Sentinel lymph node mapping in cervical cancer: the future?

Authors


Prof A Covens, Division of Gynecologic Oncology, Odette Cancer Centre, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario, M4N 3M5, Canada. Email al.covens@sunnybrook.ca

Abstract

Please cite this paper as: Eiriksson L, Covens A. Sentinel lymph node mapping in cervical cancer: the future? BJOG 2012;119:129–133.

The primary objective in the treatment of many cancers is cure. Many patients are willing to undergo morbid procedures and treatments in order to achieve cure. In recent years, however, advances in understanding and technology have allowed for less radical treatments while maintaining, and in some cases improving, oncologic outcomes. Historically, early-stage cervical cancer patients were treated with radical hysterectomy and pelvic lymph node dissection, with associated complications of urinary retention, infection, sexual dysfunction, and loss of fertility. Presently, women are benefiting from less radical procedures, both with fertility-sparing techniques, laparoscopic and robotic approaches, and most recently, sentinel lymph node biopsy.

Although lymph node status is not included in the clinical staging of cervical cancer, it is one of the most important prognostic factors. For this reason, pelvic lymph node assessment is an integral component of definitive surgical management, for both prognostication and planning adjuvant therapy. In cases with low-risk pathological features women are spared adjuvant treatment. This is particularly desirable to young women, who wish to preserve both ovarian and sexual function. Conversely, women with lymph node metastases may be offered primary or adjuvant chemo-radiation in an attempt to improve survival.

The risk of lymph node metastases in women with early-stage cervical cancer is approximately 15%.1 Consequently, over 80% of patients do not benefit from a pelvic lymphadenectomy, but may suffer from adverse complications or sequelae such as lymphoedema, lymphocyst formation, neurovascular and ureteral injury, or blood loss secondary to the dissection. Furthermore, as the cervix is a midline structure with complex lymphatic drainage,2 it is not possible to reliably predict the location of metastatic spread. An extensive pelvic dissection is necessary in order to be certain that all regional lymph nodes are removed, although lymph nodes along aberrant drainage pathways may still be missed.

The sentinel lymph node is the first lymph node to which metastatic disease will spread from a primary tumour. If the sentinel lymph node is free of metastatic disease, all other lymph nodes will also be free of disease. Sentinel lymph nodes are identified by injecting a blue dye or radioactive tracer around the primary tumour. When using dye, blue lymphatic channels are followed and blue-stained nodes are sampled. When a radioactive isotope is injected, lymph nodes with high radioactive counts (i.e. greater than five times the background count) are identified using a gamma probe.

It is our opinion that sentinel lymph node mapping should become the standard over conventional pelvic lymphadenectomy. Proponents of sentinel lymph node mapping believe that the status of regional lymph nodes is accurately represented by the status of the sentinel lymph node.3 The sentinel lymph node concept states that, in the absence of metastatic disease in the sentinel lymph node, the necessity for a complete pelvic lymphadenectomy is obviated. Consequently, adverse events associated with complete lymphadenectomy can be significantly decreased with sentinel lymph node biopsy alone, as has been seen in the treatment of breast cancer, where the risk of lymphoedema, sensory loss, and operative time have been reduced, whereas quality of life is significantly enhanced.4 In addition, oncologic outcomes are expected to be equivalent, and potentially improved, as a result of the increased localisation of lymph node metastases through the detection of aberrant lymphatic pathways, and the use of ultrastaging to aid in the identification of micrometastases.5

The technique requires not only operative skill but also logistic planning, as a portion of the procedure is often performed preoperatively in the nuclear medicine department. However, increased complexity and logistics should not preclude surgical innovation. Initially consisting of injected blue dye alone, several studies have since shown increased rates of sentinel node detection with the addition of injected radiolabelled technetium, with or without preoperative lymphoscintigraphy.6,7 Sentinel lymph node mapping protocols have been modified at various institutions, with approaches directed either by laparoscopy or laparotomy.

