• breast cancer;
  • sentinel lymph node;
  • frozen section;
  • meta-analysis;
  • intraoperative evaluation


  1. Top of page
  2. Abstract
  6. Acknowledgements


Accurate intraoperative pathologic examination of sentinel lymph nodes (SLNs) has been an important tool that can reduce the need for reoperations in patients with SLN-positive breast cancer. The objective of the current study was to determine the accuracy of intraoperative frozen section (IFS) of SLNs during breast cancer surgery.


The authors retrospectively reviewed the records of 326 patients with breast cancer who underwent IF analysis of SLNs at a single institution. Then, they conducted a meta-analysis that included 47 published studies of IFS of SLNs in patients with breast cancer.


Hematoxylin and eosin (H&E) staining revealed metastasis in SLNs in 99 patients (30.4%), including 61 patients with macrometastasis (MAM) (>2 mm) (the MAM group) and 38 patients with micrometastasis (Mi) or isolated tumor cell (ITC) deposits (the Mi/ITC group). The overall sensitivity of the institutional series was 60.6% (60 of 99 patients), and overall specificity was 100% (227 of 227 true negatives). The sensitivity of IFS was significantly lower in the Mi/ITC group (28.9%) than in the MAM group (80.3%; P < .0001). According to the meta-analysis of published studies and data from the author's institution (47 studies, for a total of 13,062 patients who underwent SLN dissection with IFS of SLNs), the mean sensitivity was 73%, and the mean specificity was 100%. The mean sensitivity was 94% for the MAM group and 40% for the Mi/ITC group.


IFS of SLNs was more reliable for detecting MAM than for detecting Mi/ITC deposits. It lacked sufficient accuracy to rule out Mi/ITC deposits. Cancer 2011. © 2010 American Cancer Society.

Sentinel lymph node (SLN) dissection (SLND) has become a standard technique for accurately determining axillary lymph node status in patients with breast cancer,1-4 in whom numerous studies have confirmed the high sensitivity and specificity of SLND.5-18 Several studies have reported that SLND significantly minimized morbidity and lymphedema compared with axillary lymph node dissection (ALND).19-21

Accurate SLN analysis requires coordinated efforts by a multidisciplinary team. If the SLN is identified as positive intraoperatively, then the patient may be a candidate for immediate ALND. Intraoperative pathologic examination of SLNs is helpful to avoid reoperations for SLN-positive patients, thus sparing them an additional operation with its risks and additional costs.

Frozen section is the most common method of intraoperative evaluation, but no pathologic method has been standardized for SLN evaluation. Multilevel sectioning has been used for pathologic evaluation of SLNs; however, institutional protocols vary widely.22, 23 The 2005 American Society of Clinical Oncology (ASCO) guidelines for the intraoperative evaluation of SLNs indicated that, although intraoperative frozen section (IFS) risks significant destruction of potentially diagnostic tissue, it may be the most desirable method for experienced teams of surgeons and pathologists.24 The objective of the current study was to determine the accuracy of IFS of SLNs during breast cancer surgery.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Patient Selection

In this retrospective study, we reviewed the records of 326 consecutive patients who with invasive breast cancer who underwent SLND with routine IFS at a single institution (the University of California at San Francisco [UCSF]) from November 1997 through April 2003. This study was approved by the institutional review board at UCSF. No patients in this series received neoadjuvant chemotherapy.

Preoperative Lymphoscintigraphy

The decision to use preoperative lymphoscintigraphy to identify the draining axillary SLNs was at the discretion of the operating surgeon. Technetium 99 sulfur colloid was injected either intradermally above the tumor, peritumorally, or at the surgical site of the previous biopsy. Patients underwent injection of radioisotope, with or without lymphoscintigraphy, either the morning of the surgery or in the late afternoon the day before surgery.

