APMIS 123: 401–409
© 2015 APMIS. Published by John Wiley & Sons Ltd. DOI 10.1111/apm.12368
Tetranectin positive expression in tumour tissue leads to longer survival in Danish women with ovarian cancer. Results from the ‘Malova’ ovarian cancer study MEL C. HEERAN,1,2 LENE RASK,1,2 CLAUS K. HØGDALL,3 SUSANNE K. KJAER,2,3 LISE CHRISTENSEN,4 ALLAN JENSEN,2 JAN BLAAKAER,5 IB JARLE CHRISTENSEN6 and ESTRID V.S. HØGDALL1,3 Department of Pathology, Herlev Hospital, University of Copenhagen, Copenhagen; 2Virus, Lifestyle and Genes, Danish Cancer Society Research Center, Copenhagen; 3The Gynaecologic Clinic, The Juliane Marie Centre, Rigshospitalet, University of Copenhagen, Copenhagen; 4Department of Pathology, Rigshospitalet, University of Copenhagen, Copenhagen; 5Department of Gynecology & Obstetrics, Skejby, Aarhus University Hospital, Aarhus; and 6The Finsen Laboratory, Rigshospitalet and Biotech Research and Innovation Centre (BRIC), University of Copenhagen, Copenhagen, Denmark 1
Heeran MC, Rask L, Høgdall CK, Kjaer SK, Christensen L, Jensen A, Blaakaer J, Jarle Christensen IB, Høgdall EVS. Tetranectin positive expression in tumour tissue leads to longer survival in Danish women with ovarian cancer. Results from the ‘Malova’ ovarian cancer study. APMIS 2015; 123: 401–409. The primary objective of this study was to analyse Tetranectin (TN) expression in tumour tissues and TN serum concentration in 758 women with epithelial ovarian tumours. The second was to evaluate, whether TN tissue expression levels correlate with clinico-pathological parameters and prognosis of the disease. Using tissue arrays we analysed the expression levels in tissues from 166 women with borderline ovarian tumours (BOTs) and 592 women with ovarian cancer (OC). A panel of three antibodies was used for immunohistochemistry: a polyclonal and two monoclonal antibodies. Serum TN was measured using the polyclonal antibody A-371. Univariate survival analyses stratified for chemotherapy showed that positive tissue TN as demonstrated by the polyclonal antibody indicated a significantly longer overall survival (OS) (p = 0.0001) as well as cancer specific survival (CSS) (p < 0.0001). High serum TN was likewise found to imply longer OS (p < 0.0001) and CSS (p < 0.0001), whereas tissue staining with the two monoclonal antibodies failed to demonstrate any significant correlation with either survival type. Univariate Kaplan–Meier survival analysis performed on all OC cases showed a significantly longer OS (p = 0.0009) and CSS (p = 0.0006) for women with TN positive tumour tissue and in women with high serum TN levels (p < 0.0001 for both). However, in the multivariate Cox regression analysis, only serum TN was found to be an independent prognostic factor for OS (p = 0.01) and not for CSS (p = 0.08). In conclusion, our results predict that a positive TN expression of both tumour tissue and serum points to a more favourable outcome for OC patients. Key words: Tissue array; ovarian cancer; tetranectin; prognosis; immunohistochemistry. Estrid Høgdall, Clinical Senior Investigator, Department of Pathology, Herlev Hospital, Herlev Ringvej 75, DK-2730 Herlev, Copenhagen, Denmark. e-mail: [email protected]
Epithelial ovarian cancer (OC) is the most common gynaecologic malignancy in the Western world, with some of the highest rates occurring in developed regions including Scandinavia (1). According to the Danish Gynecological Cancer Database (DGCD), Denmark currently has one of the highest age-standardized incidence rates of OC worldwide (11/100 000), with approximately 500 new cases Received 16 July 2014. Accepted 27 December 2014
annually (2–4). Although numbers have remained relatively constant over a generation (5), 2% of Danish women will develop OC before the age of 75 years (6, 7). Due to the paucity of early symptoms, the majority of women present with advanced-stage disease upon initial diagnosis, and this late-stage overall survival (OS) is approximately 25–30%, in contrast to women with early-stage disease, who have a survival rate of approximately 90% (8, 9). Current 401
HEERAN et al.
