- Email: [email protected]
The accuracy of magnetic resonance imaging in staging of vulvar cancer: A retrospective multi-centre study
Gynecologic Oncology 117 (2010) 82–87
Contents lists available at ScienceDirect
Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o
Original Research Article
The accuracy of magnetic resonance imaging in staging of vulvar cancer: A retrospective multi-centre study Masako Yano Kataoka a,⁎, Evis Sala a, Peter Baldwin b, Caroline Reinhold c,d, Alex Farhadi c, Turtko Hudolin e, Hedvig Hricak e a
Department of Radiology, University of Cambridge, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK Macmillan Consultant in Gynaecological Oncology, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK Department of Diagnostic Radiology, McGill University Health Centre, Montreal General Hospital, 1650 Cedar Avenue, Montreal, Quebec, Canada H3G 1A4 d Medical Director Oncology, Synarc, Inc., San Francisco, CA, USA e Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA b c
a r t i c l e
i n f o
Article history: Received 15 September 2009 Available online 21 January 2010 Keywords: MRI Vulvar cancer Lymph node metastasis Contrast enhancement Gynecology Malignancy
a b s t r a c t Objective. To retrospectively evaluate the diagnostic accuracy and clinical relevance of magnetic resonance imaging (MRI) in the management of primary and recurrent vulvar cancer and to examine the added value of contrast-enhanced MRI (CE-MRI). Methods. The research ethics committee waived informed consent for this study of 49 patients with vulvar cancer (36 primary and 13 recurrent) who underwent MRI before surgery at three major cancer centers from December 2003 to January 2008. CE-MRI was available for 31 patients (20 primary and 11 recurrent). MR images were retrospectively evaluated by three radiologists for tumor size and stage. Lymph nodes with a short axis N5 mm were measured and scored for contour, presence of necrosis, loss of fatty hilum, signal intensity relative to the vulvar lesion, and reader's diagnosis of lymph node metastasis. Scoring was repeated for CE-MRI. Histopathology constituted the reference standard. Results. The size of the vulvar lesion was accurately characterized on MRI in 83% of patients. Accuracy in staging of primary vulvar cancers on unenhanced MRI was 69.4% (n = 36). Adding CE-MRI increased lesion detection and raised staging accuracy from 75% to 85% (n = 20). For groin lymph node metastasis prediction, the ratio of the short- to the long-axis diameter and the reader's diagnosis of lymph node metastasis yielded accuracy of 85% and 87%, respectively, in groin-by-groin analysis. Conclusion. MRI can be useful in accurately assessing the size of vulvar lesion and groin lymph node metastasis. CE-MRI may be of help in improving local staging. © 2009 Elsevier Inc. All rights reserved.
Introduction Vulvar cancer accounts for 3–5% of all gynecologic malignancies [1, 2]. The disease shows a bimodal age distribution, and its incidence in young women (b 50 years old) is increasing due to the spread of the human papilloma virus (HPV) [3]. Vulvar cancer may directly invade adjacent structures or spread to inguinal and femoral lymph nodes. Early stage vulvar cancer is treated by partial vulvectomy, with either ipsilateral or bilateral inguino-femoral lymph node dissection [4]. Surgery is the treatment of choice even for locally advanced cancer, although neoadjuvant radiation and chemotherapy may be used
⁎ Corresponding author. Fax: +44 1223 330915. E-mail addresses: [email protected] (M.Y. Kataoka), [email protected] (E. Sala), [email protected] (P. Baldwin), [email protected] (C. Reinhold), [email protected] (A. Farhadi), [email protected] (T. Hudolin), [email protected] (H. Hricak). 0090-8258/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2009.12.017
either to enable surgery in cases deemed ‘inoperable’ at initial presentation or to avoid exenteration in an elderly population [5]. The most important prognostic factors in vulvar cancer are tumor size, depth of invasion and the presence of lymph node metastases, which are incorporated into the International Federation of Gynecology and Obstetrics (FIGO) staging. This classification can predict overall survival. In addition, surgical margin distance is an important predictor of local recurrence. A tumor-free pathologic margin of b 8 mm is associated with local recurrence of 23–50% [6, 7]. As the margin distance is a potentially modifiable factor, accurate preoperative evaluation of the primary tumor is important. Only one study has evaluated the role of MRI in staging primary vulvar cancer in 20 cases [8]. Contrast-enhanced MRI (CE-MRI) was performed in only 4 of the 20 patients. The accurate detection and treatment of lymph node metastases in vulvar cancer is critical and relapse in the lymph node basin is associated with an extremely poor survival [9]. Clinical examination is of limited use in the detection of abnormal inguinal lymph nodes
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
(sensitivity, 57%; specificity, 62%) [10]. For tumours with a depth of invasion of N 1 mm, surgical management will typically include formal inguino-femoral node dissection on one or both sides [11]. Although outcomes are excellent in early stage disease, lymphadenectomy is associated with considerable morbidity including wound infection, breakdown, lymphocyst formation and long-term lymphoedema. Accurate preoperative information regarding the lymphatic basin could reduce the need for unnecessary lymphadenectomy with a resultant reduction in surgical morbidity. Four studies have used MRI to evaluate lymph node metastasis in vulvar cancer [8,12-14] They report widely varying sensitivity and specificity ranging from 40% to 89% and 82% to 100%, respectively [8,12]. The purpose of this study was to retrospectively evaluate (1) the diagnostic accuracy of MRI for vulvar cancer (both the main vulvar lesion and lymph node metastasis), and (2) to examine the potential added benefit of CE-MRI above ‘standard’ MRI imaging. Material and methods Study population The research ethics committee approved and issued a waiver of informed consent for our retrospective study. Through computerized review of the gynecologic surgery and radiology databases of three institutions, consecutive patients were identified who had histologically confirmed vulvar carcinoma and underwent MR imaging within 8 weeks (mean = 3.1 weeks) before surgery from December 2003 to January 2008. A total of 49 patients (38 from institution 1, 7 from institution 2, and 4 from institution 3) formed the study population. Table 1 shows patient characteristics. MRI protocol At all three institutions, examinations were performed using a 1. 5 T whole-body MR system (GE Healthcare Technologies, Milwaukee, WI). The protocol for vulvar cancer patients included axial T1weighted spin-echo images (TR/TE, 700/16 ms), sagittal T2-weighted fast recovery fast spin-echo (FRFSE) images (4000–5000/85–90),
Table 1 Characteristics of patients with vulvar cancer (n = 49). Mean age (range)
70.4 (40–88)
Primary cancer Recurrent cancer MR examination Total Unenhanced MRI only Unenhanced + CE-MRI Primary cancer Recurrent cancer Mean interval between MRI and surgery (weeks) Surgery Wide local excision only + unilateral groin LN dissection + bilateral groin LN dissection + sentinel node dissection Radical vulvectomy only + unilateral groin LN dissection + bilateral groin LN dissection Partial/hemivulvectomy only + unilateral groin LN dissection + bilateral groin LN dissection Overall FIGO stage for primary cancer (n = 36) Stage I Stage II Stage III Stage IV
36 13
LN, lymph node.
