MRI as a tool to assess surgical margins and pseudocapsule features directly following partial nephrectomy for small renal masses

The Institutional Review Board approved this study. All patients gave written informed consent. Data were prospectively collected between November 2014 and July 2015. Inclusion criteria were a clinically suspicious tumour for RCC, scheduled PN and obtained written informed consent. No exclusion criteria were imposed. Patients were recruited and consecutively selected at the urology outpatient clinic. In total, ten specimens containing 11 tumours from nine patients were included. One specimen contained an incidentally detected second tumour that was included for analysis. One specimen consisted of a benign cyst and in one specimen perioperative disintegration of the resection margin occurred, both were excluded from analysis (Table 1).

PN was performed by open or robotic approach. Before resection, warm ischaemia time was applied by clamping the renal artery. Tumour resection was done with a minimal healthy tissue margin technique. If feasible, enucleation along the tumour PC was performed without resecting healthy parenchyma. The resection technique was macroscopically scored by one observer (TO) as pure enucleation, hybrid enucleation, pure enucleoresection or hybrid enucleoresection or resection according to the surface–intermediate–base margin score [8].

Following resection, the specimen was transported to the MR suite. After inking the parenchymal SM and renal capsule site, the specimen was fixated with wooden pins on a paraffin block within a customised Perspex holder containing rows of Perspex pins 3 mm apart from each other (Fig. 1). This holder and pins were designed to align imaging orientation with pathology slicing. The complete setup was positioned in a glass container. The specimen was immersed in perfluropolyether oil (Galden, Solvay Solexis) to avoid susceptibility artefacts. MR imaging was acquired using a 7T horizontal wide bore (200 mm) ClinScan animal scanner interfaced to a clinical Siemens Syngo VB15 console (Bruker BioSpin). MR images were recorded using an integrated circular polarized transmit/receive 1H volume coil with a free inner diameter of 154 mm. To match the MR and histopathology slides, MR sequences were aligned with water-filled tubes in the Perspex holder. The imaging protocol contained anatomical T1-weighted 3D gradient echo, T2-weighted turbo spin-echo (TSE) sequences and diffusion-weighted imaging (DWI) with b-values of 0–100–500–1,000 s/mm² (see Table 2 for imaging parameters). Apparent diffusion coefficient (ADC) maps were calculated using the standard ADC post-processing available on the console. The number of slices was chosen to entirely cover the specimen. The total work-up time, measured from the time of the specimen removal until availability of the images, was recorded.

MR images were reviewed by a radiologist (JF) blinded for the pathology outcome with 15 years of experience in the field of MR imaging. Tumour size, smallest SM (defined as smallest distance from the tumour border towards parenchymal SM), SM negativity, presence or absence of a PC, and if present, continuity of the PC and extra PC extension (EPCE) were evaluated (Table 1).

To establish which sequence was most suitable for evaluation, the quality of the MR images was assessed by the reviewing radiologist (JF), who evaluated the specimen characteristics. Visual quality assessment was done by scoring visibility of SM and PC features using a 5-point scale for all used sequences. Score 1 reflected excellent visualisation; score 5 reflected the images as being non-diagnostic.

After the MR imaging the specimen maintained its position on the board. After at least 24 h of fixation in formalin it was cut in the same direction the MR images were obtained. To enable proper correlation between the MR images and histopathological slides, the landmarks on the board were used. The specimen slides obtained were paraffin embedded after which 4 μm haematoxylin and eosin slides were prepared. All tumour-containing slides were digitally scanned (3DHistech Panoramic250, SYSMEX Belgium N.V., and Olympus Dotslide, Olympus Nederland B.V.). While being blinded for the MR imaging results, histopathological annotation on all scanned slides was done by two residents in pathology (WR/EB) under supervision of a dedicated uropathologist with 25 years of experience (CH) using commercially available software (Aperio ImageScope v11.2.0.780, Aperio Technologies, Inc. and OlyVIA, Olympus).

All features as reported by the radiologist were also histologically evaluated. In addition, histological RCC subtype according to the 2016 World Health Organization (WHO) classification and International Society of Urological Pathology (ISUP) nuclear grading were reported [9, 10]. On histopathology a PC was defined as a band of fibrous connective tissue located at the interface of the tumour and adjacent renal parenchyma [11] (Table 1). EPCE was defined as complete penetration of the tumour through the PC. The following conditions were not considered EPCE: intact PC with tumour invasion, incomplete PC but no tumour protrusion in the renal parenchyma, and PC with protruding but no penetrating tumour [12]. A positive SM was defined as presence of tumour cells in the inked surface of the parenchymal resection border.

Median scores for ability to assess SM and PC features were excellent (score 1) using T2-weighted images. T1-weighted images median scores were acceptable (score 3) with outliers to non-diagnostic for assessment of both features. DW images were least usable with median scores being acceptable (score 3) and poor (score 4) for assessment of SM and PC features, respectively (Fig. 2 and Table 3).

Median smallest SM was 2.7 mm (0.2–9.0) and 0.6 mm (0.3–1.5) on radiological and pathological review (Pearson’s r 0.667; p = 0.147). One SM was on histological evaluation found to be positive, which was identified as such on MR imaging (Fig. 3). Two tumours showed a false-positive SM on MR imaging. Presence of tumour cells in the SM was not confirmed on histopathology in one case with a smallest SM of 0.5 mm (Fig. 2). One specimen showed EPCE and ingrowth of the tumour in a blood vessel towards the parenchymal resection border. Additional histopathological slides were cut but the SM remained negative with the smallest margin being 0.5 mm. The sensitivity and specificity for assessment of negative SM with MR imaging were 100% (9/9) and 75% (6/8), respectively.

A PC was histologically confirmed in six out of nine cases. Presence or absence of a PC was correctly identified on MRI in seven out of nine cases (78%); two scans were false positive. In both cases an anatomical structure mimicking a tumour PC was present. The first case concerned a papillary RCC outlined by a thin epithelial layer of a cystic space, thereby creating a delicate space that mimicked a PC (Fig. 4). In the second specimen, the renal capsule at the perirenal site together with longitudinal cut blood vessels and compressed parenchyma towards the surgical margin mimicked a PC. Sensitivity and specificity for presence of a PC were 100% (6/6) and 33% (1/3), respectively.

Histopathologically determined PCs were discontinuous in 67% (4/6) and EPCE was present in 50% (3/6), respectively. In one case both PC discontinuity and EPCE were not identified as such on MRI. Sensitivity and specificity for continuity and EPCE were 75% (3/4) and 100% (2/2) and 67% (2/3) and 100% (3/3), respectively.