Cancer-associated fibroblasts in thyroid cancer stroma correlate with tumour progression and desmoplasia: a histopathological study Authors: Markos Christidis, Andreas C. Lazaris, Polyxeni Nicolopoulou-Stamati, Evangelia Baliou, Athanasios Zafeirakis, Georgios Vilaras, Georgia-Eleni Thomopoulou Published: 2026 ABSTRACT Aim and objectivesTumour stroma is acknowledged as a fundamental constituent of the tumour microenvironment, which drives cancer progression. Cancer-associated fibroblasts are an integral part of the cancer stroma. However, their role in thyroid cancer is not fully elucidated. The aim of this study is to determine the relationship between the expression of stromal cancer-associated fibroblasts’ immunohistochemical biomarkers in thyroid cancer and clinicopathological features of the disease.MethodsEighty-five formalin-fixed paraffin-embedded tissue sections of different thyroid cancer cases were stained with Masson’s trichrome, α-smooth muscle actin (α-SMA) and fibroblast activation protein (FAP) and scored based on the extent of the stained area and staining intensity. Statistical analysis was performed to assess any correlation with clinicopathological features.Resultsα-SMA showed a positive correlation with male gender (p = 0.02), tumour size greater than 1 cm (p = 0.0014) and histopathologic features of aggressiveness (p = 0.023). FAP was positively correlated with extent of desmoplasia (p = 0.024). Additionally, stromal desmoplasia was negatively correlated with tumour size greater than 1 cm (p = 0.0082).ConclusionCancer-associated fibroblasts are associated with increased tumour progression and desmoplasia in thyroid cancer.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s13044-026-00308-4. Introduction Thyroid cancer is the most common malignancy of the endocrine system and one of the most common malignancies worldwide, with the seventh highest incidence of all malignancies globally and with over 800.000 cases being diagnosed every year, costing the lives of 47.507 people annually [1]. Worldwide epidemiological data suggests that thyroid cancer rates are rising significantly, particularly in the young and female patients. In 2022, thyroid cancer was one of the 3 most diagnosed malignancies for females in 100 different countries, while at the same time accounting for the second most commonly diagnosed malignancy in female adolescents and young adult women [2]. Efforts to best elucidate the nature of thyroid cancer have centered on the tumour microenvironment (TME), which plays a key role in both the development and the pathogenesis of the disease. The TME, defined as the dynamic histological substrate on which the tumour grows, composed of both stromal and epithelial components, comprises a milieu of various different cells responsible for the complex immune interactions observed between primary regulators of the tumour immune response and the neoplastic cells, many of which are determinants of important clinical outcomes such as response to therapy and prognosis [3, 4]. Among the major constituents of the TME are the cancer-associated fibroblasts (CAFs). An integral component of the TME, CAFs are by themselves a heterogenous group of transformed fibroblasts originating from quiescent thyroid fibroblasts, with numerous identified subtypes, namely inflammatory CAFs, myoepithelial CAFs and antigen-presenting CAFs, each responsible for different functions. CAFs secrete growth factors, metabolites, chemokines and inflammatory cytokines such as Interleukin-6 and CXCL12, which help to establish a pro-tumorigenic niche by remodeling the tumour microenvironment [5, 6]. Moreover, they promote angiogenesis by the secretion of Vascular Endothelial Growth Factor (VEGF), aiding in increased metastatic capability of the tumour [6, 7]. Their immunomodulatory mechanisms facilitate tumour immune evasion, contributing to chemoresistance and increased aggressiveness [5, 8]. Of great clinical and scientific significance is the ability of CAFs to deposit collagen, namely Col-1, forming a dense fibrous stroma within the tumour microenvironment, a process known as desmoplasia [5]. Desmoplasia of the stromal compartment is a histologic feature seen in multiple different carcinomas and has recently become a subject of meticulous ongoing research, as it has been found to be an indicator of tumour aggressiveness and a predictive marker of lymph node metastasis in certain cancers, especially in PTC [5, 9–11]. In pancreatic ductal adenocarcinoma, which is considered as a hallmark of a highly desmoplastic tumour, there is an established association between desmoplasia and resistance to chemotherapy [12]. In addition, patients with pancreatic ductal adenocarcinoma exhibiting increased fibrosis have worse survival rates [13]. CAFs are characterized by the expression of numerous proteins, of which α-Smooth Muscle Actin (α-SMA) and Fibroblast Activation Protein (FAP) are the most widely studied in all cancers [14, 15]. Their presence in the stromal compartment of tumours has been greatly implicated in research regarding cancer pathogenesis and behaviour. Due to these emerging findings CAFs have become the target of novel anticancer therapies focused on tampering with their ability to differentiate into cancer-promoting cells [14]. Nevertheless, therapeutic outcomes of clinical trials using drugs that target CAFs have achieved poor results, with some even enhancing cancer progression, owing to our yet poor understanding of the nature of CAFs [16]. In thyroid cancer, CAFs and desmoplasia have been less well studied than in other malignancies. α-SMA and FAP are pan-CAF markers expressed in the peritumoral and