Prospective evaluation of planar bone scintigraphy, SPECT, SPECT/CT, ¹⁸F-NaF PET/CT and whole body 1.5T MRI, including DWI, for the detection of bone metastases in high risk breast and prostate cancer patients: SKELETA clinical trial

Abstract

Purpose. Detection of bone metastases in breast and prostate cancer patients remains a major clinical challenge. The aim of the current trial was to compare the diagnostic accuracy of ^99mTc-hydroxymethane diphosphonate (^99mTc-HDP) planar bone scintigraphy (BS), ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and whole body 1.5 Tesla magnetic resonance imaging (MRI), including diffusion weighted imaging, (wbMRI+DWI) for the detection of bone metastases in high risk breast and prostate cancer patients.

Material and methods. Twenty-six breast and 27 prostate cancer patients at high risk of bone metastases underwent ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI. Five independent reviewers interpreted each individual modality without the knowledge of other imaging findings. The final metastatic status was based on the consensus reading, clinical and imaging follow-up (minimal and maximal follow-up time was 6, and 32 months, respectively). The bone findings were compared on patient-, region-, and lesion-level.

Results. ^99mTc-HDP BS was false negative in four patients. In the region-based analysis, sensitivity values for ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI+DWI were 62%, 74%, 85%, 93%, and 91%, respectively. The number of equivocal findings for ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI was 50, 44, 5, 6, and 4, respectively.

Conclusion. wbMRI+DWI showed similar diagnostic accuracy to ¹⁸F-NaF PET/CT and outperformed ^99mTc-HDP SPECT/CT, and ^99mTc-HDP BS.

Breast cancer (BCa) and prostate cancer (PCa) are currently the most common cancer types in the western world [1]. Despite of advances in local and adjuvant treatment approximately 15–20% of patients will relapse and 70% of these will later develop bone metastases [2]. The patients with bone metastases are not amenable to curative therapy but will need hormonal and/or chemotherapy with non-steroid antianalgesics, opioids, bisphophonates and radiotherapy reserved for palliation [3]. The recognition of skeletal disease as an important metastatic site has elicited new forms of targeted therapies such as inhibition of receptor activator of nuclear factor-κB (RANK) ligand and radionuclide therapy using alpha emitting radium [4]. The degree of bone metastatic spread has been identified as an independent prognostic factor in PCa patients [5].

^99mTc-methyl diphosphonate or ^99mTc-hydroxymethane diphosphonate (^99mTc-HDP) bone scintigraphy (BS) is the most widely used imaging modality for the detection of bone metastases. Unfortunately, several studies have demonstrated rather limited sensitivity (46–70%) and specificity (32–57%) of BS for the detection of bone metastases in BCa and PCa patients [6,7]. The use of single photon emission computed tomography (SPECT) alone or combined with computed tomography (CT) has shown to improve both sensitivity and specificity for the detection of bone metastases [6,8]. Positron emission tomography (PET) plays an important role in cancer imaging. Several studies indicate that ¹⁸F-NaF PET/CT outperforms BS and SPECT in the detection of bone metastases [6,9–11].

Anatomical magnetic resonance imaging (MRI) has been used for the detection of bone metastases for more than 10 years [12–14]. Recent development in MR hardware and sequence optimization has resulted in whole body MRI (wbMRI) capable of surveying the entire skeleton without irradiation over imaging times of 30–50 min. Anatomical MRI with or without diffusion weighted imaging (DWI) is more sensitive and specific than BS for the detection of bone metastases [15–17]. In clinical practice anatomical MRI, with or without DWI, is commonly used to evaluate equivocal findings of nuclear medicine techniques, mainly BS. The cost of MRI is thus added to that of the nuclear medicine techniques and this approach may cause delay in the therapeutic decision process which delays treatment initiation such as the start of cancer therapy, further prolonging patients’ suffering.

