Spinal Instability Neoplastic Score as a Predictor of Vertebral Fracture in Patients Undergoing Radiation Therapy for Spinal Metastases: A Single-Institution Study

Abstract

Purpose

The Spinal Instability Neoplastic Score (SINS) is the most common method of assessing spine stability in the setting of spinal metastases. We sought to assess (1) the SINS score as a predictor of vertebral compression fracture (VCF), (2) the risk contributions of the 6 SINS individual factors, and (3) other contributors to fracture risk.

Methods and Materials

In total, 194 patients with 391 spinal lesions that had not previously been treated with vertebroplasty/kyphoplasty, radiation therapy (RT), or surgery were enrolled before RT and followed for new or worsened fracture.

Results

A total of 187 patients who were treated to 361 vertebral levels underwent post-RT follow-up. Average follow-up time for patients on study was 9.4 months (range, 0.2-38.8 months). A total of 33 new or worsened fractures (9.1% of lesions followed) were observed, with an average time to fracture of 4.4 months (range, 0.1-27.8 months). Of all 6 SINS factors, 3 were found to be individually significantly associated with increased risk of fracture, these were: lesion location in L2-L4 [hazard ratio (HR) = 2.78, P = .04], mixed or lytic appearance on computed tomography (mixed HR = 3.87, P = .01, lytic HR = 2.68, P = .02), and <50% vertebral collapse (HR = 3.52, P < .01). SINS as a summated score was significantly associated with increased risk of fracture on multivariable analysis (P < .01). Use of bone-strengthening medications such as bisphosphonates was also significantly associated with decreased risk of fracture in multivariable analysis in stable (SINS ≤6) (HR = 0.10, P = .03) and potentially unstable (SINS, 7-12) lesions (HR: 0.18, P = .03).

Conclusions

These findings support that SINS is a useful tool for estimating VCF risk, with lesion location, metastasis bone morphology, and presence of <50% collapse being the strongest predictors. Additionally, findings suggest that bone-strengthening medications such as bisphosphonates may mitigate the risk of developing VCF after RT.

Introduction

The National Cancer Institute estimated that over 2 million new cases of cancer will be diagnosed in 2024, and that over 600,000 will die from their disease. The 5-year survival rate for patients with any cancer is increasing, with recent estimates placing 5-year survival at 69%.¹ The number of people expected to be living with metastatic cancer by 2025 has been projected at 700,000.² Bone is the third most common site of metastasis after liver and lung,³ and the spine is the most common bone to which cancer is known to spread.⁴ Radiation therapy (RT) is the treatment modality of choice for spine metastases in many clinical scenarios given its noninvasive and focal mechanism of treating cancer. However, RT is not without risk. Vertebral compression fracture (VCF) is one of the most common and potentially devastating post-radiation sequelae for patients with spinal metastases. VCF frequently results in focal pain and decreased functional status.⁵ Reports show that up to 40% of patients receiving RT for spinal metastases will suffer a VCF.⁶ The societal burden of skeletal-related events includes higher rates of opioid use⁷ and increased payer cost.⁸

The Spinal Instability Neoplastic Score (SINS) was developed by the Spine Oncology Study Group in 2010 to assess risk of VCF in the setting of spine metastases. SINS comprises 6 factors that are summated to determine risk of fracture. Five of the SINS factors are radiographic in nature and 1 is dependent on patient history taking (Table 1).⁹ Using the SINS criteria, a summation score is used to stratify a spine metastasis as biomechanically stable (SINS ≤6), potentially unstable (SINS 7-12), or unstable (SINS ≥13). Patients with stable spinal lesions are often referred for more conservative management using RT alone and/or systemic therapy. Patients with unstable lesions are often referred for consideration of stabilization procedures in conjunction with RT or systemic therapy. Though some retrospective reports are available examining the predictive value of the SINS score,9, 10, 11, 12 further data are needed. The purpose of this study was to assess the utility of the summative SINS, individual components of SINS, and additional potential factors in determining VCF risk.

Table 1.