intratumoral fibrous stroma of thyroid cancer and are not found in the thyroid gland under physiological circumstances [5]. However, their exact relationship with the parameters of the tumour itself is unknown. Hence, we investigated the correlation of CAFs in the desmoplastic stroma with the clinicopathological features of thyroid cancer by assessing the expression of their primary markers α-SMA and FAP in the tumour stroma. Methods and materials A total of 85 formalin-fixed paraffin embedded tissue sections of thyroid cancer cases were collected from the archive of the Pathology Department of the Army Share Fund Hospital. Inclusion criteria were the following: Any case diagnosed histopathologically as “papillary thyroid carcinoma”, “follicular carcinoma”, or “medullary carcinoma”, cases diagnosed from 2017 until 2024. Additionally, cases whose paraffin wax block was damaged or those which yielded poor-quality sections incompatible with diagnosis were also excluded from the study. Standard evaluation of parameters such as subtype of carcinoma, tumour size, margins, extrathyroidal extension and presence of capsule was performed on the H&E slide. In multifocal tumour cases, we analyzed the largest of the tumour foci. We adopted the definition of “aggressive PTCs” from the studies of You et al. [17] and Zhang et al. [18] as those exhibiting histopathological evidence of either: a) extrathyroidal invasion, b) central lymph node metastasis or c) aggressive histopathological subtypes of PTC e.g. tall cell variant or hobnail variant. Since in our study there were no histopathological subtypes of PTC associated with increased aggressiveness, we defined as “aggressive PTCs” those that had either of the two aforementioned findings. There was only microscopic extrathyroidal invasion as opposed to macroscopic extrathyroidal invasion in our cases. Certain cases diagnosed before 2018 that utilized the 7th edition of the AJCC classification system were reclassified using the 8th edition of the AJCC classification system, which came into effect on January 1st, 2018 so that all 85 cases used the same classification system for tumour staging. The degree of desmoplasia was assessed under light microscopy on slides stained with Masson’s Trichrome [Masson Trichrome Stain Kit (Methyl Blue) Masson 1929, Atom Scientific, Product Code RRSK20-100], for better visualization of the tumour stroma. Staining was performed according to the manufacturer’s instruction kit. The degree of desmoplasia was evaluated by a pathologist and converted into a percentage for each of the 85 cases separately. Immunohistochemistry For the detection of CAFs in the tumour stroma, representative slides from the 85 formalin-fixed paraffin embedded tissue sections were stained with two antibodies, α-SMA (monoclonal mouse anti-human smooth muscle actin, clone 1A4, code number M0851, Dako, dilution 1:50) and FAP (anti-fibroblast activation protein, alpha antibody, clone EPR20021, code number ab207178, Abcam, dilution 1:250). Staining was performed on the VENTANA automated slide stainer (BenchMark Ultra Immunohistochemistry System, Roche) using the OptiView DAB Immunohistochemistry Detection kit, according to the manufacturer’s instructions. Slides were then evaluated under light microscopy using a BIOMEDX Olympus microscope on 10x magnification. We applied the method described by Zhu et al. in their study [19] to calculate the immunoreactivity score of α-SMA and FAP for each of the 85 samples. The immunoreactivity score was calculated as follows: Five areas of the tumour were randomly selected and were appointed a numerical number from 0 to 3 based on the percentage of tumour area stained by the antibody, where 0: 0%, 1: 1–10%, 2: 11–50%, 3: 51–100%. The mean of the five values was then calculated. In instances where the tumour was small enough where less than 5 areas were examined, the sum was divided by the number of areas searched. A second number was assigned based on the intensity of the staining at the same areas, where 0: no intensity, 1: faint signal intensity, 2: moderate signal intensity, 3: strong signal intensity. The five numbers were summed together and divided by 5. These two scoring systems generated two numbers, that is the average of the staining percentage and staining intensity, ranging from 0 to 3, which were then multiplied together to give the final immunoreactivity score for each case. This was done independently for α-SMA and FAP. Figures and graphs were designed using the program Julius AI. Statistical analysis All numerical values were entered in regression analysis to assess for any possible correlations [20]; Pearson’s correlation coefficient (r) was quantified to find the strength of the associations between numerical values. Continuous variables with evident normal distribution were expressed as mean ±1 standard deviation (SD); categorical variables were expressed as numbers or proportions. Comparison between all pairs of numerical values was done by the paired t-test [21]. Frequency differences for categorical data comparisons among groups were tested by chi square analysis. We set the p value for statistical significance at <0.05 [22]. Results Clinicopathological characteristics Table 1 lists the clinicopathological parameters from all 85 cases included in this study. Age of the study population ranged from 17 to 77 years. Tumour size ranged from 1 mm to 53 mm. Forty-three tumors (50.6%) were ≤ 1 cm in size, all of which were papillary microcarcinomas, apart from one which was a medullary carcinoma. No special subtypes of papillary thyroid carcinoma were included in the study. Fifteen cases had evidence of extrathyroidal