methods for detection of OC involve ultrasonography and a monitoring of the level of the serum glycoprotein CA125. However, the use of CA125 as a biomarker for OC is not optimal. It is non-specific for OC and thus, unreliable for early diagnosis (10). Prevalent therapeutic approaches to OC treatment comprise maximal surgical tumour debulking followed by combination chemotherapy with carboplatin and paclitaxel (11). Although standard chemotherapy may prolong survival, relapse often occurs resulting in chemoresistance. Therefore, due to paucity of early clinical symptoms and with the few diagnostic markers available for early detection, it is of paramount importance that OC is detected at an earlier stage so that the mechanisms of OC development is better understood so more effective and targeted treatments can be developed. Tetranectin (TN) belongs to the family of Ctype lectins that bind specifically to kringle four of plasminogen and is thought to have a role in the regulation of proteolytic and fibrinolysis processes (12, 13). Although the precise biological function of TN has not yet been clarified, it shows colocalization with plasminogen at the invasion front of melanomas, suggesting a role in cancer invasion and metastases (14). Several studies of a variety of human carcinomas including breast, colorectal, ovarian and oral tumours have reported correlation between low serum TN and a reduction in patient survival (15–18). Low serum level in combination with a high extracellular tumour tissue expression for TN has also been observed in OCs, thus highlighting TN as a potential immunohistochemistry (IHC) marker in the discrimination between malignant and benign ovarian tumours (19–23). Several antibodies have been used to evaluate the expression of TN in tumour tissue. Polyclonal antisera may contain antibody populations that react with antigens other than TN, but the commercially available polyclonal rabbit anti-human TN antibody A-371 (Dako, Glostrup, Denmark), which is known to recognize amino acid 17–181 of TN antigen, is well characterized with high specificity with respect to IHC (24). Monoclonal antibodies (MAbs) are known for producing a more specific immunostaining, and we therefore included MAb (Hyb130-11), which gives an intense staining intensity and MAb (Hyb130-12), which gives a significantly weaker staining intensity. Both of these antibodies react with non-conformational epitopes located with amino acid residues 50–181 (25). In this study, we tested the IHC performance characteristics of these anti-TN antibodies on TMA using a sizeable and well-defined Danish population-based series of OC and borderline ovarian
402
tumours (BOT). The result was correlated with selected clinicopathologic parameters and prognosis. MATERIALS AND METHODS Study population The MALOVA study (MALignant OVArian cancer study) is a multidisciplinary Danish study covering epidemiology (lifestyle factors), biochemistry and molecular biology with the purpose of identifying risk factors for OC. The design of the MALOVA study has been described elsewhere (26–29). Briefly, pre-operative blood samples and tumour tissue samples were obtained from most of the patients with a primary epithelial ovarian tumour. International Federation of Gynecology and Obstetrics (FIGO) stages were obtained from clinical records, which were reviewed by two gynaecologists, both specialized in OC. Furthermore, in the clinical records, women were either classified as radically operated with no macroscopic residual tumour or non-radically operated with macroscopic residual tumour. Histological grading and typing of the retrospectively collected paraffin-embedded tumour tissue samples were performed individually by two persons (E.H. and L.C.). The MALOVA study includes 681 women diagnosed with OC and 235 diagnosed with BOT. The study has been approved by the scientific ethical committee in the study area (KF01-384/95).
Follow-up In Denmark, all inhabitants have a personal identification number, which comprises the information regarding gender and birth date registered online in the Danish Civil Registration System. All OC women of this study were traced in the register, and date of death or emigration up to December 31st 2011, were registered. Women who died during follow-up, were also linked to The National Patient Registry and information on any hospital admissions to oncologic or gynaecologic departments was obtained. Relevant hospital files were collected, and information on the treatment (surgery and chemotherapy) and cause of death was established in October 2004. If there was uncertainty about the cause of death according to the patient file, the women were linked to the Danish Causes of Death Register. At the end of follow-up (31st December, 2011), a total of 520 OC women with OC had died (median follow-up time: 28 months, range: 1–176), and 161 OC women were still alive (median follow-up time: 176 months, range: 151–206).
