Prognostic Value of Cancer-Associated Fibroblast Marker Expression in the Intratumoral and Marginal Areas of Soft Tissue Sarcoma Open Access

Recent years have seen the acquisition of more knowledge about the microenvironment of various malignant tumors, as well as tumor cells themselves. The tumor microenvironment (TME) comprises a variety of cells, including fibroblasts and neovascular cells, as well as immune system cells such as macrophages. Among these, fibroblasts and macrophages are referred to as cancer-associated fibroblasts (CAFs) and tumor-associated macrophages (TAMs), respectively. Although many studies report various roles for CAFs and TAMs, most suggest that they play a tumor-promoting function. In vitro studies show that CAFs facilitate tumor progression by secreting soluble factors such as TGF-β1, TGF-β2, and exosomes, as well as suppressing anti-tumor immune responses by modulating the extracellular matrix [1‒3]. TAMs play a pro-tumoral role by forming a cancer-promoting inflammatory microenvironment and suppressing anti-tumor immune responses, and by promoting tumor invasion, angiogenesis, and metastasis [4‒6]. The abundance of CAFs or TAMs in the intratumoral area is associated with the clinical outcome of various cancers [7‒17]. Both CAFs and TAMs are considered major components of the TME, and both play a crucial role in tumor progression; importantly, each cell type is thought to interact with the other [18, 19].

The oncological significance of the TME may be particularly important in the context of sarcomas; indeed, this is borne out by several reports that examined the TME of sarcomas [20‒26]. Therefore, it is important to elucidate the function of the TME in sarcoma if we are to develop viable treatment approaches. A critical problem arises when assessing the TME of sarcomas: how do we distinguish sarcoma cells from non-neoplastic fibroblasts and macrophages? In general, cancer cells adhere to each other to form an epithelial nest resembling a gland or stratified squamous epithelium; therefore, it is not difficult to distinguish cancer nests from surrounding fibroblasts or macrophages. However, sarcomas do not usually form epithelioid nests, and often present with a mesenchymal appearance. These characteristics make it difficult to distinguish sarcoma cells from non-neoplastic fibroblasts or macrophages. In the case of macrophages, it may be possible to distinguish them from sarcoma cells to some extent by performing immunohistochemical (IHC) staining with relatively specific antibodies; indeed, most studies examining TAMs in soft tissue sarcomas use antibodies such as anti-CD163 to distinguish TAM from sarcoma cells [20‒26]. However, it is very difficult to distinguish CAFs from sarcoma cells, both histopathologically and immunohistochemically, because antibodies that react with alpha-smooth muscle actin (αSMA) and vimentin (which are considered typical CAF markers) also react with many sarcoma subtypes. Toda et al. [27] evaluated desmoplastic reactions in H and E sections from synovial sarcoma cases; however, very few papers have examined CAFs in soft tissue sarcoma. Previously, we reported that the number of CD163-/CD204-positive macrophages in the marginal area (i.e., non-sarcoma area) of soft tissue sarcomas of FNCLCC Grade 2/3 (grading system for soft tissue sarcomas proposed by the Fédération Nationale des Centres de Lutte Contre le Cancer) may be a better prognostic factor than intratumoral macrophages [26]. We considered that evaluating macrophage numbers in the marginal area may be more appropriate because sarcoma cells may stain positive for TAM markers and that a high density of TAMs in intratumoral area may not facilitate precise counts. Therefore, we were inspired to assess CAFs, which reside in the marginal area (non-sarcoma area) as well as the intratumoral area of the sarcoma.