Spinal Instability Neoplastic Score framework

Category	Score
Location
Junctional (Occiput-C2*, T1-T2, T11-L1, L5)	3
Mobile (L2-L4)	2
Semi-rigid (T3-T10)	1
Rigid (S2-S5)*	0
Mechanical pain
Yes†	3
Occasional pain but not mechanical‡	1
Pain-free lesion	0
Bone lesion
Lytic	2
Mixed (lytic/blastic)	1
Blastic	0
Radiographic spinal alignment
Subluxation/translation present	4
De novo kyphosis/scoliosis	2
Normal alignment	0
Vertebral body collapse
≥50% collapse	3
≤50% collapse	2
No collapse, >50% of vertebra involved	1
No collapse	0
Posterolateral involvement
Bilateral	3
Unilateral	1
None	0

Adapted from Shi et al. 2018.

^⁎Lesions at these levels were excluded from analysis.

^†Defined as increased pain with movement of spine.

^‡Defined as pain in the spinal region in which the lesion was present where pain was not associated with movement.

Methods and Materials

This study is a post-hoc analysis of a cohort of spine metastasis RT patients. Patients with metastatic solid cancers were recruited from Dana-Farber Cancer Institute or Brigham & Women’s Hospital (DFCI/BWH) between September 2020 and July 2023. Patients were eligible if they were receiving RT for a site of solid tumor metastasis in the bony thoracic and/or lumbar spine, had no prior RT to the same vertebral level, had no prior surgery with hardware to the same or adjacent vertebral level(s), had no prior kyphoplasty or vertebroplasty to the same vertebral level(s), and did not have diseases of abnormal bone metabolism such as Paget’s or Cushing’s disease. Patients were added to the study at the time of spinal RT planning and were followed for new or worsened fracture. All patients provided written informed consent as part of our departmental institutional review board-approved biorepository protocol (the BROADBAND Study, MGB IRB 2016P001582).

Clinical data

Information collected at RT consult was primary tumor, vertebral level(s) to be irradiated, performance status [Eastern Cooperative Oncology Group or Karnofsky, as reported by the treating radiation oncologist at consult], age, sex (legal sex as reported in the medical record), height, and weight. On completion of RT, information collected was number of fractions delivered, total dose, modality/energy (mV), and technique used (stereotactic body radiation therapy [SBRT], intensity modulated radiation therapy [IMRT], 3-Field, or anterior-posterior posterior-anterior [AP PA]). On identification of a vertebral fracture, information collected was vertebral level(s) fractured, imaging modality used to identify the fracture, and whether the fracture was new or a worsening of an existing fracture. For SBRT plans, volumes and margins used were consistent with the International Spine Radiosurgery Consortium Consensus Guidelines 2012.¹³ Conventional spine RT volumes and margins were defined according to typical 2-dimensional or 3-dimensional planning and setup approaches. The use of bone-strengthening drugs (ie, denosumab, romosozumab, zoledronic acid, and pamidronate) was assessed as present for a patient when given at any time and duration during the metastatic cancer diagnosis before RT and up to 6 months post-RT. Patient information was extracted from the medical record, either manually via chart review or using the DFCI’s Oncology Data Retrieval System (OncDRS).

Baseline imaging and SINS

RT simulation computed tomography (CT) scans were used for pre-RT baseline. RT simulation was performed on either a Siemens SOMATOM Confidence or GE Lightspeed CT scanner and scans were reconstructed to the thinnest slice thickness possible for a given simulation, either 0.5 mm, 1.25 mm, or 1.5 mm, both at a skin-to-skin width and a 16 cm field of view centered on the spine.

Lesions were visualized in MIM version 7.1.12 (MIM Software, Inc) and scored by a trained study member (PD) according to SINS¹⁴ using the reconstructed RT simulation imaging and the pain assessment performed at the time of RT consultation.

Follow-up clinical visits and imaging

Follow-up information was collected from office visit notes, inpatient notes, and radiologist reports of follow-up imaging, either CT, MRI, positron emission tomography-CT, or X-ray (XR), which were completed as clinically necessary—typically at 3, 6, and 12 months after RT in the absence of clinical suspicion for fracture, or as needed where a fracture or progression was suspected. Reports of back pain were collected at follow-up appointments with various providers including radiation oncology, medical oncology, neurosurgery, orthopedics, and palliative care. Patients were considered to have suffered a vertebral fracture based on radiologist assessment in follow-up imaging. New or worsened fracture was defined as the mention in a follow-up radiology report of new or increased height loss, burst fracture, new or increased wedging of the vertebra, advancement of endplate deformity, or new fracture for a scan that included the treated vertebra. Fracture information was collected for all vertebrae; however, only fractures in treated vertebrae are included in this report. Treated vertebrae were defined as those receiving at least 50% of the dose (block edge) to the vertebral body. For time-to-event analyses, the absence of a vertebral fracture was determined by all available follow-up radiographic studies of the treated vertebra not showing evidence of a new or worsened fracture. Similarly, the timing of a fracture event was determined by the first appearance on follow-up radiographic imaging of a new or worsened fracture.

