Abstract
Purpose
Advances in transoral laser microsurgery, radiotherapy, and chemotherapy (and their combinations) have reduced the indications for open partial laryngectomies, although they have replaced total laryngectomies in selected advanced or recurrent laryngeal squamous cell carcinomas (LSCCs). Tissue hypoxia in malignancies appears to be strongly associated with tumor cell invasiveness and metastases. Whether hypoxia-inducible factors can contribute to a rational recommendation of open partial laryngectomy should be investigated.
Materials and methods
Fifty consecutive patients who had undergone primary open partial laryngectomy (supraglottic and supracricoid laryngectomies) were investigated, measuring the immunohistochemical expression of the hypoxia-inducible proteins angiogenin and endoglin in their primary LSCCs also with image analysis.
Results
Univariate analysis showed a significantly higher recurrence rate ( P = .007) and shorter disease-free survival ( P = .0047) in patients with LSCC with endoglin expression more than 9.0%. Multivariate analysis found endoglin expression independently prognostic in terms of disease-free survival ( P = .012). Angiogenin expression (in carcinoma or endothelial cells) was not associated with prognosis.
Conclusions
Endoglin should be further studied as a biomarker of patients with LSCC at higher risk for recurrence after open partial laryngectomy who may benefit from more aggressive treatments. Endoglin expression in positive laryngeal biopsies may prove useful as a parameter for choosing between different surgical and multimodality approaches to controversial LSCC cases.
1
Introduction
The role of open partial laryngectomy in modern oncological head and neck surgery and its indications are definitely under debate. For early and advanced laryngeal squamous cell carcinoma (LSCC), improvements in transoral laser microsurgery, radiotherapy (RT), and chemotherapy (and their combinations) have achieved oncological results comparable with open surgery, reducing the indications for open partial laryngectomy . On the other hand, primary open partial laryngectomies offer a reliable alternative to total laryngectomy in appropriately selected patients, maintaining local control and overall survival rates while preserving voice and swallowing function . Partial procedures, in particular, supracricoid partial laryngectomies (SCPLs), have proved effective even in selected T3 and T4 carcinomas . At the ENT Unit in Vittorio Veneto, Lucioni et al found that SCPLs (including tracheohyoidopexy and tracheohyoidoepiglottopexy, as described by Rizzotto et al ) used on the strength of appropriate selection criteria (endoscopy, even with angled endoscopes, and spiral computed tomography) reduced the need for total laryngectomy in carcinomas limited to the larynx to approximately 10% of cases. Furthermore, although 70% to 80% of laryngeal carcinoma recurrences after RT are staged as rT3/T4 and although total laryngectomy is considered the treatment of choice in most of these cases, recent results suggest that, adopting appropriate selection criteria for salvage partial laryngeal surgery, laryngeal function can be preserved with comparable survival rates .
Given recent advances in our understanding of the prognostic significance of biomarkers, their potential contribution to the rational recommendation of open partial laryngectomies should be investigated. Tumor hypoxia can develop because of a greater metabolic activity and oxygen consumption by rapidly proliferating tumor cells, leading to local pH changes and consequent oxidative stress in the surrounding microenvironment. A decreasing O 2 concentration because of increasing diffusion distances from local capillaries because of an expanding cell mass also contributes to a hypoxic microenvironment . Tissue hypoxia in tumors appears to be strongly associated with tumor cell invasiveness, metastatic spread, and overall clinical outcome . Angiogenin (ANG) is a 14-kd, 123-amino acid member of the ribonuclease superfamily. It is one of a large group of angiogenic genes that are up-regulated under hypoxic conditions. Angiogenin supports primary and metastatic tumor growth through the maintenance of an adequate blood supply capable of adapting to the tumor’s increasing size . Endoglin (CD105), an accessory coreceptor for transforming growth factor β , is a hypoxia-inducible, disulfide-linked, proliferation-associated, homodimeric cell membrane glycoprotein of 180 kd expressed in angiogenic endothelial cells. In clinical studies, endoglin antibodies have shown a significant specificity for the tumor vasculature. Assessing neovascularization from endoglin staining is emerging as a potential prognostic indicator for several solid malignancies, including LSCC .
The aim of the present study was to evaluate the potential prognostic role of hypoxia-inducible factors (endoglin and ANG) in a series of patients who underwent consecutively primary open partial laryngeal surgery (supraglottic laryngectomy or SCPL) for carcinoma.
2
Materials and methods
2.1
Patients
The study was conducted on 50 patients with LSCC (41 men and 9 women, with a mean age of 63.4 ± 7.8 years) treated consecutively with open partial or reconstructive laryngectomy. Preoperatively, all patients underwent microlaryngoscopy with laryngeal biopsy, upper aerodigestive tract endoscopy, esophagoscopy, neck ultrasonography (with or without fine needle aspiration cytology), head and neck contrast-enhanced computed tomography and/or magnetic resonance imaging, chest x-rays, and liver ultrasonography.
Twenty-one patients had a primary supraglottic laryngectomy, and 29 a SCPL, all performed at the Otolaryngology Section of Padova University. Thirty-nine patients also underwent unilateral or bilateral cervical lymph node dissection; postoperative RT was administered in 6 cases. The indications for neck dissection and postoperative RT were based on currently accepted protocols . Based on the seventh edition of the TNM Classification of Malignant Tumors , the pT stage was T1 in 9 cases, T2 in 25, T3 in 12, and T4a in 4; the pN status was N 0 in 28 cases, N1 in 4, and N2 in 7 cases. No distant metastases (M) were detected at diagnosis. As for pathologic grading, 14 of the 50 cases were G1, 25 were G2, and 11 were G3. The mean follow-up was 44.3 ± 23.3 months.
2.2
Immunohistochemistry
All tissues were fixed in 4% paraformaldehyde and embedded in paraffin wax. Sections 5- μ m thick were cut for immunohistochemical examination in all 50 cases. The sections were scanned by the pathologist at magnification × 5 to select the 3 areas with the greatest degree of vascularization (hot spots) and no signs of necrosis or hemorrhage within the less differentiated areas of tumor. The same fields were exactly matched in the specimens, and for each sample, endoglin and ANG reactivity were evaluated in these defined areas.
2.3
Immunohistochemistry for ANG
Angiogenin reactivity was determined with mouse monoclonal antibody, clone MANG-1 (AbDserotech, MorphoSys, Oxford, Newcastle, UK), diluted 1:400. Immunohistochemical staining was performed using a fully automated system (Bond-maX; Vision BioSystems, UK). Tissue sections were dewaxed and rehydrated by successive incubations at 72°C in Bond Dewax Solution (Vision BioSystems), ethanol, and distilled water. Sections were pretreated with proteinase K for 10 minutes at 37°C. Endogenous peroxidase was blocked with 3% hydrogen peroxide before incubation for 15 minutes with the primary antibody. Specimens were washed with phosphate-buffered solution (PBS) (pH 7.0) and incubated with Bond Polymer Refine Detection Kit (Vision BioSystems). The sections were then dehydrated, cleared, and mounted. The color was developed using 3,3′-diaminobenzidine (Vision BioSystems) for 4 minutes. The sections were counterstained with Mayer’s hematoxylin. Samples from tonsils were used as a positive control, and serum without the primary antibody, as a negative control.
Angiogenin expression was determined in both carcinoma cells and intratumoral vessels; in carcinoma cells, it was measured as a percentage of ANG-stained cells; at magnification × 20, in intratumoral vessels, it was scored from 0 to 3 (0, < 1 stained vessel per field; 1, faint endothelial staining in 1 vessel per field; 2, endothelial staining in 2 or 3 vessels per field; and 3, endothelial staining in > 3 vessels per field).
2.4
Immunohistochemistry for endoglin
The sections were preincubated with protein block (Novocastra Laboratories Ltd, Newcastle upon Tyne, UK) for 5 minutes and stained with mouse monoclonal antibody CD105 (clone SN6h 1:10 dilution [Dako, Glostrup, Denmark]), then postprimary block (Novolink Polymer Detection System; Novocastra) was applied to the specimens for 20 minutes. The specimens were then washed with PBS (pH 7.0) for 3 minutes and incubated with Novolink Polymer for 20 minutes. The color was developed using 3,3′ diaminobenzidine (Dako) for 4 minutes. Sections were counterstained with Mayer’s hematoxylin. An angiosarcoma sample was used as a positive control for endoglin; as a negative control, the antibody was replaced with PBS. Cytoplasmic endoglin staining was determined in endothelial cells of LSCC.
2.5
Image analysis for endoglin measurements
All samples were assessed on an image analysis (IA) workstation consisting of a conventional Zeiss Axioskop light microscope (Zeiss, Jena, Germany) with a color digital, Peltier-cooled videocamera (MicroPublisher 5.0 RTV, QImaging, Surrey, BC, Canada) connected to a personal computer with the Image-Pro Plus Version 7® for Windows™ IA program (Media Cybernetics Inc, Bethesda, MD). In all cases, 1378 × 954 μ m areas of tumor tissue (hot spots) were evaluated with a 495-point sampling grid superimposed by the program on the image acquired with a × 50 field of view. After counting the points intercepting the positive and negative areas, the proportion of the endoglin-positive area was calculated and recorded as a percentage (%).
2.6
Statistical analysis
The statistical tests applied were Fisher’s exact test, the Mann-Whitney U test, the nonparametric test for trend, and Spearman rank correlation test, as appropriate. The receiver operating curve (ROC) approach (failure vs parameter) was used to set the analytically best fitting cutoff to binarize the continuous variable endoglin expression according to the highest level of the positive likelihood ratio. The Kaplan-Meier product limit estimator, the log-rank test, and Cox proportional hazards regression model were also used to compare disease-free survival (DFS) stratified by the different parameters analyzed.
In the multivariate analysis, Cox proportional hazards regression identified significant predictors of DFS. After defining a full model with the variables of interest (ie, pT [pT1-T2 vs pT3-T4], N status [N 0 vs N +], ANG score in intratumoral vessels, ANG expression in carcinoma cells, and endoglin expression) and checking for multicollinearity, a backward variable selection was used to derive a new model. Then, after estimating the maximum likelihood, a P exclusion value of P > .30 was adopted, and a final model was produced.
P < .05 was considered significant. The STATA 8 statistical package (Stata Corp, College Station, TX) was used for all analyses.
2
Materials and methods
2.1
Patients
The study was conducted on 50 patients with LSCC (41 men and 9 women, with a mean age of 63.4 ± 7.8 years) treated consecutively with open partial or reconstructive laryngectomy. Preoperatively, all patients underwent microlaryngoscopy with laryngeal biopsy, upper aerodigestive tract endoscopy, esophagoscopy, neck ultrasonography (with or without fine needle aspiration cytology), head and neck contrast-enhanced computed tomography and/or magnetic resonance imaging, chest x-rays, and liver ultrasonography.
Twenty-one patients had a primary supraglottic laryngectomy, and 29 a SCPL, all performed at the Otolaryngology Section of Padova University. Thirty-nine patients also underwent unilateral or bilateral cervical lymph node dissection; postoperative RT was administered in 6 cases. The indications for neck dissection and postoperative RT were based on currently accepted protocols . Based on the seventh edition of the TNM Classification of Malignant Tumors , the pT stage was T1 in 9 cases, T2 in 25, T3 in 12, and T4a in 4; the pN status was N 0 in 28 cases, N1 in 4, and N2 in 7 cases. No distant metastases (M) were detected at diagnosis. As for pathologic grading, 14 of the 50 cases were G1, 25 were G2, and 11 were G3. The mean follow-up was 44.3 ± 23.3 months.
2.2
Immunohistochemistry
All tissues were fixed in 4% paraformaldehyde and embedded in paraffin wax. Sections 5- μ m thick were cut for immunohistochemical examination in all 50 cases. The sections were scanned by the pathologist at magnification × 5 to select the 3 areas with the greatest degree of vascularization (hot spots) and no signs of necrosis or hemorrhage within the less differentiated areas of tumor. The same fields were exactly matched in the specimens, and for each sample, endoglin and ANG reactivity were evaluated in these defined areas.
2.3
Immunohistochemistry for ANG
Angiogenin reactivity was determined with mouse monoclonal antibody, clone MANG-1 (AbDserotech, MorphoSys, Oxford, Newcastle, UK), diluted 1:400. Immunohistochemical staining was performed using a fully automated system (Bond-maX; Vision BioSystems, UK). Tissue sections were dewaxed and rehydrated by successive incubations at 72°C in Bond Dewax Solution (Vision BioSystems), ethanol, and distilled water. Sections were pretreated with proteinase K for 10 minutes at 37°C. Endogenous peroxidase was blocked with 3% hydrogen peroxide before incubation for 15 minutes with the primary antibody. Specimens were washed with phosphate-buffered solution (PBS) (pH 7.0) and incubated with Bond Polymer Refine Detection Kit (Vision BioSystems). The sections were then dehydrated, cleared, and mounted. The color was developed using 3,3′-diaminobenzidine (Vision BioSystems) for 4 minutes. The sections were counterstained with Mayer’s hematoxylin. Samples from tonsils were used as a positive control, and serum without the primary antibody, as a negative control.
Angiogenin expression was determined in both carcinoma cells and intratumoral vessels; in carcinoma cells, it was measured as a percentage of ANG-stained cells; at magnification × 20, in intratumoral vessels, it was scored from 0 to 3 (0, < 1 stained vessel per field; 1, faint endothelial staining in 1 vessel per field; 2, endothelial staining in 2 or 3 vessels per field; and 3, endothelial staining in > 3 vessels per field).
2.4
Immunohistochemistry for endoglin
The sections were preincubated with protein block (Novocastra Laboratories Ltd, Newcastle upon Tyne, UK) for 5 minutes and stained with mouse monoclonal antibody CD105 (clone SN6h 1:10 dilution [Dako, Glostrup, Denmark]), then postprimary block (Novolink Polymer Detection System; Novocastra) was applied to the specimens for 20 minutes. The specimens were then washed with PBS (pH 7.0) for 3 minutes and incubated with Novolink Polymer for 20 minutes. The color was developed using 3,3′ diaminobenzidine (Dako) for 4 minutes. Sections were counterstained with Mayer’s hematoxylin. An angiosarcoma sample was used as a positive control for endoglin; as a negative control, the antibody was replaced with PBS. Cytoplasmic endoglin staining was determined in endothelial cells of LSCC.
2.5
Image analysis for endoglin measurements
All samples were assessed on an image analysis (IA) workstation consisting of a conventional Zeiss Axioskop light microscope (Zeiss, Jena, Germany) with a color digital, Peltier-cooled videocamera (MicroPublisher 5.0 RTV, QImaging, Surrey, BC, Canada) connected to a personal computer with the Image-Pro Plus Version 7® for Windows™ IA program (Media Cybernetics Inc, Bethesda, MD). In all cases, 1378 × 954 μ m areas of tumor tissue (hot spots) were evaluated with a 495-point sampling grid superimposed by the program on the image acquired with a × 50 field of view. After counting the points intercepting the positive and negative areas, the proportion of the endoglin-positive area was calculated and recorded as a percentage (%).
2.6
Statistical analysis
The statistical tests applied were Fisher’s exact test, the Mann-Whitney U test, the nonparametric test for trend, and Spearman rank correlation test, as appropriate. The receiver operating curve (ROC) approach (failure vs parameter) was used to set the analytically best fitting cutoff to binarize the continuous variable endoglin expression according to the highest level of the positive likelihood ratio. The Kaplan-Meier product limit estimator, the log-rank test, and Cox proportional hazards regression model were also used to compare disease-free survival (DFS) stratified by the different parameters analyzed.
In the multivariate analysis, Cox proportional hazards regression identified significant predictors of DFS. After defining a full model with the variables of interest (ie, pT [pT1-T2 vs pT3-T4], N status [N 0 vs N +], ANG score in intratumoral vessels, ANG expression in carcinoma cells, and endoglin expression) and checking for multicollinearity, a backward variable selection was used to derive a new model. Then, after estimating the maximum likelihood, a P exclusion value of P > .30 was adopted, and a final model was produced.
P < .05 was considered significant. The STATA 8 statistical package (Stata Corp, College Station, TX) was used for all analyses.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