In this study, we used fibroblast-activating protein (FAP), CD10, and podoplanin, rather than αSMA and vimentin, as CAF markers because many subtypes of sarcoma are immunohistochemically positive for αSMA and vimentin as described earlier. FAP, a member of the serine protease family of membrane proteins [28, 29], is recognized as a representative CAF marker in many studies [30‒33]. Indeed, 68Ga-FAPI positron emission tomography/computed tomography (which is based on FAP-specific inhibitors) labels CAFs and is useful for detecting various cancers with high specificity [34]. In terms of diagnostic pathology, although some histological subtypes of sarcoma are FAP-positive, FAP is not a specific marker of certain types of sarcoma [35]. CD10 is widely accepted as a positive marker for germinal center B cells and the endometrial stroma. It is often used for the diagnosis of lymphomas and endometrial stromal sarcomas, but less commonly for the diagnosis of soft tissue sarcomas. Regarding the assessment of the TME, CD10 (as well as FAP) is considered a CAF marker [9, 32, 36‒38]. A study reported that CD10-positive CAFs promote cancer formation and chemoresistance [39]. Podoplanin is a well-known marker of the lymphatic endothelium; however, it has also been evaluated in several studies as a CAF marker [8, 28, 40‒42]. In several tumors, associations of podoplanin-positive CAFs with a larger tumor size, lymphatic invasion, and the number of CD204-positive macrophages (i.e., TAMs) have been reported [43, 44].

Some sarcoma cases displayed an increase of fibroblasts around the tumor that were distinctly different from normal fibroblasts, similar to the CAFs seen around cancer nests, and these were treated as marginal CAFs. Whereas in the marginal area, we evaluated only CAFs that were positive for each CAF marker and were spindle-shaped, sarcoma cells in the intratumoral area were often positive for CAF markers, making them difficult to distinguish from CAFs. Therefore, in the intratumoral area, “all cells that were positive for any of the three CAF markers” (which would include sarcoma cells and any CAFs that may be present) were evaluated. Moreover, we also examined the number of CD163-positive macrophages (i.e., TAMs) in the intratumoral and marginal areas, and evaluated the association between the expression of CAFs-markers by sarcoma cells/CAFs and the number of TAMs in each area. The aim was to determine the clinicopathological significance of CAF marker expression in the intratumoral and marginal areas and to examine the clinicopathological relationship between CAF marker-positive cells/CAFs and TAMs.

Soft tissue sarcomas (n = 163) surgically resected initially at Akita University Hospital (Akita, Japan) between 2004 and 2018 were collected. Cases not assigned an FNCLCC grade, cases that received preoperative chemotherapy or radiotherapy, or cases judged to have poor IHC staining, were omitted (n = 30). Finally, 133 cases were included in the final analysis. The general characteristics of all enrolled cases are shown in Table 1; all underwent marginal or extended resection. Clinical and pathological data were obtained from medical records and pathological reports. Due to changes in diagnostic criteria, classification, and terminology over the past 2 decades, some cases were re-diagnosed in accordance with the current WHO classification (5th edition) [45]. The histological subtype of the enrolled cases included atypical lipomatous tumor/well-differentiated liposarcoma, myxoid liposarcoma, round cell liposarcoma, dedifferentiated liposarcoma, pleomorphic liposarcoma, myxofibrosarcoma, conventional leiomyosarcoma, poorly differentiated leiomyosarcoma, solitary fibrous tumor, fibrosarcoma, synovial sarcoma, extraskeletal myxoid chondrosarcoma, Ewings sarcoma, and undifferentiated pleomorphic sarcoma. Disease-free survival (DFS) was defined as the date of surgery to the date of identified recurrence or the date the patient was last confirmed to be recurrence-free. The study was approved by the Ethics Committee of Akita University, Graduate School of Medicine (Reference No. 2652). Informed consent was waived by the Ethics Committee. Opt-out was open to all patients who underwent surgery for soft tissue sarcoma at our institution during the above period. The study complied with the tenets of the Declaration of Helsinki.

Formalin-fixed paraffin-embedded sections were cut (4 μm thickness) and stained by a Ventana Discovery XT autostainer (Ventana Medical Systems, Tucson, AZ, USA) prior to IHC examination. Anti-mouse/rat/human CD163 (Clone EPR19518, 1:500, Abcam, Cambridge, UK) was used to stain TAMs, and anti-human FAP (Clone EPR20021, 1:200, Abcam, Cambridge, UK), CD10 (Clone 56C6, 1:1, Dako, Japan), and podoplanin (Clone D2-40, 1:1,000, BioLegend) antibodies were used to stain “CAF markers”.

