Introduction

MCO

Molecular and Clinical Oncology

2049-9450 2049-9469

D.A. Spandidos

10.3892/mco.2016.819

MCO-0-0-819

Articles

Back to the start: Evaluation of prognostic markers in gastrointestinal stromal tumors

KLIESER

ECKHARD

1 * PICHELSTORFER

MAXIMILIAN

1 2 * WEYLAND

DENIS

1 2 KEMMERLING

RALF

1 SWIERCZYNSKI

STEFAN

3 DINNEWITZER

ADAM

3 JÄGER

TARKAN

3 KIESSLICH

TOBIAS

4 5 NEUREITER

DANIEL

1 ILLIG

ROMANA

1Institute of Pathology, Paracelsus Medical University/Salzburg General Hospital (SALK), A-5020 Salzburg, Austria 2Bio- and Environmental Technology Program, University of Applied Sciences Upper Austria, A-4600 Wels, Austria 3Department of Surgery, Paracelsus Medical University/Salzburg General Hospital (SALK), A-5020 Salzburg, Austria 4Department of Internal Medicine I, Paracelsus Medical University/Salzburg General Hospital (SALK), A-5020 Salzburg, Austria 5Laboratory for Tumor Biology and Experimental Therapies (TREAT), Institute of Physiology and Pathophysiology, Paracelsus Medical University, A-5020 Salzburg, Austria

Correspondence to: Dr Daniel Neureiter, Institute of Pathology, Paracelsus Medical University/Salzburg General Hospital (SALK), 48 Müllner Hauptstrasse, A-5020 Salzburg, Austria, E-mail: d.neureiter@salk.at *

Contributed equally

05 2016

11 03 2016

4 5 763 773 09122015 26022016

2016

The aim of this study was to provide a standardized risk stratification model for gastrointestinal stromal tumors (GISTs) based on tumor localization, tumor size, involved lymph nodes and metastases, as well as mitotic activity and other morphological and molecular markers, in order to improve the risk evaluation scheme for recurrence, metastatic spread and survival for patients with GIST. A total of 201 cases of patients with GIST were investigated according to standardized morphological markers, including nuclear pleomorphism, tumor cell necrosis, mucosal infiltration, ulceration, skeinoid fibers and growth pattern. In addition, all cases were immunohistochemically analyzed using a tissue microarray platform for various markers of differentiation (CD34, CD44, CD117, desmin, discovered on GIST 1, platelet-derived growth factor receptor α, S-100 and smooth muscle actin) and proliferation (B-cell lymphoma 2, P16, P53, phosphohistone H3 and Ki-67). These findings were correlated by uni- and multivariable analyses with clinicopathological characteristics, including recurrence, metastasis and survival. The general clinicopathological parameters of this GIST specimen cohort were comparable to previous studies. While several parameters exhibited clear associations to each other and to the defined clinical endpoints, the multivariate analysis reduced the number of relevant prognostic variables to localization, margin status, growth pattern and hematoxylin and eosin-based mitosis/Ki-67-based proliferation of GISTs. With the exception of CD34, none of the applied markers of differentiation and proliferation were found to be independent prognostic markers in GIST and the classical risk factors of GIST remain important prognostic factors. Additionally, growth pattern may predict the risk of recurrence and metastasis in GIST patients. Additional independent molecular prognostic markers remain to be identified and validated.

gastrointestinal stromal tumors prognostic markers

Introduction

With an incidence of 1.4–2 cases per 100,000/year, gastrointestinal stromal tumors (GISTs) are the most frequent mesenchymal gastrointestinal neoplasms (1). GISTs mainly affect individuals aged >50 years and are more likely localized in the stomach (60%) compared with the jejunum and ileum (30%), duodenum (4–5%), rectum (4%), remaining colon and appendix (1–2%) and esophagus (<1%). Approximately 30–50% of GISTs are malignant and, in ~50% of the cases, metastasis has occurred prior to the initial diagnosis (2). Morphologically, GISTs are classified into spindle-cell (70%), epitheloid (20%) or mixed types (10%) (3). The most common symptoms of GISTs are anemia, weight loss, gastrointestinal bleeding, abdominal pain or tumor mass-related symptoms. Complete surgical resection is the main goal of therapy. A tyrosine kinase inhibitor, imatinib (STI-571), was recently approved for the (neo)adjuvant treatment of GISTs (4).

Malignant GIST lesions are more common compared with benign tumors in the intestine, whereas the opposite is true in the stomach. The main factors for risk stratification are tumor size, mitotic rate and anatomical location. Tumors sized <2.0 cm with low mitotic activity [<5 mitoses per 50 high-power fields (HPFs)] are usually benign, whereas tumors exhibiting >5 mitoses per 50 HPFs are likely to be malignant (5). This rule is applied for the classification of malignant GISTs, although there are exceptions for tumors which are considered to be benign by the accepted criteria, as lesions sized <2.0 cm and/or mitotic rate <5/50 HPFs occasionally metastasize, the reason for that remains unclear (3).

There are different risk stratification systems to classify GISTs from low- to high-risk, using parameters such as the mindbomb E3 ubiquitin protein ligase 1 index, histological subtype, alterations related to specific oncoproteins, such as P16, unfavorable kinase mutations or cytogenetic aberrations. The problem is that the different classes of GISTs are reported to overlap in a number of cases (6,7). For this reason, additional factors are required to reliably predict the behaviour of these tumors (8). These additional factors may be morphological as well as molecular markers, verified by means of immunohistochemistry (IHC).

The aim of the present study was to analyze different morphological and molecular biological markers in GISTs, using well-defined morphological criteria in hematoxylin and eosin (H&E) as well as immunohistochemical stainings based on tissue microarray platform, in order to identify further possible determining factors for recurrence, metastasis and survival.