Intraoperative Identification of Sentinel Lymph Nodes

During the operation, a hand-held gamma probe (Neoprobe Corporation, Dublin, Ohio) was used to evaluate radioactive counts. Injection of Lymphazurin dye (Hirsch Industries, Richmond, Va) was at the surgeon's discretion.25 An SLN was defined as a blue lymph node and/or a lymph node with an ex vivo radioactive count ≥10% of the ex vivo radioactive count of the hottest lymph node. In addition, any clinically suspicious, palpable lymph nodes were removed and defined as SLNs.26

Pathologic Evaluation

The radioactive count of all SLNs that were removed was measured and documented. All SLNs were submitted to the pathology department and subjected to standard frozen section evaluation with 1 or 2 hematoxylin and eosin (H&E)-stained sections.

Tumors were classified according to American Joint Committee on Cancer (AJCC) using the sixth edition of the AJCC Cancer Staging Manual; permanent sections were stained with H&E with or without immunohistochemistry (IHC) and consisted of a bisection of ≥3 levels at intervals of 40 μm to 100 μm.27, 28 All lymph node tissue specimens that were submitted for frozen section evaluation were resubmitted for permanent-section analysis (PSPA). Our protocol included 3 levels of IHC stains and 2 levels of H&E stains.22, 23

Literature Review

We systematically searched the PubMed database from January 1997 through June 2008 using the following keywords: “breast cancer,” “breast neoplasm,” “frozen section,” “sentinel lymph node,” and “SLN.” References from included articles were searched for additional studies that met our inclusion criteria. We included all articles that reported sufficient data for cross-tabulation of the results of IFS of SLNs against PSPA. We excluded all articles with missing statistical data (ie, sensitivity, specificity). We also excluded all articles in languages other than English that were not available in translated form. All studies that involved neoadjuvant therapy were excluded.

The data were collected on a per-patient basis rather than per lymph node examined. The objective of our meta-analysis was to evaluate the accuracy of IFS evaluation.

Statistical Analysis

Sensitivity, specificity, and accuracy were calculated using the results from PSPA. Because the numbers of patients with micrometastases (Mi) and with isolated tumor cell (ITC) deposits were low, we combined them into the Mi/ITC group and compared their results with those from the macrometastases (MAM) group. We used Mantel-Haenszel and/or chi-square statistics to compare the sensitivity and accuracy for different tumor sizes and to compare the sensitivity for detecting MAM and Mi/ITC. P values <.05 were considered statistically significant.

Because different studies in a meta-analysis may have used different thresholds to define positive results, we used summary receiver operating characteristic (SROC) curve analysis to account for such differences in our meta-analysis.28, 29 We also used a generalized linear mixed model to evaluate differences in diagnostic accuracy between patient groups or methods. For all of our statistical analyses, we used the statistical software package R (version 2.6.0; R Foundation for Statistical Computing, Vienna, Austria).30


  1. Top of page
  2. Abstract
  6. Acknowledgements

Patient Characteristics

The characteristics of all 326 patients in our single-institution study are summarized in Table 1. All patients were women, and their mean age at diagnosis was 55.7 years (range 23.5-88.5 years). The mean tumor size was 14.9 mm (range, 1.0-82 mm). SLND identified the SLNs in all patients. The mean number of SLNs removed was 3.2 per patient.

Table 1. Patient Characteristics (n=326)
VariableNo. of Patients (%)
  • a

    Other histologies included 10 tubular carcinomas, 5 mucinous carcinomas, 3 micropapillary carcinomas, and 1 medullary carcinoma.

Tumor classification 
 T1a31 (9.5)
 T1b83 (25.5)
 T1c151 (46.3)
 T257 (17.5)
 T34 (1.2)
Histologic findings 
 Invasive ductal carcinoma278 (85.3)
 Invasive lobular carcinoma29 (8.9)
 Othera19 (5.8)
 Mastectomy85 (26.1)
 Lumpectomy241 (73.9)

Sentinel Lymph Node Findings by Method and Tumor Size

Of the 326 patients who underwent IFS of SLNs, 99 patients (30.4%) had positive SLNs diagnosed by PSPA. Of those 99 patients, 39 patients (39.4%) had negative results according to the IFS. IFS yielded a sensitivity of 60.6%, a specificity of 100%, and an accuracy of 88%. The positive predictive value was 100%, and the negative predictive value was 83.2%. There were no false-positive IFS results.