One representative slide per case that allowed adequate observation of the tumor area and tumor margins was selected. Each slide was scanned by a pathology digital imaging system (NanoZoomer virtual slide system, Hamamatsu Photonics, Shizuoka, Japan) at an absolute magnification of ×20. The intensity of CAF marker expression and the percentage area that was CAF marker-positive were scored (referred to as the proportion score and the intensity score, respectively). Next, the “CAF score” (note that in this manuscript, the “CAF score” is used to refer to any of the three individual marker scores, i.e., the FAP score, the CD10 score, or the podoplanin score) for the marginal and intratumoral regions of each case was calculated. With respect to intensity, a score of 3 denoted strong staining recognizable at ×4 objective, a score of 1 denoted staining recognizable at ×10 objective, and a score of 2 denoted staining intermediate between these two (Fig. 1a–c). None of the CAF markers was able to completely distinguish CAFs from sarcoma cells in the intratumoral area (Fig. 1a–c, lower left panels); therefore, we evaluated all CAF marker-positive cells in the intratumoral area. To score the proportion of CAFs, the area in which CAF marker-positive cells were found relative to the entire tumor area (intratumoral), and the margins in which CAFs are found relative to the entire tumor border (marginal) were scored as follows: score 0, none; score 1, less than 1/3; score 2, 1/3 to 2/3; and score 3, 2/3 or more (Table 2). The final sum of the intensity score and the proportion score was used as the CAF score (FAP score, CD10 score, and podoplanin score) for each area and each CAF marker. Representative samples are shown in Figure 2. The number of CD163-positive macrophages in the digital images (scanned by the pathology digital imaging system) was also counted. Briefly, the intratumoral and marginal areas in five representative fields (each field was 0.2 mm² and chosen at random, avoiding extreme bias) were evaluated, and the number of CD163-positive cells was counted manually (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000539855). With respect to CAF marker expression and assessment of CD163-positive cells, the “marginal area” refers to the non-sarcoma area adjacent to the sarcoma. In cases in which the “marginal area” could not be identified due to marginal resection, the marginal tumor area was used instead of the marginal non-tumor area. In cases where the borderline between the sarcoma and non-sarcoma tissue was ambiguous, sarcoma cells were distinguished from non-sarcoma tissue based on the overall view of the tumor and the presence or absence of cellular atypia; the non-sarcoma side of the border between the two was designated as the “marginal area” around the sarcoma.

The Shapiro-Wilk test revealed that neither the proportion scores nor the number of macrophages showed a normal distribution; therefore, non-parametric analyses were performed. Clinical factors were dichotomized and assessed as follows: sex: male versus female; age: <65 vs. ≥65 years; tumor size: ≤10 cm vs. >10 cm; and location: extremities vs. other. In addition, cases were graded according to the FNCLCC grading system (i.e., tumor differentiation, mitotic count, and tumor necrosis) [46]. The correlation between clinical factors and each CAF score (the FAP, CD10, or podoplanin score) and the number of macrophages was evaluated using the Mann-Whitney U test, the Kruskal-Wallis test, and the Steel-Dwass test (Table 3; Fig. 3). The log-rank test and a Cox proportional hazards regression model were used to evaluate the relationship between each score/number and clinical outcome (Fig. 4-9; Table 4). p values <0.05 were considered significant. Optimal cut-off values for the CAF scores (except for the intratumoral FAP score) and macrophage counts were defined by the Youden index. Optimal cut-off values for the intratumoral FAP score (DFS for Grade 1/2/3 and DFS/metastasis-free survival [MFS]/local recurrence-free survival [LFS] for Grade 2/3) were defined by the maximum value (score 6) because IHC staining was weaker for FAP than for the other two CAF markers (CD10 and podoplanin) and difficult to distinguish from non-specific staining in the intratumoral area (Fig. 1a, top left and middle left). For Grade 1 DFS, optimal cut-off values of the intratumoral FAP score were defined by the Youden index because only 1 case showed the maximum value (score 6). Receiver operating characteristic curves and the Youden index are shown in online supplementary Figure S2. The correlations between clinical outcome and CAF scores were evaluated using the Mann-Whitney U test (Table 5). Correlation coefficients between 0.4 and 0.7 were defined as intermediate, whereas those between 0.2 and 0.4 were defined as weak. All data were analyzed by EZR (Saitama Medical Center, Jichi Medical University), which is based on R (The R Foundation for Statistical Computing, Vienna, Austria, version 4.1.2), and R commander [47].