49 18 31 20 11 3.1 8 8 9 8 2 1 8 0 1 4 7 9 11 8
83
axial and coronal T2-weighted fat-suppressed FRFSE images (4000– 5000/85–90). In 31 cases, an intravenous bolus injection of gadopentate dimeglumine (0. 1 mmol per kilogram body weight) was given at 3 mL. Axial T1-weighted fat-suppressed spin-echo images (600/16.5) were acquired before and after the injection. All sequences were obtained with a slice thickness of 3–5 mm. Image evaluation Images were displayed using GE PACS system. Two genitourinary MR radiologists (_ and _) with 6 and 8 years of experience, respectively, evaluated the MRI examinations. Both radiologists were aware of the diagnosis of vulvar cancer but were blinded to the final surgical and histopathological results when scoring the MRI examinations. Evaluation of vulvar lesion A five-point scoring system was used for the detection of vulvar lesions: 1, tumor definitely absent; 2, tumor probably absent; 3, tumor possibly present; 4, tumor probably present; 5, tumor definitely present. Vulvar lesion size was measured in antero–posterior, left– right, and cranio–caudal direction using two orthogonal views. The maximum length of the three measurements was recorded as the size of the lesion. MRI stage was classified according to the FIGO classification system for 36 patients with primary vulvar cancer (Appendix 1). In addition, for 31 patients who underwent CE-MRI, the above evaluations were repeated. CE-MRI was evaluated after the evaluation of their non CE-MRI images. In 20 of the 31 patients who underwent both unenhanced and CE-MRI, the incremental value of CE-MRI in MRI stage according to FIGO classification was assessed. Evaluation of groin lymph node metastasis Groin lymph nodes measuring more than 5 mm in the short-axis diameter on MRI were considered as potentially metastatic lymph nodes, based on reported specificity b 70% with short-axis diameter of ≤ 5 mm in the literature [8]. Both the long- and short-axis diameters were measured and scores were assigned for contour (1, smooth; 2, lobular; 3, irregular; and 4, spiculated), presence of cystic changes/ necrosis, loss of fatty hilum, and similarity of signal intensity to the primary lesion (5-point scoring system reflecting five levels of diagnostic confidence). In addition to individual findings described above, a five-point scoring system was also used to evaluate reader's diagnostic confidence for presence of metastasis (1, metastasis definitely absent; 2, metastasis probably absent; 3, metastasis possibly present; 4, metastasis probably present; 5, metastasis definitely present) [15]. Groin lymph node metastasis was assessed using “groin-by-groin” analysis. Since superficial and deep inguinal lymph nodes were combined on histopathology assessment, these two regions were jointly evaluated on imaging. The lymph node with the highest score (most malignant) per groin provided the overall score for that groin. Statistical analysis The accuracy levels of both unenhanced and CE-MRI for detection and staging of vulvar cancer were evaluated. Due to non-normal distribution, the Spearman correlation coefficient was used to evaluate the correlation between the size of the lesion on MRI and histopathology in both primary and recurrent disease. Wilcoxon's rank sum test was used to compare reader's confidence for detection of vulvar lesions using unenhanced MRI with that using CE-MRI. The hypothesis of no disproportionate change between stages (i.e., correct or incorrect stage) determined with unenhanced MRI alone and stage determined with unenhanced and CE-MRI was tested using McNemar's test.
84
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
Specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy were calculated for the evaluation of groin lymph node metastasis. Receiver operating curve (ROC) analyses were performed for long axis, short axis and short axis/long axis (S/L) ratio. The cut-off point for contour was determined as between 1 and 2 (absence of metastasis) and 3 and 4 (presence of metastasis). For all the features evaluated with the 5-point scoring system, the cut-off point was determined as between 1 to 3 (absence of metastasis) and 4 and 5 (presence of metastasis). Analysis of the MR features of groin lymph nodes was performed once using only groins with significant lymph nodes (i.e., nodes with a short axis N 5 mm) and again using all groins. A P value of less than 0.05 was considered statistically significant. All analysis was performed using STATA (version 8.0; STATA, College Station, TX). Results Surgery and histopathology Surgical and histopathologic findings were compared with MRI findings (Table 1). Invasion to lower urethra, vagina and anus was histopathologically confirmed in three, two and one patients, respectively. No patient had invasion to upper urethra, bladder, rectum, or pelvic bones. Histopathological information about groin lymph node metastasis, obtained either by lymph node dissection or sentinel node biopsy, was available in 61 groin nodes. The remaining 37 groin lymph nodes were not dissected or biopsied and therefore excluded from analysis. Groin nodes were positive for presence of metastases in 32 groins and negative in 29 groins. Evaluation of vulvar lesion: unenhanced MRI The vulvar lesion was visible on MRI in all 36 patients with primary vulvar cancer and 11 of 13 patients with recurrent tumor. The mean size of the vulvar lesion on MRI was 34. 7 mm (primary 35. 2 mm, recurrent 33. 1 mm). Twelve patients (primary 8, recurrent 4) had vulva lesion ≤ 2 cm and 35 patients (primary 28, recurrent 7) had vulva lesion N 2 cm. The mean histopathological size was 36. 2 mm (primary 36. 8 mm, recurrent 34. 2 mm). Size on MRI and size on histopathology showed moderate correlation (Spearman correlation coefficient: 0.72, primary 0.72, recurrent 0.57). MRI accurately classified 31 out of 36 primary vulvar cancers (accuracy 86%) and 8 out of 11 recurrent vulvar cancers (accuracy 72%) into the correct size category (overall accuracy of 83%).
Table 2 Diagnostic accuracy of staging using unenhanced MRI and CE-MRI: correlation with histopathology (n = 20). Histopathology
MRI Stage I Stage II Stage III Stage IV Total Overall accuracy CE-MRI Stage I Stage II Stage III Stage IV Total Overall accuracy
Stage I
Stage II
Stage III
Stage IV
Total
3 2 0 0 5 75.0%
2 3 0 0 5
0 1 4 0 5
0 0 0 5 5
5 6 4 5 20
3 2 0 0 5 85.0%
0 5 0 0 5
0 1 4 0 5
0 0 0 5 5
3 8 4 5 20
Fig. 1. ROC curves of long axis (mm) (a), short axis (b), and S/L ratio (c) in diagnosing groin lymph node metastasis, with areas under the ROC curves of 0.67 (95% CI: 0.50– 0.84), 0.69 (95% CI: 0.52–0.86), and 0.86 (95% CI: 0.74–0.98) respectively.