Materials and methods <sec> <title>Patient and sample characteristics

The present study included 201 formalin-fixed and paraffin-embedded (FFPE) tissue samples of primary GIST specimens resected between 1997 and 2014 [115 (57.2%) from female and 86 (42.8%) from male patients], with complete histopathological records (spindle-shaped, epitheloid or mixed histomorphological subtypes) and tumor classification according to the TNM staging system (9). The general characteristics of the investigated GISTs are summarized in Table I. The range of fixation time for the obtained specimens was determined between 12 and 24 h in 4% phosphate-buffered saline-formalin solution to avoid false-positive or -negative immunohistochemical staining patterns due to under- or over-fixation (10,11).

Morphology

A formally well-defined list of histomorphological criteria (7,12) was used to retrospectively investigate all GIST cases in terms of nuclear pleomorphism (NP; no, low, moderate, or severe), tumor cell necrosis (TCN; absence or presence), mucosal infiltration (MI; absence or presence), mucosal ulceration (MU; absence or presence), skeinoid fibers (SF; absence or presence) and growth pattern (GP; type I-luminal/intramural or type II-extramural).

Based on 5-µm H&E-stained FFPE sections, mitotic figures were counted in 50 consecutive HPFs as previously described (3,9) under a Leica DM2000 microscope (Leica Microsystems, Vienna, Austria) by two independent investigators (RK and DN).

IHC

IHC was routinely performed on an Autostainer Plus (Dako Austria GmbH, Vienna, Austria), according to the manufacturer's recommendations (13), using heat-induced epitope retrieval in pH 9.0 antigen retrieval buffer (Dako Austria GmbH) at 95°C for 40 min for the applied antibodies (Table II).

Immunohistochemical stainings were performed on whole slides for phosphohistone H3 (PHH3) and Ki-67, whereas all other immunohistochemical stainings were performed on tissue microarrays (TMAs), as previously described in detail (14). In brief, TMAs were prepared from paraffin-embedded tissues with a designed grid of 2.0-mm black circles leaving a 2.0-mm space between them on plain paper fixed to stainless steel moulds. Dried cores of paraffin-embedded GIST tissue blocks obtained by a 2.0-mm Harris Uni-Core punch (Ted Pella Inc., Redding, CA, USA) were aligned to the grid. Once all cores were attached to the mould, melted paraffin was gently poured into the mould and the TMA was handled according to routine histopathological procedures from this point onwards.

Interpretation of IHC

PHH3-positive cells were manually counted in 50 HPFs. The Ki-67-based proliferation rate was assessed by the optimized particle analysis module according to the software manual (Image Access 9 Enterprise; Imagic Bildverarbeitung AG, Glattbrugg, Switzerland) on three digitized hot spot areas and related to the total number of cells, as previously described (15).

All other immunohistochemical stainings were evaluated using a modified quickscore method by multiplication of the intensity [rated as 0 (negative), 1 (low), 2 (moderate) or 3 (strong)] and extent of positive cells (0–100%), which yields values ranging from 0 (negative) to a maximum of 300 (staining intensity of 3 and positive cells 100%) (16). Image evaluation was independently performed by two investigators (MP and DN). Differences in assessment were discussed until consensus was reached.

Ethics

This study was conducted following our National and Institutional guidelines and in accordance with the Declaration of Helsinki (1964); the samples are exclusively available for research purposes in retrospective studies. Based on the retrospective nature of this study and full anonymization of the patient data, the present study is not subject to formal approval by the appropriate local Ethics Committee (http://www.salzburg.gv.at/ethikkommission).

Statistical analysis

Statistical analysis was performed with IBM SPSS 20.0 software (IBM Corp., Armonk, NY, USA). The χ²-test or Student's t-test was used to compare data of nominal or interval level. The Student's t-test and univariate analysis of variance (ANOVA) using Bonferroni's post-hoc test to adjust for multiple comparisons were applied for differences between two and more groups of tissue samples, respectively. For survival analysis, cases with missing date of recurrence, metastasis or death were excluded. Multivariate analysis was performed using time-independent binary logistic regression or time-dependent Cox-regression technique for recurrence, metastasis or death. Based on the predicted probabilities of the regression analysis, dichotomous variables were calculated with optimized cluster analysis for additional survival analysis using the Kaplan-Meier method comparing the curves of recurrence, metastasis and survival with the log-rank test. For all calculations, P<0.05 and P<0.01 were considered to indicate statistically significant or highly significant differences, respectively.

Results <sec> <title>Major clinicopathological, morphological and immunohistochemical characteristics in GISTs Clinicopathological characteristics (<xref rid="tI-mco-0-0-819" ref-type="table">Table I</xref>)

Overall, 201 patients [115 women (57.2%) and 86 men (42.8%); median age at the time of diagnosis, 68.9 years] were included in the present study. The most common localization of GISTs was the stomach (n=127) followed by the small intestine (n=67), the rectum (n=5) and the esophagus (n=2), with a median tumor size of 4.0 cm (range, 0.1–32.0 cm), corresponding to the main tumor status of pT2, according to the current TNM guidelines (9), whereas the more locally advanced GISTs (pT3 and pT4) were limited to the stomach and the small intestine. Only 1 case (0.5%) exhibited lymph node metastasis, whereas the lymph node detection rate was overall very low [n=46 (22.9%)]. Multifocality [n=9 (4.5%)] was observed in the small intestine (n=6) and the stomach (n=3). The intention of surgery was mainly due to preoperatively diagnosed or suspected GIST [n=180 (89.6%)], or incidentally discovered in the remaining cases [n=21 (10.4%)]. Local organ-sparing GIST resection was the most frequently selected type of surgery [n=147 (73.1%)], with an incomplete resection (R1) in 30 cases (14.9%), mostly observed in the small intestine (20.9%). Recurrence of the GIST was observed in 18 cases (9.0%; mean/median, 35.4/24.6 months) and metastasis in 17 cases (8.5%; mean/median, 45.0/41.7 months). Finally, to date, 15 patients (7.5%; mean/median, 49.6/38.5 months) included in this study have succumbed to the disease. The cut-off date of the events was July 1st, 2015.

The statistical analysis revealed that the localization of the GISTs, resection status, tumor status (according to TNM), tumor size and H&E-based grading are significantly associated with recurrence of GISTs. Furthermore, tumor size and H&E-based grading are significantly associated with metastasis of GISTs (P<0.05, χ² and t-test). Finally, a significant association was observed between intention and type of surgery, as well as between multifocality and survival (P<0.05, χ² test), whereas all other clinicopathological characteristics did not affect recurrence, metastasis or patient survival.

