| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Tristán Associates, 4518 Union Deposit Road, Harrisburg, PA 17111
Editor:
The article by Dr Liberman and colleagues in the June 1999 issue of Radiology (1) regarding sentinel lymph node biopsy in breast carcinoma was interesting and informative. However, I have concerns about how patients are being treated after these biopsies are complete and the results are known. At Dr Liberman's institution and many others, the protocol is to perform sentinel node biopsy in carefully selected patients; a conventional surgical lymph node dissection is subsequently performed when the sentinel node biopsy result is positive and is not performed when the sentinel node is found to be free of metastases. This approach strikes me as counterintuitive.
There is considerable debate about the independent value of lymph node dissection in prolonging life (2). There is no debate that it carries considerable potential for morbidity.
In Dr Liberman and colleagues' series, sentinel nodes were found at surgery in 30 women. The sentinel node was positive in seven (23%). It is presumed that all of these patients were going to undergo chemotherapy anyway once the sentinel node proved positive. In six of these seven patients (86%), no tumor was found at conventional surgical axillary node dissection; these patients gained no benefit from the axillary node dissection. In the other patient, tumor was found in the axillary nodes and in the sentinel node; whether this patient derived any benefit from the surgical dissection is arguable.
In the 23 patients (77%) with negative sentinel nodes, conventional surgical axillary node dissection was not performed. The negative predictive value is the relevant number to use in considering what to do with these patients and represents how often a negative sentinel node biopsy result is correct: TN/(FN + TN), where TN is true-negative findings and FN is false-negative findings. This particular study did not address the issue of the negative predictive value. To do so would have required surgical axillary node dissection in every patient with a negative sentinel node biopsy result. However, Dr Liberman and colleagues' study immediately followed another study performed at the same institution by O'Hea et al (3), in which a negative predictive value of 92% was found. Stated another way, 8% of patients with negative sentinel nodes in fact had axillary node metastases at surgery.
If the results of the study by O'Hea et al are applicable to those of the study by Dr Liberman and colleagues (and they ought to be, given that the two studies were conducted consecutively at the same institution and shared five authors), then statistically two of the 23 women (8%) in Dr Liberman's study who had a negative sentinel node biopsy result in fact have axillary node metastases and are unaware of it.
A negative predictive value less than 100% is the Achilles' heel of this procedure—how strongly do we rely on a negative sentinel node? By definition, there never will be a false-positive finding, and those patients for whom there were technical failures or a failure to identify any sentinel node at least carry with them the safety net of a surgical node dissection.
I believe the sentinel node biopsy protocol should be revised so that a negative sentinel node biopsy result leads to a surgical axillary node dissection, until we get the false-negative rate of sentinel node biopsy down to 0%–1%. Furthermore, all women with positive sentinel nodes should forego standard axillary node dissection and proceed directly to chemotherapy until clear evidence exists that surgical dissection itself yields prolonged survival or at least disease-free survival.
All of this raises a philosophical question for us as radiologists: What role do we play in the use of the information we provide with our studies? All my concerns about the article by Dr Liberman and colleagues focus not on imaging but on issues outside the realm of radiology and under the control of surgical and medical oncologists. While I am glad that the oncology community has quickly embraced the merits of the sentinel node biopsy procedure, I have concerns that they are using the information improperly. Having 8% of women with negative sentinel node biopsy results forego chemotherapy because they do not know they have regional metastatic disease is far too high a number for my comfort.
Let's put it personally—if your wife, sister, mother, daughter, or significant other had a small infiltrating cancer and a negative sentinel node biopsy result, would you advise her to forego chemotherapy, knowing that there is an 8% chance that she has unsuspected axillary metastatic disease? I'm not sure I would.
References
Breast Imaging Section, Department of Radiology,* and Breast Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021
We thank Dr Guenin for his interest in our article (1). We would like to address his concerns.
Our study was not a validation study of sentinel lymph node biopsy with correlative axillary dissection but an assessment of the implementation of sentinel node biopsy in women with percutaneously proved, nonpalpable breast cancer. The validation of sentinel node biopsy was provided in the 18 published articles on studies correlating the results of sentinel lymph node biopsy with axillary dissection in more than 2,000 women, with a sensitivity of 94%, specificity of 100%, positive predictive value of 100%, negative predictive value of 96%, and accuracy of 98% (1).
Dr Guenin disagrees with our protocol to forego axillary dissection if the sentinel node is negative. He argues that a negative sentinel lymph node biopsy result should lead to a surgical axillary dissection until we get the false-negative rate of sentinel node biopsy down to 0%–1%. While a 0%–1% false-negative rate for sentinel node biopsy certainly is desirable, it is not likely to happen. The false-negative rate is the proportion of women with axillary metastases in whom the sentinel node is free of tumor. In the 18 published articles on validation studies of sentinel node biopsy, the overall false-negative rate of sentinel node biopsy was 6% (range, 0%–15%) (1).