The aim of the current study was to compare the diagnostic accuracy of ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and whole body 1.5 Tesla (T) MRI, including DWI (wbMRI+DWI) for the detection of bone metastases in high risk BCa and PCa patients. Our goal was to reduce redundant bone imaging by studying the added value of tomographic and hybrid imaging techniques and wbMRI+DWI to standard approach which has been BS for many decades.

Material and methods

Patients

Between February 2011 and March 2013, 26 BCa and 27 PCa patients were prospectively enrolled. The clinical trial registration number is NCT01339780. The study was approved by the local ethics committee and each patient gave written inform consent. Patients with high risk of bone metastases were enrolled if they met at least one of the following criteria: 1) high clinical suspicion of bone metastases: localized pain over bone area; 2) laboratory findings: elevated alkaline phosphates (> 105 U/l), elevated PSA or high PSA doubling time after prostatectomy, radiotherapy; 3) histopathologic findings: stage N3a or higher in BCa patients, stage T3a or higher and/or Gleason score of 4 + 3 or higher in PCa patients. All BCa and PCa patients were referred for the evaluation of possible bone metastases after mastectomy/prostatectomy and/or external radiotherapy. Fifteen (58%, 15/26) BCa and 13 (48%, 13/27) PCa patients had hormonal therapy at the time of imaging. All enrolled patients underwent ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F NaF PET/CT and wbMRI+DWI within two weeks.

Bone scintigraphy, SPECT and SPECT/CT

Bone scan, SPECT and SPECT/CT examinations were performed using Symbia T6, True Point SPECT/CT scanner (Siemens, Erlangen, Germany). The anterior and posterior views were collected 3 h after injection of 670 MBq ^99mTc-HDP using low-energy high-resolution (LEHR) collimators, scan speed 13 cm/min and matrix size 256 × 1024. Single-photon emission computed tomography combined with CT was acquired immediately after BS with the following parameters: three bed positions, LEHR collimators, 90 views, 9 s scanning time per view, matrix 128 × 128, zoom 1.0, energy window 140 keV ± 15% with lower scatter window, reconstruction using Flash 3D (Siemens, Erlangen, Germany) with five iterations and 10 subsets and CT with an effective mAs 10 and 130 kVp. SPECT and SPECT/CT were performed from the tip of the head to the midthighs.

¹⁸F-NaF PET/CT

The patients received intravenous injection of 209 ± 7 (mean± SD) MBq of ¹⁸F-NaF diluted in 3–5 ml of saline as a 60 s bolus which was promptly flushed with saline. The PET data were acquired 64 ± 6 (mean± SD) min after the tracer injection using Discovery VCT PET/CT or PET/CT 690 scanner (General Electric Medical Systems, Milwaukee, WI, USA). A static emission scan was acquired over whole body with 3 min (PET/CT VCT scanner) and 2 min (PET/CT 690 scanner) acquisition time per bed position. The CT scan had a noise index of 25. The sinogram data were corrected for deadtime, decay, photon attenuation and reconstructed in a 128 × 128 matrix. Image reconstruction followed a fully three-dimensional (3D) maximum likelihood ordered subsets expectation maximization algorithm incorporating CT attenuation, random and scatter corrections with two iterations and 28 subsets.

Whole body MRI

Coronal T1-weighed (T1wi), coronal short tau inversion recovery (STIR) images and DWI data were acquired using a 1.5 T MR system (Avanto, Siemens, Erlangen, Germany) and surface array coils. The coronal images were collected from six table positions (table movement 250 mm) covering whole body with the following imaging parameters, T1wi: repetition time/echo time (TR/TE) 709/10 ms, field of view (FOV) 500 mm, slice thickness 7.0 mm, voxel size 2.2 × 2.2 × 7.0 mm³; STIR: TR/TE 14790/97 ms, inversion time 130 ms, FOV 500 mm, slice thickness 7.0 mm, voxel size 1.9 × 1.3 × 7.0 mm³. The axial DWI was performed using single shot spin echo sequence with 10 table positions (table movement 100 mm), TR/TE 6300/80 ms, SPAIR fat suppression, FOV 500 mm, slice thickness 5.0 mm, voxel size 2.6 × 2.6 × 5.0 mm³, three optimized diffusion directions (three-scan trace option on), b-values of 0, 150, 1000 s/mm². The lower limbs were covered in coronal orientation with 2–3 table positions (table movement 250 mm) and similar imaging parameters as axial DWI except of voxel size 3.3 × 2.6 × 6.0 mm³. The overall imaging time was approximately 50–60 min.

