Annals of Surgical Oncology Sign the Guestbook
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
10.1245/ASO.2006.03.078
Annals of Surgical Oncology 13:310-320 (2006)
© 2006 Society of Surgical Oncology
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lambert, L. A.
Right arrow Articles by Meric-Bernstam, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lambert, L. A.
Right arrow Articles by Meric-Bernstam, F.

Original Article

Validation of a Breast Cancer Nomogram for Predicting Nonsentinel Lymph Node Metastases After a Positive Sentinel Node Biopsy

Laura A. Lambert, MD1, Gregory D. Ayers, MS2, Rosa F. Hwang, MD1, Kelly K. Hunt, MD1, Merrick I. Ross, MD1, Henry M. Kuerer, MD, PhD1, S. Eva Singletary, MD1, Gildy V. Babiera, MD1, Frederick C. Ames, MD1, Barry Feig, MD1, Anthony Lucci, MD1, Savitri Krishnamurthy, MD3 and Funda Meric-Bernstam, MD1

1 Department of Surgical Oncology, Unit 444, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas 77030
2 Department of Biostatistics and Applied Math, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas 77030
3 Department of Pathology, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Texas 77030

Correspondence: Address correspondence and reprint requests to: Funda Meric-Bernstam, MD; E-mail: fmeric{at}mdanderson.org.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: Although completion lymph node dissection (CLND) is the standard of care for breast cancer patients with sentinel lymph node (SLN) metastases, the SLN is the only node with tumor in 40% to 60% of cases. To assist with decision-making regarding CLND, investigators at Memorial Sloan-Kettering Cancer Center devised and validated a nomogram for predicting the likelihood of non-SLN metastases. To assess the generalizable use of this nomogram, validation analysis was performed by using an external database.

Methods: Eight clinicopathologic variables for 200 consecutive breast cancer patients at the University of Texas M. D. Anderson Cancer Center with SLN metastases and CLND were entered into the nomogram. The accuracy of the nomogram to predict non-SLN metastases was assessed by the receiver operating characteristic (ROC) curve and linear regression analysis. The accuracy of the nomogram with touch-imprint cytology (TIC) as a substitute variable for frozen section was also evaluated.

Results: The linear correlation coefficient of the nomogram-predicted probabilities correlated with the observed incidence of non-SLN metastases for all patients (.97). The accuracy of the nomogram as measured by the area under the ROC curve was .71. When applied solely to patients who had TIC assessment of the SLN, the area under the ROC curve was .74.

Conclusions: This study validated the Memorial Sloan-Kettering Cancer Center breast cancer nomogram by using an external database. TIC seems to be an acceptable substitute for frozen section as a nomogram variable. The nomogram may help predict an individual’s risk of non-SLN metastases and assist in patient decision making regarding the benefit of CLND.

Key Words: Sentinel node • Prediction • Nomogram • Breast cancer • Axillary lymph node dissection


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Sentinel lymph node (SLN) biopsy has been shown by many studies to be a valid method of assessing regional lymph node involvement in breast cancer patients.1–5 Because of the low risk of unrecognized regional lymph node metastases for patients in whom the SLN is histopathologically negative, many institutions no longer perform completion lymph node dissection (CLND) for these patients.6,7 For patients in whom the SLN is histopathologically positive, CLND remains the current standard of care. However, the SLN has been shown to be the only positive lymph node in 40% to 60% of these patients.8 For these patients, CLND offers no additional diagnostic, prognostic, or therapeutic benefit and carries a significant risk of morbidity. Furthermore, even for patients in whom there are residual regional lymph node metastases, the additional benefit of CLND is questioned because these patients will generally receive systemic therapy based on the presence of SLN metastases.

Patients and physicians dealing with the diagnosis of an SLN metastasis must weigh a patient’s risk of non-SLN metastases with the benefit of CLND to decide whether a patient should undergo CLND. To assist this decision-making process, by using their large institutional database and a multivariate logistic regression analysis, Van Zee et al.9 at the Memorial Sloan-Kettering Cancer Center (MSKCC) devised a nomogram to estimate an individual patient’s risk of non-SLN metastases after a positive SLN biopsy result (Fig. 1Go). Formulated on eight readily available histopathologic factors related to the primary tumor (pathologic size, histological type, nuclear grade, presence or absence of lymphovascular invasion [LVI], multifocality, and estrogen receptor status), the SLN (number of positive and negative SLNs), and the method of detection of the SLN metastasis (including intraoperative frozen section [FS] analysis), this nomogram has been prospectively validated by MSKCC. However, a recent study by Kocsis et al.10 who used an independent external database, failed to validate the nomogram, and the authors therefore warned against its use. Thus, although the data from MSKCC are promising, it remains unknown whether the nomogram is useful for predicting the risk of non-SLN metastases in patients at other institutions who have positive SLN biopsy findings.


