Air space invasion or aerogenous spread of lung adenocarcinoma is a relatively new concept and has been implicated as a potential prognostic factor as well as has been added as an exclusion point in the diagnosis of minimally invasive adenocarcinoma. Potential role for Ki-67 immunostaining as a diagnostic and prognostic marker in pulmonary carcinoid tumors has been suggested in the literature, given the significant interobserver variability and the difficulty in predicting their clinical behavior.
To review the concept of air space invasion in lung adenocarcinoma and the current controversies regarding the role of Ki-67 immunostaining on pulmonary carcinoid tumors
PubMed search of English literature.
Pathologists need to recognize air space invasion with a critical evaluation to differentiate it from artifacts that are commonly seen in sections. Currently, Ki-67 immunostaining is not recommended for routine use in the diagnosis of pulmonary carcinoid tumors or for predicting their prognosis, except for the differential diagnosis from small cell carcinomas or large cell neuroendocrine carcinomas in small biopsy specimens with crush artifacts.
Air space invasion or aerogenous spread of tumor is a relatively new concept and has become increasingly important for pathologists to recognize. Several terms including tumor islands (TIs) and spread through air spaces (STAS) have been used to describe air space invasion by the tumors.1,2 Recent studies2–5 proposed to include air space invasion as a factor for upstaging in squamous cell carcinomas as well as in adenocarcinomas of the lung, given the worse recurrence-free survival (RFS) in those with STAS. The presence of STAS is an exclusion point in the diagnosis of minimally invasive adenocarcinoma (MIA), a small, solitary adenocarcinoma (≤3 cm) with predominantly lepidic pattern and invasion extending 5 mm or less.6,7 MIA is now an official category adopted in the 2015 World Health Organization (WHO) classification, which can be potentially treated with a limited resection.7
Neuroendocrine tumors (NETs) of the lung are divided into carcinoid tumors, small carcinoma, and large cell neuroendocrine carcinoma (LCNEC). While Ki-67 index has been incorporated into the grading system of NETs occurring in the gastrointestinal tract and the pancreas since the 2010 WHO classification, the diagnostic or prognostic role of Ki-67 in pulmonary NETs has been debatable.
This review is focused on the current concepts and controversies on air space invasion in lung adenocarcinomas and on the role of Ki-67 immunostaining in pulmonary carcinoid tumors.
AIR SPACE INVASION IN LUNG ADENOCARCINOMAS
Invasion in lung adenocarcinoma is defined as (1) the presence of nonlepidic patterns such as acinar, papillary, solid, or micropapillary; (2) infiltration of stroma; and (3) lymphovascular or pleural invasion.6 Owing to the unique lung anatomy with abundant air spaces surrounded by capillary network in the alveolar septa, lung tumor cells may undergo aerogenous spread without parenchymal destruction or subsequent stroma reaction such as desmoplasia or angiogenesis. Aerogenous tumor spread is now an increasingly recognized pattern of tumor invasion especially in lung adenocarcinomas and defined as the presence of tumor cells within air spaces in the lung parenchyma beyond the outer edge of the main tumor.2,7 Detached intraalveolar tumor cells are very commonly seen within the main tumor and should be differentiated from the air space invasion, which refers to the tumor cells that spread via air spaces in the lung parenchyma surrounding the outer edge of the tumor, but not within the main tumor.
Jin et al8 showed that frequent aerogenous spread with decreased E-cadherin expression in ROS-1 rearranged lung cancer predicts poor disease-free survival. This study showed a high association of their ROS-1–rearranged tumors with micropapillary component (P < .001), aerogenous spread (P = .002), and E-cadherin loss (P = .049). It suggested that decreased membranous E-cadherin expression and aerogenous spread may be associated with worse disease-free survival. This study suggests, albeit indirectly, that loss of E-cadherin expression might be associated with aerogenous spread.
