Cancer Stem Cells in Sarcomas: In Vitro Isolation and Role as Prognostic Markers: A Systematic Review

Simple Summary Cancer stem cells (CSCs) are responsible for the great challenge in the treatment of sarcomas due to their high prediction to form metastases. This systematic review focuses on collecting existing research on the expression of CSC markers in different types of sarcomas in both in vitro cell lines and patient samples. The results show a great heterogeneity of the studied markers, with ALDH being the only marker commonly used in sarcomas. This broader view may help to develop new CSC characterization assays to advance in more personalized treatments. Abstract Sarcomas are a diverse group of neoplasms with an incidence rate of 15% of childhood cancers. They exhibit a high tendency to develop early metastases and are often resistant to available treatments, resulting in poor prognosis and survival. In this context, cancer stem cells (CSCs) have been implicated in recurrence, metastasis, and drug resistance, making the search for diagnostic and prognostic biomarkers of the disease crucial. The objective of this systematic review was to analyze the expression of CSC biomarkers both after isolation from in vitro cell lines and from the complete cell population of patient tumor samples. A total of 228 publications from January 2011 to June 2021 was retrieved from different databases, of which 35 articles were included for analysis. The studies demonstrated significant heterogeneity in both the markers detected and the CSC isolation techniques used. ALDH was identified as a common marker in various types of sarcomas. In conclusion, the identification of CSC markers in sarcomas may facilitate the development of personalized medicine and improve treatment outcomes.

Despite current treatments (a combination of neoadjuvant and adjuvant chemotherapy with surgery), sarcoma prognosis and survival remain poor due to the high propensity of sarcomas to form metastases, even at the time of diagnosis, with undetectable "Ewing's sarcoma"[Title/Abstract] OR "uterine sarcoma"[Title/Abstract] sarcoma"[Title/Abstract] OR "pleomorphic sarcoma"[Title/Abstract] coma"[Title/Abstract] OR "angiosarcoma"[Title/Abstract] OR coma"[Title/Abstract] OR "Myxofibrosarcoma"[Title/Abstract]) AND cells[MeSH Terms] OR cancer stem cell[MeSH Terms])) AND (biological m  Terms] OR "biomarkers" [MeSH Terms]) AND ("2011/ "2021/05/28"[PDAT]))). The same strategy was followed in all the database equation when necessary. In addition, the bibliography of the selected a viewed to include other works of interest that did not appear in the initial

Study Selection
Authors M.C. and C.M. conducted a bibliographic search in various performed a first screening based on the titles and abstracts. The second s complete reading of the research articles that passed the first screening. In articles were evaluated based on the pre-established inclusion and exclus vitro studies were the focus of this review, and articles that only cond studies were excluded at this stage. Figure 1 depicts the flow diagram o process.

Data Extraction
Following the selection process, M.C. and C.M. independently extrac interest from each research article. According to Cohen's Kappa statistic was good correlation between M.C. and C.M. Any disagreements were res discussion until a consensus was reached. Otherwise, a third experienced the final decision. Each research article was subjected to a quality test for in which was divided into two phases. The first phase contained filters on acteristics that an in vitro study should meet (score ≥5). Studies that did score were excluded. The second phase comprised questions on method and conclusion. The studies were classified based on their score: low qual medium quality (score [6][7][8][9][10][11][12][13][14][15], and high quality (score [16][17][18][19][20]. Tables 1 an obtained data for ease of understanding. Both tables display the reference article, the type of sarcoma studied, the type of sample, the methodology tion and characterization, and the main results.

