Current Evidence Does Not Warrant Frozen Section Evaluation for the Presence of Tumor Spread Through Alveolar Spaces Open Access

After approval by our Institutional Review Board, the surgical pathology database at our medical center was searched for resected pulmonary pT1–2 adenocarcinomas that had been diagnosed intraoperatively by FS and resected during the same operation. Specimens from patients with previously treated primary lung tumors, metastatic tumors, and cases in which tumor cells were observed beyond the confines of tissue sections on slide review (indicating probable contamination in the histology laboratory) were excluded. The study cohort included 2 groups of randomly selected cases: tumors resected via video-assisted thoracoscopic surgery (VATS; n = 25) and tumors resected via open thoracotomy (OT; n = 23). The 2 groups were selected to investigate whether the VATS procedure, requiring more tissue manipulation to remove a lung specimen through a considerably smaller space than via OT, resulted in a higher incidence of STAS. Seven different thoracic surgeons performed the lobectomies (n = 27) and sublobar resections (n = 21). If the results provided a suitable data set, survival data were available for analysis from our tumor registry. The adenocarcinomas were reclassified according to the 2015 World Health Organization classification, and the predominant histologic subtype was determined using the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society international multidisciplinary classification of lung adenocarcinoma criteria.^12,13  Using a double-headed microscope, 2 experienced pathologists searched for STAS on hematoxylin-eosin–stained FS slides, slides from the FS control block (FSC), and all other slides with lung tissue adjacent to tumor (AdLT). The criteria proposed by Warth et al,⁷  Kadota et al,⁹  and Ohe et al¹⁴  were used to diagnose STAS, and tumor cells that were attached to alveoli or present within lymphatic or vascular spaces or in the stroma were excluded. Immunostains for cytokeratin 7 (Leica Biosystems Inc, Buffalo Grove, Illinois) and CD163 (Leica Biosystems) were used in selected cases to distinguish tumor cells from clustered alveolar macrophages (Figure). The following features were recorded: presence/absence of STAS, number of detached tumor cell groups observed in alveolar spaces adjacent to the tumor (0 groups, 1–5 groups or >5 groups), and the maximum distance (<3 alveolar spaces or ≥3 alveolar spaces) from the tumor edge to the detached tumor cells in the FS, FSC, and AdLT slides of each case. In cases with more than 1 FS, FSC, and/or AdLT slide, the total number of STAS groups observed in slides of each type and the single greatest distance of STAS from the tumor edge were recorded per case. The STAS incidence, number of groups, and distance from the tumor edge were compared by FS versus FSC versus AdLT, VATS versus OT, and by lobectomy versus sublobar resection using 2-tailed χ² statistics and P values <.05 as significant (software available online at www.Medcalc.org; accessed March 27, 2017). Sensitivity, specificity, and positive and negative predictive values of FS and other slides to diagnose STAS were also calculated.

The specific questions shown in Table 1 were formulated to perform a systematic literature review of the English-language literature and query for the incidence of STAS on FS and its possible prognostic or predictive value, using PubMed (National Library of Medicine), Web of Science, and ScienceDirect. The following search terms were used: “STAS,” “tumor spread through air spaces”, “intra-alveolar tumor spread,” “displaced tumor,” “air space invasion,” “aerogenous tumor,” “aerogenous spread,” “tumor islands,” and “free-floating tumor cells” in conjunction with “lung” and “pulmonary.” The abstracts of all retrieved references were reviewed, and case reports, abstracts, and editorial articles were excluded from the study. The level of evidence of each study was assessed using previously published criteria.^10,11 

The demographics and tumor characteristics of the patients in our study are shown in Table 2. The median age of patients in the VATS group was 5.5 years older than that of patients in the OT group. The VATS group contained a smaller proportion of lobectomies (P = .004) and a larger proportion of pT1–2 N0 lesions than the OT group (P = .001). No significant differences were observed in sex, median tumor size, distribution of predominant histologic subtypes, or STAS incidence between the 2 groups.

Tumor spread through alveolar spaces was seen in association with tumors ranging from 0.6 to 5.4 cm in diameter and from pT1N0 to pT2N2, across all histologic subtypes of adenocarcinoma, and in cases operated by each of the 7 participant thoracic surgeons. Tumor spread through alveolar spaces was identified in 46 of the 48 cases (95.8%) and was present in 23 of 48 FSs (47.9%), 32 of 48 FSCs (66.7%), and 43 AdLTs (89.6%). However, concordant presence of STAS in the FS, FSC, and AdLT from the same case was seen in only 18 of the 46 STAS⁺ cases (39.1%). The incidence of STAS positivity was significantly higher in AdLT than in FSC (P = .007) and FS (P < .001). There was no significant difference in STAS⁺ incidence by FSC versus FS (P = .06), lobectomy versus sublobar resection (P = .11), or VATS versus OT (P = .17). In more than 75% of the 46 STAS⁺ cases, 5 or more tumor cell clusters were seen in alveolar spaces, and/or STAS was present more than 3 alveolar spaces from the tumor edge. No STAS was seen in 25 FSs, but only 2 of these cases were true negatives, with absence of STAS in both the corresponding FSC and AdLT.

Table 3 shows that FS yielded 50% sensitivity, 100% specificity, 100% positive predictive value, and 8% negative predictive value for the detection of STAS. Combining the FS findings with those on the corresponding FSC increased the sensitivity for the diagnosis of STAS from 50% to 76.1% (P < .001) and increased the negative predictive value from 8% to 15.4% (Table 3). Because all but 2 of our 48 cases were STAS⁺, the dataset was not suitable to analyze for possible prognostic or predictive value.