At our institution, sentinel lymph nodes are detected using technetium sulfur colloid, with or without lymphazurin blue dye. Between 2 and 4 hours preoperatively 0.2–0.4 mci of filtered sulfur colloid technetium is injected submucosally at 3 and 9 o’clock of the exocervix peripheral to the tumour (if visible). Lymphoscintigraphy is performed 20 minutes after injection. If required, 4 ml of lymphazurin blue dye is injected into the submucosa at the start of the operation, 2 ml each in the same sites as the technetium injections. The lymphazurin dye is used only if bilateral sentinel lymph nodes are not identified on the preoperative scintogram. Intraoperatively, sentinel lymph nodes are detected using a laparoscopic gamma probe (Navigator GPS; Tyco Healthcare) to look for nodes with radioactive counts greater than five times the background count. The pelvic sidewalls, presacral area, and para-aortic lymphatic beds are scanned entirely, and ‘hot spots’ are identified by gamma probe and/or by the visualisation of the blue-stained lymphatic channels and lymph nodes. Identified sentinel lymph nodes are selectively removed and sent to pathology for intraoperative review. The pelvis is then rescanned and considered negative if no further ‘hot spots’ are identified.

A chief concern in the adoption of the sentinel lymph node concept is the rate of false negatives. This apprehension is well founded, as unidentified lymphatic metastases, left untreated, present an increased risk of recurrence and consequently an increased morbidity of attempted salvage, when salvage is possible. However, with strict definitions and criteria for the implementation of sentinel lymph node mapping, false-negative rates have proven to be minimal.

Whereas computed tomography (CT) and magnetic resonance imaging (MRI) may be used to identify grossly metastatic lymph nodes, these techniques are typically not able to identify lymph nodes in patients with anything more than macroscopic disease. Meanwhile, positron emission tomography (PET) is limited to identifying metastatic deposits greater than 6 mm in diameter. As such, smaller tumour deposits are typically associated with negative PET scans. Lymph nodes that appear to be grossly abnormal should be removed, whether ‘hot’, blue, or not, as the lymphatic channels may be obstructed by tumour, and the lymphatic drainage and tracer may be bypassing such nodes. In these cases the sentinel lymph node may be falsely recognised as an alternative node, in which no evidence of metastases would lead to an inappropriate designation of ‘false negative’.

As the cervical lymphatics drain bilaterally,2 sentinel lymph nodes must be detected and interpreted per hemipelvis in order to be considered acceptable as an isolated procedure. If no sentinel lymph node is identified, a complete pelvic lymph node dissection must be performed on that side, as ipsilateral sentinel lymph node status does not represent nodal status in the contralateral hemipelvis. A negative sentinel lymph node contralateral to a metastatic lymph node is not a ‘false negative’. Neither does the detection of one sentinel lymph node per patient represent a detection rate of 100%. Rather, false negatives and detection rates must be interpreted per side in order to be meaningful in terms of prognostication and patient outcome.

These concepts form the basis for understanding the detection rate, sensitivity and negative predictive value of sentinel lymph node mapping. When previous studies are revisited (Table 1), detection rates, when interpreted per side, are lower, whereas false negative rates, considering each hemipelvis in isolation, are improved. As such, both sensitivity and negative predictive values approach 100%. These ideas have recently been compiled in an algorithm to optimise the detection rate and sensitivity of the procedure.8 Sentinel lymph node mapping has the potential for enhanced sensitivity in the detection of metastatic disease. Ultrastaging is the process by which sentinel lymph nodes are examined for metastases. This process of serial sectioning is time consuming and costly, and is not typically feasible for all lymph nodes obtained through a complete pelvic lymphadenectomy. The examination of sentinel lymph nodes is ideal, however, as ultrastaging is required for only one or two lymph nodes per hemipelvis. Furthermore, it has been found that in up to 80% of cases the sentinel lymph node is the only lymph node harbouring metastatic disease.6 Serial sectioning and the use of immunohistochemical staining may result in a higher detection of micrometastases (tumours of 0.2–2 mm in diameter) as well as isolated tumour cells (<0.2 mm in diamater), which are rarely identified using conventional pathology sections and staining.9 In our experience, the odds of detecting lymph node metastases are increased 2.8-fold through sentinel lymph node mapping.5

Table 1.   Published experience of sentinel lymph node mapping in cervical cancer
AuthorYearNumber of ptsTechniqueDR of SLN per patient, %DR of SLN per hemipelvis, %Number of metastases, (%)Sensitivity*, %NPV*, %
  1. B, blue dye; DR, detection rate; N/A, not available; NPV, negative predictive value; SLN, sentinel lymph node; Tc, radiolabeled technetium.

  2. *Sensitivity and NPV not available when laterality of SLN and ‘false negative’ not stated.