Tissue array analysis Paraffin-embedded tissue from primary surgery was used for TA analyses. A diagnostic representative tissue block was used for the TA production, and the presence of epithelial tumour cells in all cores was assured by H&E staining prior to immunostaining. A total of 108 samples had to be excluded due to poor tissue quality, lack of tumour tissue in the original paraffin blocks, lack of tumour tissue in the selected cylinders, loss of tissue during the staining procedure or folding/tearing of sections during processing. A final 592 OC and 166 BOT cases were eligible for TN
© 2015 APMIS. Published by John Wiley & Sons Ltd
TETRANECTIN EXPRESSION IN OVARIAN CANCER
tissue expression and survival analyses. Kidney and liver tissue (four cores) were placed on each TA, not only as staining control but also to ensure orientation of the TA. The control tissues were stained with equal intensity on each run, and as the same staining results were obtained in two neighbouring slides from the same TA, which were used in both runs, we found the method reliable. The high performance may be a consequence of the fact that we used relatively large cores of 2 mm each.
Immunohistochemical analysis
Tissue preparation and immunostaining Two-micrometre sections (Section Transfer System, STS, Ergostar HM200; MICROM International GmbH, Walldorf, Germany) from the TA blocks were transferred to glass slides (Dako Chem-Mate Capillary Cap Microscope Slides, 57 mm; Dako A ⁄ S). Slides were stored at 4 °C for a maximum of 8 days until staining for TN. Prior to staining, the sections were deparaffinized in xylene and rehydrated in graded dilutions of alcohol. To expose antigens within the tissue, sections were pretreated in protease buffer (Protease type XXIV, P8038, 1:20,000, Sigma, St. Louis, MO, USA) at 37 °C for 5 min. The staining procedure for TN overexpression was performed on a Dako TechmateTM 500PLUS instrument using the polyclonal rabbit anti-human TN antibody (Code no. A-371; Dako), and two monoclonal mouse, Anti-human TN antigens (HYB 130-11 and 130.12; BioPorto, Hellerup, Denmark), diluted 1:100 followed by incubation for 1 h at room temperature. Other conditions were performed according to the manufacturer’s instructions.
TN scoring of tissue expression Two observers experienced in evaluating IHC stained tissue, simultaneously assessed the pattern of TN staining for each tumour sample. Standardization of scoring was achieved by comparison of scores, and any discrepancies were resolved by consensus. Scoring for TN expression was based on the immune-positivity of staining of a given tumour specimen, which exhibited a specific staining of tumour cells and a specific staining of a subgroup of stromal cells. As no international consensus on cut-off value for TN positivity has been published, a two-tiered scale, negative and positive was used for the description of TN expression in BOT and OC. In survival analyses, we used the same two-tiered scale, for the examination of overall survival (OS) and cancer specific survival (CSS).
Statistical analyses Statistical comparisons between groups were carried out using chi-square tests for categorical data or rank sum tests for ordinal variables. Survival rates were estimated using the Kaplan–Meier method. The impact of TN expression on OC mortality was estimated on the basis of the Cox proportional hazards model (30), applying the TN two-tiered scale using time since primary surgery as the time axis. The assumption of proportionality for the Cox model was assessed using Schoenfeld and martingale residuals as well as graphical methods. The assumption of proportional hazards between women not
© 2015 APMIS. Published by John Wiley & Sons Ltd
receiving chemotherapy, those treated with platin-based therapy and those treated with other anti-cancer drugs was not fulfilled. Therefore, a stratified proportional hazards model was used for multivariate analysis, with strata defined by the therapy group. Model assessment demonstrated that the proportional hazards assumption as well as linearity where applicable were fulfilled for other covariates included in this study. The regression analyses were adjusted for FIGO stage (I, II, III and IV), residual tumour after primary surgery, histological type of tumour (serous, mucinous, endometrioid and other histological types), age at diagnosis (linear) and histological grade (high, moderate and low). Confidence intervals (95% CI) for the corresponding parameters in the multivariate COX regression model were included. Overall survival and cancer specific survival were defined as length of time from primary surgery until death from any cause or death from OC, respectively. Results were considered statistically significant at the p < 0.05 level. The three antibodies, serum and age were analysed by the Spearman rank correlation, comparing rank order number rather than actual values, of two sets of variables to calculate a correlation coefficient. If a monotonic relationship existed between the two variables, one variable was invariably associated with an increase or decrease of the other (31). The analyses were performed, using the using the SAS, v9.2 (SAS Institute, Cary, NC, USA).
