Article Text

Download PDFPDF
False negative rate for intraoperative sentinel lymph node frozen section in patients with breast cancer: a retrospective analysis of patients in a single Asian institution
  1. Jolene Wong1,
  2. Wei Sean Yong2,
  3. Aye Aye Thike3,
  4. Jabed Iqbal3,
  5. Ahmed Syed Salahuddin3,
  6. Gay Hui Ho2,
  7. Preetha Madhukumar2,
  8. Benita Kiat Tee Tan2,
  9. Kong Wee Ong2,
  10. Puay Hoon Tan3
  1. 1Department of General Surgery, Singapore General Hospital, Singapore
  2. 2Department of Surgical Oncology, National Cancer Centre, Singapore
  3. 3Department of Pathology, Singapore General Hospital, Singapore
  1. Correspondence to Jolene Wong, 11 Hospital Drive, Singapore 169610;{at}


Background and objective Intraoperative frozen section of the sentinel lymph node (SLN) in clinically node negative breast cancer patients detects metastatic disease and enables axillary lymph node dissection to be performed in the same operative setting. Internationally, the false negative rate (FNR) for SLN biopsy ranges from 5.5% to 43%. The size of SLN metastasis has been identified as a key factor affecting FNR. We review our institutional experience on the accuracy of intraoperative SLN biopsy.

Methods Data were collected retrospectively from patients undergoing SLN biopsy performed at Singapore General Hospital. The SLN was identified using blue dye, radioisotope or both. Frozen section was performed intraoperatively. When SLN was positive for metastasis on frozen section, completion axillary clearance was performed. False negative cases were defined as patients in whom a negative frozen section result was obtained, whose final permanent paraffin section was positive. We determined the FNR of SLN frozen section and evaluated the factors associated with it.

Results A total of 2202 SLN biopsies were performed between January 2005 and June 2012. There were 89 false negative cases, of which there were 23 (25.8%) cases of isolated tumour cells (ITCs), 49 (55.1%) cases of micrometastasis, and 17 (19.1%) cases of macrometastasis. The overall FNR was 13.5%. FNR was 79.3% in ITCs, 59.8% in micrometastasis, and 3.1% in macrometastatic disease. Non-ductal histological subtype, absence of lymphovascular invasion and the size of SLN metastasis were identified as significant independent factors associated with a higher FNR.

Conclusions FNRin our institution is acceptable when compared to other large centres. Failure to detect metastasis in frozen section in more than half of our patients was due to ITCs and micrometastasis.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In the context of breast cancer, the sentinel lymph node (SLN) refers to the lymph node that first receives lymph from the area of the breast containing the tumour. It is well established that in the presence of a negative SLN, the likelihood of further axillary involvement is very low.1–6 Long term results from randomised trials have demonstrated that overall survival rates in patients who received no axillary dissection when the SLN was negative was similar to patients who received upfront complete axillary dissection.7–10 As such, with the rationale of minimising the morbidity associated with axillary lymph node dissection (ALND),11–14 it is performed only when the SLN is positive for metastatic disease.

Intraoperative frozen section (IFS) analysis of SLNs allows prompt detection of metastatic disease so that immediate ALND may be performed in the same operative setting. Internationally, false negative rates (FNRs) of IFS for SLNs range from 5.5% to 43%.15–19 This large inter-institutional variability may be attributed to differing institutional protocols and expertise available.15 ,19

The goal of this study is to determine the FNR of IFS of SLNs in breast cancer surgery in our institution and to define factors associated with false negativity.

Materials and methods

Patient selection

The study was carried out under the approval of the Centralised Institutional Review Board of the Singapore Health Service. Data were retrospectively collected from patients who underwent SLN biopsy (SLNB) at Singapore General Hospital (SGH) between January 2005 and June 2012. All patients had a histopathologically proven diagnosis of breast cancer and clinically negative axilla. Patients with ductal carcinoma in situ or prior neo-adjuvant chemotherapy were excluded from this study. Consent was taken for either partial or total mastectomy with SLNB; and patients were counselled for the likelihood of ALND in the same operative setting should the SLN be positive for metastatic disease. A total of 2202 SLN biopsies were performed during the assigned time period.

Identification of SLNs

SLNs were identified using a combination of both vital blue dye and/or technetium-99m sulfur colloid.20 These were injected either intradermally above the tumour or peritumorally. Preoperative lymphoscintigraphy was used to identify draining axillary SLNs in selected cases at the surgeon's discretion. A hand-held gamma probe was used to evaluate radioactive counts where technetium-99m sulfur colloid was utilised.

