Hans Blaauwgeers

HISTOPATHOLOGICAL ASPECTS OF RESECTED NON-SMALL CELL LUNG CANCER with emphasis on Spread Through Air Spaces and collapsed Adenocarcinoma in Situ Hans Blaauwgeers

Histopathological aspects of resected non-small cell lung cancer, with emphasis on spread through air spaces and collapsed adenocarcinoma in situ Hans Blaauwgeers

Histopathological aspects of resected non-small cell lung cancer, with emphasis on spread through air spaces and collapsed adenocarcinoma in situ Thesis, Free University, Amsterdam, The Netherlands Provided by thesis specialist Ridderprint, ridderprint.nl Printing Ridderprint Layout and design Indah Hijmans, persoonlijkproefschrift.nl Cover design Indah Hijmans ISBN 978-94-6483-237-2 DOI http://doi.org/10.5463/thesis.317 Financial support for the production and printing of this thesis was kindly provided by the Stichting Wetenschappelijk Onderzoek OLVG (SWOO) Copyright © Hans Blaauwgeers

Table of contents Chapter 1 General introduction and outline of the thesis 9 Part I: Histopathologic factors in the outcome and prognosis of NSCLC Chapter 2 Complete pathological response is predictive for clinical outcome after trimodality therapy for carcinomas of the superior pulmonary sulcus 19 Chapter 3 The prognostic value of proliferation, PD-L1 and nuclear size in patients with superior sulcus tumours treated with chemoradiotherapy and surgery. 35 Chapter 4 A population-based study of outcomes in surgically resected T3N0 nonsmall cell lung cancer in the Netherlands, defined using TNM-7 and TNM-8; justification of changes and an argument to incorporate histology in the staging algorithm. 47 Part II: Spread through air spaces, an artifact or reality? Chapter 5 Ex vivo artifacts and histopathologic pitfalls in the lung 65 Chapter 6 A prospective study of loose tissue fragments in non-small cell lung cancer resection specimen. An alternative view to “spread through airspaces” 83 Chapter 7 Pulmonary loose tumor tissue fragments and spread through air spaces (STAS): invasive pattern or artifact? A critical review. 95 Chapter 8 To the editor: “Spread through air spaces (STAS) is prognostic in atypical carcinoid, large cell neuroendocrine carcinoma, and small cell carcinoma of the lung” 107 Chapter 9 To the editor: “Spread through air spaces (STAS); can an artifact really be excluded?”. 111 Chapter 10 Loose tumor cells in pulmonary arteries of lung adenocarcinoma resection specimen; no correlation with survival, despite high prevalence. 115 Chapter 11 Is Spread Through Air Spaces an in vivo phenomenon or an inducible artifact? 131

Part III: Collapsed adenocarcinoma in situ; the challenge of defining invasion Chapter 12 Elastin in pulmonary pathology: relevance in tumors with lepidic and/or papillary appearance. A comprehensive understanding from a morphological viewpoint. 147 Chapter 13 Incorporating surgical collapse in the pathological assessment of resected adenocarcinoma in situ of the lung. A proof of principle study. 165 Chapter 14 3-dimensional reflection aids in understanding the fog of 2-dimensional pattern recognition in pulmonary adenocarcinomas. Proposal for a modified classification. 173 Chapter 15 Invasion measurement and adenocarcinoma in situ (AIS) in pulmonary adenocarcinomas of 3 cm or less. A reproducibility study according to the WHO and a modified classification. 201 Chapter 16 Arguments for non-mucinous adenocarcinoma in situ of the lung larger than 7 cm. 235 Chapter 17 General discussion and future perspectives 251 Appendices Supplementary files 276 References 284 Summary 308 Nederlandse samenvatting (Dutch summary) 314 List of publications 320 Dankwoord (Acknowledgements) 326 Curriculum Vitae 329 CRediT authors statements 330

1 General introduction and outline of the thesis

10 Chapter 1 Introduction Lung cancer is the leading cancer type among men worldwide and, after prostate and skin cancer, the most common malignancy in the Netherlands. In recent decades, there has been an increase in the incidence of lung cancer among women. Each year, over 14,000 men and women in the Netherlands are diagnosed with lung cancer, that includes non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). Annually, about 10,000 patients die from their disease1. The 5-year survival rate is relatively low at 17%, although small improvements have been observed recently2. Pathology plays a crucial role in the diagnosis, staging, and treatment of NSCLC, which is the most common type of lung cancer (70%). For an adequate diagnosis, pathologists use, beside cytology specimen, tissue samples obtained through biopsy or surgical resection to generate a diagnosis of lung cancer. The morphologic examination includes examination of cells and the structural relation of these cells. To this end, the initial diagnostic step is the conventional hematoxylin and eosin (H&E) staining and is often employed along with supplemental histochemical and/or immunohistochemical stainings. This pathological analysis is critical in determining the type of lung cancer and in differentiation from metastases to the lung, derived from other organs. At the time of diagnosis of NSCLC, the extent of the disease, i.e., the stage is determined. Pathological examination can be supportive in determining the size of the tumor, and depending on where the sample is taken from, it can also help in assessing the presence of metastases, either locoregional in lymph nodes or at distant places for instance in other organs. Pulmonologists and oncologists need this information in determining the stage of the disease and treatment plan. Classification and staging The most prevalent type of lung cancer is NSCLC, which constitutes 70% of all pathologically diagnosed cases. Treatment for lower stage NSCLC involves a combination of surgery and other therapies. SCLC accounts for 11% of cases and is typically already metastasized at the time of diagnosis, ruling out surgical intervention. Carcinoids are rare, comprising only 1% of cases. The remaining 17% of clinically diagnosed cases are suspected to be early-stage NSCLC and are referred for treatment by stereotactic radiotherapy without a pathologic diagnosis2. The recognition of the diversity of NSCLC has led to further subclassification, which was published in the 2004 and 2015 World Health Organization (WHO) classification3, 4. The major types of NSCLC include adenocarcinoma (AdC), squamous cell carcinoma (SqCC), and large cell carcinoma (LCC). Adenocarcinomas can be further subdivided into the most common non-mucinous variant and the mucinous variant. The WHO classification of non-mucinous lung adenocarcinoma in 2015 states that the predominant pattern observed determines the subtype of the tumor4. This assessment

