1179

Ex vivo MRI evaluation of vulvar cancer in fresh wide local excision specimens for cancer localization and prediction of the surgical tumour free margins.

Jan Heidkamp¹, Petra Zusterzeel², Andor Veltien¹, Arie Maat³, Ilse van Engen-van Grunsven³, Maroeska Rovers⁴, and Jurgen Fütterer¹

¹Radiology and Nuclear Medicine, Radboud university medical center, Nijmegen, Netherlands, ²Obstetrics and Gynaecology, Radboud university medical center, Nijmegen, Netherlands, ³Pathology, Radboud university medical center, Nijmegen, Netherlands, ⁴Operating Rooms, Radboud university medical center, Nijmegen, Netherlands

Synopsis

Currently, perioperative information on the surgical margin status of wide local excision specimens containing vulvar cancer is only based on the surgeon’s estimation. In this study we performed both a non-blinded and a blinded annotation of vulvar cancer location and surgical tumour free margins in the ex vivo MRI obtained from fresh wide local excision specimens. Annotated whole-mount section histopathology slides obtained from totally included specimens were adhered as gold standard. There was high correlation and agreement between ex vivo MRI, and high NPV and PPV were obtained for vulvar cancer localization and identification of margins <8 mm.

Introduction

Wide local excision (WLE) combined with a sentinel node procedure or inguinal lymph node dissection is the preferred treatment of the primary tumour in vulvar squamous-cell carcinoma (VSCC)^1,2. Tumour free margins of ≥8 mm are considered the most important predictive factor for local recurrences^3–5. The main challenge in treating VSCC is to keep the surgical margins large enough to minimize the chance of local recurrence and simultaneously preserve as much genital tissue as possible to diminish morbidity to urethra, clitoral or anal sphincter. The surgeon is hampered by the fact that no accurate and topical information on the margin status is available during or directly after the surgery. Previously optimal imaging was investigated⁶, whereas this study assessed the value of ex vivo MRI in providing information on VSCC location and the surgical tumour free margin status in fresh WLE specimens. Both technical feasibility and clinical applicability were investigated.

Methods

After approval of our Institutional Review Board, written informed consent from all patients (n=9) was obtained. Consecutive patients with biopsy proven VSCC who were scheduled for WLE were prospectively included. Following surgery, the entire intact vulva specimen was positioned in a specifically designed container and was scanned using a 7 Tesla preclinical MR scanner interfaced to a Siemens console. The scan protocol was described previously⁶. After imaging, the specimens were completely cut into 3 mm thick slices and whole-mount H&E-stained slides were obtained for histological examination. A pathologist annotated VSCC area and minimal histological tumour free margins (3- and 9 o’clock, and basal) on each digitalized histological slide. Next, individual ex vivo MRI slices were correlated with the histological slides. An observer with the histological images at his disposal annotated VSCC area and minimal tumour free margins on the correlated ex vivo MRI (the non-blinded annotation). Subsequently, a radiologist performed annotation of the same features as well, however, without knowledge of the histology (the blinded annotation). The ex vivo MRI measurements were corrected for formalin fixation induced shrinkage, whereupon linear correlation (Spearman’s rho) and agreement (Bland-Altman analysis) with histology were assessed. Diagnostic performance of ex vivo MRI for the localization of VSCC and identification of margins <8 mm was investigated by calculating positive- and negative predictive values (PPV, NPV). The Dice similarity coefficient (DSC) was calculated to determine the spatial overlap between VSCC areas of the non-blinded and blinded annotations.

Results

In total 153 of 159 histological slides were correlated to ex vivo MRI. In the non-blinded annotation 91 margins were identified as <8 mm, 79 were confirmed by histology (PPV = 87%). Likewise, 110 out of 124 margins were correctly identified as ≥8 mm (NPV = 89%). Ex vivo MRI measurements were significantly linearly correlated and demonstrated good agreement with histology. In the blinded annotation, absence and presence of VSCC was correctly confirmed in 73 out of 81 (NPV = 90%) and in 65 out of 72 (PPV = 90%) ex vivo MRI slices respectively. A total of 64 out of 90 margins were correctly identified as <8 mm (PPV = 71%) and 83 out of 102 margins were correctly identified as ≥8 mm (NPV = 81%). The blindly annotated VSCC area and margins were also significantly linearly correlated and demonstrated good agreement with histology. Compared to the non-blinded annotation, the linear correlation between ex vivo MRI and histology was less strong and the 95% limits of agreement in the Bland-Altman analysis were wider. The mean DSC between non-blinded and blinded annotation of ex vivo MRI was 0.73.

Discussion

The non-blinded annotation of ex vivo MRI demonstrated that it is technically feasible to obtain VSCC location and surgical tumour free margin status in fresh WLE specimens with high correlation and agreement with histology using ex vivo MRI. Furthermore, the blinded annotation demonstrates the potential clinical applicability of ex vivo MRI by showing high PPV and NPV for prediction of VSCC and margins <8 mm. Limitations are the small number of patients and the fact that this was a one reader study. The ultimate goal is to obtain perioperative information on the surgical tumour free margins of WLE specimens using ex vivo MRI. In this respect the results are promising, but more research is required in the form of a randomized controlled trial.

