Background
Solitary fibrous tumors of the pleura (SFTP) are rare neoplasms of the chest cavity with an estimated incidence of 2.8 new cases/100,000 individuals/year [
1,
2]. They are responsible for less than 5 % of all tumors arising from the pleura, but have been reported to occur in other anatomical locations [
3‐
5]. Klemperer et al. were the first to describe SFTP in 1931, and since then only 900 cases have been reported up to 2005 [
1,
6].
According to their distinct histological pattern, SFTP used to be called ‘hemiangiopericytoma’. A variety of other names were used, which made it difficult to classify them into one entity [
7]. Histologically, they typically show alternate areas with varying hypercellularity. Characteristic in those less cellular areas is a dense collagenous background with variable assembled spindle cells, the so-called “patternless pattern” and haemangiopericytoma-like branching blood vessels. By using immunohistochemical (IHC) staining including CD34 and CD99, it has been shown that SFTP seem to originate from mesenchymal tissue [
8,
9]. However, the major breakthrough in the firm diagnosis of SFTP was the detection of the NAB-STAT6 gene fusion which is present in up to 100% of these tumors and, thus, has been shown to help diagnosing these rare neoplasms and distinguish them from histologic mimics [
10‐
12]. Furthermore, some studies suggested that a certain fusion type might be associated with different variations of SFTP [
13,
14].
The majority of SFTP seem to follow a benign behavior. However, 10–25% can relapse or present signs of malignancy, i.e. metastases [
15]. In 1989, England et al. proposed morphological and histological characteristics to distinguish malignant from benign variants of SFTP [
16]. These included tumor size, atypical localization, sessile rather than pedunculated tumor, necrosis or hemorrhage, more than four mitoses per 10 high power fields (HPF), and nuclear pleomorphism. These features built the basis for subsequent scoring systems which have been introduced to better predict the biological behavior of SFTP [
17,
18]. However, these scoring systems do not consider IHC markers as possible prognosticators of the biological behavior. The aim of this study was to investigate the role of clinical, morphological and, in particular, IHC staining as possible prognosticator of the outcome in patients with SFTP.
Methods
Subjects and data
Patients were recruited at seven hospitals in Switzerland and Germany. All patients with a histologically proven SFTP who were diagnosed and operated between 2000 and 2015 at one of these centers and had a follow-up of at least one month were eligible for the study. Patients’ clinical and follow-up data were drawn from the medical records and, if required, obtained from their general practitioners.
Formalin-fixed and paraffin-embedded (FFPE) tissue blocks and representative slides of every eligible patient were collected from all participating hospitals. Accuracy of the SFTP diagnosis was achieved using IHC for STAT6 c-terminus epitope as a surrogate marker for the NAB-STAT6 fusion protein which was performed in every sample. All NAB-STAT6 negative tumors were microscopically re-examined by an expert lung pathologist (A.S.). Patients with incorrect diagnosis of SFTP or missing FFPE tissue block were excluded from the study.
Written informed consent was obtained from all patients or their relatives. The study was approved by the Ethics committee of the Canton of Zurich (KEK-ZH 2012–0279), and the study was notified at
ClinicalTrials.gov (Identifier: NCT01694654).
Hybrid cytology/tissue microarray
To reproduce a representative profile of the whole tumor we obtained four cores for every patient and selected two less proliferative and two more proliferative areas of the parrafin block. From each of these regions of the donor block, a paraffin core of 0.6 mm diameter and 3–4 mm height was taken and precisely arrayed into a new recipient paraffin block using a custom-made, semiautomatic tissue arrayer (taucherinsel-neuenburg@t-online.de). Of these, 2.5 μm sections were cut for IHC staining which was performed under standardized conditions. Commercially available antibodies were used for IHC staining. The following antibodies were used: MIB-1 (Rabbit anti-Ki-67, mib 1, clone 30–9, Ventana-Roche, Tucson, AZ, USA), CD 34 (Confirm anti-CD34, clone QBEnd/10, Ventana-Roche, Tucson, AZ, USA), bcl-2 (Mouse anti-bcl-2, clone 124 Ventana-Roche, Tucson, AZ, USA), CD99 (Mouse anti-human CD99, clone HO36–1.1, Novocastra laboratories Ltd., Newcastle upon Tyne, UK), p53 (Mouse anti-human p53 Protein, clone DO-7, DAKO A/S, Glostrup, Denmark), Vimentin (Mouse anti-human Vimentin, clone VIM3B4, DAKO A/S, Glostrup, Denmark), Insulin-like-growth-factor IGF2 (Mouse anti-Insulin like growth factor 2, Cell Signaling Technology, Cambridge, UK) and NAB- STAT-6 (Stat6 S-20, Santa Cruz Biotechnology Inc., Dallas, TX, USA).