Data analysis and interpretation

Each imaging modality was interpreted without the knowledge of other imaging results by three highly experienced nuclear medicine physicians and two radiologists. The reviewers were only aware of the fact that the patients are at high risk of bone metastases. Lesions were graded as highly suspicious for being metastases, equivocal or benign. On BS and SPECT scans, lesions were categorized as benign when they were located around joints, hot osteophytes, vertically involving several ribs (suggesting fracture), H-shaped pelvic abnormal, bursitis, avulsion injury, tendinitis [9,10]. The vertebral lesions were considered as highly suspicious when they involved posterior aspect, pedicle or the whole vertebral body. Anatomical CT images of PET/CT and SPECT/CT were used to categorize uptakes as benign or highly suspicious due to corresponding morphologic findings. Typical benign lesions according to CT data were bone cysts, degenerative lesions (e.g. around joints), and fractures. When the tracer uptake was located on osteoblastic, osteolytic or mixed lesion, the lesion was marked as highly suspicious based on PET/CT and/or SPECT/CT. Lesions associated with a tracer uptake and not typical benign or malignant changes on CT were considered as equivocal.

A lesion was considered as highly suspicious on MRI if a focal or diffusion low signal intensity (SI) was present on T1wi with the corresponding intermediate or high SI on STIR and/or restricted diffusion on DWI. Typical benign lesions and/or sclerosis were interpreted according to previously published criteria [14,18]. Trace images (mean of three diffusion directions) were evaluated visually in conjunction with anatomical T1wi and STIR. No quantitative cut-off values were used for DWI.

The bone findings were compared on patient-, region- and lesion-level. In the region-based analysis, the skeleton was divided into five regions: head, thorax and ribs, spine, pelvis and limbs. In the lesion-based analysis, only lesions which were highly suspicious or equivocal on at least one imaging modality were included. In addition, maximum of five lesions with the highest agreement between modalities per anatomical location (five locations as defined in the region-based analysis) were included in the lesion-based analysis.

Best valuable comparator (BCV)

The findings of each imaging modality were compared with best valuable comparator (BVC) in order to define their nature [7,19]. Consensus reading of all imaging modalities and follow-up data of clinical, imaging and laboratory results were used to define BVC. The mean± standard deviation follow-up time of 53 enrolled patients was 15 ± 7 months while the range was 6–32 months. In total, 74 follow-up imaging examinations were performed consisting of 21 ^99mTc-HDP BS, one ^99mTc-HDP SPECT/CT, 38 CT, four ¹⁸F-NaF PET/CT, three ¹⁸F-NaF PET/MRI, and seven MRI examinations. Three BCa and two PCa patients died during the follow-up period. Imaging follow-up data were available for 14 BCa and 11 PCa patients. These studies were mainly performed in the patients with disconcordant findings among the different imaging modalities while no follow-up imaging was done in 10 BCa and 10 PCa patients with no highly suspicion lesions on any of the imaging modalities. These patients did not show any signs of progression during the follow-up period and were considered as true negative. Three PCa patients had no imaging follow-up examinations while having highly suspicious lesion(s) on BS and/or SPECT which were considered as false positive based on the consensus imaging findings of the initial imaging examinations. Highly suspicious lesions were considered as true positive if the lesions on the consensus reading of the initial imaging examinations and/or follow-up examinations were positive for bone metastases. Lesions were considered as false positive if rated as highly suspicious on one modality but as benign on consensus reading of the initial imaging examinations and/or no signs of active disease in any of the follow-up examinations was discovered. False negative finding was made when a reader found no lesion or marked as a benign lesion and the consensus reading of the initial imaging examinations and/or follow-up examinations showed bone metastases. If no imaging follow-up data were available (12 BCa and 16 PCa patients), patients with clinical, laboratory follow-up data suggesting progression of the metastatic disease, as well as consensus reading of the initial imaging examinations suggesting metastases, were considered to have metastatic disease at the time of initial imaging (two BCa and three PCa patients).