Histomorphological characteristics (<xref rid="tIII-mco-0-0-819" ref-type="table">Table III</xref>)

As regards histomorphological pattern, spindle-cell morphology was predominantly observed (61.2%) compared with the epitheloid and mixed cell types. The median number of mitoses of 3 per 50 HPFs revealed that more cases of GISTs are of low mitotic rate according to the current TNM guidelines, whereby the mitosis count varies widely (≤266 per 50 HPFs), reflecting the mitotic heterogeneity of GISTs. Notably, NP was low, MI and MU were detectable in approximately one-third of all cases, whereas TCN and SF were most frequently found in GISTs. Finally, the luminal/intramural GP was predominant in this cohort of GISTs.

Comparing the different GIST sub-localizations, significant differences in grade of NP (P=0.044, ANOVA), TCN, MI and luminal/intramural GP (P<0.05, χ²-test) were observed in the small intestine.

Immunohistochemical findings (<xref rid="tIV-mco-0-0-819" ref-type="table">Table IV</xref>)

Based on the applied quickscore method by multiplication of the intensity and extent of the staining, the obtained scores for markers of differentiation (in increasing order, S-100<desmin<SMA<<<PDGFRα <<DOG-1<CD34<CD117) were significantly higher compared with those of cell proliferation/apoptosis markers (in increasing order, P53<P16<<CD44<Bcl-2).

Comparing the different GIST sub-localizations, a significant correlation between localization and expression of SMA, C-kit, DOG-1, desmin, CD34 and Bcl-2 were observed (P<0.01, ANOVA).

Univariate statistical analysis of association between clinicopathological, histomorphological and immunohistochemical characteristics Histomorphological vs. clinicopathological characteristics

The univariate analysis revealed that the majority of the morphological findings were significantly associated with low or high mitotic activity (NP, TCN, MI, MU and SF) (P<0.05, χ² test) and pT-stage (NP, TCN, MI, MU and GP) (P<0.01, χ² test), whereby only multifocality and resection status were significantly associated with the GIST GP (P<0.01, χ² test).

Compared with the events of recurrence, metastasis and survival, NP, TCN and GP exhibited a significant association with recurrence and metastasis (P<0.05, χ² test), whereas no morphological characteristics were associated with survival.

Immunohistochemical vs. clinicopathological characteristics

The statistical analysis revealed that Ki-67-based proliferation rate was significantly associated with all the clinicopathological findings, whereas only certain markers of differentiation (S-100, DOG-1, CD34, CD44 and desmin) and proliferation (P16, P53 and Bcl-2) were significantly associated clinicopathological parameters, such as size (S-100, DOG-1, CD34, CD44 and Bcl-2), multifocality (DOG-1), grading (PHH3, CD-117, desmin, CD44 and Bcl-2) and tumor stage (desmin, P16, P53 and Bcl-2) using t-test or ANOVA (P<0.05).

As regards recurrence, metastasis and survival status, Ki-67, CD117 and CD34 scores were significantly associated with events of recurrence, and Ki-67 and CD117 scores were associated with detected metastases. Finally, Ki-67, CD34 and PDGFRα scores were significantly different in surviving and non-surviving patients (P<0.05, t-test).

Histomorphological vs. immunohistochemical characteristics

Heterogeneous sets of immunohistochemical factors are statistically associated with NP (Ki-67, PHH3, S-100, PDGFRα and Bcl-2; P<0.01, ANOVA), TCN (Ki-67, PHH3, desmin, PDGFRα, P53 and Bcl-2) and SF (Ki-67, CD-34, PDGFRα, P53 and Bcl-2) as well as GP (Ki-67, PHH3, desmin and P53; P<0.05, t-test). Additionally, Ki-67-based proliferation rate was significantly associated with the histomorphological characteristics of MI and MU in GISTs (P<0.05, t-test).

Multivariate analysis of clinicopathological, histomorphological and immunohistochemical characteristics according to recurrence, metastasis and overall survival (<xref rid="tV-mco-0-0-819" ref-type="table">Table V</xref>)

The independent variables for multivariate analysis [binary logistic regression (BR) and Cox regression (CR) analysis, summarized in Table V] are listed in descending order of their odds ratio (OR).

Recurrence

GP, margin status, localization of the GIST and immunhistochemical CD34 score were time-independent predictors (BR), whereas the same variables (without CD34 score) were identified as time-dependent predictors (CR) for recurrence.

Metastasis

GP was the most significant independent predictor for both mathematical models (BR and CR). H&E-based mitosis count (per 50 HPFs) was identified as an additional independent predictor for BR and maximum size of the GIST was an independent predictor for CR.

Survival

Multifocality, Ki-67-based proliferation rate (per mm²) and immunhistochemical CD34 score were identified as independent survival predictors using BR. In comparison, only Ki-67-based proliferation rate (per mm²) was an independent factor for survival when applying CR.

Focusing on independent variables with an OR >5, only margin status, GP, multifocality and localization of GISTs were identified as independent variables with the highest effect (Table V).

Kaplan-Maier-analyses (<xref rid="f1-mco-0-0-819" ref-type="fig">Fig. 1</xref>)

Based on the predicted probabilities, dichotomous variables were calculated to perform further graphical assessment for the prognosis of recurrence, metastasis and survival in GISTs using the Kaplan-Meier method. Based on an optimized cluster analysis, high selective cut-offs for the probability of recurrence (0.1), metastasis (0.05) and survival (0.025) were calculated. The calculated models had the ability to significantly differentiate the risk for all time-to-event procedures for patients with GIST based on the evaluated independent risk factors (Kaplan-Meier method, log-rank test <0.05). Comparing the significance levels of the log-rank tests, the best prognostic model was obtained for recurrence, followed by metastasis and survival. Comparing the graphs, the event recurrence of patients with GIST occurred earlier compared with metastasis or death.