11 Introduction and aim of thesis is based on a combination of microscopic examination, immunohistochemical analysis, and molecular/genetic analysis. Regarding NSCLC, several histopathological criteria have been linked to prognosis. For example, adenocarcinomas as a type, is associated with a better prognosis than the squamous cell carcinoma type of NSCLC5. Within the adenocarcinomas some subtypes, such as micropapillary and solid are associated with a worse prognosis6, 7. Also within the subtypes, certain patterns are associated with worse prognosis, such as the cribriform variant of acinar adenocarcinoma8. In 1960, Liebow introduced the term “bronchioloalveolar carcinoma” (BAC) to describe a well-differentiated form of adenocarcinoma that presents with one of three growth patterns: (1) a single nodular pattern, (2) a multinodular pattern, or (3) a diffuse pneumonic pattern9. However, the definition of bronchioloalveolar carcinoma remained somewhat unclear in terms of pathological criteria. As radiology, histology, and immunophenotyping knowledge advanced, it became increasingly crucial to distinguish certain subtypes of non-small cell lung cancer with a better prognosis and less invasive treatment from the conglomerate of cases. In 1995, Noguchi et al. identified subtypes that exhibited 100% long term survival. Type A demonstrated growth through replacement of alveolar lining cells, accompanied by minimal or slight thickening of the alveolar septa, and an absence of fibrotic foci within the tumors, called localized bronchioloalveolar carcinoma (LBAC). Type B followed a comparable pattern, but fibrotic foci due to alveolar collapse were evident in the tumors and was called bronchioloalveolar carcinoma with foci of collapse of alveolar structure10. In 1997, Silver and Askin examined cases that were diagnosed as having papillary or bronchioloalveolar characteristics. They identified certain cases with a worse prognosis when stricter criteria were applied, such as over 75% papillary pattern, complicated papillary growth in secondary and tertiary branches, greater cytonuclear atypia, and distortion or destruction of underlying pulmonary architecture11. However, in the 1999 edition of the WHO classification, the papillary subtype is not so strictly defined, namely as “an adenocarcinoma with a predominance of papillary structures that replace the underlying alveolar architecture”12. In the 2015 edition of the WHO classification, the term bronchioloalveolar carcinoma (BAC) was abandoned, and instead, the concepts of adenocarcinoma in situ (AIS) and minimally invasive adenocarcinoma (MIA) were introduced. These new concepts have been increasingly diagnosed, particularly in lesions discovered through lung cancer screening13. When surgically removed, both AIS and MIA exhibit nearly 100% 5-year survival rates. A precise and accurate diagnosis distinguishing non-invasive adenocarcinomas such as AIS and MIA from invasive carcinomas is crucial. This is because invasive adenocarcinomas may require additional therapy, while non-invasive non-mucinous adenocarcinomas can be cured through surgical resection alone, eliminating the need for further treatment. 1

12 Chapter 1 In 2012 a reproducibility study on simple power point images of different adenocarcinoma subtypes showed that one group of pathologists consistently judged a subset of adenocarcinomas to be invasive (Invasive Group, ING), while another group of pathologists consistently judged the same subset to be non-invasive (non-ING)14. This clearly points toward a necessity for calibration or re-evaluation of the diagnostic criteria. The concept of “loose tumor tissue fragments” as a pattern of invasion in lung carcinoma has recently been proposed and is included in the 2015 WHO fascicle on the classification of lung tumors, and was specifically called ‘spread through air spaces” or STAS4. This inclusion is at least somewhat controversial, as an alternative explanation is that “loose tumor tissue fragments” represents an artifact. In the current WHO classification of lung cancer “STAS” is a criterium of invasion. The difference in consequences of both concepts warrants further research. The main function of staging lung cancer, is to determine the extent of disease at time of diagnosis, because of the associated prognosis. The staging system for cancers is based upon anatomic primary tumor characteristics, lymph node metastasis, and distant metastasis (TNM) at time of diagnosis and has been developed to predict survival outcome in cancer patients. Stage is the main denominator for treatment planning in patients with lung cancer NSCLC15. As for many other tumors, the descriptors of the T, N and M for NSCLC evolve with time. The most recent transition from the 7th (TNM7) to the 8th (TNM-8) editions dates in 201716 17 18. The accurate determination of the maximum diameter of the invasive part of the tumor is crucial for pathologists with regards to the T-descriptor. This is because the 8th edition of the Union for International Cancer Control (UICC)/American Joint Committee on Cancer (AJCC) TNM classification system for NSCLC recommends that the measurement of the primary tumor’s size should be based only on its invasive components19. This emphasizes the need of precise and accurate diagnosis of a non-invasive (AIS and MIA) as opposed to an invasive carcinoma. Another aspect of staging is the examination of the response after neoadjuvant therapy in resection specimen. This may provide insight in the assessment of the efficacy of neoadjuvant treatment in NSCLC. Aim and outline of the thesis This thesis investigates three histopathological aspects in resected NSCLC specimens, relevant to diagnosis, prognosis, and staging. Firstly, we aimed to establish criteria for the pathologic response effect after neoadjuvant therapy by examining whether histopathological findings such as the percentage of tumor rest, proliferative activity, and morphometric properties may lead to a refinement in the prognosis of patients with NSCLC after chemoradiation therapy.

