- © 2002 by American Society of Clinical Oncology
Lymphatic Mapping and Sentinel Node Identification in Patients With Cervix Cancer Undergoing Radical Hysterectomy and Pelvic Lymphadenectomy
- Charles Levenback,
- Robert L. Coleman,
- Thomas W. Burke,
- W. Michael Lin,
- William Erdman,
- Michael Deavers and
- Ebrahim S. Delpassand
- From the Department of Gynecologic Oncology, Department of Diagnostic Imaging, and Department of Pathology, The University of Texas M.D. Anderson Cancer Center, Houston; and Division of Gynecologic Oncology and Department of Diagnostic Imaging, The University of Texas Southwestern Medical Center, Dallas, TX.
- Address reprint requests to Charles Levenback, MD, Department of Gynecologic Oncology—440, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Houston, TX 77030; email: clevenba{at}mdanderson.org
Abstract
PURPOSE: The purpose of this study was to determine the feasibility of sentinel node identification in patients with invasive cervix cancer undergoing radical hysterectomy and pelvic lymphadenectomy using preoperative and intraoperative lymphatic mapping.
PATIENTS AND METHODS: Thirty-nine patients at two institutions were enrolled onto this institutional review board–approved study. All underwent preoperative lymphoscintigraphy and intraoperative lymphatic mapping with blue dye and a handheld gamma probe. Radical hysterectomy was aborted in four patients because metastatic disease was discovered on frozen section analysis of the sentinel node.
RESULTS: Preoperative lymphoscintigraphy revealed at least one sentinel node in 33 patients (85%), including 21 (55%) with bilateral sentinel nodes. All 39 patients had at least one sentinel node identified intraoperatively. Eighty percent of sentinel nodes were in three pelvic locations: iliac, obturator, and parametrial (in descending order of frequency). The remaining sentinel nodes were in the common iliac and para-aortic nodal basins. A total of 132 nodes were identified clinically as sentinel nodes; 65 (49%) were both blue and hot, 35 (27%) were blue only, and 32 (24%) were hot only. Eight patients (21%) had metastatic disease. In five of these patients, sentinel nodes were the only positive lymph nodes. One patient had false-negative sentinel nodes. She had four microscopically positive parametrial nodes that were resected in continuity with the uterus. The sensitivity of the sentinel node was 87.5% and the negative predictive value was 97%.
CONCLUSION: Preoperative lymphoscintigraphy and intraoperative lymphatic mapping were highly successful at identifying sentinel nodes in patients undergoing radical hysterectomy.
CERVIX CANCER is a good candidate disease for lymphatic mapping for several reasons. First, the incidence of positive nodes in the most favorable group of patients, stage IB1, is approximately 15%.1-3 This means that more than 80% of these patients who undergo lymphadenectomy derive no benefit from the procedure yet must endure the associated increase in operative time, blood loss, risk of lymphocyst, and lymphedema. Second, the cervix is a midline structure with complex lymphatic drainage, and thus an extensive pelvic dissection is necessary to be certain that all regional lymph nodes are removed. Third, the location of positive nodes cannot be determined on the basis of clinical factors such as the size of the primary, morphology (exophytic or endophytic), or location (ectocervical, endocervical). Finally, accurate determination of nodal status in patients with cervix cancer would allow triage of node-positive patients to chemoradiation without radical pelvic surgery.
The purpose of this study was to determine the feasibility of sentinel node identification with a combination of preoperative lymphoscintigraphy and intraoperative lymphatic mapping with blue dye and a radionuclide in patients with invasive cervix cancer undergoing radical hysterectomy and total pelvic lymphadenectomy.
PATIENTS AND METHODS
This study was approved by the institutional review boards at The University of Texas M.D. Anderson Cancer Center in June 1999 and at the University of Texas Southwestern Medical School in August 1999. Patients were eligible if they were scheduled to undergo surgical exploration with the intent of undergoing radical hysterectomy and total pelvic lymphadenectomy for cervical cancer. All histologic tumor types were allowed. Preoperative testing was at the discretion of the surgeon.