Discussion

Histomorphological parameters for predicting the biological behavior of GISTs have been analyzed and validated for several years leading to consensual statements. However, it remains difficult to distinguish benign from malignant tumors (3). In our study, we presented a large series of 201 GIST cases with complete clinicopathological evaluation and analyzed these cases for classical histomorphological criteria and immunohistochemical markers using TMAs. The clinical data of age, gender, localization, grading (mitotic activity), as well as surgical procedures and different outcome parameters (recurrence, metastasis and survival) are largely comparable to published GIST series (4), indicating absence of special bias factors.

In the univariate analysis, first of all, a number of variables were associated with recurrence, metastasis and survival, such as NP, TCN and GP (for recurrence and metastasis), CD117 and CD34 scores (for recurrence), CD117 score (for metastasis), and CD34 and PDGFRα scores (for survival). Of note, a considerable amount of the published clinicopathological investigations reveal heterogeneous subsets of prognostic and predictive markers, ranging from morphological parameters, such as gastrointestinal bleeding and tumor size, to molecular parameters, such as expression of SMA, CD34, DOG-1 or CD117, as well as novel markers, such as S-phase kinase-associated protein 2 or carbonic anhydrase II (17–20). However, the majority of these studies, as well as ours, were limited by the fact that most of the statistically significant variables based on univariate statistics are not statistically strong enough to ‘survive’ a further multivariate analysis, leading to the heretic question of whether IHC does actually provide additional prognostic data in GISTs (18). Furthermore, we were unable to confirm the role of differentiation and cell-cycle regulatory proteins, such as Bcl-2, P16 and P53, as well as CD44, as prognostic factors, similar to previous studies (8,21–24). The reasons may be different antibodies, antigen retrieval or inter-observer variations of the immunohistochemical staining (10,11,15).

In the multivariate analysis, the number of variables in a time-independent/time-dependent analysis was significantly reduced from a total of 31 (100%) to 4 (12.9%)/3 (9.6%) for recurrence, 2 (6.4%)/2 (6.4%) for metastasis and 3 (9.6%)/1 (3.2%) for survival (Table V), indicating the rarity of suitable endpoint markers available for GISTs to date. Interestingly, classical and ‘conservative’/‘well-established’ parameters, such as tumor localization, margin status, multifocality, as well as H&E-based mitosis and Ki-67-based proliferation rate of GISTs, were significantly associated with defined clinical endpoints, as already described by the European Sarcoma Network Working Group in 2012 (25). This, however, confirms earlier clinical studies (26,27) and indicates that powerful prognostic molecular markers (with the exception of CD34) are still not available to date. Of note, in contrast to the findings of Demir et al (18), lower immunhistochemical CD34 score was associated with higher recurrence rate and shorter survival, whereas this significant effect of CD34 expression was lost when performing a multivariate Cox regression analysis.

Additionally, a ‘new’ and prognosis-relevant parameter as reported by Agaimy et al (7) is GP, namely type I (luminal/intramural) or type II (extramural), which was found to be an independent prognostic factor for recurrence and metastasis (with an OR of ≤13.715 for recurrence). Therefore, preoperative radiological analysis and GP evaluation is essential for the surgical strategies of local or extensive resection. Furthermore, GP should be macroscopically described in detail by the surgeons and pathologists and should be included in further tumor stratifications of GISTs, such as the classical risk stratification scheme of Miettinen et al and others (2,28–30).

Comparing the current uni- and multivariate statistical analyses, it becomes evident that more parameters may be associated with the clinicopathological endpoints (recurrence, metastasis and survival) in the univariate rather than in the multivariate statistical analysis. This is clearly caused by the applied statistical approaches of excluding variables using ANOVA with forward WALD procedures. Therefore, the statistical power of the univariate analysis is very limited, as it was similarly found in most previous GIST studies on prognostic makers, whereas all these mathematical procedures may include bias to disregard the biological impact of possible predictive variables (31). However, we were able to calculate statistically significant models based on the predicted probabilities of the regression analysis to provide prognostic information for recurrence, metastasis and survival of patients with GISTs. Such an approach may be helpful in developing such ‘heat maps’, as reported by Joensuu et al, to provide clinicians and patients with better prognostic information for relevant endpoints (32).

In summary, our investigations demonstrated that i) classical parameters of GISTs, such as localization, multifocality and H&E-based mitosis/Ki-67-based proliferation, have the most relevant prognostic value; ii) GP of GISTs must be taken into consideration as new prognostic factor and further validated; and iii) there remains a need for efficient molecular markers for GISTs.

Acknowledgements

The expert technical assistance of Mrs. Berta Lechner, Mrs. Monika Prinz, Mrs. Ines Grob-Achleitner and Mr. Brian van Merkestijn is gratefully acknowledged. The present study was supported by the Wissenschaftlicher Verein der Pathologie Salzburg/Austria.

Abbreviations

ANOVA

analysis of variance

Bcl-2

B-cell lymphoma 2

confidence interval

DOG-1

discovered on GIST 1

FFPE

formalin-fixed and paraffin-embedded

GIST

gastrointestinal stromal tumor

growth pattern

H&E

hematoxylin and eosin

HPF

high-power field

IHC

immunohistochemistry

mucosal infiltration

mucosal ulceration

nuclear pleomorphism

odds ratio

PDGFRα

platelet-derived growth factor receptor α

PHH3

phosphohistone H3

skeinoid fibers

SMA

smooth muscle actin

TCN

tumor cell necrosis

TMA

tissue microarray

References 1

Nilsson

Bümming

Meis-Kindblom

Odén

Dortok

Gustavsson

Sablinska

Kindblom

Gastrointestinal stromal tumors: The incidence, prevalence, clinical course, and prognostication in the preimatinib mesylate era - a population-based study in western Sweden