The FAP score, the CD10 score, the podoplanin score, and the number of CD163-positive macrophages in each area were analyzed statistically (Table 3). As mentioned above, some cases harbored more or fewer sarcoma cells in the intratumoral area that were positive for any of the three CAF markers. It was difficult to distinguish sarcoma cells from CAFs; therefore, we evaluated all CAF marker-positive cells, which would include sarcoma cells and CAFs, in the intratumoral area. Regarding age, there were significant differences between the older and younger groups (<65 vs. ≥65) with respect to the intratumoral FAP score and intratumoral CD10 score (p = 0.0358 and p = 0.00248, respectively). Regarding tumor size, there were significant differences between the groups with different-sized tumors (i.e., ≤10 cm vs. >10 cm) with respect to the number of intratumoral CD163-positive macrophages (p = 0.0238). Regarding FNCLCC grade, there were significant differences in the intratumoral FAP score, the marginal FAP score, the intratumoral CD10 score, the marginal CD10 score, the intratumoral podoplanin score, the marginal podoplanin score, the number of intratumoral CD163-positive macrophages, and the number of marginal CD163-positive macrophages (all p < 0.001) (Table 3). Next, we performed a post hoc analysis of each CAF score and the CD163-positive macrophage counts for each FNCLCC grade (Fig. 3). Overall, samples with a higher FNCLCC grade tended to have higher CAF scores for each area. There were significant differences between each grade with respect to the intratumoral FAP score, the intratumoral CD10 score, and the marginal FAP score (Grade 1 vs. Grade 2; Grade 1 vs. Grade 3; and Grade 2 vs. Grade 3 for the intratumoral FAP score [p < 0.001, p < 0.001, and p = 0.044604958539948, respectively]; for the intratumoral CD10 score [p = 0.024720291403324, p < 0.001, and p < 0.001, respectively]; and for the marginal FAP score [p = 0.0016037657730, p < 0.001, and p = 0.0081406082766, respectively]). Regarding the intratumoral podoplanin score and the marginal CD10 score, there were significant differences between Grade 1 and Grade 3 tumors (p < 0.001). There were also significant differences in the marginal podoplanin score between Grade 1 and Grade 2, and between Grade 1 and Grade 3, tumors (p = 0.004286301485 and p < 0.001, respectively). Regarding CD163-positive macrophages, there were significant differences in the numbers in the intratumoral and marginal areas between Grade 1 and Grade 3 and between Grade 1 and Grade 2 tumors (Grade 1 vs. Grade 2 and Grade 1 vs. Grade 3 [intratumoral area]: p < 0.001; Grade 1 versus Grade 2 and Grade 1 versus Grade 3 [marginal area]: p < 0.001).