Preoperative Lymphoscintigraphy
Preoperative lymphoscintigraphy was performed by injection of 1.0 to 1.5 mL of filtered technicium-99 radiocolloid into the cervix. This was divided into four injections around the tumor, or at the 12, three, six, and nine o’clock positions around the cervix for midline tumors. Most patients reported mild to moderate pain with the injection but reported that the pain passed quickly. Injections were performed with a 25-gauge needle with constant gentle pressure to try to prevent spillage into the vagina. In most cases the first lymphoscintigrams were obtained within 30 minutes of radionuclide injection. Lymphoscintigrams were repeated as necessary for up to 3 more hours to identify sentinel nodes. Anterior/posterior and lateral views were obtained with markers placed on bony landmarks, such as the pubic symphysis and the anterior superior iliac crests. In some cases, transmission scans were obtained to show the outline of the body at the discretion of the nuclear medicine specialist Fig 1.
Fig 1. (A) Preoperative lymphoscintigram in a patient with stage IB1 squamous carcinoma of the cervix. A sentinel node is visible in the right pelvis. The liver edge is visible at the top right border. (B) Intraoperative photograph of the right pelvis taken of the same patient after injection of blue dye. The efferent lymphatic channel is visible traversing the parametrium and entering the sentinel node. The bladder is behind the retractor on the left.
The half-life of technetium-99 is approximately 6 hours. For this reason, if preoperative lymphoscintigraphy was performed more than 18 hours (three half-lives) before surgery, the injection of radionuclide was repeated the morning of the procedure, 1 to 6 hours before surgery.
Intraoperative Lymphatic Mapping
On the day of surgery, after induction of general anesthesia, patients were placed in low lithotomy position and examined. Routine preparation and draping was performed. The skin incision was made, the abdomen was explored, and the bowel was packed out of the pelvis. At this point a speculum was placed in the vagina and isosulfan blue (Lymphazurin 1%, US Surgical Co., Norwalk, CT) was injected into the cervix as has previously been described.4 A needle extender or spinal needle was used with gentle pressure to minimize spillage of the blue dye. The injection sites were chosen as described for lymphoscintigraphy. Before, during, or after the injection of blue dye depending on the surgeon’s preference, incisions were made in the pelvic peritoneum to expose the retroperitoneum. The retroperitoneum was then visually inspected and scanned with a hand-held gamma counter to locate sentinel nodes. Both the pelvic and para-aortic regions were surveyed. The probe was collimated and angled laterally as much as possible to reduce detection of residual radioactivity from the primary cervical tumor. The preoperative lymphoscintigram was available in the operating room to help guide the search for sentinel nodes.
Sentinel Node Identification
Any node that was blue was considered a sentinel node. Radioactive lymph nodes identified with the gamma probe were removed and the radioactivity measured ex vivo. If the counts were at least 10-fold above background radiation levels, the node was considered sentinel. Sentinel nodes were labeled as blue, hot (radioactive), or blue/hot. The radioactive counts and location of each sentinel node were recorded.
Pathologic Processing
Sentinel lymph nodes were visually inspected by the pathologist. If gross metastatic disease was visible, simple sections were submitted. Normal-appearing sentinel nodes were cut perpendicular to the long axis into 3- to 4-mm sections and submitted for routine hematoxylin and eosin staining (H&E). If the initial section was negative, five 40-μm wide interval sections were obtained for H&E staining and unstained slides were prepared for cytokeratin immunohistochemical analysis. Frozen section was performed only if requested by the operating gynecologic oncologist.