Echt199913B15 (2/13)N/A3 (23)100 (3/3)100 (10/10)
Dargent200035BNS86 (59/69)8 (23)100 (8/8)100 (27/27)
Verheijen200010Tc + B80 (8/10)60 (12/20)1 (10)100 (1/1)100 (9/9)
O’Boyle200020B60 (12/20)43 (17/40)4 (20)100 (4/4)100 (16/16)
Lantzsch200114Tc93 (13/14)64 (18/28)1 (7)100 (1/1)100 (13/13)
Malur200150Tc + B, Tc, B78 (39/50)N/A10 (20)N/AN/A
Rhim200226Tc + BN/AN/A5 (19)N/AN/A
Levenback200239Tc + B100 (39/39)97 (76/78)8 (21)88 (7/8)97 (31/32)
Plante200370Tc + B, B87 (61/70)74 (103/140)12 (17)100 (12/12)100 (58/58)
Chung200326Tc + B100 (26/26)60 (31/52)6 (23)N/AN/A
Barranger200313Tc + B92 (12/13)58 (15/26)2 (15)100 (2/2)100 (11/11)
Buist200325Tc + B100 (25/25)94 (47/50)10 (40)90 (9/10)94 (15/16)
Lambaudie200312Tc + B92 (11/12)88 (21/24)3 (25)100 (3/3)100 (9/9)
Van Dam200325Tc84 (21/25)5427/505 (20)100 (5/5)100 (20/20)
Marchiole200429B100 (29/29)95 (55/58)8 (28)N/AN/A
Pijpers200434Tc + B76 (26/34)74 (50/68)12 (35)N/AN/A
Li200428Tc96 (27/28)82 (46/56)6 (21)100 (6/6)100 (22/22)
Holub20047B100 (7/7)86 (12/14)1 (14)100 (1/1)100 (6/6)
Martinez-Palones200425Tc + B100 (25/25)82 (41/50)3 (12)100 (3/3)100 (22/22)
Niikura200420Tc + B90 (18/20)82 (33/40)3 (15)100 (3/3)100 (17/17)
Rob2005100B80 (80/100)71 (142/200)20 (20)100 (20/20)100 (80/80)
Rob200583Tc + B96 (80/83)93 (155/166)15 (18)100 (15/15)100 (68/68)
Roca200540Tc + B100 (40/40)N/A4 (10)100 (4/4)100 (36/36)
Lin200530Tc100 (30/30)N/A7 (23)100 (7/7)100 (23/23)
Silva200556Tc93 (52/56)65 (73/112)17 (30)100 (17/17)100 (39/39)
Angioli200535Tc74 (26/35)48 (34/70)6 (17)100 (6/6)100 (29/29)
Di Stefano200550B90 (45/50)72 (72/100)10 (20)90 (9/10)98 (40/41)
Wydra2006100Tc + B84 (84/100)75 (150/200)22 (22)100 (22/22)100 (78/78)
Hauspy200739Tc + B, Tc, B98 (38/39)85 (66/79)3 (8)100 (3/3)100 (36/36)
Altgassen2008590Tc + B, Tc, B90 (529/590)N/A106 (18)N/AN/A
Ogawa201082Tc88 (72/82)77 (126/164)15 (6)100 (15/15)100 (67/67)
Darlin2010105Tc90 (94/105)74 (156/210)18 (17)94 (17/18)99 (87/88)
Du201168Tc94 (64/68)68 (92/136)8 (12)100 (8/8)100 (60/60)
Diaz201181Tc + B95 (77/81)83 (135/162)26 (32)N/AN/A
Roy2011211Tc + B99 (209/211)92 (390/422)33 (16)100 (33/33)100 (178/178)
Lecuru2011139Tc + B98 (136/139)86 (240/278)26 (19)96 (25/26)99 (113/114)
Total 2333    98.2 (269/274)99.6 (1220/1225)

Despite the absence of lymph node involvement and clear surgical margins, pelvic recurrences occur in up to 15% of lymph node-negative patients.10 Theoretically, these patients may have had unidentified lymph node micrometastases, thereby explaining the increased risk of recurrence noted particularly in patients with deep invasion and lymph vascular space involvement. Horn et al.11 and Fregnani et al.12 demonstrated that patients with lymph node micrometastases experience survival rates between those with macroscopic lymph node metastases and negative lymph nodes.

Given the increased risk of recurrence, and decrease in disease-free status and overall survival, many authors believe that patients with micrometastatic disease should be offered adjuvant treatment. If the treatment of micrometastatic disease is effective, the improved detection of micrometastases using sentinel lymph mapping and ultrastaging could lead to a survival advantage.