For the analysis of all cases (Grade 1/2/3), we assigned the cases into “low” and “high” groups according to the FAP score (intratumoral: 0–5 vs. 6, respectively; marginal: 0–2 vs. 3–, respectively), the CD10 score (intratumoral: 0–4 vs. 5–, respectively; marginal: 0–3 vs. 4–, respectively), the podoplanin score (intratumoral; 0–3 vs. 4–, respectively; marginal 0–2 vs. 3–), and the number of CD163-positive macrophages (intratumoral: 0–295 vs. 296–, respectively; marginal: 0–206 vs. 207–, respectively), and then compared the DFS for each using the log-rank test. All cases in the high group with respect to all scores and numbers (with the exception of the marginal CD10 score) had significantly shorter DFS than those in the low group (Fig. 4; FAP score, CD10 score, podoplanin score, and the number of CD163-positive macrophages in the intratumoral area: p < 0.001, p < 0.001, p < 0.001, and p = 0.00522, respectively. FAP score, CD10 score, podoplanin score, and the number of CD163-positive macrophages in the marginal area: p < 0.001, p = 0.104, p < 0.001, and p < 0.001, respectively). Next, we examined DFS of those with FNCLCC Grade 1 and FNCLCC Grade 2/3 tumors. We found no significant differences in DFS between the high and low groups (intratumoral FAP score: 0–2 vs. 3–, respectively; marginal FAP score: 0–2 vs. 3–, respectively; intratumoral CD10 score: 0–4 vs. 5–, respectively; marginal CD10 score: 0–3 vs. 4–, respectively; intratumoral podoplanin score: 0–3 vs. 4–, respectively; marginal podoplanin score: 0–2 vs. 3–, respectively; intratumoral CD163-positive macrophage counts: 0–295 vs. 296–, respectively; marginal CD163-positive macrophage counts: 0–206 vs. 207–, respectively) with Grade 1 tumors (Fig. 5a–h). To analyze Grade 2/3 cases, we assigned the cases into “low” and “high” groups according to the FAP score (intratumoral: 0–5 vs. 6, respectively; marginal: 0–2 vs. 3–, respectively), the CD10 score (intratumoral: 0–4 vs. 5–, respectively; marginal: 0–3 vs. 4–, respectively), the podoplanin score (intratumoral: 0–3 vs. 4–, respectively; marginal: 0–2 vs. 3–, respectively), and the number of CD163-positive macrophages (intratumoral: 0–295 vs. 296, respectively; marginal: 0–206 vs. 207–, respectively). For Grade 2/3 cases, the group with a high CAF score (excluding the marginal FAP and CD10 scores) showed significantly worse DFS than the group with a low score (Fig. 6a–c, 6e–g; FAP score, CD10 score, and podoplanin score in the intratumoral area: p = 0.0423, p = 0.0033, and p = 0.0484, respectively; FAP score, CD10 score, podoplanin score in the marginal area: p = 0.0776, p = 0.824, and p = 0.0222, respectively). Cases with a high number of CD163-positive macrophages in each area did not show significantly worse DFS (Fig. 6d, h). In addition, we evaluated MFS and LFS for Grade 2/3 cases. Those with a high intratumoral FAP score, a high intratumoral CD10 score, and a high marginal podoplanin score had significantly worse MFS (Fig. 7a, g; p = 0.0472, p = 0.0113, and p < 0.001, respectively). Cases with a high number of CD163-positive macrophages in the marginal area tended to have worse MFS (Fig. 7h; p = 0.06). Regarding LFS, those with a high intratumoral CD10 score had a worse clinical outcome (Fig. 8b; p = 0.0457).

Next, we conducted a multivariate analysis to evaluate the utility of the CAF scores (excluding the marginal CD10 score) and the number of marginal CD163-positive macrophages as predictive factors (Table 4). Sex (male vs. female), age (<65 vs. ≥65), tumor size (≤10 cm vs. >10 cm), tumor location (extremities vs. others), and FNCLCC grade (Grade 1 vs. Grade 2/3) were included as variables. With respect to DFS, the marginal FAP score, the intratumoral CD10 score, the intratumoral podoplanin score, and the marginal podoplanin score were identified as independent prognostic factors (p = 0.01639, p = 0.00851, p = 0.007191, and p = 0.003213, respectively), whereas the intratumoral FAP score, the marginal FAP score, the intratumoral CD10 score, the marginal podoplanin score, and the number of marginal CD163-positive macrophages were identified as independent prognostic factors for MFS (p = 0.03392, p = 0.01793, p = 0.006431, p < 0.001, and p = 0.03868, respectively). Finally, the intratumoral podoplanin score was identified as an independent prognostic factor for LFS (p = 0.0179500).

CAF scores were compared between the recurrent and non-recurrent groups of well-differentiated liposarcoma and undifferentiated pleomorphic sarcoma, for which we collected a relatively large number of cases (Table 5). For the cases of well-differentiated liposarcoma, CAF scores did not significantly differ between the recurrent and non-recurrent groups. By contrast, for the cases of undifferentiated pleomorphic sarcoma, the intratumoral CD10 and podoplanin scores were significantly higher in the recurrent group than in the non-recurrent group (p = 0.0136 and p = 0.0102, respectively). A log-rank test was performed on undifferentiated pleomorphic sarcoma cases in the high and low CD10 and podoplanin score groups. Cases in the high groups of the CD10 and podoplanin scores had significantly shorter DFS (Fig. 9a, b; p = 0.00445 and p = 0.00798, respectively).