Data Extraction
Following the selection process, M.C. and C.M. independently extracted the data of interest from each research article. According to Cohen's Kappa statistic test [23], there was good correlation between M.C. and C.M. Any disagreements were resolved through discussion until a consensus was reached. Otherwise, a third experienced author made the final decision. Each research article was subjected to a quality test for in vitro studies, which was divided into two phases. The first phase contained filters on the basic characteristics that an in vitro study should meet (score ≥ 5). Studies that did not meet this score were excluded. The second phase comprised questions on methodology, results, and conclusion. The studies were classified based on their score: low quality (score 0-5), medium quality (score [6][7][8][9][10][11][12][13][14][15], and high quality (score [16][17][18][19][20]. Tables 1 and 2 show the obtained data for ease of understanding. Both tables display the reference of the selected article, the type of sarcoma studied, the type of sample, the methodology of CSC extraction and characterization, and the main results.
The viability of cultures with Vincristine, cyclophosphamide, and etoposide is higher in the ALDH1 + cells than the ALDH1 − cells. - The expression of ALDH1A3, ALDH1B1, ALDH1L2, Sox2, ABCG2/BCRP, ABCB1/MDR1, and ABCA2 in the ALDH1 + cells increased compared to ALDH1 − cells. There was no difference in ALDH1A1 and c-myc expression. - The sample after chemotherapy exhibited a greater ALDH1 expression than before chemotherapy. There

Results
A total of 228 articles were identified in the initial search across the different databases consulted. After eliminating non-original articles (n = 106), duplicates between databases (n = 15), and articles that did not meet the inclusion criteria, 45 articles were retained for detailed analysis. Subsequently, 9 articles that did not meet the inclusion criteria and 1 that scored very low in the quality test were discarded. Finally, 35 articles were included in this systematic review. The flow diagram representing the search process and the screening of the articles is presented in Figure 1.

Results
A total of 228 articles were identified in the initial search across the different databases consulted. After eliminating non-original articles (n = 106), duplicates between databases (n = 15), and articles that did not meet the inclusion criteria, 45 articles were retained for detailed analysis. Subsequently, 9 articles that did not meet the inclusion criteria and 1 that scored very low in the quality test were discarded. Finally, 35 articles were included in this systematic review. The flow diagram representing the search process and the screening of the articles is presented in Figure 1.
The number of articles published on CSCs in sarcomas decreased over time. The years with the highest number of publications were 2012 and 2015 (6 and 7 per year, respectively), whereas the fewest publications were observed in 2018 (1 article) and 2019 (no articles) (Figure 2). Analysis of the type of sarcoma studied in the 35 articles published in the last 10 years revealed that 22 articles focused on BS, 7 on STS, and 3 on both types (i.e., STS and BS). GISTs have been less extensively studied with only 3 articles published in the last decade. Interestingly, more articles were published on BS than on STS, except in 2013 and 2014. However, articles on STS included a larger number of different subtypes, and the percentage of STS was similar to that of BS studied ( Figure 3A). In the category of BS, most studies focused on osteosarcoma (75%), while on STS, synovial sarcoma (24%) and fibrosarcoma (24%) were the most studied subtypes, followed by rhabdomyosarcoma (12%) and liposarcoma (12%) ( Figure 3B,C).

Bone Sarcoma
Of the 35 articles selected for this systematic review, 27 studied CSC markers for BS. Specifically, 21 isolated CSCs: 16 from osteosarcoma, 2 from chondrosarcoma, and 3 from Ewing's sarcoma ( Table 1). The percentage of the total studies collected on BS was 57%, 7%, and 11%, respectively ( Figure 3B), with osteosarcoma being the most studied BS for isolating CSCs in vitro.
CSC markers in osteosarcoma were studied in 16 articles ( Table 1). The most commonly used osteosarcoma cell lines were MG63, Saos-2, and U2OS, although most articles (10) utilized primary tumor samples from osteosarcoma patients to isolate CSCs. In addition, one article obtained samples from osteosarcoma lung metastasis [29]. Only four studies specified that patients had not been previously treated with chemotherapy [29,30,32,35], while patients underwent chemotherapy and radiotherapy in one study [46].
The most frequently used CSC isolation technique in osteosarcoma was Fluorescent-Activated Cell Sorter (FACS) combined with subsequent culture of the isolated and sorted cells in induction media for maintenance [25,27,31,37,46]. Four articles carried out the sphere formation assay using an induction culture medium [24,28,35,36], one article used FACS only without induction medium [33], two articles used the Magnetic-Activated Cell Sorting (MACS) technique and then induction medium [26,29], one article used MACS without induction medium [34], and finally, the Side Population (SP) technique with subsequent induction medium was used in three articles [13,30,32].