Consider the implications of a 6% false-negative rate of sentinel lymph node biopsy. The expected frequency of axillary metastases in stage T1 breast cancer is 25%–30%. Assume 100 women have technically successful sentinel lymph node biopsy, of whom 30 (30%) have axillary metastases. If the false-negative rate of sentinel lymph node biopsy is 6%, the sentinel node will be false-negative in two (6% of 30) women, or 2% of the 100 women who underwent the sentinel node biopsy procedure. The sentinel node biopsy result would be true-positive in 28 (28%) and true-negative in 70 (70%); the latter 70 women would be spared the axillary lymph node dissection, with its risk of numbness, scarring, and lymphedema. A woman may reasonably choose to accept a 2% risk of having a false-negative sentinel node to have a 70% chance of being spared the potential morbidity of axillary dissection.
There are several factors that can decrease the false-negative rate of sentinel node biopsy, and that probably made our false-negative rate lower than that reported by O'Hea et al (2) in an earlier study conducted at our hospital. First, the false-negative rate decreases with experience. At our institution, the false-negative rate in 104 cases with planned "back-up" axillary dissection was 10.6% (3). When the first six cases of every surgeon were excluded, the false-negative rate decreased to 5%; eliminating the first 15 cases of each surgeon reduced the false-negative rate to 2% (3). The study by O'Hea et al included our first 60 sentinel node biopsy cases (all performed with back-up axillary dissection) and represents our learning curve. By the time our article (1) was submitted, we had conducted over 500 sentinel node biopsies; we have now conducted more than 1,500.
Second, the false-negative rate can be lowered by using immunohistochemical analysis of sentinel nodes (4). Immunohistochemistry was used in our study but not in the early experience reported by O'Hea et al (2).
Third, the false-negative rate is lowest in the smallest tumors (2,3,5,6). In an analysis of 104 cases with planned back-up axillary dissection conducted at our hospital, there were no false-negative findings among stage T1a or T1b cancers (3). Our study consisted exclusively of stage T1 cancers, while the study by O'Hea et al (2) also included stage T2 lesions and stage T3 lesions.
The effect of a false-negative sentinel node is diminished by the use of current treatment protocols. At our institution, chemotherapy is given for node-positive breast cancers and for all node-negative breast cancers measuring 1 cm or larger. The only scenario in which a false-negative sentinel node would lead a woman to forego chemotherapy is if the sentinel node is falsely negative in a cancer smaller than 1 cm. We have had no false-negative sentinel nodes in cancers smaller than 1 cm that had planned back-up axillary dissection (3), nor have others (5,6). It is therefore likely that of the few false-negative sentinel nodes encountered, most will occur in women with tumors larger than 1 cm who will still receive chemotherapy.
Axillary dissection, the standard for histologic analysis of the axilla, has a false-negative rate higher than the 0%–1% suggested by Dr Guenin. In retrospective studies in women with lymph nodes interpreted as negative at conventional axillary dissection, careful repeat analysis with the use of serial sections, immunohistochemistry, or both disclosed metastases in an average of 15%–20% of cases; there is evidence that these metastases were associated with significantly lower survival rates (7). Histologic analysis of lymph nodes removed at axillary dissection is often limited to one or a small number of hematoxylin-eosin–stained sections; such an approach can result in missed axillary metastases. Sentinel lymph node biopsy allows the pathologist to focus more intensive histologic analysis on the few nodes most likely to contain tumor. For this reason, the false-negative rate of sentinel node biopsy with immunohistochemistry, serial sections, or both actually may be lower than the false-negative rate of conventional axillary dissection (8).
Dr Guenin also takes issue with our protocol to perform axillary dissection if tumor is found in sentinel nodes. The 18 published articles on validation studies of sentinel node biopsy reviewed in our article demonstrated that approximately half of women with tumor in sentinel nodes also have metastases in other axillary nodes (1). Relying on chemotherapy to sterilize the nodes in these women is an experimental approach that is not yet proved to be effective. The appropriateness of performing sentinel node biopsy without axillary dissection in this setting will be evaluated in an upcoming National Surgical Adjuvant Breast and Bowel Project trial that will randomly assign women to sentinel node biopsy with or without axillary dissection.
For all patients, we advocate use of the procedure with the highest accuracy and the lowest morbidity. The rapidly accruing evidence suggests that for women with small infiltrating breast carcinomas and clinically negative nodes, that procedure may be sentinel lymph node biopsy. Further work is necessary to optimize techniques for sentinel node identification and analysis and to assess long-term outcome.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