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
The overall staging accuracy of primary vulvar cancers on unenhanced MRI was 69.4% (n = 36). For 36 patients with primary vulvar cancer, staging by MRI was as follows: stage I: 6, stage II: 9, stage III: 13, stage VI: 8. Six patients were overstaged, and five patients were understaged by MRI. Evaluation of vulvar lesion: the added value of CE-MRI Among 47 patients whose main vulvar lesion was visible on MRI, 30 patients (primary 20, recurrent 10) had CE-MRI. The mean size of the vulvar lesion on CE-MRI was 34. 1 mm (primary 34. 0 mm, recurrent 34. 2 mm). Five patients were reclassified after CE-MRI; three patients (primary 1, recurrent 2) were correctly reclassified as having a lesion “size N 2 cm,” and in the other two patients (primary 1, recurrent 1) CE-MRI overestimated the size of the vulvar lesion. CEMRI accurately classified lesion size in 25 out of 30 (85.1%) patients. The addition of CE-MRI did not change the accuracy of the size category of primary vulvar cancers (accuracy 85.0%); however, the accuracy of the size category for recurrent vulvar cancers changed from 70% to 80%. In 10 patients (primary 6, recurrent 4), reader confidence in the detection of the vulvar tumor was higher on CE-MRI (mean score 4.4, primary 4.5, recurrent 4.8) than on unenhanced MRI (mean score, 3.9; p = 0.004, primary 2.8, recurrent 3.5). In one of these patients with recurrent cancer, no residual invasive cancer was found in the surgical specimen and therefore this was considered as false positive. Table 2 shows the diagnostic accuracy of CE-MRI in 20 patients with primary vulvar cancer. CE-MRI accurately staged 17 patients, for an overall staging accuracy of 85%; it overstaged 2 patients and understaged 1 patient. The overall changes in staging accuracy, although not significant (p = 0.16), was due to the accurate staging of patients with stage II disease, who were understaged on unenhanced MRI. CE-MRI did not change staging in patients with stage III and VI disease. Evaluation of lymph node metastases One or more significant groin lymph nodes were found on MRI in 46 groins. Among these 46 groins, histopathology was positive for groin lymph node metastasis in 30 groins and negative in 16. Of the 15 groins without significant lymph nodes on MRI, 2 were positive for groin lymph node metastasis on histopathology. The area under the
85
ROC curve was largest for the S/L ratio (0.86; 95% CI: 0.74–0.98), followed by short axis and long axis (Fig. 1). An S/L ratio of ≥ 0.75 resulted in the highest accuracy of 85% for prediction of lymph node metastases. Table 3 summarizes the sensitivity, specificity, PPV, NPV, and accuracy of MR features in predicting groin lymph node metastasis. Reader's diagnostic confidence for presence of lymph node metastasis showed the highest sensitivity (93%) and accuracy (87%) for both “groins with significant lymph nodes” and “all groins.” An S/L ratio of ≥ 0.75 showed a sensitivity 87% and specificity of 81% for diagnosis of groin lymph node metastasis. Necrosis within a lymph node showed the highest specificity, but low sensitivity. Representative cases are shown in Figs. 2 and 3. Discussion Accurate assessment of the extent of vulvar disease and the detection of lymph node metastases are essential in planning the treatment of vulvar cancer. In contrast to other imaging modalities such as CT or ultrasound, MRI has the potential to demonstrate both the extent of the vulvar lesion and assess the groin lymph nodes simultaneously, thereby enabling the surgeons to tailor the extent of the procedure to the individual patient, hence reducing surgical morbidity. Despite this advantage, few studies of the MRI assessment of vulvar lesions and groin lymph nodes have been published. To our knowledge, this study is the largest to examine the staging accuracy of vulvar cancer using MRI, and the first study to examine the added value of CE-MRI. We have demonstrated that MRI is highly accurate in evaluating tumor size and presence of groin lymph node metastasis in patients with vulvar carcinoma. CE-MRI improved detection and staging of the vulvar tumor, although the improvement was not statistically significant. In our study, the vulvar lesion was visible in all patients with primary vulvar cancer, whereas in a previous study, only 10 out of 20 vulvar lesions were visible on MRI [8]. The higher tumor detection rate in our study may be explained by our use of fat suppression on T2-weighted images. The perineal region is rich in fat, and tumors showing intermediate-high signal on T2-weighted images are difficult to recognize without fat suppression. Although small vulvar cancers can typically be evaluated clinically in the outpatient department, assessment may be hindered by pain and body habitus. The traditional solution to this problem has been for the patient to undergo
Table 3 MR feature analysis in predicting groin lymph node metastasis. Features
Sensitivity % (n)
Specificity % (n)
PPV % (n)
NPV % (n)
Accuracy % (n)
Groins with significant LN S/L ratio ≥ 0.75 Contoura Necrosis Loss of fatty hilum Similarity of SI to vulva lesionb Readers' confidence of metastasis Readers' confidence of metastasis CEc
86.7 70.0 40.0 80.0 88.5 93.3 85.7
81.3 53.3 87.5 50.0 27.3 75.0 77.8
89.7 75.0 85.7 75.0 74.2 87.5 85.8
76.5 47.1 43.8 57.1 50.0 85.7 77.8
84.8 64.4 56.5 69.6 70.3 87.0 82.6
All groins S/L ratio ≥ 0.75 Contour Necrosis Loss of fatty hilum Similarity of SI to vulva lesion† Readers' confidence of metastasis Readers' confidence of metastasis CE††
81.3 (26/32) 65.6 (21/32) 37.5 (12/32) 75.0 (24/32) 82.1 (23/28) 87.5 (28/32) 80.0 (12/15)
(26/30) (21/30) (12/30) (24/30) (23/26) (28/30) (12/14)
(13/16) (8/15) (14/16) (8/16) (3/11) (12/16) (7/9)
89.7 (26/29) 75.0 (21/28) 93.1 (27/29) 72.4 (21/29) 60.0 (12/20) 86.2 (25/29) 88.2 (15/17)
(26/29) (21/28) (12/14) (24/30) (23/31) (28/32) (12/14)
81.3 (26/32) 75.0 (21/28) 85.7 (12/14) 75.0 (24/32) 74.2 (23/31) 87.5 (28/32) 85.7 (12/14)
(13/17) (8/17) (14/32) (8/14) (3/6) (12/14) (7/9)
89.7 (26/29) 65.6 (21/32) 57.4 (27/39) 72.4 (21/29) 70.6 (12/17) 86.2 (25/29) 83.3 (15/18)
(39/46) (29/45) (26/46) (32/46) (26/37) (40/46) (19/23)
85.2 (52/61) 70.0 (42/60) 63.9 (39/61) 73.8 (45/61) 72.9 (35/48) 86.9 (53/61) 84.4 (27/32)
In all groin analysis, 2 false positive and 13 true negative groins were added to the negative category, except for the following two features: 2 false positive and 9 true negative groins were added to the negative category (†) and 1 false positive and 8 true negative groins were added to the negative category (††). a One groin was excluded from the analysis due to motion artefact. b Nine groins were excluded because the main vulva lesion was difficult to be recognized. c Twenty-three groins from cases without CE were excluded from analysis.
86
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
Fig. 3. A woman (87 years old) with vulvar cancer. Anterior central vulvar lesion is difficult to see on axial fat-suppressed FRFSE image (a), but is enhanced and easier to recognize on axial T1-weighted fat-suppressed spin-echo image (b) (arrows).
Fig. 2. A woman (78 years old) with vulvar cancer. Axial fat-suppressed FRFSE image (a) shows a 7 × 3 cm lobulated mass in the vulva. Invasion to lower urethra and vagina was suspected. A left groin node was enlarged (15 mm) with S/L ratio of 0.9, showing signal intensity similar to that of vulvar lesion, with slightly irregular contour (arrow), highly suspicious for metastasis. (b) An image slightly cranial to (a) shows bilateral enlarged groin nodes (arrows). The round shape, signal intensity similar to that of the vulvar lesion, and small high-signal-intensity area suggestive of necrosis all suggest metastasis. Sagittal FRFSE image (c) demonstrates a large vulvar tumor invading to the lower urethra and possibly to vagina (arrows). On histopathology, lower urethral invasion was confirmed and tumor reached very close to vagina, with multiple bilateral groin node metastases.