Bone Sarcoma
Of the 35 articles selected for this systematic review, 27 studied CSC markers for BS. Specifically, 21 isolated CSCs: 16 from osteosarcoma, 2 from chondrosarcoma, and 3 from Ewing's sarcoma ( Table 1). The percentage of the total studies collected on BS was 57%, 7%, and 11%, respectively ( Figure 3B), with osteosarcoma being the most studied BS for isolating CSCs in vitro.
CSC markers in osteosarcoma were studied in 16 articles ( Table 1). The most commonly used osteosarcoma cell lines were MG63, Saos-2, and U2OS, although most articles (10) utilized primary tumor samples from osteosarcoma patients to isolate CSCs. In addition, one article obtained samples from osteosarcoma lung metastasis [29]. Only four studies specified that patients had not been previously treated with chemotherapy [29,30,32,35], while patients underwent chemotherapy and radiotherapy in one study [46].
The most frequently used CSC isolation technique in osteosarcoma was Fluorescent-Activated Cell Sorter (FACS) combined with subsequent culture of the isolated and sorted cells in induction media for maintenance [25,27,31,37,46]. Four articles carried out the sphere formation assay using an induction culture medium [24,28,35,36], one article used FACS only without induction medium [33], two articles used the Magnetic-Activated Cell Sorting (MACS) technique and then induction medium [26,29], one article used MACS without induction medium [34], and finally, the Side Population (SP) technique with subsequent induction medium was used in three articles [13,30,32].
In the case of FACS and MACS techniques which were used in nine osteosarcoma articles, CSC markers were necessary for their isolation. The most commonly used method (three/nine articles) was double labeling with CD133 and CD44 [26,27,29], followed by labeling with CD133 only [34]. Other markers used were CD24 [25], CD49f [33], Sox2 together with Sca-1 [37] and CD109 [46]. Of the 13 osteosarcoma articles using induction and sphere-forming media, all utilized serum-free medium supplemented with factors, with EGF (epidermal growth factor) and FGF (fibroblast growth factor) being the most important, although in varying proportions. The most frequently used base medium was DMEM/F12, with a total of nine articles. One article used Ham's F12 [24], while DMEM [27], RPMI [35], and N2B27 [37] were used depending on the cell line.
All articles conducted characterization of CSCs after isolation. The most commonly used methods were RT-qPCR, Western blot, flow cytometry, IF, and ALDEFLUOR assay to check the expression of CSC markers in the isolated cells. In addition, in vivo tumorigenicity assays were performed. The most highly expressed markers of CSCs isolated from established cell lines were Oct4, CD133, SOX2, Nanog, and CD44 ( Figure 4A). On the other hand, the most highly expressed markers in CSCs isolated from primary tumor samples from patients coincided with those mentioned for the cell lines. In the case of the only article that studied pulmonary metastasis, the CD133 and CD44 markers coincided in both the cell lines and patient samples [29]. Furthermore, several CSC markers have been associated with poor patient prognosis, including CD24 [25], DANCR [27], and CD49 [33]. Some articles also examined the modulation of certain markers after treatment. Zhou et al. [25] observed an increase in CD24 expression after treatment with Cisplatin, Epirubicin, and Hydrochloride. Epigallocatechin gallate (EGCG) reduced several CSC markers, namely CD44, CD133, SOX2, OCT4, c-Myc, and Nanog (CD133+/CD44+) [26]. Co-culture with mesenchymal stem cells (MSCs) also increased the expression of SOX2, OCT4, Nanog, and CXCR4 markers [28]. CSCs with high expression of CD248, CD133, OCT4, Nanog, and Nestin were more resistant to Doxorubicin, Cisplatin, and Methotrexate [30]. Verapamil treatment increased the sensitivity of CSCs to Doxorubicin [36].
Ewing's sarcoma was the second most studied sarcoma subtype in terms of CSC isolation, with a total of three articles (11% of all BS) included in this review. There was no common cell line among these articles. Sphere formation using induction medium was the primary method of CSC isolation in two articles [38,39], while FACS isolation with subsequent culture in induction medium was applied in Emori et al. [46], using the CD109 marker. The medium composition was DMEM/F12 in two articles and IMDM in one [38]. The most common methods used for characterization were FACS, in vivo tumorigenicity, and RT-PCR, although SP and ALDEFLUOR assays were also applied. In Ewing's sarcoma, CD57 was the most highly expressed marker in CSCs derived from patient samples [39], followed by OCT4 and Nanog [38] ( Figure 4B). Additionally, the CD109 marker was expressed in both cell lines and patient samples [46].
