Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Neuropathology and Applied Neurobiology (2015), 41, e41–e55 doi: 10.1111/nan.12185 Differential role of angiogenesis and tumour cell proliferation in brain metastases according to primary tumour type: analysis of 639 cases A. S. Berghoff*†‡, A. Ilhan-Mutlu†‡, C. Dinhof†‡, M. Magerle†‡, M. Hackl§, G. Widhalm†¶, J. A. Hainfellner*†, K. Dieckmann†**, J. Pichler††, M. Hutterer‡‡§§, T. Melchardt¶¶, R. Bartsch†‡, C. C. Zielinski†‡, P. Birner†*** and M. Preusser†‡ *Institute of Neurology, Medical University of Vienna, Vienna, Austria, †Comprehensive Cancer Center CNS Tumors Unit, Medical University of Vienna, Vienna, Austria, ‡Department of Medicine I, Medical University of Vienna, Vienna, Austria, §Austrian National Cancer Registry, Statistics Austria, Vienna, Vienna, Austria, ¶Department of Neurosurgery, Medical University of Vienna, Vienna, Austria, **Department of Radiotherapy, Medical University of Vienna, Vienna, Austria, ††Department of Medicine and Neurooncology, Landes-Nervenklinik Wagner-Jauregg, Linz, Austria, ‡‡Department of Neurology, Christian-Doppler-Klinik, Paracelsus Medical University, Salzburg, Austria, §§Department of Neurology and Wilhelm-Sander Neurooncology Therapy Unity, University Hospital and Medical School Regensburg, Regensburg, Germany, ¶¶3rd Medical Department, Paracelsus Medical University Hospital Salzburg, Salzburg, Austria, and ***Department of Pathology, Medical University of Vienna, Vienna, Austria A. S. Berghoff, A. Ilhan-Mutlu, C. Dinhof, M. Magerle, M. Hackl, G. Widhalm, J. A. Hainfellner, K. Dieckmann, J. Pichler, M. Hutterer, T. Melchardt, R. Bartsch, C. C. Zielinski, P. Birner and M. Preusser (2015) Neuropathology and Applied Neurobiology 41, e41–e55 Differential role of angiogenesis and tumour cell proliferation in brain metastases according to primary tumour type: analysis of 639 cases Aim: We aimed to characterize angiogenesis and proliferation and their correlation with clinical characteristics in a large brain metastasis (BM) series. Methods: Ki67 proliferation index, microvascular density (MVD) and hypoxiainducible factor 1 alpha (HIF-1 alpha) index were determined by immunohistochemistry in BM and primary tumour specimens. Results: Six hundred thirty-nine BM specimens of 639 patients with lung cancer (344/639; 53.8%), breast cancer (105/639; 16.4%), melanoma (67/ 639; 10.5%), renal cell carcinoma (RCC; 52/639; 8.1%) or colorectal cancer (CRC; 71/639; 11.1%) were available. Specimens of the corresponding primary tumour were available in 113/639 (17.7%) cases. Median Ki67 index was highest in CRC BM and lowest in RCC BM (P < 0.001). MVD and HIF-1 alpha index were both highest in RCC BM and lowest in melanoma BM (P < 0.001). Significantly higher Ki67 indices, MVD and HIF-1 alpha indices in the BM than in matched primary tumours were observed for breast cancer, non-small cell lung cancer (NSCLC) and CRC. Correlation of tissue-based parameters with overall survival in individual tumour types showed a favourable and independent prognostic impact of low Ki67 index [hazard ratio (HR) 1.015; P < 0.001] in NSCLC BM and of low Ki67 index (HR 1.027; P = 0.008) and high angiogenic activity (HR 1.877; P = 0.002) in RCC. Conclusion: Our data argue for differential pathobiological and clinical relevance of Ki67 index, HIF1-alpha index and MVD between primary tumour types in BM patients. An independent prognostic impact of tissue-based characteristics was observed in patients with BM from NSCLC and RCC, supporting the incorporation of these tissue-based parameters into diagnosis-specific prognostic scores. Keywords: brain metastases, HIF-1 alpha index, ki67 proliferation index, microvascular density Correspondence: Matthias Preusser, Department of Medicine I and Comprehensive Cancer Center, Central Nervous System Tumours Unit (CCC-CNS), Medical University of Vienna, Waehringer Guertel 18-20,1090 Vienna, Austria. Tel: +43 1 40400 4457; Fax: +43 1 40400 6088; E-mail: [email protected] © 2014 British Neuropathological Society e41 e42 A. S. Berghoff et al. Introduction Brain metastases (BM) are a frequent and devastating complication in cancer patients. The prognosis of BM patients is generally poor, although long-term survivors exist. Treatment strategies rely mainly on local approaches such as radiotherapy and neurosurgery. So far, progress in clinical approaches to BM patients has been hampered by a lack of research projects aiming at elucidating basic principles of BM formation. However, better understanding of BM pathobiology may lead to improvements in clinical patient care, for example, by enabling development of improved tools for prognostic evaluation or identification of targets for prevention and treatment of BM using specific inhibitor drugs [1–5]. Proliferation and hypoxia-induced angiogenesis are hallmarks of cancer and have repeatedly been implicated to be relevant in BM formation and growth [6]. For example, Ki67 proliferation index was postulated as a prognostic factor in a smaller series of BM [7]. Furthermore, early angiogenesis, especially via vascular co-option, was pointed out as an early mandatory mechanism in the establishment of micrometastases [8]. However, BM originate from diverse types of primary tumours and the expression profiles of hypoxia-associated factors and angiogenic patterns and their clinical relevance may differ between cancer subtypes. For example, outgrowth of BM of nonsmall cell lung cancer (NSCLC) has been shown to depend on early angiogenesis in an animal model, while melanoma cells showed a tendency for perivascular growth (‘vascular co-option’) without prominent neo-angiogenesis [9]. However, systematically collected data on human tissue samples is