statistical analyses were performed using R software (version 3.3). Descriptive statistics were used to characterize subjects, and data were expressed as mean ± SD. Correlations between FD-related ¹⁸F-NaF PET/CT parameters and continuous clinical outcome measures were quantified using Pearson’s (r) or Spearman’s (rho) correlation coefficient calculations with bootstrap 95% confidence intervals (CIs) calculated from 100 iterations where appropriate. The strength of each correlation was distinguished as “very weak”, “weak”, “moderate”, “strong” and “very strong” for an absolute correlation coefficient lying in range of [0.0 – 0.19], [0.2 – 0.39], [0.4 – 0.59], [0.6 – 0.79] and [0.8 – 1], respectively ²⁴. Scatterplots with a linear regression and 95% confidence intervals were generated to assess the relationship between age, clinical outcome measures, and bone turnover markers with ¹⁸F-NaF PET/CT parameters. Statistical diagnostic tests regarding linearity, homoscedasticity, normality of residuals and outliers were performed for validation of the regression analysis results. Mann-Whitney-Wilcoxon test was employed to assess potential differences based on categorical FD variables (impaired ambulation, scoliosis, optic neuropathy, hearing loss, pain), in association with ¹⁸F-NaF PET/CT parameters, with boxplots depicting the different subject groups. For all tests, a p-value of less than 0.05 was considered to indicate statistical significance. P-values were adjusted to account for multiple hypothesis testing using Benjamini and Hochberg false discovery rate correction.

RESULTS

Study population

Subject characteristics are included in Table 1. The study population was clinically heterogeneous, representing a wide range of ages, clinical outcome measures, and FD involvement. The subjects’ racial identities included non-Hispanic White (12/15, 80%), Black or African-American (2/15, 13%), and Asian (1/15, 7%). All subjects had polyostotic FD, defined as the presence of 2 or more non-contiguous FD lesions. The mean age and standard deviation were 27 ± 12 years, range 6–57, and included 10 female subjects and 5 male subjects. No subjects had clinical evidence of fractures or active malignancies at the time of scanning. One 28-year-old female subject had a history of FD-associated osteosarcoma of the left mandible diagnosed at age 12, for which she underwent hemimandibulectomy. She did not require further treatment for her malignancy and has had no signs of recurrence or metastases.

Table 1.

Subject characteristics

Age (years) & Sex	Race	MAS endocrine features	FD location	Lifetime fractures/mean rate per year1	Lifetime orthopedic surgeries/mean rate per year1	Lifetime craniofacial surgeries/mean rate per year1	Other clinical outcome measures
6F	W	GH	Cf	0	0	0	none
17F	W	PP, GH	Cf, Ax	1/0.06	0	1/0.06	Hearing loss, pain
18F	W	none	Cf	0	0	0	Optic neuropathy
22F	W	PP	Cf, Ax, Ap	2/0.1	0	0	Pain
22F	W	PP, HT, PW	Cf, Ax, Ap	100/4.6	52/2.4	3/0.1	Impaired ambulation, scoliosis, pain
23M	W	PW	Cf, Ax, Ap	4/0.2	0	0	Scoliosis, pain
24M	W	none	Cf, Ap	0	0	0	Pain
25M	W	GH	Cf, Ax, Ap	10/0.4	0	1/0.04	Scoliosis, pain
27F	B	PP	Cf, Ax, Ap	20/0.8	6/0.2	6/0.2	Scoliosis, pain
28F	A	PP, HT, GH, PW	Cf, Ax, Ap	6/0.2	9/0.3	6/0.2	Hearing loss, pain
30M	W	PW	Cf, Ax, Ap	36/1.2	6/0.2	2/0.07	Impaired ambulation, scoliosis, pain
31F	B	GH	Cf, Ap	0	0	1/0.03	Optic neuropathy
39F	W	none	Ap	1/0.03	2/0.05	0	Pain
47F	W	PP, HT, PW	Cf, Ax, Ap	50/1.1	7/0.2	0	Impaired ambulation, scoliosis, optic neuropathy, pain
57M	W	PP, PW	Cf, Ax, Ap	40/0.7	34/0.6	0	Impaired ambulation, scoliosis, pain