Statistical analysis

Equivocal findings of the imaging modalities were classified either as suggestive for metastases (“pessimistic analysis”) or suggestive for non-metastatic origin (“optimistic analysis”). Sensitivity and specificity values of patient-, region- and lesion-based analyses were compared using McNemar test [20] and two-sided p-values were calculated. In region-based analysis, diagnostic accuracy values for the detection of bone metastases [sensitivity, specificity, accuracy and area under the curve (AUC)] were calculated from all ROIs which were pooled into one group. Moreover, receiver operation characteristic curves (ROC) analysis was performed using 60 000 bootstraps [21] to account for within-patient correlations. AUC values were calculated using the trapezoid rule and compared using a method described by Hanley and McNeil [22], two-sided p-values were calculated. Bootstrap samples were constructed by stratifying patients based on overall cancer level (BCa/PCa present or not) and drawing patients as the independent units with replacement from these groups (BCa/PCa present or not) [21]. P values < 0.05 were considered statistically significant. All statistical analyses were performed using in-house written Matlab codes (Mathworks Inc., Natick, MA, USA). The Matlab codes as well as all MRI sequences are freely available upon request.

Results

Patient-based analysis

In patient-based analysis, 19 (36%, 19/53) patients, 11 (42%, 11/26) BCa and eight (30%, 8/27) PCa patients, had presence of bone metastases based on BVC. BS had significantly lower sensitive and AUC values than ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI (Table I; Figures 1, 2). When equivocal lesions were considered as suggestive for bone metastases (pessimistic analysis), ^99mTc-HDP BS and ^99mTc-HDP SPECT had significantly lower specificity values than ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI. Differences in sensitivity, specificity, AUC values of ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI did not reach statistical significance.

Figure 1. ^99mTc-hydroxymethane diphosphonate bone scintigraphy, anterior (A) and posterior view (B), and SPECT (C) of 69-year-old prostate cancer patient with serum PSA values of 97 ng/ml and Gleason score of 5 + 4 based on systematic biopsy. One equivocal lesions (10th rib, marked by black arrows) was present on ^99mTc-hydroxymethane diphosphonate bone scintigraphy while focal uptake was present on SPECT in the 10th rib on the left side and right lower part of sacral bone (marked by black arrows) suggesting bone metastases.

Figure 2. Coronal ¹⁸F-NaF PET and DWI, inverted signal intensity of trace image b-value of 1000 s/mm², of the same 69-year-old prostate cancer patient as displayed in Figure 1. The lesions in the 10th rib on the left side and right lower part of sacral bone (marked by black arrows) were highly suspicious for being bone metastases similarly to SPECT (Figure 1) and SPEC/CT (not shown) while the lesion in sacral bone was missed and the in the 10th rib was considered as equivocal on bone scintigraphy. These highly suspicious lesions were suggestive to be bone metastases based on clinical and imaging follow-up, demonstrating a limitation of bone scintigraphy for the detection of bone metastases in prostate cancer patients.

Table I. Patient-based analysis.

	Optimistic analysis				Pessimistic analysis
	Sensitivity	Specificity	Accuracy	AUC	Sensitivity	Specificity	Accuracy	AUC
BS	79 (p < 0.001)	91 (p = 0.16)	87	0.85 (p < 0.05)	85 (p = 0.08)	59 (p < 0.01)	67	0.72 (p < 0.01)
SPECT	89 (p = 0.16)	80 (p < 0.01)	83	0.84 (p < 0.01)	95 (p = 0.32)	56 (p < 0.01)	69	0.75 (p < 0.01)
SPECT/CT	89 (p = 0.16)	94 (p = 0.32)	93	0.92 (p = 0.12)	95 (p = 0.32)	88 (NA)	90	0.92 (p = 0.12)
PET/CT	95 (p = 0.32)	97 (NA)	96	0.96 (p = 0.42)	100 (NA)	82 (p = 0.32)	89	0.91 (p = 0.15)
wbMRI+DWI	100 (*)	97 (*)	98	0.99 (*)	100 (*)	88 (*)	93	0.94 (*)