Recent studies2,9,10 have shown that TIs and STAS have strong prognostic implications for recurrence and patient survival, especially among small early-stage tumors undergoing limited resection. Tumor islands refer to large collections of tumors cells isolated within alveolar spaces and can be seen in lung adenocarcinomas.1 While the term tumor islands was introduced first, STAS has been used more widely in the literature and practically became the term interchangeably used for any form of air space invasion (including TIs) in all types of lung cancers. Given the fact that the studies on TIs first introduced the concept of air space invasion in lung adenocarcinomas, the initial studies on TIs are reviewed first, followed by subsequent studies using the term spread through air spaces.
In 2012, Onozato et al1 evaluated the role of 3-dimensional reconstruction of histology sections in the classification of lung adenocarcinoma, a technique that was not novel, but has not been widely used in the clinical setting because of its cumbersome nature. They were able to reconstruct the 3D models of resected lung adenocarcinomas. Using this methodology, they assessed the cases with large nests of tumor cells (ie, TIs) located within the alveolar spaces and detached from the main tumor. The interest in these island-like structures began from the observation that TIs within the air spaces do not fit the definition of micropapillary pattern. Since the International Association for Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS) classification does not identify TIs as a pattern, they may have been classified as a micropapillary component or considered as an artifact secondary to surgical manipulation and/or processing. Remarkably, the 3D reconstruction study demonstrated that these TIs were interconnected, with the main tumor mass at multiple points, suggesting that TIs represent a form of tumor extension through air spaces. Importantly, these points of connection between the TIs and the main tumor are not apparent in conventional 2D histologic assessment, which was well reiterated by a subsequent publication by the same group.9
In a follow-up study by this group, Onozato et al10 evaluated 261 patients with stage I or II lung adenocarcinomas who underwent surgical resection with curative intent in order to assess the clinical and prognostic significance of TIs. In this second study, TIs were identified in 57 patients (22%) of the cohort (n = 261) and ranged from 35 to 366 μm in greatest dimension with a mean of 155 μm. There was also a wide variation in the number of TIs observed in a given tumor, from 2 to 29. Interestingly, the presence of TIs, irrespective of their numbers, was significantly associated with adverse clinicopathologic features: positive smoking, a solid predominant pattern of growth, and higher nuclear grade. There was a trend toward a higher proportion of micropapillary predominant tumors and a lower proportion of lepidic predominant tumors in those with TIs. Importantly, in this study only 3.5% of tumors with TIs were classified as MIA, as opposed to 14.3% of those without. It would have been an exclusion criterion against the diagnosis of MIA, if one followed the current WHO classification criteria, however.7 The presence of TIs was strongly associated with KRAS mutation. The TIs were identified in 46% (17 of 37) of KRAS mutants, but not in any of the EGFR mutants.10 Importantly, patients with the adenocarcinoma with TIs had significantly worse outcomes than those without, for stage I to II tumors as well as stage IA tumors alone.10 For patients who underwent isolated wedge resection, early recurrence after resection was associated with the presence of TIs, while the 5-year RFS was not significantly different between those with and without TIs. Thus, this study proposed aggressive surveillance and/or further intervention for the patients whose tumors exhibit TIs.10
Spread Through Air Spaces
STAS may occur with micropapillary clusters, solid nests (ie, TI), or single cells. Because this represents a manifestation of tumor spread, it is not included in the percentage measurements of subtype patterns in the current guidelines.7 However, STAS is now recognized as a form of invasion and listed as one of the exclusion criteria along with the presence of visceral pleural invasion, tumor necrosis, and lymphovascular invasion in the diagnosis of MIA.7
Kadota et al2 reviewed 411 cases of resected small stage I lung adenocarcinomas (≤2 cm) during the 1995–2006 period. Tumor STAS was evaluated, and estimated risk of disease recurrence and its association with clinicopathologic risk factors were studied. STAS was found in 155 of 411 cases (38%). In their 120 cases of limited resection group, the risk of recurrence was significantly higher in the patients with STAS-positive tumors than in the patients with STAS-negative tumors. There was no association between STAS and recurrence in the lobectomy group (n = 291), however. In a multivariate analysis, the presence of tumor STAS remained independently associated with the risk of developing recurrence in limited resection group (hazard ratio, 3.08; P = .01). Based on these results, they concluded that the presence of STAS is a significant risk factor of recurrence in small lung adenocarcinomas treated with limited resection.