only available from some smaller studies. Therefore, we investigated in the presented study the Ki67 tumour cell proliferation index, expression of hypoxia inducible factor 1 alpha (HIF-1 alpha), and microvascular density (MVD) and their prognostic value in a large and well-characterized cohort of BM patients. Our analyses included comparative evaluation of these factors between cases with lung cancer, breast cancer, melanoma, renal cell carcinoma or colorectal cancer. Methods Patients All patients with BM of histologically proven lung cancer, breast cancer, melanoma, renal cell carcinoma or © 2014 British Neuropathological Society colorectal cancer and neurosurgical resection of BM between January 1990 and February 2011 were identified from the Neuro-Biobank, Medical University of Vienna. Only patients with availability of at least one tissue block containing formalin-fixed and paraffinembedded viable BM tissue for research purposes were eligible for inclusion. Clinical data were retrieved by chart review, and survival data were obtained from the National Cancer Registry of Austria and the Austrian Brain Tumor Registry [10,11]. Diagnosis-specific graded prognostic assessment (DSGPA) was calculated according to previously published clinical characteristics. DS-GPA for lung cancer included age (>60, 50–60, <50), Karnofsky performance score (<70, 70–80, 90–100), status of extracranial metastatic disease (present, absent) and number of BM (>3, 2–3, 1). DS-GPA for melanoma and renal cell carcinoma was calculated based on Karnofsky performance score (<70, 70–80, 90–100) and number of BM (>3, 2–3, 1). DS-GPA for breast cancer included Karnofsky performance score (<50, 60, 70–80, 90–100), breast cancer subtype (triple negative, luminal A, HER2, luminal B) and age (>60, <60). For colorectal cancer, DS-GPA was calculated only based on Karnofsky performance score (<70, 70, 80, 90, 100) [12]. Prognostic score for SCLC was calculated according to previously published clinical characteristics including Karnofsky performance score (<70, ≥70), number of BM (1–3, ≥4) and status of extracranial disease (present, absent) [13]. The ethics committee of the Medical University of Vienna approved the study (Vote 078/2004). Tissue-based analysis Immunohistochemistry was performed on full slides by using an automated horizontal slide-processing system (AutostainerPlusLink, Dako, Glastrup, Denmark; Ki67) and a fully automated multi-modal slide-staining system (Benchmark ULTRA, Ventana, Tuscon, AZ, USA; HIF1alpha, CD34) according to previously published standard protocol [14,15]. Immunohistochemical staining results were analysed using light microscopy and a counting graticule. For evaluation of Ki67 proliferation index, area with the highest density of cells showing a specific, strong nuclear immunohistochemical signal was identified. Within this ‘hot spot’, 500 cells were counted, giving the percentage of positive cells (0–100%) [16]. HIF-1 alpha index was calculated according to the modified NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases H-score. Here, intensity of the specific, nuclear immunohistochemical signal was defined as follows: 0 = no appreciable staining in the tumour cell nucleus; 1 = barely detectable staining intensity in the nucleus; 2 = moderate staining intensity distinctly in the tumour cell nucleus; and 3 = strong staining intensity of the tumour cell nucleus. For each intensity group, the fraction of cells (0–100%) was recorded. By multiplying the intensity of the nuclear staining and the fraction of cells producing this particular intensity, the HIF-1 alpha index was calculated producing a total range of 0–300. The MVD was evaluated by counting the number of vascular structures, as defined by immunohistochemical staining for CD34, using a counting graticule with an area of 0.7 mm2 at a 200-fold magnification and within the area of the highest density (‘hot spot’) [17]. Using the immunohistochemical staining for CD34, the angiogenic pattern was semiquantitatively analysed. The angiogenic pattern was scored by two independent observers to belong to the ‘angiogenic’, the ‘silent’ or the ‘balanced’ angiogenic pattern. The ‘angiogenic type’ was defined by the predominance of sprouting vessels with multi-layered endothelium as investigated in the CD34 immunohistochemical staining. In contrast, the ‘silent type’ was defined by the predominance of vessels with thin, mono-layered endothelium. Specimens with an equal distribution between angiogenic and silent type were defined as the ‘balanced type’ [14]. Statistical analysis Kruskal–Wallis and χ2-test were used to assess group differences as appropriate. Spearman’s rank correlation coefficient was used to evaluate monotone associations between two continuous variables. A Spearman’s rank correlation coefficient between 0.7 and 1 (−0.7 to −1) was considered to indicate a strong correlation, whereas a correlation coefficient of 0.3–0.5 (−0.3 to 0.5) was considered to indicate a very low correlation, and a correlation coefficient of 0–0.3 (0 to −0.3) was considered to basically denote no correlation. A two-tailed significance level of 0.05 was applied. Overall survival (OS) was defined from diagnosis of BM to death or date of last follow-up. Estimation of OS was performed using the Kaplan–Meier product limit method. Log rank test was used to investigate group differences. Scale variables were divided in two groups according to the median in order to analyse the survival impact in the univariate analysis using the log rank test. © 2014 British Neuropathological Society e43 Therefore, Ki67 proliferation index, MVD and HIF-1 alpha index were cut at the median in order to define low (below median) and high (above median) and