Statistical analysis

Time to new or worsened vertebral fracture was defined as time from the start date of RT to date of new or worsened vertebral fracture, or censored at the date of last assessment for those without a new or worsened vertebral fracture. Fine and Gray regression was used to assess associations between predictors and risk of new/worsened vertebral fracture. Mixed effects were also used to account for multiple lesions within a given patient. Univariable analyses were first conducted for each covariate, and predictors significant at a 0.05 level were included in the multivariable model along with the SINS composite score. Individual SINS components were not included in multivariable modeling to prevent bias from collinearity. Hazard ratios (HRs) reflect the risk of new/worsened vertebral fracture when comparing to reference levels. All analyses were done on R 4.1.0 [R Core Team (2024). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/].

Results

Information from 194 patients was collected at baseline, with 391 unique vertebral levels eligible to be followed for new or worsened fracture. The majority of the patients were male and the most common cancer types were prostate, lung, and renal (Table 2). For the 391 unique vertebral levels involved, the mean SINS score was 5.5.

Table 2.

Characteristics of 194 patients and of 391 spine metastases included for analysis

Patient characteristics (N = 194)	Descriptive statistic
Age (y)
Mean (SD)	64.2 (12.6)
Sex
Male	129 (66.5%)
Female	65 (33.5%)
Primary tumor type
Bladder	1 (0.3%)
Breast	53 (13.6%)
GYN	9 (2.3%)
H&N	13 (3.3%)
Liver	5 (1.3%)
Lower GI	11 (2.8%)
Lung	69 (17.6%)
Pancreatic	6 (1.5%)
Prostate	139 (35.5%)
Renal	53 (13.6%)
Sarcoma	23 (5.9%)
Thymic	3 (0.8%)
Thyroid	2 (0.5%)
Upper GI	4 (1.0%)
Number of vertebral bodies involved
Median (range)	1 (1-9)
Body mass index
Mean (SD)	27.2 (5.4)
Spine metastases characteristics (N = 391)
SINS components
Lesion location
Mean (SD)	2.1 (0.9)
Median (range)	2.0 (1.0-3.0)
Pain
Mean (SD)	1.1 (1.2)
Median (range)	1.0 (1.0-3.0)
Bone lesion
Mean (SD)	0.8 (0.9)
Median (range)	0.0 (0.0-2.0)
Alignment
Mean (SD)	0.3 (0.7)
Median (range)	0.0 (0.0-2.0)
Collapse
Mean (SD)	0.4 (0.8)
Median (range)	0.0 (0.0-3.0)
Involvement of posterolateral   spinal elements
Mean (SD)	0.8 (1.0)
Median (range)	0.0 (0.0-3.0)
SINS summated score
Mean (SD)	5.5 (2.8)
Median (range)	5.0 (1.0-14.0)
Prior vertebral fracture
No	341 (87.2%)
Yes	50 (12.8%)
Radiation therapy technique
3D	194 (49.6%)
IMRT	18 (4.6%)
SBRT	179 (45.8%)
Radiation therapy total dose,   median (range)	30 Gy (8-40)
Radiation therapy fractions,   median (range)	5 (1-10)
BED*
Mean (SD)	82.4 (38.2)
Median (range)	60.0 (29.3-153.3)
BMI†
Mean (SD)	26.4 (5.1)
Median (range)	25.5 (15.4-49.3)

Abbreviations: GI = gastrointestinal; GYN = gynecologic; H&N = head & neck; IMRT = intensity modulated radiation therapy; SBRT = stereotactic body radiation therapy; 3D = 3-dimensional; BED = biologically effective dose; BMI = body mass index; RT = radiation therapy; SINS = Spinal Instability Neoplastic Score.

^⁎α/β of 3 used to determine impact of RT on bone.