13 Introduction and aim of thesis Secondly, we sought to induce the phenomenon of “STAS” by handling the gross specimen, to investigate whether it is plausible that “loose tumor fragments” interpreted as “STAS” is not a biological way of metastasizing, but rather an artifact. We investigate the potential association between possible artifacts such as spreading tumor cells or clusters caused by mechanical knife force or collapse of lung tissue, and classification, staging, or prognosis. Thirdly, we aimed to refine diagnostic tools that can aid in distinction between invasion and non-invasion in non-mucinous pulmonary adenocarcinoma. To achieve this, we examined the impact of iatrogenic collapse on pulmonary adenocarcinomas and its effect on invasion assessment and prognosis. Part I of this study, comprising chapters 2-4, focuses on describing the histopathologic findings in resected specimens of patients with NSCLC, with the aim of refining pTNM staging. In chapter 2, we try to describe the histological changes after neoadjuvant therapy, analyzing a study cohort of 46 patients with sulcus superior tumors who received chemoradiation treatment followed by surgical resection. In chapter 3, we try to identify additional morphologic prognostic characteristics of residual tumors in this patient group, and examined whether proliferation, PDL-1, and nuclear size after chemoradiation in comparison to pretreatment measurements are related to prognosis. The conversion from TNM-7 to TNM-8 harbors a shift of some T-descriptors to another category. Chapter 4 presents the results of a nationwide study on T3N0 NSCLC, initiated in preparation of chapter 16, which encompasses various tumor categories, including those with parietal pleura invasion or a diameter exceeding 7 cm. The aim of the study in chapter 4 was twofold: to assess the validity of this shift in the Dutch population and to investigate whether the inclusion of additional morphologic factors could improve staging accuracy. Part II concerns with the phenomenon of Spread Through Air Spaces (STAS). In the chapters 5 – 11 we focus on this phenomenon, that in the most recent WHO fascicle is classified as a new way of metastases, but is possibly an artifact. Chapter 5 describes in general several histopathologic artifacts related to specimen handling, such as spreading (tumor) tissue through a knife, while cutting the resected specimen. Chapter 6 describes a prospective multicentered study on the investigation of the phenomenon of STAS being an inducible artifact or a new way of metastasizing. Chapter 7 is a critical review on the subject of STAS and in chapter 8 and 9 we comment on this issue, related to a study on neuroendocrine tumors of the lung and to the findings in a comparable study to ours as described in chapter 6. In chapter 10 we examine another possible artifact, namely the presence of individual tumor cells and tumor cell clusters in pulmonary artery branches in histologic sections of pulmonary resection specimen. Chapter 11 summarizes the most recent arguments about STAS as an artifact in the CON part of a Pro-Con editorial. 1

14 Chapter 1 In part III with the chapters 12-16, we focus on iatrogenic and biological collapse as a possible pitfall in the assessment of invasion in small adenocarcinomas. In chapter 12 we describe the role of elastin in pulmonary pathology and its possible usefulness in the recognition of non-invasive patterns. Chapter 13 discusses a proof-of-principle study that explores the feasibility of considering surgical collapse when diagnosing non-invasive cases of lung cancer. This study examines whether the use of cytokeratin 7 as an immunohistochemical marker can facilitate the recognition of surgical collapse and enable the diagnosis of more non-invasive cases with excellent prognoses. Chapter 14 details the morphological characteristics of iatrogenic and biologic collapsed AIS, incorporating lessons learned from a mathematical model and utilizing cytokeratin 7 and elastin staining to assess invasiveness. Two independent cohorts of resected adenocarcinomas measuring 3 cm or smaller were examined to investigate these aspects and their relationship to recurrence-free and overall survival was analyzed. Chapter 15 concerns a large international interobserver study, determining whether incorporating iatrogenic and biological collapse in the assessment of lung adenocarcinoma invasion using elastin and cytokeratin 7 stainings could diagnose (collapsed) AIS and reduce the interobserver variation of invasive patterns in small pulmonary adenocarcinomas. According to the WHO for AIS has a maximum size limit of 3 cm. We wondered whether AIS may occur as a larger tumor. In Chapter 16, we examined in the nationwide subgroup of T3N0 NSCLC for the possible presence of AIS with a size larger than 7 cm.

15 Introduction and aim of thesis 1

Part I Histopathologic factors in the outcome and prognosis of NSCLC

Hans Blaauwgeers, Ingrid Kappers, Houke Klomp, José Belderbos, Lea Dijksman, Egbert Smit, Pieter Postmus, Rick Paul, Jan Oosterhuis, Koen Hartemink, Cornelis Vos, Sjaak Burgers, Max Dahele, Erik Phernambucq, Birgit Lissenberg-Witte, Erik Thunnissen Virchows Archives (2013) 462:547–556 DOI 10.1007/s00428-013-1404-6 2 Complete pathological response is predictive for clinical outcome after tri-modality therapy for carcinomas of the superior pulmonary sulcus

20 Chapter 2 Abstract The objective was to define the relationship between histopathologic changes after pre-operative chemo-radiotherapy (CRT) and clinical outcome following tri-modality therapy in patients with superior sulcus tumors. A retrospective analysis of tumour material was performed in a series of 46 patients who received tri-modality therapy between 1997 and 2007. Median follow-up was 34 months (5 – 154). Pathologic complete response (pCR) was present in 20/46 tumors (43%). The most common RECIST score after CRT in patients with pCR was a partial response (PR; 10/17, 3 unknown) whereas in patients without a pCR, stable disease was the most common (22/26) (p=0.002). In 26 specimens with residual tumour, this was mainly located in the periphery of the lesion rather than the center (Spearman’s correlation - 0.67, p<0.001). Prognosis was significantly better after a pCR compared to residual tumour (70% 5-years overall survival vs. 20%; p=0.001) and in patients with fewer than 10% vital tumour cells as compared to those with >10% (65% 5-years overall survival vs 18%; p<0.001). A low mitotic count was associated with a longer disease-free survival (p=0.02). Complete pathological response and the presence of fewer than 10% vital tumour cells after preoperative CRT are both associated with a more favorable prognosis. A modification of the pathological staging system after radiotherapy, incorporating the percentage of vital tumour cells, is proposed.