entered as a dummy variable in univariate analysis. Variables with significant univariate results were entered into a multivariable Cox proportional hazards model. For analysis in the multivariate analysis scale, variables were entered as such. Due to the exploratory and hypothesisgenerating design of the present study, no adjustment for multiple testing was applied [18]. All statistical analysis was performed with statistical package for the social sciences (spss) 20.0 software (SPSS Inc., Chicago, IL, USA). Results Patients’ characteristics Six hundred thirty-nine BM specimens of 639 patients (336 male, 303 female) with histologically proven BM were available for this study. Median age at first diagnosis of BM was 57 years (range 25–82). Primary tumour was lung cancer in 344/639 (53.8%) patients, breast cancer in 105/639 (16.4%), melanoma in 67/639 (10.5%), renal cell carcinoma in 52/639 (8.1%) and colorectal cancer in 71/639 (11.1%) patients. Table 1 lists the patients’ characteristics. Matched primary tumour samples were available in 113/639 (17.7%) patients; 63/113 (55.8) NSCLC; 8/113 (7.1%) SCLC; 19/113 (16.8%) breast cancer; 3/113 (2.7%) melanoma; 8/113 (7.1%) renal cell carcinoma; and 12/113 (10.6%) colorectal cancer). Tissue-based findings in BM and corresponding primary tumours Examples of immunostaining results for Ki67, CD34 and HIF-1 alpha are shown in Figure 1. The median Ki67 proliferation index in the entire BM cohort was 39.2% (range 0–97%), the median HIF-1 alpha index was 50 (range 0–300) and the median MVD was 71/0.7 mm2 (range 3–302/0.7 mm2). Predominance of microvascular sprouting (angiogenic type) was evident in 288/619 (46.5%) evaluable specimens, while 169/619 (27.3%) presented without signs of angiogenesis (silent type), and in 162/619 (25.4%), specimens showed an equal distribution between microvascular sprouting and silent angiogenesis (balanced type). The MVD showed a significant association with the vascular pattern, as specimens of the angiogenic type NAN 2015; 41: e41–e55 e44 A. S. Berghoff et al. Table 1. Patients’ characteristics Entire population (n = 639) Characteristic Median age at first diagnosis, years (range) Primary tumour Lung cancer NSCLC SCLC Breast cancer Melanoma Renal cell carcinoma Colorectal cancer Stage IV of primary tumour at BM diagnosis Yes No Unknown Surgery of primary tumour before diagnosis of BM Yes No Unknown Number of extracranial metastatic sites 0 1 ≥2 Unknown Visceral metastases before diagnosis of BM Yes No Unknown Lung metastases before diagnosis of BM yes no Unknown Liver metastases before diagnosis of BM Yes No Unknown BM first metastatic site/no extracranial metastases Yes No Unknown Number of chemotherapy lines before diagnosis of BM 0 1 ≥2 Unknown Time from diagnosis of primary tumour until diagnosis of BM, months (range) Median age at first diagnosis of BM, years (range) DS-GPA class at diagnosis of BM I II III IV unknown SCLC BM score according to Rades et al. Group A Group B Group C © 2014 British Neuropathological Society n % 55 (20–80) 344 291 53 105 67 52 71 53.8 45.5 8.2 16.4 10.5 8.1 11.1 284 350 5 42.1 57.3 0.8 366 269 4 57.3 42.4 0.6 371 151 116 1 58.2 23.6 18.1 0.2 178 443 18 27.7 69.3 2.8 121 500 18 18.9 78.2 2.8 74 547 18 11.6 85.6 2.8 371 267 1 58.1 41.8 0.2 373 183 74 9 23 (1–299) 57 (25–82) 58.4 28.6 11.5 1.4 114 266 139 32 88 17.8 41.6 21.8 5.0 13.8 4 18 31 7.5 34.0 58.5 NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases e45 Table 1. (Continued) Entire population (n = 639) Characteristic Number of BM at diagnosis of BM 1 2–3 >3 Unknown Localization of BM Supratentorial Infratentorial Both Unknown Status of primary tumour at t diagnosis of BM No evidence of disease Partial response Stable disease Progressive disease Synchronous 1st diagnosis of primary tumour and BM Unknown First-line treatment for BM Gamma knife Chemotherapy Surgery Whole brain radiation therapy Best supportive care Chemotherapy after diagnosis of BM Yes No Unknown Alive at last follow-up Yes No Median overall survival from diagnosis of primary tumour, months (range) Median overall survival from diagnosis of BM, months (range) n % 419 158 57 5 65.6 24.7 8.9 0.8 399 141 93 6 62.4 22.1 14.6 1.0 185 22 138 69 220 5 29.0 2.4 21.6 10.8 34.4 0.8 28 3 573 31 4 4.4 0.5 89.7 4.9 0.7 219 394 26 34.3 61.7 4.1 93 546 23 (0–334) 8.0 (0–207) 14.6 85.4 BM, brain metastasis; DS-GPA, diagnosis-specific graded prognostic assessment; NSCLC, non-small cell lung cancer; SCLC, small cell lung cancer. had higher MVD values compared with specimens with silent type (P < 0.001; Kruskal–Wallis test). No correlation was observed between Ki67 proliferation index and MVD (P = 0.154; Spearman correlation coefficient −0.056). Further, no correlation was observed between MVD and HIF-1 alpha index (Spearman correlation coefficient 0.107) or between HIF-1 alpha index and Ki67 proliferation index (Spearman correlation coefficient 0.179). We detected significant differences in MVD and expression of Ki67 and HIF-1 alpha between BM of different origins. Ki67 proliferation index was highest in colorectal cancer BM and lowest in renal cell carcinoma BM (P < 0.001; Kruskal–Wallis test; Figure 2A). MVD was © 2014 British Neuropathological Society highest in renal cell carcinoma BM and lowest in melanoma BM (P < 0.001; Kruskal–Wallis test; Figure 2B). Accordingly, renal cell carcinoma BM had the highest fraction of specimens with angiogenic sprouting and melanoma the highest fraction of specimens of the silent angiogenic subtype (P < 0.001; χ2-test). In line with these data, HIF-1 alpha index was highest in renal cell carcinoma BM and lowest in melanoma BM (P < 0.001; Kruskal–Wallis test; Figure 2C). Comparing MVD and expression of Ki67 and HIF-1 alpha