examination under anaesthesia (EUA). However, as large tumours are frequently found in elderly patients with significant comorbidities, accurate preoperative staging of local disease without recourse to EUA would be an attractive option. To date, CE-MRI has not been considered useful in vulvar cancers, perhaps because of very limited data. The only published data is based on four patients, of whom only two showed tumor enhancement [8]. Our study includes a substantially larger number of patients, and suggests that CE-MRI may help readers to more confidently detect vulvar lesions. The use of fat suppression in CE-MRI increases the conspicuity of vulvar lesions and thus may have contributed to the improvement in diagnostic accuracy. The ability of preoperative imaging to accurately detect local extension to perineal structures would have significant prognostic, if not clinical impact, as reflected by recent proposals for the change in the FIGO staging classification [16]. A potential pitfall of CE-MRI includes overestimation of lesion size. Lymph node status in vulvar cancer is of crucial prognostic importance and will dictate both the extent of surgery in the lymphatic basin and the need for adjuvant (chemo) radiation. The most commonly-used MRI criterion for groin lymph node metastasis is the short-axis diameter of the node. However the reported sensitivity of this criterion is low, ranging from 40% to 50% [8, 12]. Other criteria, including S/L ratio, contour, and signal intensity, have
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
been examined [13, 14] for better sensitivity and accuracy. Our study demonstrated that S/L ratio was a good quantitative marker for diagnosing lymph node metastasis with sensitivity and specificity of 87% and 81%, respectively. Bipat et al. reported a sensitivity of 52% and specificity of 85–89% for detection of lymph node metastases. The contrasting results between the two studies may be explained by the differences in the study population (i.e., more advanced stage in our study), ratio of groins which were histologically examined (46 out of 98 groins in our study compared to 119 out of 126 in theirs), and the incidence of histologically positive lymph nodes (30 out of 46 in our study compared to 23 out of 119 in theirs) [12]. We examined other qualitative characteristics including those suggested by other studies [12-14], but no single criterion was as good as the overall diagnostic score, perhaps suggesting that experienced genitourinary radiologists make their diagnosis based on a combination of several MR features. The desire to reduce surgical morbidity for early stage disease has been the driving factor behind the evolution of sentinel node dissection (SLND) in vulvar cancer. Whilst this procedure appears to be safe and reproducible in early stage disease [17], there is a clear learning curve and several potential pitfalls that can influence the safe application of the technique [18]. The risk of false negative results secondary to lymphatic stasis has lead many authorities to recommend the use of cross-sectional imaging such as MRI in selecting cases for SLND. Identifying metastatic disease on preoperative MRI might reduce false negative results of SLND in obese patients. Furthermore, many SLND protocols rely on intraoperative frozen section assessment of the sentinel nodes. The sensitivity of this technique is typically between 80–90% [19]. Information from preoperative MRI, perhaps coupled with image guided fine needle aspiration, might avoid an unnecessary sentinel procedure and reduce the risk of requiring further surgery to the groins where lymph node metastasis is identified on routine pathological assessment after a false negative intraoperative frozen section. Our results would suggest that MRI may have a valuable role to play in case selection prior to definitive surgery. Our study was limited by its retrospective nature and by its reliance on a relatively small sample size. The number of patients with primary vulvar cancer who underwent CE-MRI was only 20, which may have lead to the statistically non-significant results regarding the added value of CE-MRI in staging vulvar cancer. Furthermore, proper sample size calculation was not possible due to variations in the methodology and the results of existing studies along with lack of published literature on the possible added role of CE-MRI in the evaluation of vulvar cancer. Another limitation was the approach to the analysis of lymph node metastasis; we performed per groin analysis rather than per lymph node analysis, and superficial and deep inguinal (femoral) lymph nodes were evaluated together. Per lymph node analysis is ideal in evaluating diagnostic accuracy, but correlating each lymph node on MRI with histopathological specimens is impossible in a retrospective study. In conclusion, in patients with vulvar cancer, MRI may be useful in evaluating the size of both primary and recurrent vulvar lesions and predicting the presence of groin lymph node metastasis. CEMRI may improve staging of primary vulvar tumors. New techniques may be necessary to further improve the diagnostic accuracy of MRI and enable accurate preoperative staging of vulvar cancer.
Acknowledgments We are grateful to the MRIS radiographers for assistance with patient scans. We also thank the NIHR: Cambridge Biomedical Research Centre and ACT for funding.
87
Appendix 1 International Federation of Gynecology and Obstetrics (FIGO) classification of vulvar cancer Definition Stage I
The cancer is confined to the vulva or the perineum. The tumor is 2 cm or less in diameter and has not spread to lymph nodes or distant sites. Stage II The cancer is confined to the vulva or perineum or both, and is larger than 2 cm. It has not spread to lymph nodes or distant sites Stage III The cancer extends into the anus, vagina, or lower urethra but there are no distant metastases. Or cancer is found in the vulva or perineum or both and has spread to nearby lymph nodes on one side of the groin but has not spread to distant sites Stage IV The cancer has spread to lymph nodes on both sides of the groin or it has spread beyond nearby tissues to the upper part of the urethra, bladder, rectum, or pelvic bone but has not spread to distant sites (stage IVA), or Cancer cancer has spread to distant organs or lymph nodes (including pelvic lymph nodes) (stage IVB)
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ygyno.2009.12.017. References [1] UK Vulva cancer incidence statistics. Cancer Research. UK: Cancer Stats; 2009. [2] Cancer Statistics,Vulva. Surveillance Epidemiology and End Results. National Cancer Institute; 2009. [3] Hampl M, Deckers-Figiel S, Hampl JA, Rein D, Bender HG. New aspects of vulvar cancer: changes in localization and age of onset. Gynecol Oncol 2008;109:340–5. [4] Burger MPM, Hollema H, Bouma J. The side of groin node metastases in unilateral vulvar carcinoma. Int J Gynecol Cancer 1996;6:318–22. [5] van Doorn HC, Ansink A, Verhaar-Langereis M, Stalpers L. Neoadjuvant chemoradiation for advanced primary vulvar cancer. Cochrane Database Syst Rev 2006;3:CD003752. [6] 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;38:309–14. [7] Chan JK, Sugiyama V, Pham H, Gu M, Rutgers J, Osann K, et al. Margin distance and other clinico-pathologic prognostic factors in vulvar carcinoma: a multivariate analysis. Gynecol Oncol 2007;104:636–41. [8] Sohaib SA, Richards PS, Ind T, Jeyarajah AR, Shepherd JH, Jacobs IJ, et al. MR imaging of carcinoma of the vulva. AJR Am J Roentgenol 2002;178:373–7. [9] Gonzalez Bosquet J, Magrina JF, Gaffey TA, Hernandez JL, Webb MJ, Cliby WA, et al. Long-term survival and disease recurrence in patients with primary squamous cell carcinoma of the vulva. Gynecol Oncol 2005;97:828–33. [10] Piura B, Rabinovich A, Cohen Y, Friger M, Glezerman M. Squamous cell carcinoma of the vulva in the south of Israel: a study of 50 cases. J Surg Oncol 1998;67: 174–81. [11] Hacker NF, Berek JS, Lagasse LD, Nieberg RK, Leuchter RS. Individualization of treatment for stage I squamous cell vulvar carcinoma. Obstet Gynecol 1984;63: 155–62. [12] Bipat S, Fransen GA, Spijkerboer AM, van der Velden J, Bossuyt PM, Zwinderman AH, et al. Is there a role for magnetic resonance imaging in the evaluation of inguinal lymph node metastases in patients with vulva carcinoma? Gynecol Oncol 2006;103:1001–6. [13] Hawnaur JM, Reynolds K, Wilson G, Hillier V, Kitchener HC. Identification of inguinal lymph node metastases from vulval carcinoma by magnetic resonance imaging: an initial report. Clin Radiol 2002;57:995–1000. [14] Singh K, Orakwue CO, Honest H, Balogun M, Lopez C, Luesley DM. Accuracy of magnetic resonance imaging of inguinofemoral lymph nodes in vulval cancer. Int J Gynecol Cancer 2006;16:1179–83. [15] Sala E, Akin O, Moskowitz CS, Eisenberg HF, Kuroiwa K, Ishill NM, et al. Endorectal MR imaging in the evaluation of seminal vesicle invasion: diagnostic accuracy and multivariate feature analysis. Radiology 2006;238:929–37. [16] Tantipalakorn C, Robertson G, Marsden DE, Gebski V, Hacker NF. Outcome and patterns of recurrence for International Federation of Gynecology and Obstetrics (FIGO) stages I and II squamous cell vulvar cancer. Obstet Gynecol 2009;113: 895–901. [17] Van der Zee AG, Oonk MH, De Hullu JA, Ansink AC, Vergote I, Verheijen RH, et al. Sentinel node dissection is safe in the treatment of early-stage vulvar cancer. J Clin Oncol 2008;26:884–9. [18] de Hullu JA, Oonk MH, Ansink AC, Hollema H, Jager PL, van der Zee AG. Pitfalls in the sentinel lymph node procedure in vulvar cancer. Gynecol Oncol 2004;94:10–5. [19] Brunner AH, Polterauer S, Tempfer C, Joura E, Reinthaller A, Horvat R, et al. The accuracy of intraoperative frozen section of the inguinal sentinel lymph node in vulvar cancer. Anticancer Res 2008;28:4091–4.