extension, all of which were only visible microscopically, either to perithyroidal fat or strap muscles. Lymph node metastasis was present in 7 out of 85 cases, four of which had concurrent extrathyroidal invasion. Lymph nodes with metastatic disease ranged from 1 to 13 in a single case. Mean extent of desmoplasia was 0.32 ± 0.21 (range = 0.1 to 0.98). Table 1Clinicopathological characteristics of sampled casesClinicopathological factorsTotal/Mean (±SD)Gender Male21 (24.7%) Female64 (75.3%)Age (years)49.8 ± 12.8Tumour size (mm)13.18 ± 9.5Tumour capsule Presence23 Absence62Number of tumour foci 132 226 318 44 54 6- 71Types of Thyroid Carcinoma Papillary81 Follicular1 Medullary3Aggressive tumours18Non-aggressive tumours67Extrathyroidal extension Presence15 Absence70Lymph Node Metastasis Presence7 Absence78Tumour Stage (TNM, AJCC, 8th edition) pT1a43 pT1b25 pT215 pT3a1 pT3b1Underlying thyroid pathology Goiter27 Autoimmune thyroiditis17 Lymphocytic thyroiditis7 Lymphocytic thyroiditis + goiter5 Autoimmune thyroiditis + goiter4Desmoplasia (%)32α-SMA immunoreactivity score3.02 ± 2.26FAP immunoreactivity score1.66 ± 2.39 Table 1Clinicopathological characteristics of sampled casesClinicopathological factorsTotal/Mean (±SD)Gender Male21 (24.7%) Female64 (75.3%)Age (years)49.8 ± 12.8Tumour size (mm)13.18 ± 9.5Tumour capsule Presence23 Absence62Number of tumour foci 132 226 318 44 54 6- 71Types of Thyroid Carcinoma Papillary81 Follicular1 Medullary3Aggressive tumours18Non-aggressive tumours67Extrathyroidal extension Presence15 Absence70Lymph Node Metastasis Presence7 Absence78Tumour Stage (TNM, AJCC, 8th edition) pT1a43 pT1b25 pT215 pT3a1 pT3b1Underlying thyroid pathology Goiter27 Autoimmune thyroiditis17 Lymphocytic thyroiditis7 Lymphocytic thyroiditis + goiter5 Autoimmune thyroiditis + goiter4Desmoplasia (%)32α-SMA immunoreactivity score3.02 ± 2.26FAP immunoreactivity score1.66 ± 2.39 Clinicopathological characteristics of sampled cases α-SMA and FAP staining patterns in CAFs We identified CAFs as those cells expressing either α-SMA, FAP, or both within the peritumoural or intratumoural fibrous stroma. α-SMA and FAP exhibited moderate differences in their staining patterns regarding CAFs. While both α-SMA and FAP stained the cytoplasm of CAFs present within the peritumoral or the intratumoural stroma, albeit to varying degrees, CAFs were more often positive for α-SMA rather than for FAP. CAFs stained more intensely for α-SMA than for FAP. The mean immunoreactivity scores for α-SMA and FAP were 3.02 ± 2.26 and 1.66 ± 2.39, respectively (both ranges = 0 to 9). In addition to CAFs, α-SMA also stained smooth muscle fibers, particularly those found on the vessel walls, as well as smooth muscle cells of the extrathyroidal soft tissues, serving as a good internal control. As for FAP, apart from red blood cells in the vessel lumina, it did not stain any other normal histological structure, which was in accordance to the manufacturer’s guidelines. The number of cases which stained negative for α-SMA was one (1.17%), whereas those that did not stain for FAP were 33 (38.8%). Figure 1 displays different staining patterns of both α-SMA and FAP based on immunoreactivity scores. Fig. 1Immunohistochemical staining patterns of α-SMA and FAP in PTC. (A – C): α-SMA; (D – F): FAP; (A) Low α-SMA immunoreactivity score with limited staining area and moderate intensity; (B) Moderate α-SMA immunoreactivity score with both intermediate staining area and staining intensity; (C) High α-SMA immunoreactivity scores such as in this case, exhibited increased area of staining as well as strong staining intensity; (D) Low FAP immunoreactivity scores as seen in this case had limited area of staining as well as low staining intensity; (E) Moderate FAP immunoreactivity score. CAFs demonstrate intermediate area of staining and marked staining intensity; (F) Cases with markedly increased staining area and pronounced staining intensity such as this one, had high FAP immunoreactivity scores. Magnification: 40x Immunohistochemical staining patterns of α-SMA and FAP in PTC. (A – C): α-SMA; (D – F): FAP; (A) Low α-SMA immunoreactivity score with limited staining area and moderate intensity; (B) Moderate α-SMA immunoreactivity score with both intermediate staining area and staining intensity; (C) High α-SMA immunoreactivity scores such as in this case, exhibited increased area of staining as well as strong staining intensity; (D) Low FAP immunoreactivity scores as seen in this case had limited area of staining as well as low staining intensity; (E) Moderate FAP immunoreactivity score. CAFs demonstrate intermediate area of staining and marked staining intensity; (F) Cases with markedly increased staining area and pronounced staining intensity such as this one, had high FAP immunoreactivity scores. Magnification: 40x Correlations of α-SMA with clinicopathological factors Tables 2 and 3 summarize all correlations between α-SMA, FAP and their respective clinicopathological factors. Table 2Correlations of α-SMA and FAP with categorical clinicopathological factorsClinicopathological factors (n)α-SMAFAPMeant statisticp valueMeant statisticp valueGenderMale (21)4 ± 2.43−2.3690.02*2.5 ± 0.31−0.2320.84Female (64)2.7 ± 2.122.07 ± 0.45Age>50 (42)3.12 ± 2.40−0.4190.681.57 ± 2.310.3550.72≤50 (43)2.92 ± 2.142.49 ± 0.38Tumour capsulePresence (23)2.92 ± 20.2380.812.1 ± 2.84−1.0220.31Absence (62)3.06 ± 2.361.51 ± 2.21FociUnifocal (32)3.45 ± 2.071.3750.1731.99 ± 2.730.9760.33Multifocal (53)2.76 ± 2.351.47 ± 2.17Underlying thyroid pathologyLymphocytic/Hashimoto thyroiditis (33)2.64 ± 2.15−1.2360.221.86 ± 2.420.5860.56Goiter/no underlying pathology (52)3.26 ± 2.311.67 ± 2.39*indicates