The Application of Sentinel Lymph Node Biopsy in Cervical Cancer

Lymph Node Involvement in Cervical Cancer

Although the FIGO staging system does not include lymph node evaluation, lymph node status is the main prognostic factor of ECC and a crucial factor in determining treatments. The 5-year overall survival (OS) rate for patients with ECC is 88% in the absence of nodal spread. This value drops significantly to 57% when lymph node involvement is demonstrated. The number and sites of metastatic nodes play a major role, especially when the paraaortic area is reached. The estimated 5-year survival rate decreased from 84.9% to 33.1% if more than one positive node was present. Similarly, the 10-year survival rate dropped from 84.9% to 26.5%. In terms of nodal distribution, in general, the survival rate drops when the common iliac and upper areas are invaded. The addition of radiation therapy or, more recently, chemoradiation has been proved to improve local control and even the survival rate. Consequently, systematic lymphadenectomy may be performed in patients with ECC.

On the other hand, the morbidity of lymphatic dissections has been underestimated for a long time. Longer operative times, perioperative complications such as vascular or neurologic injuries, and delayed morbidities such as lymphedema and lymphocysts occur frequently. In addition, postoperative ileus, venous thromboembolism, and prolonged hospital stay have been reported. The occurrence of these iatrogenic complications is not rare. Twenty percent of patients undergoing lymphadenectomy subsequently develop lower limb lymphedema that not only is an incurable condition but also carries a heavy psychological burden including anxiety, depression, and adjustment disorders. Quality of life of patients cured of their cancer is now a major issue.

The reported frequency of nodal involvement ranges from 0% to 31% in patients with stage IA1 to IIA disease. Thus in the best case scenario, more than 70% of patients are undergoing unnecessary lymphadenectomy with all its potential complications, without any staging or therapeutic benefit. Another consideration is the low nodal burden. The median number of metastatic nodes is two in ECC, and in 22% to 38% of patients the nodal metastasis measures less than 2 mm. Most nodal metastases are located in the obturator and external iliac area. This is explained by the physiologic drainage of the uterine cervix, with a pathway going through the parametrium and terminating in the obturator and iliac nodes. However, alternative pathways, with direct drainage to the presacral and paraaortic areas, have also been described. The low prevalence of nodal involvement in these anatomic areas has made it difficult for authors to draw conclusions regarding the necessity of systematically sampling these areas. Finally, 10% to 15% of patients who are considered to be free of nodal metastasis will develop relapses emerging from lymphatic territories, showing that the positive node was not retrieved.

For all these reasons, a systematic lymphadenectomy in which nodes are harvested between anatomic landmarks and not according to tumor spread may not be appropriate for obtaining accurate information regarding nodal involvement.

SLN sampling is not an emerging concept. Cabanas and co-workers introduced it in 1977 for penile cancer; then it was applied to melanoma and breast cancer. In gynecology, Levenback and colleagues promoted the technique for use in vulvar cancer. A pioneer work by Echt and colleagues integrated the principle into the management of cervical cancer tumors. Later, many studies tried to evaluate the feasibility and diagnostic value of this technique for treatment of uterine cervical malignancies. Researchers subsequently reported on diagnostic accuracy and the use of safety rules to limit the risk of false-negative results.

Cervical Cancer: A Candidate for Sentinel Lymph Node Mapping

The role of imaging techniques in the assessment of nodal metastasis and SLN detection was studied for a long time as researchers sought the optimal modality. In a major meta-analysis, Choi and co-workers compared all these techniques. The patient-based sensitivities of computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET) or PET-CT were reported to be 50%, 56%, and 82%, respectively, and the specificities of the same techniques were 92%, 91%, and 95%, respectively. A region- or node-based comparison of the same modalities showed sensitivities of 52%, 58%, and 54% and specificities of 92%, 97%, and 97%, respectively. Despite the superiority of PET or PET-CT over the other imaging techniques, performance remains modest when one considers the potentially broad role of this diagnostic modality in determining management and prognosis. The poor performance of these noninvasive techniques in the evaluation of nodal metastasis establishes the need for other methods.