IFS analysis of SLNs

Identified SLNs were submitted for IFS evaluation. The lymph nodes were sliced at 2 mm intervals, frozen in liquid nitrogen and cryosectioned to produce 3–6 levels of 5 μm thick frozen sections (FSs). These were then stained with H&E and examined for the presence of metastasis. Metastases were classified as either micrometastatic (tumour deposits of size up to 2 mm in the greatest dimension) or macrometastatic (deposits more than 2 mm in size). Isolated tumour cells (ITCs) were defined as tumour foci measuring up to 0.2 mm. The frozen material as well as any unfrozen SLN tissue were fixed and embedded in paraffin. Serial sections were stained with H&E and examined microscopically. Immunohistochemical (IHC) stains for CAM5.2 and AE1:AE3 were used only when the presence of cells suspicious of metastases were seen on H&E sections, needing IHC verification.

Statistical analysis

Outcomes of the IFS of SLN were compared with that based on final histopathological analysis of the paraffin section of SLN for standard computation of FNR, sensitivity (Se), negative predictive value (NPR) and false NPR. False negative cases were defined as patients in whom a negative FS result was obtained, and whose final permanent paraffin section was positive. FNR was defined as (false negative)/(true positive+false negative). The sensitivity of IFS was defined as (true positive)/(true positive+false negative).

Logistic regression models were fitted to estimate the ORs to assess the association of various variables with a false negative diagnosis of IFS of SLN. Multivariable analyses were performed on variables with p<0.05 from the univariable analyses. Forward selection, backward elimination and stepwise selection algorithms were applied to identify independent predictors. For the final derived model, the goodness of fit between the observed and predicted number of outcomes was assessed based on the Hosmer–Lemeshow test. A 2-tailed p value <0.05 was considered statistically significant. All analyses were performed using SAS V.9.3.


A total of 2202 SLN biopsies were performed for 2174 patients between January 2005 and June 2012; 28 (2.2%) patients had two specimens due to recurrent or bilateral disease. The median age was 55 years (range 23–88) and the median tumour size was 22 mm (range 0.5–500); clinicopathological characteristics are summarised in table 1.

Table 1

Clinicopathological characteristics of patients

Nodal metastasis was found in 660 (30%) final paraffin section (FPS) specimens. FS was negative in 89 and positive in 571 specimens. FNR was 13.5% (table 2). There were three cases in which FPS was negative, yet FS was positive. These were not considered true false positives as a review of the FS slides did reveal metastatic tumour foci. We postulate that the final negative histopathology results were likely due to tissue loss during paraffin section processing.

Table 2

Comparison of frozen section versus final paraffin section results of sentinel lymph node (SLN)

Factors associated with false negative IFS result

Of the 89 false negative IFS specimens, 23 (25.8%) were ITCs, 49 (55.1%) micrometastases, and 17 (19.1%) macrometastases. Overall FNR was 13.5%. FNR was 79.3%, 59.8% and 3.1% in ITCs, micrometastatic and macrometastatic disease, respectively. The smaller the SLN metastases, the higher the odds of a false negative diagnosis (p<0.001, table 3).

Table 3

False negative rate (FNR) of frozen section of sentinel lymph node (SLN) by size of SLN metastases

Other factors that were significantly associated with false negative diagnosis included non-invasive ductal carcinoma (non-IDC) histology and absence of lymphovascular involvement (table 4). There was no evidence from the data to support that invasive lobular carcinoma (ILC) was significantly associated with higher odds of false negative diagnoses than IDC (OR=2.0; 95% CI 0.8 to 4.7) or non-ILC (OR=1.3; 95% CI 0.6 to 3.1).

Table 4

Univariable logistic regression of false negative rate (FNR) of frozen section of sentinel lymph node

Multivariable logistic regression analysis using forward selection, backward elimination and stepwise selection algorithms identified identical predictors for false negative diagnosis of IFS of SLN (table 5). Size of SLN metastases and histology were independent predictors for false negative diagnosis in the final fitted model. Between the two variables, size of SLN metastases was more important than histology as a predictor for false negative diagnosis (OR for ITC vs macrometastases=123.5; OR for micrometastases vs macrometastases=46.1; OR for non-IDC vs IDC=3.1).