Sensitivity, specificity and accuracy values are displayed in %. In each group, two-sided p-values are displayed with reference to the modality marked by (*). BS, ^99mTc-hydroxymethane diphosphonate bone scintigraphy; NA, not applicable; PET/CT, ¹⁸F-NaF positron emission tomography combined with computed tomography; SPECT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography; SPECT/CT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography combined with computed tomography; wbMRI+DWI, whole body MRI (T1-weighted imaging, STIR) including diffusion weighted imaging.

Region-based analysis

In region-based analysis, 58 (22%, 58/265) ROIs, 39 (30%, 39/130) BCa and 19 (14%, 19/135) PCa ROIs, had presence of bone metastases based on BVC. ^99mTc-HDP BS and ^99mTc-HDP SPECT had significantly lower sensitive and AUC values than ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI in both pessimistic and optimistic analyses (Table II). In pessimistic analysis, ¹⁸F-NaF PET/CT and wbMRI+DWI were significantly more sensitive than ^99mTc-HDP SPECT/CT while differences in specificity and AUC values did not reach the level of statistically significance. In contrast, when equivocal lesions were considered as not suggestive for bone metastases (optimistic analysis), the differences in sensitivity, specificity and AUC values of ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI did not reach statistical significance.

Table II. Region-based analysis.

	Optimistic analysis				Pessimistic analysis
	Sensitivity	Specificity	Accuracy	AUC	Sensitivity	Specificity	Accuracy	AUC
BS	62 (p < 0.001)	98 (p = 0.32)	90	0.80 (p < 0.001)	72 (p < 0.01)	89 (p < 0.01)	86	0.81 (p < 0.001)
SPECT	74 (p < 0.01)	94 (p < 0.01)	89	0.83 (p < 0.01)	86 (p < 0.05)	87 (p < 0.001)	87	0.87 (p < 0.01)
SPECT/CT	85 (p = 0.05)	99 (p = 0.32)	96	0.92 (p = 0.24)	85 (p < 0.01)	98 (NA)	90	0.91 (p = 0.07)
PET/CT	93 (*)	99 (*)	98	0.96 (*)	97 (*)	97 (p = 0.52)	97	0.97 (*)
wbMRI+DWI	91 (p = 0.65)	99 (p = 0.32)	97	0.95 (p = 0.92)	91 (p = 0.25)	98 (*)	96	0.95 (p = 0.36)

Sensitivity, specificity and accuracy values are displayed in %. In each group, two-sided p-values are displayed with reference to the modality marked by (*). BS, ^99mTc-hydroxymethane diphosphonate bone scintigraphy; NA, not applicable; PET/CT, ¹⁸F-NaF positron emission tomography combined with computed tomography; SPECT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography; SPECT/CT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography combined with computed tomography; wbMRI+DWI, whole body MRI (T1-weighted imaging, STIR) including diffusion weighted imaging.

Lesion-based analysis

In total, 234 lesions, 172 in BCa and 62 in PCa patients, were highly suspicious or equivocal in at least one imaging modality and no more than five lesions per region with the highest agreement between modalities, as defined in the region-based analysis, were included. Of these 234 lesions, 159 (68%, 159/234) lesions, 123 (72%, 123/172) in BCa and 36 (58%, 36/62) in PCa patients, were considered to be metastatic bone lesions based on BVC. ¹⁸F-NaF PET/CT and wbMRI+DWI were significantly more sensitive in the lesion-based analysis than ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT (Table 3). Moreover, ¹⁸F-NaF PET/CT and wbMRI+DWI had significantly higher specificity values than ^99mTc-HDP BS and ^99mTc-HDP SPECT. ¹⁸F-NaF PET/CT and wbMRI+DWI had similar sensitivity, specificity, accuracy and AUC values (Figures 3–5).