Immunohistochemistry
Immunoreactivity of the samples was scored visually by M.D. and a dedicated pathology technician. The staining of all IHC markers except MIB-1 were semi-quantitatively scored from 0 (negative) to 3 (strong reaction) in all four cores. Finally, the mean of these scores was used for statistical analysis. For MIB-1 proliferation index, all the cells were counted visually and the percentage of cells having any positivity was calculated. The mean percentage of positive cells of all four cores was used for statistical analysis.
Definition of adverse outcome
Adverse outcome was defined as recurrence of SFTP, development of metastases or SFTP-related death. Thus, event-free survival time was defined as the time between surgery and the occurrence of an adverse outcome.
Statistical analysis
All statistical analyses were performed using SPSS Statistics for Windows, version 20.0 (IBM Corporation, Armonk, NY). χ2-test (categorical data) and Mann-Whitney U test (continuous data) were used to determine a significant association between the variables and an adverse outcome. Overall-survival and event-free survival were calculated using Kaplan-Meier curves. To estimate a prognostic value of each variable we used the Cox regression model. Each variable with a p-value less than 0.1 was entered in a multivariable analysis, and the hazard ratio (HR) and confidence interval (CI) were computed. A p-value less than 0.05 was considered to be significant. We tested morphological, clinical and IHC staining variables for their possible prognostic value. Furthermore, sensitivity and specificity of the currently used scores were computed.
Discussion
In this multicenter study, we collected clinical data and FFPE tissue blocks of patients with histologically proven SFTP, aiming to investigate IHC markers which may improve predicting the biological behavior of SFTP. Eleven percent of patients had an adverse outcome. SFTP-related lethality was 0.04%, whereas recurrence rate was 7.7% compared to 10–25% in other series [
15,
16,
18]. This could be related to a shorter follow-up time in our study with a median follow-up time of 36 months compared to a median follow-up time of 70 months in the study of Schirosi et al. [
15].
While the majority of SFTP follow a benign course, few SFTP will behave aggressively and lead to recurrence, tumor-related death or metastatic disease. However, many SFTP originally classified as malignant follow a benign course, and vice-versa [
16,
19]. Therefore, England et al. proposed six histomorphological features to characterize malignant variants [
16]. These features are widely accepted and used in many centers. It is not clearly defined though, which and how many of these features have to be fulfilled to define a malignant variant of SFTP. In our study, from the variables included in the criteria proposed by England et al., [
16] only tumor size greater than 10 cm, nuclear pleomorphism and increased mitotic activity (more than four mitoses per 10 HPF) were significantly associated with an adverse outcome in univariable analysis. Yet, in the multivariable analysis, only two of these features (mitotic count and tumor diameter) remained significant. This finding was already confirmed by other studies demonstrating that only mitotic count and tumor size were associated with impaired outcome [
20,
21]. Furthermore, the specificity of the England criteria was challenged by Schirosi et al. who showed that at least one of these variables is present in 59% of SFTP, whereas mortality and recurrence rate are only 10% and 18%, respectively [
15].
