Introduction
Renal cell carcinoma (RCC) accounts for approximately 5% of adult cases of cancer in men and 3% in women1. It is known for its strong metastatic potential. Metastatic disease is diagnosed at presentation in up to 18% of patients with RCC1 and during follow-up, post nephrectomy in 50% of patients2. Bone is the second most common site of metastasis in metastatic RCC and is involved in 20-60% of patients3-6.
A bone metastasis from a clinically occult primary tumour is frequently the mode of presentation in patients with RCC7-9, as seen in 48% of patients with RCC bone metastases10. Solitary bone metastases are relatively frequent, occurring in 2.5% of all patients with RCC3,11, with bone being the most common site of solitary metastasis. Imaging features of many bone tumours are nonspecific. A RCC bone metastasis, particularly if solitary, can mimic a malignant primary bone tumour, and its diagnosis on imaging often relies on the availability of adequate clinical information.
The “flow-void” sign is a distinctive imaging feature in RCC initially described by Choi et al12 in 2003. However, little attention has been paid to this sign since then, with just one citation on PubMed and 24 citations on Google Scholoar, 8 of which related to RCC bone metastases. Flow voids result from signal loss due to blood flowing out of slice before signal can be sampled and are therefore a manifestation of rapid flow in arteries and also turbulence-related dephasing13. They are demonstrated as serpiginous, tubular areas without signal on all pulse sequences and correlate well with the presence of pathological vessels12,1. Flow voids have also been described in other bone lesions, including benign lesions, such as haemangiomata, and in rare vascular tumours such as angiosarcoma and haemangiopericytoma.
The purpose of this study was to evaluate the frequency of the “flow-void” sign in a large series of pathologically proven RCC bone metastases referred to a specialist unit. From our experience we note that the morphology of the “flow-void” sign can vary. Therefore, we also aim to describe a proposed grading system for the description of the “flow-void” sign.
Materials and methods
Patients
A retrospective search was performed using the radiology information system in our department to identify patients with RCC bone metastases. Patients with magnetic resonance (MR) imaging performed between September 2007 and December 2017 were included in the study. Patient demographics, the site and dimensions of bone lesions and the presence of single or multiple lesions were recorded. The history of known RCC prior to presentation with the index bone lesion was also recorded using the oncology database in our institution. Institutional review board approval was not required for the retrospective data collection used in this study.
Imaging
MR imaging performed in our institution was performed with a 3.0 Tesla MRI (Magnetom Skyra; Siemens, Erlengen, Germany). MR images were obtained in at least three orthogonal planes in each patient with a slice thickness of 3-10 mm. Intravenous gadolinium was not routinely administered in our institution. We also included patients with MR imaging performed in external institutions in our study.
Image Evaluation
MR imaging was reviewed by 5 readers (JM, AMD, SLJ, RB AP) for the presence of the “flow-void” sign. The “flow-void” sign was defined as the presence of multiple dot-like or tubular structures with low signal intensity located within or around the lesion. A grading system was applied for the “flow-void” sign when present by one reader (JM) (figure 1). Type 1 was defined as multiple punctate areas without signal on all pulse sequences within or surrounding the lesion. Type 2 was defined as serpiginous, tubular areas of signal void on all pulse sequences visible on < 3 consecutive slices, measuring < 3mm in diameter. Type 3 was defined as multiple serpiginous, tubular areas of signal void on all pulse sequences visible on 3 consecutive slices, measuring 3mm in diameter.
Figure 1: Grading system used for evaluation of the "flow-void" sign on MRI.
Radiographs were also reviewed to evaluate for matrix mineralisation or trabeculation which could simulate the “flow-void” sign.
Statistical Analysis
Statistical analysis was performed with SPSS software (IBM Corp. Released 2016. IBM SPSS Statistics for Mac, Version 24.0. Armonk, NY: IBM Corp). Descriptive statistics were calculated. Kruskall-Wallis H test was used to calculate the difference between lesion size and the presence and type of “flow-void” sign. The association between the presence and type of “flow-void” sign and type of MRI sequence was calculated using a Pearson chi-square test. A p-value of <0.05 was considered statistically significant.
Results
One hundred and forty bone lesions in one hundred and twenty-three patients with histologically-proven metastatic RCC were reviewed. Patient demographics are detailed in Table 1 . The mean age in our patient cohort was 62.9 (range 36-84) years and seventy-nine (64.2%) patients were male. In our patient cohort, 50.4% presented with symptoms of bone metastases or pathological fracture without a known prior history of RCC. One hundred and four (84.6%) patients presented with a solitary lesion at diagnosis, while the remaining patients presented with 2 or more lesions.