Figure 3. ^99mTc-hydroxymethane diphosphonate bone scintigraphy, anterior (A) and posterior view (B), and SPECT (C) of 62-year-old breast cancer patient with elevated alkaline phosphatase and localized pain in upper back. Lesions in the 5th and 7th rib (marked by black arrows) on the left side are highly suspicions of being bone metastases based on bone scintigraphy and SPECT. The lesions (marked by half filled arrows) in the fifth lumbal vertebra and 9th rib on the left side were considered as equivocal on prospective SPECT reading while the increased uptake (marked by black arrowhead) in the left side of thoracic spine at the level of X was considered as a benign uptake on prospective SPECT reading.

Figure 4. Coronal ^99mTc-hydroxymethane diphosphonate SPECT/CT (A) ¹⁸F-NaF PET/CT (B) and T1-weighted image (C) of the same 62-year-old breast cancer patient as displayed in Figure 3. In addition to the lesions in the 5th and 7th rib on the left side detected on ^99mTc-hydroxymethane diphosphonate bone scintigraphy and SPECT, lytic lesion was present in the body of the 5th lumbal vertebra (marked by white arrow).

Figure 5. Maximum intensity projection of ¹⁸F-NaF PET (A) and DWI (B), inverted signal intensity of trace image b-value of 1000 s/mm², of the same 62-year-old breast cancer patient as displayed in Figures 3 and 4. In addition to the lesions in the 5th, 7th rib on the left side and L5 (marked by black arrows) detected by SPECT/CT, two additional highly suspicion lesions (marked by black arrows) in the 6th and 9th rib on the right side were detected by ¹⁸F-NaF PET/CT and whole body MRI, including DWI. All of the highly suspicious lesions were suggestive to be bone metastases based on clinical and imaging follow-up, showing limited accuracy of bone scintigraphy for the detection of bone metastases in breast cancer patients. Magnetic susceptibility related artifacts are present on DWI in the right hip due to presence of metal hip prosthesis.

Table III. Lesion-based analysis.

	Optimistic analysis				Pessimistic analysis
	Sensitivity	Specificity	Accuracy	AUC	Sensitivity	Specificity	Accuracy	AUC
BS	54 (p < 0.001)	88 (p < 0.01)	65	0.71 (p < 0.001)	69 (p < 0.001)	72 (p < 0.001)	71	0.71 (p < 0.001)
SPECT	71 (p < 0.001)	79 (p < 0.01)	74	0.75 (p < 0.001)	77 (p < 0.001)	52 (p < 0.001)	70	0.65 (p < 0.001)
SPECT/CT	81 (p < 0.001)	96 (NA)	85	0.88 (p = 0.04)	82 (p < 0.001)	94 (NA)	85	0.88 (p = 0.02)
PET/CT	94 (p = 0.78)	96 (*)	95	0.95 (*)	96 (*)	89 (p = 0.20)	94	0.93 (p = 0.38)
wbMRI+DWI	95 (*)	95 (p = 0.65)	95	0.95 (p = 0.91)	95 (p = 0.76)	94 (*)	95	0.95 (*)

Sensitivity, specificity and accuracy values are displayed in %. In each group, two-sided p-values are displayed with reference to the modality marked by (*). BS, ^99mTc-hydroxymethane diphosphonate bone scintigraphy; NA, not applicable; PET/CT, ¹⁸F-NaF positron emission tomography combined with computed tomography; SPECT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography; SPECT/CT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography combined with computed tomography; wbMRI+DWI, whole body MRI (T1-weighted imaging, STIR) including diffusion weighted imaging.