^†Includes one patient with multiple lesions who had a leg amputation.

In total, 7 patients who had a combined 30 vertebral levels treated were without imaging follow-up because of death and were excluded, leaving 187 patients who were treated at 361 vertebral levels that were followed for new or worsened fracture. The most common dose regimen used for SBRT patients was 35 Gy in 5 fractions (110 vertebrae), and the most common conventional RT dose regimen was 20 Gy in 5 fractions (115 vertebrae). From the start of RT, the median follow-up was 9.4 months (range, 0.2-38.8 months). Among the 361 treated metastatic vertebral sites, 9.1% developed a new or worsened fracture at a median time of 4.4 months after RT (range, 0.1-27.8 months) (Table 3). In the group treated with SBRT, 18 out of 179 lesions developed a new or worsened fracture (10.1%, median time to fracture 9.7, range, 0.1-27.8, months), and in the group treated not with SBRT, 15 out of 182 lesions developed a new or worsened fracture (8.2%, median time to fracture 2.8, range, 0.4-8.4, months).

Table 3.

New or worsened vertebral fractures among 361 treated vertebral metastases

	Summary
Number of events among total lesions (N = 361), N (%)	33 (9.1)
Number censored* among total lesions (N = 361), N (%)	328 (90.9)
Number of patients with new/worsened vertebral fracture (N = 187), N (%)	26 (13.9)
Time to new/worsened vertebral fracture (months)†, median (range)	4.4 (0.1–27.8)
Time to last assessment for patients alive at data compilation (months)‡, median (range)	9.4 (0.2-38.8)

^⁎Censored at date of last scan.

^†Not a Kaplan-Meier estimate; includes multiple lesions for some patients.

^‡Not a Kaplan-Meier estimate; among those lesions without a new/worsened vertebral fracture; includes multiple lesions for some patients.

Clustered Fine and Gray regression analysis was performed with death as a competing risk for each component of SINS, SINS as a summated score, and several potential clinical covariates as detailed in Table 4. On univariable analysis, lesions located within the L2-L4 vertebrae (HR = 2.78, 95% CI, 1.07-7.24, P = .04), lesions with a mixed (lytic/blastic) or lytic appearance on CT (HR = 3.87, 95% CI, 1.34-11.18, P = .01; HR = 2.68, 95% CI, 1.13-6.35, P = .02, respectively), and lesions causing <50% vertebral collapse (HR = 3.52, 95% CI, 1.61-7.71, P < .01) were found to be associated with a higher risk of vertebral fracture. The summated SINS score was found to be statistically significant on multivariable analysis (univariable P = .07, multivariable P < .01).

Table 4.

Results of clustered Fine and Gray regression for time to new/worsened vertebral fracture with death as competing risk (N = 361^*)

	Univariate		Multivariable
	HR† (95% CI)	P	HR† (95% CI)	P
Age			–	–
≤67	Ref.
>67	1.28 (0.56-2.89)	.56
Sex			–	–
Male	Ref.
Female	0.71 (0.32-1.61)	.42
Lesion location			–	–
1	Ref.
2	2.78 (1.07-7.24)	.04
3	1.77 (0.65-4.80)	.26
Pain			–	–
0	Ref.
1	1.59 (0.65-3.91)	.31
3	1.03 (0.34-3.19)	.95
Bone lesion			–	–
0	Ref.
1	3.87 (1.34-11.18)	.01
2	2.68 (1.13-6.35)	.02
Alignment			–	–
0	Ref.
2	1.64 (0.69-3.92)	.26
4	0 (-)	–
Collapse‡			–	–
0	Ref.
1	0.99 (0.24-4.06)	.99
2	3.52 (1.61-7.71)	<.01
3	– (–)
PL involvement			–	–
0	Ref.
1	0.73 (0.30-1.76)	.48
3	0.89 (0.28-2.84)	.84
SINS Score	1.14 (0.99-1.30)	.07	1.20 (1.06-1.37)	<.01
SINS Score (factor)			–	–
1-3	Ref.
4-5	0.83 (0.32-2.14)	.70
6	0.60 (0.12-2.88)	.52
7	1.91 (0.53-6.90)	.32
>7	2.07 (0.79-5.42)	.14
Technique‡			–	–
3D	Ref.
IMRT	1.04 (0.17-6.29)	.96
SBRT	1.23 (0.54-2.82)	.62
BED			–	–
≤69	Ref.
>69	1.20 (0.53-2.68)	.66
BMI§			–	–
≤26	Ref.
>26	0.56 (0.25-1.22)	.14
Bisphosphonate
Negative	Ref.		Ref.
Positive	0.32 (0.12-0.83)	.02	0.25 (0.11-0.60)	<.01