Contents lists available at ScienceDirect
Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o
Original Research Article
The accuracy of magnetic resonance imaging in staging of vulvar cancer: A retrospective multi-centre study Masako Yano Kataoka a,⁎, Evis Sala a, Peter Baldwin b, Caroline Reinhold c,d, Alex Farhadi c, Turtko Hudolin e, Hedvig Hricak e a
Department of Radiology, University of Cambridge, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK Macmillan Consultant in Gynaecological Oncology, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK Department of Diagnostic Radiology, McGill University Health Centre, Montreal General Hospital, 1650 Cedar Avenue, Montreal, Quebec, Canada H3G 1A4 d Medical Director Oncology, Synarc, Inc., San Francisco, CA, USA e Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA b c
a r t i c l e
i n f o
Article history: Received 15 September 2009 Available online 21 January 2010 Keywords: MRI Vulvar cancer Lymph node metastasis Contrast enhancement Gynecology Malignancy
a b s t r a c t Objective. To retrospectively evaluate the diagnostic accuracy and clinical relevance of magnetic resonance imaging (MRI) in the management of primary and recurrent vulvar cancer and to examine the added value of contrast-enhanced MRI (CE-MRI). Methods. The research ethics committee waived informed consent for this study of 49 patients with vulvar cancer (36 primary and 13 recurrent) who underwent MRI before surgery at three major cancer centers from December 2003 to January 2008. CE-MRI was available for 31 patients (20 primary and 11 recurrent). MR images were retrospectively evaluated by three radiologists for tumor size and stage. Lymph nodes with a short axis N5 mm were measured and scored for contour, presence of necrosis, loss of fatty hilum, signal intensity relative to the vulvar lesion, and reader's diagnosis of lymph node metastasis. Scoring was repeated for CE-MRI. Histopathology constituted the reference standard. Results. The size of the vulvar lesion was accurately characterized on MRI in 83% of patients. Accuracy in staging of primary vulvar cancers on unenhanced MRI was 69.4% (n = 36). Adding CE-MRI increased lesion detection and raised staging accuracy from 75% to 85% (n = 20). For groin lymph node metastasis prediction, the ratio of the short- to the long-axis diameter and the reader's diagnosis of lymph node metastasis yielded accuracy of 85% and 87%, respectively, in groin-by-groin analysis. Conclusion. MRI can be useful in accurately assessing the size of vulvar lesion and groin lymph node metastasis. CE-MRI may be of help in improving local staging. © 2009 Elsevier Inc. All rights reserved.
Introduction Vulvar cancer accounts for 3–5% of all gynecologic malignancies [1, 2]. The disease shows a bimodal age distribution, and its incidence in young women (b 50 years old) is increasing due to the spread of the human papilloma virus (HPV) [3]. Vulvar cancer may directly invade adjacent structures or spread to inguinal and femoral lymph nodes. Early stage vulvar cancer is treated by partial vulvectomy, with either ipsilateral or bilateral inguino-femoral lymph node dissection [4]. Surgery is the treatment of choice even for locally advanced cancer, although neoadjuvant radiation and chemotherapy may be used
⁎ Corresponding author. Fax: +44 1223 330915. E-mail addresses: [email protected] (M.Y. Kataoka), [email protected] (E. Sala), [email protected] (P. Baldwin), [email protected] (C. Reinhold), [email protected] (A. Farhadi), [email protected] (T. Hudolin), [email protected] (H. Hricak). 0090-8258/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2009.12.017
either to enable surgery in cases deemed ‘inoperable’ at initial presentation or to avoid exenteration in an elderly population [5]. The most important prognostic factors in vulvar cancer are tumor size, depth of invasion and the presence of lymph node metastases, which are incorporated into the International Federation of Gynecology and Obstetrics (FIGO) staging. This classification can predict overall survival. In addition, surgical margin distance is an important predictor of local recurrence. A tumor-free pathologic margin of b 8 mm is associated with local recurrence of 23–50% [6, 7]. As the margin distance is a potentially modifiable factor, accurate preoperative evaluation of the primary tumor is important. Only one study has evaluated the role of MRI in staging primary vulvar cancer in 20 cases [8]. Contrast-enhanced MRI (CE-MRI) was performed in only 4 of the 20 patients. The accurate detection and treatment of lymph node metastases in vulvar cancer is critical and relapse in the lymph node basin is associated with an extremely poor survival [9]. Clinical examination is of limited use in the detection of abnormal inguinal lymph nodes
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
(sensitivity, 57%; specificity, 62%) [10]. For tumours with a depth of invasion of N 1 mm, surgical management will typically include formal inguino-femoral node dissection on one or both sides [11]. Although outcomes are excellent in early stage disease, lymphadenectomy is associated with considerable morbidity including wound infection, breakdown, lymphocyst formation and long-term lymphoedema. Accurate preoperative information regarding the lymphatic basin could reduce the need for unnecessary lymphadenectomy with a resultant reduction in surgical morbidity. Four studies have used MRI to evaluate lymph node metastasis in vulvar cancer [8,12-14] They report widely varying sensitivity and specificity ranging from 40% to 89% and 82% to 100%, respectively [8,12]. The purpose of this study was to retrospectively evaluate (1) the diagnostic accuracy of MRI for vulvar cancer (both the main vulvar lesion and lymph node metastasis), and (2) to examine the potential added benefit of CE-MRI above ‘standard’ MRI imaging. Material and methods Study population The research ethics committee approved and issued a waiver of informed consent for our retrospective study. Through computerized review of the gynecologic surgery and radiology databases of three institutions, consecutive patients were identified who had histologically confirmed vulvar carcinoma and underwent MR imaging within 8 weeks (mean = 3.1 weeks) before surgery from December 2003 to January 2008. A total of 49 patients (38 from institution 1, 7 from institution 2, and 4 from institution 3) formed the study population. Table 1 shows patient characteristics. MRI protocol At all three institutions, examinations were performed using a 1. 5 T whole-body MR system (GE Healthcare Technologies, Milwaukee, WI). The protocol for vulvar cancer patients included axial T1weighted spin-echo images (TR/TE, 700/16 ms), sagittal T2-weighted fast recovery fast spin-echo (FRFSE) images (4000–5000/85–90),
Table 1 Characteristics of patients with vulvar cancer (n = 49). Mean age (range)
70.4 (40–88)
Primary cancer Recurrent cancer MR examination Total Unenhanced MRI only Unenhanced + CE-MRI Primary cancer Recurrent cancer Mean interval between MRI and surgery (weeks) Surgery Wide local excision only + unilateral groin LN dissection + bilateral groin LN dissection + sentinel node dissection Radical vulvectomy only + unilateral groin LN dissection + bilateral groin LN dissection Partial/hemivulvectomy only + unilateral groin LN dissection + bilateral groin LN dissection Overall FIGO stage for primary cancer (n = 36) Stage I Stage II Stage III Stage IV
36 13
LN, lymph node.