Recently new scoring systems have been proposed to predict either the risk of recurrence or the combined risk of adverse outcome, namely recurrence, development of metastasis or SFTP-related death [
17,
18]. De Perrot et al. proposed a model based on morphologic and histologic features (morphology, high cellularity, cellular pleomorphism, high mitotic count, necrosis or stromal/vascular invasion) with a treatment plan and recommendation of adjuvant therapy and follow up [
18]. Likewise, Tapias et al. suggested a scoring system to predict the risk of recurrence based on certain histologic and morphologic features (pleural origin, morphology, size, hypercellularity, presence of necrosis or haemorrhage, number of mitosis per 10 HPFs) [
17]. However, we could not confirm those scores proposed by Tapias et al. [
17] in our cohort, since it was not associated with the risk of recurrence or adverse outcome.
In the end, none of the grading systems based on histomorphological features is sufficiently accurate to predict the biological behavior of SFTP at the time of diagnosis. Thus, attempts to find new biomarkers or to expand the conventional grading systems by new biomarkers is warranted. However, none of the above mentioned studies investigated the prognostic value of IHC markers. A reason for this could be that IHC staining for SFTP is not standardized and, thus, not uniformly used in different centers. Likewise, MIB-1 proliferation index (Ki-67) has not been routinely measured, although some studies showed a trend towards a promising prognostic value of MIB-1 [
17,
20,
22]. We collected FFPE tissue blocks of patients with histologically proven SFTP from seven hospitals in two countries and uniformly performed IHC staining on them. Loss of CD34 expression has been shown to be more prevalent in malignant SFTP [
23]. We could not confirm this finding, since all SFTP in our study were positive for CD34. Furthermore, positivity of the tumor suppressor protein p53 has been shown to be associated with malignant features [
15,
23]. In our study, p53 expression was not significantly more prevalent in patients with adverse outcome. NAB-STAT6 gene fusion has been shown to be present in the vast majority of SFTP and is used for diagnosis of these tumors with a sensitivity and specificity of 98% and 85%, respectively [
10‐
12,
24]. In our study only three SFTP showed negativity for NAB-STAT6 resulting in a sensitivity of 96% in diagnosing SFTP, which is similar to other published series [
10,
11]. MIB-1 is an antibody against Ki-67 which is a protein associated with cell proliferation and ribosomal RNA synthesis. Thus, MIB-1 is a reliable IHC marker of cell proliferation, and MIB-1 proliferations index (Ki-67 expression) is associated with poor prognosis in various cancers, especially breast and prostate cancer [
25,
26]. However, the prognostic value of MIB-1 proliferation index has not yet been systematically investigated in SFTP. To date, there is only one retrospective study in a relatively small cohort published by our group, revealing that a high proliferation index assessed with Ki-67 was associated with poor prognosis [
20]. To the best of our knowledge, the present study is the largest to systematically investigate the role of MIB-1 in SFTP. We found that a high proliferation index MIB-1 of more than 10% was independently associated with adverse outcome.
Upon this finding, we established a new score using the variables which have been shown to correlate with impaired outcome. This score, using mitosis count, tumor size, MIB-1 and necrosis performed better than traditional scores in predicting adverse outcome of SFTP in our cohort. Compared to the criteria proposed by England et al. [
16] or the score proposed by Tapias et al. [
17] this score is the first to use an IHC marker. We excluded histomorphological features, such as pleomorphism, nuclear atypia and hemorrhage, since they are subject to inter-observer variability and lack clear-cut objectiveness. Certainly, this score has to be validated in a larger population, but, nevertheless we highly recommend including MIB-1 in any future score.
A clear strength of our study is a considerably high patient number drawn from multiple centers despite the low incidence of SFTP. Furthermore, the design of the study allowed a systematical and uniform investigation of IHC staining and their prognostic value. However, some limitations merit consideration. First, there was a relatively short follow-up time. Secondly, due to lack of variety of proliferation rates we could not perform a receiver operating curve (ROC) analysis to evaluate the best cut-off value of MIB-1. Finally, considering the histological structure of SFTP with alternate areas of more or less proliferation, it has to be further evaluated, whether our scoring of MIB-1 using two less and two more proliferative areas is reproducible in other series.