Main

Sarcomas are rare, accounting for ~1% of all tumor types. Nevertheless, the heterogeneity within sarcomas is very large; currently, soft tissue and bone sarcomas are classified into 25 main categories by the World Health Organization (WHO),1 and in total >180 different sarcoma subtypes are distinguished. Because of this large diversity, patient cohorts are generally small and the available amount of patient tumor material is scarce. Hence, progress in sarcoma research is relatively slow, and bona fide breakthroughs (eg, the introduction of chemotherapy in the 1970s, or more recently, Imatinib in c-kit positive gastrointestinal tumors) are rare. As a result, patient survival of many types of sarcomas has not been significantly improved over the last few decades.2

Present-day preclinical sarcoma research, like many other cancer research areas, is largely relying on commercially available ‘standard’ cell lines, many of which who were developed in the 1950s and 1960s with largely unknown patients history. Recent genetic analyses have shown that with continued passaging of these cells, new genetic changes are introduced as a consequence of the inherent genetic or phenotypic instability of these cell lines.3, 4 Therefore, it is possible that some characteristics of these cells no longer reflect the cellular processes occurring early in the patient. Consequently, it is important to focus more on primary, patient-derived material with known patient’s history in addition to these commercially available standard cell lines.

So far, cell lines have been obtained from sarcomas mainly by two methods; seeding material from pleural effusions5, 6, 7, 8, 9, 10, 11, 12 or seeding of cells after enzymatic digestion of tumor material.13, 14, 15, 16, 17, 18, 19 As we assumed it to be difficult to optimize enzymatic digestion protocols for each single tissue type affected by sarcoma and because pleural effusion is mainly performed in patients with end-stage lung metastases only, we explored here the potential to obtain patient-derived cell lines from sarcoma patients using the explant technique. Notably, apart from incidental case reports,20, 21, 22, 23, 24, 25 this technique has so far not been investigated thoroughly for sarcomas in general or for specific sarcoma entities. Here, we show that this technique yields enriched cancer cells at early passages in 41% of explanted specimens. These patient-derived cell lines can be used for in vitro studies as drug testing for personalized medicine, but also for establishing additional animal models by orthotopic injection into immunocompromised mice, or for linkage studies when the clinical history of the samples is known.

MATERIALS AND METHODS

Explant Culture

The collection and use of human tissue samples has been approved by the local ethic committees (KEK-10/2007). Samples were processed in a level 2 laboratory according to the regulations of the Swiss authorities.

Samples (711) were from patients in Switzerland with >95% collected at Balgrist University Hospital, and were explanted immediately after removal without prior freezing. Twenty-one osteosarcoma and two chondrosarcoma samples were collected at The Denver Clinic for Extremities at Risk (Denver, CO, USA), frozen in DMEM containing 10% DMSO and shipped on dry ice to Balgrist University Hospital. Upon arrival, samples were thawed quickly and cultured as described below.

Shortly, tissue obtained from diagnostic biopsies (BX) or resections (RX) from primary tumors, recurrences or metastases were placed in tissue culture medium (DMEM, 4.5 g/l glucose:F12 (1:1); 10% FCS; 1% antibiotic–antimycotic (Gibco, Basel, Switzerland)). Upon arrival in the lab, usually <1 h after excision, the material was cut into smaller pieces of about 3x3x3 mm, transferred to 25 cm2 tissue culture flasks and cultured at 37 °C, 5% CO2. Outgrowth of cells was monitored weekly. When cells were distributed equally within the flask, the cells were fed weekly until the flask became almost confluent. When spot-like outgrowth was observed, a trypsinization step (0.05% trypsin and 0.02% EDTA in PBS) was performed and cells were seeded back in the same flask to allow an even distribution of the outgrown cells. After the 25 cm2 flask reached near confluence it was split (passage 1) by trypsinization in four (osteosarcoma), three (chondrosarcoma and Ewing’s sarcoma) or two (all other entities) 75 cm2 flasks. After the passage 1 cells reached confluence, the cells were detached by trypsinization and frozen in tissue culture medium (90%) and DMSO (10%) at 37.5 cm2 equivalents per ampoule.

Tumor samples from which outgrown cells were successfully isolated were classified as ‘growing’, all others as ‘non-growing’. Of these non-growing samples, >80% showed outgrowth of viable cells, but thereafter did not proliferate enough to subsequently be frozen. Success rates were calculated as percentage of samples containing growing cells compared with the total amount of samples within each entity. The time period (Δ; days) between explant and freezing of the outgrowing cells (time to freezing) was used as a parameter to compare differences in proliferation speed between individual sarcoma entities. This time period is influenced by several components; the amount, size distribution (total surface area) and viability of the explanted material, the ability of the cells to outgrow, and finally the proliferation rate of the outgrown cells on plastic dishes.