Figure 1
View larger version (22K):
[in this window]
[in a new window]
  		FIG. 1. Nomogram devised by investigators at the Memorial Sloan-Kettering Cancer Center to predict the likelihood of additional non–sentinel lymph node (SLN) metastases in a patient with a positive SLN. NUCGRADE, tumor type and nuclear grade (ductal, nuclear grade I; ductal, nuclear grade II; ductal, nuclear grade III; lobular); LVI, lymphovascular invasion; MULTIFOCAL, multifocality of primary tumor; ER, estrogen receptor status; NUMNEGSLN, number of negative SLNs; NUMSLNPOS, number of positive SLNs; PATHSIZE, pathologic size in centimeters; METHDETECT, method of detection of SLN metastases (frozen section, routine hematoxylin and eosin [HE], serial HE, or immunohistochemistry [IHC]). The first row (Points) is the point assignment for each variable. Rows 2 to 9 represent the variables included in the model. For an individual patient, each variable is assigned a point value (uppermost scale, Points) on the basis of the histopathologic characteristics. A vertical line is made between the appropriate variable value and the Points line. The assigned points for all eight variables are summed, and the total is found in row 10 (Total Points). Once the total is located, a vertical line is made between Total Points and the final row, row 11 (Predicted Probability of + Non-SLN). (Reprinted with permission from Van Zee et al.9 Copyright © 2003 The Society of Surgical Oncology, Inc.)

 
The aims of this study were 2-fold: (1) to determine whether we could validate the nomogram by using a large, external, and independent cancer center database and (2) to assess the substitutability of touch-imprint cytology (TIC) for frozen section (FS) analysis as a nomogram variable.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
In accordance with the institutional review board, we retrospectively reviewed a prospective database at the University of Texas M. D. Anderson Cancer Center that includes all patients who have undergone intraoperative lymphatic mapping and SLN biopsy for invasive breast cancer. Between January 1, 1993, and June 30, 2003, we identified 200 consecutive patients with invasive breast cancer who had a clinically negative axilla by physical examination, a positive SLN shown by TIC, hematoxylin and eosin (H&E) staining, or immunohistochemistry (IHC) analysis for cytokeratin and who subsequently underwent CLND. Patients with a positive fine-needle biopsy of an axillary lymph node before SLN biopsy were excluded. Intraoperative assessment of the SLN was performed with TIC in 94 cases (47%).

Technique of Intraoperative Lymphatic Mapping and SLN Biopsy
SLN biopsy was performed as described by Breslin et al.11 Briefly, intraoperative lymphatic mapping was performed with peritumoral injection of blue dye alone, 99mTc-labeled sulfur colloid alone, or a combination of the two. When lymphoscintigraphy was performed, patients received filtered 99mTc-labeled sulfur colloid, which was injected into the breast parenchyma surrounding the tumor or the biopsy cavity on the day of surgery (.5 mCi) or the day before surgery (2.5 mCi). When the tumor was not palpable, the injection was performed with mammographic or sonographic guidance. On the day of surgery, 5 mL of 1% isosulfan blue (Lymphazurin; US Surgical, Norwalk, CT) was injected into the breast peritumorally, and the breast was massaged for 5 minutes while the patient was under general anesthesia. For patients who received sulfur colloid injection, a handheld gamma detection probe (NeoProbe 2000; US Surgical) was used to scan the axilla transcutaneously and identify the most radioactive area. Through an axillary incision over this area, SLNs were identified as nodes with blue dye uptake, radiotracer uptake, or both.

During the SLN biopsy technique learning phase, each surgeon in this study performed immediate level I and II CLND for the first 30 SLN biopsies. After this teaching period, CLND was performed only when the SLN was found to be positive on pathologic examination. In addition, if the SLN was confirmed to be positive at the time of the SLN biopsy, CLND was performed during the same operation.

Pathologic Evaluation
Pathologic evaluation of SLNs at the M. D. Anderson Cancer Center has evolved with changes in technology. Before April 2000, SLNs were serially sectioned along the short axis at 2- to 3-mm intervals; sections were embedded in paraffin blocks, and one level from each block was stained with H&E. Beginning in April 2000, SLNs were grossly processed in the same manner as before, and each paraffin block was then serially sectioned at 5-µm intervals to produce 10 levels. Levels 1, 2, and 4 were evaluated by routine H&E staining, and level 3 was analyzed for cytokeratin by IHC.12 Non-SLNs from the CLND specimen were analyzed by routine H&E staining only.