¹Mean rate per year determined by number of total lifetime events divided by years of life. F = female, M = male, W = non-Hispanic White, B = Black or African-American, A = Asian, MAS = McCune-Albright syndrome, GH = growth hormone excess, PP = precocious puberty, HT = hyperthyroidism, PW = phosphate wasting, n/a = not available, Cf = craniofacial skeleton, Ax = axillary skeleton, Ap = appendicular skeleton

Correlation between ¹⁸F-NaF PET/CT parameters and skeletal outcomes

Extracranial clinical outcomes

Correlations were evaluated between clinical outcomes involving the extracranial skeleton (i.e., entire skeleton excluding the skull) and ¹⁸F-NaF PET/CT parameters for extracranial FD (Table 2) (Fig 4). TV and TA of extracranial FD showed very strong correlation with lifetime fractures and mean fractures per year (Fig 4A–D). Strong correlations were also observed between TV and TA of extracranial FD and lifetime orthopedic surgeries and mean orthopedic surgeries per year (Fig 4 E–H). Extracranial SUV_max or SUV_mean did not correlate with the prevalence of fractures or surgeries.

Table 2.

Correlation between continuous clinical outcome measures and ^F18-NaF PET/CT parameters

	Lifetime fractures1	Mean fractures per year1	Lifetime orthopedic surgeries1	Mean orthopedic surgeries per year1	Lifetime craniofacial surgeries2	Mean craniofacial surgeries per year2
SUVmax	rho = 0.513, p = 0.113	rho = 0.458, p = 0.184	rho = 0.303, p = 0.442	rho = 0.254, p = 0.546	rho = 0.008, p = 1.000	rho = −0.004, p = 1.000
SUVmean	rho = 0.114, p = 0.824	rho = 0.083, p = 0.904	rho = 0.175, p = 0.694	rho = 0.171, p = 0.694	rho = 0.028, p = 0.998	rho = 0.019, p = 0.998
TV	rho = 0.897, p < 0.001	rho = 0.873, p < 0.001	rho = 0.825, p = 0.001	rho = 0.805, p = 0.002	rho = 0.831, p < 0.001	rho = 0.835, p < 0.001
TA	rho = 0.870, p < 0.001	rho = 0.844, p < 0.001	rho = 0.801, p = 0.002	rho = 0.775, p = 0.003	rho = 0.855, p < 0.001	rho = 0.860, p < 0.001

SUV_max = standardized uptake value of the fibrous dysplasia lesion with the highest uptake, SUV_mean = average SUV of all ¹⁸F-NaF positive fibrous dysplasia lesions, TV = volumes of all ¹⁸F-NaF avid FD lesions, TA = total lesional activity

¹Analyses performed for the extracranial skeleton.

²Analyses performed for the cranial skeleton.

Values are Spearman’s rank correlation coefficients and corresponding p-values adjusted for false discovery rate.

Figure 4. Correlation between ¹⁸F-NaF PET/CT parameters and extracranial skeletal outcomes.

Scatter plot and linear regression with 95% confidence intervals depicting TA and TV values, respectively, for FD-related activity in the extracranial skeleton in relation to lifetime fractures (A & B), mean fractures per year (C & D), lifetime orthopedic surgeries (E & F), and mean orthopedic surgeries per year (G & H).

Subjects with impaired ambulation had significantly higher TV and TA of extracranial FD in comparison to patients with full mobility (Table 3). No associations were identified with extracranial SUV_max or SUV_mean.

Table 3.

Correlation between categorical clinical outcome variables and ¹⁸F-NaF PET/CT parameters

	Impaired ambulation1	Scoliosis2	Pain3	Optic neuropathy4	Hearing loss4
SUVmax	p = 0.357	p = 0.558	p = 0.694	p = 0.202	p = 0.649
SUVmean	p = 0.998	p = 1.000	p = 0.112	p = 0.242	p = 0.270
TV	p = 0.043	p = 0.012	p = 0.077	p = 0.933	p = 0.649
TA	p = 0.043	p = 0.014	p = 0.077	p = 0.933	p = 0.221

SUV_max = standardized uptake value of the fibrous dysplasia lesion with the highest uptake, SUV_mean = average SUV of all ¹⁸F-NaF positive fibrous dysplasia lesions, TV = volumes of all ¹⁸F-NaF avid FD lesions, TA = total lesional activity

¹Analyses performed for the extracranial skeleton.