Abbreviations: IMRT = intensity modulated radiation therapy; PL = posterolateral; SBRT = stereotactic body radiation therapy; 3D = 3-dimensional; BED = biologically effective dose; BMI = body mass index; RT = radiation therapy; SINS = Spinal Instability Neoplastic Score.

^⁎Excludes lesions without any follow-up.

^†HR represents hazard ratio comparing group to reference group.

^‡Very few lesions in certain groups.

^§Excludes one patient who had amputation (this patient had 4 lesions in this analysis).

The relationships of several non-SINS characteristics with subsequent fracture were also analyzed, such as RT technique used, primary tumor histology, biologically effective dose, patient body mass index, and use of bone-strengthening agents (such as bisphosphonates). Of these, the use of bone-strengthening agents was found to be significant on univariable analysis (P = .02) and on multivariable analysis (P < .01). Furthermore, clustered Cox regression analysis with competing risks was performed and showed no significant interaction between SINS composite score and RT technique when comparing conventional RT and SBRT/intensity modulated radiation therapy (IMRT).

To better elucidate the effects of bone-strengthening agents on vertebral fracture risk, patients were stratified by SINS score and mixed effects Cox regression analysis was performed for patients who did or did not have a history of bone-strengthening agent use, as shown in Table 5. Of 361 treated vertebral levels, 176 (48.8%) vertebral levels were treated among patients who had received and/or were to receive bone-strengthening agents within 6 months of RT. The risk of vertebral fracture in patients with SINS score of 0-6 was significantly greater in patients who were not on a bone-strengthening agent (HR = 0.10, 95% CI, 0.01-0.82, P = .03). Similarly, among patients with a SINS score of 7-12, patients not receiving bone-strengthening agents had a higher fracture risk (HR = 0.18, 95% CI, 0.04-0.85, P = .03).

Table 5.

Mixed effects cox regression by bisphosphonate use in SINS subgroups

	HR* (95% CI)	P
SINS Score 0-6 (N = 242)†
Bisphosphonate
Negative	Ref.
Positive	0.10 (0.01-0.82)	0.03
SINS Score 7-12 (N= 112)†
Bisphosphonate
Negative	Ref.
Positive	0.18 (0.04-0.85)	0.03
SINS Score 13+ (N= 7)‡
Bisphosphonate
Negative	–	–
Positive

Abbreviation: SINS = Spinal Instability Neoplastic Score.

^⁎HR represents hazard ratio comparing group to reference group.

^†Some patients can be included in all 3 subgroups.

^‡No events in negative group; cannot compute.

Discussion

This study of patients receiving RT—approximately half with conventional RT and half SBRT—for spine metastases from solid tumors demonstrated a low rate of vertebral fractures after RT at 9.1%. This study provides further data to support the utility of the SINS score to assess risk of fracture. Among the 6 components of SINS, those most predictive of vertebral fracture were lesions within the L2-L4 vertebrae, mixed (lytic/blastic) or purely lytic lesion appearance on CT, and lesions causing partial (<50%) vertebral collapse.

Additionally, the use of bone-strengthening agents (eg, bisphosphonates) was shown to significantly decrease the risk of new or worsened vertebral fracture in patients with either stable (SINS 0-6) or potentially unstable (SINS 7-12) lesions. Bone-strengthening agents such as bisphosphonates and denosumab have long been known to reduce the incidence of skeletal events, and were recommended by the European Society for Medical Oncology’s 2020 Clinical Practice Guidelines on bone health in patients with cancer indefinitely as soon as a bony metastasis is detected.¹⁵ Our finding corroborates this guideline.