Each primary tumor was evaluated in terms of the size of the invasive component, the histological type, Black’s nuclear grade, estrogen and progesterone receptor status, HER-2/neu status, and the presence of LVI. For hormonal receptor status, >10% staining of the cells by IHC was considered positive.

TIC was performed as described by Lee et al.13 Briefly, as soon as the SLNs were received in the FS suite, each lymph node was serially sectioned into 1- to 2-mm slices perpendicular to or along the lymph node’s long axis. Touch imprints were made of each cut surface. The imprints were either alcohol fixed (95%) and stained by using the Papanicolaou technique or air-dried and stained with Diff Quik (Anapath, Lewisville, TX), as determined by the pathologist. All lymph node sections were then fixed in formalin and embedded in paraffin for routine histological evaluation.

Nomogram Analysis
The nomogram produced by MSKCC for predicting the likelihood of non-SLN metastases in breast cancer patients with a positive SLN was downloaded from the institution’s Web site (http://www.mskcc.org/nomograms). The eight data variables required by the nomogram include the pathologic size of the primary tumor in centimeters, tumor type and nuclear grade (ductal with grade I to III or lobular), the number of positive SLNs, the number of negative SLNs, the method of detection of the SLN (FS, routine H&E, serial H&E, or IHC), estrogen receptor status, multifocality, and LVI. In this study, TIC was substituted for FS for the intraoperative assessment of the SLNs. The nomogram was used retrospectively for the 200 patients with positive SLN biopsies who underwent CLND during the study period. Nonductal/nonlobular tumor types were lumped together with ductal cancers of similar grade. For mixed lobular and ductal cancers, the predominant histology (as determined by highest estimated percentage) was accepted as the histological type. When LVI or estrogen receptor status was unknown, it was considered to be negative during the initial calculation of nomogram scores and validation analysis. Validation was then later performed on the basis of positive LVI and estrogen receptor status. For patients who received preoperative chemotherapy, the final pathologic tumor size was regarded as the primary tumor size.

The predicted risk of non-SLN metastases was calculated for each patient by using the nomogram. The discrimination of the nomogram was assessed by calculating the area under the receiver operating characteristic (ROC) curve by the method of Hanley and McNeil.14 The ROC curve assesses the relationship between the sensitivity and the false-positive rate (1 – specificity) of a test across all possible threshold values that define the positivity of a disease or condition. The area under the curve (AUC) is a summary measure of the ROC that in a single statistic summarizes the inherent capacity of a test to discriminate a diseased from a nondiseased subject across all possible levels of positivity. It is generally accepted that AUC values of .7 to .8 represent reasonable discrimination, whereas AUC values exceeding .8 represent good discrimination. Ninety-five percent confidence regions constructed from 95% confidence intervals for sensitivity and 1 – specificity at predicted probability levels of 10%, 15%, 25%, and 50% were calculated to determine the significance of the ROC curve relative to the AUC value of .5 (the equivalent of the results having no discriminatory power). Logistic regression was used to estimate the increased odds of a non-SLN metastasis per 10-unit increase in the prediction by nomogram.

The patients were then divided into 10 groups by predicted percentage risk (0%–10%, 11%–20%, 21%–30%, 31%–40%, 41%–50%, 51%–60%, 61%–70%, 71%–80%, 81%–90%, and 91%–100%). The mean predicted percentage risk of non-SLN metastases and the percentage of observed patients with non-SLN metastases were calculated for each group. The observed probability of non-SLN metastases per group was then compared with the average nomogram-predicted probability of non-SLN metastases per group. To assess the validity of substituting TIC for FS in the nomogram, similar analyses were applied to the subset of patients who underwent TIC.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Descriptive characteristics of the total study population and of the patients who underwent TIC for intraoperative assessment are listed in Table 1Go. Table 2Go shows the incidence of non-SLN metastases by primary tumor and SLN pathologic characteristics for the entire study group. Sixty-four (32%) of the 200 patients had non-SLN metastases. The overall predictive accuracy of the nomogram, as measured by the AUC, was .71 (Fig. 2Go). The AUC can be interpreted as the probability of a correct assignment of disease presence in random pairs of patients, one patient in each pair who actually has disease and one who does not. On average, the probability of correct assignments would be .5 by flipping a coin to make each assignment. By comparison, the probability of correct disease assignment by the nomogram was .21 higher than would be expected from random chance alone. Ninety-five percent confidence regions at predicted probabilities of 10%, 15%, 25%, and 50% showed the ROC to be statistically significantly different from the AUC of .5 (P = .05). Logistic regression analysis comparing the nomogram-predicted probability and the actual probability of non-SLN involvement for each patient demonstrated an odds ratio of 1.53 (95% confidence interval, 1.31–1.81; P ≤ .0001). The sensitivity of the nomogram in identifying patients with non-SLN metastases at threshold-predicted probabilities of 10%, 15%, and 50% was 94%, 80%, and 32%, respectively. The specificity of the nomogram in identifying patients with non-SLN metastases at threshold-predicted probabilities of 10%, 15%, and 50% was 26%, 45%, and 93%, respectively.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Descriptive characteristics of the patient population
 

