Diagnostic value of intraoperative ultrasonography to assess para-aortic lymph nodes in women with ovarian and uterine corpus malignancy

Authors


Abstract

Objectives

To examine the ability of intraoperative ultrasonography to detect enlarged para-aortic lymph nodes, and to assess its potential use in reducing the number of unnecessary para-aortic lymphadenectomies performed in women with ovarian and uterine corpus malignancies.

Methods

Computed tomography (CT), palpation during surgery, and intraoperative ultrasonography were used to assess whether para-aortic lymph nodes were enlarged in 163 women with ovarian and uterine corpus malignancy. All the women underwent para-aortic lymphadenectomy, and nodes were assessed for metastasis.

Results

Thirty-five women had pathological para-aortic node metastasis. The sensitivity, specificity, and positive and negative predictive values of CT for the diagnosis of metastasis were 42.9, 96.1, 75.0 and 86.0%, respectively. These values were 60.0, 82.0, 47.7 and 88.2% for palpation, and 91.4, 69.5, 45.1 and 96.7% for intraoperative ultrasonography, respectively. If para-aortic lymphadenectomy had been performed only when enlarged lymph nodes were detected on CT then the number performed would have been reduced from 163 to 20 (12.3%); however, node metastasis would have been missed in 20 out of 35 women. On the same basis, the number of lymphadenectomies performed would have been 44 (27.0%) and metastasis would have been missed in 14 women on palpation during surgery, and 71 lymphadenectomies (43.6%) would have been performed and metastasis would have been missed in three women on intraoperative ultrasonography.

Conclusions

Intraoperative ultrasonography is a highly sensitive tool with which to diagnose lymph node metastasis. Its high negative predictive value allows avoidance of unnecessary para-aortic lymphadenectomy in women with ovarian and uterine corpus malignancy. Copyright © 2008 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Para-aortic lymph nodes are regional lymph nodes of ovarian and endometrial carcinomas, and determining whether or not they are metastasized is very important in assessing spread of the cancer. Up to now, there has been no reliable procedure other than systematic para-aortic lymphadenectomy for assessment; however, routine lymphadenectomy increases surgical morbidity and its therapeutic significance is uncertain.

In a preliminary report1 it was suggested that intraoperative ultrasonography has a superior negative predictive value (NPV) and sensitivity to preoperative computed tomography (CT) and palpation during surgery in detecting para-aortic node metastasis in gynecological malignancies. The purpose of this study was to further assess the ability of intraoperative ultrasonography to detect enlarged para-aortic lymph nodes, and to assess its potential use in reducing the number of unnecessary para-aortic lymphadenectomies performed in women with ovarian and uterine corpus malignancies, again comparing its effectiveness with those of preoperative CT and palpation during surgery.

Methods

According to the protocols of the hospitals in which this study was carried out, pelvic and para-aortic lymphadenectomy are performed when a woman is less than 75 years old and has a good performance status and no distant metastasis. Under these conditions, all women with ovarian carcinoma undergo pelvic and para-aortic lymphadenectomy up to the renal vessels when abdominal residual tumors are less than 1 cm in diameter. Similarly, all women with endometrial carcinoma undergo pelvic and para-aortic lymphadenectomy up to the renal vessels. One exception is that para-aortic lymphadenectomy is not performed in cases of well-differentiated endometrial carcinoma in which no muscular invasion is demonstrated by either preoperative magnetic resonance imaging (MRI) or intraoperative macroscopic findings of the removed uterus. Women with uterine sarcoma also undergo pelvic and para-aortic lymphadenectomy when the diagnosis of sarcoma is confirmed before surgery.

According to these protocols 163 women were recruited into this study—29 consecutive women at Musashino Red Cross Hospital in 1998–2000, and 134 consecutive women at Teikyo University Hospital in 2002–2006—all of whom underwent para-aortic lymphadenectomy. The mean age (SD) of the women was 55.6 ( ± 9.74) years. There were 73 cases of ovarian carcinoma, 78 cases of endometrial carcinoma, seven cases of uterine sarcoma (six of carcinosarcoma and one of leiomyosarcoma) and five cases of endometrial and ovarian synchronous carcinomas. Ten women with ovarian carcinoma underwent para-aortic lymphadenectomy in interval-debulking surgery after chemotherapy. No anticancer therapy was performed before surgery in the remaining 153 women. Table 1 shows the incidence of each type of gynecological malignancy in the study population.

Table 1. Characteristics of the study population
DiseaseNumber of patients
  1. FIGO, International Federation of Gynecology and Obstetrics.