statistically significant valueTable 3Correlations of α-SMA and FAP with numerical clinicopathological factorsHistopathologic parametersMeanα-SMAFAPr coefficientp valuer coefficientp valueDesmoplasia0.32 ± 0.210.0840.4450.2430.024*α-SMA immunoreactivity score3.02 ± 2.25--0.1190.27*indicates statistically significant value Table 2Correlations of α-SMA and FAP with categorical clinicopathological factorsClinicopathological factors (n)α-SMAFAPMeant statisticp valueMeant statisticp valueGenderMale (21)4 ± 2.43−2.3690.02*2.5 ± 0.31−0.2320.84Female (64)2.7 ± 2.122.07 ± 0.45Age>50 (42)3.12 ± 2.40−0.4190.681.57 ± 2.310.3550.72≤50 (43)2.92 ± 2.142.49 ± 0.38Tumour capsulePresence (23)2.92 ± 20.2380.812.1 ± 2.84−1.0220.31Absence (62)3.06 ± 2.361.51 ± 2.21FociUnifocal (32)3.45 ± 2.071.3750.1731.99 ± 2.730.9760.33Multifocal (53)2.76 ± 2.351.47 ± 2.17Underlying thyroid pathologyLymphocytic/Hashimoto thyroiditis (33)2.64 ± 2.15−1.2360.221.86 ± 2.420.5860.56Goiter/no underlying pathology (52)3.26 ± 2.311.67 ± 2.39*indicates statistically significant value Correlations of α-SMA and FAP with categorical clinicopathological factors *indicates statistically significant value Table 3Correlations of α-SMA and FAP with numerical clinicopathological factorsHistopathologic parametersMeanα-SMAFAPr coefficientp valuer coefficientp valueDesmoplasia0.32 ± 0.210.0840.4450.2430.024*α-SMA immunoreactivity score3.02 ± 2.25--0.1190.27*indicates statistically significant value Correlations of α-SMA and FAP with numerical clinicopathological factors *indicates statistically significant value α-SMA immunoreactivity score was correlated with tumour size. α-SMA showed both increased staining intensity and increased area of staining in tumours greater than 1 cm in diameter than those of 1 cm or less. As shown in Table 4, tumours greater than 1 cm expressed a higher α-SMA immunoreactivity score than those of 1 cm in size or less. This was shown to be statistically significant (p = 0.0014). Figure 2 shows a comparison between the distribution of each of the two tumour size groups. Figure 3 shows the α-SMA immunohistochemical staining pattern of two tumours of different size diameter. Table 4Correlation of tumour size with histopathologic parametersHistopathologic parameterstumour size ≤ 10 mm (43)tumour size > 10 mm (42)t valuep valueα-SMA Immunoreactivity score (mean)2.26 ± 1.913.79 ± 2.34−3.3020.0014*FAP Immunoreactivity score (mean)1.58 ± 2.321.75 ± 2.48−0.3320.7Desmoplasia (%)38 ± 1925 ± 202.7140.0082**indicates statistically significant valueFig. 2Empirical cumulative distributions function (ECDF) comparing directly each distribution of α-SMA immunoreactivity score per tumour sizeFig. 3α-SMA staining in tumours of different size groups. (A) In this tumour measuring 4 mm, CAFs showed mild staining area and staining intensity; (B and C) in tumours measuring greater than 10 mm, CAFs staining occupied an increased area (B) as well as increased staining intensity (C), thus contributing to higher α-SMA immunoreactivity scores overall. These two tumours measured 53 and 28 mm, respectively. Magnification: 40x Table 4Correlation of tumour size with histopathologic parametersHistopathologic parameterstumour size ≤ 10 mm (43)tumour size > 10 mm (42)t valuep valueα-SMA Immunoreactivity score (mean)2.26 ± 1.913.79 ± 2.34−3.3020.0014*FAP Immunoreactivity score (mean)1.58 ± 2.321.75 ± 2.48−0.3320.7Desmoplasia (%)38 ± 1925 ± 202.7140.0082**indicates statistically significant value Correlation of tumour size with histopathologic parameters *indicates statistically significant value Empirical cumulative distributions function (ECDF) comparing directly each distribution of α-SMA immunoreactivity score per tumour size α-SMA staining in tumours of different size groups. (A) In this tumour measuring 4 mm, CAFs showed mild staining area and staining intensity; (B and C) in tumours measuring greater than 10 mm, CAFs staining occupied an increased area (B) as well as increased staining intensity (C), thus contributing to higher α-SMA immunoreactivity scores overall. These two tumours measured 53 and 28 mm, respectively. Magnification: 40x α-SMA immunoreactivity score was correlated with PTC aggressiveness. Overall, aggressive PTCs had statistically higher values of α-SMA immunoreactivity scores, demonstrating stronger intensity and greater staining area compared to those which did not show any histological features of increased aggressiveness (p = 0.023) (Table 5). Figure 4 illustrates the comparative distribution of α-SMA immunoreactivity scores between the aggressive and non-aggressive tumour groups. Table 5Correlation of tumour aggressiveness with pathological parametersPathological parametersAggressive tumours (18)Non-Aggressive tumours (67)t valuep valueα-SMA Immunoreactivity score (mean)4.12 ± 2.172.73 ± 2.20−2.3810.023*FAP Immunoreactivity score (mean)1.79 ± 2.201.63 ± 2.45−0.2360.81Desmoplasia (%)35% ± 22%31% ± 21%−0.7050.48*indicates statistically significant valueFig. 4Comparative distribution of α-SMA immunoreactivity scores between the aggressive and non-aggressive tumour groups Table 5Correlation of tumour aggressiveness with pathological parametersPathological parametersAggressive tumours (18)Non-Aggressive tumours (67)t valuep valueα-SMA Immunoreactivity score (mean)4.12 ± 2.172.73 ± 2.20−2.3810.023*FAP Immunoreactivity score (mean)1.79 ± 2.201.63 ± 2.45−0.2360.81Desmoplasia (%)35% ± 22%31% ± 21%−0.7050.48*indicates statistically significant value Correlation of tumour aggressiveness with pathological parameters *indicates statistically significant value Comparative distribution of α-SMA immunoreactivity scores between the aggressive and non-aggressive tumour groups There was a statistically significant correlation between α-SMA and gender. The α-SMA immunoreactivity scores were higher in males than in females (p = 0.02). No statistically significant correlations were found with age, presence of tumour capsule, number of tumour foci, underlying thyroid pathology, FAP immunoreactivity scores or desmoplasia. Correlations of FAP with clinicopathological factors FAP was positively correlated with degree of desmoplasia per tumour area. Tumours with higher percentage of desmoplasia had higher FAP immunoreactivity scores than those with lower percentage of desmoplasia (Table 3). This finding was statistically significant (p = 0.024). There were no statistically significant correlations between FAP and age, gender, the presence of fibrous capsule, number of foci or underlying thyroid pathology. When compared with tumour aggressiveness, mean FAP showed similar immunoreactivity scores between aggressive and non-aggressive histological groups (p = 0.81). FAP did not show any correlation with tumour size or α-SMA. When comparing FAP+ to FAP- CAFs, no association was observed with tumour size, aggressiveness, presence of capsule, underlying thyroid pathology, number of foci, desmoplasia or α-SMA immunoreactivity score. Correlation of desmoplasia with clinicopathological features Tumour size was negatively correlated with degree of desmoplastic reaction. Tumours of 1 cm or less in diameter exhibited a higher degree of desmoplastic reaction per tumour area than tumours greater than 1 cm in diameter (Table 4). This difference in desmoplasia between the two tumour size groups was a statistically significant finding (p = 0.0082). Figure 5 demonstrates the mean percentage of desmoplastic reaction per tumour area of each tumour size group, while Fig. 6 shows a comparison of the degree of desmoplasia between two tumours of different size diameter, as seen on Masson’s Trichrome stain. There was no association between desmoplasia and gender, age, presence of tumour capsule, number of foci, underlying thyroid pathology or tumour aggressiveness. Fig. 5Comparison of the mean value of the area of desmoplastic reaction (%) of the two tumour size groups. On the left side, tumours up to 10 mm in diameter exhibited a higher mean value of desmoplasia in their stroma compared to those on the right measuring greater than 10 mm in diameterFig. 6Desmoplasia as visualized under light microscopy with H&E and Masson’s trichrome staining of 4 cases of PTCs of different size groups. Figures a, c, e and g demonstrate four different cases of PTC stained in H&E, while Fig. 6b, d, f and h show the same cases, respectively, stained with Masson’s trichrome. In H&E the desmoplastic stroma is evident as the paucicellular eosinophilic area intersecting the neoplastic thyroid follicles. In Masson’s trichrome desmoplasia is visualized in blue colour, while thyroid follicular cells and tumour cells are stained in pink. Colloid retains its bright eosinophilic colour as in H&E staining. (a and b) PTC with <50% of the tumour comprised of a desmoplastic stroma. The desmoplastic stroma is seen here forming intersecting bands and fascicles between groups of tumour cells, compartmentalizing them in the process. In this case, the tumour was 53 mm in diameter. (c and d) PTC with a diameter of 21 mm exhibiting scant desmoplasia. In this case, the desmoplastic reaction was calculated at 10% relative to the tumour size, with neoplastic cells forming the greater part of the tumour. (e and f) markedly increased desmoplastic reaction relative to the neoplastic cells is seen here in this 9 mm size tumour. g and h) almost complete effacement of the tumour architecture by the desmoplastic reaction is observed here, with only minimal isolated neoplastic cells found within the stroma. This was a 6 mm size PTC with 80% of the tumour exhibiting desmoplastic reaction. Magnification: 10x Comparison of the mean value of the area of desmoplastic reaction (%) of the two tumour size groups. On the left side, tumours up to 10 mm in diameter exhibited a higher mean value of desmoplasia in their stroma compared to those on the right measuring greater than 10 mm in diameter Desmoplasia as visualized under light microscopy with H&E and Masson’s trichrome staining of 4 cases of PTCs of different size groups. Figures a, c, e and g demonstrate four different cases of PTC stained in H&E, while Fig. 6b, d, f and h show the same cases, respectively, stained with Masson’s trichrome. In H&E the desmoplastic stroma is evident as the paucicellular eosinophilic area intersecting the neoplastic thyroid follicles. In Masson’s trichrome desmoplasia is visualized in blue colour, while thyroid follicular cells and tumour cells are stained in pink. Colloid retains its bright eosinophilic colour as in H&E staining. (a and b) PTC with <50% of the tumour comprised of a desmoplastic stroma. The desmoplastic stroma is seen here forming intersecting bands and fascicles between groups of tumour cells, compartmentalizing them in the process. In this case, the tumour was 53 mm in diameter. (c and d) PTC with a diameter of 21 mm exhibiting scant desmoplasia. In this case, the desmoplastic reaction was calculated at 10% relative to the tumour size, with neoplastic cells forming the greater part of the tumour. (e and f) markedly increased desmoplastic reaction relative to the neoplastic cells is seen here in this 9 mm size tumour. g and h) almost complete effacement of the tumour architecture by the desmoplastic reaction is observed here, with only minimal isolated neoplastic cells found within the stroma. This was a 6 mm size PTC with 80% of the tumour