The SLN is, by definition, the first chain node that receives primary lymphatic flow from a solid tumor. Accordingly, SLN mapping is based on the fact that lymphatic drainage occurs in a defined pattern away from a tumor. Following this reasoning suggests that if the SLN is negative for metastasis, then the node beyond that point should also be negative.

SLN mapping cannot be applied to all solid tumors, but cervical cancer appears to be a reasonable candidate for many reasons. First, nodal metastasis in ECC is present in 0% to 16% of tumors smaller than 2 cm, a rate that reaches 15% to 31% in stage IB disease, implying that a vast majority of patients will not benefit from a systematic lymphadenectomy. Second, routine preoperative imaging still fails to enable accurate determination of nodal status in such patients, owing to the small size of the metastases. Third and most important, the lymphatic drainage in cervical cancer mostly follows well-defined pathways with some checkpoints that were once defined as “interrupting nodes” and are known today as SLNs. But alternative pathways do exist. For all these reasons, cervical cancer is an area of opportunity for the application of targeted SLN biopsy.

In addition, the tracers can be easily injected into the cervix—that is, around the tumor—making the SLN actually representative of the tumor drainage and not only of the organ.

Anatomy of the Cervix and Its Lymphatic Channels

The lymphatic drainage of the uterine cervix, which is a centrally located organ, was studied by anatomists and surgeons by use of a variety of dyes. Reports from Reiffenstuhl and from Plental and Friedman described the lymphatic drainage system based on pathways. The most important route extends along the lateral parametrium to the obturator, external iliac, internal iliac, and common iliac lymph nodes. A second route is via the anterior channel that follows the vesicouterine ligament and extends to the interiliac lymph nodes, ending in the external iliac chain. Another route is the posterior channel that runs along the uterosacral ligament and drains in the common iliac, sacral, and paraaortic nodes. The normal drainage through these channels occurs in a stepwise progression, although exceptional variations do exist. Skip lymph node drainage in cervical cancer is considered to be a rare event, and paraaortic involvement comes after the sequential spread to pelvic and common iliac nodes. Spread through aberrant lymphatic channels is a potential explanation for this phenomenon.

Rouviere observed that there was a collector of the internal iliac pedicle that could pool directly to the common iliac area at the level of L5, and he theorized about the potential presence of an anastomosis between the uterine and cervical drainage systems that could spread metastasis at the level of L4 through the infundibulopelvic ligament. These ideas were supported by many other researchers who reported direct extension to the paraaortic nodes. Roughly, typical drainage systems were defined as those that followed the external iliac and interiliac basins, whereas other drainage routes were considered atypical. These anatomic data were also confirmed by many other authors, among them those who performed systematic lymphadenectomy and those who tried to standardize the surgical procedure. The conclusion from these studies is that removal of the nodes from the external iliac, interiliac, obturator, and common iliac territories will permit identification of most of the involved nodes. The information that lymphadenectomy provided over the years was at the basis of SLN sampling ( Fig. 6.1 ).

Efficiency of Sentinel Lymph Node Technique

To be considered efficient, SLN biopsy should have both a high detection rate and a high negative predictive value. The detection rate can be calculated as follows: at least one SLN detected per patient divided by the total number of patients injected. The negative predictive value per patient is the result of the following formula: true-negative SLNs divided by the sum of the true-negative and false-negative SLNs.

The development of the lymphatic system starts at the pelvic side walls at around 10 weeks of gestation; because the cervix is at the midline, lymphatic drainage occurs from the organ to the laterally situated plexuses. For this reason, SLN found in a hemipelvis will be predictive of the status of the remaining nodes on the ipsilateral side and does not represent the contralateral hemipelvis. This concept caused many oncologic surgeons to consider the hemipelvis to be a separate entity and to suggest that the sensitivity of SLN mapping techniques be calculated based on the hemipelvis as the unit of analysis, rather than on a per-patient basis.