21 Outcomes of resected SST after CRT Introduction In patients with clinically resectable Non-Small Cell Lung Cancer (NSCLC), the survival rates vary, ranging from an estimated median survival of 19 months for stage IIIA to 95 months for stage IA 20. In locoregionally-advanced NSCLC, combined modality treatments using concurrent chemo-radiotherapy (CRT) have been associated with an increase in survival rates 21,22. Superior Sulcus Tumours (SST), also called Pancoast tumours, are a rare sub-set comprising 3-5% of all NSCLC and are located in the lung apex. They are locally advanced with invasion of the chest wall, adjacent structures or vertebrae, reducing the likelihood of a radical (R0) resection with primary surgery. Pre-operative induction CRT followed by resection is now considered the standard treatment option in medically fit patients 23,24. This combined-modality approach is associated with 95% complete resection (R0) rates and up to 50% of tumours demonstrates pathologic complete regression (pCR) 25,26. Limited information is available about the histopathologic changes of such neoadjuvant regimes on tumour and adjacent normal lung parenchyma and their relation to prognosis 27,28,29. The aim of this study was to describe the histopathologic pulmonary and tumour related changes after concurrent CRT and relate these to clinical outcome. Patients and methods A retrospective analysis was performed on a consecutive series of patients with SST treated by concurrent CRT and subsequent surgical resection in the period 1997-2007 at the VU University Medical Center (VUmc) and the Netherlands Cancer Institute (NKI-AVL), both located in Amsterdam, the Netherlands. SST was defined as a tumour growing into the thoracic wall at the apex of the lung, above the level of the second rib 30. A total of 50 patients were identified, after excluding 6 patients (13%) who started CRT, but had progressive disease or were no longer clinically fit for surgery. Four cases had to be excluded because of irretrievable histological slides and tissue blocks. All but one of the 19 NKI-AVL patients received an accelerated hypo-fractionated radiation schedule of 66 Gy delivered in 24 fractions and most of them (n=17, 89%) had concurrent single agent chemotherapy consisting of daily low dose Cisplatin. The treatment at the VUmc was one cycle of cisplatin-gemcitabine, followed by concurrent CRT consisting of cisplatin-etoposide and radiation doses ranging from 39 Gy in 13 fractions to 50 Gy in 25 fractions (mean 44 Gy). All VUmc patients received chemotherapy. Combining the data from both institutes the mean absolute radiation dose was 53 Gy. After completion of CRT, restaging (typically with some combination of CT and/or MRI of the thorax and upper abdomen, MRI scan of the brain, FDG-PET scanning and invasive mediastinal staging) and discussion in a multidisciplinary meeting, surgical resection was attempted in patients without disease progression. 2

22 Chapter 2 Pre- and post-induction treatment tumour diameters were measured on CT/MRI scans. Response evaluation was performed according to the RECIST criteria 31. In the surgical resection specimens, tumour size was measured on gross pathological exam and the tumour diameter was retrieved from the pathology report. Histological slides of all available paraffin blocks together with the pathology reports of the resection specimens were reviewed. In each case the number of blocks related to the tumour area was estimated. Tumour type was based on the most classifiable specimen, either the pre-treatment histology or cytology, or the resection specimen. For example, a case with a pretreatment biopsy diagnosis of non-small cell carcinoma not otherwise specified, and a resection specimen diagnosis of squamous cell carcinoma was classified as squamous cell carcinoma. In cases without viable tumour in the resection specimen, the tumour was classified according to the pre-treatment histology or cytology. In order to evaluate the histopathologic changes within the tumour mass, items such as amount of viable tumour, fibrotic changes, vasculopathy and intra- and peritumoural inflammatory reaction were identified and for applicable items a semi-quantitatively scoring system was used as shown in Table 1. Table 1. Criteria for scoring histology Item/category 0 1 2 3 41 Tumour type2 Squamous cell Adenocarcinoma Large cell Other, specify Viable tumour3 no vital tumour cells <10% vital tumour cells 10-50% vital tumour cells >50%% vital tumour completely vital tumour Tumour location within mass Not relevant (no tumour) Randomly distributed Mainly peripherally Mainly centrally Diameter macroscopical mass4 ≤ 2 cm > 2 and ≤ 3 cm > 3 and ≤ 5 cm > 5 and ≤ 7 cm > 7 cm Dominant tumour growth pattern Not relevant (no tumour) Mainly confluent nests Randomly distributed nests or cells Mainly perivascular Other, specify Cytonuclear atypia tumour5 Not relevant (no tumour) Slight (uniform nuclei, small nucleoli) Moderate (variation nuclear size, prominent nucleoli) Severe (bizarre nuclei, macronucleoli) Extreme (mainly bizarre nuclei) Mitotic activity7 Not applicable (no tumour) Low (less then 1/ HPF) Moderate (1-3/HPF) High (more then 3/ HPF) Atypical mitosis7 Not applicable (no tumour) None Few (1/HPF) Several (more then 1/ HPF) Amount of fibrosis - necrosis Just fibrosis Mainly fibrosis (> 75%) Equal fibrosisnecrosis Mainly necrosis (> 75%) Just necrosis Presence of cholesterol clefts None Just in alveolar spaces Focal Extensive Presence of foamy macrophages None Focal Extensive

23 Outcomes of resected SST after CRT Table 1. Criteria for scoring histology (continued) Item/category 0 1 2 3 41 Presence of dyselastosis None Focal Extensive Acute vasculopathy8 (present in 1 or more vessels) None Mainly neutrophils in vessel wall, no thrombi Mainly neutrophils in vessel wall with thrombi or necrosis Mainly lymphocytes in the vessel wall, fibroblastic intima thickening, no thrombi Mainly lymphocytes in vessel wall, fibroblastic intima thickening with thrombi Chronic vasculopathy9 (present in 1 or more vessels) None Slight intima thickening Clear intima thickening Nearly obliterated vessels (pinpoint lumina) Obliterated vessels; no recanalization Boundary tumourlung tissue Sharp all around Mainly sharp (> 75%) Mainly irregular (> 75%) Completely irregular Amount of reactive changes lung parenchyma None Focal (close to mass) Multifocal (one segment) Diffuse Changes lung parenchyma next to mass None Slight (septal thickening, mild inflammation) Moderate septal thickening or inflammation, no structure loss Severe reactive or inflammation, structure loss10 Cytonuclear atypia epithelial cells in lung parenchyma5 None Slight (uniform nuclei, small nucleoli) Moderate (variation nuclear size, stratification, prominent nucleoli) Severe (bizarre nuclei, macronucleoli) 1 Items having more than 4 scores are mentioned below 2 According to WHO classification 200432, not in a semi-quantative order 3 Modified from Dworak grading for colorectal cancer33 4 Cut-off points according to the proposed IASLC staging for lung cancer 200934 Longest axis was used. 5 Most severe atypia accounts for the score 6 Pre-operative specimen not available or cytological 7 At least 3x HPF (1HPF=2mm2) counted35 8 Scores 5-9 as follows: Score 5: mainly lymphocytes within the vessel wall with fibrotic intima thickening without thrombi Score 6: mainly lymphocytes within the vessel wall with fibrotic intima thickening with thrombi Score 7: extensive inflammatory infiltrate within the vessel wall with fibrinoid necrosis Score 8: vasculitis distant from the tumour mass including score 1-7 Score 9: vasculitis distant from the tumour mass including score 0 9 Score 5 as follows: Score 5: obliterated vessels with recanalization (eccentric lumina) 10 Or classifiable as interstitial lung disease pattern To estimate the percentage of vital tumour cells, the area of vital tumour cells was related to the total tumour visible in the microscopy slides, including areas of fibrosis, inflammation and necrosis (Figure 1a-e). 2