²Analyses performed for the spine.

³Analyses performed for the total skeleton.

⁴Analyses performed for the craniofacial skeleton.

Values are Spearman’s rank correlation coefficients and corresponding p-values adjusted for false discovery rate.

¹⁸F-NaF PET/CT parameters were separately assessed for spinal FD lesions and correlated with the presence of spinal deformity (Fig. 5) (Table 3). Spinal TV and TA were significantly higher in subjects with scoliosis versus subjects without scoliosis. No associations were identified with spinal SUV_max or SUV_mean.

Figure 5. ¹⁸F-NaF PET spinal imaging.

(A) ¹⁸F-NaF PET image of the torso showing normal radiotracer distribution in a spine unaffected by FD. (B) ¹⁸F-NaF PET image of the torso showing spinal deformity with diffusely increased activity and the semi-automated generated VOI encompassing all spinal FD-related ¹⁸F-NaF activity.

Craniofacial outcomes

Correlations were evaluated between craniofacial outcomes and ¹⁸F-NaF PET/CT parameters for FD affecting the skull (Table 2) (Fig 6). Craniofacial TV and TA were strongly associated with lifetime craniofacial surgeries (Fig 6 A&B) and mean craniofacial surgeries per year (Fig 6 C&D). Craniofacial SUV_max or SUV_mean did not correlate with the prevalence of craniofacial surgeries. There were no associations between craniofacial ¹⁸F-NaF PET/CT parameters and the presence of optic neuropathy or hearing loss.

Figure 6. Correlation between ¹⁸F-NaF PET/CT parameters and craniofacial outcomes.

Scatter plot and linear regression with 95% confidence intervals depicting TA and TV values, respectively, for FD-related activity in the craniofacial skeleton in relation to lifetime craniofacial surgeries (A & B) and mean craniofacial surgeries per year (C & D).

Pain

There were no significant differences in ¹⁸F-NaF PET/CT parameters in the entire skeleton between subjects with and without bone pain (Table 3).

Correlation between ¹⁸F-NaF PET/CT parameters and bone turnover markers

TV and TA of FD in the entire skeleton strongly correlated with serum levels of alkaline phosphatase, osteocalcin, and NTX)(Table 4)(Fig. 7)Neither skeletal SUV_max nor SUV_mean showed significant correlation with bone turnover markers (Table 4).

Table 4.

Correlation between ¹⁸F-NaF PET/CT parameters, bone turnover markers and age

	Alkaline phosphatase (U/L)	Osteocalcin (ng/mL)	N-telopeptides (nmol/mmol)	Age (years)
SUVmax	r = −0.252, p = 0.546	r = −0.355, p = 0.333	r = −0.287, p = 0.540	r = 0.141, p = 0.769
SUVmean	r = −0.135, p = 0.774	r = −0.445, p = 0.200	r = −0.072, p = 0.933	r = −0.519, p = 0.112
TV	r = 0.685, p = 0.015	r = 0.760, p = 0.005	r = 0.670, p = 0.037	r = 0.380, p = 0.287
TA	r = 0.758, p = 0.005	r = 0.707, p = 0.012	r = 0.739, p = 0.014	r = 0.229, p = 0.587

SUV_max = standardized uptake value of the fibrous dysplasia lesion with the highest uptake, SUV_mean = average SUV of all ¹⁸F-NaF positive fibrous dysplasia lesions, TV = volumes of all ¹⁸F-NaF avid FD lesions, TA = total lesional activity

Values are Pearson’s coefficients of correlation and corresponding p-values adjusted for false discovery rate.

Figure 7. Correlation between ¹⁸F-NaF PET/CT parameters and bone turnover markers.

Scatter plot and linear regression with 95% confidence intervals depicting TV and TA values for FD-related activity in the entire skeleton in relation to bone turnover markers, including alkaline phosphatase (A & B), urine-NTX (C & D), and osteocalcin (E & F). U/L = units per liter, nmol/mmol = nanomole per millimole, ng/mL = nanograms per milliliter.

Correlation between ¹⁸F-NaF PET/CT parameters and age

There were no significant associations between 18F-NaF PET/CT parameters and subject age at the time of scanning (Table 4).