RESULTS
Patient Characteristics
Thirty-nine patients were accrued to this trial and underwent preoperative lymphoscintigraphy and intraoperative lymphatic mapping. The median age of the cohort was 40 years (range, 27 to 70 years). Patient and tumor characteristics are presented in Table 1. The median tumor size was 3 cm (range, 0.3 to 5 cm). In four cases, radical hysterectomy was aborted when the sentinel node proved positive for metastatic disease on frozen section evaluation. This has been standard practice at our institution for many years. These patients had total pelvic lymphadenectomy completed by the time the frozen section result was available (Table 1).
Tumor and Treatment Characteristics
Lymphoscintigraphy
Preoperative lymphoscintigrams were obtained before the day of surgery in 23 patients (median, 1 day before surgery; range, 1 to 15 days before surgery) and on the day of surgery in 16 patients. Overall, 33 patients (85%; 95% confidence interval, 69% to 94%) had at least one sentinel node localized on lymphoscintigraphy. Twenty-one patients (55%; 95% confidence interval, 37% to 70%) had bilateral sentinel nodes identified. The likelihood of sentinel node localization on lymphoscintigraphy was unrelated to how long before surgery the study was obtained (P = .15). Six patients had no sentinel node localized on lymphoscintigraphy. The likelihood of a negative study was not related to whether the study was done before or on the day of surgery (P = .58). All six patients without sentinel nodes localized on lymphoscintigraphy were found in the operating room to have bilateral hot nodes.
Intraoperative Lymphatic Mapping
On the day of surgery blue lymph nodes were identified in the pelvis a median of 7 minutes (interquartile range [IQR], 5 to 16.5 minutes) after isosulfan blue injection. These nodes remained blue for a median of 21 minutes (IQR, 9.5 to 30 minutes) after identification.
At least one sentinel node was located in each of the 39 patients. Thirty-four patients (87%) had blue sentinel nodes in the pelvis or para-aortic area either bilaterally (24 patients) or unilaterally (10 patients). Thirty-nine patients (100%) had a hot sentinel node in the pelvis or para-aortic area identified with the gamma probe. Twenty-eight patients had bilateral hot sentinel nodes and 11 had unilateral hot sentinel nodes. Figure 1 demonstrates the results of combining the blue dye and radionuclide localization techniques. All five patients in whom no blue node was seen intraoperatively had hot sentinel nodes identified. Three of these five had bilateral sentinel nodes, and two had unilateral sentinel nodes.
In two patients (5.0%), only a single unilateral sentinel node was found. One of these was blue only and the other was hot only. In eight additional patients (21%), a solitary sentinel node was retrieved from each hemipelvis (bilateral solitary sentinel nodes). In four of these patients, both nodes were blue and hot. The remaining 29 patients had bilateral sentinel nodes and had a total of at least three sentinel nodes removed.
Sixteen patients underwent cervical conization before lymphatic mapping and surgical extirpation. The performance of this procedure did not affect the likelihood that a sentinel node would be identified at the time of exploration (P = .68) or affect whether the sentinel node would be blue (P = .2) or hot (P = .49). Likewise, we were unable to demonstrate a significant relationship between unilateral or bilateral sentinel node identification and the size of the primary lesion when evaluated either as a continuous variable (P = .36) or a categorical (≤ 4 cm v > 4 cm) variable (P = .3).
Sentinel Node Characteristics
A total of 132 lymph nodes were identified in the operating room as sentinel. Operating surgeons did not attempt to dissect sentinel nodes from the surrounding fat pad. Our pathologists identified a total of 185 lymph nodes among the 132 sentinel lymph node specimens. The median number of sentinel nodes identified in the operating room per patient was four (IQR, three to five), with a median two sentinel nodes identified in each hemipelvis.
The descriptive characteristics of the 132 sentinel nodes identified are summarized in Table 2. Sixty-five (49%) were both blue and hot. Thirty-two nodes (24%) were identified using the hand-held gamma probe but were not stained blue. All of these nodes were confirmed to be hot by ex vivo counting as described in the materials and methods section. In 35 nodes (26%), only blue dye was observed. One hundred five sentinel nodes (80%) were found in three pelvic sites: obturator, iliac below the bifurcation of the common iliac artery, and parametrial. The most common locations for sentinel nodes were the interiliac and obturator basins (Table 3). Surprisingly, 12 sentinel nodes (9%) were found in the para-aortic area: nine on the right side and three on the left. The locations and characteristics of sentinel nodes are shown in Fig 2.
