WHAT IS ALREADY KNOWN ON THIS TOPIC

Touch preparation (TP) is an easy and rapid method to detect metastasis during intraoperative evaluation of sentinel lymph nodes (SLNs) in patients with breast cancer who undergo primary surgery. TP may not be the appropriate method to evaluate SLNs in patients treated with neoadjuvant systemic therapy (NST). We compared TP and frozen section (FS) in both primary surgery and NST settings.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYINTRODUCTION

For several decades, intraoperative evaluation (IOE) of sentinel lymph nodes (SLN) to determine the need for an immediate axillary lymph node dissection (ALND) has been standard practice in the surgical management of patients with breast cancer. More recently, this practice has been increasingly used in the evaluation of SLNs in patients treated with neoadjuvant systemic therapy (NST).1

The two most common methods used for IOE of SLNs include frozen section (FS) and touch preparation (TP). Like any other assay, multiple factors influence the accuracy of results in the IOE of SLNs. Some of the factors include the method of evaluation, quality of the TP or FS slides, the size of the metastasis, tumour type and grade, whether the patient received treatment before surgery and the experience of the pathologists interpreting the slides.2 3

In the setting of primary surgery (non-NST setting), detection of lymph node metastasis intraoperatively by TP alone has an overall sensitivity of 63% (range from 34% to 95%),4 whereas the overall sensitivity for FS is 86% (range from 44% to 99.8%).5 In contrast, IOE of SLN metastasis in patients treated with NST can be challenging due to therapy effects. Common changes encountered in both the primary site and in the lymph node metastases after NSTs are loss of tumour cellularity, altered morphology of the residual tumour cells, presence of fibrosis, necrosis and macrophages.6 Since the IOE of SLNs in the NST setting is a recent development in surgical treatment of patients with breast cancer, there is limited literature on the accuracy of the IOE of SLNs using TP or FS.7–14

At our institution, most patients with breast cancer undergo clinical and imaging evaluation of their axillary lymph nodes during their initial cancer diagnosis. Any lymph node suspicious for metastasis is biopsied under image guidance. Patients scheduled to receive NST, a clip is placed to mark the lymph node at the time of the core biopsy or fine-needle aspiration biopsy. Most of our patients treated with NST undergo SLN biopsy (SLNB) at the time of definitive surgery and, thus, IOE of SLNs is commonly performed to determine the need for an immediate ALND. Patients whose tumours are resistant to therapy (non-responders) and continue to have a clinically positive lymph node undergo ALND at the time of definitive surgery. The aim of this study was to assess the accuracy of TP and FS in the IOE of SLNs in the neoadjuvant setting and to investigate factors that contribute to low accuracy when compared with non-NST settings.

Materials and methods

This retrospective study was undertaken at the University of Texas Southwestern Medical Center from the cases collected between January 2016 through January 2019. The institutional review board approved the study. At our institution, the pathology department has a subspecialty practice model. However, IOEs are performed by pathologists from all subspecialties, including pathologists who are in breast subspecialty.

Patients with invasive breast carcinoma who had an SLNB procedure and IOE of the SLNs either by touch prep (TP), FS or by both methods were included in this study. Based on our standard operative procedure, each SLN is serially sectioned at 2–3 mm intervals perpendicular to the long axis of the node. For TP, slides are gently pressed into the cut surfaces of lymph nodes once (one slide per block). For FS, all slices of the LN are submitted (unless a lymph node is grossly positive). We routinely cut one FS slide unless the pathologist requests additional slides for technical issues. The TP and FS slides are stained with H&E stain and interpreted by a single pathologist assigned for frozen service. The pathologists performing IOE include all subspecialties of surgical pathology, including a few who also practice cytopathology. Touch preps and FS were interpreted by the assigned pathologist irrespective of their cytopathology training and did not require a second review by a cytopathologist. For permanent sections, two levels of H&E sections were cut from each paraffin block. Routine cytokeratin stains were not performed in SLNs unless pathologists requested to confirm atypical cells or to determine the extent of metastasis in difficult cases.