Comparison between ¹⁸F-NaF PET/CT and 99mTc-MDP scintigraphy for FD lesion detection

Analyses were performed to compare the ability of ¹⁸F-NaF PET/CT and ^99mTc-MDP scintigraphy (considered the present standard of care for FD lesion identification) to detect total skeletal FD involvement. Analyses were limited to adult subjects ≥ age 18 years, after which total skeletal disease burden is expected to remain stable ²⁵. Skeletal Burden Scores were determined for 7 subjects who had previously undergone ^99mTc-MDP scans and compared to Skeletal Burden Scores determined from ¹⁸F-NaF PET/CT scans, using previously described methodology ¹³. The median ^99mTc-MDP-derived Skeletal Burden Score was 52.5 (interquartile range 44.8, 67.9)(range 10.9–75), and the median ¹⁸F-NaF-derived Skeletal Burden Score was 59.5 (interquartile range 41.2, 71.8)(range 9.95–75). The median difference between the scans was 0 (interquartile range −2.9, 3.9)(range −3.6–7.0). Linear regression analyses showed very strong correlation between Skeletal Burden Scores derived from both scan types (R² = 0.98, p<0.0001) (Supplemental Fig 1).

DISCUSSION

Accurate detection and quantification of skeletal lesions is important to determine prognosis, guide management, and serve as endpoints for clinical studies in patients with FD. This is the first study to demonstrate that FD-related ¹⁸F-NaF uptake correlates with skeletal outcomes in this disorder. Strong and consistent correlations were observed between TV and TA of all skeletal ¹⁸F-NaF avid FD lesions with bone turnover markers, suggesting this imaging technique is capable of accurately reflecting underlying bone metabolism in FD. Clinically-relevant outcomes such as fractures and surgeries were very strongly associated with volume and activity of lesions assessed by ¹⁸F-NaF PET/CT. Finally, ¹⁸F-NaF parameters TV and TA extracted from ¹⁸F-NaF positive FD lesions in the extracranial skeleton, craniofacial skeleton, spine, and entire skeleton were able to differentiate patients with critical disease-related features involving these areas, including impaired ambulation, scoliosis, and pain.

SUV-based measures normalized to patient body weight, lean body mass, or body surface area, are the most commonly used semi-quantitative tools in PET-imaging ²⁶. However, in our study the obtained SUVmetrics (SUV_mean and SUV_max) did not show any correlation with clinical or biochemical measures of disease. FD is a benign disorder in which complications arise due to structural instability of abnormal bone; it is therefore intuitive that prognosis is determined in large part by the overall volume of affected tissue, which is not directly accounted for using SUV-based metrics. This is consistent with a growing body of evidence that radiotracer-avid lesion volume and lesion activity may also outperform SUV-based measurements in other polyostotic conditions, such as metastatic bone disease ^21,27–29.

The optimal SUV-threshold values for application of ¹⁸F-NaF PET/CT have not been standardized and should be determined based on the underlying pathologic condition and the goals of the evaluation. In several oncologic studies of patients with prostate cancer, researchers used an SUV-threshold >15 for 18F-NaF uptake in order to segment bone metastases ^22,30. In other oncologic studies utilizing ¹⁸F-NaF PET/CT scans to evaluate metastatic diseases from a variety of cancer types, an SUV threshold of >10 was applied ^20,21. In the current study, SUV-thresholds ranged between 7–8. The lower thresholds in this study likely reflect the benign nature of FD, where normal bone and marrow are replaced by fibro-osseous rather than malignant tissue.

¹⁸F-NaF PET/CT offers considerable advantages over current clinical assessment tools in FD. Clinical trials have been severely limited by a lack of surrogate markers capable of reflecting FD activity. Until now, Skeletal Burden Score using ^99mTc-MDP scintigraphy has been the only imaging technique capable of assessing disease burden and associating with ambulation outcomes ¹³. A direct comparison between ¹⁸F-NaF and ^99mTc-MDP scans in this study showed that both methodologies appear to be equally effective in determining Skeletal Burden Score. Because Skeletal Burden Score measures only the extent of affected tissue, it has been shown to remain static throughout adulthood after FD lesions have been fully established ²⁵. It is therefore not a useful endpoint for evaluating response to bone-altering therapies, which may potentially decrease lesion activity and improve bone quality but are unlikely to affect the size of established lesions. Bone turnover markers are typically elevated in patients with FD and may also serve as potential quantitative endpoints ⁴. However, FD-specific serum markers have not been identified, and it is therefore unfeasible to use bone turnover markers to quantify response to bone-altering therapies that affect both FD and non-FD bone. The unique quantitative capabilities offered by ¹⁸F-NaF PET/CT imaging, together with the vastly superior resolution of PET-imaging compared to conventional gamma camera imaging, represent a considerable advantage and may potentially have the capacity to serve as a clinically relevant, quantitative surrogate endpoint for clinical trials in FD.