Chondrosarcoma was the least studied BS, with a total of two articles (7% of all BS) included in this review. In both cases, FACS isolation was carried out using the ALDH marker [40,48], but only Granger et al. [40] subsequently cultured the isolated cells in induction medium. Common characterization methods included Western blot and RT-PCR, while IF was also applied in Lohberger et al. [48]. Both articles only used established cell lines. Granger et al. [40] found high expression of ALDH, CD44, STRO1, and STAT3 in the JJ012 line, while Lohberger et al. [48] found a high expression of ALDH, c-Myc, SOX2, B-Catenin, ABCG2, and ABCB1 in the SW1353 cell line ( Figure 4C). Additionally, ALDH+ CSCs were found to be resistant to chemotherapy in Lohberger et al., 2012 [48], while treatment with PRP-1 (proline-rich polypeptide-1) had no effect in Granger et al. [40]. Ewing's sarcoma was the second most studied sarcoma subtype in terms of CSC isolation, with a total of three articles (11% of all BS) included in this review. There was no common cell line among these articles. Sphere formation using induction medium was the primary method of CSC isolation in two articles [38,39], while FACS isolation with subsequent culture in induction medium was applied in Emori et al. [46], using the CD109 marker. The medium composition was DMEM/F12 in two articles and IMDM in one [38]. The most common methods used for characterization were FACS, in vivo tumorigenicity, and RT-PCR, although SP and ALDEFLUOR assays were also applied. In Ewing's sarcoma, CD57 was the most highly expressed marker in CSCs derived from patient samples [39], followed by OCT4 and Nanog [38] (Figure 4B). Additionally, the CD109 marker was expressed in both cell lines and patient samples [46].
Chondrosarcoma was the least studied BS, with a total of two articles (7% of all BS) included in this review. In both cases, FACS isolation was carried out using the ALDH marker [40,48], but only Granger et al. [40] subsequently cultured the isolated cells in induction medium. Common characterization methods included Western blot and RT-PCR, while IF was also applied in Lohberger et al. [48]. Both articles only used established cell lines. Granger et al. [40] found high expression of ALDH, CD44, STRO1, and STAT3 in the JJ012 line, while Lohberger et al. [48] found a high expression of ALDH, c-Myc, SOX2, B-Catenin, ABCG2, and ABCB1 in the SW1353 cell line ( Figure 4C). Additionally, ALDH+ CSCs were found to be resistant to chemotherapy in Lohberger et al., 2012 [48], while treatment with PRP-1 (proline-rich polypeptide-1) had no effect in Granger et al. [40].
The isolation of CSCs from synovial sarcoma was described in four articles ( Table 1). The most commonly used cell line was Fuji. FACS with induction media and the CD109 [46] and CD271 markers [47] were the most commonly used isolation methods, followed by FACS without induction media with the ALDH marker [48]. Sphere formation assay was also employed by Kimura et al. [41]. The induction medium used in all three articles was based on DMEM/F12. Characterization methods included RT-PCR, flow cytometry, Western blot, immunohistochemistry (IHC), and IF. From already established cell lines, Kimura et al. [41] found increased expression of the markers Nanog, Oct4, SOX2, and CXCR4 in CSCs from SYO-I, Fuji, and HS-SYII lines. Lohberger et al. [48] used the SW982 line, whose CSCs expressed ALDH, c-Myc, Sox2, B-catenin, and ABCG2 ( Figure 4D). Emori et al. [46] did not find expression of CD119 in FUJI and YaFuss. One article showed expression of the unusual markers CD105, 2E4B4, 58B1, W5C4, and CD109 in both the SW982 line and in primary tumor samples from patients ( Figure 5). In addition, the tumors expressed CD271 and TNAP [47]. Regarding chemotherapy treatment, SW982 ALDH+ cells were found to be resistant to Doxorubicin [48]. Furthermore, CD271+ cells derived from patient primary tumors and the SW982 line were found to be resistant to Doxorubicin and Epirubicin [47].
Emori et al. [46] did not find expression of CD119 in FUJI and YaFuss. One article showed expression of the unusual markers CD105, 2E4B4, 58B1, W5C4, and CD109 in both the SW982 line and in primary tumor samples from patients ( Figure 5). In addition, the tumors expressed CD271 and TNAP [47]. Regarding chemotherapy treatment, SW982 ALDH+ cells were found to be resistant to Doxorubicin [48]. Furthermore, CD271+ cells derived from patient primary tumors and the SW982 line were found to be resistant to Doxorubicin and Epirubicin [47]. Fibrosarcoma, along with synovial sarcoma, was one of the most studied STS in the articles reviewed. Table 1 lists four articles on this topic. The most commonly used cell Fibrosarcoma, along with synovial sarcoma, was one of the most studied STS in the articles reviewed. Table 1 lists four articles on this topic. The most commonly used cell line was HT1080 [44,47] (Figure 