24 Chapter 2 Figure 1. Histological overview slides: a complete vital tumour, b >50% vital tumour cells (within black lines), c 10-50% vital tumour cells, d <10% vital tumour cells, e no vital tumour cells (i.e. pathologic complete response) Two experienced lung pathologists (JB, ET) independently scored all tumour area slides with respect to the items of amount of viable tumour and distribution of vital tumour cells. There were no discrepancies in their scores. A continuous area of >6 mm tumour in one slide was sufficient for scoring > 10% vital tumour cells. The percentage of vital tumour cells < 10% was defined as small focal areas in one or more sections with an estimated total area of less than 10% of the gross size of the lesion. For the other pathologic items, both pathologists independently evaluated 10 cases, also revealing no discordance. One pathologist (JB) subsequently evaluated the remaining cases. Followup clinical data were retrieved from the patient archives by the relevant clinicians.

25 Outcomes of resected SST after CRT Statistics Statistical analyses were carried out by SPSS for Windows and Mac version 18.0. (IBM, New York, USA). Associations between all histological items, except for those with complete regression in which tumour characteristics were omitted, were tested using the Fisher’s exact test in case of dichotomous variables with a significant level <0.05. Spearman’s correlations were calculated in case of ordinal variables with a significance level of <0.05. Overall survival rates were estimated by Kaplan-Meier curves, and differences between curves were tested by the log-rank test. Where appropriate threshold levels of significance were adjusted for multiple comparisons by the Bonferroni’s correction 36. Results Clinicopathologic characteristics of the 46 patients are shown in Table 2. Most patients were male (74%), almost all were smokers (87%) and the predominant histological subtype was adenocarcinoma (46%). Nineteen cases were staged as stage IIB (41%), 20 as stage IIIA (44%) and 7 as stage IIIB (15%). At the time of diagnosis 9 patients had clinical N2 disease and an additional 2 had N1 disease (table 2). After the induction treatment none of the patients had clinical N2 disease and 4 had N1 disease. Table 2. Characteristics of all eligible patients (n=46) Age (yrs) mean (range) 56.4 (31-78) Gender Male 34 Female 12 Hospital VUmc 27 NKI-AVL 19 Smoking history Pack-years (yrs) mean (range) 30.7 (0-68) Smoker at time of diagnosis Yes 40 No 3 Unknown 3 Smoker at time of treatment Yes 20 2

26 Chapter 2 Table 2. Characteristics of all eligible patients (n=46) (continued) No 18 Unknown 8 cTNM (stage) cT3N0M0 (IIB) 19 cT3N1M0 (IIIA) 2 cT3N2M0 (IIIA) 2 cT4N0M0 ((IIIA) 16 cT4N1M0 (IIIA) 0 cT4N2M0 (IIIB) 7 Histology Adenocarcinoma 21 Large cell carcinoma 14 Squamous cell carcinoma 7 Unknown 4a a Treatment based on clinical evidence without pathology diagnosis and without residual tumour in the resection specimen Tumour related histopathologic findings No significant differences were found between the two hospitals for the histologic items, allowing combined data analysis. The resection specimens showed that all lesions were at least 1 cm in size. No vital tumour cells (= pCR) were found in 20 cases (43%). In these 20 specimens, the presumed tumour area showed either necrosis, fibrosis with or without dyselastosis or a combination of the two. Cholesterol clefts were focally or extensively present in the majority of both the pCR and non-pCR cases. All cases showed vascular changes ranging from vasculitis-like changes to complete luminal obliteration and recanalization. The results of the statistical analysis between all scored histological items, are shown in the table 3.

27 Outcomes of resected SST after CRT Table 3. The relationship between morphologic tumor and lung tissue characteristics is illustrated in this table. Tumour location in mass Diameter macroscopic mass Dominant growth pattern Mitotic activity Amount of fibrosis/ necrosis Foamy macrophages Dyselastosis Vasculitis-like changes Chronic vasculitis Amount of reactive changes Type of reactive changes Cytonuclear atypia lung parenchyma Number available slides macroscopic mass Tumor type combined prepostOK 0,14 0,03 -0,09 0,15 0,17 -0,05 -0,23 -0,16 -0,25 0,18 -0,06 -0,09 -0,18 Regression -0,67 0,34 -0,41 0,00 0,15 -0,39 -0,15 0,08 -0,10 0,28 0,14 0,24 0,35 pathologic Complete Response (pCR) . 0,33 . . 0,09 -0,34 -0,19 0,12 -0,10 0,27 0,18 0,21 0,31 Diameter macroscopic mass -0,03 1,00 0,23 0,50 0,49 -0,04 -0,38 0,03 -0,13 -0,02 -0,06 -0,17 0,47 Cytonuclear atypia within tumour -0,29 0,36 0,08 0,23 0,58 0,24 -0,19 -0,12 -0,18 0,18 0,14 0,18 0,10 Amount of fibrosis/ necrosis 0,13 0,49 -0,16 0,36 X 0,07 -0,52 -0,06 -0,19 0,12 -0,01 0,15 0,18 Cholesterol clefts -0,07 0,10 0,17 -0,04 0,27 0,40 0,02 -0,23 0,06 -0,03 0,00 0,14 0,10 Dyselastosis -0,25 -0,38 0,05 -0,19 -0,52 0,15 X 0,36 0,41 0,06 0,17 0,10 -0,33 Vasculitis like changes 0,02 0,03 0,23 0,03 -0,06 0,11 0,36 X 0,33 0,15 0,31 0,12 0,04 Boundary tumour-lung -0,25 0,12 -0,13 0,38 0,05 -0,25 0,04 0,04 0,21 0,31 0,26 0,34 -0,05 Amount reactive changes lung -0,10 -0,02 0,01 0,10 0,12 0,08 0,06 0,15 0,26 X 0,32 0,48 0,11 Type of reactive changes lung -0,05 -0,06 0,18 0,05 -0,01 -0,13 0,17 0,31 0,25 0,32 X 0,40 0,02 Correlation coefficients between scored items are shown (bold/light grey cell=significant P-values <0.05; bold italic/dark grey cell=significant P-values <0.01). Items without significance as well as those with a skewed distribution are not shown, except for tumor type. 2