Dai et al5 reported that tumor STAS affects the recurrence and overall survival in patients with lung adenocarcinoma greater than 2 to 3 cm. They suggested that STAS could be considered as a factor in the staging system to predict prognosis more precisely in those with pT1b (according to the American Joint Committee on Cancer's AJCC Cancer Staging Manual, 7th edition).11 Uruga et al4 showed that one-third of resected small lung adenocarcinomas had high STAS. They purported higher STAS was predictive of worse RFS from their semiquantitative study using the surgically resected small pT1a stage I lung adenocarcinomas during 2003–2009 at a single institution. They defined low STAS as 1 to 4 single cells or clusters of STAS and high STAS as 5 or more single cells or clusters of STAS by using an ×200 field. They found no STAS in 109 of 208 cases (52.4%), low STAS in 38 cases (18.3%), and high STAS in 61 cases (29.3%). There was significant association between increasing STAS and shorter RFS in multivariate as well as univariate analysis. Kadota et al3 recently also showed that tumor STAS is an independent predictor of RFS in patients with resected lung squamous cell carcinoma.
While the importance of STAS has been documented in many studies, one cannot overemphasize the fact that differentiation between STAS and various forms of artifacts (eg, floater, carryover, tangential section) might be extremely difficult and virtually impossible in some cases. A recent prospective study of loose tissue fragments in non–small cell lung cancer resection specimens purported that STAS might represent a mechanical artifact.12 In that multi-institutional study, they showed that up to 93% of the loose tissue fragments could be explained by mechanical forces associated with tissue handling caused by spread through a knife surface. This study postulated that STAS may be an ex vivo artifact, mainly seen in more discohesive poorly differentiated tumors, and also argued that STAS may be simply an epiphenomenon, offering a biologic explanation rather than representing a true novel pattern of tumor invasion. This view requires more validation by independent studies but would be a strong reminder that one should be always mindful about differentiating STAS from any artifacts (Figures 1 through 3). It is already very challenging to handle potential MIA cases during intraoperative frozen section practice. Evaluation of STAS will add another layer of difficulty in the evaluation of possible MIA cases during frozen section, given the fact that the presence of STAS may become a potential decision point between limited resection and lobectomy.
Ki-67 LABELING INDEX IN PULMONARY CARCINOID TUMORS
Neuroendocrine tumors of the lung include carcinoid tumors, small cell carcinomas, and LCNECs. Though categorized as a group of NETs of the lung, small cell carcinomas and LCNECs of the lung share little or no common clinicopathologic features with pulmonary carcinoid tumors, which would argue against a classification system of grade 1 to 3 as in gastroenteropancreatic (GEP) NETs. Furthermore, the terms carcinoid tumors and small cell carcinomas have been well-established terminologies in the literature that allow good communication among pathologists, clinicians, and biologists. On the other hand, interobserver variability for the WHO classification of pulmonary carcinoid tumors has been relatively high.13 Thus, additional biomarkers such as Ki-67 or other immunomarkers may be needed to improve consistency of classification and prediction of prognosis.13
Carcinoid tumors of the lung are divided into 2 categories according to the current WHO criteria: (1) Typical carcinoids (TCs) are carcinoid tumors at least 0.5 cm or larger in size with fewer than 2 mitoses per 2 mm2, and lacking necrosis. (2) Atypical carcinoids (ACs) are carcinoid tumors with 2 to 10 mitoses per 2 mm2 and/or foci of necrosis.14 The terms used in the GEP NETs as well-differentiated or moderately differentiated or grade 1 or 2 NET, are not recommended to be used for pulmonary NETs. Currently, Ki-67 proliferation index, either by manual examination or digital image analysis of Ki-67 immunohistochemistry, is not implemented for classification, in contrast to the GEP NETs that are classified as grade 1, 2, or 3, on the basis of mitotic count and Ki-67 labeling index (LI).