exhibiting desmoplastic reaction. Magnification: 10x Discussion In this study, we sought any possible link between CAFs and clinical or histological parameters in thyroid cancer. We showed that tumours larger than 1 cm in diameter had on average higher α-SMA immunoreactivity scores than thyroid microcarcinomas. This finding is corroborated by other existing studies. Zhu et al. found that α-SMA expression in CAFs was positively correlated with tumour size in PTC [19]. In a similar study done in pancreatic tissue, Winter et al. found that pancreatic ductal adenocarcinomas greater than 3 cm in diameter had higher α-SMA expression than smaller sized tumours [23]. Chuaysri et al. reported that cholangiocarcinomas with larger tumour size had significantly increased α-SMA expression of CAFs [24]. The study of Sinn et al., however, did not establish any correlation between α-SMA expression in CAFs on pancreatic cancer stroma and tumour size [25]. Our decision to group thyroid cancer cases using 1 cm as a cut-off value for tumour size diameter is not arbitrary. According to the 8th and current edition of the TNM classification system, thyroid cancers of 1 cm in diameter or less are designated as pT1a and are termed “microcarcinomas”, whereas those larger than 1 cm in diameter are designated as pT1b or higher. It is therefore evident that such a difference in tumour size is decisive in classifying a thyroid carcinoma according to the TNM classification system as they reflect the clinical behaviour of the tumour. Accordingly, some authors consider tumour size in thyroid cancer, as incorporated into the TNM classification to be an indicator of prognostic significance [26]. In their 2018 article, Kim et al. argue that the TNM classification system is useful in predicting disease-specific survival of patients with PTC [27]. PTCs greater than 1 cm have higher rates of extrathyroidal and lymphovascular invasion than papillary microcarcinomas, as well as higher rates of lymph node metastasis [28]. Regarding papillary thyroid microcarcinomas, Lucandri et al. raise the issue in their 2024 article that papillary thyroid microcarcinomas may not actually have any impact whatsoever on survival and are indolent enough as to merit active surveillance as opposed to total thyroidectomy [29]. This is relevant, as stromal CAFs may be implicated in their behaviour as slowly advancing or even non-progressive tumours. In our study, aggressive PTCs also showed higher α-SMA immunoreactivity scores than non-aggressive tumours. This is in line with previously published results. Zhu et al. confirmed in their own study that α-SMA is overexpressed in CAFs of PTCs with extrathyroidal invasion and lymph node metastasis [19]. They also showed that PTCs harbouring BRAF mutations had increased expression of α-SMA [19]. It is well established in the medical literature that extrathyroidal invasion, lymph node metastasis and/or BRAF mutation in PTC are associated with increased aggressiveness and poorer survival [7, 30, 31]. In fact, CAFs exert an important influence on the metastatic potential of PTC cells. This is most evident in the 2024 study by Wang et al., who showed that patients suffering from PTC who had lymph node metastasis had significantly higher density of fibrosis at the tumour invasive front [11]. Furthermore, there was increased α-SMA immunostaining at the invasive fibrotic front of such tumours, affirming that PTCs with lymph node metastasis have a higher composition of stromal CAFs than those without lymph node metastasis [11]. It is known that high stromal density of CAFs is associated with worse prognosis in some cancers because it is linked to increased tumour resistance to therapy [32]. An important finding that further adds to the fact that CAFs are associated with PTC aggressiveness is that dedifferentiated thyroid carcinoma, namely anaplastic thyroid carcinoma, demonstrates consistently higher α-SMA immunoreactivity scores than differentiated thyroid carcinoma [7]. The strong correlation observed in numerous studies between CAFS and PTC lymph node metastasis is reflected by how some authors have suggested that CAFs may be used to predict metastasis of PTC in cervical lymph nodes [33]. This observation seems to be validated by studies done in other solid tumours as well. In mismatch repair-proficient colorectal carcinomas there was a clear association between CAF density in the tumour periphery and higher N stages, as evidenced by the increased α-SMA immunoreactivity scores at the said area [34]. These patients also had increased risk of recurrence. In non-small cell lung carcinoma, α-SMA+ CAFs are associated with higher N stage and more advanced tumour stage overall [35]. The spatial distribution of CAFs within the tumour stroma also seems to be a contributing factor in cancer aggressiveness, as evidenced by the study of Bo et al., who observed that in patients with pancreatic ductal adenocarcinoma, CAFs that were distributed mostly in the peripheral stroma of the tumour conferred a higher probability of lymph node metastasis, poorer survival rates and were associated with higher TNM stages [36]. CAFs contribute in increased tumour aggressiveness in a number of ways. Their main product of synthesis Col-1 is a type 1 collagen that augments the motility of PTC cells and increases cancer progression in PTCs harbouring BRAFv600E mutation and loss of PTEN [37]. Col-1 is also found in abundance in poorly differentiated thyroid cancers and their levels are related to poorer survival rates in PTC patients [37]. Inflammatory CAFs secrete CXCL12, a pro-inflammatory cytokine that binds to CXCR4, activating ERK and AKT signaling pathways, which in turn induce the synthesis of matrix metalloproteinases (MMPs). MMPs degrade the extracellular matrix of the surrounding tissues, remodel the TME, promote angiogenesis and collagen synthesis, all of which enhance tumour proliferation and invasiveness [38, 39]. MMPs are also responsible for creating a dense and fibrotic extracellular matrix within the tumour stroma that forms a barrier against chemotherapeutic agents, rendering cancer cells resistant to therapy [38, 39]. Recently, it was found that pancreatic neuroendocrine tumour cells neighbouring α-SMA-expressing stromal CAFs have upregulation of the MMP9 gene [40]. Furthermore, CAFs impair immune surveillance through the secretion of immunosuppressing cytokines TGF-β, IL-6, IL-10, IL-35 and by inhibiting the antigen presenting abilities of macrophages and dendritic cells [38]. B7–H3, also known as CD276, is an important transmembrane protein expressed by CAFs that has been reported to amplify the synthesis of pro-tumorigenic proteins CAIX and ILK, which drive cancer progression and metastasis in the most aggressive types of thyroid cancer [41]. A notable finding in our study was that microcarcinomas exhibited higher percentages of desmoplasia per tumour area than those greater than 1 cm in diameter. While it may seem contradictory that larger tumours are less desmoplastic than smaller ones, there is evidence to believe that higher degree of desmoplasia may not in fact be a characteristic of more advanced tumours. In one study, depletion of α-SMA+ myofibroblastic CAFs in the stroma of ductal adenocarcinoma of the pancreas, which secrete type 1 collagen, a primary constituent of the desmoplastic stroma, resulted in increased aggressiveness and decreased survival, indicating, according to the authors, that desmoplasia may confer a protective effect rather than sustaining tumour progression [42]. In similar fashion, when scientists deleted Sonic Hedgehog protein from pancreatic ductal adenocarcinoma cells, the tumour desmoplastic stroma was reduced. This, in turn, rendered the cancer cells more aggressive, with increased risk of metastasis and poorer survival [43]. In fact, one study showed that highly desmoplastic pancreatic ductal adenocarcinomas are associated with better overall survival [10]. Comparable results have also been observed in colorectal and liver cancer metastasis [44, 45]. Nonetheless, in PTC, low fibrosis scores correlate with improved survival rates and prognosis [11, 46]. In our study, aggressive carcinomas did indeed display a slightly higher average of desmoplasia percentage compared to non-aggressive carcinomas, albeit the difference was not statistically significant. While tumour size or stage is not an indicator of aggressiveness per se, it is important to recognize that the stroma of larger-sized tumours is more abundant in α-SMA+ CAFs, which, owing to their complex heterogeneity, can be both supportive of the TME or protective against cancer progression. Increased α-SMA expression of CAFs in thyroid cancer in males is not an unexpected finding, as men tend to have more advanced disease at the time of diagnosis, with larger size, higher rates of extrathyroidal invasion, lymph node metastasis and worse prognosis than women overall [47]. FAP is one of the most widely used biomarkers employed in the study CAFs. It is a transmembrane protein with an extracellular and an intracellular domain, the former of which has catalytic properties owing to a triad of serine proteases. FAP is found on the cell surface of certain fibroblasts, particularly those involved in wound healing, inflammation, embryogenesis and, under pathological conditions, in fibrotic diseases such as liver cirrhosis or idiopathic pulmonary fibrosis [15]. What makes it unique is that it is never expressed in thyroid tissue under physiological circumstances. Its rich expression and abundance in the tumour stroma has made it an ideal target of multiple novel therapeutic drugs, which are being studied in ongoing clinical trials [15, 48]. FAP’s role in accelerating cancer progression and altering its biological behavior is well documented. By virtue of its enzymatic function, FAP acts through the PTEN/PI3K/AKT and RAS-ERK pathways to enhance tumour cell invasion. Moreover, it promotes angiogenesis and neovascularization by remodeling the extracellular matrix. This is achieved in a number of ways. For instance, FAP+ CAFs secrete cytokines and chemokines, namely CXCL12, which attracts stromal cells in the tumour site, forming a barrier that prevents entry of CD8+ T cytotoxic cells and target pharmaceutical agents. FAP+ CAF-secreted cytokines interact with T regulatory cells and effector T cells, while also interfering with antigen presentation by macrophages and dendritic cells, altering their phenotype in favour of tumour growth, all of which help establish an immunosuppressive niche, further contributing to a pro-tumorigenic environment [48, 49]. An important mechanism by which FAP+ CAFs contribute to tumour immune evasion is by oversecretion of PD-L1, which binds to PD-1 on T lymphocytes, inhibiting T cell proliferation, resulting in immunosuppression [48]. FAP has been associated with increased cancer aggressiveness and poorer clinicopathological outcomes in pancreatic adenocarcinoma, extrahepatic cholangiocarcinoma, ovarian cancer and thyroid cancer, whereas the opposite has been observed in squamous cell carcinoma of the lung [19, 50–54]. In our study we established neither such an association, as aggressive and non-aggressive differentiated thyroid malignancies had insignificant differences in FAP immunoreactivity scores. FAP and α-SMA also did not exhibit