between BM and matched primary tumours, we found significant differences between tumour sites in several tumour types (Figure 3). In brief, Ki67 index was higher in breast cancer BM compared with the matched NAN 2015; 41: e41–e55 e46 A. S. Berghoff et al. Figure 1. Examples of immunostaining results. (A and B) Ki67 proliferation index in a primary NSCLC specimen (A) and the matched BM specimen (B); (C and D) silent angiogenic pattern and low MVD in a primary colorectal cancer specimen (C) and the matched BM specimen with sprouting angiogenic pattern and high MVD (D); (E and F) low HIF-1 alpha index in a primary NSCLC specimen (E) and the matched BM specimen (F). Colour figures are available from the authors upon request. © 2014 British Neuropathological Society NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases Figure 2. Comparison of Ki67 proliferation index (A), microvascular density (MVD, B) and HIF-1 alpha index (C) in BM of different primary tumour types. © 2014 British Neuropathological Society e47 Figure 3. Comparison of Ki67 proliferation index (A), microvascular density (B) and HIF-1 alpha (C) between primary tumours and matched brain metastases for all tumour types. NAN 2015; 41: e41–e55 e48 A. S. Berghoff et al. primary tumours (46.6% vs. 38.9%; P = 0.032; paired t-test), while MVD was higher in central nervous system lesions than in the primary tumours in NSCLC (74/ 0.7 mm2 vs. 65/0.7 mm2; P = 0.050; paired t-test). Also, HIF-1 alpha index was higher in BM than in matched primary tumours in NSCLC (50 vs. 30; P = 0.026; paired t-test), breast cancer (60 vs. 30; P = 0.041; paired t-test) and colorectal cancer (70 vs. 20; P = 0.004; paired t-test). In NSCLC BM, higher Ki67 proliferation index (49.2% vs. 36.4%; P = 0.002; Mann–Whitney test) and higher HIF-1 alpha index (80 vs. 60; P = 0.021; Mann–Whitney test) were observed in patients with squamous compared with patients with non-squamous NSCLC histology. However, higher MVD in NSCLC BM was observed in nonsquamous compared with squamous NSCLC histology (71/0.7 mm2 vs. 63/0.7 mm2; P = 0.038; Mann–Whitney test). Survival analyses Survival analyses in the overall cohort Clinical parameters In the entire cohort of 639 BM patients, median OS from diagnosis of BM was 8 months (range 0–207). The GPA prognostic score showed a statistically significant correlation with OS. Median OS in patients with class I was 19 months, 10 months in class II, 7 months in class III and 2 months in class IV (P < 0.001; log rank test; Figure 4A). In SCLC BM patients, the prognostic score according to Rades et al. was applied and showed significant impact on OS [13]. Median OS in patients of group A was 1 month, 8 months in group B patients and 12 months in patients belonging to group C (P < 0.001; log rank test). Furthermore, the histology of the primary tumour showed a significant prognostic value with the longest survival among patients with breast cancer and renal cell carcinoma, both with a median OS of 11 months, followed by patients with NSCLC (median OS 9 months), SCLC (median OS 10 months), melanoma (median OS 6 months) and colorectal cancer (median OS 5 months; P = 0.001; log rank test; Figure 4B). Long-term survival was associated with primary tumour histology as 9/52 (17.3%) patients with renal cell carcinoma compared with 38/344 (11.0%) patients with lung cancer, 12/105 (11.4%) patients with breast cancer, 1/67 (1.5%) patient with melanoma and 0/71 (0.0%) patients with colorectal cancer survived longer than 36 months (P = 0.001; χ2-test). © 2014 British Neuropathological Society Tissue-based parameters Patients with high Ki67 proliferation index presented with significantly impaired OS prognosis compared with patients with low Ki67 proliferation index (7 vs. 11 months; P < 0.001; log rank test; Figure 4C). In line, long-term survival (>36 months) was more common among patients with low Ki67 proliferation index (41/322, 12.7%) compared with patients with high Ki67 proliferation index (19/317, 6.0%). Furthermore, patients with high MVD had an improved OS prognosis compared with patients with low MVD (11 vs. 8 months; P = 0.004; log rank test; Figure 4D). Long-term survival (>36 months) was more common among patients with high MVD as 40/317 (12.6%) patients with high MVD experienced long-term survival compared with 20/322 (6.2%) patients with low MVD (P = 0.005; χ2-test). However, no impact of angiogenic pattern on OS was observed as patients with angiogenic, sprouting pattern presented with a median OS of 10 months compared with 8 months in patients with silent angiogenesis and 8 months in patient with balanced angiogenesis (P = 0.256; log rank test). No difference in survival prognosis was observed for patients with high HIF-1 alpha index (median OS 8 months) compared with patients with low HIF-1 alpha index (median OS 9 months; P = 0.724; log rank test). Survival analyses in DS-GPA groups Impact of tissue-based parameters was analysed in each DS-GPA group separately. Here, high Ki67 proliferation index correlated with unfavourable median OS in patients with DS-GPA class II (11 vs. 7 months; P = 0.010; log rank test) and in patients with DS-GPA class III (8 vs. 5 months; P = 0.017; log rank test). Further, high MVD correlated with favourable median OS in patients with DS-GPA class I (13 vs. 22 months; P = 0.010; log rank test) and in patients with DS-GPA class III (5 vs. 8 months; P = 0.050; log rank test). High HIF-1 alpha index associated with unfavourable median OS only in patients with DS-GPA class IV (2 vs. 11 months; P = 0.005; log rank test). In multivariate analysis, Ki67 proliferation index [hazard ratio (HR) 1.007; P = 0.001], MVD (HR 0.997; P = 0.004) and DS-GPA class (HR 1.530; P < 0.001) remained significant (Cox regression model). Survival analyses in tumour subtypes We correlated Ki67 index, HIF-1 alpha index and MVD with OS in the sub-cohorts of patients with NSCLC (n = 291), SCLC (n = 53), breast cancer (n = 105), NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases e49 Figure 4. Kaplan–Meier plots for overall survival from diagnosis of BM to death according to diagnosis-specific graded prognostic assessment class (A), histology of the primary tumour (B), the Ki67 proliferation index (C) and microvascular density (D). © 2014 British Neuropathological Society NAN 2015; 41: e41–e55 e50 A. S. Berghoff et al. Table 2. Tissue-based findings in brain metastases Entire cohort (n = 639) OS in < median OS in ≥ median Non-small cell lung cancer (n = 291) OS in < median OS in ≥ median Small cell lung cancer (n = 53) OS in < median OS in ≥ median Breast cancer (n = 105) OS in < median OS in ≥ median Melanoma (n = 67) OS in < median OS in ≥ median Renal cell carcinoma (n = 52) OS in < median OS in ≥ median Colorectal caner (n = 71) OS in < median OS in ≥ median Median ki67 proliferation index (%), (range) Median microvascular density/ 0.7 mm2 (range) Angiogenic pattern Angiogenic (%) Balanced (%) Silent (%) 39.2 11* 7* 38.3 11* 7* 55.40 11 8 39.2 12 11 28.8 8 5 17.0 19* 3* 64.0 9 4 71 8* 11* 70 8 11 72 10 11 75 9 13 50 6 7 133 6 15 65 5 6 (3–302) 288/619 (46.5) 10 162/619 (26.2) 8 169/619 (27.3) 8 (7–298) 126/281 (44.8) 9 70/281 (24.9) 9 85/281 (30.2) 8 (16–187) 26/51 (51.0) 11 13/51 (25.5) 7 12/51 (23.5) 12 (7–240) 57/104 (54.8) 11 31/104 (29.8) 16 16/104 (15.4) 11 (3–197) 16/62 (25.8) 8 17/62 (27.4) 6 29/62 (46.8) 7 (34–302) 37/51 (72.5) 15* 11/51(21.6) 3* 3/51 (5.9) 2* (13–251) 26/70 (37.1) 5 20/70 (28.6) 4 24/70 (34.3) 6 (0–97) (1–97) (17–86) (2–88) (0–82) (2–76) (16–91) Median HIF-1 alpha index; (range) 50 9 8 60 11* 7* 60 11 7 50 11 13 10 6 6 120 8 15 50 7 5 (0–300) (0–270) (0–180) (0–230) (0–160) (0–300) (0–180) *Marks significant (P < 0.05) group differences. HIF-1 alpha, hypoxia-inducible factor 1 alpha; OS, overall survival. melanoma (n = 67), renal cell carcinoma (n = 52) and colorectal cancer (n = 71; Table 2). Significant correlations of these tissue-based parameters with OS from BM diagnosis were found in NSCLC and renal cell cancer patients. NSCLC Patients with low median Ki67 proliferation index (median OS 11 months) and BM from NSCLC presented with an improved prognosis compared with patients with high median Ki67 proliferation index (median OS 7 months; P = 0.001; log rank test; Figure 5A). Furthermore, patients with low HIF-1 alpha index (median OS 11 months) and BM from NSCLC survived longer than patients with high HIF-1 alpha index (median OS 7 months; P = 0.015; log rank test; Figure 5B). No differences in Ki67 proliferation index (P = 0.901; Kruskal–Wallis test), MVD (P = 0.166; Kruskal–Wallis test) or HIF-1 alpha index (P = 0.140; Kruskal–Wallis test) according to DS-GPA class were observed in patients with NSCLC BM. Further, tissue-based parameters were investigated for their prognostic impact in each DS-GPA group. Here, low Ki67 proliferation index correlated with favourable median OS in patients with DS-GPA class II (12 vs. 6 © 2014 British Neuropathological Society months: P = 0.002; log rank test) and in patients with DS-GPA class III (7 vs. 5 months; P = 0.027; log rank test). Further, low HIF-1 alpha index was a favourable prognostic factor in patients presenting with DS-GPA class I (22 vs. 11 months; P = 0.030; log rank test). Thus, Ki67 proliferation index, HIF-1 alpha index and DS-GPA were entered in multivariate analysis. Here, Ki67 proliferation index [HR 1.015; 95% confidence interval (CI) 1.008–1.022; P < 0.001; Cox regression model] and DS-GPA (HR 1.879; 95% CI 1.566–2.254; P < 0.001; Cox regression model) remained as independent prognostic factors. Renal cell carcinoma Low median Ki67 proliferation index correlated with favourable OS in patients with BM from renal cell carcinoma (median OS 19 vs. 3 months; P < 0.001; log rank test; Figure 5C). Furthermore, patients with BM showing angiogenic pattern (median OS 15 months) survived significantly longer compared with patients with silent angiogenic pattern (median OS 2 months) or with balanced angiogenesis (median OS 3 months: P = 0.002; log rank test; Figure 5D). Thus, Ki67 proliferation index, angiogenic pattern and DS-GPA were entered in multivariate analysis. Here, Ki67 proliferation index (HR 1.027; 95% CI 1.007–1.048; P = 0.008, Cox NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases e51 Figure 5. Kaplan–Meier plots for overall survival from diagnosis of BM to death according to Ki67 index (A) and HIF-1 alpha index (B) in NSCLC patients. Kaplan–Meier plots for overall survival from diagnosis of BM to death according to Ki67 index (C) and angiogenic pattern (D) in renal cell carcinoma patients. © 2014 British Neuropathological Society NAN 2015; 41: e41–e55 e52 A. S. Berghoff et al. regression model), angiogenic pattern (HR 1.877; 95% CI 1.088–3.236; P = 0.024; Cox regression model) and DS-GPA (HR 1.956; 96% CI 1.291–3.003; P = 0.002; Cox regression model) remained as independent prognostic parameters. Discussion In this project, we indicate that BM of the most relevant solid tumour types typically show signs of prominent neo-angiogenesis and tumour cell proliferation. However, we document differences in the expression profiles of angiogenesis- and hypoxia-related factors between BM of different cancers and furthermore show that these parameters have distinct prognostic implications between tumour types. In general, our data underscore the notion that BM are a heterogeneous group of tumour manifestations, and more consideration should be given to this heterogeneity in the design