Table 1. Clinical characteristics and TN expression in tumour tissues from women diagnosed with borderline ovarian tumour (BOT) and ovarian cancer (OC) Characteristics Ovarian cancer Borderline (n = 592) ovarian tumour (n = 166) Total 592 166 FIGO stage I 164 (28%) 142 (85%) II 64 (11%) 5 (3%) III 310 (52%) 19 (11%) IV 54 (9%) 0 (0%) Tumour cell type Adenocarcinoma, 32 (5%) NOS Serous 365 (62%) adenocarcinoma Mucinous 54 (9%) adenocarcinoma Endometrioid 83 (14%) adenocarcinoma Clear cell carcinoma 47 (8%) Undifferentiated 11 (2%) Treatment No chemotherapy 146 (25%) Chemotherapy 48 (8%) without platinum Chemotherapy 398 (67%) with platinum Age median (range) 59 (35–79) 55 (35–79) FIGO, International Federation of Gynecology and Obstetrics, NOS, not otherwise specified.
403
HEERAN et al.
130-11
130-12
A371
A
B
C
D
E
F
G
H
I
J
K
L
Fig. 1. Tetranectin-stained tissues from Ovarian Cancer (OC) patients (objective x20) showing (A) serous adenocarcinoma OC stained with 130-11 antibody, (B) serous adenocarcinoma OC stained with 130-12 antibody, (C) serous adenocarcinoma OC stained with A-371 antibody, (D) clear cell carcinoma OC stained with 130-11 antibody, (E) clear cell carcinoma OC stained with 130-12 antibody, (F) clear cell carcinoma OC stained with A-371 antibody, (G) mucinous adenocarcinoma OC stained with 130-11 antibody, (H) mucinous adenocarcinoma OC stained with 130-12 antibody, (I) mucinous adenocarcinoma OC stained with A-371 antibody, (J) endometrioid adenocarcinoma OC stained with 130-11 antibody, (K) endometrioid adenocarcinoma OC stained with 130-12 antibody and (L) endometrioid adenocarcinoma OC stained with A-371 antibody. All three images (J, K and L) were from the exact same position on the TMA core.
RESULTS Clinicopathological characteristics of OC and BOT
Results from the analysis of TN by TA of the 758 tumours originating from our population404
based collection are shown in Table 1. The age of OC women ranged 35–79 years, with an average of 59 years and the age of BOT patients ranged 35–79 years, with an average of 55 years old. Of all 592 OC tissue samples 310 (52%) were © 2015 APMIS. Published by John Wiley & Sons Ltd
TETRANECTIN EXPRESSION IN OVARIAN CANCER
found within FIGO group III, and of all 166 BOT tissue samples 142 (85%) were found in FIGO group I. Most of the OC tumours were serous adenocarcinoma 62% (365), and a majority of women diagnosed with OC received chemotherapy with platinum 67% (398). Details of patient characteristics and descriptive statistics for the tumours are shown in Table 1. Representative photomicrographs of tumour tissue showing positive staining for the specific antigen are presented in Fig. 1. Correlations between the TN markers and to age
To investigate whether the prognostic value of TN expression in ovarian tumours was dependent on the antibody used for IHC detection, a comparison was made between the three different antibodies recognizing TN within tumour tissue on one side and a comparison between these and serum TN on the other. This was done to evaluate comparability as well as prognostic value of these markers. The Spearman’s rank correlation was used to examine the relationship between the four TN identification measures. Weak associations were observed among them all with absolute values of p = 0.3. Correlations with age were weak as well (p = 0.1) (Table 2). Consecutive tissue sections had been used in order to demonstrate TN staining in exactly the same tissue position of the sample (Fig. 1).