28 Chapter 2 Histological tumour type was not associated (p>0.05) with any of the scored items. The smaller the tumour diameter and the lower the number of tumour blocks taken for histological exam, the higher chance of pCR (Spearman’s correlation r=-0.33, p=0.03 and r=0.31, p=0.04, respectively). In other words, from larger tumors more blocks were examined and residual tumour was more likely to be found. The numbers of blocks examined per patient varied from three to 15 with a mean of 6,8 blocks. In cases of residual tumour, most (20/26) had less than 50% vital tumour cells. These vital cells were more frequently found in the periphery than in the center of the pathologic lesion (16/20) (p<0.001). The cytonuclear atypia of tumour cells was more prominent in cases with necrosis compared to cases with mainly fibrosis (r=0.58, p=0.002). A high amount of dyselastosis was associated with a smaller tumour diameter (r=-0.38, p=0.009), as well as with more fibrosis than necrosis (r=-0.52 p<0.001). Dyselastosis was also correlated with chronic vascular changes (i.e., from intimal fibrosis to complete luminal obstruction, r=0.41, p=0.005). In addition, cases with chronic vascular changes frequently occurred together with acute vasculitis-like changes (r=0.33, p=0.03). More type II pneumocytic cytonuclear atypia was found in lung parenchyma that showed more reactive changes (r=0.48, p=0.001). No correlation was found between vascular changes and histological tumour regression. When threshold levels of significance were adjusted for multiple comparisons by the Bonferroni’s correction, none of the stromal related correlations showed p-values < 0.05 36 . The tumour-lung boundary was predominantly sharp in nearly all cases (44/46) and no tumour foci were found outside this boundary. Treatment response Tumour response after CRT and before surgery could be determined on CT scans and/or MRI’s in all but 3 cases. The median time between pre- and post-CRT response evaluation was 2.0 months (range 0-12 months). The median time between end of induction CRT and surgical resection was 4.0 weeks (range 0-26 weeks). Data for the pairwise comparisons using clinical RECIST criteria for pre-treatment, post-CRT and pathological tumour sizes are shown in Table 4. More partial responses were found when the RECIST criteria were applied to pre-treatment imaging and pathological gross specimen tumour size than to pre- and post-treatment imaging (p<0.005). Pathological complete response (pCR) was more frequently found in cases with clinical partial response compared to clinical stable disease and progressive disease (p=0.001). Other histological changes were not associated with the extent of response.

29 Outcomes of resected SST after CRT Table 4. Number of patients with tumour response based on pre and post CRT imaging using RECIST criteria, and a comparison of pre-CRT imaging and post CRT gross pathology. The number of patients in these groups with pathologic complete response (pCR) is also described. RECIST (Number; %) (Pathologic Complete Response: Number; %) Partial Response Stable Disease Progressive disease Missing Total Pre – post- CRT imaging 14; 30% (10; 50%) 29; 63% (7; 35%) 0 3; 7% (3; 15%) 46 (20; 100%) Pre-CRT imaging – postCRT gross pathology 27; 59% (16; 80%) 15; 33% (3; 15%) 4; 9% (1; 5%) 0 46 (20; 100%) CRT chemoradiotherapy Follow-up data Follow-up data was available for all patients. There were no differences in follow-up status between the two hospitals or between patients with different clinical stages. Table 5 shows the follow-up status of all eligible patients according to tumour type and tumour response. Table 5. Follow up status related to tumour type and pathological response Follow up status AND DND AWD DOD Tumour type Squamous cell carcinoma 2 1 0 4 Adenocarcinoma 8 2 1 10 Large cell carcinoma 6 1 0 7 Unknown 2 0 0 2 Total 18 4 1 23 Tumour response Complete regression 12 1 1 6 < 10% vital tumour 1 0 0 2 10-50% vital tumour 3 3 0 11 >50% vital tumour 1 0 0 4 Total 18 4 1 23 AND = Alive No Disease; DND = Dead No Disease; AWD = Alive With Disease; DOD = Dead Of Disease The median follow-up time was 34 months (range 5 – 154 months). Prognosis was significantly better after a pCR compared to residual tumour (70% 5-years overall survival vs. 20%; p=0.001) and in patients with fewer than 10% vital tumour cells as compared to those with >10% (65% 5-years overall survival vs 18%; p<0.001; Figure 2

30 Chapter 2 2). There were no differences in cases with 10-50% compared to >50% vital tumour cells. A low mitotic count (< 1 per high power field) was associated with a longer diseasefree survival (p=0.02). None of the other associations between histological scores and follow-up status were statistically significant. Figure 2. Kaplan-Meier survival curves showing a significantly better prognosis in patients with less than 10% vital tumor cells in their resection specimen. Discussion This is the largest series reporting a detailed histopathologic evaluation of resected Pancoast tumours after tri-modality therapy. Twenty-three out of 46 patients (50%) with SST showed pathological complete response (n=20) or <10% vital tumour cells (n=3) after CRT. Pathologic complete regression or the presence of <10% vital tumour cells was associated with survival. A low mitotic rate was associated with a longer disease-free survival37. The majority of patients with radiological partial response according to RECIST criteria had a pathologic complete response. There were no statistically significant differences between the patients or outcomes from the two institutes, even though their treatment regimens differed. Tumour regression in the lesions was associated with histologic features of necrosis and/or fibrosis, as well as variable histiocytic reactions such as foam cells and cholesterol crystal-related giant cells. These findings are similar to those of Liu-Jarin et al, who analyzed the histological patterns in 30 surgical resection NSCLC specimens after neo-adjuvant therapy28. In their study, consisting of 15 cT3 tumours, the extent of fibrosis correlated with the radiologic response in patients that showed more than