Sentinel Node Blue/Hot Characteristics
Methods of Identification of Sentinel lymph Nodes by Nodal Location
Fig 2. Location, number, and characteristics of sentinel lymph nodes.
Metastatic Disease
A total of 996 sentinel and nonsentinel lymph nodes were retrieved from the patients in this cohort (median, 25 nodes per patient; range, seven to 60 nodes). Metastatic disease was found in eight patients (21%). A total of 25 positive nodes were removed from these patients, of which 21 were sentinel and four were nonsentinel. Seven of the eight patients with positive nodes had positive sentinel nodes; the number of positive sentinel nodes per patient ranged from one to five. In two of these seven patients, there were positive nonsentinel nodes in addition to the positive sentinel nodes. In the remaining five patients, the only positive nodes were the sentinel nodes. One patient of the eight with positive nodes had false-negative sentinel nodes (12%). This patient had four sentinel nodes identified on preoperative lymphoscintigraphy. Intraoperatively, all four of these nodes were both hot and blue. In this case, the pathologist identified microscopic disease in four nonpalpable parametrial lymph nodes removed with the radical hysterectomy specimen. The sensitivity of the sentinel node was 87.5% and the negative predictive value was 97%.
Pathologic Evaluation
In 31 patients with no positive sentinel nodes on routine H&E processing, serial step sectioning was performed. No micrometastases were found using this method. Ten patients with negative sentinel nodes on routine H&E and serial step sectioning had their sentinel nodes submitted to cytokeratin immunohistochemical analysis. No metastases were identified with this technique. One patient with a positive sentinel node on routine H&E staining had a micrometastasis found by serial sectioning of the contralateral sentinel node.
Complications
Overall tolerance for the mapping procedure was excellent. No patients experienced anaphylaxis during injection of either isosulfan blue or technetium-99 radiocolloid. Transient decrease in oxygen saturation as measured by pulse oximetry was noted in a few patients as described previously.5 Some patients complained the radionuclide injection was painful; however, they reported the discomfort was very brief (1 or 2 minutes).
DISCUSSION
Several lymphatic mapping studies have been performed in patients with cervical cancer.4,6-11 Previous experience with blue dye alone4 was disappointing, with a sentinel node detection rate of only 60%. This led us to design this study using preoperative lymphoscintigraphy and intraoperative lymphatic mapping with both blue dye and radiocolloid. In addition, we focused on patients with smaller tumors that should have a lower chance of having lymph nodes replaced with tumor. In this study, we increased our sentinel node identification rate to 100%. We found that the combined technique revealed more complexity to the lymphatic drainage of the cervix.
Verheijen et al6 used a combined technique similar to ours in a series of 10 patients and succeeded in identifying the sentinel node in eight patients. Dargent et al7 used blue dye alone for laparoscopic sentinel node identification. Once the technique was perfected, there was a high rate of successful sentinel node identification. All sentinel nodes were found in the pelvis. Malur et al11 reported a series of 50 patients, including 18 with stage II or higher disease, who underwent lymphatic mapping with a variety of techniques. They identified sentinel nodes in 90% of patients using a combination of blue dye and radiocolloid, compared with 55% using blue dye alone and 76% using radiocolloid alone. They used a laparoscopic gamma probe to successfully identify sentinel nodes in an unspecified number of patients.