Detection rates varied significantly from one study to another, ranging from 15% to 100% per patient and from 43% to 97% when the hemipelvis was used as the unit of analysis. Major contributing factors were the FIGO stage, mapping technique, and surgeon’s experience; a history of previous cone biopsy or preoperative brachytherapy did not affect this rate. Neoadjuvant chemotherapy (NAC) remains in a gray zone, with conflicting reports. The surgeon’s experience is known to play a major role. Plante and colleagues and Seong and colleagues reported a relationship between the surgeons’ experience and an increase in SLN detection. Khoury-Collado and colleagues further supported this idea by stating that more than 30 cases are needed for the detection rate to increase from 77% to 94% in endometrial cancer. A retrospective analysis by Dargent and Enria showed that the average time to retrieve the SLN was 58.7 minutes for the first 35 patients, dropping to 35.5 for subsequent patients ( P < .05). A study conducted by Hwang and co-workers showed that around 40 and 57 cases are needed per surgeon to achieve a turning point in operation time and complication rates, respectively.

The learning curve applies not only to the intraoperative detection of sentinel nodes, but also to the initial injection technique, lymphoscintigraphy interpretation, and pathologic assessment. As an example, Lantzsch and colleagues and Li and colleagues reported detection failure in patients who had received an inappropriate injection of the tracer substance.

False-negative results and negative predictive value are sparsely assessed in the literature. Some of the findings may be related to the characteristic of the tumor and pelvic anatomy, and others to the application and analysis of the results. Hauspy and co-workers performed a review of the literature to assess the false-negative rate. When their results were combined with those of the previous studies, the rate of false-negative results was less than 2%. With regard to tumor characteristics, the first parameter was the initial tumor size and the second was lymphovascular space invasion (LVSI). Darlin and colleagues, in a study that included 105 patients with ECC (IA1–IIA) and in which the isotopic technique was followed, reported 100% negative predictive value in tumors smaller than 2 cm. A multicenter cohort study by Altgassen and co-workers included 590 eligible patients and followed both the isotopic and colorimetric techniques; after systematic lymphadenectomy plus or minus paraaortic lymph node dissection was performed, the negative predictive value was 99.1% for tumors smaller than 20 mm, whereas it was a disappointing 94.3% for tumors larger than 20 mm. Slama and colleagues found an increase in false-negative rates in tumors measuring more than 20 cm3 and those with LVSI in a study that included 225 patients with FIGO stage IA2 to IIB cervical cancer and in which ultrastaging was performed on SLNs.

In light of these findings, this technique was considered to be fully reliable only when the nodes are detected bilaterally, thus achieving the so-called “optimal mapping.” Cibula and colleagues, in a retrospective multicenter cohort study that included 645 patients with FIGO stage IA to IIB disease who underwent SLN biopsy and ultrastaging, described a 1.3% rate of false-negative findings when SLNs are detected bilaterally, thus reinforcing the results of SENTICOL I.

Based on this principle, an algorithm was described by Cormier and co-workers and the team at Memorial Sloan Kettering Cancer Center (MSKCC). In this algorithm, all mapped SLNs are submitted for pathologic analysis; when results of routine hematoxylin and eosin staining are negative, ultrastaging is performed. Suspicious nonsentinel lymph nodes are also removed and sent for analysis. If mapping is not achieved in a hemipelvis, a complete side-specific lymphadenectomy becomes necessary, not forgetting the parametrectomy that is performed by removal of tissue en bloc with the primary tumor specimen. This algorithm relies on use of the hemipelvis as a unit of analysis in order to improve the negative predictive value of SLN biopsy (96.8%) and to decrease false-negative results (7.4%) (see Fig. 6.1 ). The importance of the negative predictive value and false-negative rate resides in the fact that if a patient is incorrectly labeled as N0, she will not benefit from postoperative radiation therapy and will be at higher risk of recurrence.