View this table:
[in this window]
[in a new window]
  		TABLE 2. Incidence of non-SLN metastases by the pathologic characteristics of the primary tumor and SLN
 

Figure 2
View larger version (13K):
[in this window]
[in a new window]
  		FIG. 2. Receiver operating characteristic curve with 95% confidence regions at 10%, 15%, 25%, and 50% for the entire study population (n = 200). The vertical axis represents the sensitivity of each nomogram score, and the horizontal axis represents the false-positive rate (1 – specificity) of each nomogram score. The 45° line reflects the characteristics of a test with no discriminating power. The area under the curve (AUC) summarizes the trade-off between the sensitivity and 1 – specificity of each nomogram score.

 
The patients were divided into 10 groups by predicted percentage risk (e.g., 0%–10% and 11%–20%). The observed probability of non-SLN metastases per risk group is shown in Fig. 3Go. The mean predicted percentage risk of non-SLN metastases for each risk group was highly correlated with the observed risk of non-SLN metastases (r = .97; P < .0001).


Figure 3
View larger version (16K):
[in this window]
[in a new window]
  		FIG. 3. Bar graph depicting the relationship between the observed percentage of non–sentinel lymph node (SLN) metastases (vertical axis) and the average predicted probability of non-SLN metastases per nomogram-calculated risk group (horizontal axis; n = number of patients per nomogram-calculated risk group) for the entire study population (n = 200).

 
LVI was unknown for 35 patients, and estrogen receptor status was unknown in 4 patients. In the analysis of patients with unknown LVI or estrogen receptor status, no significant difference was seen in the discrimination of the nomogram whether the status was considered positive or negative (data not shown).

Thirty (32%) of the 94 patients who underwent intraoperative SLN assessment with TIC had non-SLN metastases (Table 2Go). The overall predictive accuracy of the nomogram for patients undergoing TIC as measured by the AUC was .74 (Fig. 4Go). Logistical regression analysis comparing the nomogram predicted probability and the observed probability of non-SLN metastases for each patient demonstrated an odds ratio of 1.58 (95% confidence interval, 1.26–2.05; P = .0002). The sensitivity of the nomogram for TIC patients at predicted probabilities of 10%, 15%, and 50% was 97%, 87%, and 43%, respectively. The specificity of the nomogram for TIC patients at predicted probabilities of 10%, 15%, and 50% was 20%, 38%, and 91%, respectively.


Figure 4
View larger version (14K):
[in this window]
[in a new window]
  		FIG. 4. Receiver operating characteristic curve with 95% confidence regions at 10%, 15%, 25%, and 50% for the patients who had touch-imprint cytology assessment of the sentinel lymph node (n = 94). The vertical axis represents the sensitivity of each nomogram score, and the horizontal axis represents the false-positive rate (1 – specificity) of each nomogram score. The 45° line reflects the characteristics of a test with no discriminating power. The area under the curve (AUC) summarizes the trade-off between the sensitivity and 1 – specificity of each nomogram score.

 
The patients who underwent TIC were divided into 10 groups by predicted percentage risk. The observed probability of non-SLN metastases per risk group is shown in Fig. 5Go. The mean predicted percentage risk of non-SLN metastases for patients in each risk group was highly correlated with the observed risk of non-SLN metastases (r = .94; P < .001).


Figure 5
View larger version (14K):
[in this window]
[in a new window]
  		FIG. 5. Bar graph depicting the relationship between the observed incidence of non–sentinel lymph node (SLN) metastases (vertical axis) and the average predicted probability of non-SLN metastases per nomogram-calculated risk group (horizontal axis; n = number of patients per nomogram-calculated risk group) for patients who had touch-imprint cytology assessment of the SLN (n = 94).