In our study, we found sentinel nodes that were hot and blue, hot only, or blue only. Several factors might account for these differences. First, the radionuclide injection was performed with the patient awake and blue dye with the patient under general anesthesia. Under anesthesia more manipulation of the cervix was possible with the use of a larger speculum or tenaculum and perhaps this resulted in slightly different injection sites. The volumes of the two injections were different, 1 to 1.5 mL of radionuclide compared with 3 to 4 mL of blue dye. Finally, the half-life of the materials is different, minutes for blue dye and 6 hours for the radionuclide. The radionuclide was, therefore, more likely to be detected in second echelon nodes.
The sentinel node is defined as the first draining lymph node. When a patient has more than one node that picks up blue dye or radiocolloid are they both sentinel or is one sentinel and the other a second echelon lymph node? The lymphatic drainage of the cervix is thought to be sequential, as shown on the left in Fig 3. In this case the sentinel node is always in the pelvis. It is also possible that there could be direct lymphatic drainage from the cervix to common iliac or even para-aortic lymph nodes as shown on the right in Fig 3. The mapping techniques we used cannot reliably distinguish between these two possibilities, therefore we labeled all nodes that took up dye or radiocolloid as sentinel.
Fig 3. Two possible drainage lymphatic drainage patterns of the cervix. Mapping techniques do distinguish between these two models, although nonmapping data suggest the sequential model is most common.
One patient in our series had negative sentinel nodes and positive nonsentinel parametrial nodes removed with the primary tumor. Strictly speaking, the sentinel nodes in this case were false-negative. Medial parametrial nodes are very difficult to detect on lymphoscintigraphy because they are so close to the cervix, which retains a large amount of radioactivity after radiocolloid injection. Medial parametrial nodes are difficult to detect with blue dye for the same reason. The afferent lymphatic channel is short and not visible at the start of the dissection. The dye is gone by the time the parametrial dissection is performed. Finally, these nodes may be quite small, as was the case with our patient.
There were patients in our series who did not have sentinel nodes detected on preoperative lymphoscintigraphy yet had hot nodes detected in the operating room with the hand-held gamma counter. Possible explanations include the hot node being very close to the primary tumor and indistinguishable on the scintigram, premature scanning (before uptake of the radiotracer by the sentinel node), and slightly different injection sites in patients with a second radionuclide injection just before surgery.
We believe that our learning curve with in this study was brief. The participating gynecologic oncologists had experience with lymphoscintigraphy and blue dye in patients with vulvar cancer and blue dye alone in patients with cervix cancer. Gynecologic oncologists with no prior mapping experience might have a longer learning period. Our nuclear medicine specialists did not have any recent experience performing pelvic examinations. Gynecologic oncologists performed the majority of injections.
A recent study used an analysis of the location of lymph node metastases found at the time of radical hysterectomy to argue that the sentinel node concept is not applicable in cervix cancer.12 Like these authors, we too found positive lymph nodes throughout the pelvis and para-aortic regions. However, the fact that we could select these positive nodes with our mapping techniques indicates to us that the sentinel node concept most likely applies to the cervix as it does to other solid tumor systems.
It is interesting to speculate on how lymphatic mapping will be incorporated into clinical practice. We found that the preoperative lymphoscintigram had little or no impact on clinical decision making and do not foresee this as standard practice. On the other hand, new imaging technologies such as positron-emission tomography could be combined with lymphatic mapping to help identify patients with early metastatic disease. New molecular techniques could be combined with lymphatic mapping to identify biologic markers of metastatic disease.13
Our data is the largest series of patients with cervix cancer subjected to combined intraoperative lymphatic mapping using a consistent technique and case selection criteria in a multi-institutional setting. These data are not sufficient to recommend abandoning standard practice, pelvic lymphadenectomy, in favor of sentinel node biopsy. We do believe that there are enough data to proceed with a large multi-institutional trial and have proposed such a study to the Gynecologic Oncology Group.
Assuming lymphatic mapping proves valid in cervix cancer patients, phase III trials with survival end points will be necessary to determine of sentinel node biopsy can replace pelvic lymphadenectomy in patients with early cervix cancer.
- Received May 7, 2001.
- Accepted September 17, 2001.