The number of equivocal lesions was 50, 44, 5, 6 and 4 in ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI readings, respectively. These lesions were present in 22 (42%, 22/53), 20 (38%, 20/53), 5 (9%, 5/53), 4 (11%, 4/53), and 3 (8%, 3/53) patients of ^99mTc-HDP BS, ^99mTc-HDP SPECT, SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI+DWI readings, respectively. Furthermore, in 13 (25%, 13/53), 9 (17%, 9/53), 4 (8%, 4/53), 4 (8%, 4/53) and 2 (4%, 2/53) patients only equivocal lesions with or without benign lesions were present in ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT, wbMRI+DWI readings, respectively.

Change in patient management

In two PCa patients, the detection of bone metastases discovered only by SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI resulted in the change of treatment (start of chemotherapy). In one BCa patient, hormonal treatment was initiated due to detection of bone metastases detected by ¹⁸F-NaF PET/CT and wbMRI+DWI. Due to detection of liver and mesenteric metastases only by wbMRI+DWI, treatment plan was changed in one BCa and one PCa patients, respectively. These lesions were confirmed to be metastases by the imaging and clinical follow-up.

Estimates of price range for each of the imaging modalities, patient examination time, and reading time are shown in Table IV. Please note that the price ranges are just estimates based on data available from Finland's healthcare system. Nevertheless, it can be seen that ¹⁸F-NaF PET/CT is associated with the higher cost which is about four times the cost of ^99mTc-HDP SPECT/CT and wbMRI+DWI. Moreover, the reading times (time a reader spends reporting the exam) are comparable between ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI+DWI.

Table IV. Estimated price ranges, examination times, and reading times of the imaging modalities.

	Estimated price ranges of the imaging modalities (EUR)	Examination time (min)	Reading time (min)
BS	300–500	30	5–10
SPECT	400–600	30	15–20
SPECT/CT	450–650	30	15–30
PET/CT	1800–2200	30	15–30
wbMRI+DWI	450–650	50	20–30

BS, ^99mTc-hydroxymethane diphosphonate bone scintigraphy; PET/CT, ¹⁸F-NaF positron emission tomography combined with computed tomography; SPECT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography; SPECT/CT, ^99mTc-hydroxymethane diphosphonate single photon emission computed tomography combined with computed tomography; wbMRI+DWI, whole body MRI (T1-weighted imaging, STIR) including diffusion weighted imaging.

Discussion

Accurate detection of bone metastases in BCa and PCa patients is vital for treatment planning and patient prognosis. In the current study, we have evaluated diagnostic accuracy of ^99mTc-HDP BS, SPECT, SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI, including DWI, for the detection of bone metastases in high risk BCa and PCa patients. ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI, including DWI, had significantly higher AUC values for the detection of bone metastases than ^99mTc-HDP BS and SPECT. Furthermore, ¹⁸F-NaF PET/CT and wbMRI+DWI demonstrated significantly higher sensitivity and AUC values in lesion-based analysis than conventional nuclear medicine techniques (Figures 3, 4 and 5). To our knowledge, this is the first prospective clinical trial directly comparing the diagnostic accuracy of ^99mTc-HDP BS, SPECT, SPECT/CT, ¹⁸F-NaF PET/CT, and wbMRI, including DWI, for the detection of bone metastases in high risk BCa and PCa patients.

The current study enrolled high risk BCa and PCa with the aim to compare diagnostic accuracy for the detection of bone metastases rather than to evaluate the impact of different imaging modalities on diagnosis and treatment planning. Similarly to recent study by Lecouvet et al. [19], our end points were pure diagnostic with special focus on bone metastases. Whole body MRI, including DWI, demonstrated high diagnostic performance for the detection of bone metastases and was significantly more sensitive than BS, which is still the most widely used imaging modality for the detection of bone metastases.