 
Twenty-nine (15%) of the 200 patients received preoperative chemotherapy before undergoing SLN biopsy and CLND, and 14 (48%) of these patients had non-SLN metastases. Only two patients (7%) had a nomogram-predicted probability of <10%, and neither had non-SLN metastases. Six patients (21%) had a predicted probability of <20%, and only one (17%) of these patients had non-SLN metastases. For the patients treated with preoperative chemotherapy, the mean predicted percentage risk of non-SLN metastases in each risk group correlated with the observed risk (r = .92). There was no significant difference between the linear correlation coefficients of the entire study population either with or without the patients who received neoadjuvant chemotherapy (r = .97 vs. .95, respectively).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Several pathologic characteristics of both the primary tumor and its metastases influence the risk of non-SLNs metastases in breast cancer patients. The size of the primary tumor15–17 and the size of the largest SLN metastasis15–18 have been shown in numerous studies to be independent prognostic factors for non-SLN metastases. Other reported independent prognostic factors include LVI and the number of positive SLNs.17,19–21 The estrogen receptor status and the method of detection of the SLN have been suggested to influence the risk of non-SLN metastases.22,23 However, because of the variations in study methods, sample sizes, and study populations, there is no consensus within the literature on the true risk associated with each of these variables. Consequently, estimating an individual patient’s risk of non-SLN metastases on the basis of findings in the literature alone is very difficult.

The overall findings of this study support the predictive accuracy of the nomogram with a statistically significant ROC measurement of .71, a 10-unit odds ratio of 1.53, and a linear correlation coefficient of .97. Some loss of discrimination of the nomogram for the patients in our database relative to that seen by Van Zee et al.9 may be related to the fact that some of the patients with positive SLN biopsies at our institution did not undergo CLND. For example, if these patients did not undergo CLND because they were considered to be at low risk for non-SLN metastases, this could contribute to the relative underpredicting of non-SLN metastases seen in our study among patients in the 11% to 30% predicted risk groups. Other possible causes include differences across institutions in the way the SLN is pathologically assessed, the frequency of application of various SLN-detection methods, the surgical approach to SLN mapping, and the methods of analysis used to validate the nomogram.

In the nomogram-validation study by Kocsis et al.10 which evaluated 140 patients who underwent SLN biopsy and CLND at the Bacs-Kiskun County Teaching Hospital at the University of Sciences at Szeged Medical School in Hungary, the researchers performed a linear regression analysis that compared the average nomogram-predicted probability of non-SLN metastases per patient decile with the observed percentage of non-SLN metastases per decile (r = .84). These results failed to validate the nomogram, and the researchers warned against its use. In addition, Kocsis et al. identified an increased incidence of non-SLN metastases in patients with the lowest predicted risk. The researchers theorized that because the nomogram uses the method of SLN detection as a surrogate for the size of SLN metastasis (with FS detecting the largest metastases and IHC alone detecting the smallest), differences in the pathologic assessment of the SLN were the likely cause of the discrepancy with the results of the investigators at MSKCC. Specifically, as proposed by Kocsis et al. thinner sections and more detailed sampling could reduce the number of metastases detected by IHC alone, thus potentially contributing to a loss of accuracy. Further study is needed to determine whether using the size of SLN metastases as a variable, rather than method of detection, would improve the interinstitution reproducibility of the accuracy of the nomogram.

The differences in the accuracy of the nomogram among institutions may also be related to variations in the frequency of the application of the methods of detection of SLN metastases. At MSKCC, FS was used in 354 (95%) of 373 patients, and this was also the most frequent method of detection of the SLN metastasis (73%). At the Bacs-Kiskun County Teaching Hospital, TIC was used in all 140 patients, and this was also the most frequent method of detection of SLN metastasis (53%). In our study, serial sectioning with H&E staining was the most frequent method of detection of SLN metastasis (50%), and this may account in part for the slightly lower discriminatory power of the nomogram in our study. Because routine (nonserial) H&E staining is not the first step in the processing of SLNs at our institution, nomogram scores for patients in our study whose SLN metastasis could have been detected with routine H&E staining will be artificially lower than nomogram scores that would be calculated at institutions that use routine H&E in the SLN processing.

Another significant difference among the institutions may be the surgical approach to SLN mapping. The appropriate number of SLNs to remove at the time of operation has not been clearly defined in the literature. Some groups advocate stopping after a certain number of SLNs are removed (e.g., three), whereas investigators from MSKCC have advocated removing all blue nodes and all hot nodes.24 Although the mean number of SLNs retrieved was not stated in the article by Kocsis et al.10 previous reports from the same institution have reported a mean of 1.3 to 1.4 SLNs and a median of 1 SLN removed,25 compared with a mean of 2.7 SLNs and a median of 2 SLNs removed in our series. Studies show that in 98% of node-positive patients who have multiple SLNs removed, metastasis is detected within the first three SLNs.24 One would expect that with fewer SLNs removed, there would be a corresponding increase in the number of positive non-SLNs, which may in part account for the underestimation of non-SLN metastases in the Kocsis et al. series.