Table 2. Relationship between TN markers and Spearman rank correlation Covariate TN poly TN_30-11 TN_30-12 TN_Serum 0.13 0.007 0.003 0.004 0.87 0.96 486 489 400 TN_Poly 0.02 0.02 0.68 0.71 731 606 TN_30-11 0.26
© 2015 APMIS. Published by John Wiley & Sons Ltd. DOI 10.1111/apm.12368
Tetranectin positive expression in tumour tissue leads to longer survival in Danish women with ovarian cancer. Results from the ‘Malova’ ovarian cancer study MEL C. HEERAN,1,2 LENE RASK,1,2 CLAUS K. HØGDALL,3 SUSANNE K. KJAER,2,3 LISE CHRISTENSEN,4 ALLAN JENSEN,2 JAN BLAAKAER,5 IB JARLE CHRISTENSEN6 and ESTRID V.S. HØGDALL1,3 Department of Pathology, Herlev Hospital, University of Copenhagen, Copenhagen; 2Virus, Lifestyle and Genes, Danish Cancer Society Research Center, Copenhagen; 3The Gynaecologic Clinic, The Juliane Marie Centre, Rigshospitalet, University of Copenhagen, Copenhagen; 4Department of Pathology, Rigshospitalet, University of Copenhagen, Copenhagen; 5Department of Gynecology & Obstetrics, Skejby, Aarhus University Hospital, Aarhus; and 6The Finsen Laboratory, Rigshospitalet and Biotech Research and Innovation Centre (BRIC), University of Copenhagen, Copenhagen, Denmark 1
Heeran MC, Rask L, Høgdall CK, Kjaer SK, Christensen L, Jensen A, Blaakaer J, Jarle Christensen IB, Høgdall EVS. Tetranectin positive expression in tumour tissue leads to longer survival in Danish women with ovarian cancer. Results from the ‘Malova’ ovarian cancer study. APMIS 2015; 123: 401–409. The primary objective of this study was to analyse Tetranectin (TN) expression in tumour tissues and TN serum concentration in 758 women with epithelial ovarian tumours. The second was to evaluate, whether TN tissue expression levels correlate with clinico-pathological parameters and prognosis of the disease. Using tissue arrays we analysed the expression levels in tissues from 166 women with borderline ovarian tumours (BOTs) and 592 women with ovarian cancer (OC). A panel of three antibodies was used for immunohistochemistry: a polyclonal and two monoclonal antibodies. Serum TN was measured using the polyclonal antibody A-371. Univariate survival analyses stratified for chemotherapy showed that positive tissue TN as demonstrated by the polyclonal antibody indicated a significantly longer overall survival (OS) (p = 0.0001) as well as cancer specific survival (CSS) (p < 0.0001). High serum TN was likewise found to imply longer OS (p < 0.0001) and CSS (p < 0.0001), whereas tissue staining with the two monoclonal antibodies failed to demonstrate any significant correlation with either survival type. Univariate Kaplan–Meier survival analysis performed on all OC cases showed a significantly longer OS (p = 0.0009) and CSS (p = 0.0006) for women with TN positive tumour tissue and in women with high serum TN levels (p < 0.0001 for both). However, in the multivariate Cox regression analysis, only serum TN was found to be an independent prognostic factor for OS (p = 0.01) and not for CSS (p = 0.08). In conclusion, our results predict that a positive TN expression of both tumour tissue and serum points to a more favourable outcome for OC patients. Key words: Tissue array; ovarian cancer; tetranectin; prognosis; immunohistochemistry. Estrid Høgdall, Clinical Senior Investigator, Department of Pathology, Herlev Hospital, Herlev Ringvej 75, DK-2730 Herlev, Copenhagen, Denmark. e-mail: [email protected]
Epithelial ovarian cancer (OC) is the most common gynaecologic malignancy in the Western world, with some of the highest rates occurring in developed regions including Scandinavia (1). According to the Danish Gynecological Cancer Database (DGCD), Denmark currently has one of the highest age-standardized incidence rates of OC worldwide (11/100 000), with approximately 500 new cases Received 16 July 2014. Accepted 27 December 2014
annually (2–4). Although numbers have remained relatively constant over a generation (5), 2% of Danish women will develop OC before the age of 75 years (6, 7). Due to the paucity of early symptoms, the majority of women present with advanced-stage disease upon initial diagnosis, and this late-stage overall survival (OS) is approximately 25–30%, in contrast to women with early-stage disease, who have a survival rate of approximately 90% (8, 9). Current 401
HEERAN et al.