31 Outcomes of resected SST after CRT 10% tumour regression. We found more cytonuclear atypia in cases with more necrosis than fibrosis, in line with Yamane et al29. The pCR or minimal residual disease rate of 50% (23/46 cases), is broadly in keeping with that reported by others (Table 6), however there is some variation between studies. For example, Rusch et al26 described a series of 88 patients and found up to 69% pCR or minimal residual tumour. One possible explanation for the differences might be variation in the how the pathological assessment was made (which is often incompletely described) or in histopathologic definition of response. Table 6. Selected series of patients with non-small cell lung cancer and histopathologic evaluation after CRT Reference No. evaluable cases No. patients with CRT (%) Mean radiation dose No. of patients with pCR (%) No. of patients with pCR or minimala residual tumour cells This article 46 46 (100%) 53 Gy 20 (43%) 23 (50%) Pourel25 71b 71 (100%) 45 Gy n.m. 28 (39.4%) Rusch26 88 88 (100%) 45 Gy 32 (36%) 61 (69%) Junker27 40 40 (100%) 45 Gy 7 (17%) 27 (67%) Liu28 30 20 (66%) 50 Gy 5 (25%) 13 (65%) Fischer38 44 44 (100%) 45 Gy 13 (30%) 28 (64%) CRT chemo-radiotherapy, pCR pathologic complete response, n.m. not mentioned a minimal residual cells is variably defined in different studies. For example, in this study and the report by Fischer et al it was <10% b107 pts in the study, 105 completed CRT, 72 underwent surgery, 71 were pathologically evaluable The results add to the growing body of evidence indicating that a complete or nearcomplete pathologic response is associated with a more favourable prognosis in patients with SST. This is in line with Yamane et al. who report that a smaller area of residual tumour is associated with a better prognosis29. Significantly more pCR was found in cases with a partial imaging response compared to cases with stable disease. Although the RECIST criteria were not designed to compare radiological tumour diameters with diameters measured on pathologic resection specimen, we found significantly more cases with ‘partial response’ when we compared the diameters on pre-treatment imaging with those in the resection specimens. The latter also tended to be smaller compared to post-treatment imaging diameters. These data suggest that pCR is more likely to be found in cases with larger diameters with subsequent significant regression on combined modality treatment, than in cases that were stable on imaging despite a smaller initial size. Junker et al27 describe a 3-grade regression scoring system for patients receiving induction therapy i.e. grade I, no or slight regression; IIA, marked but incomplete 2

32 Chapter 2 response (>10% vital tumour); IIB, less than 10% vital tumour and grade III for complete response. We divided their grade IIA in 2 separate grades, i.e., one for more than 50% vital tumour and one for 10-50% vital tumour. This 4-tiered grading system is similar to the one used by Dworak for colorectal cancers33. We examined whether this 4-tiered approach was associated with prognosis. This turned out not to be the case. Our data support a slight modification of the Junker classification by combining grade I and IIA into one group, which is associated with a poorer prognosis, the group with <10% vital tumour cells (=Junker IIB) and those with pCR (Junker grade III). The prognostically relevant 10% threshold has also been reported for esophageal cancer39. Interestingly, residual vital tumour was found significantly more often at the periphery of the pathologic lesion in the resection specimen. Although the center of the solid tumour is more likely to be sensitive to hypoperfused and hypoxic, a microenvironment associated with an increased chance for development of resistance to radiotherapy and anticancer chemotherapy40, prolonged hypoxia of the tumour tissue may also lead to necrosis, a frequent finding in larger solid tumours41. In patients without pCR the margins of the surgical resection were more likely to contain vital tumour cells. The limitations of this study include the following: (a) the total number of cases is relatively small, (b) only the available slides could be studied and (c) it is a retrospective study; therefore, the tumour lesions, for example, were sampled for reporting purposes and not more systematically for research, and in daily practice more samples may be taken from the periphery of tumour lesions than from its centre or from necrotic appearing areas. For future studies, we advocate sampling more blocks from the resected tumour mass in order to more reliably determine the absence or presence of vital tumour cells. In addition, it should be noted that the presence of histologically vital tumour cells does not necessarily predict their biological capacity to metastasize. Finally, (d) we acknowledge that there was variation in the time between pre- and postinduction CRT imaging, and the end of CRT and surgery. In future studies of patients with neo-adjuvant CRT and surgery, we propose taking the post-CRT fraction of vital tumour cells into account. Although the IASLC Staging Manual in Thoracic Oncology advises categorization of ypT, the extent of tumour actually present at the time of pathology examination32 35, detailed information on how to evaluate vital tumour cells with relevant portions of necrotic and reactive changes is not provided. For better comparison, in future studies, we suggest the following approach: firstly, record gross size of pathological lesion (including reactive changes such as fibrosis and necrosis, and if present vital tumour) to determine the size of pT. This will also allow comparison with the RECIST data with the size on the resection specimen. Secondly, for the vital tumour component, the adjunct VT is used with ranges as shown in table 7, comparable to the regression system proposed by Becker et al for gastric cancer 42. For example, a pathologic tumour mass at gross examination of 4 cm with no vital tumor cells, less than 10% and more than 50% vital tumour cells would then be

33 Outcomes of resected SST after CRT classified as ypT2a (VT0), ypT2a(VT1), ypT2a (VT3), respectively. We realize that the current recommendation for no vital tumour cells should be ypT0. Table 7. Proposed classification of the presence of vital tumour cells within a tumour after CRT % of vital tumour cells Notation (as superscript) 0 VT0 <10% VT1 10-50% VT2 > 50% VT3 VT vital tumor cells, CRT chemoradiotherapy In conclusion, we identified a pCR in over 40% of patients with SST undergoing induction CRT a complete or near-complete pathologic regression was associated with a more favorable prognosis. 2

3 The prognostic value of proliferation, PD-L1 and nuclear size in patients with superior sulcus tumours treated with chemoradiotherapy and surgery Hans Blaauwgeers, Birgit Lissenberg-Witte, Chris Dickhoff, Sylvia Duin, Erik Thunnissen Journal of Clinical Pathology. 2023;76(2):111–5.