Conclusion

Accurate localization of VSCC and measurements of the surgical tumour free margins in fresh WLE specimens using ex vivo MRI is technically feasible. The high NPV and PPV for localization of VSCC and identification of margins <8 mm demonstrate clinical applicability of the technique.

Acknowledgements

Ex vivo MRI was performed using equipment of the Preclinical Imaging Centre (PRIME) within the Radboud University Medical Center, Nijmegen, the Netherlands.

References

1. Ansink A, van der Velden J. Surgical interventions for early squamous cell carcinoma of the vulva. Cochrane Database Syst Rev. 2000;(2):CD002036.

2. de Hullu JA, van der Zee AGJ. Surgery and radiotherapy in vulvar cancer. Crit Rev Oncol Hematol. 2006;60(1):38–58.

3. Heaps JM, Fu YS, Montz FJ, Hacker NF, Berek JS. Surgical-pathologic variables predictive of local recurrence in squamous cell carcinoma of the vulva. Gynecol Oncol. 1990;38(3):309–14.

4. de Hullu JA, Hollema H, Lolkema S, et al. Vulvar carcinoma: The price of less radical surgery. Cancer. 2002;95(11):2331–8.

5. Chan JK, Sugiyama V, Pham H, et al. Margin distance and other clinico-pathologic prognostic factors in vulvar carcinoma: A multivariate analysis. Gynecol Oncol. 2007;104:636–41.

6. Heidkamp J, Zusterzeel P, Veltien A, Maat A, Van Engen-Van Grunsven I, Fütterer J. Ex vivo MRI evaluation of vulvar cancer to predict resection margins in fresh wide local excision specimens: a pilot study (0101). In: Proc Intl Soc Mag Reson Med 25. 2017.

Figures

The vulva specimen obtained after performing wide local excision of a vulva carcinoma on a 93-year-old patient. On the left, a slight oblique/top view showing the specimen pinned down on the paraffin layer using 5 wooden pins. Furthermore, the two opposing rows of rods can be appreciated. On the right, top view of the pinned down specimen. Sutures were applied to demarcate the 12 o’clock side of the specimen and the position at which the specimen was transected for surgical resection purposes.

Ex vivo MRI and histology from a fresh WLE specimen; (a) T1W image with limited contrast between tumour and surrounding tissue; (b-d) T2W images acquired using TEs of 13, 27, and 54 milliseconds. With increasing TE the contrast between tumour and healthy tissue increases, however, at the expense of signal; (e-f) ADC-map and b1000 image, the tumour is easily spotted. The images were unsuitable for tumour delineation and surgical margin assessment due to limited resolution and noise; (g) T2W image (TE, 14 milliseconds) in higher resolution providing high anatomical detail advantageous for tumour demarcation; (h) histology confirming FIGO-stage 3A VSCC.

VSCC and surgical margin annotation on ex vivo MRI and histological slide; (a) non-annotated ex vivo MRI for reference; (b) histological slide with VSCC of 14.3. Furthermore, 3 o’clock -, 9 o’clock -, basal margin of 5.3, 12.1, and 12.1 mm; (c) non-blinded annotation of VSCC with area of 16.0 mm2 and 3 o’clock-, 9 o’clock-, basal margins of 4.3, 13.4, 12.9 mm (formalin induced shrinkage corrected); (d) blinded annotation with VSCC area of 22.3 mm2 and 3 o’clock-, 9 o’clock-, basal margins of 3.2, 13.8, 12.6 mm. The DSC of VSCC area between (c) and (d) was 0.75.

For the non-blinded annotation, scatter plots (a-d) and Bland-Altman plots (e-h) showing the linear correlation and agreement between histology and ex vivo MRI. With (a, e) showing the VSCC-area variable and (b-d, f-h) showing the 3 o’clock, 9 o‘clock, and basal margin variables respectively. In (a-d): linear regression equation, Spearman’s rho (rs), regression line (solid line), unity line (dashed line), and two reference lines at 8 mm cut off (red dotted lines) are provided as reference. In (e-f): mean difference (solid line) and 95% limits of agreement (dashed lines) are provided as reference.

For the blinded annotation, scatter plots (a-d) and Bland-Altman plots (e-h) showing the linear correlation and agreement between histology and ex vivo MRI. With (a, e) showing the VSCC-area variable and (b-d, f-h) showing the 3 o’clock, 9 o‘clock, and basal margin variables respectively. In (a-d): linear regression equation, Spearman’s rho (rs), regression line (solid line), unity line (dashed line), and two reference lines at 8 mm cut off (red dotted lines) are provided as reference. In (e-f): mean difference (solid line) and 95% limits of agreement (dashed lines) are provided as reference.

Proc. Intl. Soc. Mag. Reson. Med. 26 (2018)

1179