methods for detection of OC involve ultrasonography and a monitoring of the level of the serum glycoprotein CA125. However, the use of CA125 as a biomarker for OC is not optimal. It is non-specific for OC and thus, unreliable for early diagnosis (10). Prevalent therapeutic approaches to OC treatment comprise maximal surgical tumour debulking followed by combination chemotherapy with carboplatin and paclitaxel (11). Although standard chemotherapy may prolong survival, relapse often occurs resulting in chemoresistance. Therefore, due to paucity of early clinical symptoms and with the few diagnostic markers available for early detection, it is of paramount importance that OC is detected at an earlier stage so that the mechanisms of OC development is better understood so more effective and targeted treatments can be developed. Tetranectin (TN) belongs to the family of Ctype lectins that bind specifically to kringle four of plasminogen and is thought to have a role in the regulation of proteolytic and fibrinolysis processes (12, 13). Although the precise biological function of TN has not yet been clarified, it shows colocalization with plasminogen at the invasion front of melanomas, suggesting a role in cancer invasion and metastases (14). Several studies of a variety of human carcinomas including breast, colorectal, ovarian and oral tumours have reported correlation between low serum TN and a reduction in patient survival (15–18). Low serum level in combination with a high extracellular tumour tissue expression for TN has also been observed in OCs, thus highlighting TN as a potential immunohistochemistry (IHC) marker in the discrimination between malignant and benign ovarian tumours (19–23). Several antibodies have been used to evaluate the expression of TN in tumour tissue. Polyclonal antisera may contain antibody populations that react with antigens other than TN, but the commercially available polyclonal rabbit anti-human TN antibody A-371 (Dako, Glostrup, Denmark), which is known to recognize amino acid 17–181 of TN antigen, is well characterized with high specificity with respect to IHC (24). Monoclonal antibodies (MAbs) are known for producing a more specific immunostaining, and we therefore included MAb (Hyb130-11), which gives an intense staining intensity and MAb (Hyb130-12), which gives a significantly weaker staining intensity. Both of these antibodies react with non-conformational epitopes located with amino acid residues 50–181 (25). In this study, we tested the IHC performance characteristics of these anti-TN antibodies on TMA using a sizeable and well-defined Danish population-based series of OC and borderline ovarian
402
tumours (BOT). The result was correlated with selected clinicopathologic parameters and prognosis. MATERIALS AND METHODS Study population The MALOVA study (MALignant OVArian cancer study) is a multidisciplinary Danish study covering epidemiology (lifestyle factors), biochemistry and molecular biology with the purpose of identifying risk factors for OC. The design of the MALOVA study has been described elsewhere (26–29). Briefly, pre-operative blood samples and tumour tissue samples were obtained from most of the patients with a primary epithelial ovarian tumour. International Federation of Gynecology and Obstetrics (FIGO) stages were obtained from clinical records, which were reviewed by two gynaecologists, both specialized in OC. Furthermore, in the clinical records, women were either classified as radically operated with no macroscopic residual tumour or non-radically operated with macroscopic residual tumour. Histological grading and typing of the retrospectively collected paraffin-embedded tumour tissue samples were performed individually by two persons (E.H. and L.C.). The MALOVA study includes 681 women diagnosed with OC and 235 diagnosed with BOT. The study has been approved by the scientific ethical committee in the study area (KF01-384/95).
Follow-up In Denmark, all inhabitants have a personal identification number, which comprises the information regarding gender and birth date registered online in the Danish Civil Registration System. All OC women of this study were traced in the register, and date of death or emigration up to December 31st 2011, were registered. Women who died during follow-up, were also linked to The National Patient Registry and information on any hospital admissions to oncologic or gynaecologic departments was obtained. Relevant hospital files were collected, and information on the treatment (surgery and chemotherapy) and cause of death was established in October 2004. If there was uncertainty about the cause of death according to the patient file, the women were linked to the Danish Causes of Death Register. At the end of follow-up (31st December, 2011), a total of 520 OC women with OC had died (median follow-up time: 28 months, range: 1–176), and 161 OC women were still alive (median follow-up time: 176 months, range: 151–206).