Just as radical trachelectomy is not appropriate for all patients who desire fertility preservation, sentinel lymph node mapping may not be possible, nor appropriate, for all patients. With further experience in the technique and outcome data, further defined selection criteria will probably emerge. Sentinel lymph node detection seems to be dependent on FIGO stage and primary cervical tumour size.7 Studies have demonstrated that sentinel lymph nodes are less likely to be identified if the tumour size is greater than 2 cm.7 There may be limited tracer uptake because of the obstruction of cervical stromal lymphatic channels or because of the complete involvement of the sentinel lymph nodes by tumour. Furthermore, the tracer must be injected into normal cervical stroma, thereby precluding large tumours replacing the exocervix. The use of neo-adjuvant chemotherapy may also decrease sentinel lymph node detection. Prior cervical conisation may distort the cervix, thereby complicating tracer injection; however, overall it has not been clearly associated with inferior rates of sentinel lymph node detection.

Not all sentinel lymph nodes are readily accessible. Difficulty is encountered in the detection of sentinel lymph nodes in the parametrium. They are not identified on lymphoscintigram because they are too close to the point of tracer injection, and they are difficult to identify intraoperatively by the visualisation of blue dye or by the use of the gamma probe because of the increased intensity of signal close to the cervix. Parametrial lymph nodes may contain metastatic disease.6 The identification of these lymph nodes is probably immaterial, however, as the sentinel lymph node procedure is designed to replace pelvic sidewall lymphadenectomy, whereas the parametrial lymph nodes are excised with the surgical specimen at the time of radical hysterectomy or trachelectomy. Furthermore, the incidence of metastatic parametrial lymph nodes with negative pelvic sidewall nodes is rare. If sentinel lymph node status reflects parametrial node status, sentinel lymph node mapping and frozen sectioning could be used to influence the degree of parametrial resection.

Traditional pelvic lymphadenectomy is used to guide adjuvant treatment. However, the interpretation of frozen section analysis of sentinel lymph nodes may be used to guide intraoperative management. For some oncologists, whether to proceed with radical resection, or to defer in favour of chemo-radiotherapy is dependent on the nodal status at the time of surgery. However, ‘false-negative’ frozen sections do occur,5,6 usually in otherwise normal-appearing lymph nodes. Ultimately, the accuracy of the frozen section analysis will determine whether patients are spared a second operation for the complete lymphadenectomy for pelvic and para-aortic nodal status, or whether they have to undergo multimodality treatment when lymph node metastases are detected only on final pathology.

Adverse events are relatively uncommon in sentinel lymph node mapping, although the injection of blue dye may result in anaphylaxis with vasomotor collapse on rare occasions. Pre-medication with antihistamines may decrease the risk of allergic reaction; however, it has been suggested that lymphoscintigraphy should be the technique of choice because of its superior safety profile.6 Alternatively, the use of blue dye can be minimized by only using it for patients in which lymphoscintigraphy has been unsuccessful in identifying bilateral sentinel lymph nodes, as is the protocol at our centre.

Proficiency in the execution of sentinel lymph node mapping increases with experience. In breast cancer and melanoma, between 20 and 30 cases should be performed in order to achieve competence. In vulva, ten cases may be sufficient, whereas in cervical cancer 30 cases may be necessary, as intracervical injection is technically challenging and the lymphatic drainage is more complex.6

Although we embrace and endorse the sentinel lymph node concept, further research is needed. Prospective, observational, multicentre studies are important for many clinicians to confirm the safety of omitting complete lymph node dissections in patients with negative sentinel lymph nodes. The role of sentinel lymph node mapping as a pre-therapeutic work-up versus an intraoperative technique and the usefulness of SPECT/CT lymphoscintigraphy and molecular quantification remain to be determined. Uncertainties remain regarding the ideal size of radiocolloid, and the false-positive rate of immunohistochemistry and cytokeratins. Multiple studies are in progress.

In our experience, with the combined use of tracers and ultrastaging, the sensitivity of sentinel lymph node mapping is superior to complete lymph node dissections.5 In due course, as has been seen in the staging of breast cancer, malignant melanoma, and vulvar cancer, we anticipate that sentinel lymph node mapping will become the standard of care, replacing pelvic lymphadenectomy in the surgical management of early-stage cervical cancer.

Disclosure of interests

The authors have no conflicts of interest and nothing to disclose.

Contribution to authorship

AC conceived and designed the article. AC and LE collected and assembled the data, and discussed and interpreted the findings. LE composed the initial manuscripts. AC and LE revised and approved the final article.

Details of ethics approval

No ethical approval was required for this commentary.

Funding

None.

Acknowledgements

None.

Ancillary