Differences in the methods of analysis used to validate the nomogram may also contribute to the differences in the nomogram’s accuracy. Kocsis et al.10 performed a linear regression analysis on both their patient population and the patient population from the Van Zee et al.9 series by dividing patients into deciles based on nomogram scores (14 vs. 37 patients per decile) and comparing the average predicted percentage of non-SLN metastases per decile with the observed percentage of non-SLN metastases per decile (r = .84 vs. .97). Concerns about this approach include the difference in the number of patients per decile in the two studies and the random separation of patients with similar nomogram scores possibly resulting in loss of information. To avoid random separation of patients with similar nomogram scores, patients can be divided into groups on the basis of ranges of predicted probabilities; this was the method used in our study, and it is similar to the methodology used in other nomogram-validation studies.26,27

TIC is used over FS for intraoperative assessment of SLNs in our institution because all cut surfaces of the lymph node can be used for imprints and, most importantly, because there is better preservation of the histology of the SLN before permanent processing. Numerous studies have shown the sensitivity, specificity, and accuracy of TIC to be clinically acceptable for the intraoperative assessment of SLNs.13,28,29 In studies directly comparing FS with TIC, the sensitivities of the two methods are roughly equivalent (FS, 52%–91%; TIC, 62%–96%).30–32 As in our clinical practice, in the series analyzed by Kocsis et al.10 TIC was also used for intraoperative SLN assessment. Kocsis et al. stated that, because of the reported similar performance rates of these two methods of intraoperative SLN assessment, their failure to validate the nomogram was not because of their use of TIC. Our study supports the validity of TIC as a substitute for FS analysis; TIC had a statistically significant ROC of .74, a 10-unit odds ratio of 1.58, and a correlation coefficient of .94.

Although deciding to undergo CLND after a positive SLN biopsy with a nomogram-predicted probability of ≥50% may not be difficult for patients and physicians, it is the patient faced with a predicted probability of non-SLN metastases of ≤20% for whom the accuracy of the nomogram is most significant. Our findings show that the nomogram performs significantly better at predicting non-SLN metastases with sensitivities of 94% and 80% and specificities of 26% and 43% for patients with a predicted probability of 10% and 15%, respectively. Thus, within our study population, a sensitivity of 94% at a predicted probability level of 10% means that if all patients with a nomogram-predicted probability of ≥10% undergo CLND, 94% of all patients with non-SLN metastases will have CLND, and only 6% of patients with non-SLN metastases will not have CLND. With regard to specificity, a specificity of 26% at a predicted probability level of 10% means that if all patients with a nomogram-predicted probability of ≥10% undergo CLND, 74% of all patients without non-SLN metastases will have CLND, and 26% of patients without non-SLN metastases will not have CLND. For the patients in whom SLNs were assessed by TIC, the sensitivity of the nomogram increased to 97% and 84% at predicted probabilities of 10% and 15%, respectively.

Because of the 40% to 60% likelihood of no additional metastases in patients with positive SLN biopsy results and because all of these patients are at risk for adverse sequelae from the procedure, significant debate surrounds the true benefit of CLND for all patients. Although the nomogram devised by MSKCC does not recommend which patients should or should not have CLND, it does provide patients and their physicians with a meaningful tool for assessing an individual’s risk of non-SLN metastases. Although our study lends support to the validity of the nomogram and to the suitability of substituting TIC for FS within the nomogram, given the institutional variation in SLN techniques and pathologic processing, we recommend that the nomogram be validated at each institution before it is used for patient counseling. Finally, although the performance of the nomogram among the small group of patients who had primary chemotherapy in our study was similar to its performance in those who did not, dedicated validation of the nomogram for this patient population is needed before its use in the neoadjuvant setting.


   ACKNOWLEDGMENTS
 
The authors thank Martha Morrison for the critical reading of this manuscript and Kimberly Van Zee, MD, for her helpful comments. Supported by the National Institutes of Health (F.M.-B.).


   FOOTNOTES
 
Presented at the Annual Meeting of the Society of Surgical Oncology, Atlanta, Georgia, March 3–6, 2005.