Cancer1038218292005

10.1002/cncr.20862

15648083

Miettinen

Lasota

Gastrointestinal stromal tumors: Pathology and prognosis at different sites

Semin Diagn Pathol2370832006

10.1053/j.semdp.2006.09.001

17193820

Fletcher

Berman

Corless

Gorstein

Lasota

Longley

Miettinen

O'Leary

Remotti

Rubin

Diagnosis of gastrointestinal stromal tumors: A consensus approach

Hum Pathol334594652002

10.1053/hupa.2002.123545

12094370

Sorour

Kassem

Ghazal

Ael-H

El-Riwini

Abu

Nasr A

Gastrointestinal stromal tumors (GIST) related emergencies

Int J Surg122692802014

10.1016/j.ijsu.2014.02.004

24530605

Miettinen

El Rifai

Sobin

Lasota

Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: A review

Hum Pathol334784832002

10.1053/hupa.2002.124123

12094372

Rossi

Miceli

Messerini

Bearzi

Mazzoleni

Capella

Arrigoni

Sonzogni

Sidoni

Toffolatti

Natural history of imatinib-naive GISTs: A retrospective analysis of 929 cases with long-term follow-up and development of a survival nomogram based on mitotic index and size as continuous variables

Am J Surg Pathol35164616562011

10.1097/PAS.0b013e31822d63a7

21997685

Agaimy

Vassos

Wunsch

Hohenberger

Hartmann

Croner

Impact of serosal involvement/extramural growth on the risk of synchronous and metachronous peritoneal spread in gastrointestinal stromal tumors: Proposal for a macroscopic classification of GIST

Int J Clin Exp Pathol512222012

22295142

Panizo-Santos

Sola

Vega

de Alava

Lozano

Idoate

Pardo-Mindán

Predicting metastatic risk of gastrointestinal stromal tumors: Role of cell proliferation and cell cycle regulatory proteins

Int J Surg Pathol81331442000

10.1177/106689690000800208

11493978

Tannapfel

Wittekind

The current TNM system for gastrointestinal tumors part II

Pathologe313483522010(In German)

10.1007/s00292-010-1305-2

20798945

Illig

Fritsch

Schwarzer

Breaking the seals: Efficient mRNA detection from human archival paraffin-embedded tissue

RNA:15158815962009

10.1261/rna.1278109

19549718

Illig

Fritsch

Schwarzer

Spatio-temporal expression of HOX genes in human hindgut development

Dev Dyn24253662013

10.1002/dvdy.23893

23073994

Brainard

Goldblum

Stromal tumors of the jejunum and ileum: A clinicopathologic study of 39 cases

Am J Surg Pathol214074161997

10.1097/00000478-199704000-00006

9130987

Kemmerling

Stintzing

Mühlmann

Dietze

Neureiter

Primary testicular lymphoma: A strictly homogeneous hematological disease?

Oncol Rep23126112672010

20372839

Kemmerling

Alinger

Dietze

Bösmüller

Ocker

Wolkersdörfer

Berr

Neureiter

Kiesslich

Association of stem cell marker expression pattern and survival in human biliary tract cancer

Int J Oncol415115222012

22614781

Kemmerling

Weyland

Kiesslich

Illig

Klieser

Jäger

Dietze

Neureiter

Robust linear regression model of Ki-67 for mitotic rate in gastrointestinal stromal tumors

Oncol Lett77457492014

24527082

Kiesslich

Alinger

Wolkersdörfer

Ocker

Neureiter

Berr

Active Wnt signalling is associated with low differentiation and high proliferation in human biliary tract cancer in vitro and in vivo and is sensitive to pharmacological inhibition

Int J Oncol3649582010

19956832

Tian

Guan

Zhao

Dong

Hao

SKP2 high expression, KIT exon 11 deletions, and gastrointestinal bleeding as predictors of poor prognosis in primary gastrointestinal stromal tumors

PLoS One8e629512013

10.1371/journal.pone.0062951

23690967

Demir

Ekinci

Erten

Kucukzeybek

Alacacioglu

Somali

Can

Dirican

Bayoglu

Akyol

Does immunohistochemistry provide additional prognostic data in gastrointestinal stromal tumors?

Asian Pac J Cancer Prev14475147582013

10.7314/APJCP.2013.14.8.4751

24083738

Liu

Zhao

Chen

Overexpression of carbonic anhydrase II and Ki-67 proteins in prognosis of gastrointestinal stromal tumors

World J Gastroenterol19247324802013

10.3748/wjg.v19.i16.2473

23674848

Zhao

Wang

Meng

Tumor sites and microscopic indicators are independent prognosis predictors of gastrointestinal stromal tumors

Tohoku J Exp Med23365722014

10.1620/tjem.233.65

24827382

Rudolph

Gloeckner

Parwaresch

Harms

Schmidt

Immunophenotype, proliferation, DNA ploidy, and biological behavior of gastrointestinal stromal tumors: A multivariate clinicopathologic study

Hum Pathol297918001998

10.1016/S0046-8177(98)90447-6

9712419

Gumurdulu

Erdogan

Kayaselcuk

Seydaoglu

Parsak

Demircan

Tuncer

Expression of COX-2, PCNA, Ki-67 and p53 in gastrointestinal stromal tumors and its relationship with histopathological parameters

World J Gastroenterol134264312007

10.3748/wjg.v13.i3.426

17230613

Fujita

Yamamoto

Imamura

Nakamura

Maehara

Tsuneyoshi

Oda

Expression level of the mitotic checkpoint protein and G2-M cell cycle regulators and prognosis in gastrointestinal stromal tumors in the stomach

Virchows Arch4601631692012

10.1007/s00428-011-1181-z

22190007

Meara

Cangiarella

Simsir

Horton

Eltoum

Chhieng

Prediction of aggressiveness of gastrointestinal stromal tumours based on immunostaining with bcl-2, Ki-67 and p53

Cytopathology182832892007

10.1111/j.1365-2303.2007.00505.x

17883690

ESMO/European Sarcoma Network Working Group: Gastrointestinal stromal tumors: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up

Ann Oncol23Suppl 7vii49vii552012

22997454

Emory

Sobin

Lukes

Lee

O'Leary

Prognosis of gastrointestinal smooth-muscle (stromal) tumors: Dependence on anatomic site