of research protocols and ultimately also in the clinical setting [19,20]. Comparing the microvascularization between the different tumour types, we saw the highest MVD and a predominance of angiogenic active vessels in BM of renal cell carcinoma. Renal cell carcinoma BM were also characterized by the highest expression values of HIF-1 alpha, which is a known driver of neo-angiogenesis in cancer [21]. Although HIF-1alpha might be only expressed for a short time upon hypoxic stimuli, it is a widely used surrogate marker for hypoxic conditions in various cancer types [21– 26]. In renal cell carcinoma, HIF-1 alpha expression has been shown to be expressed not only under hypoxic but also under normoxic conditions as a result of oncogenic activation due to the loss of the Von Hippel–Lindau gene [25]. We did not investigate in our study the mechanistic link between HIF-1 alpha expression and hypoxia or (epi-)genetic aberrations in BM, and this issue should be subject to further studies. However, our findings may be of clinical relevance for patients with BM of renal cell cancer as they suggest that angiogenesis-targeting agents that are routinely used for extracranial disease (e.g. VEGF pathway inhibitors such as sunitinib, pazopanib, bevacizumab) may also be of therapeutic value for patients with brain metastatic disease [27–29]. Moreover, we demonstrate a strong and independent prognostic value of angiogenic patterns and also of the Ki67 tumour cell proliferation index on patients’ OS from BM diagnosis. In line with our data, both Ki67 index and microvascularization were repeatedly described as prognostic factor in extracranial manifesta© 2014 British Neuropathological Society tions of renal cell carcinoma [30–34]. Pending validation of our results in independent data sets, incorporation of these tissue-based parameters into prognostic scores (e.g. the DS-GPA) may be considered for patients with BM of renal cell carcinoma in order to provide more accurate approximation of survival times in the cohort of patients treated by neurosurgical resection. Of note, current prognostic scores have been shown to predict BM patient survival with inadequate accuracy [35]. In NSCLC, we found significantly higher MVD and HIF-1 alpha indices in BM than in corresponding primary tumours and an independent prognostic impact of the Ki67 index on OS from BM diagnosis. These findings corroborate similar results that were elaborated by our group in a smaller patient cohort [14,36]. Hopefully, future clinical trials on NSCLC patients undergoing resection of BM will incorporate the Ki67 index into the study protocol to validate its prognostic value in a prospective manner. Inhibition of neo-angiogenesis has been identified as promising strategy for prevention and treatment of NSCLC CNS manifestations in preclinical studies [9,19,37,38]. Besides, some other immunohistochemical markers including the transcription factors LEF1/TCF4, netrin-1, insulin-like growth factor 1 or integrins were postulate to correlate with survival prognosis, further underscoring that tissuebased parameter do add valuable information to the prognostic assessment of BM patients [7,36,39,40]. Indeed, our findings of higher MVD and HIF1 alpha expression in BM provide further support for clinical studies on such approaches. We did not find prognostic value of tissue-based parameters assessed in this study in patients with BM of breast cancer, melanoma and colorectal cancer. It must be acknowledged that our data do not definitely exclude prognostic implications of these parameters in BM of these tumour types due to the limitations associated with the retrospective nature of our study and sample size issues limiting the statistical power. However, our cohorts in these diagnostic categories are among the largest and best characterized that are available in the literature (n = 105 for breast cancer, n = 71 for colorectal cancer, n = 67 for melanoma), and moderate to strong prognostic effects should have been detectable. Nevertheless, further studies, ideally in larger and prospectively collected patient series, would certainly be of interest. Interestingly, BM of melanoma were characterized by the lowest MVD and HIF-1 alpha values in comparison with the other tumour types. This finding may be related to prior observations that brain NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases metastatic melanoma BM invade the brain mainly through vascular co-option without significant induction of neoangiogenesis and again reinforce the need for differentiated investigation of pathobiological mechanisms in BM of various cancer types [9,41]. In conclusion, our data argue for differential pathobiological and clinical relevance of Ki67 index, HIF1alpha index and MVD between primary tumour types in BM patients. Specifically, we found independent prognostic value of these tissue-based parameters in patients with BM of NSCLC and renal cell carcinoma, thus making them candidate biomarkers in these indications. Acknowledgements We thank Irene Leisser, Elisabeth Dirnberger and Bettina Jesch for excellent technical assistance. This study was performed within the PhD thesis project of Anna Sophie Berghoff in the PhD program ‘Clinical Neuroscience (CLINS)’ at the Medical University Vienna. The costs for this project were covered by the research budget of the Medical University of Vienna and the grant ‘Hochschuljubiläumsstiftung der Stadt Wien’ for the project ‘Klinische Relevanz von Entzündung in Hirnmetastasen’ (grant number: H-263727/2013). e53 Karin Dieckmann: data collection, data interpretation, manuscript writing, approval of final manuscript version. Josef Pichler: data collection, data interpretation, manuscript writing, approval of final