Cases were divided into two groups: patients who received NST (endocrine therapy or chemotherapy) and patients who had primary surgery (non-NST group). This included patients who may have had a positive core biopsy of an axillary lymph node prior to NST. In patients in the NST group with a prior positive lymph node biopsy, if feasible, lymph nodes are localised using radioactive seed prior to surgery to ensure removal of the prior positive node during SLNB. If biopsied pretreatment node was benign, that node is not localised for the SLNB procedure. However, the surgeons often use intraoperative specimen radiograph if a previously biopsied node is within the sentinel nodes. This is done to capture any suspicious node that was false negative in initial biopsy. In the non-NST group, only patients who had invasive carcinoma were included in the study. The results of IOE and final pathologic diagnosis were reviewed in all cases. In discordant cases, the TP and FS slides were reviewed (by SS) to investigate the reason for false-positive or false-negative results. Other parameters analysed are the results of lymph node status prior to NST, number of SLNs removed, tumour grade, histologic subtype, size of metastasis and response to treatment. SLN metastases were classified as isolated tumour cells (≤0.2 mm ≤200 cells), micrometastases (>0.2 mm to 2 mm) or macrometastases (>2 mm).

The sensitivity, specificity, positive predictive value (PPV) and likelihood ratios for each method in NST and non-NST settings were calculated in two different manners: (1) on the level of individual lymph node, regardless of the patient’s overall lymph node status and (2) on the patient level irrespective of number of lymph nodes retrieved during SLNB. Atypical or deferred to permanent results were excluded from the calculations of sensitivity, specificity and PPV. The findings in these nodes were reported descriptively.

We used Fisher test to determine the association of a false-negative result for categorical factors and Mann-Whitney U test for continuous factors. A p value of <0.05 was considered statistically significant. All the statistics were performed with Stata 16.

Results

Figure 1 lists the distribution of cases in each category. Of the 871 SLNs with available IOE, 615 SLNs were from the NST setting (166 patients) and 256 SLNs from the non-NST setting (66 patients). Patients and tumour characteristics are summarised in table 1.

Table 1

Patient and tumour characteristics

Figure 1

Distribution of the study groups.

Neoadjuvant setting

Of the 615 SLNs from 166 patients, 366 SLNs (75 patients) were evaluated by TP alone, 90 SLNs (19 patients) by FS alone and 135 SLNs (72 patients) by both TP and FS. Four patients with atypical results and 24 SLNs interpreted as atypical or deferred to permanent sections were excluded from the statistical analysis of accuracy.

Touch preparation

Of the 366 SLNs examined by TP alone, 349 were interpreted as negative and 17 as positive for metastases. There were 19 false-negative TP results (18 invasive ductal carcinomas and 1 invasive lobular carcinoma). The majority had micrometastasis (11/19, 57.9%) or isolated tumor cells (ITC) (3/19, 15.79%) and the remaining (5/19, 26.3%) had macrometastasis (measuring from 2.5 mm to 6 mm) on permanent sections (figure 2). The majority (68.42%, 13/19) of the SLNs with false-negative TP had therapy-related changes in the residual metastases. One of the 17 SLNs interpreted as positive for metastasis was a false positive. Review of the false-positive lymph node TP revealed reactive changes including multinucleated giant cells and histiocytes from a prior core biopsy site (figure 3). However, the false-positive result in this case was not clinically significant as a separate SLN with macrometastasis was interpreted correctly.

Figure 2

A prior positive lymph node, evaluated by touch preparation alone, revealed patchy cellular lymphoid cells in parts of the smear with giant cells (A). Other parts show paucicellular areas with multinucleated giant cells (inset, high power of the giant cells) (B). Low power view of the permanent section reveals prior biopsy site and partially response to therapy (C). Medium power demonstrates tumour cells embedded in a fibrous background (D).

Figure 3

Touch preparation of a sentinel lymph node reveals numerous multinucleated giant cells that can be misinterpreted as tumour cells if one is unaware of the presence of a prior biopsy clip. (A) Low power magnification and (B) high power magnification.