Table 5

Multivariable logistic regression of false negative rate of frozen section of sentinel lymph node (SLN)

Review of false negative specimens

An independent pathological review of all false negative specimens was carried out. IFS slides were retrieved and reanalysed. Of the 89 recorded false negative specimens, 69 FS slides were retrievable. On review, 9 (13%) slides yielded a positive result, that is, SLN metastasis was identified in the original FS slide. Of the 60 negative IFS slides, a review of the FPS was performed. Thirty-three FPS were retrievable, and the characteristics of SLN metastasis in these sections were studied. Ninety-four per cent (31 out of 33 specimens) of metastatic foci were found in the subcapsular sinus of the lymph node (figure 1).

Figure 1

Micro-photograph of single subcapsular sinus metastasis in the sentinel lymph node.


Discordance between IFS analysis and final histopathology of SLN occurred in 13.5% of our study cohort. Size of SLN metastases as well as histological subtype were identified as significant predictors of FNR. This is consistent with results from multiple published studies,2 ,18 ,19 ,21 with many authors concluding that the failure of FS diagnosis was in the detection of micrometastatic disease and ITCs.

In our study, a non-IDC histological subtype was associated with a significantly higher rate of false negative IFS results. It was initially postulated that nodal metastases from ILC were more difficult to visualise than those of IDC subtype, hence the higher FNR. Loss of E-cadherin, a cell–cell adhesion molecule in ILC, may result in a tendency for SLN metastasis to occur as discohesive cells, mimicking histiocytes, and therefore becoming detected only in FPS with the help of IHC.22 However, when sub-analysis was performed, ILC did not prove to be significantly associated with a false negative diagnosis. This was consistent with findings from Chao et al17 and Turner et al,23 where a lobular histology did not result in a difference in FNRs. It has been reported that in patients with favourable histological subtypes—pure mucinous, medullary, cribriform or tubular—a higher FNR is expected.24 These subtypes were included into the non-IDC group, and as such could possibly have resulted in a significant difference in FNR between the IDC versus non-IDC group on multivariable analysis (table 5).

In a review of the 89 false negative specimens encountered in our study, 94% had metastatic foci located in the subcapsular sinus of the SLN. Initial studies on the clinical implications of the micro-anatomical location of metastatic deposits in the SLN were conducted for melanomas.25 ,26 Dewar et al26 concluded that the microanatomic location of SLN metastasis predicted non-SLN involvement, and proposed that lymph node dissection could be safely avoided in patients with tumour deposits limited to the subcapsular sinus.27 This was based on the principle that tumour cells follow an orderly route of spread within the SLN, arriving first at the subcapsular sinus before invading into the parenchyma.28 A large proportion of our false negative cohort had disease limited to the subcapsular sinus, and if this was predictive of minimal involvement of non-SLN lymph nodes, immediate axillary clearance in this group might be of questionable benefit.

In recent years, the need for axillary clearance in women with early breast cancer has been called to question. The IBCSG 23-01 trial indicated that in women with micrometastatic disease in the SLN, ALND was no longer indicated as it did not result in added survival benefit.29 This was further supported by results from the ACOSOG Z0011 trial.30 While the need to detect micrometastatic disease diminishes in relevance in the light of recent publications, their true significance remains a matter for debate.31 The use of IFS to detect macrometastatic disease in the axilla continues to remain important. The FNR was found to be 3.1% for macrometastases, therefore most of our patients would have received the necessary axillary clearance during the same operative setting, and only a very small number would have required a second operation; thereby saving on time, cost, additional morbidity and inconvenience.


FNR in our institution is acceptable when compared to other large centres. Failure to detect metastasis in FS in more than half of our patients was due to ITCs and micrometastasis.

Take home messages

  • Intraoperative frozen section (FS) is a reliable modality to assess the presence of sentinel lymph node (SLN) metastases.

  • Sensitivity of FS is dependent on the size of the SLN metastases as well as histology.



  • Correction notice This article has been corrected since it was published Online First. The provenance and peer review statement has been amended.

  • Handling editor Cheok Soon Lee

  • Contributors JW: first author; WSY: second author; AAT, JI, ASS: assisted in collection of data; GHH, PM, BKTT, KWO: assisted in providing the database for data collection; PHT: principal investigator.

  • Competing interests None.

  • Ethics approval Singapore General Hospital Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.