Received for publication March 15, 2005. Accepted for publication September 16, 2005.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Giuliano AE, Jones RC, Brennan M, Statman R. Sentinel lymphadenectomy in breast cancer. J Clin Oncol 1997; 15:2345–50.[Abstract/Free Full Text]
  2. Veronesi U, Paganelli G, Galimberti V, et al. Sentinel-node biopsy to avoid axillary dissection in breast cancer with clinically negative lymph-nodes. Lancet 1997; 349:1864–7.[CrossRef][Medline]
  3. Veronesi U, Paganelli G, Viale G, et al. Sentinel lymph node biopsy and axillary dissection in breast cancer: results in a large series. J Natl Cancer Inst 1999; 91:368–73.[Abstract/Free Full Text]
  4. O’Hea BJ, Hill AD, El-Shirbiny AM, et al. Sentinel lymph node biopsy in breast cancer: initial experience at Memorial Sloan-Kettering Cancer Center. J Am Coll Surg 1998; 186:423–7.[CrossRef][Medline]
  5. Krag D, Weaver D, Ashikaga T, et al. The sentinel node in breast cancer—a multicenter validation study. N Engl J Med 1998; 339:941–6.[Abstract/Free Full Text]
  6. Giuliano AE, Haigh PI, Brennan MB, et al. Prospective observational study of sentinel lymphadenectomy without further axillary dissection in patients with sentinel node-negative breast cancer. J Clin Oncol 2000; 18:2553–9.[Abstract/Free Full Text]
  7. Turner RR, Ollila DW, Krasne DL, Giuliano AE. Histopathologic validation of the sentinel lymph node hypothesis for breast carcinoma. Ann Surg 1997; 226:271–6.[CrossRef][Medline]
  8. Grube BJ, Giuliano AE. Modification of the sentinel node technique: it was a hit in New York, but will it play in Poughkeepsie? Ann Surg Oncol 2001; 8:3–6.[Free Full Text]
  9. Van Zee KJ, Manasseh DM, Bevilacqua JL, et al. A nomogram for predicting the likelihood of additional nodal metastases in breast cancer patients with a positive sentinel node biopsy. Ann Surg Oncol 2003; 10:1140–51.[Abstract/Free Full Text]
  10. Kocsis L, Svebis M, Boross G, et al. Use and limitations of a nomogram predicting the likelihood of non-sentinel node involvement after a positive sentinel node biopsy in breast cancer patients. Am Surg 2004; 70:1019–24.[Medline]
  11. Breslin TM, Cohen L, Sahin A, et al. Sentinel lymph node biopsy is accurate after neoadjuvant chemotherapy for breast cancer. J Clin Oncol 2000; 18:3480–6.[Abstract/Free Full Text]
  12. Yared MA, Middleton LP, Smith TL, et al. Recommendations for sentinel lymph node processing in breast cancer. Am J Surg Pathol 2002; 26:377–82.[CrossRef][Medline]
  13. Lee A, Krishnamurthy S, Sahin A, et al. Intraoperative touch imprint of sentinel lymph nodes in breast carcinoma patients. Cancer 2002; 96:225–31.[CrossRef][Medline]
  14. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148:839–43.[Abstract/Free Full Text]
  15. Hwang RF, Krishnamurthy S, Hunt KK, et al. Clinicopathologic factors predicting involvement of nonsentinel axillary nodes in women with breast cancer. Ann Surg Oncol 2003; 10:248–54.[Abstract/Free Full Text]
  16. Weiser MR, Montgomery LL, Tan LK, et al. Lymphovascular invasion enhances the prediction of non-sentinel node metastases in breast cancer patients with positive sentinel nodes. Ann Surg Oncol 2001; 8:145–9.[Abstract/Free Full Text]
  17. Turner RR, Chu KU, Qi K, et al. Pathologic features associated with nonsentinel lymph node metastases in patients with metastatic breast carcinoma in a sentinel lymph node. Cancer 2000; 89:574–81.[CrossRef][Medline]
  18. Chu KU, Turner RR, Hansen NM, et al. Do all patients with sentinel node metastasis from breast carcinoma need complete axillary node dissection? Ann Surg 1999; 229:536–41.[CrossRef][Medline]
  19. Abdessalam SF, Zervos EE, Prasad M, et al. Predictors of positive axillary lymph nodes after sentinel lymph node biopsy in breast cancer. Am J Surg 2001; 182:316–20.[CrossRef][Medline]
  20. Rahusen FD, Torrenga H, van Diest PJ, et al. Predictive factors for metastatic involvement of nonsentinel nodes in patients with breast cancer. Arch Surg 2001; 136:1059–63.[Abstract/Free Full Text]
  21. Sachdev U, Murphy K, Derzie A, et al. Predictors of non-sentinel lymph node metastasis in breast cancer patients. Am J Surg 2002; 183:213–7.[CrossRef][Medline]
  22. Reynolds C, Mick R, Donohue JH, et al. Sentinel lymph node biopsy with metastasis: can axillary dissection be avoided in some patients with breast cancer? J Clin Oncol 1999; 17:1720–6.[Abstract/Free Full Text]
  23. Teng S, Dupont E, McCann C, et al. Do cytokeratin-positive-only sentinel lymph nodes warrant complete axillary lymph node dissection in patients with invasive breast cancer? Am Surg 2000; 66:574–8.[Medline]
  24. McCarter MD, Yeung H, Fey J, et al. The breast cancer patient with multiple sentinel nodes: when to stop? J Am Coll Surg 2001; 192:692–7.[CrossRef][Medline]
  25. Cserni G, Rajtar M, Boross G, et al. Sentinel node biopsy for breast cancer patients at the Bacs-Kiskun County Teaching Hospital (in Hungarian). Orv Hetil 2002; 143:437–46.[Medline]
  26. Greene KL, Meng MV, Elkin EP, et al. Validation of the Kattan preoperative nomogram for prostate cancer recurrence using a community based cohort: results from cancer of the prostate strategic urological research endeavor (capsure). J Urol 2004; 171:2255–9.[CrossRef][Medline]
  27. Yanke BV, Gonen M, Scardino PT, Kattan MW. Validation of a nomogram for predicting positive repeat biopsy for prostate cancer. J Urol 2005; 173:421–4.[CrossRef][Medline]
  28. Aihara T, Munakata S, Morino H, Takatsuka Y. Touch imprint cytology and immunohistochemistry for the assessment of sentinel lymph nodes in patients with breast cancer. Eur J Surg Oncol 2003; 29:845–8.[Medline]
  29. Rubio IT, Korourian S, Cowan C, et al. Use of touch preps for intraoperative diagnosis of sentinel lymph node metastases in breast cancer. Ann Surg Oncol 1998; 5:689–94.[Abstract]
  30. Aihara T, Munakata S, Morino H, Takatsuka Y. Comparison of frozen section and touch imprint cytology for evaluation of sentinel lymph node metastasis in breast cancer. Ann Surg Oncol 2004; 11:747–50.[Abstract/Free Full Text]
  31. Liang R, Craik J, Juhasz ES, Harman CR. Imprint cytology versus frozen section: intraoperative analysis of sentinel lymph nodes in breast cancer. Aust N Z J Surg 2003; 73:597–9.[CrossRef]
  32. Nagashima T, Suzuki M, Yagata H, et al. Intraoperative cytologic diagnosis of sentinel node metastases in breast cancer. Acta Cytol 2003; 47:1028–32.[Medline]