SLNs evaluated by TP alone showed sensitivity, specificity and PPV of 45.7%, 99.7% and 94.1%, respectively. The sensitivity of TP increased to 72.2% when ITC and micrometastases were excluded from the analysis (table 2).

Table 2

Accuracy of intraoperative evaluation of sentinel lymph nodes by different methods

When analysed at the patient level, the sensitivity, specificity and PPV were 64.3%, 100% and 100%, respectively.

Frozen section

Of the 90 SLNs evaluated in this group, 80 were interpreted as negative and 10 as positive. We found two false-negative results; in one, a 5 mm metastasis associated with prior biopsy site and therapy-related changes and in the second case metastatic tumour clusters in the lymphatic spaces of the capsule of the lymph node (the largest focus measured 0.5 mm) was missed on FS. In the second case, a different SLN was interpreted correctly as positive on FS. There were no false positive results in this group.

Overall, FS alone in the NST setting had a sensitivity, specificity and PPV of 83.3%, 100% and 100%, respectively, at the lymph node level. The sensitivity increased to 88.9% when micrometastases were excluded from the analysis (table 2).

When analysed at the patient level, the sensitivity, specificity and PPV of FS alone were 85.7%, 100% and 100%, respectively.

Touch preparation and frozen Section

Of the 135 SLNs examined in this group, 81 were interpreted as negative and 54 as positive. Thirteen of 81 SLNs were false negative (eight micrometastases, one ITC and four macrometastases that ranged from 5 mm to 10 mm). The majority (11/13, 84.61%) of the false-negative lymph nodes showed therapy-related changes (fewer scattered cells interspersed with fibrosis) on permanent sections (figure 4). Of the 54 SLN called positive on IOE, one was false positive. The false-positive node had a prior core biopsy associated with macrophages and multinucleated giant cells, which was misinterpreted as malignant cells. In the permanent section, no metastasis was identified on H&E stain or with cytokeratin stain.

Figure 4

Sentinel lymph node of a patient after neoadjuvant chemotherapy. (A) Lower power view of a frozen section slide with a subcapsular fibrotic scar that was interpreted as negative for metastasis during intraoperative consultation, (B) retrospective review of the frozen slide reveals a few clusters of tumour cells (arrows) that was overlooked, (C) permanent H&E stained slide reveals multiple nests of tumour cells (circle) embedded in a fibrous stroma, adjacent to the subcapsular scar, (D) cytokeratin AE1/3 stain was performed to determine the extent of the metastasis.

SLNs evaluated by TP and FS had sensitivity, specificity and PPV of 80.3%, 98.6% and 98.1%, respectively. As in the previous groups, the sensitivity was higher (90.9%) when micrometastases were excluded from the analysis (table 2).

Of 72 patients evaluated by TP and FS, the sensitivity, specificity and PPV at the patient level were 80%, 95.2% and 97.3%, respectively.

Cases with atypical or deferred to permanent section

Of the 24 (3.9%) SLNs interpreted as atypical or deferred to permanent, 9 were interpreted as atypical and 15 were deferred to permanent sections because prior SLNs were already reported as positive. Eight atypical SLNs had therapy-related changes or exuberant histiocytic reactions around the prior biopsy site, and one had few tumour cells in a vascular space in the perinodal adipose tissue.

Factors associated with false-negative results in the neoadjuvant setting

We analysed clinical and pathological features in both discordant result groups (false-negative results) and concordant result groups (true-positive and true-negative results). SLNs with false-negative results had a higher frequency of micrometastasis, pretreatment-positive LN biopsy, estrogen receptor (ER)-positive/human epidermal growth factor receptor 2 (HER2)-negative tumour type and lower histological grade (p<0.05). The number of SLNs removed and histologic subtypes were not statistically significant (table 3).

Table 3

Factors associated with intraoperative false-negative results in the neoadjuvant setting at individual lymph node level

Non-neoadjuvant setting

Of the 256 SLNs (66 patients) included in this group, 193 SLNS (40 patients) were evaluated with TP alone, 22 SLNs (7 patients) with FS and 34 SLNS (19 patients) with both TP and FS. Seven SLNs interpreted as atypical or deferred to permanent sections were excluded from the statistical analysis.