Tissue array analysis Paraffin-embedded tissue from primary surgery was used for TA analyses. A diagnostic representative tissue block was used for the TA production, and the presence of epithelial tumour cells in all cores was assured by H&E staining prior to immunostaining. A total of 108 samples had to be excluded due to poor tissue quality, lack of tumour tissue in the original paraffin blocks, lack of tumour tissue in the selected cylinders, loss of tissue during the staining procedure or folding/tearing of sections during processing. A final 592 OC and 166 BOT cases were eligible for TN
© 2015 APMIS. Published by John Wiley & Sons Ltd
TETRANECTIN EXPRESSION IN OVARIAN CANCER
tissue expression and survival analyses. Kidney and liver tissue (four cores) were placed on each TA, not only as staining control but also to ensure orientation of the TA. The control tissues were stained with equal intensity on each run, and as the same staining results were obtained in two neighbouring slides from the same TA, which were used in both runs, we found the method reliable. The high performance may be a consequence of the fact that we used relatively large cores of 2 mm each.
Immunohistochemical analysis
Tissue preparation and immunostaining Two-micrometre sections (Section Transfer System, STS, Ergostar HM200; MICROM International GmbH, Walldorf, Germany) from the TA blocks were transferred to glass slides (Dako Chem-Mate Capillary Cap Microscope Slides, 57 mm; Dako A ⁄ S). Slides were stored at 4 °C for a maximum of 8 days until staining for TN. Prior to staining, the sections were deparaffinized in xylene and rehydrated in graded dilutions of alcohol. To expose antigens within the tissue, sections were pretreated in protease buffer (Protease type XXIV, P8038, 1:20,000, Sigma, St. Louis, MO, USA) at 37 °C for 5 min. The staining procedure for TN overexpression was performed on a Dako TechmateTM 500PLUS instrument using the polyclonal rabbit anti-human TN antibody (Code no. A-371; Dako), and two monoclonal mouse, Anti-human TN antigens (HYB 130-11 and 130.12; BioPorto, Hellerup, Denmark), diluted 1:100 followed by incubation for 1 h at room temperature. Other conditions were performed according to the manufacturer’s instructions.
TN scoring of tissue expression Two observers experienced in evaluating IHC stained tissue, simultaneously assessed the pattern of TN staining for each tumour sample. Standardization of scoring was achieved by comparison of scores, and any discrepancies were resolved by consensus. Scoring for TN expression was based on the immune-positivity of staining of a given tumour specimen, which exhibited a specific staining of tumour cells and a specific staining of a subgroup of stromal cells. As no international consensus on cut-off value for TN positivity has been published, a two-tiered scale, negative and positive was used for the description of TN expression in BOT and OC. In survival analyses, we used the same two-tiered scale, for the examination of overall survival (OS) and cancer specific survival (CSS).
Statistical analyses Statistical comparisons between groups were carried out using chi-square tests for categorical data or rank sum tests for ordinal variables. Survival rates were estimated using the Kaplan–Meier method. The impact of TN expression on OC mortality was estimated on the basis of the Cox proportional hazards model (30), applying the TN two-tiered scale using time since primary surgery as the time axis. The assumption of proportionality for the Cox model was assessed using Schoenfeld and martingale residuals as well as graphical methods. The assumption of proportional hazards between women not
© 2015 APMIS. Published by John Wiley & Sons Ltd
receiving chemotherapy, those treated with platin-based therapy and those treated with other anti-cancer drugs was not fulfilled. Therefore, a stratified proportional hazards model was used for multivariate analysis, with strata defined by the therapy group. Model assessment demonstrated that the proportional hazards assumption as well as linearity where applicable were fulfilled for other covariates included in this study. The regression analyses were adjusted for FIGO stage (I, II, III and IV), residual tumour after primary surgery, histological type of tumour (serous, mucinous, endometrioid and other histological types), age at diagnosis (linear) and histological grade (high, moderate and low). Confidence intervals (95% CI) for the corresponding parameters in the multivariate COX regression model were included. Overall survival and cancer specific survival were defined as length of time from primary surgery until death from any cause or death from OC, respectively. Results were considered statistically significant at the p < 0.05 level. The three antibodies, serum and age were analysed by the Spearman rank correlation, comparing rank order number rather than actual values, of two sets of variables to calculate a correlation coefficient. If a monotonic relationship existed between the two variables, one variable was invariably associated with an increase or decrease of the other (31). The analyses were performed, using the using the SAS, v9.2 (SAS Institute, Cary, NC, USA).