Am J Surg Pathol2382871999

10.1097/00000478-199901000-00009

9888707

Yan

Marchettini

Acherman

Gething

Brun

Sugarbaker

Prognostic assessment of gastrointestinal stromal tumor

Am J Clin Oncol262212282003

10.1097/01.COC.0000018296.45892.CE

12796588

Miettinen

Lasota

Gastrointestinal stromal tumors: Review on morphology, molecular pathology, prognosis and differential diagnosis

Arch Pathol Lab Med130146614782006

17090188

Miettinen

Lasota

Histopathology of gastrointestinal stromal tumor

J Surg Oncol1048658732011

10.1002/jso.21945

22069171

Reichardt

Gastrointestinal stromal tumors (GIST)

Dtsch Med Wochenschr138201320162013(In German)

10.1055/s-0033-1349575

24065407

Lovell

Biological importance and statistical significance

J Agric Food Chem61834083482013

10.1021/jf401124y

23909755

Joensuu

Vehtari

Riihimäki

Nishida

Steigen

Brabec

Plank

Nilsson

Cirilli

Braconi

Risk of recurrence of gastrointestinal stromal tumour after surgery: An analysis of pooled population-based cohorts

Lancet Oncol132652742012

10.1016/S1470-2045(11)70299-6

22153892

Figure 1.

Kaplan-Maier curves for (A1) recurrence, (B1) metastasis and (C1) survival stratified on the predicted probabilities (A2, B2 and C2, respectively) for these events based on the logistic regression analysis. The cut-off values for the predicted probabilities were added to the graphs A1, B1 and C1.

Table I.

General characteristics of investigated GIST series.

	Tumor site

Characteristics	Esophagus	Stomach	Small intestine	Rectum	All sites
Total cases, n (%)	2 (1.0)	127 (63.2)	67 (33.3)	5 (2.5)	201
Sublocation^a, n (%)		14 (11.0)-52 (41.0)-28 (22.0)-33 (26.0)	20 (30.0)-14 (21.0)-6 (9.0)-27 (40.0)
Age, years
Mean (median)	65.7 (65.7)	68.7 (69.6)	64.9 (68.5)	69.8 (64.5)	68.5 (68.9)
Range	64–67	29–98	20–92	50–92	20–98
Intention of surgery, n (%)^b
GIST	2 (100.0)	114 (89.8)	60 (89.6)	4 (80.0)	180 (89.6)
Others^c	0 (0.0)	13 (10.1)	7 (10.8)	1 (20.0)	21 (10.4)
Margin status, n (%)^b
R0	2 (100.0)	112 (88.2)	53 (79.1)	4 (80.0)	171 (85.1)
R1	0 (0.0)	15 (11.8)	14 (20.9)	1 (20.0)	30 (14.9)
Metastases, n (%)^b
Yes	0 (0.0)	8 (6.3)	9 (13.4)	0 (0.0)	17 (8.5)
No	2 (100.0)	119 (93.7)	58 (86.6)	5 (100.0)	184 (91.5)
Mitotic activity^d, n (%)^b
Low	2 (10.0)	87 (67.4)	45 (69.2)	4 (80.0)	138 (68.7)
High	0 (0.0)	42 (32.6)	20 (30.8)	1 (20.0)	63 (31.3)
Multifocality, n (%)^b
Yes-median	0 (0.0)-0	3 (2.4)-2	6 (9.0)-4	0 (0.0)-0	9 (4.5)-3
No	2 (100.0)	124 (97.6)	61 (91.0)	5 (100.0)	192 (95.5)
Nodal status, n (%)^b
NX^e	2 (100.0)	103 (81.1)	46 (68.7)	4 (80.0)	155 (77.1)
N0	0 (0.0)	23 (18.1)	21 (31.3)	1 (20.0)	45 (22.4)
N1	0 (0.0)	1 (0.8)	0 (0.0)	0 (0.0)	1 (0.5)
Recurrence, n (%)^b
Yes	0 (0.0)	5 (3.9)	12 (17.9)	1 (20.0)	18 (9.0)
No	2 (100.0)	122 (96.1)	55 (82.1)	4 (80.0)	183 (91.0)
Gender, n (%)^b
Male	2 (100.0)	58 (45.7)	25 (37.2)	1 (20.0)	86 (42.8)
Female	0 (0.0)	69 (54.3)	42 (62.7)	4 (80.0)	115 (57.2)
Size, cm
Mean (median)	2.2 (2.2)	4.5 (4.0)	4.7 (3.8)	2.1 (1.5)	4.5 (4.0)
Range	1.4–3	0.1–32	0.7–27	0.2–5	0.1–32.0
Survival, n (%)^b
Yes	2 (100.0)	120 (94.5)	59 (88.1)	5 (100.0)	186 (92.5)
No	0 (0.0)	7 (5.5)	8 (11.9)	0 (0.0)	15 (7.5)
Tumor status, n (%)^b
T1	1 (50.0)	34 (26.8)	18 (26.9)	3 (60.0)	56 (27.9)
T2	1 (50.0)	60 (47.2)	32 (47.8)	2 (40.0)	95 (47.3)
T3	0 (0.0)	24 (18.9)	10 (14.9)	0 (0.0)	34 (16.9)
T4	0 (0.0)	9 (7.1)	7 (10.4)	0 (0.0)	16 (8.0)
Type of surgery, n (%)^b
Local resection	2 (100.0)	100 (78.7)	41 (61.2)	4 (80.0)	147 (73.1)
Organ resection	0 (0.0)	11 (8.7)	11 (16.4)	1 (20.0)	23 (11.4)
Multiorgan resection	0 (0.0)	12 (9.4)	14 (20.9)	0 (0.0)	26 (12.9)
Others^f	0 (0.0)	4 (3.1)	1 (1.5)	0 (0.0)	5 (2.5)

Stomach, fundus-corpus-antrum-unknown; small intestine, duodenum-jejunum-ileum-unknown.

Percentage per column, i.e., per tumor site.

Resection of a known carcinoma.

According to the current TNM guidelines (7th edition, 2010).

No lymph nodes were detected in the specimen, since lymph node resection is not mandatory in GISTs.

Incidental findings during autopsy. GIST, gastrointestinal stromal tumors.