In the intratumoral area, the FAP and CD10 scores showed an intermediate correlation with the number of CD163-positive macrophages, while the podoplanin score showed a weak correlation (Fig. 10a–c; r = 0.577, r = 0.553, and r = 0.288, respectively). In the marginal area, each of these scores showed an intermediate correlation with the number of CD163-positive macrophages (Fig. 10d–f; r = 0.53, r = 0.48, and r = 0.553, respectively).

Here, we performed IHC examination of FAP, CD10, and podoplanin as CAF markers, as well as CD163 as a TAM marker, in 133 sarcoma cases, and evaluated CAF scores and macrophages counts in the intratumoral and marginal areas. Statistical analysis of clinical factors revealed that groups with multiple high CAF scores in the intratumoral and marginal areas had a significantly worse clinical outcome than groups with low scores in these areas. In addition, we found a weak-to-intermediate correlation between CAF scores and macrophage counts in the intratumoral and marginal areas.

Several studies reported that malignant epithelial tumors with abundant FAP-positive CAFs have a poor outcome [30‒33]. Few studies have examined FAP in sarcoma; however, Yuan et al. [48] reported that cases of osteosarcoma overexpressing FAP showed a poor prognosis. Several studies have examined the expression of FAP by malignant epithelial tumors and reported that those with high FAP expression in cancer cells have a worse outcome [30, 49].

Malignant epithelial tumors with abundant CD10-positive CAFs have a poor outcome or worse clinical characteristics [9, 32, 36‒38]. Studies also suggest that malignant epithelial tumors with high CD10 expression have a poor outcome [36, 38, 50]. To the best of our knowledge, the only reported sarcoma with high CD10 expression was high grade, but the link between CD10 expression and prognosis has not been studied [51].

Malignant epithelial tumors harboring abundant podoplanin-positive CAFs, as well as other CAF markers, have a poor outcome [8, 28, 40‒42]. Several studies conducted podoplanin staining of tumor cells themselves and found that high expression of podoplanin is associated with poor outcome [41, 49, 52].

As mentioned above, FAP, CD10, and podoplanin are CAF markers; however, we found cases in this study in which the sarcoma cells themselves were positive for these CAF markers. In all cases with high expression of any of the three CAF markers (i.e., FAP or CD10 or podoplanin), it was the sarcoma cells that were positive. We were not able to completely distinguish CAFs from sarcoma cells in the intratumoral area; therefore, all cells (sarcoma cells and CAFs) that were positive for CAF markers were evaluated together.

There was a tendency for each score and the number of CD163-positive macrophages in each area to correlate with FNCLCC grade; this was particularly true for the intratumoral CD10 score and the intratumoral/marginal FAP scores, with significant differences observed between each FNCLCC grade. Reflecting this, Grade 1/2/3 cases with a high CD10/FAP score and a high number of CD163-positive macrophages (excluding the marginal CD10 score) had significantly shorter DFS than those with low scores/numbers. However, we found no significant differences between the high-score and low-score groups, or the number of CD163-positive macrophages, in the cases limited to Grade 1 (Fig. 5); we assume that this is due to the low recurrence rate of Grade 1 sarcomas.

In the intratumoral area of Grade 2/3 cases, we found significant differences between the high and low groups with respect to the prognostic impact of the CD10, FAP, and podoplanin scores on DFS, that of the CD10 and FAP scores on MFS, and that of the intratumoral CD10 score on LFS. The Cox proportional hazards model identified multiple CAF scores in the intratumoral area as independent prognostic factors for DFS, MFS, and LFS. Although this study included a variety of histological tumor types, the results indicate that Grade 2/3 sarcomas with high expression of CAF markers have a poor outcome. These results may be interpreted as indicating that high-grade sarcomas have CAF-like properties, although further in vitro studies are needed to clarify this hypothesis. Many cases showing high expression of CAF markers overlapped with each other, but some did not (online suppl. Table 1). This is consistent with the heterogeneity of CAFs, which have different cellular origins and functions [2, 3, 28]. We also found that undifferentiated pleomorphic sarcoma cases in the high groups of intratumoral CD10 and podoplanin scores had significantly shorter DFS, but the number of undifferentiated pleomorphic sarcoma cases was not large; therefore, the results must be confirmed. To the best of our knowledge, this is the first report to evaluate the prognostic utility of CAF marker (CD10, FAP, and podoplanin) expression by sarcomas.