False-negative results have been observed in biopsies of SLNs in all areas. However, the occurrence of false-negative findings in the parametrium has been a matter of debate. The problem starts with the difficult interpretation of the preoperative lymphoscintigraphy images, includes the intraoperative detection and the false isotopic signals, and ends with the microscopic analysis of these nodes. It is important to mention that debates are rising as to whether parametrial nodes should be considered part of the sentinel nodes. Frumovitz and co-workers performed a “triple injection” using radiocolloids, blue dye, and India ink to assess the lymphatic drainage in 20 patients with ECC who were treated with radical hysterectomy or trachelectomy. Pathologic processing of the parametrium and the removed nodes was also done. By following the patterns obtained from these injections, and because there were no parametrial nodes in 25% of the patients, the authors concluded that there was a direct route of drainage extending from the cervix to the pelvic nodes and bypassing the parametrial nodes. Thus these nodes may not be always considered as sentinel in women with ECC.

Whether the unit of analysis is per patient or per side is also important. A true false-negative result occurs whenever a metastatic lymph node is detected but the sentinel node is negative for tumor cells. Conversely, a negative SLN contralateral to a positive SLN is not necessarily a false-negative finding, because SLN status on one side of the pelvis does not predict the nodal status on the contralateral side. For this reason, many authors have underlined the importance of interpreting the results of SLN biopsy per side and not per patient.

The aim of the SLN technique is not only to replace the aggressive lymphadenectomy procedure but also to decrease the morbidities of the baseline hysterectomy or trachelectomy; this is another implicit advantage that is often forgotten. A study by Strnad and co-workers showed that the risk of parametrial involvement is minimal if cervical cancer infiltrates less than two-thirds of the cervical stroma and if SLNs are negative, whereas the risk reaches 28% when the result is positive. In other words, the SLN technique can be used as a guide to calibrate the extent of radical hysterectomy, making it possible to consider a modified radical hysterectomy, a nerve-sparing procedure, or a simple trachelectomy when SLN biopsy results are negative.

The efficiency of the SLN technique is not an absolute fact; rather, well-defined rules need to be applied in order to reach a reliable application of this technique and what is referred to as quality assurance. The first rule is related to preoperative selection of patients. A thorough clinical examination and accurate imaging and pathologic analysis are the most basic requirements. The aim of this exploration is to select tumors smaller than 2 cm without suspected nodal involvement and to stratify them based on their histologic characteristics: squamous versus glandular. The second rule is related to the medical team. The surgical and analytical team members must be trained in the interpretation of the preoperative images, such as lymphoscintigraphic images or, more recently, single-photon emission computed tomography–computed tomography (SPECT-CT) scans; the injection of the tracer, whether isotopic, colorimetric, or fluorescent; perioperative decision making according to the MSKCC algorithm; and, finally, the definitive pathologic analysis that is the last factor affecting subsequent management ( Fig. 6.2 ).

Methodologies

Three main techniques are used for the completion of SLN biopsy. They are classified according to the tracer’s nature: radioactive tracers, colorimetric dyes, and fluorescent dyes. These tracers can be injected preoperatively or intraoperatively based on their diffusion and detection times. For instance, the isotopic radioactive tracers can be injected at any time between 1 and 24 hours preoperatively, whereas the colorimetric and fluorescent dyes are detectable 10 to 20 minutes after the injection of the tracer. The amount of tracer injected ranges from 1 mL to 4 mL. The number of injections in the cervical stroma varies from two to four, and their locations may be at the 3- and 9-o’clock positions, the 3-, 6-, 9-, and 12-o’clock positions, or the 2-, 4-, 8-, and 10- o’clock positions based on the surgeon’s preferences and training ( Fig. 6.3 ). Authors have also used superficial injections, deep injections, or both. Results appear similar whichever technique is used, making SLN biopsy a robust technique.

Experience is needed for the injection of the tracer. A wrongly performed injection can result in improper diffusion of the tracer—from no diffusion to extended diffusion that may mask the surgical field—and therefore suboptimal detection. The reported failure time is the time after which it becomes useless to look for labeled nodes; this ranges from 70 to 150 minutes.