Ovarian carcinoma73
 FIGO stage 
I33
II8
III30
IV2
 Histological study 
Serous26
Mucinous8
Endometrioid8
Clear21
Other10
Endometrial carcinoma78
 FIGO stage 
I44
II10
III23
IV1
 Histological study 
G137
G218
G313
Other10
Synchronous carcinoma5
Uterine sarcoma7
Total163

All the women underwent the standard course of treatment usually performed in the participating hospitals, with the only deviation from normal procedure being the addition of intraoperative ultrasonography. All participants gave informed consent, and the study was approved by the institutional review board.

Within 2 weeks before the surgery, CT was performed using a helical scanner (Hispeed NX, Lightspeed Plus, GE and Yokogawa Medical Systems, Tokyo, Japan) in 1999–2000 and a multidetector-row eight-channel helical CT unit (Light Speed QX/I, GE Medical Systems, Milwaukee, WI, USA) in 2002–2006. Transverse images 5 mm in thickness were obtained at 5-mm intervals at the level between the aortic bifurcation and renal vessels. All the women received 2 mL/kg nonionic intravenously administered contrast material at a rate of 2–2.5 mL/s in 1999–2000 (Optiray 320, Yamonouchi, Tokyo, Japan) and 1.5 mL/s in 2002–2006 (Omnipaque, Daiichi, Tokyo, Japan). Scanning was performed 90–120 s after the initiation of intravenous contrast administration. A structure adjacent to the aorta or vena cava was identified as an enlarged node if it had a well-defined round shape with soft-tissue attenuation and its diameter was more than 5 mm (a previous CT imaging study2 had found that normal nodes at the upper and lower para-aortic regions had mean diameters ( ± SD) of 3.7 ( ± 0.2) and 3.4 ( ± 0.1) mm, respectively). These assessments were performed by at least two radiologists, each with more than 10 years' experience.

The procedures during surgery were as follows: the woman was placed in the lithotomy position under general anesthesia and the lower abdomen was opened via a midline incision. After laparotomy the surgeon, who was a gynecological specialist with more than 10 years' experience and blinded to the results of the CT scans, palpated the para-aortic region and assessed whether there were enlarged para-aortic nodes of more than 5 mm in diameter. The para-aortic region was defined as a rectangular area from the level of the aortic bifurcation to that of the left renal vein and laterally from 6 cm to the right to 3 cm to the left of the aorta. The surgeon could recognize the aorta but not the left renal vein by palpation; therefore its level was set at about 5 cm up from the lower pole of the left kidney.

Next, an examiner, also blinded to the results of the CT and palpation, inserted an intraoperative ultrasound probe into the lower abdominal incision towards the upper abdomen. By making contact with the retroperitoneum, the examiner scanned the para-aortic region, and assessed enlarged nodes. First, the para-aortic region was delineated by identifying the aortic bifurcation, the renal arteries and veins, and the bilateral ovarian veins that could be visualized on intraoperative ultrasonography. Then, the left lateroaortic area was scanned by moving the ultrasound probe from the aortic bifurcation to the renal vessels in the transverse plane. In the same way, the preaortic and retroaortic areas, intra-aortocaval area, precaval and retrocaval areas, and right laterocaval area were scanned. If a hypoechoic, well-defined and apparently round structure was found, its shape was confirmed by further observations from more than two different directions to exclude vessels. A hypoechoic well-defined round structure adjacent to the aorta or the vena cava was identified as a lymph node, and it was assessed as enlarged if its diameter in the transverse plane was more than 5 mm. All ultrasound scans were performed by one examiner, who was an ultrasound specialist, using an SSD-2000 machine with a UST-995-7.5-MHz probe (Aloka, Tokyo, Japan) in 1999–2000 and a NEMIO SSA 550A machine with a PVF-738H probe (Toshiba Medical System, Tokyo, Japan) in 2002–2006.

Following the lower abdominal procedures including pelvic lymphadenectomy, the abdominal incision was extended to the subxyphoidal region. We mobilized the ascending colon, descending colon, and duodenum, and displaced them to the right, to the left and upwards, respectively, so that the para-aortic retroperitoneal space was opened up to the renal vessels. Lymphatic tissue surrounding the retroperitoneal vessels was removed completely and was microscopically examined for node metastasis by pathologists in the hospitals.