36 Chapter 3 Abstract Aims The aim of this study was to determine the relationship between proliferative activity, PD-L1 status and nuclear size changes after pre-operative chemo-radiotherapy (CRT) and clinical outcome in patients with superior sulcus tumours. Methods Proliferative activity (MIB-1) and PD-L1 status was estimated by immunohistochemistry in the tumour cells of resection specimen in a series of 33 patients with residual tumour after tri-modality therapy for a sulcus superior tumour between 2005 and 2014. A morphometric analysis of both pre- and post-treatment tumour material was also performed. Results were related to disease free survival (DFS) and overall survival (OS). Results Low proliferative activity (<20% MIB-1) was associated with better OS: 2-year OS of 73% compared to 43% and 25%, respectively for moderate (MIB-1 20-50%) and high proliferative activity (MIB-1 >50%) (p=0.016). A negative PD-L1 status (<1% positive tumour cells) was also associated with a better OS (p=0.021). The mean nuclear size of normal lung tissue pneumocytes was significantly smaller compared to the mean nuclear size of tumour cells of the resection specimens (median difference -38.1; range -115.2 - 16.0; p<0.001). Mean nuclear size of tumour cells did not differ between pre-treatment biopsies and resection specimens (median difference -4.6; range -75.2 - 86.7; p=0.14). Nuclear size was not associated with survival (p= 0.82). Conclusions Low proliferative activity determined by MIB-1 as well as a negative PDL-1 expression are significantly associated with a better overall survival in patients with residual tumour after CRT for superior sulcus tumour.

37 Prognostic value of MIB-1, PD-L1 and nuclear size in SST after CRT Introduction In patients with resectable non-small cell lung cancer (NSCLC), survival rates drop with higher clinical stage, ranging from an estimated median survival of 95 months for stage IA to 19 months for stage IIIA 20. In locally-advanced NSCLC (LA-NSCLC), combined modality treatments using concurrent chemo-radiotherapy (CRT) have led to increased survival rates21,22. For superior sulcus tumours (SST), also called Pancoast tumours, comprising 3-5% of all NSCLC, and typically invade the chest wall, induction CRT followed by resection is considered the standard treatment in medically fit patients23,24,30. This combined modality approach is associated with 95% complete resection (R0) rates and up to 50% of tumours demonstrate pathologic complete regression25,26. The presence of complete pathological regression was found predictive for favourable clinical outcome after tri-modality therapy in these patients43. Other morphological characteristics found in pretreatment biopsies and resection specimen, such as Ki-67, MIB-1, nuclear size and PD-L1 expression, have been investigated for prognostic value in resected NSCLC, but their role in patients treated with induction and resection (trimodality treatment), and their relation to clinical outcome, is currently unknown44. This lack of knowledge was noticed by the members of the International Association Lung Cancer (IASLC), who recently published multidisciplinary recommendations for pathologic assessment of lung cancer resection specimens after neoadjuvant therapy, stating that future studies to explore the role of immunohistochemistry on resected tumours in the neoadjuvant setting are encouraged45. Against this background, the aim of this study was to examine whether changes in nuclear size, expression of MIB-1 and PD-L1 in the residual tumour compared to baseline, are prognostically relevant for patients with SST treated with concurrent CRT, followed by resection. Patients and methods A retrospective analysis was performed on a consecutive series of patients with SST treated with concurrent CRT and subsequent surgical resection in the period 2005-2014 at Amsterdam University Medical Center, location VU Medical Center in Amsterdam. SST was defined as a tumour growing into the thoracic wall at the apex of the lung, above the level of the second rib 30. Patients were selected on basis of residual tumour being present in the resection specimen. For all 33 patients, a pre-treatment diagnostic biopsy was included for morphometric evaluation. Clinical data on follow-up were retrieved from the patient medical records. Part of this patient population has previously been described43. 3

38 Chapter 3 Morphometry The method of nuclear size quantitation was performed on haematoxylin- eosin stained sections as described before46. The measuring system was a commercially available interactive video overlay-based system (Q-PRODIT; Leica, Cambridge, UK). The microscopic image (100x objective) was recorded by a video camera and shown on a computer screen (final magnification 3000x). Biopsy area was selected by pathologist (E.T.) and nuclear area measured with manual delineation. At least 25 nuclei of neoplastic cells, and nuclei of type II pneumocytes, if present, were measured. The mean, median, 75th and 90th percentile of the nuclear size was estimated in all patients. Immunohistochemistry The paraffin blocks of the preoperative biopsies contained too few residual tumour tissue to perform the additional immunohistochemical analysis for MIB-1 and PD-L1. For this reason, immunohistochemistry analysis was only performed on the resection specimen. In short, immunohistochemistry for Ki67 (clone MIB-1clone (Dako/Agilent, Glostrup, Denmark) was performed in the Ventana Benchmark Ultra (Tucson, USA) diluted 1/50 and incubated for 32 minutes after antigen retrieval with Cell Conditioning Buffer 1 (CC1) 24 minutes at 100°C detection with Optiview DAB Detection Kit. The proliferation fraction of the residual tumour in the resection specimen was estimated by scoring the overall number of positive staining tumour cells of one whole tumour section in one of three categories: <20%, 20-50% and >50%47,48. PD-L1 immunohistochemistry with the 22C3 clone was performed in a laboratory developed test (LDT) as previously described49. The tumour proportion score (TPS) of PDL-1 positive tumour cells was assigned to one of 3 categories: <1% (negative), 1-49% (weak positivity) and ≥ 50% (strong positivity)50. Statistics Statistical analyses were carried out by SPSS for Windows and Mac version 26 (IBM Corp., Armonk, NY, USA). Kaplan-Meier curves with log-rank test were used to compare overall survival (OS) and disease-free survival (DFS) between the different categories of MIB1 and PD-L1 status. OS was defined as time from diagnosis to time of death, DFS was defined as time from diagnosis to date of recurrence. Patients alive or without recurrence were censored at their last date of follow-up. Nuclear measurements in biopsies were compared to those in the resection specimens within patients using by the Wilcoxon sign rank test. Nuclear measurements in the resection specimen and normal tissue in the resection specimen from patients with and without recurrent disease were compared, using the Mann-Whitney U test. Data are described by median and range. A p-value < 0.05 was considered significant.