manuscript version. Markus Hutterer: data collection, data interpretation, manuscript writing, approval of final manuscript version. Thomas Melchardt: data collection, data interpretation, manuscript writing, approval of final manuscript version. Rupert Bartsch: study design, data collection, data interpretation, manuscript writing, approval of final manuscript version. Christoph C. Zielinski: study design, data collection, data interpretation, manuscript writing, approval of final manuscript version. Peter Birner: study design, data collection, data interpretation, manuscript writing, approval of final manuscript version. Matthias Preusser: study design, data collection, data interpretation, manuscript writing, approval of final manuscript version. Conflicts of interest The authors declare no conflicts of interest. Author contribution Anna S. Berghoff: study design, data collection, data interpretation, manuscript writing, approval of final manuscript version. Aysegül Ilhan-Mutlu: data collection, data interpretation, manuscript writing, approval of final manuscript version. Carina Dinhof: data collection, data interpretation, manuscript writing, approval of final manuscript version. Manuel Magerle: data collection, data interpretation, manuscript writing, approval of final manuscript version. Monika Hackl: data collection, data interpretation, manuscript writing, approval of final manuscript version. Georg Widhalm: data collection, data interpretation, manuscript writing, approval of final manuscript version. Johannes A. Hainfellner: data collection, data interpretation, manuscript writing, approval of final manuscript version. © 2014 British Neuropathological Society References 1 Jenkinson MD, Haylock B, Shenoy A, Husband D, Javadpour M. Management of cerebral metastasis: evidence-based approach for surgery, stereotactic radiosurgery and radiotherapy. Eur J Cancer 2011; 47: 649–55 2 Kalkanis SN, Kondziolka D, Gaspar LE, Burri SH, Asher AL, Cobbs CS, Ammirati M, Robinson PD, Andrews DW, Loeffler JS, McDermott M, Mehta MP, Mikkelsen T, Olson JJ, Paleologos NA, Patchell RA, Ryken TC, Linskey ME. The role of surgical resection in the management of newly diagnosed brain metastases: a systematic review and evidence-based clinical practice guideline. J Neurooncol 2010; 96: 33–43 3 Kienast Y, Winkler F. Therapy and prophylaxis of brain metastases. Expert Rev Anticancer Ther 2010; 10: 1763–77 4 Scoccianti S, Ricardi U. Treatment of brain metastases: review of phase III randomized controlled trials. Radiother Oncol 2012; 102: 168–79 5 Soffietti R, Cornu P, Delattre JY, Grant R, Graus F, Grisold W, Heimans J, Hildebrand J, Hoskin P, Kalljo M, NAN 2015; 41: e41–e55 e54 6 7 8 9 10 11 12 13 14 15 16 A. S. Berghoff et al. Krauseneck P, Marosi C, Siegal T, Vecht C. EFNS Guidelines on diagnosis and treatment of brain metastases: report of an EFNS Task Force. Eur J Neurol 2006; 13: 674–81 Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011; 144: 646–74 Harter PN, Zinke J, Scholz A, Tichy J, Zachskorn C, Kvasnicka HM, Goeppert B, Delloye-Bourgeois C, Hattingen E, Senft C, Steinbach JP, Plate KH, Mehlen P, Schulte D, Mittelbronn M. Netrin-1 expression is an independent prognostic factor for poor patient survival in brain metastases. PLoS ONE 2014; 9: e92311 Carbonell WS, Ansorge O, Sibson N, Muschel R. The vascular basement membrane as ‘soil’ in brain metastasis. PLoS ONE 2009; 4: e5857 Kienast Y, von Baumgarten L, Fuhrmann M, Klinkert WE, Goldbrunner R, Herms J, Winkler F. Real-time imaging reveals the single steps of brain metastasis formation. Nat Med 2010; 16: 116–22 Wohrer A, Waldhor T, Heinzl H, Hackl M, Feichtinger J, Gruber-Mosenbacher U, Kiefer A, Maier H, Motz R, Reiner-Concin A, Richling B, Idriceanu C, Scarpatetti M, Sedivy R, Bankl HC, Stiglbauer W, Preusser M, Rossler K, Hainfellner JA. The Austrian Brain Tumour Registry: a cooperative way to establish a population-based brain tumour registry. J Neurooncol 2009; 95: 401– 11 Woehrer A. Brain tumor epidemiology in Austria and the Austrian Brain Tumor Registry. Clin Neuropathol 2013; 32: 269–85 Sperduto PW, Berkey B, Gaspar LE, Mehta M, Curran W. A new prognostic index and comparison to three other indices for patients with brain metastases: an analysis of 1,960 patients in the RTOG database. Int J Radiat Oncol Biol Phys 2008; 70: 510–14 Rades D, Dziggel L, Segedin B, Oblak I, Nagy V, Marita A, Schild SE. The first survival score for patients with brain metastases from small cell lung cancer (SCLC). Clin Neurol Neurosurg 2013; 115: 2029–32 Berghoff AS, Ilhan-Mutlu A, Wohrer A, Hackl M, Widhalm G, Hainfellner JA, Dieckmann K, Melchardt T, Dome B, Heinzl H, Birner P, Preusser M. Prognostic significance of Ki67 proliferation index, HIF1 alpha index and microvascular density in patients with non-small cell lung cancer brain metastases. Strahlenther Onkol 2014; 190: 676–85 Birner P, Piribauer M, Fischer I, Gatterbauer B, Marosi C, Ambros PF, Ambros IM, Bredel M, Oberhuber G, Rossler K, Budka H, Harris AL, Hainfellner JA. Vascular patterns in glioblastoma influence clinical outcome and associate with variable expression of angiogenic proteins: evidence for distinct angiogenic subtypes. Brain Pathol 2003; 13: 133–43 Preusser M, Heinzl H, Gelpi E, Hoftberger R, Fischer I, Pipp I, Milenkovic I, Wohrer A, Popovici F, Wolfsberger S, Hainfellner JA. Ki67 index in intracranial ependymoma: © 2014 British Neuropathological Society 17 18 19 20 21 22 23 24 25 26 27 28 a promising histopathological candidate biomarker. Histopathology 2008; 53: 39–47 Weidner N. Current pathologic methods for measuring intratumoral microvessel density within breast carcinoma and other solid tumors. Breast Cancer Res Treat 1995; 36: 169–80 Bender R, Lange