Table II.

Antibodies and conditions used for immunohistochemistry.

Antibody	Catalogue no.	Clone	Species	Dilution/incubation	Pretreatment^a
Bcl-2	M0887^b	124	Mouse	1:100/30 min	WB/pH 9
CD34	IR632^b	QB END10	Mouse	Ready to use/30 min	WB/pH 9
CD44	NCL-CD44–2^c	DF 1485	Mouse	1:100/30 min	WB/pH 9
CD117 (C-kit)	A4502^b	C-kit	Rabbit	1:200/30 min	WB/pH 9
Desmin	IR606^b	D33	Mouse	Ready to use/30 min	WB/pH 9
DOG-1	NCL-L-DOG-1^c	K9	Mouse	1:100/30 min	WB/pH 9
Ki67	M7240^b	MIB-1	Mouse	1:500/30 min	WB/pH 9
P16	725 4713^d	EGH4	Mouse	Ready to use/30 min	pH 9
P53	M7001^b	DO7	Mouse	1:200/30 min	WB/pH 9
PDGFRα	RB-9027-P1^e	Not reported	Rabbit	1:50/30 min	pH 9
PHH3	369A-18^f	Not reported	Rabbit	Ready to use/30 min	WB/pH 9
S-100	Z 0311^b	Not reported	Rabbit	1:2000/30 min	WB/pH 6
SMA	M0851^b	1A4	Mouse	1:600/30 min	WB/pH 9

pH 6 or 9, heat-induced epitope retrieval in pH 6 or 9 antigen retrieval buffer (Dako,Glostrup, Denmark) and WB in a Pascal pressurized heating chamber (Dako). Purchased from

Dako

Novocastra/Leica Microsystems (Vienna, Austria)

Ventana (Tucson, AZ, USA)

Thermo Scientific (Waltham, MA, USA) and

Cell Morque (Rocklin, CA, USA). Bcl-2, B-cell lymphoma 2; DOG-1, discovered on GIST 1; PDGFRα, platelet-derived growth factor receptor α; PHH3, phosphohistone H3; SMA, smooth muscle actin; WB, waterbath; GIST, gastrointestinal stromal tumor; MIB-1, mindbomb E3 ubiquitin protein ligase 1.

Table III.

Histomorphological characteristics of the GIST series.

	Tumor site

Characteristics	Esophagus	Stomach	Small intestine	Rectum	All sites
Growth pattern^a,b, n (%)^c
Luminal/intramural	2 (100.0)	107 (84.3)	46 (68.7)	5 (100.0)	160 (79.6)
Extramural	0 (100.0)	20 (15.7)	21 (31.3)	0 (100.0)	41 (20.4)
Histomorphological pattern, n (%)^c
Spindle-cell	2 (100.0)	72 (56.7)	45 (67.2)	4 (80.0)	123 (61.2)
Epitheloid	0 (0.0)	29 (22.8)	9 (13.4)	1 (20.0)	39 (19.4)
Mixed type	0 (0.0)	26 (20.5)	13 (19.4)	0 (0.0)	39 (19.4)
Mitoses/50 HPFs, n (%)^c
Median (range)	1 (1–1)	3 (0–266)	3 (0–117)	2 (0–6)	3 (0–266)
Mean ± SD	1.00±1.0	7.83±24.10	12.76 ±23.6	2.60±2.4	9.27±23.6
Mucosal infiltration^b, n (%)^c
Yes	2 (100.0)	85 (66.9)	34 (50.7)	5 (100.0)	126 (62.7)
No	0 (0.0)	42 (33.1)	33 (49.3)	0 (0.0)	75 (37.3)
Mucosal ulceration, n (%)^c
Yes	2 (100.0)	87 (68.5)	43 (64.2)	4 (80.0)	136 (67.7)
No	0 (0.0)	40 (31.5)	24 (35.8)	1 (20.0)	65 (32.3)
Nuclear pleomorphism^b,d
Median (range)	0.5 (0–1.0)	1.0 (0–2,5)	1.5 (0–2,5)	1.0 (0–1,5)	1.0 (0–2,5)
Mean ± SD	0.50±0.707	1.20±0.543	1.33±0.519	0.90±0.548	1.23±0.543
Skeinoid fibers, n (%)^c
Yes	1 (50.0)	56 (44.1)	39 (58.2)	2 (40.0)	98 (48.8)
No	1 (50.0)	71 (55.9)	28 (41.8)	3 (60.0)	103 (51.2)
Tumor cell necrosis^b, n (%)^c
Yes	2 (100.0)	42 (33.1)	22 (32.8)	4 (80.0)	70 (34.8)
No	0 (0.0)	85 (66.9)	45 (67.2)	1 (20.0)	131 (65.2)

According to Agaimy et al (7).

Significant differences between GIST sub-localizations using χ² test or analysis of variance.

Percentage per column, i.e., per tumor site.

0, no; 1, low; 2, moderate; and 3, strong. GIST, gastrointestinal stromal tumor; HPFs, high-power fields; SD, standard deviation.

Table IV.

Immunohistochemical characteristics of the GIST series.