49 18 31 20 11 3.1 8 8 9 8 2 1 8 0 1 4 7 9 11 8
83
axial and coronal T2-weighted fat-suppressed FRFSE images (4000– 5000/85–90). In 31 cases, an intravenous bolus injection of gadopentate dimeglumine (0. 1 mmol per kilogram body weight) was given at 3 mL. Axial T1-weighted fat-suppressed spin-echo images (600/16.5) were acquired before and after the injection. All sequences were obtained with a slice thickness of 3–5 mm. Image evaluation Images were displayed using GE PACS system. Two genitourinary MR radiologists (_ and _) with 6 and 8 years of experience, respectively, evaluated the MRI examinations. Both radiologists were aware of the diagnosis of vulvar cancer but were blinded to the final surgical and histopathological results when scoring the MRI examinations. Evaluation of vulvar lesion A five-point scoring system was used for the detection of vulvar lesions: 1, tumor definitely absent; 2, tumor probably absent; 3, tumor possibly present; 4, tumor probably present; 5, tumor definitely present. Vulvar lesion size was measured in antero–posterior, left– right, and cranio–caudal direction using two orthogonal views. The maximum length of the three measurements was recorded as the size of the lesion. MRI stage was classified according to the FIGO classification system for 36 patients with primary vulvar cancer (Appendix 1). In addition, for 31 patients who underwent CE-MRI, the above evaluations were repeated. CE-MRI was evaluated after the evaluation of their non CE-MRI images. In 20 of the 31 patients who underwent both unenhanced and CE-MRI, the incremental value of CE-MRI in MRI stage according to FIGO classification was assessed. Evaluation of groin lymph node metastasis Groin lymph nodes measuring more than 5 mm in the short-axis diameter on MRI were considered as potentially metastatic lymph nodes, based on reported specificity b 70% with short-axis diameter of ≤ 5 mm in the literature [8]. Both the long- and short-axis diameters were measured and scores were assigned for contour (1, smooth; 2, lobular; 3, irregular; and 4, spiculated), presence of cystic changes/ necrosis, loss of fatty hilum, and similarity of signal intensity to the primary lesion (5-point scoring system reflecting five levels of diagnostic confidence). In addition to individual findings described above, a five-point scoring system was also used to evaluate reader's diagnostic confidence for presence of metastasis (1, metastasis definitely absent; 2, metastasis probably absent; 3, metastasis possibly present; 4, metastasis probably present; 5, metastasis definitely present) [15]. Groin lymph node metastasis was assessed using “groin-by-groin” analysis. Since superficial and deep inguinal lymph nodes were combined on histopathology assessment, these two regions were jointly evaluated on imaging. The lymph node with the highest score (most malignant) per groin provided the overall score for that groin. Statistical analysis The accuracy levels of both unenhanced and CE-MRI for detection and staging of vulvar cancer were evaluated. Due to non-normal distribution, the Spearman correlation coefficient was used to evaluate the correlation between the size of the lesion on MRI and histopathology in both primary and recurrent disease. Wilcoxon's rank sum test was used to compare reader's confidence for detection of vulvar lesions using unenhanced MRI with that using CE-MRI. The hypothesis of no disproportionate change between stages (i.e., correct or incorrect stage) determined with unenhanced MRI alone and stage determined with unenhanced and CE-MRI was tested using McNemar's test.
84
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
Specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy were calculated for the evaluation of groin lymph node metastasis. Receiver operating curve (ROC) analyses were performed for long axis, short axis and short axis/long axis (S/L) ratio. The cut-off point for contour was determined as between 1 and 2 (absence of metastasis) and 3 and 4 (presence of metastasis). For all the features evaluated with the 5-point scoring system, the cut-off point was determined as between 1 to 3 (absence of metastasis) and 4 and 5 (presence of metastasis). Analysis of the MR features of groin lymph nodes was performed once using only groins with significant lymph nodes (i.e., nodes with a short axis N 5 mm) and again using all groins. A P value of less than 0.05 was considered statistically significant. All analysis was performed using STATA (version 8.0; STATA, College Station, TX). Results Surgery and histopathology Surgical and histopathologic findings were compared with MRI findings (Table 1). Invasion to lower urethra, vagina and anus was histopathologically confirmed in three, two and one patients, respectively. No patient had invasion to upper urethra, bladder, rectum, or pelvic bones. Histopathological information about groin lymph node metastasis, obtained either by lymph node dissection or sentinel node biopsy, was available in 61 groin nodes. The remaining 37 groin lymph nodes were not dissected or biopsied and therefore excluded from analysis. Groin nodes were positive for presence of metastases in 32 groins and negative in 29 groins. Evaluation of vulvar lesion: unenhanced MRI The vulvar lesion was visible on MRI in all 36 patients with primary vulvar cancer and 11 of 13 patients with recurrent tumor. The mean size of the vulvar lesion on MRI was 34. 7 mm (primary 35. 2 mm, recurrent 33. 1 mm). Twelve patients (primary 8, recurrent 4) had vulva lesion ≤ 2 cm and 35 patients (primary 28, recurrent 7) had vulva lesion N 2 cm. The mean histopathological size was 36. 2 mm (primary 36. 8 mm, recurrent 34. 2 mm). Size on MRI and size on histopathology showed moderate correlation (Spearman correlation coefficient: 0.72, primary 0.72, recurrent 0.57). MRI accurately classified 31 out of 36 primary vulvar cancers (accuracy 86%) and 8 out of 11 recurrent vulvar cancers (accuracy 72%) into the correct size category (overall accuracy of 83%).
Table 2 Diagnostic accuracy of staging using unenhanced MRI and CE-MRI: correlation with histopathology (n = 20). Histopathology
MRI Stage I Stage II Stage III Stage IV Total Overall accuracy CE-MRI Stage I Stage II Stage III Stage IV Total Overall accuracy
Stage I
Stage II
Stage III
Stage IV
Total
3 2 0 0 5 75.0%
2 3 0 0 5
0 1 4 0 5
0 0 0 5 5
5 6 4 5 20
3 2 0 0 5 85.0%
0 5 0 0 5
0 1 4 0 5
0 0 0 5 5
3 8 4 5 20
Fig. 1. ROC curves of long axis (mm) (a), short axis (b), and S/L ratio (c) in diagnosing groin lymph node metastasis, with areas under the ROC curves of 0.67 (95% CI: 0.50– 0.84), 0.69 (95% CI: 0.52–0.86), and 0.86 (95% CI: 0.74–0.98) respectively.