S. Adjusting for multiple testing–when and how? J Clin Epidemiol 2001; 54: 343–9 Preusser M, Winkler F, Collette L, Haller S, Marreaud S, Soffietti R, Klein M, Reijneveld JC, Tonn JC, Baumert BG, Mulvenna P, Schadendorf D, Duchnowska R, Berghoff AS, Lin N, Cameron DA, Belkacemi Y, Jassem J, Weber DC. Trial design on prophylaxis and treatment of brain metastases: lessons learned from the EORTC Brain Metastases Strategic Meeting 2012. Eur J Cancer 2012; 48: 3439–47 Kondziolka D, Kalkanis SN, Mehta MP, Ahluwalia M, Loeffler JS. It is time to reevaluate the management of patients with brain metastases. Neurosurgery 2014; 75: 1–9 Ruan K, Song G, Ouyang G. Role of hypoxia in the hallmarks of human cancer. J Cell Biochem 2009; 107: 1053–62 Bachtiary B, Schindl M, Potter R, Dreier B, Knocke TH, Hainfellner JA, Horvat R, Birner P. Overexpression of hypoxia-inducible factor 1alpha indicates diminished response to radiotherapy and unfavorable prognosis in patients receiving radical radiotherapy for cervical cancer. Clin Cancer Res 2003; 9: 2234–40 Kim SJ, Rabbani ZN, Dewhirst MW, Vujaskovic Z, Vollmer RT, Schreiber EG, Oosterwijk E, Kelley MJ. Expression of HIF-1alpha, CA IX, VEGF, and MMP-9 in surgically resected non-small cell lung cancer. Lung Cancer 2005; 49: 325–35 Park S, Ha SY, Cho HY, Chung DH, Kim NR, Hong J, Cho EK. Prognostic implications of hypoxia-inducible factor1alpha in epidermal growth factor receptor-negative non-small cell lung cancer. Lung Cancer 2011; 72: 100–7 Shen C, Kaelin WG Jr. The VHL/HIF axis in clear cell renal carcinoma. Semin Cancer Biol 2013; 23: 18–25 Zhong H, De Marzo AM, Laughner E, Lim M, Hilton DA, Zagzag D, Buechler P, Isaacs WB, Semenza GL, Simons JW. Overexpression of hypoxia-inducible factor 1alpha in common human cancers and their metastases. Cancer Res 1999; 59: 5830–5 Besse B, Lasserre SF, Compton P, Huang J, Augustus S, Rohr UP. Bevacizumab safety in patients with central nervous system metastases. Clin Cancer Res 2010; 16: 269–78 Cochran DC, Chan MD, Aklilu M, Lovato JF, Alphonse NK, Bourland JD, Urbanic JJ, McMullen KP, Shaw EG, Tatter SB, Ellis TL. The effect of targeted agents on outcomes in patients with brain metastases from renal cell carcinoma treated with Gamma Knife surgery. J Neurosurg 2012; 116: 978–83 NAN 2015; 41: e41–e55 Differential role of angiogenesis and tumor cell proliferation in brain metastases 29 Massard C, Zonierek J, Gross-Goupil M, Fizazi K, Szczylik C, Escudier B. Incidence of brain metastases in renal cell carcinoma treated with sorafenib. Ann Oncol 2010; 21: 1027–31 30 Gayed BA, Youssef RF, Bagrodia A, Darwish OM, Kapur P, Sagalowsky A, Lotan Y, Margulis V. Ki67 is an independent predictor of oncological outcomes in patients with localized clear-cell renal cell carcinoma. BJU Int 2014; 113: 668–73 31 Toma MI, Weber T, Meinhardt M, Zastrow S, Grimm MO, Fussel S, Wirth MP, Baretton GB. Expression of the Forkhead transcription factor FOXP1 is associated with tumor grade and Ki67 expression in clear cell renal cell carcinoma. Cancer Invest 2011; 29: 123–9 32 Dudderidge TJ, Stoeber K, Loddo M, Atkinson G, Fanshawe T, Griffiths DF, Williams GH. Mcm2, Geminin, and KI67 define proliferative state and are prognostic markers in renal cell carcinoma. Clin Cancer Res 2005; 11: 2510–17 33 Iakovlev VV, Gabril M, Dubinski W, Scorilas A, Youssef YM, Faragalla H, Kovacs K, Rotondo F, Metias S, Arsanious A, Plotkin A, Girgis AH, Streutker CJ, Yousef GM. Microvascular density as an independent predictor of clinical outcome in renal cell carcinoma: an automated image analysis study. Lab Invest 2012; 92: 46–56 34 Saroufim A, Messai Y, Hasmim M, Rioux N, Iacovelli R, Verhoest G, Bensalah K, Patard JJ, Albiges L, Azzarone B, Escudier B, Chouaib S. Tumoral CD105 is a novel independent prognostic marker for prognosis in clear-cell renal cell carcinoma. Br J Cancer 2014; 110: 1778– 84 35 Kondziolka D, Parry PV, Lunsford LD, Kano H, Flickinger JC, Rakfal S, Arai Y, Loeffler JS, Rush S, Knisely JP, Sheehan J, Friedman W, Tarhini AA, Francis L, Lieberman F, Ahluwalia MS, Linskey ME, McDermott M, Sperduto P, © 2014 British Neuropathological Society 36 37 38 39 40 41 e55 Stupp R. The accuracy of predicting survival in individual patients with cancer. J Neurosurg 2014; 120: 24–30 Bleckmann A, Siam L, Klemm F, Rietkotter E, Wegner C, Kramer F, Beissbarth T, Binder C, Stadelmann C, Pukrop T. Nuclear LEF1/TCF4 correlate with poor prognosis but not with nuclear beta-catenin in cerebral metastasis of lung adenocarcinomas. Clin Exp Metastasis 2013; 30: 471–82 Winkler F, Reck M, Miles D, Mariani P, Lutiger B, Nendel V, Srock S, Perez-Moreno P, Wick W. Incidence of brain metastases in patients treated with bevacizumab. p. ESMO 2013, Amsterdam, Poster #3349, 2013 Berghoff AS, Preusser M. Biology in prevention and treatment of brain metastases. Expert Rev Anticancer Ther 2013; 13: 1339–48 Wu PF, Huang WC, Yang JC, Lu YS, Shih JY, Wu SG, Lin CH, Cheng AL. Phosphorylated insulin-like growth factor-1 receptor (pIGF1R) is a poor prognostic factor in brain metastases from lung adenocarcinomas. J Neurooncol 2013; 115: 61–70 Berghoff AS, Kovanda AK, Melchardt T, Bartsch R, Hainfellner JA, Sipos B, Schittenhelm J, Zielinski CC, Widhalm G, Dieckmann K, Weller M, Goodman SL, Birner P, Preusser M. Alphavbeta3, alphavbeta5 and alphavbeta6 integrins in brain metastases of lung cancer. Clin Exp Metastasis 2014 (in press) Berghoff AS, Rajky O, Winkler F, Bartsch R, Furtner J, Hainfellner JA, Goodman SL, Weller M, Schittenhelm J, Preusser M. Invasion patterns in brain metastases of solid cancers. Neuro-Oncol 2013; 15: 1664–72 Received 11 July 2014 Accepted after revision 22 September 2014 Published online Article Accepted on 24 September 2014 NAN 2015; 41: e41–e55