	Tumor site

Characteristics	Esophagus	Stomach	Small intestine	Rectum	All sites
Bcl-2^a
I	0.5 (0.5)	1.9 (2.0)	2.6 (3.0)	1.8 (1.5)	2.1 (2.0)
E	5 (5.0)	70 (80.0)	86 (90.0)	58 (60.0)	74 (90)
S	5±7	158±98	236±84	123±117	178±102
CD117 (C-kit)^a
I	1.8 (1.8)	2.0 (2.0)	2.5 (3.0)	2.6 (2.8)	2.2 (2.0)
E	58 (57.5)	79 (95.0)	93 (100.0)	99 (100.0)	84 (100.0)
S	109±101	183±101	236±89	260±49	202±99
CD34^a
I	1.8 (1.8)	2.4 (3.0)	1.7 (2.0)	2.5 (2.5)	2.2 (2.5)
E	50 (50.0)	81 (100.0)	52 (45.0)	96 (97.5)	72 (95.0)
S	95±92	220±106	126±117	239±45	189±117
CD44
I	0.5 (0.5)	0.7 (1.0)	1.0 (1.0)	0.5 (0.5)	0.8 (1.0)
E	3 (2.5)	25 (10.0)	34 (30.0)	13 (10.0)	27 (20.0)
S	3±4	39±64	62±92	13±15	46±75
Desmin^a
I	1.0 (1.0)	0.3 (0.0)	0.2 (0.0)	0.4 (0.0)	0.3 (0.0)
E	48 (47.5)	4 (0.0)	3 (0.0)	2 (0.0)	4 (0.0)
S	95±134	6±19	7±37	2±3	7±29
DOG-1²
I	1.5 (1.5)	1.6 (1.5)	2.2 (2.0)	2.5 (2.5)	1.8 (2.0)
E	48 (47.5)	70 (80.0)	86 (100.0)	91 (97.5)	75 (90.0)
S	87±103	133±92	209±94	231±75	160±99
Ki-67 (/mm²)^b
Median (range)	47.2 (44–50)	122 (16–2,011)	133 (22–1,322)	161 (0–377)	127 (0–2,011)
Mean ± SD	47.2±3.8	165±210	227±255	152±149	184±226
PDGFRα
I	1.0 (1.0)	1.8 (2.0)	1.9 (2.0)	0.9 (1.0)	1.8 (2.0)
E	45 (45.0)	77 (85.0)	83 (90.0)	58 (70.0)	78 (85.0)
S	45±49	150±88	154±79	68±55	148±85
PHH3 per 50 HPFs
Median (range)	3.5 (0–7)	11 (0–605)	15 (1–219)	7 (3–15)	13 (0–605)
Mean ± SD	3.5±4.9	27.3±77.3	29.6±42.6	8.3±6.1	27.6±66.3
P16
I	0.5 (0.5)	0.6 (0.0)	0.9 (1.0)	0.7 (1.0)	0.7 (0.5)
E	20 (20.0)	11 (0.0)	21 (10.0)	15 (5.0)	14 (2.5)
S	20±28	17±38	36±69	15±22	23±50
P53
I	0.0 (0.0)	0.4 (0.0)	0.5 (0.0)	0.3 (0.0)	0.4 (0.0)
E	0 (0.0)	6 (0.0)	11 (0.0)	5 (0.0)	8 (0.0)
S	0±0	7±15	17±41	5±10	10±27
SMA^a
I	1.5 (1.5)	0.6 (0.0)	1.0 (1.0)	0.5 (0.5)	0.7 (1.0)
E	50 (50.0)	11 (0.0)	23 (10.0)	4 (2.5)	15 (5.0)
S	150±212	18±43	39±56	4±5	26±52
S-100
I	0.0 (0.0)	0.2 (0.0)	0.2 (0.0)	0.4 (0.0)	0.2 (0.0)
E	0 (0.0)	3 (0.0)	3 (0.0)	2 (0.0)	3 (0.0)
S	0±0	5±22	6±38	2±3	5±28

I and E are expressed as mean (median); S is expressed as mean ± SD.

Significant differences between GIST sub-localizations using analysis of variance.

Analyzed as previously described (15). GIST, gastrointestinal stromal tumor; I, intensity; E, extent; S, quickscore; HPFs, high-power fields; SD, standard deviation; Bcl-2, B-cell lymphoma 2; DOG-1, discovered on GIST 1; PDGFRα, platelet-derived growth factor receptor α; PHH3, phosphohistone H3; SMA, smooth muscle actin.

Table V.

Assessments of recurrence, metastasis and survival by either BR or CR analyses.

		Recurrence			Metastasis			Survival

Variables	BR or CR	Variable(s)	OR (95% CI)	n	Variable(s)	OR (95% CI)	n	Variable(s)	OR (95% CI)	n
Clinicopathological (n=10)
Age, intention of surgery,	BR	Localization	3.523
localization, margin			(1.204–10.314)	2			0	Multifocality	8.402	1
status, grade (TNM),		Margin status	5.486						(0.903–78.144)
multifocality, gender, size,			(0.924–32.569)
tumor status (TNM),	CR	Localization	5.373			1.161
type of surgery			(1.742–16.571)	2	Size	(0.982–1.374)	1			0
		Margin status	5.132
			(1.109–23.747)
Morphological (n=8)
Growth pattern,	BR	Growth	6.043		Mitoses/50 HPFs	1.023
histomorphological		pattern	(1.255–29.098)	1		(0.996–1.050)	2			0
pattern, mitoses/50 HPFs,					Growth pattern	4.726
mucosal infiltration,						(1.223–18.253)
mucosal ulceration,	CR	Growth	13.715		Growth pattern	6.632
nuclear pleomorphism,		pattern	(2.556–73.605)	1		(0.974–45.158)	1			0
skeinoid fibers, tumor
cell necrosis
Immunohistochemical (n=13)
Bcl-2, CD117 (C-kit),	BR	CD34	0.991					Ki-67/mm²	1.002
CD34, CD44, desmin,			(0.983–0.999)	1			0		(1.000–1.004)	1
DOG-1, Ki-67/mm²,								CD34	0.992
PDGFRα, PHH3, P16,									(0.984–1.000)
P53, SMA, S-100	CR			0			0	Ki-67/mm²	1.003
									(1.001–1.004)	1
Total (n=31)	BR/CR			4/3			2/2			3/1
		Further risk classification of GIST cases based on predicted probabilities of margin status, localization, multifocality and growth pattern (see Fig.1 for Kaplan-Meier analyses).

All analysed variables are listed and the bold print indicates statistical significance. In the column of recurrence, metastasis and survival, the variables identified via multivariate statistical analysis are listed and the bold print indicates OR ≥5.0. BR, binary logistic regression; CR, Cox regression; OR, odds ratio; CI, confidence interval; HPFs, high-power fields; Bcl-2, B-cell lymphoma 2; DOG-1, discovered on GIST 1; PDGFRα, platelet-derived growth factor receptor α; PHH3, phosphohistone H3; SMA, smooth muscle actin.