Macrometastases Versus Micrometastases/Isolated Tumor Cell Deposits

Of the 99 SLN-positive patients, 61.6% had MAM (Table 2), and 38.4% had Mi/ITC (Table 3). The sensitivity of IFS was significantly lower in the Mi/ITC group (28.9%) than in the MAM group (80.3%; P < .0001). Sensitivity was not correlated significantly with tumor size in either the MAM group (P = .22) or the Mi/ITC group (P = .74).

Table 2. Studies From the Sentinel Lymph Node Meta-Analysis
StudyNo. of PatientsNo. of True- Positive FindingsNo. of False- Negative FindingsNo. of False- Positive FindingsNo. of True- Negative FindingsSensitivity, %Specificity, %Accuracy, %
Veronesi 19974010732180576410083
Flett 199839561830358610095
Hill 199942405681003278710098
Noguchi 199943621990346810085
Turner 199944278535801674810079
Van Diest 199938542120319110096
Morgan 19994532550225010084
Rahusen 20004610021160635710084
Weiser 2000478901359616585899.889
Imoto 200048521720338910096
Motomura 20003510113120765210088
Gemignani 200049375433402985610091
Liu 200050381160216510084
Gulec 200151157182301164410085
van der Loo 200152275641301988310095
Tanis 200153265712511687499.490
Zurrida 200154192552601116810086
Chao 200155203361711496899.391
Nagashima 2003561242360957910095
Leidenius 2003573751162322348399.293
Liang 20035820530126310085
Menes 2003596721120346410082
Nahrig 200336301430138210090
Krogerus 200441 (Method A)102280074100100100
Krogerus 200441 (Method B)1023660608610094
D'Errico 200460681410539310099
Lauridsen 20046112136290565510076
Holck 200462263752801607310089
Wada 2004635691332604108410095
Aihara 2004641072340808510096
Reitsamer 2004653281042402008110093
Khalifa 200466961830758610097
Mitchell 2005678581539406116210089
Hung 20056810534161546898.284
Cao 200569221661101448610095
Perez 200570872570557810092
Al-Shibli 200571701540517910094
Grabau 200572272612311877399.591
Broqi 20057313329210835810084
Schrenk 2005742326513282015316510088
Celebioglu 20067510236130537410087
Choi 200637791460597010092
Leung 200776262631101888510096
Langer 2009316481498304166410087
Arora 200832327783012186399.590
McLaughlin 20083393117013606255610085
Shimazu 200834301662732057198.690
Current study326603902276110088
Table 3. Macrometastasis and Micrometastasis Results: Studies From the Sentinel Lymph Node Meta-Analysis
No. of PatientsNo. of False-Negative FindingsSensitivity, %No. of PatientsNo. of False-Negative FindingsSensitivity, %
Turner 19994464571147198
Rahusen 20004618132819384
Weiser 2000471158724109992
Nagashima 2003568450210100
Leidenius 20035745176294694
Menes 2003591082023483
Nahrig 200336626750100
Krogerus 200441 (Method A)70100210100
D'Errico 200460518070100
Lauridsen 20046130222735780
Wada 200463211624771087
Perez 2005703167230100
Grabau 20057228202956395
Broqi 20057325202025196
Celebioglu 200675201335270100
Choi 2006374250120100
Langer 200931686111145398
Shimazu 20083432223161592

In the Mi/ITC group, 13 patients (34.2%) had metastasis identified in an SLN when PSPA was done using H&E staining, whereas 12 patients (31.6%) had metastasis identified in an SLN using IHC. The Mantel-Haenszel statistical method to control for tumor size indicated a significant difference in the sensitivity of IFS between the MAM group and the Mi/ITC group (P < .00001). In the Mi/ITC group, 8 patients had only ITC deposits in SLNs, and the sensitivity was 0% (0 of 8 patients).