Karyotyping by GTG Banding

Four to five T25 cell culture flasks with 5 ml cell culture medium at ~60% confluence were obtained for each sample. Cultures were incubated with 25 μl colcemid (KaryoMAX Colcemid Solution, 10 μg/μl; Life Technologies, Zug, Switzerland) at 37 °C and 5% CO2 for 6, 24, 48, 72 and 96 h before harvest. Chromosome preparation was done using standard techniques.26 Briefly, the adherent cells were detached using incubation with trypsin, followed by hypotonic treatment, fixation with methanol–acetic acid, slide preparation and GTG banding. Metaphases were searched and captured using an automated microscope equipped with a scanning software (Metafer, MetaSystems GmbH, Altlussheim, Germany). At least 10 metaphases were analyzed using the Metasystems IKAROS software and the karyotypes were reported according to the latest ISCN.27 Normal samples (0% abnormality) were distinguished from samples with low (0–20%), intermediate (20–60%), and high (60–100%) presence of abnormal metaphases.

FISH Analysis

Fluorescence in-situ hybridization (FISH) was performed according to the protocol of the manufacturer using the MDM2 Amplification probe and the EWSR1 Breakapart probe (Cytocell, Cambridge, UK). Briefly, cytospins with patient cells and 2 μl probes were co-denatured at 75 °C and hybridized overnight at 37 °C. After two post-hybridization washes, fluorescent signals were viewed and analyzed with an Axio Imager Z.2 fluorescence microscope equipped with filters for DAPI, FITC, TRITC, and DEAC (Carl Zeiss AG, Feldbach, Switzerland). Images were captured with a Cool Cube 1 camera and the ISIS software (MetaSystems GmbH, Altlussheim, Germany). Two-hundred nuclei were analyzed for each hybridization.

Immunocytochemistry

Cells were seeded on gelatin-coated (0.1%) glass slides in 24-well plates at a density of 2500–10 000 cells per well. Cells were allowed to attach and spread on the slides and were then fixed with 4% formalin for 5 min at room temperature. Cells were washed with PBS and remaining formalin was neutralized by incubation with cell culture medium containing 10% FCS for 30 min at room temperature. Cells were washed with PBS and incubated with anti-CD99 antibody (Abcam, Cambridge, UK; clone 12E7; ab8855; 1:100; 2 μg/ml) or isotype control (mouse IgG1, kappa (Becton Dickinson, San Jose, CA, USA); #557273; 1:250; 2 μg/ml) in PBS, 1% bovine serum albumin for 2 h at room temperature. Cells were washed three times with PBS and incubated with Alexa546 anti-mouse (Invitrogen, Carlsbad, CA, USA; A11030; 1:400) and Hoechst dye (Sigma, St. Louis, MO, USA; H3570; 5 μg/ml) in PBS, 1% bovine serum albumin for 30 min at room temperature. Cells were washed three times with PBS and slides mounted with Immu-mount (Thermo Scientific, Waltham, MA, USA). Slides were viewed with the 10x objective and respective fluorescence filter blocks using an Axiocam MRm camera attached to a Zeiss Observer.Z1 microscope (Carl Zeiss AG, Feldbach, Switzerland). For estimating the percentage of CD99-positive cells, cells were detached with trypsin/EDTA and staining was performed in suspension under the conditions described above. Images were analyzed with ImageJ (http://imagej.nih.gov/ij/).

Statistical Analysis

Success rates were compared with χ2 test if n were >5, and otherwise with two-tailed Fisher exact probability test using VassarStats (http://vassarstats.net/). Differences between means were compared with paired or unpaired t-test using PRISM 5.

RESULTS

Sample Description

Between June 2005 and November 2014 (9.5 years) a total of 734 tissue samples from sarcoma patients were cultured by the explant technique as described in Materials and Methods section. Table 1 divides these samples into the different sarcoma types according to the 2013 WHO sarcoma classification. Samples (552) were classified as ‘growing’, whereas 182 samples were viable with outgrowing cells, but stopped proliferation and did not expand sufficiently to be considered for freezing and storage (ie, ‘non-growing’). Of all, 323 samples (44%) were from BX, 382 (52%) from surgical RX, 16 (2%) from metastases (M) and 13 (2%) from recurrences (RC).