This article has been cited by other articles:


Home page
 Ann. Surg. Oncol.Home page
H. S. Cody III and K. J. Van Zee
Predicting Nonsentinel Node Metastases in Sentinel Node-Positive Breast Cancer: What Have We Learned, Can We Do Better, and Do We Need To?
Ann. Surg. Oncol., November 1, 2008; 15(11): 2998 - 3002.
[Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
A. Pinero, M. Canteras, E. Ortiz, E. Martinez-Barba, and P. Parrilla
Validation of a Nomogram to Predict the Presence of Sentinel Lymph Node Metastases in Melanoma
Ann. Surg. Oncol., October 1, 2008; 15(10): 2874 - 2877.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
E. Poirier, L. Sideris, P. Dube, P. Drolet, and S. H. Meterissian
Analysis of Clinical Applicability of the Breast Cancer Nomogram for Positive Sentinel Lymph Node: The Canadian Experience
Ann. Surg. Oncol., September 1, 2008; 15(9): 2562 - 2567.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
S. Samphao, J. M. Eremin, M. El-Sheemy, and O. Eremin
Management of the Axilla in Women With Breast Cancer: Current Clinical Practice and a New Selective Targeted Approach
Ann. Surg. Oncol., May 1, 2008; 15(5): 1282 - 1296.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
K. J. Van Zee and M. W. Kattan
Validating a Predictive Model for Presence of Additional Disease in the Non Sentinel Lymph Nodes of a Woman with Sentinel Node Positive Breast Cancer
Ann. Surg. Oncol., August 1, 2007; 14(8): 2177 - 2178.
[Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
L. A. Lambert, G. D. Ayers, and F. Meric-Bernstam
Validation of a Breast Cancer Nomogram for Predicting Nonsentinel Lymph Node Metastases after a Positive Sentinel Node Biopsy
Ann. Surg. Oncol., August 1, 2007; 14(8): 2422 - 2423.
[Full Text] [PDF]

Home page
 Ann. Surg. Oncol.Home page
A. Soran, T. Evrensel, G. Ahrendt, B. Kocer, and R. Johnson
Use of the Breast Cancer Nomogram to Predict Non-Sentinel Nodal Positivity in Patients Having Received Pre-Operative Chemotherapy
Ann. Surg. Oncol., January 1, 2007; 14(1): 165 - 165.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lambert, L. A.
Right arrow Articles by Meric-Bernstam, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lambert, L. A.
Right arrow Articles by Meric-Bernstam, F.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS