Stefan M. Willems
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Stefan M. Willems has a total of 115 co-authors. Here are the top co-authors by number of shared publications:
- Joost Bart
Department of Pathology, University Medical Centre Groningen, Groningen, The Netherlands
3 shared publications - Robert J. J. van Es
Department of Head and Neck Surgical Oncology, University Medical Center Utrecht, Utrecht, The Netherlands
3 shared publications - Willem W. B. de Kort
Department of Pathology, University Medical Center Utrecht, Utrecht, The Netherlands
3 shared publications A.J. van der Wekken
Department of Pulmonary Diseases and Tuberculosis, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands
2 shared publications- Alfio Ferlito
Coordinator of the International Head and Neck Scientific Group Padua Italy
2 shared publications - Annemieke H. van der Hout
Department of Genetics, University Medical Center Groningen, University of Groningen, 9700 RB Groningen, The Netherlands
2 shared publications
Latest Publications
- 24 February 2026
- Published by Elsevier in European Journal of Cancer
- Vol. 237, 116597
Corrigendum to “Regional differences in predictive biomarker testing rates for patients with metastatic NSCLC in the Netherlands” [Eur J Cancer 205 (2024)114125]
- 24 February 2026
- Published by Elsevier in European Journal of Cancer
- Vol. 237, 116597
Corrigendum to “Regional differences in predictive biomarker testing rates for patients with metastatic NSCLC in the Netherlands” [Eur J Cancer 205 (2024)114125]
- 29 November 2025
- Published by Springer Nature in Virchows Archiv
- Vol. 488 (6) , 1283-1291
Large language model-based approach for low-cost, rapid and accurate automated extraction of predictive biomarker testing results from Dutch pathology reports
National cancer registries and pathology report databases contain valuable real-world information that may be used for assessment of healthcare quality or retrospective observational studies. As many of these data are currently non- or semi-structured, manual information extraction is needed, which is labor-intensive, costly and causes significant time delay to data analysis. This study aimed to develop a large language model (LLM)-based method to extract biomarker testing data for two routinely tested predictive biomarkers — EGFR and KRAS — from pathology reports of patients with non-small cell lung cancer (NSCLC). Patient cohorts and pathology reports were derived from the Netherlands Cancer Registry (NCR) and the Dutch nationwide pathology databank (Palga). Manually captured data regarding EGFR and KRAS testing in 3887 patients diagnosed with metastatic NSCLC in the Netherlands in 2019 were used for training, testing and validation. Annotated data included biomarker testing status for EGFR and KRAS, the use of next-generation sequencing (NGS), and test results including specific mutation(s). In the test set of the 2019 cohort, the model yielded (micro) F1 scores ≥ 0.98 for all variables (overall biomarker testing status, KRAS/EGFR testing, KRAS/EGFR test results). The trained model was then applied to pathology reports of 4122 patients diagnosed with NSCLC between July 2022 and June 2023, with manual annotation for 410 randomly selected cases to determine model accuracy. In this test set, (micro) F1 scores ≥ 0.95 were achieved. None of the manually annotated positive molecular test results were missed. Standardized notation of reported mutations was correct in 98.7% and 100.0% of KRAS and EGFR mutations, respectively. In the entire 2022–2023 cohort, model output revealed overall test rates of 88.1% and 86.4% for KRAS and EGFR, respectively. Among the tested patients, the model described positive KRAS and EGFR test results in 40.5% and 11.5%, respectively. This study illustrates the possibility to train and use an LLM-based model to accurately extract biomarker testing results from pathology reports of patients with lung cancer. This application enables rapid, low-cost assessment of biomarker testing, which can be used for evaluation of guideline adherence and retrospective biomarker follow-up studies at a nationwide level of the Netherlands.
- 29 November 2025
- Published by Springer Nature in Virchows Archiv
- Vol. 488 (6) , 1283-1291
Large language model-based approach for low-cost, rapid and accurate automated extraction of predictive biomarker testing results from Dutch pathology reports
National cancer registries and pathology report databases contain valuable real-world information that may be used for assessment of healthcare quality or retrospective observational studies. As many of these data are currently non- or semi-structured, manual information extraction is needed, which is labor-intensive, costly and causes significant time delay to data analysis. This study aimed to develop a large language model (LLM)-based method to extract biomarker testing data for two routinely tested predictive biomarkers — EGFR and KRAS — from pathology reports of patients with non-small cell lung cancer (NSCLC). Patient cohorts and pathology reports were derived from the Netherlands Cancer Registry (NCR) and the Dutch nationwide pathology databank (Palga). Manually captured data regarding EGFR and KRAS testing in 3887 patients diagnosed with metastatic NSCLC in the Netherlands in 2019 were used for training, testing and validation. Annotated data included biomarker testing status for EGFR and KRAS, the use of next-generation sequencing (NGS), and test results including specific mutation(s). In the test set of the 2019 cohort, the model yielded (micro) F1 scores ≥ 0.98 for all variables (overall biomarker testing status, KRAS/EGFR testing, KRAS/EGFR test results). The trained model was then applied to pathology reports of 4122 patients diagnosed with NSCLC between July 2022 and June 2023, with manual annotation for 410 randomly selected cases to determine model accuracy. In this test set, (micro) F1 scores ≥ 0.95 were achieved. None of the manually annotated positive molecular test results were missed. Standardized notation of reported mutations was correct in 98.7% and 100.0% of KRAS and EGFR mutations, respectively. In the entire 2022–2023 cohort, model output revealed overall test rates of 88.1% and 86.4% for KRAS and EGFR, respectively. Among the tested patients, the model described positive KRAS and EGFR test results in 40.5% and 11.5%, respectively. This study illustrates the possibility to train and use an LLM-based model to accurately extract biomarker testing results from pathology reports of patients with lung cancer. This application enables rapid, low-cost assessment of biomarker testing, which can be used for evaluation of guideline adherence and retrospective biomarker follow-up studies at a nationwide level of the Netherlands.
- Open Access
- 17 October 2024
- Published by Elsevier in JTO Clinical and Research Reports
- Vol. 5 (12) , 100745
Prognostic Value of KRAS/TP53 Status for Overall Survival in First-Line Monoimmunotherapy and Chemoimmunotherapy Treated Patients With Nonsquamous NSCLC in the Netherlands: A Brief Report
- 17 October 2024
- Published by Elsevier in JTO Clinical and Research Reports
- Vol. 5 (12) , 100745
Open AccessPrognostic Value of KRAS/TP53 Status for Overall Survival in First-Line Monoimmunotherapy and Chemoimmunotherapy Treated Patients With Nonsquamous NSCLC in the Netherlands: A Brief Report
- Open Access
- 20 May 2024
- Published by Elsevier in European Journal of Cancer
- Vol. 205, 114125
Regional differences in predictive biomarker testing rates for patients with metastatic NSCLC in the Netherlands
- 20 May 2024
- Published by Elsevier in European Journal of Cancer
- Vol. 205, 114125
Open AccessRegional differences in predictive biomarker testing rates for patients with metastatic NSCLC in the Netherlands
- Open Access
BRCA1/2 Testing Landscape in Ovarian Cancer: A Nationwide, Real-World Data Study
Analyzing BRCA1/2 tumor pathogenic variants (TPVs) in epithelial tubal/ovarian cancers (EOCs) has become an essential part of the diagnostic workflow in many centers to guide treatment options and genetic cascade testing. However, there is no standardization of testing procedures, including techniques, gene assays, or sequencers used, and data on the execution of tumor tests remains scarce. Therefore, we evaluated characteristics of BRCA1/2 tumor testing in advanced-stage EOC with real-world national data. Pathology reports of patients diagnosed with EOC in 2019 in the Netherlands were obtained from the Dutch Pathology Registry (PALGA), and data regarding histological subtype and BRCA1/2 tumor tests were extracted. A total of 999 patients with advanced-stage EOC were included. Tumor tests were performed for 502 patients (50.2%) and BRCA1/2 TPVs were detected in 14.7%. Of all tests, 48.6% used hybrid capture techniques and 26.5% used PCR-based techniques. More than half of the tests (55.0%) analyzed other genes in addition to BRCA1/2. Overall, this study highlights the heterogeneity in the execution of BRCA1/2 tumor tests. Despite a lack of evidence of quality differences, we emphasize that adequate reporting and internal and external quality monitors are essential for the high-quality implementation and execution of reliable BRCA1/2 tumor testing, which is crucial for identifying all patients with BRCA1/2 TPVs.
Open AccessBRCA1/2 Testing Landscape in Ovarian Cancer: A Nationwide, Real-World Data Study
Analyzing BRCA1/2 tumor pathogenic variants (TPVs) in epithelial tubal/ovarian cancers (EOCs) has become an essential part of the diagnostic workflow in many centers to guide treatment options and genetic cascade testing. However, there is no standardization of testing procedures, including techniques, gene assays, or sequencers used, and data on the execution of tumor tests remains scarce. Therefore, we evaluated characteristics of BRCA1/2 tumor testing in advanced-stage EOC with real-world national data. Pathology reports of patients diagnosed with EOC in 2019 in the Netherlands were obtained from the Dutch Pathology Registry (PALGA), and data regarding histological subtype and BRCA1/2 tumor tests were extracted. A total of 999 patients with advanced-stage EOC were included. Tumor tests were performed for 502 patients (50.2%) and BRCA1/2 TPVs were detected in 14.7%. Of all tests, 48.6% used hybrid capture techniques and 26.5% used PCR-based techniques. More than half of the tests (55.0%) analyzed other genes in addition to BRCA1/2. Overall, this study highlights the heterogeneity in the execution of BRCA1/2 tumor tests. Despite a lack of evidence of quality differences, we emphasize that adequate reporting and internal and external quality monitors are essential for the high-quality implementation and execution of reliable BRCA1/2 tumor testing, which is crucial for identifying all patients with BRCA1/2 TPVs.
- Open Access
- 9 April 2024
- Published by MDPI AG in Preprints.org
BRCA1/2 Testing Landscape in Ovarian Cancer: A Nationwide, Real-World Data Study
Analyzing BRCA1/2 tumor pathogenic variants (TPVs) in epithelial tubal/ovarian cancers (EOC) has become an essential part of the diagnostic workflow in many centers to guide treatment options and genetic cascade testing. However, there is no standardization of testing procedures, including techniques, gene assays, or sequencers used, and data on the execution of tumor tests remains scarce. Therefore, we evaluated characteristics of BRCA1/2 tumor testing in advanced-stage EOC with real-world national data. Pathology reports of patients diagnosed with EOC in 2019 in the Netherlands were obtained from the Dutch Pathology Registry (PALGA), and data regarding histological subtype and BRCA1/2 tumor tests were extracted. A total of 999 patients with advanced-stage EOC were included. Tumor tests were performed for 502 patients (50.2%) and BRCA1/2 TPVs were detected in 14.7%. Of all tests, 48.6% used hybrid capture techniques and 26.5% used PCR-based techniques. More than half of the tests (55.0%) analyzed other genes in addition to BRCA1/2. Overall, this study highlights the heterogeneity in the execution of BRCA1/2 tumor tests. Despite a lack of evidence of quality differences, we emphasize that adequate reporting and internal and external quality monitors are essential for high-quality implementation and execution of reliable BRCA1/2 tumor testing, which is crucial for identifying all patients with BRCA1/2 TPVs.
- 9 April 2024
- Published by MDPI AG in Preprints.org
Open AccessBRCA1/2 Testing Landscape in Ovarian Cancer: A Nationwide, Real-World Data Study
Analyzing BRCA1/2 tumor pathogenic variants (TPVs) in epithelial tubal/ovarian cancers (EOC) has become an essential part of the diagnostic workflow in many centers to guide treatment options and genetic cascade testing. However, there is no standardization of testing procedures, including techniques, gene assays, or sequencers used, and data on the execution of tumor tests remains scarce. Therefore, we evaluated characteristics of BRCA1/2 tumor testing in advanced-stage EOC with real-world national data. Pathology reports of patients diagnosed with EOC in 2019 in the Netherlands were obtained from the Dutch Pathology Registry (PALGA), and data regarding histological subtype and BRCA1/2 tumor tests were extracted. A total of 999 patients with advanced-stage EOC were included. Tumor tests were performed for 502 patients (50.2%) and BRCA1/2 TPVs were detected in 14.7%. Of all tests, 48.6% used hybrid capture techniques and 26.5% used PCR-based techniques. More than half of the tests (55.0%) analyzed other genes in addition to BRCA1/2. Overall, this study highlights the heterogeneity in the execution of BRCA1/2 tumor tests. Despite a lack of evidence of quality differences, we emphasize that adequate reporting and internal and external quality monitors are essential for high-quality implementation and execution of reliable BRCA1/2 tumor testing, which is crucial for identifying all patients with BRCA1/2 TPVs.
- Open Access
- 18 September 2023
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 12 (18) , 6035
Jaw Bone Invasion of Oral Squamous Cell Carcinoma Is Associated with Osteoclast Count and Expression of Its Regulating Proteins in Patients and Organoids
Aims: Oral squamous cell carcinoma (OSCC) frequently invades the jaw. The exact mechanism of bone invasion remains unclear. This study investigates (premature) osteoclasts and the expression of its differentiation regulating proteins RANKL, OPG and RANK in patients with OSCC. Methods: Resection specimens from OSCC patients were divided into NI group (No Invasion), E group (Erosion) or I group (bone Invasion). Tissue sections were stained with Cathepsin K (osteoclast-counting), RANKL, OPG and RANK. The staining intensity was scored on different regions of the tumor: front, center, back and normal mucosa. Immunohistochemistry and qPCR for RANKL/OPG/RANK were performed on five head and neck squamous cell carcinoma (HNSCC) organoids. Results: The mean number of osteoclasts (I group) and premature osteoclasts (E group) was significantly higher compared to the NI group (p = 0.003, p = 0.036). RANKL expression was significantly higher in the tumor front and tumor center compared to normal mucosa (all groups). In the I group, RANKL and RANK expression was significantly higher in the tumor front compared to the tumor back and there was a trend of higher RANKL expression in the tumor front compared to the E group and NI group. qPCR showed a 20–43 times higher RANKL mRNA expression in three out of five tumor organoids compared to a normal squamous cell organoid line. There was no correlation between protein and mRNA expression in the HNSCC organoids. Conclusions: These findings suggest that OSCCs induce bone invasion by stimulating osteoclast activation by regulating the production of RANKL and RANK proteins.
- 18 September 2023
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 12 (18) , 6035
Open AccessJaw Bone Invasion of Oral Squamous Cell Carcinoma Is Associated with Osteoclast Count and Expression of Its Regulating Proteins in Patients and Organoids
Aims: Oral squamous cell carcinoma (OSCC) frequently invades the jaw. The exact mechanism of bone invasion remains unclear. This study investigates (premature) osteoclasts and the expression of its differentiation regulating proteins RANKL, OPG and RANK in patients with OSCC. Methods: Resection specimens from OSCC patients were divided into NI group (No Invasion), E group (Erosion) or I group (bone Invasion). Tissue sections were stained with Cathepsin K (osteoclast-counting), RANKL, OPG and RANK. The staining intensity was scored on different regions of the tumor: front, center, back and normal mucosa. Immunohistochemistry and qPCR for RANKL/OPG/RANK were performed on five head and neck squamous cell carcinoma (HNSCC) organoids. Results: The mean number of osteoclasts (I group) and premature osteoclasts (E group) was significantly higher compared to the NI group (p = 0.003, p = 0.036). RANKL expression was significantly higher in the tumor front and tumor center compared to normal mucosa (all groups). In the I group, RANKL and RANK expression was significantly higher in the tumor front compared to the tumor back and there was a trend of higher RANKL expression in the tumor front compared to the E group and NI group. qPCR showed a 20–43 times higher RANKL mRNA expression in three out of five tumor organoids compared to a normal squamous cell organoid line. There was no correlation between protein and mRNA expression in the HNSCC organoids. Conclusions: These findings suggest that OSCCs induce bone invasion by stimulating osteoclast activation by regulating the production of RANKL and RANK proteins.
- Open Access
- 24 July 2023
- Published by Springer Nature in Head and Neck Pathology
- Vol. 17 (3) , 697-707
p-mTOR, p-ERK and PTEN Expression in Tumor Biopsies and Organoids as Predictive Biomarkers for Patients with HPV Negative Head and Neck Cancer
Background: Survival rates of head and neck squamous cell carcinoma (HNSCC) have only marginally improved in the last decades. Hence there is a need for predictive biomarkers for long-time survival that can help to guide treatment decisions and might lead to the development of new therapies. The phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR signaling pathway is the most frequently altered pathway in HNSCC, genes are often mutated, amplificated and overexpressed causing aberrant signaling affecting cell growth and differentiation. Numerous genetic alterations of upstream and downstream factors have currently been clarified. However, their predictive value has yet to be established. Therefore we assess the predictive value of p-mTOR, p-ERK and PTEN expression. Methods: Tissue microarrays (TMA’s) of HPV-negative patients with oropharyngeal (n = 48), hypopharyngeal (n = 16) or laryngeal (n = 13) SCC, treated with primary chemoradiation (cisplatin/carboplatin/cetuximab and radiotherapy), were histologically stained for p-mTOR, PTEN and p-ERK. Expression was correlated to overall survival (OS), disease free survival (DFS) and locoregional control (LRC). Also p-mTOR was histologically stained in a separate cohort of HNSCC organoids (n = 8) and correlated to mTOR-inhibitor everolimus response. Results: High p-mTOR expression correlated significantly with worse OS in multivariate analysis in the whole patient cohort [Hazar Ratio (HR) 1.06, 95%CI 1.01–1.11, p = 0.03] and in the cisplatin/carboplatin group with both worse OS (HR 1.09, 95%CI 1.02–1.16, p = 0.02) and DFS (HR 1.06, 95%CI 1.00–1.12, p = 0,04). p-ERK expression correlated significantly with DFS in univariate analysis in the whole patient cohort (HR 1.03, 95%CI 1.00–1.05, p = 0.04) and cisplatin/carboplatin group (HR 1.03, 95%CI 1.00–1.07, p = 0.04). PTEN-expression did not correlate with OS/DFS/LRC. Better organoid response to everolimus correlated significantly to higher p-mTOR expression (Rs = − 0.731, p = 0.04). Conclusions: High p-mTOR expression predicts and high p-ERK expression tends to predict worse treatment outcome in HPV negative HNSCC patients treated with chemoradiation, providing additional evidence that these markers are candidate prognostic biomarkers for survival in this patient population. Also this study shows that the use of HNSCC organoids for biomarker research has potential. The role of PTEN expression as prognostic biomarker remains unclear, as consistent evidence on its prognostic and predictive value is lacking.
- 24 July 2023
- Published by Springer Nature in Head and Neck Pathology
- Vol. 17 (3) , 697-707
Open Accessp-mTOR, p-ERK and PTEN Expression in Tumor Biopsies and Organoids as Predictive Biomarkers for Patients with HPV Negative Head and Neck Cancer
Background: Survival rates of head and neck squamous cell carcinoma (HNSCC) have only marginally improved in the last decades. Hence there is a need for predictive biomarkers for long-time survival that can help to guide treatment decisions and might lead to the development of new therapies. The phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR signaling pathway is the most frequently altered pathway in HNSCC, genes are often mutated, amplificated and overexpressed causing aberrant signaling affecting cell growth and differentiation. Numerous genetic alterations of upstream and downstream factors have currently been clarified. However, their predictive value has yet to be established. Therefore we assess the predictive value of p-mTOR, p-ERK and PTEN expression. Methods: Tissue microarrays (TMA’s) of HPV-negative patients with oropharyngeal (n = 48), hypopharyngeal (n = 16) or laryngeal (n = 13) SCC, treated with primary chemoradiation (cisplatin/carboplatin/cetuximab and radiotherapy), were histologically stained for p-mTOR, PTEN and p-ERK. Expression was correlated to overall survival (OS), disease free survival (DFS) and locoregional control (LRC). Also p-mTOR was histologically stained in a separate cohort of HNSCC organoids (n = 8) and correlated to mTOR-inhibitor everolimus response. Results: High p-mTOR expression correlated significantly with worse OS in multivariate analysis in the whole patient cohort [Hazar Ratio (HR) 1.06, 95%CI 1.01–1.11, p = 0.03] and in the cisplatin/carboplatin group with both worse OS (HR 1.09, 95%CI 1.02–1.16, p = 0.02) and DFS (HR 1.06, 95%CI 1.00–1.12, p = 0,04). p-ERK expression correlated significantly with DFS in univariate analysis in the whole patient cohort (HR 1.03, 95%CI 1.00–1.05, p = 0.04) and cisplatin/carboplatin group (HR 1.03, 95%CI 1.00–1.07, p = 0.04). PTEN-expression did not correlate with OS/DFS/LRC. Better organoid response to everolimus correlated significantly to higher p-mTOR expression (Rs = − 0.731, p = 0.04). Conclusions: High p-mTOR expression predicts and high p-ERK expression tends to predict worse treatment outcome in HPV negative HNSCC patients treated with chemoradiation, providing additional evidence that these markers are candidate prognostic biomarkers for survival in this patient population. Also this study shows that the use of HNSCC organoids for biomarker research has potential. The role of PTEN expression as prognostic biomarker remains unclear, as consistent evidence on its prognostic and predictive value is lacking.
- Open Access
Tumor Markers and Their Prognostic Value in Sinonasal ITAC/Non-ITAC
One of the rare tumor entities present in the nose and paranasal sinuses is sinonasal (non-) intestinal-type adenocarcinoma (ITAC/non-ITAC). Currently, surgery with postoperative radiotherapy is the cornerstone of the treatment of these tumors. Systemic treatment is usually applied in a palliative setting. The prognosis of these tumors is very diverse. Biomarkers that may have prognostic value in these rare malignancies could help clinicians in decision-making. A systematic search of the literature was performed using the PubMed database. All studies investigating the prognostic significance of biomarkers in paranasal sinus ITAC/non-ITAC were retrieved. The findings were categorized within the hallmarks of cancer, to gain an understanding of the functions of possible prognostic biomarkers in the development of ITAC/non-ITAC. There were twenty-one studies reporting on twenty-one possible biomarkers included in the review. The expression of Mucin antigen sialosyl-Tn, C-erbB-2 oncoprotein, TIMP3 methylation, TP53, VEGF, ANXA2, MUC1 and the mucinous histological subtype were found to have a significant negative effect on survival. None of the biomarkers were found to have a positive effect on prognosis. The hallmarks ‘activating invasion and metastasis’ and ‘sustaining proliferative signaling’ seem to play the largest role in sinonasal (non-)ITAC. It could be concluded that there are multiple biomarkers foreboding a negative prognosis for ITAC/non-ITAC patients.
Open AccessTumor Markers and Their Prognostic Value in Sinonasal ITAC/Non-ITAC
One of the rare tumor entities present in the nose and paranasal sinuses is sinonasal (non-) intestinal-type adenocarcinoma (ITAC/non-ITAC). Currently, surgery with postoperative radiotherapy is the cornerstone of the treatment of these tumors. Systemic treatment is usually applied in a palliative setting. The prognosis of these tumors is very diverse. Biomarkers that may have prognostic value in these rare malignancies could help clinicians in decision-making. A systematic search of the literature was performed using the PubMed database. All studies investigating the prognostic significance of biomarkers in paranasal sinus ITAC/non-ITAC were retrieved. The findings were categorized within the hallmarks of cancer, to gain an understanding of the functions of possible prognostic biomarkers in the development of ITAC/non-ITAC. There were twenty-one studies reporting on twenty-one possible biomarkers included in the review. The expression of Mucin antigen sialosyl-Tn, C-erbB-2 oncoprotein, TIMP3 methylation, TP53, VEGF, ANXA2, MUC1 and the mucinous histological subtype were found to have a significant negative effect on survival. None of the biomarkers were found to have a positive effect on prognosis. The hallmarks ‘activating invasion and metastasis’ and ‘sustaining proliferative signaling’ seem to play the largest role in sinonasal (non-)ITAC. It could be concluded that there are multiple biomarkers foreboding a negative prognosis for ITAC/non-ITAC patients.
- Open Access
- 3 November 2022
- Published by Elsevier in Oral Oncology
- Vol. 135, 106227
The prognostic role of tumor associated macrophages in squamous cell carcinoma of the head and neck: A systematic review and meta-analysis
- 3 November 2022
- Published by Elsevier in Oral Oncology
- Vol. 135, 106227
Open AccessThe prognostic role of tumor associated macrophages in squamous cell carcinoma of the head and neck: A systematic review and meta-analysis