Touch preparation

Of the 193 SLNs (from 40 patients) evaluated in this group, 185 were interpreted as negative and eight as positive. Of the 185 negative SLNs, four were false negative (two ITC, one micrometastasis of 1 mm, one macrometastasis of 3 mm). The SLN with 1 mm metastasis contained silicone granulomas and the SLN with 3 mm metastasis had other SLN identified correctly as positive. There were no false-positive results in this group. TP alone in the non-NST setting showed a sensitivity, specificity and PPV of 66.7%, 100% and 100%, respectively, both at the individual lymph node level and patient level.

Frozen Section

Of the 22 SLNs evaluated in this group, 21 were interpreted as negative and 1 as positive. There were no false-negative or false-positive results, therefore the sensitivity, specificity and PPV were 100% both at the individual lymph node level and patient level.

Touch preparation and frozen section

Thirty-four SLNs were evaluated using both TP and FS. Twenty-four were interpreted as negative and 10 as positive. There were no false-negative results. Of the 10 positive SLNs, one lymph node had a cluster of tumour cells on the TP slide only but not on the FS slide or on the permanent section. In this case, technical difficulty in cutting the FS due to a fatty lymph node was documented in the report. TP plus FS in the non-NST setting showed sensitivity, specificity and PPV of 100% both at the individual node and patient level.

Cases with atypical or deferred to permanent section

A total of seven (2.7%) SLNs were interpreted as atypical or deferred to permanent; two were interpreted as atypical (both cases were negative on permanent sections) and five were deferred to the permanent section because prior SLNs were interpreted as positive.

Discussion

A wide range of accuracy reported in the IOE of SLNs both in the non-NST and NST settings can be attributed to factors such as the method used (TP vs FS), proportion of micrometastasis in the study group, the use of immunohistochemistry in the permanent slide and the institutional experience in handling SLNs (both technical and professional components). Despite the variable accuracy, many pathologists, particularly those familiar with cytology preparations, prefer TP as it is technically easier to perform and offers faster turnaround time. Other reasons TP may be favoured over FS is that it is technically challenging to cut fatty lymph nodes, and FS can deplete tissue and, therefore, possibly smaller metastasis.

Our study revealed a lower sensitivity rate when only TP is used for IOE of SLNs in the NST setting (45.7%) compared with the non-NST setting (66.7%), even though the specificity is high in both settings (99.7% in NST and 100% in non-NST). The sensitivity of TP in the NST setting in our study is within the range (38.6% to 87.9 %) reported by other studies.7 8 15–18 Our results for using TP in the non-NST setting are also consistent with a pooled estimated sensitivity of 63% (range from 34% to 95%) and a pooled estimated specificity of 99% (range from 94% to 100%).4 The accuracy of TP detecting metastases irrespective of the prior treatment status is much higher when micrometastases are excluded from the analysis.7 9 However, the clinical implications of missing a micrometastasis during IOE in the primary surgery setting versus neoadjuvant setting are different. For example, results from the ACOSOG Z0011 trial in patients who undergo primary surgery (partial mastectomy) and have 1–2 positive SLNs may not require ALND.19 In contrast, the presence of ITC or micrometastasis after NST is indicative of a larger metastasis that has undergone partial response to therapy and, therefore, based on the current practice, most surgical oncologists would perform an immediate ALND if ITC or micometastasis is detected during IOE. Additionally, we recognised that fibrosis that often develops secondary to treatment in the neoadjuvant setting prevents adequate transfer of tumour cells to the TP slide, leading to decreased sensitivity. Conversely, FS has an advantage, as fibrotic lymph nodes will usually result in a complete section than, for example, a fatty lymph node.

In the NST setting, though we demonstrated a higher sensitivity with FS alone (83.3%) compared with TP (45.7%), when compared with the non-NST setting, the sensitivity of FS was lower in the NST setting (83.3% vs 100%). When compared with the results by previous studies using FS, our FS sensitivity (83% in NST and 100% in non-NST settings) is comparable to the results reported in the NST setting (a range of 71.9% to 93.8%)8–13 and in the non-NST setting (a range of 44% to 100%).5 Furthermore, a high specificity of utilising FS in IOE is extremely important in avoiding unnecessary ALND and associated complications.