4F). Three articles used FACS in different ways: with the ALDH marker and induction medium [43], with ALDH without subsequent culture in induction medium [48], or with the CD271 marker and subsequent induction medium [47]. MACS with the CD133 marker and subsequent induction medium was used in Feng et al., 2013 [44]. RPMI was used as induction medium in Li et al. [43], while the rest used DMEM/F12. The prevailing characterization methods were RT-PCR, Western blot, in vivo tumorigenesis, IF, and IHC. Only one article used primary tumor samples from patients in the study. Lohberger et al. [48] used the SW982 and SW684 lines, Wirths et al. [47] used SW982, SE872, and HT1080, Feng et al. [44] used HT1080, and Li et al. [43] used the NMFH-1 line. Three articles found a high expression of c-Myc and SOX2 markers in CSCs. Nanog, Oct4, and ABCG2 markers were found in two articles. Other markers expressed in cell lines include STAT3, CD133, BMi-1, ALDH, and B-catenin. Notably, BMi-1 was overexpressed in CD133+ CSCs in HT1080 [44], but not in ALDH+ NMFH-1 cells [43]. Both ALDH+ and CD133+ cells mentioned were found to be resistant to treatment with Doxorubicin and Cisplatin. Additionally, CD105, 2E4B4, 58B1, W5C4, and CD109 were expressed in cell lines and samples.
Isolation of CSCs from rhabdomyosarcoma and liposarcoma was carried out in two articles each once ( Table 1). The studies used different cell lines: TE-671 [48], and RD and KYM-1 [42] (Figure 4E). None of the studies used patient samples, but both used FACS-based CSC isolation methods with the ALDH marker. The common characterization methods were RT-PCR, Western blot, IF, and in vivo tumorigenicity. One of the articles reported no results using the cellular line of rhabdomyosarcoma [48], and the other found overexpression of ALDHA3, ALDHB1, ALDHL2, SOX2, ABCG2, ABCB1, and ABCA2 markers. Additionally, ALDH+ cells were found to be more resistant to treatment with Vincristine, Cyclophosphamide, and Etoposide. Table 1 includes two articles that evaluated CSC isolation in liposarcoma [47,48]. Both studies used the SW872 cell line. Wirths et al. [47] isolated CSCs using FACS (CD271) and subsequently maintained the cells in induction medium. They characterized CSCs using IHC. On the other hand, Lohberger et al. [48] used FACS (ALDH) for isolation and IF Western blot and RT-PCR for characterization. The markers studied in both articles did not coincide. In Wirths et al. [47], both CSCs derived from the cell line and primary tumor samples from patients expressed CD105, 2E4B4, 58B1, W5C4, and CD109. The primary tumor tissue also expressed CD271 and TNAP, with CD271+ cells being associated with Doxorubicin resistance. On the other hand, Lohberger et al. [48] only isolated CSCs from the cell line, and these cells expressed ALDH, c-Myc, Sox2, B-catenin, and ABCG2 markers.
Finally, one article related to the isolation of CSCs from epithelioid sarcoma, malignant fibrous histiocytoma, leiomyosarcoma and chordoma was found. Table 1 presents a single article reporting the isolation of CSCs in epithelioid sarcoma [46]. The study employed two cell lines, FU-EPS-1 and VA-ES-BJ, in addition to primary tumor samples. Isolation of CSCs was carried out using FACS with the CD109 marker and subsequent culture in induction medium. CSCs were characterized using RT-PCR and the ALDEFLUOR assay. The CSCs obtained from the established cell lines and tumor samples expressed ALDH, SOX2, OCT4, Nanog, and CD109 markers. It is worth noting that the expression of CD109 correlated with low survival. On the other hand, the study by Emori et al. [46] used the malignant fibrous histiocytoma cell lines, MFH2033 and MFH2004, in addition to primary tumor samples. Isolation of CSCs was carried out using FACS with the CD109 marker and subsequent culture in induction medium. The CSCs were characterized using RT-PCR and the ALDEFLUOR assay. The CSCs obtained from the cell lines and tumor samples expressed CD109, which also correlated with poor survival. In relation to leiomyosarcoma, Table 1 includes the only article that investigated this tumor [47]. The study employed the SK-LMS1l cell line and tissue samples from the primary tumor. Isolation of CSCs was carried out using FACS with CD271, followed by induction media and characterization with IHC. The CSCs obtained from the cell lines and tumor samples expressed CD105, 2E4B4, 58B1, W5C4, and CD109 markers. The primary tumor tissue also expressed CD271 and TNAP, with CD271+ cells being associated with Doxorubicin resistance. Finally, Lohberger et al. [48] studied chordoma using the MUG-Chor-1 cell line and did not use patient samples. Isolation of CSCs was carried out using FACS with ALDH and subsequent characterization of CSCs with IF, Western blot, and RT-PCR. The isolated CSCs expressed ALDH, Sox2, cMyc, β-catenin, and ABCG2 markers.