There have been a few studies proposing to use Ki-67 in classification of pulmonary NETs.15,16 Clay et al15 examined the diagnostic and prognostic values of the Ki-67 LI in pulmonary carcinoid. They studied 94 consecutive patients with a confirmed diagnosis of TC (n = 75) or AC (n = 19) during a 14-year period at a single institution and found that the combination of mitotic index, Ki-67, and necrosis led to the classification of NETs into 4 different prognostic groups (very low, low, intermediate, and high risks of relapse). They proposed the incorporation of Ki-67 LI in the prognostic classification of pulmonary carcinoid tumors, based on their result that Ki-67 LI and mitotic index have continuous effect on prognosis. They also suggested that prognostic models incorporating multiple cutoffs of Ki-67 and mitotic index might better predict outcome and inform clinical decisions.
Pelosi et al17 did an extensive review for the 5 relevant issues to Ki-67 LI in practice, addressed by using a question-answer methodology, with relevant key points discussing major interpretation issues. Their conclusion was that Ki-67 is a feasible and potentially meaningful marker in lung NETs but more data are needed to determine its ideal function in this setting of tumors.
Technical Issues to Ki-67 Immunohistochemistry and Evaluation of Results
Currently, there is no uniform methodology for Ki-67 immunohistochemistry (IHC) and evaluation of results, but most studies used MIB-1 clone on paraffin sections after antigen retrieval procedures optimized within each laboratory and the assessment of a Ki-67 LI. Unlike the GEP system, there have been no comparative studies that evaluated different methods for Ki-67 results in lung NETs. However, most published studies measured Ki-67 LI in hot spots, after visual scrutiny of the entire tumor. This would apply especially to TC or AC, whereas Ki-67 decoration is usually much more uniform in small cell carcinomas and LCNECs. For practical purposes, Ki-67 LI should be calculated in surgical specimens by counting at least 2000 consecutive tumor cells in hot spot fields at ×40 lens or 2 mm2 for consistency with histologic classification, possibly in the same tumor areas as those used for assessing mitotic count. Biopsy or cytology samples may not fulfill the 2000 cells or the 2 mm2 criterion, given the limited number of tumor cells or areas present on the sections. In such cases, it could be reasonable to calculate Ki-67 LI on all evaluable tumor cells on the sections. Additional work and reproducibility studies are needed to address the optimal procedure for evaluating Ki-67 in lung NETs.
Diagnostic Role for Ki-67 LI in Lung NETs
It is generally agreed that there is no diagnostic role per se for Ki-67 LI in lung NETs. However, it could be very useful in separating TC and AC from high-grade NETs in small biopsy specimens with crush artifacts.18
Possible Prognostic Role for Ki-67 LI in Excised Specimens for TC and AC
Role for Ki-67 LI in Tumor Grading
There is no established role for Ki-67 LI in tumor grading, and mitotic count is the most important histologic parameter to grade lung NETs.
Predictive Role for Ki-67 LI in Therapeutic Decisions
No predictive role for Ki-67 is known in therapeutic decisions. There have been no randomized trials documenting an established role for Ki-67 in lung NETs to guide therapy.
While there is no clear role for Ki-67 as either a diagnostic or predictive marker currently, given the insufficient data for clinical use, pathologists will continue to confront the requests for Ki-67 IHC from many directions. Thus, it would be important for practicing pathologists to be familiar with the current status for effective communication with oncologists or other involved clinicians. Conflicting results may be stemming from several factors, including selection of patients, number and type of tumors under evaluation, histologic criteria used for classification, variability in the choice of antibodies and immunostaining protocols, Ki-67 LI cutoff thresholds, assessment criteria (eg, automated analysis, manual counting, eyeball estimation, field and cell selection, number of analyzed cells), length and accuracy of follow-up, and/or clinical parameters under evaluation, which may have prevented direct cross-study comparisons.17,21 More studies are warranted to establish a lung-specific and clinically meaningful grading system based on Ki-67 LI, alone or in combination with other parameters, with defined cutoff thresholds and uniform procedures for assessing Ki-67 LI.