In our cohort, the combination of FS and TP showed a sensitivity of 80.3% in the NST setting and 100% in the non-NST setting. A few studies in the non-NST setting have suggested that performing both TP and FS can increase the sensitivity compared with TP or FS alone.3 20 In our study, the sensitivity and specificity for using FS alone versus combined FS and TP are similar, suggesting no advantage of adding TP if FS is planned.

As reported in previous studies, our study also had high false-negative results in the nodes that had micrometastasis after NST.9 11 21 The residual metastasis with partial response to NST often have few residual tumour cells distributed in a larger fibro-inflammatory background that are difficult to recognise on a FS slide. Therefore, it may not be possible to attain the same level of sensitivity in the IOE of SLN after NST when compared with metastases that are not treated prior to surgery (non-NST settings). Additionally, metastasis from invasive lobular carcinoma is known to have high false negative rates. Two studies have also confirmed high false negative rates in the NST setting for invasive lobular carcinoma using TP for IOE of nodes.7 22 In our study, the histological subtypes were not statistically significant as we had fewer cases of invasive lobular carcinoma (only 5% of cases in the NST setting). Nonetheless, the overall sensitivity of detecting metastases after NST can be improved with experience and modifications to the standard operating procedure. Below we share the evolution in our practice and the improvements made in handling SLN in NST settings.

For decades, we have used TP alone for the IOE of SLN(s), although, mostly in the non-NST setting. However, for the last several years, the number of breast cancer patients treated with NST has increased exponentially. Many of our treated patients undergo SLNB at the time of definitive surgery with the rare exception where elective ALND is performed because of minimal or no response to therapy in positive lymph node(s). Consequently, the number of IOE of the SLNs in the NST setting has been increasing. In the initial period, we continued using only TP for IOE in the NST setting. However, the breast and non-breast subspecialty pathologists encountered frequent false negative results using TP alone in the NST setting. A decision was taken then to implement routine use of FS with an option by the individual pathologist to use TP. Currently, our practice has adopted using only FS for all NST cases.

Our experience with IOE of SLNs in the NST setting has also shaped our practice in other ways. Since pCR rates are low after chemotherapy in ER+/HER2− tumours, these cases are examined more carefully with higher suspicion for a persistent positive lymph node. Conversely, patients treated for HER2+ tumours are more likely to achieve pCR and therefore suspicion for a persistent positive lymph node is lower. We have also required the pathology team involved in the IOE of SLNs (pathology assistants and trainees) to obtain relevant information such as history of NST, tumour profile and any prior lymph node biopsy and its result prior to examining a SLN. At our institution, patients who have a prior positive lymph node and have good response to therapy are referred to radiology for localisation of the node by a radioactive seed or another localised marker prior to surgery (this is common practice at our institution for LN retrieval by the surgeon). When several SLNs are received for IOE, gross identification of any prior positive LN containing a biopsy clip or localised marker is prioritised for FS to save time and increase efficiency. If the prior positive node still has residual tumour, one may forgo IOE of the remaining SLNs. However, one ought to be cautious when examining a prior biopsied LN with HydroMARK clip to avoid interpreting exuberant histiocytic reaction and multinucleated giant cells as tumour cells.

The limitations of our study are that it was non-randomised and retrospective; therefore, patient and lymph node characteristics of each group were not controlled. In general, IOE of SLNs in invasive lobular carcinoma cases have lower sensitivity than invasive ductal carcinomas. However, in our study, very few invasive lobular carcinoma cases were included in the NST setting as these tumours are much less likely to undergo neoadjuvant therapy.

In conclusion, while TP and FS are both considered as acceptable methods for IOE of SLNs in the primary surgery setting, FS is recommended in the neoadjuvant setting.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. NA.

Ethics statementsPatient consent for publication

Not applicable.