It would have been difficult to correlate the results obtained for each individual node as evaluated by CT, palpation, intraoperative ultrasonography and pathological examination. Therefore, just the relationships between the detection of enlarged nodes by each of the three assessment methods and the presence of para-aortic node metastasis, as assessed by pathological examination, in each woman were examined. The sensitivity, specificity, and positive and negative predictive values (PPV and NPV) for each assessment method were calculated and compared using the Chi-square test. The potential clinical usefulness of the three assessments in decreasing the number of para-aortic lymphadenectomies performed, while maintaining high sensitivity for node metastasis, was then evaluated.

Results

Scanning by intraoperative ultrasonography took less than 5 min and had no adverse effect on any woman. The mean ( ± SD) number of para-aortic nodes removed per woman was 25.6 ( ± 12.90). Para-aortic lymph node metastasis was found microscopically in 18 women with ovarian carcinoma, 15 women with endometrial carcinoma, and two women with uterine sarcoma. The number of positive nodes per woman showing node metastasis was 4.6 ( ± 7.0). Figure 1 shows examples of enlarged nodes detected by intraoperative ultrasonography in two women with para-aortic node metastasis.

Figure 1.

Intraoperative ultrasound images (transverse sections) showing lymph node 9 mm in diameter located in the intra-aortocaval area (a), and lymph node 22 mm in diameter located in the left lateroaortic area (b). Computed tomography and palpation could detect the lymph node shown in (b), but not that in (a).

Table 2 shows the sensitivity, specificity, PPV and NPV of CT, palpation during surgery and intraoperative ultrasonography for the detection of para-aortic lymph node metastasis and Table 3 shows the P-values for pair-wise comparisons of the modalities.

Table 2. Sensitivity, specificity and positive and negative predictive values (PPV and NPV) of computed tomography, palpation during surgery, and intraoperative ultrasonography for the detection of para-aortic lymph node metastasis
MethodSensitivity (% (n))Specificity (% (n))PPV (% (n))NPV (% (n))
Computed tomography42.9 (15/35)96.1 (123/128)75.0 (15/20)86.0 (123/143)
Palpation60.0 (21/35)82.0 (105/128)47.7 (21/44)88.2 (105/119)
Intraoperative ultrasonography91.4 (32/35)69.5 (89/128)45.1 (32/71)96.7 (89/92)
Table 3. P-values for pairwise comparisons of modalities
Modality pair Sensitivity SpecificityPPVNPV
  1. CT, computed tomography; IU, intraoperative ultrasonography; NPV, negative predictive value; Palp, palpation; PPV, positive predictive value.

CT vs. Palp0.23180.00070.07730.7278
IU vs. CT< 0.0001< 0.00010.03440.0133
IU vs. Palp0.00530.02860.93070.0458

A total of 35 women had actual node metastasis. Enlarged para-aortic lymph nodes were detected on CT in 20 women, of whom 15 had metastasized nodes; 20 women had metastasized para-aortic nodes that were not identified as enlarged on CT. If para-aortic lymphadenectomy had been performed only when enlarged lymph nodes were detected on CT, the number performed could have been reduced from 163 to 20 (12.3%); however, metastatic nodes would have been missed in 20 out of 35 women. Similarly, enlarged nodes were palpated in 44 women during surgery, and metastatic nodes were found in 21 of these. Intraoperative ultrasonography detected enlarged nodes in 71 women, of whom 32 had metastatic nodes. If para-aortic lymphadenectomy had been performed only when enlarged lymph nodes were detected, the number performed would have been 44 (27.0%) and metastasis would have been missed in 14 women on palpation, while the number performed would have been 71 (43.6%) and metastasis would have been missed in three women on intraoperative ultrasonography.

Discussion

According to the International Federation of Gynecology and Obstetrics surgical staging criteria, patients with ovarian or endometrial cancer and with positive para-aortic lymph nodes are classified as Stage IIIC. However, there is no agreement as to how to assess the lymph node status.

Tempany et al.3 compared ultrasonography, MRI and CT for staging advanced ovarian cancer, and all three modalities had relatively high specificities but only mediocre sensitivities for para-aortic lymph node metastasis. Similar results were obtained by Ozalp et al.4, who compared lymphoscintigraphy, Ga-67 scintigraphy and CT. The current prevailing notion is that enlarged lymph nodes must be palpated and resected if necessary. However, several lines of evidence suggest that the results of intraoperative palpation are frequently incorrect5–7. Thus clinicomorphological factors and palpation at surgery cannot be relied upon to predict the lymph node status. Young et al.8 reported that incomplete surgical procedures often caused inadequate down staging in ovarian cancer, and Trimbos et al.9 reported that complete surgical staging with lymphadenectomy improved the prognosis of early ovarian cancer. However, systematic lymphadenectomy is not yet regarded as the standard surgical procedure because it increases surgical morbidity and its therapeutic significance is uncertain. There have been several reports suggesting that lymphadenectomy improves the prognosis for women with ovarian carcinoma9–11, but others remain unconvinced12, 13. Panici et al.14 carried out a prospective randomized study of women with optimally debulked advanced ovarian carcinoma, and reported that systematic lymphadenectomy improved progression-free but not overall survival compared to the resection of bulky nodes only.