One other important issue in pulmonary NET classification system is that there are some tumors, albeit uncommon, showing carcinoid morphology but with mitotic rate exceeding 10 mitoses/2 mm2. The current WHO scheme recommends such tumors to be classified as LCNECs but their clinical and pathologic features are not well known. Quinn et al23 studied 12 cases of such borderline tumors and found that those tumors have more in common with the carcinoid tumor group than LCNECs. This result is not entirely surprising in that the median mitotic count for LCNECs is approximately 70 per 2 mm2; the cases showing carcinoid morphology with much lower mitotic rate (eg, <30 per 2 mm2) would technically belong to the high-grade NET category (ie, small cell carcinoma or LCNEC). However, those cases might be different in their clinical behavior from that of high-grade NETs including small cell carcinoma and LCNEC. Thus, an additional category might be needed for this in-between tumor that has higher mitotic count, unacceptable for AC but still more in keeping with the morphology of AC than LCNEC or small cell carcinoma.
A similar problem as in lung NETs showing carcinoid morphology with high mitotic rate (presumably with increased Ki-67 LI) has been recognized in pancreatic NETs. WHO grade 3 pancreatic NET category is morphologically and biologically heterogeneous and includes both well-differentiated and poorly differentiated neoplasms. Basturk et al24 demonstrated that the mitotic rate and Ki-67 LI-based grades of pancreatic NETs can be discordant. They demonstrated that well-differentiated pancreatic NETs that are grade 3 by Ki-67 LI are significantly less aggressive than bona fide poorly differentiated NECs showing carcinoma morphology, suggesting that the current WHO grade 3 category in pancreatic NETs is heterogeneous and contains 2 distinct neoplasms. Thus, they proposed to further separate these entities into well-differentiated NET with an elevated proliferation rate and poorly differentiated neuroendocrine carcinoma.
A body of literature embraced the concept of air space invasion as a new form of invasion in lung cancers mainly in adenocarcinomas but also in other types of carcinomas including squamous cell carcinomas, provided that a possibility of artifact is excluded by a careful and critical evaluation. Tumor islands was introduced first as a form of air space invasion by an elaborate 3D study but this term has been subsequently absorbed by a newer term, spread through air spaces, that encompasses solid nests (ie, TIs) as well as micropapillary and single tumor cells in the air spaces. STAS practically became the official terminology that is now almost exclusively used in the literature. STAS has been associated with a significant increase in recurrences after limited resection and is now listed as an exclusion criterion in the diagnosis of MIA. Thus, recognition of STAS would be crucial during intraoperative frozen section practice, which poses a significant challenge to practicing pathologists.
Ki-67 LI is often requested by clinicians for lung carcinoid tumors, given its established role in the WHO grading system of GEP NETs since 2010. However, Ki-67 is not recommended to be used either as a diagnostic or prognostic marker in pulmonary carcinoid tumors given the insufficient data available in the literature. Ki-67 might be still useful in the differential diagnosis between carcinoid tumors and high-grade NEC of the lung such as small cell carcinoma and LCNEC, especially in small biopsy specimens with crush artifacts. More studies may be warranted to establish a lung-specific and clinically meaningful grading system based on Ki-67 LI, alone or in combination with other parameters, with defined cutoff thresholds and uniform procedures for assessing Ki-67 LI.
The authors have no relevant financial interest in the products or companies described in this article.
This article was presented in part at the 16th Spring Seminar of Korean Pathologists Association of North America (KOPANA); March 2–4, 2017; San Antonio, Texas.