GIST
Only one of the three articles (33%) on GIST reported the isolation of CSCs ( Figure 3D) [45]. The study employed four cell lines, namely, GIST882, GIST48, GIST62, and GIST-T1 ( Figure 4G). Isolation of CSCs was performed by FACS using CD133 and CD44 markers, followed by characterization using IF, flow cytometry, and SP assay. The CSCs showed high expression of CD44 and CD133 markers, especially in GIST located in the stomach. The high expression of these markers was associated with resistance to Imatinib.

Bone Sarcoma
This systematic review includes 27 articles on BS, of which 8 only measured CSC marker expression on the entire cell population without performing in vitro treatment on cell lines or patient tumor samples. Specifically, 18% of the articles focused on osteosarcoma, and 7% on chondrosarcoma. No articles were found on Ewing's sarcoma. The results are summarized in Table 2.
Analysis of CSC markers in osteosarcoma without isolation was found in four articles ( Table 2). The commonly used cell lines were MG63 and U2OS, while two articles only used patient samples [49,57]. One article exclusively used patient samples [51]. The marker expression analysis was primarily based on Western blot, followed by RT-qPCR. One article performed FACS and ALDEFLUOR assay [57], and another article used IHC [49]. CD133 and CXR4 markers were expressed in 26% and 36% of the samples, respectively, and were found to be correlated with pulmonary metastasis [49]. In contrast, Avdonkina et al. [57] did not observe CD133 expression. Moreover, the modulation of markers by the SOX2lncRNA gene was studied; its overexpression increased OCT4, ALDH, CD133, and CD44 markers in Saos-2 cells, while the opposite was observed in U2OS cells. Overexpression of SOX2lncRNA was also associated with poor survival [50] ( Figure 6A). Additionally, Zoledronate-resistant cells showed high expression of Nanog, c-Myc, Oct-4, and Sox2 markers [51]. Only one article studied CSC markers in chondrosarcoma [52] using the JJ012 cell line, without using patient samples. The authors analyzed Nanog expression using Western blot. Comparison of the baseline with the line after PRP-1 treatment revealed a significant decrease in the Nanog marker.