The situation for endometrial carcinoma is similar to that for ovarian carcinoma. A study by the Gynecologic Oncology Group15 stated that palpation of the nodal area should not be relied upon to determine whether lymphadenectomy should be performed, and preoperative imaging assessments have been reported to be a poor predictor of nodal disease4, 16–18. In many hospitals pathological factors guide clinicians in their decision as to whether or not to remove nodes. Also, in the hospitals participating in this study, para-aortic lymphadenectomy is not performed in cases of well-differentiated endometrial carcinoma in which no muscular invasion has been demonstrated. However, only a few women fit these conditions. Moreover, preoperative pathological findings and macroscopic impression of muscular invasion sometimes differ from the final pathological diagnosis. Only systematic lymphadenectomy is an accurate method for evaluating nodal status, but it is not the standard procedure in endometrial carcinoma. Several reports have suggested that lymphadenectomy improves the prognosis for women with endometrial carcinoma19–24, but this has been disputed25, 26.

In summary, the assessment of para-aortic lymph nodes is very important in determining the spread of ovarian and endometrial carcinoma, but there is no reliable procedure for doing this other than lymphadenectomy, the therapeutic significance of which is uncertain. Moreover, it increases operating time and blood loss14, and sometimes causes complications such as para-aortic lymph cyst, chylous ascites and bowel complications. Therefore, it would be better to have some criteria for the selection of only those women who actually require para-aortic lymphadenectomy.

In this study, both specificity and PPV were highest for CT. If para-aortic lymphadenectomy had been performed only when enlarged lymph nodes were detected on CT, its high specificity would have enabled most women without metastasis to avoid lymphadenectomy. However, node metastasis would have been missed in more than half of the women in which it was present. Therefore CT would not have been a practical approach for selecting women who require para-aortic lymphadenectomy. Both sensitivity and NPV were highest with intraoperative ultrasonography, which could have reduced the number of para-aortic lymphadenectomies performed by around 60%, while maintaining the sensitivity for node metastasis at around 90%.

One of the advantages of ultrasound imaging is its high resolution, which is less than 1 mm, and ultrasonography is the best current clinical imaging modality. However, transabdominal ultrasonography overlooks metastatic nodes as frequently as CT because ultrasound is attenuated and scattered by obstacles such as the abdominal wall and bowel3. Bringing an ultrasound probe close to a target is a very good method of avoiding this, and the intraoperative approach makes it possible. In daily clinical practice, most gynecologists are convinced of the superiority of transvaginal ultrasonography over the other imaging modalities in observations of the uterus and ovaries near the vagina. Para-aortic areas are flat and the branching pattern of the vessels is simple compared with that of the intrapelvic cavity. In some cases, color Doppler imaging is useful for making a distinction between lymph nodes and vessels by observing blood flow in the latter. Para-aortic lymph nodes are therefore suited to observation by intraoperative ultrasonography.

The diagnostic accuracy of an examination depends upon the criteria that are used. However, when a new method is being tested there are initially no data available on which to base a decision as to the criteria that should be used. In this study, a lymph node was defined as abnormal simply when its diameter in the transverse plane was more than 5 mm, a value that was based upon data from a CT imaging study2. This cut-off level appears to be relatively effective, judging from the results of this study. Obtaining further imaging data, such as size, number, shape and internal echo of lymph nodes, would allow better criteria to be established for diagnosing lymph node metastasis.

In conclusion, intraoperative ultrasonography can detect node metastasis with high sensitivity, allowing avoidance of unnecessary para-aortic lymphadenectomy in women with ovarian and uterine corpus malignancy.

Acknowledgements

I would like to thank Toshiharu Yasugi, Katsumi Mizutani (Tokyo Metropolitan Komagome Hospital), Shigeki Takeshita, Takuya Ayabe (Teikyo University), Yuji Taketani (Tokyo University) and my other colleagues, without whose help this work would not have been possible.

Ancillary