Table 1. Clinical characteristics and TN expression in tumour tissues from women diagnosed with borderline ovarian tumour (BOT) and ovarian cancer (OC) Characteristics Ovarian cancer Borderline (n = 592) ovarian tumour (n = 166) Total 592 166 FIGO stage I 164 (28%) 142 (85%) II 64 (11%) 5 (3%) III 310 (52%) 19 (11%) IV 54 (9%) 0 (0%) Tumour cell type Adenocarcinoma, 32 (5%) NOS Serous 365 (62%) adenocarcinoma Mucinous 54 (9%) adenocarcinoma Endometrioid 83 (14%) adenocarcinoma Clear cell carcinoma 47 (8%) Undifferentiated 11 (2%) Treatment No chemotherapy 146 (25%) Chemotherapy 48 (8%) without platinum Chemotherapy 398 (67%) with platinum Age median (range) 59 (35–79) 55 (35–79) FIGO, International Federation of Gynecology and Obstetrics, NOS, not otherwise specified.
403
HEERAN et al.
130-11
130-12
A371
A
B
C
D
E
F
G
H
I
J
K
L
Fig. 1. Tetranectin-stained tissues from Ovarian Cancer (OC) patients (objective x20) showing (A) serous adenocarcinoma OC stained with 130-11 antibody, (B) serous adenocarcinoma OC stained with 130-12 antibody, (C) serous adenocarcinoma OC stained with A-371 antibody, (D) clear cell carcinoma OC stained with 130-11 antibody, (E) clear cell carcinoma OC stained with 130-12 antibody, (F) clear cell carcinoma OC stained with A-371 antibody, (G) mucinous adenocarcinoma OC stained with 130-11 antibody, (H) mucinous adenocarcinoma OC stained with 130-12 antibody, (I) mucinous adenocarcinoma OC stained with A-371 antibody, (J) endometrioid adenocarcinoma OC stained with 130-11 antibody, (K) endometrioid adenocarcinoma OC stained with 130-12 antibody and (L) endometrioid adenocarcinoma OC stained with A-371 antibody. All three images (J, K and L) were from the exact same position on the TMA core.
RESULTS Clinicopathological characteristics of OC and BOT
Results from the analysis of TN by TA of the 758 tumours originating from our population404
based collection are shown in Table 1. The age of OC women ranged 35–79 years, with an average of 59 years and the age of BOT patients ranged 35–79 years, with an average of 55 years old. Of all 592 OC tissue samples 310 (52%) were © 2015 APMIS. Published by John Wiley & Sons Ltd
TETRANECTIN EXPRESSION IN OVARIAN CANCER
found within FIGO group III, and of all 166 BOT tissue samples 142 (85%) were found in FIGO group I. Most of the OC tumours were serous adenocarcinoma 62% (365), and a majority of women diagnosed with OC received chemotherapy with platinum 67% (398). Details of patient characteristics and descriptive statistics for the tumours are shown in Table 1. Representative photomicrographs of tumour tissue showing positive staining for the specific antigen are presented in Fig. 1. Correlations between the TN markers and to age
To investigate whether the prognostic value of TN expression in ovarian tumours was dependent on the antibody used for IHC detection, a comparison was made between the three different antibodies recognizing TN within tumour tissue on one side and a comparison between these and serum TN on the other. This was done to evaluate comparability as well as prognostic value of these markers. The Spearman’s rank correlation was used to examine the relationship between the four TN identification measures. Weak associations were observed among them all with absolute values of p = 0.3. Correlations with age were weak as well (p = 0.1) (Table 2). Consecutive tissue sections had been used in order to demonstrate TN staining in exactly the same tissue position of the sample (Fig. 1).
Table 2. Relationship between TN markers and Spearman rank correlation Covariate TN poly TN_30-11 TN_30-12 TN_Serum 0.13 0.007 0.003 0.004 0.87 0.96 486 489 400 TN_Poly 0.02 0.02 0.68 0.71 731 606 TN_30-11 0.26