The overall rate of VCF in this cohort is on the lower end of similarly sized studies previously reported, which have ranged from 8% to 40.5% in a recent meta-analysis.¹⁶ The SINS summative score has previously been shown to be a significant predictor of new or worsened vertebral fracture in prior largely retrospective data sets.^9,10,11,12 Similar analyses of the individual components of SINS have shown comparable results, such as study by Kim et al.¹², a meta-analysis which showed that lytic morphology and <50% vertebral collapse were associated with vertebral compression fracture. Sahgal et al.¹⁰ found that lytic tumor morphology, spinal alignment, and partial vertebral collapse were significant predictors of VCF. In addition, in the report by Sahgal et al.,¹⁰ SBRT dose ≥24 Gy/fraction was a significant predictor of VCF. Together with these prior reports, these data support the use of SINS as a summative score, and additionally suggest that the most critical components include lytic morphology and pre-existing partial fracture. Further study is needed to determine if the SINS might benefit from amendment, such as to weigh these factors in the SINS score more highly or whether components that are not significantly associated with increased VCF risk can be omitted.

Other work has focused on incorporating additional factors into the prediction of VCF in patients treated with RT for spinal metastases, such as Kowalchuk et al.¹¹ This work examined patients undergoing SBRT for spinal metastases exclusively and reported significant VCF predictive value in the following variables: SINS > 6, lumbar spine location, gross tumor volume >10 cc, and epidural extension. Our analysis found that the use of bone-strengthening agents is independently associated with a decreased risk of VCF in patients with SINS < 13. This finding is unique in that it provides an additional, modifiable factor determining VCF risk after RT. This finding necessitates further study. However, data regarding agents such as bisphosphonates and denosumab¹⁵ already support the use of these agents as an important tool to reduce the incidence of skeletal events in patients with metastatic cancer. In clinical practice, the benefits of bone-strengthening agents, particularly for patients with spine metastases, need to be carefully weighed against the potential side effects (such as hypocalcemia and jaw necrosis).

This study has several limitations. First, our study cohort contained more patients with prostate cancer (35.5%) than any other primary tumor histology. Second, most patients at BWH/DFCI are treated as part of a multidisciplinary spine tumor program consisting of radiologists, radiation oncologists, neurosurgeons, orthopedic surgeons, palliative care physicians, and others. Thus, patients with high SINS scores were often treated with surgery or vertebral augmentation before they could receive RT, making them ineligible for the study, skewing our patient population to those with lower SINS scores. A third limitation is the exclusion of cervical and sacral spine lesions, which may limit the application of these findings to lesions outside of the thoracic and lumbar spine. Fourth, when including competing risks in our analysis of the relationship between RT technique and SINS composite score, we did not consider frailty, which may have limited our ability to analyze within-cluster effects. Finally, our assessment of bone-strengthening agent use was limited given its categorization of any use versus none.

Conclusions

These data indicate that SINS is a useful predictor of VCF, with certain SINS components being most strongly associated with VCF risk. Additional work refining and simplifying the SINS framework may be beneficial when SINS is used to predict the risk of post-RT VCF. This study also suggests that the use of bone-strengthening agents is correlated with decreased risk of vertebral fracture in patients who received RT to uncomplicated spinal metastases. Expansion of the use of bisphosphonates should be considered in this population.

Disclosures

No artificial intelligence (AI) or AI-assisted technologies were used in the creation of this article, not including the use of basic tools for checking grammar, spelling, references, etc. SC, NK, KS, MDM, TJ, HK, TP, CN, MAH, EK, ST, CR, RA, and TB report no conflicts of interest. Patrick Doyle: consulting fees from Merck. Michael Groff: royalties or licenses from Globus Spine. Alexander Spektor: grant funding from the National Cancer Institute and the Burroughs-Wellcome Fund Career Award for Medical Scientists; support for attending meetings and/or travel from the International Congress of Radiation Research and the Korean Society for Biochemistry and Molecular Biology. Wenxin (Vincent) Xu: grant funding from the Department of Defense; consulting fees from Aveo, Merck, Celdara, and Deciphera; Scientific Advisory or Data Safety Monitoring Board for Eisai, Exelixis, Xencor, and Jazz; research support (paid to institution) from Merck, Oncohost, and Arsenal Biosciences. David Kozono: consulting fees from Genentech/Roche; payment or honoraria from UpToDate/Wolters Kluwer; leadership or fiduciary role on the Alliance for Clinical Trials in Oncology. David Hackney: leadership or fiduciary role on the BIDMC Radiology Foundation Board. All other authors have no disclosures to report.