Radiological findings are detailed in table 2. In our patient cohort, the mean size of RCC bone metastases at presentation was 67.4 mm, with a range of 16 – 63 mm. Ninety-nine (69.7%) lesions were present within long bones, 57.6% of which occurred in the proximal long bone. The pelvis was the next most common site (24 (16.9%) lesions). One hundred and thirty-two (94.3%) lesions demonstrated the “flow-void” sign on at least one sequence in each study. A statistically significant difference was demonstrated between lesion size and the presence and type of “flow-void” sign (table 3). Lesions demonstrating type 3 “flow-void” sign had mean dimensions of 82.2 mm compared to 47.3 mm for lesions that did not demonstrate the “flow-void” sign.
Six-hundred and thirty-six sequences were reviewed in this patient cohort. MRI sequences performed are detailed in table 4. Axial and coronal T1-weighted and coronal STIR sequences were performed most frequently. The association between the presence and type of “flow-void” sign and MRI sequence is detailed in table 5. In our patient cohort, a statistically significant association was found between the presence of the “flow-void” sign and both the type of sequence and the use of fat saturation (2 =23.0; p=0.001 and 2 =30.6; p=0.00 respectively). The “flow-void” sign was more frequently demonstrated on T2-weighted and proton density imaging rather than T1-weighted imaging. In addition, proton density imaging demonstrated type 2 and type 3 “flow-void” sign most frequently (75.4% of proton density sequences). The “flow-void” sign was also seen more frequently in fat saturated compared to non-fat saturated imaging. Type 2 and type 3 “flow-void” sign was demonstrated more frequently in fat saturated imaging (70.6% of fat saturated sequences). A statistically significant association was not found between the plane of imaging and presence of the “flow-void” sign (2 =4.7; p=0.58).
Discussion
Bone is the second most common site of metastasis from RCC and is associated with a poor prognosis. Shortened survival has been reported in patients with RCC bone metastases compared to other metastatic sites14,15. Like other bone tumours, imaging of RCC bone metastases can be nonspecific. When the suspicion for RCC bone metastases is raised, further investigations can be initiated to evaluate for an occult primary renal tumour. Targeted treatment such as surgical resection, radiotherapy and percutaneous transcatheter arterial embolisation have been shown to be safe and reliable for restoring mechanical bone stability, relieve pain, improve mobility and provide good function in most patients with RCC bone metastases5,10,16-22.
The “flow-void” sign in RCC bone metastases was initially described in 2003 by Choi et al12. In this small study of 16 patients, the “flow-void” sign was identified in up to 80% of lesions and correlated well with the presence of pathological vessels on CT and digital subtraction angiography. A further study of 13 patients with RCC bone metastases published in 2009 also reported the presence of flow voids in 80% of lesions23. In our patient cohort of 123 patients, the “flow-void” sign was present in 94.3% of RCC bone metastases. Although the specificity of the “flow-void” sign is unknown, its presence on MRI should raise the suspicion of RCC bone metastases. Thus, the “flow-void” sign should additionally serve as an indicator of hypervascularity when considering biopsy or other intervention of the lesion.
We also described a grading system for the “flow-void” sign. This proposed grading system is the first of its kind for the “flow-void” sign to our knowledge. In our patient cohort, a statistically significant difference was demonstrated between the type of “flow-void” sign and lesion size. With increasing size of a lesion, the signal voids within the lesion became more conspicuous and more closely resembled vessels. In addition, T2-weighted, proton density and fat saturated imaging also demonstrated the “flow-void” sign more frequently, with the latter two sequences also demonstrating type 2 and type 3 “flow-void” sign most frequently.
While this study evaluating the “flow-void” sign in RCC bone metastases constitutes the largest patient cohort to our knowledge, there are limitations to our study. The sensitivity and specificity of the “flow-void” sign was not determined as other bone tumours were not included in the study. In addition, as some patients in this cohort were referred from external institutions with imaging performed elsewhere, there was some heterogeneity across the MRI sequences performed.
In conclusion, the results of our study indicate that the “flow-void” sign is a common imaging feature within RCC bone metastases. When observed, this sign can be a useful imaging feature in the diagnosis of RCC bone metastases, particularly in cases of occult or previously treated and forgotten RCC.