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
The overall staging accuracy of primary vulvar cancers on unenhanced MRI was 69.4% (n = 36). For 36 patients with primary vulvar cancer, staging by MRI was as follows: stage I: 6, stage II: 9, stage III: 13, stage VI: 8. Six patients were overstaged, and five patients were understaged by MRI. Evaluation of vulvar lesion: the added value of CE-MRI Among 47 patients whose main vulvar lesion was visible on MRI, 30 patients (primary 20, recurrent 10) had CE-MRI. The mean size of the vulvar lesion on CE-MRI was 34. 1 mm (primary 34. 0 mm, recurrent 34. 2 mm). Five patients were reclassified after CE-MRI; three patients (primary 1, recurrent 2) were correctly reclassified as having a lesion “size N 2 cm,” and in the other two patients (primary 1, recurrent 1) CE-MRI overestimated the size of the vulvar lesion. CEMRI accurately classified lesion size in 25 out of 30 (85.1%) patients. The addition of CE-MRI did not change the accuracy of the size category of primary vulvar cancers (accuracy 85.0%); however, the accuracy of the size category for recurrent vulvar cancers changed from 70% to 80%. In 10 patients (primary 6, recurrent 4), reader confidence in the detection of the vulvar tumor was higher on CE-MRI (mean score 4.4, primary 4.5, recurrent 4.8) than on unenhanced MRI (mean score, 3.9; p = 0.004, primary 2.8, recurrent 3.5). In one of these patients with recurrent cancer, no residual invasive cancer was found in the surgical specimen and therefore this was considered as false positive. Table 2 shows the diagnostic accuracy of CE-MRI in 20 patients with primary vulvar cancer. CE-MRI accurately staged 17 patients, for an overall staging accuracy of 85%; it overstaged 2 patients and understaged 1 patient. The overall changes in staging accuracy, although not significant (p = 0.16), was due to the accurate staging of patients with stage II disease, who were understaged on unenhanced MRI. CE-MRI did not change staging in patients with stage III and VI disease. Evaluation of lymph node metastases One or more significant groin lymph nodes were found on MRI in 46 groins. Among these 46 groins, histopathology was positive for groin lymph node metastasis in 30 groins and negative in 16. Of the 15 groins without significant lymph nodes on MRI, 2 were positive for groin lymph node metastasis on histopathology. The area under the
85
ROC curve was largest for the S/L ratio (0.86; 95% CI: 0.74–0.98), followed by short axis and long axis (Fig. 1). An S/L ratio of ≥ 0.75 resulted in the highest accuracy of 85% for prediction of lymph node metastases. Table 3 summarizes the sensitivity, specificity, PPV, NPV, and accuracy of MR features in predicting groin lymph node metastasis. Reader's diagnostic confidence for presence of lymph node metastasis showed the highest sensitivity (93%) and accuracy (87%) for both “groins with significant lymph nodes” and “all groins.” An S/L ratio of ≥ 0.75 showed a sensitivity 87% and specificity of 81% for diagnosis of groin lymph node metastasis. Necrosis within a lymph node showed the highest specificity, but low sensitivity. Representative cases are shown in Figs. 2 and 3. Discussion Accurate assessment of the extent of vulvar disease and the detection of lymph node metastases are essential in planning the treatment of vulvar cancer. In contrast to other imaging modalities such as CT or ultrasound, MRI has the potential to demonstrate both the extent of the vulvar lesion and assess the groin lymph nodes simultaneously, thereby enabling the surgeons to tailor the extent of the procedure to the individual patient, hence reducing surgical morbidity. Despite this advantage, few studies of the MRI assessment of vulvar lesions and groin lymph nodes have been published. To our knowledge, this study is the largest to examine the staging accuracy of vulvar cancer using MRI, and the first study to examine the added value of CE-MRI. We have demonstrated that MRI is highly accurate in evaluating tumor size and presence of groin lymph node metastasis in patients with vulvar carcinoma. CE-MRI improved detection and staging of the vulvar tumor, although the improvement was not statistically significant. In our study, the vulvar lesion was visible in all patients with primary vulvar cancer, whereas in a previous study, only 10 out of 20 vulvar lesions were visible on MRI [8]. The higher tumor detection rate in our study may be explained by our use of fat suppression on T2-weighted images. The perineal region is rich in fat, and tumors showing intermediate-high signal on T2-weighted images are difficult to recognize without fat suppression. Although small vulvar cancers can typically be evaluated clinically in the outpatient department, assessment may be hindered by pain and body habitus. The traditional solution to this problem has been for the patient to undergo
Table 3 MR feature analysis in predicting groin lymph node metastasis. Features
Sensitivity % (n)
Specificity % (n)
PPV % (n)
NPV % (n)
Accuracy % (n)
Groins with significant LN S/L ratio ≥ 0.75 Contoura Necrosis Loss of fatty hilum Similarity of SI to vulva lesionb Readers' confidence of metastasis Readers' confidence of metastasis CEc
86.7 70.0 40.0 80.0 88.5 93.3 85.7
81.3 53.3 87.5 50.0 27.3 75.0 77.8
89.7 75.0 85.7 75.0 74.2 87.5 85.8
76.5 47.1 43.8 57.1 50.0 85.7 77.8
84.8 64.4 56.5 69.6 70.3 87.0 82.6
All groins S/L ratio ≥ 0.75 Contour Necrosis Loss of fatty hilum Similarity of SI to vulva lesion† Readers' confidence of metastasis Readers' confidence of metastasis CE††
81.3 (26/32) 65.6 (21/32) 37.5 (12/32) 75.0 (24/32) 82.1 (23/28) 87.5 (28/32) 80.0 (12/15)
(26/30) (21/30) (12/30) (24/30) (23/26) (28/30) (12/14)
(13/16) (8/15) (14/16) (8/16) (3/11) (12/16) (7/9)
89.7 (26/29) 75.0 (21/28) 93.1 (27/29) 72.4 (21/29) 60.0 (12/20) 86.2 (25/29) 88.2 (15/17)
(26/29) (21/28) (12/14) (24/30) (23/31) (28/32) (12/14)
81.3 (26/32) 75.0 (21/28) 85.7 (12/14) 75.0 (24/32) 74.2 (23/31) 87.5 (28/32) 85.7 (12/14)
(13/17) (8/17) (14/32) (8/14) (3/6) (12/14) (7/9)
89.7 (26/29) 65.6 (21/32) 57.4 (27/39) 72.4 (21/29) 70.6 (12/17) 86.2 (25/29) 83.3 (15/18)
(39/46) (29/45) (26/46) (32/46) (26/37) (40/46) (19/23)
85.2 (52/61) 70.0 (42/60) 63.9 (39/61) 73.8 (45/61) 72.9 (35/48) 86.9 (53/61) 84.4 (27/32)
In all groin analysis, 2 false positive and 13 true negative groins were added to the negative category, except for the following two features: 2 false positive and 9 true negative groins were added to the negative category (†) and 1 false positive and 8 true negative groins were added to the negative category (††). a One groin was excluded from the analysis due to motion artefact. b Nine groins were excluded because the main vulva lesion was difficult to be recognized. c Twenty-three groins from cases without CE were excluded from analysis.
86
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
Fig. 3. A woman (87 years old) with vulvar cancer. Anterior central vulvar lesion is difficult to see on axial fat-suppressed FRFSE image (a), but is enhanced and easier to recognize on axial T1-weighted fat-suppressed spin-echo image (b) (arrows).
Fig. 2. A woman (78 years old) with vulvar cancer. Axial fat-suppressed FRFSE image (a) shows a 7 × 3 cm lobulated mass in the vulva. Invasion to lower urethra and vagina was suspected. A left groin node was enlarged (15 mm) with S/L ratio of 0.9, showing signal intensity similar to that of vulvar lesion, with slightly irregular contour (arrow), highly suspicious for metastasis. (b) An image slightly cranial to (a) shows bilateral enlarged groin nodes (arrows). The round shape, signal intensity similar to that of the vulvar lesion, and small high-signal-intensity area suggestive of necrosis all suggest metastasis. Sagittal FRFSE image (c) demonstrates a large vulvar tumor invading to the lower urethra and possibly to vagina (arrows). On histopathology, lower urethral invasion was confirmed and tumor reached very close to vagina, with multiple bilateral groin node metastases.