In total, 47 articles (including our own data) reported sufficient data for cross-tabulation of the results of IFS of SLNs against PSPA (Table 2).31-76 Of those articles, 2 evaluated metastasis in IFS of SLNs using H&E staining and ultrarapid cytokeratin IHC.36, 37 In those 2 studies, we included only the results from H&E staining, because few patients underwent ultrarapid IHC. Most IFS SLNs were stained with H&E or toluidine blue. One study reported 2 methods of intraoperative SLN evaluation.41

The total number of enrolled patients in the 47 studies was 13,062; 32% of those patients had positive SLNs, and 68% had negative SLNs. When we combined the data from all 47 studies in our meta-analysis, the overall sensitivity was 68%, and the overall accuracy was 90%.

Only 18 studies reported comparing IFS results between patients with MAM and patients with Mi (Table 3). The sensitivity in those 18 studies ranged from 80% to 100% for patients with MAM and from 11% to 100% for patients with Mi. Only 5 studies reported the results from IFS of ITC deposits (Table 4), and 3 of those studies reported 0% sensitivity for detecting ITC deposits.31, 37, 70

Table 4. Isolated Tumor Cell Results: Studies From the Sentinel Lymph Node Meta-Analysis
StudyNo. of PatientsSensitivity, %
ITC DepositsFalse-Negative Findings
  1. ITC indicates isolated tumor cell; SLN, sentinel lymph node.

Nahrig 2003366183
D'Errico 20046030100
Perez 200570660
Choi 200637440
Langer 20093119190

According to our meta-analysis, the mean sensitivity for detecting metastasis using the SROC model was 73%, and the mean specificity was 100%. The mean sensitivity for detecting Mi, however, was substantially lower at 40%, whereas the mean sensitivity for detecting MAM was 94%. The mean specificity was 99% regardless of the type or size of metastasis. Frozen section was significantly more sensitive for MAM than for Mi in our meta-analysis (P < .0001).

We also analyzed the sectioning and staining methods of both IFS and PSPA. For the different intraoperative sectioning methods, we observed no significant differences in diagnostic accuracy measured by SROC curves (data not shown). In contrast, for the different PSPA staining methods, the combination of H&E and IHC provided the best diagnostic accuracy according to the SROC curve analysis (P < .004).


  1. Top of page
  2. Abstract
  6. Acknowledgements

Although an accurate method for analyzing SLNs intraoperatively is needed to prevent a second operation for completion ALND in patients with breast cancer who have lymph node-positive disease, no universally accepted method exists. Many surgeons use IFS; however, its use is controversial because its sensitivity varies widely, as our study demonstrates.

We observed that IFS was reliable for detecting MAM and could help avoid reoperations in patients with breast cancer; however, IFS typically failed to detect Mi/ITC. Multiple published studies also indicated that IFS failed to detect Mi/ITC.40, 46, 47, 55 In addition to its low sensitivity for detecting Mi/ITC, IFS has other potential disadvantages, including tissue consumption, artifacts from freezing and thawing of specimens,77-79 increased surgical time, and cost. The benefit of avoiding a second surgical procedure must be weighed against the risk of obtaining a false-negative result, which could increase the risk of an inaccurate lymph node assessment. Despite these concerns, our meta-analysis validated the routine use of IFS given its outstanding sensitivity for MAM.

IFS reduces the need for secondary operation in patients who have been recommended to undergo completion axillary dissection for a positive SLN. Recently, results reported from the American College of Surgeons Clinical Oncology Group (ACOSOG) Z0011 trial indicated that there was no benefit from ALND for women who had clinically lymph node-negative breast cancer and a positive SLN.80 Our findings support the routine clinical use of IFS of SLNs. However, we recognize that the appropriate use of IFS will continue to be an evolving issue given findings from the ACOSOG Z0011 trial. If those findings are validated and incorporated into clinical practice, then fewer completion ALNDs may be recommended for the very patients in whom IFS previously played an important role.