Soft Tissue Sarcoma
Synovial sarcoma was studied in two articles focusing solely on patient samples and analyzing the markers using IHC [57] or FACS with CD133 and the ALDEFLUOR assay [53]. According to this study, CD133 was the most frequently expressed marker in 85% of Only one article studied CSC markers in chondrosarcoma [52] using the JJ012 cell line, without using patient samples. The authors analyzed Nanog expression using Western blot. Comparison of the baseline with the line after PRP-1 treatment revealed a significant decrease in the Nanog marker.

Soft Tissue Sarcoma
Synovial sarcoma was studied in two articles focusing solely on patient samples and analyzing the markers using IHC [57] or FACS with CD133 and the ALDEFLUOR assay [53]. According to this study, CD133 was the most frequently expressed marker in 85% of the samples, followed by CD24 and CD44 in 55%, Nestin in 30%, and ALDH in 25%. ALDH was found to be correlated with poor survival. Interestingly, ALDH was expressed more frequently in STS than in BS [57]. On the other hand, only one article investigated undifferentiated cardiac sarcoma [54]. The study analyzed CD44 and OCT3/4 markers in patient samples using BS, and only CD44 expression was detected. Finally, we identified one article that studied several types of STS, including liposarcoma, fibrosarcoma, leiomyosarcoma, rhabdomyosarcoma, schwannoma, alveolar sarcoma, clear cell sarcoma, and dermatofibrosarcoma [57]. The study only analyzed patient samples using FACS and the ALDEFUOR assay. The results showed a higher expression of ALDH compared to BS, but no difference in CD133.

GIST
GIST was investigated in two articles. One study analyzed GIST882 and GIST48b cell lines, as well as patient samples [55], while the other only analyzed patient samples [56]. Immunohistochemistry was the main method used to investigate marker expression. The study by Geddert et al. [55] found that treatment with 5-aza-dC (5-aza-2'-deoxycytidine) reduced CD133 reactive hypomethylation and inversely correlated CD133 expression with survival, while directly correlating CD133 expression with gastric localization and KIT mutation. The relationship between KIT mutation and CD133 expression was also supported in the study by Bozzi et al. [56], as well as CD90, CD44, and CD34 markers. The expression of Kit, CD133, CD90, and CD34 was found to decrease or increase with the combination of KIT mutation and Imatinib treatment.