Table 4 shows the results of the systematic literature review. Only retrospective case series, describing fairly heterogenous cases in terms of TNM category and other features, were found, providing only level 4 evidence. The incidence of STAS in T1 to T2 primary lung adenocarcinomas ranged from 80 of 500 cases (16%) to 155 of 411 cases (38%).^3,9  Tumor spread through alveolar spaces was present in 288 of 569 cases (50.6%) in the study by Warth et al,⁷  which included resection specimens from patients with stages I to IV primary lung adenocarcinomas. In the study by Shiono et al,⁴  STAS was observed in 49 of 96 resected pulmonary metastases from patients with colorectal primaries (51%). Several studies as well as editorials note the potential clinical importance of reporting STAS intraoperatively at a time when it could help guide the extent of resection, but none provide any data regarding the incidence of STAS in FS slides or its predictive value.^9,15–18  Indeed, Morimoto et al¹⁷  state that “it is difficult to diagnose STAS in frozen section because the lung is not sufficiently inflated,” and “the diagnosis of STAS is only possible in permanent sections.”

Our retrospective study demonstrates that although STAS was present in 46 of our 48 cases (95.8%), FS could identify its presence in only 23 of these 46 cases (50%). The incidence of STAS detection was higher in FSC slides and AdLT slides than in FS slides of comparable cases, suggesting that future prospective studies evaluating intraoperative FS for the presence of STAS may benefit from the selection of samples that include larger portions of nonneoplastic lung tissue adjacent to the tumor and/or microscopic examination of additional deeper levels prepared from the frozen tissue block. Because almost all cases in our study exhibited the presence of STAS, the data were not suitable for statistical analysis regarding the impact of STAS presence on survival, recurrence rate, or other prognostic and predictive variables. Systematic literature review identified no data on the incidence of STAS in FS slides and no information supporting the intraoperative use of STAS presence as a predictive feature.

We cannot explain the substantially higher incidence of STAS observed in our cases than in previous studies (Table 4). We followed the guidelines published by Warth et al⁷  and Kadota et al⁹  to diagnose STAS and confirmed cells as either epithelial cells or alveolar macrophages with immunohistochemical stains in selected cases. Although the number of cases we studied is small, we attempted to avoid selection bias by randomly choosing cases spanning more than 2 decades that were operated on by several different surgeons who performed lobectomies and variously sized sublobar resections, which were sampled by various pathologists and pathology residents. There was no significant difference in the incidence of STAS by OT and VATS or by surgical procedure (lobectomy, wedge resection, segmentectomy, and trisegmentectomy). These results were somewhat surprising, because we had hypothesized that the VATS group would show a higher incidence of STAS resulting from the detachment of tumor cells during the manipulation required to remove the resected specimen through a smaller chest wall incision. There is limited information about the incidence of STAS by surgical approach or procedure. The surgical approach, OT versus VATS, was mentioned in a study by Shiono and Yanagawa¹⁹  but was not evaluated for its statistical significance. Their study found no significant difference in the incidence of STAS by lobectomy versus segmentectomy or other surgical procedures. Similarly, Onozato et al²⁰  found no significant difference in STAS incidence when they compared wedge resections with “anatomic” resections. More than 75% of our STAS⁺ cases showed 5 or more tumor cell clusters in alveolar spaces and/or STAS 3 or more alveolar spaces from the tumor edge. Warth et al⁷  found no significant difference in overall survival or disease-free survival when they compared patients whose tumors showed “limited” STAS (<3 alveolar spaces from the main tumor) with those whose tumors showed “extensive” STAS (>3 alveolar spaces from the main tumor).

The concept that STAS is a form of intrapulmonary tumor spread rather than an artifact secondary to tissue manipulation during surgery or processing in the pathology laboratory is supported by studies showing that patients whose resection specimens are STAS⁺ have a significantly worse prognosis than those whose specimens lack this finding. However, in the English-language literature we found only retrospective studies of relatively small patient cohorts, providing only level 4 evidence. Moreover, these studies used somewhat different criteria to define STAS⁺, evaluated assorted variables—separately and/or in different combinations (such as patient age, tumor TMN, histologic subtype, extent of STAS, distance of STAS from tumor edge, and tumor proliferation index [Ki-67]), and/or evaluated different prognostic parameters (such as overall survival, disease-free survival, risk of recurrence, and/or time to locoregional, distant, and/or any recurrence). Hence they do not provide comparable data that could be analyzed with meta-analysis. The results are also somewhat variable. For example, some studies have shown an increased risk for tumor recurrence only in patients treated with “limited” (sublobar) resection but not in those undergoing lobectomy, whereas others have shown no significant difference in incidence of recurrence by surgical procedure.^5,8,9,19  Prognostic data regarding the presence of STAS could also have been biased by the histologic subtype(s) of the adenocarcinomas that were included in each study, particularly by the incidence of micropapillary adenocarcinoma, a tumor subtype associated with worse prognosis.^7,8,17,20,21  Morimoto et al¹⁷  showed that in tumors with a micropapillary component, the recurrence-free survival in STAS⁺ tumors was similar to that in patients with pT3 tumors, whereas recurrence-free survival in STAS⁻ tumors was similar to that in patients with pT2 tumors.

In summary, our findings and the lack of substantial information in the literature regarding the intraoperative evaluation of STAS with FS suggest that currently there is insufficient evidence supporting the need for routine reporting of STAS in intraoperative FS. Although pathologists may elect to mention the presence of this feature during intraoperative consultation in cases where STAS is seen in FS slides, they should advise thoracic surgeons that to date there are no prospective data showing that this information is useful to help stratify patients for lobectomy versus sublobar resections.