Recently, a commercially available, real-time, reverse-transcriptase polymerase chain reaction (the GeneSearch BLN assay; Veridex LLC, Warren, NJ) was used to detect SLN metastasis. Viale et al81 compared results from the BLN assay with results from serial IFS in a series of 293 SLNs from patients with breast cancer. The BLN assay correctly identified 51 of 52 SLNs with MAM and 5 of 20 SLNs with Mi; its sensitivity was 98.1% for detecting MAM (94.7% for MAM >1 mm and 77.8% for MAM >0.2 mm). Those authors concluded that the sensitivity of the BLN assay was comparable to that of histopathologic examination of the entire SLN by serial sectioning at 1.5 mm to 2 mm.81 Compared with IFS, the BLN assay for intraoperative SLN analysis yielded better results than those produced by our single-institution study and our meta-analysis.

The value of imprint cytology (IC) for the intraoperative analysis of SLNs is controversial. A meta-analysis of 31 studies that evaluated intraoperative IC for SLNs by Tew et al82 indicated that the pooled sensitivity of IC was 63%, and its specificity was 99%. Pooled sensitivity was 81% for MAM and 22% for Mi. The authors concluded that IC is simple and rapid and has good sensitivity for MAM but not for Mi. The sensitivity of frozen section for detecting MAM (94%) and Mi (40%) in our meta-analysis was better than the sensitivity for IC in published reports.35, 82, 83

Some have argued that tissue loss is a potential concern of IFS. The term “false-negative rate” implies a difference between IFS and PSPA, because tissue loss cannot be measured exactly in IFS. Studies that have carefully and exhaustively sampled SLNs through multiple, deeper level sections have indicated that metastatic deposits may continue to be discovered throughout the lymph node.84, 85 Tissue loss is 1 of the factors cited by various authors who argue against routine IFS analysis of SLNs.77, 78 Some authors contend that grossly benign, diminutive SLNs should be sent for PSPA without IFS evaluation.79 However, a metastatic deposit so small that it would be exhausted at the time of IFS indicates a very small lymph node tumor burden that would be questionable in terms of clinical significance.

In contrast, others would answer this concern regarding tissue loss as a theoretical source of false-negative results by suggesting that pathologists examine technical variables as a source of tissue loss. Some examples of such variables would be 1) the expertise of the technician performing sectioning; 2) the variability of fatty tissue present within and around the lymph node(s), which creates challenges in achieving suitable sections; and 3) the pathologist's emphasis (whether on preserving tissue for PSPA or on rendering a definitive diagnosis by performing multiple levels of IFS).

An important limitation of both our study and our meta-analysis was the lack of any standard for defining clinically suspicious lymph nodes in need of IFS. Specht et al86 defined clinically suspicious lymph nodes as firm, shotty, and more prominent than lymph nodes on the contralateral side. This lack of a standard definition for clinically suspicious lymph nodes, in addition to different pathologic techniques and variable institutional experience, may explain in part the wide range of sensitivity reported for IFS.

In most situations, IFS evaluation increases the efficiency of surgical treatment for patients with early stage breast cancer. However, in select situations, it is potentially problematic (fatty lymph nodes that are difficult to section, diminutive lymph nodes that may result in substantial tissue loss if subjected to IFS). IFS generally is helpful for avoiding reoperations for complete ALND in patients who undergo SLND; like all approaches, there are some potential drawbacks.

IFS is performed readily by most pathologists without additional equipment or specialized training in cytopathology. Our meta-analysis demonstrates that, although results vary between institutions, the overall sensitivity of IFS of SLNs is excellent for MAM at 94%. Currently, no intraoperative method of SLN analysis is considered sensitive enough to rule out Mi/ITC. IFS reduces the need for secondary operations, at least for most patients who have breast cancer with identifiable positive SLNs and unequivocal evidence of positive lymph node disease.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank Pamela Derish, MA (University of California at San Francisco) and Mary Knatterud, PhD (University of Arizona) for assistance with grammatical review.


  1. Top of page
  2. Abstract
  6. Acknowledgements