The detailed morphological characterization provided by the multisection CT component of ¹⁸F-NaF PET/CT imaging offers another important advantage. Unlike ¹⁸F-NaF PET/CT, traditional ^99mTc-MDP scans are unable to provide sufficient anatomical information for a thorough clinical assessment of FD lesions, and must be accompanied by additional imaging studies such as radiographs or CT scans. The combination of disease quantification and anatomical imaging may be particularly advantageous in evaluating craniofacial and spinal FD, where TA and TV of these skeletal compartments correlated with region-specific complications (i.e. craniofacial surgeries and scoliosis). The presence of skeletal deformities does not confound the ability of ¹⁸F-NaF to demarcate FD lesions, because this process is fully automated upon application of the SUV-threshold. In addition, the favorable ¹⁸F-NaF pharmacokinetic characteristics compared to ^99mTc- MDP, including minimal plasma protein binding and rapid soft-tissue clearance, result in shorter imaging times (<1 hour post intravenous injection) and increased bone-to-background, leading to twice as great bone uptake and enhanced diagnostic accuracy compared to ^99mTc- MDP scintigraphy ¹⁷. This combination of quantitative capabilities, superior anatomical characterization, and improved diagnostic accuracy suggests that ¹⁸F-NaF PET/CT imaging is the modality of choice for evaluation and monitoring of FD activity.

The primary limitations of this study are the relatively small subject numbers and the clinical heterogeneity between subjects. This is an inherent obstacle to rare disease research, where studies are frequently underpowered to detect statistically significant effects. However, in this study, strong correlations were observed despite the small number of subjects, and the clinical heterogeneity demonstrated the utility of this technique across a spectrum of disease. Additional studies with larger numbers of subjects are needed to verify the reproducibility of these findings, including inter-reader reproducibility analyses. Another important limitation was the cross-sectional design. Prospective studies with serial scans are needed to determine whether ¹⁸F-NaF PET/CT scans can detect longitudinal changes in lesion activity over time in individual patients. Prospective studies are also needed to determine whether ¹⁸F-NaF PET/CT scans are predictive of skeletal complications. In this study bone pain was evaluated categorically, and additional studies using validated quantitative pain measures are needed to determine the relationship between ¹⁸F-NaF PET/CT parameters and pain. It is unknown if ¹⁸F-NaF PET/CT uptake related to FD lesions can reliably be distinguished from uptake related to fractures and secondary malignancies, both of which are known complications in this disorder. In this study no subjects had clinical evidence of fractures or active malignancies at the time of scanning. However, if patients have concerning symptoms, such as acute-onset focal pain or rapid expansion of an FD lesion, clinical evaluation for fractures and/or malignancies should be performed prior to nuclear medicine scanning. Application of these findings therefore should involve collaboration between metabolic bone specialists and nuclear medicine specialists experienced in interpreting radionuclear scans.

¹⁸F-NaF PET/CT imaging parameters demonstrate strong correlations with clinically relevant skeletal outcomes in patients with FD. These findings establish ¹⁸F-NaF PET/CT as the current imaging modality of choice for evaluation of FD activity. In addition to its immediate clinical application, this technique holds potential as a surrogate endpoint for clinical trials, a critical area of need to improve outcomes for patients with FD.

Supplementary Material

Supp FigS1

Supplemental Figure 1. Comparison between ¹⁸F-NaF PET/CT and ^99mTc-MDP scintigraphy for determination of total Skeletal Disease Burden. Skeletal Burden Scores were obtained using both ¹⁸F-NaF PET/CT and ^99mTc-MDP scintigraphy for 7 adult subjects who had undergone both scan types, using previously described methodology (reference 13). Very strong correlations were observed between scores from both scan types.