any association. Klivaer et al., who reported similar results, interpret this as having a co-existence of different subtypes of CAFs within the desmoplastic stroma, with each subtype expressing a different biomarker [51]. We did, however, observe a statistically significant correlation between FAP and tumour desmoplasia. This finding is identical to that of Koperek et al., who studied the expression of FAP and Tenascin C in CAFs located in the fibrous stroma of medullary thyroid carcinoma. The authors found that both FAP and Tenascin C were more abundantly expressed in highly desmoplastic carcinomas [55]. The importance of FAP as a causative agent of desmoplasia was further corroborated by Lo et al. in their 2015 paper. In it, the authors proved that the extent of desmoplasia of different tumours was dependent on the presence of FAP+ cancer-associated stromal cells, which were primarily responsible for the secretion of hyaluronan, a pivotal constituent of the tumour stroma. When these cells were depleted, stromal desmoplasia significantly decreased [56]. While it is possible that FAP+ CAFs are the main drivers of stromal desmoplasia in thyroid cancer, such a postulate may be an oversimplification of their role, as FAP+ CAFs have been shown in multiple studies to resolve fibrosis in the same tissues where they actively synthesize the fibrotic meshwork at a different point in time [49]. Furthermore, modern therapeutic agents aimed at eliminating FAP, namely FAP inhibitors, have proven to have adverse clinical outcomes, with increased tumour aggressiveness and shorter survival [16, 48, 49]. Based on the existing evidence, it is safe to assert that FAP and desmoplasia are strongly, yet intricately interrelated, as it can either induce and maintain desmoplasia or resolve it, depending on the reciprocal influence of the TME at a given point in time. This study is not without its limitations, the biggest one being its small sample size. Sample size is an important determinant of the precision and accuracy of the data. Small sample sizes may reduce the statistical power whereas large sample sizes increase the reliability of the results and improve the quality of the data. Given the relatively high frequency of thyroid cancer in the general population, in particular PTC, a sample size of 81 such cases could be considered as small. However, it should be acknowledged that in the aforementioned cases we were able to gather available clinicopathological data, as well as to secure sufficient tissue material for processing. Nevertheless, future studies investigating the relationship between CAFs and thyroid cancer progression should aim for larger sample size to increase precision of their findings. Additionally, the diversity of the sample size is not homogenous, with PTC cases vastly outnumbering follicular and medullary carcinomas. This is not unexpected, as PTC is by far the most common type of thyroid cancer in surgical pathology practice. As such, a conclusion on the relationship of CAFs and each histological subtype of thyroid carcinoma could not be reached in this study. Another limitation in this study is the fact that we did not evaluate the spatial distribution of CAFs separately for each immunostain in the intratumoral or peritumoral compartment as was done in other studies. We intend to address this shortcoming in our future studies by including additional CAF immune biomarkers, namely CXCL12 and Periostin, which are non-specific biomarkers expressed in two major subsets of CAFs, inflammatory and myofibroblastic CAFs, located primarily in the peritumoral and intratumoural compartment, respectively [57]. We hope that this will increase both specificity and sensitivity of our methodology, enabling a safer approach in analyzing the spatial distribution of different CAF subsets in thyroid cancer and the potential elucidation of their role in cancer progression, as well as the significance of their heterogeneity in thyroid cancer. Furthermore, we did not collect follow-up data from the patients; hence we could not assess for survival and recurrence rates. Analysis of immunohistochemical staining of our biomarkers according to tumour size based on TNM classification was not possible, since the distribution of cases was unequal, owing to the downgrading of a large proportion of our pT3 cases due to the implementation of the 8th edition. Conclusion CAFs possess the ability to exert separate and at times opposing influences on thyroid cancer homeostasis. The biomarkers that were used in this study highlight the existence of two functionally different CAF subpopulations within the tumour stroma of thyroid cancer, each possessing distinct properties. α-SMA predominant CAFs appear to be associated with larger tumour size, increased tumour aggressiveness and male gender and are more closely implicated with thyroid cancer progression, while FAP predominant CAFs seem to be more closely associated with higher degree of desmoplasia, affecting the composition of the tumour stroma and consequently the tumour microenvironment. Current limitations of this study do not allow for a definite conclusion to be drawn regarding CAF heterogeneity in thyroid cancer stroma and its significance in thyroid cancer progression. Future studies should focus on exploring functional CAF heterogeneity employing the use of additional biomarkers that further identify key subpopulations of CAFs and their association with clinicopathological parameters. The findings of this study indicate that CAFs in the desmoplastic stroma of thyroid cancer may influence tumour progression and desmoplasia. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1