Discussion
The aim of this systematic review was to provide a detailed description of the existing literature on the expression of CSC markers in sarcomas, both in vitro and in the heterogeneous population of cells present in samples. To our knowledge, this is the first systematic review of CSC markers in sarcomas. After screening, 35 articles were selected. A great imbalance was found when comparing the types of sarcomas studied in each article. Specifically, 22 articles focused only on BS, while 7 focused on STS. Three articles studied both BS and STS, and the remaining studied GIST. These data do not agree with the classification of sarcomas since the largest group is composed of STS with 80%, followed by BS with 15% and GIST with 5% [3,4]. Therefore, although more research is needed in general in this field, specifically more studies are needed in STS.
Focusing on STS, the most frequent subtype in children is rhabdomyosarcoma [58]. It is a very aggressive tumor with a skeletal muscle phenotype [59]. Two subtypes of rhabdomyosarcoma have been described: alveolar and embryonal, with the former being much more aggressive [60]. They present a 5-year survival rate of 28.9% and 74%, respectively [61]. On the other hand, the most frequent BS is osteosarcoma, which presents a peak incidence in adolescents, between 15 and 18 years of age, and another in adults at 42 years of age [62]. The 5-year survival rate in localized osteosarcoma is 65% and less than 20% in metastatic disease [63]. In this systematic review, only three articles on rhabdomyosarcoma have been found, so more studies are needed due to its high frequency. However, osteosarcoma is the most studied BS, which is consistent with its high prevalence.
This review specifically focuses on articles studying CSC markers in sarcomas. CSCs are defined as a subpopulation of cells in the tumor mass with the capacity for tumor initiation, progression, metastasis, and drug resistance [64,65]. Currently, the strategy followed by most studies is to first isolate CSCs and then analyze the expression levels of certain markers in these cells. CSCs can be isolated from existing cell lines or by generating cell cultures derived from patient tumor samples, which is not always an effective technique. In fact, the success rate in establishing cultures derived from patient samples is 82% in bone tumors and 68% in STS [66]. There are different techniques for the isolation of CSCs based on their unique characteristics, including drug resistance, dye explosion or SP assay, separation using surface markers with FACS or MACS, or ALDH activity. CSCs can also be enriched due to their ability to form tumor spheres in the absence of adherence and serum. However, there was no common method used in the studies reviewed, and even when the same method was used in sarcomas, there was much disparity in the methodology used. The CSC culture medium presented very different compositions, and no common marker was found to be used for isolation by FACS or MACS. In general, these processes select a population of CSCs with a specific characteristic, such as a particular marker, being separated from other subpopulations of CSCs with less expression of the same marker. Therefore, this CSC population is distanced from the heterogeneous population present in tumors in vivo. In conclusion, the studies were not carried out in the most appropriate way. Another strategy is based on analyzing CSC markers on the entire tumor population, either on in vitro cultures or on tumor samples using IF, IHC, or qPCR. This way, the entire CSC population present in the tumor can be analyzed. Of the 35 articles reviewed, only 9 used this approach, which is the most novel.
On the other hand, few studies analyzed the expression of CSC markers in tumor samples, so more studies are needed. It is necessary to compare the results of marker expression in isolated CSCs and in the whole cell population since the common markers could be the ones that should be used to corroborate that CSCs have been isolated in vitro similar to those present in the clinic. When analyzing the results of the included studies, the different homogeneity/heterogeneity between the markers detected in the different techniques used in the isolation of CSCs is very interesting. For example, in the case of GIST, CD44 and CD133 are constant in CSCs of this type of tumor, both in isolation using induction media and in biopsy studies. However, in the case of osteosarcoma and synovial sarcoma, a different marker profile can be seen in the case of biopsy studies compared to studies with induction media on cells. Thus, in the case of osteosarcoma, CD44 and Nanog are outstanding markers when isolation is carried out with induction media, while markers such as ALDH and CXCR5 are found on biopsy samples ( Figure 6). Other markers such as CD133 and SOX2 are constant between studies. When we consider synovial sarcoma, CD44 is detected in biopsies but not in studies with induction media on the populations obtained. However, these results could be due either to a difference between the induced CSCs and those existing in the primary tumors, or it is possible that there is a bias related to the type of marker studied in each case, and the non-existence of results is due to the fact that such studies have not been carried out.
One of the most important findings from the reviewed studies is that ALDH is shown to be an important pan-CSC marker in the case of sarcomas, since it has been detected as positive in a multitude of sarcomas analyzed. Additionally, the use of chemotherapy seems to select some populations of CSCs with respect to others, with subpopulations of greater or lesser chemosensitivity and aggressiveness within the same CSCs analyzed, which could help to personalize therapies. However, there is a general lack of studies on which markers are more sensitive to different chemotherapeutic regimens. Only 30 articles in this review employed the use of chemotherapy treatment. This is important because it would enable us to determine which chemotherapy or radiotherapy to use according to the expression of each specific marker and to study resistance.
Our results are based on studies of primary and metastatic tumor samples, with the aim of studying CSCs to achieve a treatment to prevent metastasis or otherwise prevent further spread of the disease. Other reviews in this field focus on more novel aspects such as the study of markers in circulating CSCs in blood, with the aim of developing a new therapeutic target for these circulating cells responsible for metastasis. These studies also highlight the significant problem that we are addressing and show that there is also great heterogeneity in the use of CSC markers in circulating cells [67][68][69][70][71][72]. Ultimately, the goal in the future is to achieve personalized medicine.

Conclusions
The role of CSCs in the prognosis of sarcomas has gained great importance in recent years. However, currently very few studies have investigated the expression of CSC markers in the diverse range of existing sarcoma subtypes. Therefore, further research is urgently needed to identify more sensitive CSC biomarkers and to establish a comprehensive set of markers that can accurately characterize CSCs in sarcomas. Ultimately, this will lead to the development of personalized treatments tailored to the specific CSC markers present in each patient's tumor.