In the marginal area of Grade 2/3 tumors, some factors had a detrimental effect on survival (i.e., the podoplanin scores on DFS, and the podoplanin score and the number of CD163-positive macrophages on MFS). In particular, multivariate analysis of MFS identified all scores and the number of CD163-positive macrophages (excluding the CD10 score) as independent prognostic factors. Many studies report that high CAF numbers are associated with poor clinical outcomes for various tumors, and the results of our study are consistent with these [30, 31, 33, 42, 49]. In theory, evaluating CAFs in the marginal area would seem to be the more appropriate method because it is difficult to distinguish CAFs in the intratumoral area from the sarcoma cells themselves. However, there are some problems with this method. For example, there were some cases in which diffuse and extensive CD10 staining was seen in areas quite distant from the tumor; this is because anti-CD10 antibodies also stain inflammatory myofibroblasts, making precise evaluation of the marginal CD10 score difficult. For this reason, we think that the marginal CD10 score should not be used as a prognostic indicator. By contrast, although anti-FAP did not stain inflammatory myofibroblasts as extensively as anti-CD10, staining was somewhat weak, making it difficult to evaluate. Podoplanin also stained lymphatic vessels, but it was relatively easy to distinguish lymphatic vessels from CAFs; indeed, CAFs stained with anti-podoplanin were the easiest to observe (Fig. 1c, upper left and middle right panels). We used only a single representative section of the tumor limbus in this study; therefore, it is possible that the amount of CAFs varied depending on the type of marginal tissue (muscle, adipose tissue, periosteum). With respect to TAMs, cases with a high number of marginal CD163-positive macrophages tended to have significantly shorter MFS than those with a low number (p = 0.06), which is consistent with the results of our previous report [26].

Finally, we confirmed a moderate correlation between each score and the number of CD163-positive macrophages in the intratumoral and marginal areas. These results suggest crosstalk between CAFs and TAMs in the marginal area. Indeed, studies demonstrate crosstalk between TAMs and CAFs in epithelial tumors; a similar mechanism may exist in sarcomas [19, 53].

Sarcomas are rarer than carcinomas; in addition, sarcomas encompass many histologic types. Therefore, studying a single histologic type may be difficult. The sarcoma cases in this study were not limited to a single histologic type. Knowledge of the sarcoma microenvironment remains limited. A few studies have examined sarcomas and TAMs, but very few have examined sarcomas and CAFs. Therefore, although this study is not limited to one histological type, we still believe that it is significant, as we consider it to provide fundamental information about the TME in sarcoma.

In summary, sarcomas with high CAF marker expression in the intratumoral area, and high CAF marker expression in the marginal area, have a poorer clinical outcome. We also found a weak-to-moderate correlation between each score and CD163-positive macrophage counts.

The opt-out consent protocol was used for use of participant data for research purposes. This study and consent procedure were reviewed and approved by the Ethics Committee of Akita University, Graduate School of Medicine (Approval No. 2652, date of decision: May 27, 2021).

The authors have no conflict of interests.

This study was funded by a JSPS KAKENHI Grant (No. 21K15380). The authors declare no competing interests.

Michinobu Umakoshi: conceptualization, data curation, formal analysis, funding acquisition, investigation, and writing original draft. Yukitsugu Kudo-Asabe: data curation and investigation. Hiroyuki Tsuchie: resources and investigation. Zhuo Li and Daichi Maeda: methodology. Kei Koyama, Ken Miyabe, and Makoto Yoshida: investigation. Hiroyuki Nagasawa, Hiroshi Nanjo, and Naohisa Miyakoshi: resources. Kyoji Okada: methodology and resources. Masamitsu Tanaka and Akiteru Goto: conceptualization and supervision.

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to restrictions, for example, they contain information that could compromise the privacy of research participants.