examination under anaesthesia (EUA). However, as large tumours are frequently found in elderly patients with significant comorbidities, accurate preoperative staging of local disease without recourse to EUA would be an attractive option. To date, CE-MRI has not been considered useful in vulvar cancers, perhaps because of very limited data. The only published data is based on four patients, of whom only two showed tumor enhancement [8]. Our study includes a substantially larger number of patients, and suggests that CE-MRI may help readers to more confidently detect vulvar lesions. The use of fat suppression in CE-MRI increases the conspicuity of vulvar lesions and thus may have contributed to the improvement in diagnostic accuracy. The ability of preoperative imaging to accurately detect local extension to perineal structures would have significant prognostic, if not clinical impact, as reflected by recent proposals for the change in the FIGO staging classification [16]. A potential pitfall of CE-MRI includes overestimation of lesion size. Lymph node status in vulvar cancer is of crucial prognostic importance and will dictate both the extent of surgery in the lymphatic basin and the need for adjuvant (chemo) radiation. The most commonly-used MRI criterion for groin lymph node metastasis is the short-axis diameter of the node. However the reported sensitivity of this criterion is low, ranging from 40% to 50% [8, 12]. Other criteria, including S/L ratio, contour, and signal intensity, have
M.Y. Kataoka et al. / Gynecologic Oncology 117 (2010) 82–87
been examined [13, 14] for better sensitivity and accuracy. Our study demonstrated that S/L ratio was a good quantitative marker for diagnosing lymph node metastasis with sensitivity and specificity of 87% and 81%, respectively. Bipat et al. reported a sensitivity of 52% and specificity of 85–89% for detection of lymph node metastases. The contrasting results between the two studies may be explained by the differences in the study population (i.e., more advanced stage in our study), ratio of groins which were histologically examined (46 out of 98 groins in our study compared to 119 out of 126 in theirs), and the incidence of histologically positive lymph nodes (30 out of 46 in our study compared to 23 out of 119 in theirs) [12]. We examined other qualitative characteristics including those suggested by other studies [12-14], but no single criterion was as good as the overall diagnostic score, perhaps suggesting that experienced genitourinary radiologists make their diagnosis based on a combination of several MR features. The desire to reduce surgical morbidity for early stage disease has been the driving factor behind the evolution of sentinel node dissection (SLND) in vulvar cancer. Whilst this procedure appears to be safe and reproducible in early stage disease [17], there is a clear learning curve and several potential pitfalls that can influence the safe application of the technique [18]. The risk of false negative results secondary to lymphatic stasis has lead many authorities to recommend the use of cross-sectional imaging such as MRI in selecting cases for SLND. Identifying metastatic disease on preoperative MRI might reduce false negative results of SLND in obese patients. Furthermore, many SLND protocols rely on intraoperative frozen section assessment of the sentinel nodes. The sensitivity of this technique is typically between 80–90% [19]. Information from preoperative MRI, perhaps coupled with image guided fine needle aspiration, might avoid an unnecessary sentinel procedure and reduce the risk of requiring further surgery to the groins where lymph node metastasis is identified on routine pathological assessment after a false negative intraoperative frozen section. Our results would suggest that MRI may have a valuable role to play in case selection prior to definitive surgery. Our study was limited by its retrospective nature and by its reliance on a relatively small sample size. The number of patients with primary vulvar cancer who underwent CE-MRI was only 20, which may have lead to the statistically non-significant results regarding the added value of CE-MRI in staging vulvar cancer. Furthermore, proper sample size calculation was not possible due to variations in the methodology and the results of existing studies along with lack of published literature on the possible added role of CE-MRI in the evaluation of vulvar cancer. Another limitation was the approach to the analysis of lymph node metastasis; we performed per groin analysis rather than per lymph node analysis, and superficial and deep inguinal (femoral) lymph nodes were evaluated together. Per lymph node analysis is ideal in evaluating diagnostic accuracy, but correlating each lymph node on MRI with histopathological specimens is impossible in a retrospective study. In conclusion, in patients with vulvar cancer, MRI may be useful in evaluating the size of both primary and recurrent vulvar lesions and predicting the presence of groin lymph node metastasis. CEMRI may improve staging of primary vulvar tumors. New techniques may be necessary to further improve the diagnostic accuracy of MRI and enable accurate preoperative staging of vulvar cancer.
Acknowledgments We are grateful to the MRIS radiographers for assistance with patient scans. We also thank the NIHR: Cambridge Biomedical Research Centre and ACT for funding.
87
Appendix 1 International Federation of Gynecology and Obstetrics (FIGO) classification of vulvar cancer Definition Stage I
The cancer is confined to the vulva or the perineum. The tumor is 2 cm or less in diameter and has not spread to lymph nodes or distant sites. Stage II The cancer is confined to the vulva or perineum or both, and is larger than 2 cm. It has not spread to lymph nodes or distant sites Stage III The cancer extends into the anus, vagina, or lower urethra but there are no distant metastases. Or cancer is found in the vulva or perineum or both and has spread to nearby lymph nodes on one side of the groin but has not spread to distant sites Stage IV The cancer has spread to lymph nodes on both sides of the groin or it has spread beyond nearby tissues to the upper part of the urethra, bladder, rectum, or pelvic bone but has not spread to distant sites (stage IVA), or Cancer cancer has spread to distant organs or lymph nodes (including pelvic lymph nodes) (stage IVB)
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ygyno.2009.12.017. References [1] UK Vulva cancer incidence statistics. Cancer Research. UK: Cancer Stats; 2009. [2] Cancer Statistics,Vulva. Surveillance Epidemiology and End Results. National Cancer Institute; 2009. [3] Hampl M, Deckers-Figiel S, Hampl JA, Rein D, Bender HG. New aspects of vulvar cancer: changes in localization and age of onset. Gynecol Oncol 2008;109:340–5. [4] Burger MPM, Hollema H, Bouma J. The side of groin node metastases in unilateral vulvar carcinoma. Int J Gynecol Cancer 1996;6:318–22. [5] van Doorn HC, Ansink A, Verhaar-Langereis M, Stalpers L. Neoadjuvant chemoradiation for advanced primary vulvar cancer. Cochrane Database Syst Rev 2006;3:CD003752. [6] 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;38:309–14. [7] Chan JK, Sugiyama V, Pham H, Gu M, Rutgers J, Osann K, et al. Margin distance and other clinico-pathologic prognostic factors in vulvar carcinoma: a multivariate analysis. Gynecol Oncol 2007;104:636–41. [8] Sohaib SA, Richards PS, Ind T, Jeyarajah AR, Shepherd JH, Jacobs IJ, et al. MR imaging of carcinoma of the vulva. AJR Am J Roentgenol 2002;178:373–7. [9] Gonzalez Bosquet J, Magrina JF, Gaffey TA, Hernandez JL, Webb MJ, Cliby WA, et al. Long-term survival and disease recurrence in patients with primary squamous cell carcinoma of the vulva. Gynecol Oncol 2005;97:828–33. [10] Piura B, Rabinovich A, Cohen Y, Friger M, Glezerman M. Squamous cell carcinoma of the vulva in the south of Israel: a study of 50 cases. J Surg Oncol 1998;67: 174–81. [11] Hacker NF, Berek JS, Lagasse LD, Nieberg RK, Leuchter RS. Individualization of treatment for stage I squamous cell vulvar carcinoma. Obstet Gynecol 1984;63: 155–62. [12] Bipat S, Fransen GA, Spijkerboer AM, van der Velden J, Bossuyt PM, Zwinderman AH, et al. Is there a role for magnetic resonance imaging in the evaluation of inguinal lymph node metastases in patients with vulva carcinoma? Gynecol Oncol 2006;103:1001–6. [13] Hawnaur JM, Reynolds K, Wilson G, Hillier V, Kitchener HC. Identification of inguinal lymph node metastases from vulval carcinoma by magnetic resonance imaging: an initial report. Clin Radiol 2002;57:995–1000. [14] Singh K, Orakwue CO, Honest H, Balogun M, Lopez C, Luesley DM. Accuracy of magnetic resonance imaging of inguinofemoral lymph nodes in vulval cancer. Int J Gynecol Cancer 2006;16:1179–83. [15] Sala E, Akin O, Moskowitz CS, Eisenberg HF, Kuroiwa K, Ishill NM, et al. Endorectal MR imaging in the evaluation of seminal vesicle invasion: diagnostic accuracy and multivariate feature analysis. Radiology 2006;238:929–37. [16] Tantipalakorn C, Robertson G, Marsden DE, Gebski V, Hacker NF. Outcome and patterns of recurrence for International Federation of Gynecology and Obstetrics (FIGO) stages I and II squamous cell vulvar cancer. Obstet Gynecol 2009;113: 895–901. [17] Van der Zee AG, Oonk MH, De Hullu JA, Ansink AC, Vergote I, Verheijen RH, et al. Sentinel node dissection is safe in the treatment of early-stage vulvar cancer. J Clin Oncol 2008;26:884–9. [18] de Hullu JA, Oonk MH, Ansink AC, Hollema H, Jager PL, van der Zee AG. Pitfalls in the sentinel lymph node procedure in vulvar cancer. Gynecol Oncol 2004;94:10–5. [19] Brunner AH, Polterauer S, Tempfer C, Joura E, Reinthaller A, Horvat R, et al. The accuracy of intraoperative frozen section of the inguinal sentinel lymph node in vulvar cancer. Anticancer Res 2008;28:4091–4.