Accurate evaluation of bone metastatic spread is becoming more important with introduction of novel therapeutic option [4]. In the current study, BS and SPECT missed four and two patients with bone metastases of 53 in per patient analysis, respectively. However, only two of those four patients had change in the treatment management (initiation of hormonal therapy) since the remaining two patients had hormonal therapy at the time of imaging. This could potentially raise a question, what is the current clinical additive value of scanning high risk BCa and PCa patients with more advanced methods? The major difference, however, is in the diagnostic certainty or uncertainty with BS and SPECT due to high numbers of equivocal lesions compared with SPECT/CT, PET/CT or wbMRI+DWI. Equivocal lesions of BS were present in 39% (22/53) of patients and in 25% (13/53) of patients only equivocal lesions with or without benign lesions were present. Similarly, 38% (20/53) of patients had equivocal lesions on SPECT and 17% (9/53) of patients had only equivocal lesions with or without benign lesions on SPECT. This group of patients causes a major diagnostic dilemma and therefore often require further imaging studies to reveal the nature of lesions. This leads to unnecessary delays in diagnosis causing patient discomfort and extra costs. Repeated imaging studies aiming to detect bone metastases are commonly needed in treatment planning of patients and cumulative radiation dose could become an issue, at least among the youngest patients.

Our finding of wbMRI+DWI and ¹⁸F-NaF PET/CT being superior to conventional nuclear imaging techniques for discovering bone metastases is in line with previous studies [6,7,9,19]. Considering the cost (Table IV), availability and radiation dose, wbMRI+DWI may be a preferred choice in comparison with ¹⁸F-NaF PET/CT. Whole body MRI, including DWI, was as accurate as ¹⁸F-NaF PET/CT for the detection of bone metastases in the current study. Moreover, wbMRI+DWI can potentially provide useful information concerning soft tissues [19]. Our findings support the use of wbMRI+DWI as a “single-step” imaging method for the detection of both soft and bone metastases in high risk BCa and PCa patients. The strategy of “single-step” detection of metastases using wbMRI+DWI in high risk PCa was previously suggested by Lecouvet et al. [19]. The wide use of this imaging modality for the detection of bone metastases in high risk BCa and PCa patient could still be partly limited by rather long imaging time of 40–50 min. However, further development of MR hardware and sequences could allow performing robust wbMRI+DWI in less than 30 min.

In the current study, DWI was evaluated in conjunction with T1wi and STIR to reduce number of false positive lesions [15] and only visual approach was used for DWI data, trace images. Quantitative evaluation of DWI could potentially allow monitoring of therapy response and early detection of responders from those patients who need change in treatment [23], potentially strengthening a role of wbMRI+DWI as “single-step” imaging modality for detection of bone metastases in high risk BCa and PCa patients.

Correct assessment of true nature of the lesions is the main limitation of the current study because histological confirmation was not available. In addition to consensus reading of all imaging modalities, clinical, and imaging follow-up of at least six months were used to define true nature of the lesions detected by each of the modalities. This approach is similar to many other studies focusing on the detection of bone and/or lymph node metastases [7,19,24,25]. However, this can result in overestimation of diagnostic accuracy of the most accurate imaging modality [19]. Differences in experience of ^99mTc-HDP BS, ^99mTc-HDP SPECT, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI readers could potentially also have influence on the results. However, all of the readers are highly experienced nuclear medicine physicians and/or radiologists with at least five years of experiences in the detection of bone metastases using that particular imaging modality. The current study is further limited by relatively small number of patients.

In the current study, ^99mTc-HDP SPECT/CT, ¹⁸F-NaF PET/CT and wbMRI+DWI were significantly more sensitive than ^99mTc-HDP BS, ^99mTc-HDP SPECT for the detection of bone metastases in high risk BCa and PCa patients. Whole body MRI, including DWI, was as accurate as ¹⁸F-NaF PET/CT for the detection of bone metastases in high risk BCa and PCa patients. In the context of nuclear medicine techniques, ^99mTc-HDP SPECT/CT was superior to ^99mTc-HDP BS, and SPECT, especially SPECT/CT having less equivocal findings.

Acknowledgments

This study was financially supported by grants from the Instrumentarium Research Foundation, Sigrid Jusélius Foundation, Turku University Hospital, TYKS-SAPA research fund, Finnish Cancer Society, and Finnish Cultural Foundation.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.