Table 1 WHO classification of sarcomas

Success Rates and Times to Freezing in Soft Tissue vs Bone Sarcomas

We first compared the overall success rates, defined as percent of samples that yielded enough cells to be frozen, between bone and soft tissue sarcoma. In bone sarcomas, 317 out of 388 samples (success rate: 82%) were classified as ‘growing’, which was significantly higher (P<0.0001) compared with 235 out of 346 soft tissue sarcomas (success rate 68%: Table 2). The mean success rate of all entities was 81% with a 95% confidence interval (CI) of 73–88%. In soft tissue sarcoma three (38%) entities (adipocytic and nerve sheath tumors and undifferentiated/unclassified sarcomas) had lower success rates than the lower CI, and two (25%) entities (so-called fibrohistiocytic and pericytic (perivascular) tumors) had higher success rates than the upper CI of the mean success rate. On the other hand, in bone sarcoma one (11%) entity (haemotopoietic neoplasms) had lower success rate than the lower CI, and four (44%) entities (osteogenic, osteoclastic giant cell-rich, myogenic, lipogenic, and epithelial tumors, and undifferentiated high-grade pleomorphic sarcoma) had higher success rates than the upper CI of the mean success rate.

Table 2 Success rates and time to freezing in individual sarcoma entities

The mean time to freezing (Δ) of all entities was 74 days with a CI of 64–83 days (Table 2). The mean time to freezing of the bone sarcomas (65 days) was significantly (P<0.03) lower than that of soft tissue sarcomas of 84 days, respectively. In soft tissue sarcoma 4 (50%) entities (adipocytic and fibroblastic/myofibroblastic tumors and undifferentiated/unclassified sarcomas) had higher times to freezing then the upper CI of the mean time to freezing and none (0%) had lower times to freezing than the lower CI of the mean time to freezing. On the other hand, in bone sarcoma four entities (osteogenic tumors, Ewing sarcoma, myogenic, lipogenic, and epithelial tumors and undifferentiated high-grade pleomorphic sarcoma) showed times to freezing below the lower CI and only one (11%) entity (tumor syndromes) had higher time to freezing than the upper CI of the mean time to freezing.

In conclusion, the explant technique is suitable to obtain cells from patients’ tissue for most of the sarcoma entities investigated, and was particularly applicable in samples from bone tumors.

Dependency of Success Rates and Times to Freeze on Disease Malignancy

Some sarcoma entities are further subcategorized into benign, intermediate, or malignant based on the ability of the tumor to only grow locally or build metastases. We determined whether these subcategories could explain the variance in success rate, defined as percent of samples that yielded enough cells to be frozen, and time to freeze. Success rates were independent from malignancy in adipocytic, fibroblastic/myofibroblastic tumors, and osteogenic tumors and in tumors of uncertain differentiation (not shown) irrespective whether benign disease or benign plus intermediate disease was compared with malignant disease. In chondrogenic tumors, malignant chondrosarcoma had a significant (P<0.03) higher (85%) success rate than benign plus intermediate disease (65%).

The times to freezing were independent on the malignancy in adipocytic, fibroblastic/myofibroblastic, and osteogenic tumors, in tumors of uncertain differentiation and in undifferentiated/unclassified sarcomas (Figure 1a) when benign disease was compared with malignant disease. In chondrogenic tumors, the time to freezing was significantly lower in chondrosarcoma (53 days) compared with benign disease (79 days; P<0.003) and intermediate disease (73 days; P<0.05; Figure 1b). A similar trend was observed in osteogenic tumors (P<0.11) and tumors of uncertain differentiation (P<0.14) when benign plus intermediate disease was compared with malignant disease (Figure 1a).

Figure 1
figure 1

Relation between malignancy and time to freezing. (a) Number (n) of samples and P for time to freezing (Δ) values for benign versus malign comparison (†), and benign plus intermediate versus malign comparison (††). Values in black indicate significant differences. (b) The distribution of values in chondrogenic tumors with means and 95% confidence intervals.

In hematopoietic neoplasms of the bone, with malignant disease only, the lowest success of 33% was obtained with plasma cell myeloma and highest success (73%) with primary non-Hodgkin lymphoma.

Dependency of Success Rates and Times to Freeze on the Source of Material

When all sarcoma subtypes were considered, the overall success rates and time to freezing did not differ between BX- and RX-derived material (not shown), however, within specific sarcoma entities significant differences were observed. Success rates were significantly (P<0.05) lower in RX (48%) compared with BX (67%) in adipocytic tumors (Figure 2a). Conversely, in chondrogenic tumors the success rates were lower (P<0.02) in BX (59%) compared with RX (83%).

Figure 2
figure 2

Relation between source of material and success and time to freezing. (a) Time to freezing (Δ). Number of samples and P-values for total comparison (†) and paired comparison (††) of biopsies and resections. Values in black indicate significant differences. The distribution of values in adipocytic tumors (b) and chondrogenic tumors (c) with means and 95% confidence intervals are shown as scatter plots, and paired alignment in so-called fibrohistiocytic tumors (d) and tumors of uncertain differentiation (e).

The time to freezing was significantly (P<0.05) higher in RX (Δ=140 days) when compared to BX (Δ=106 days) in adipocytic tumors, whereas in chondrogenic tumors the time to freezing was lower (P<0.02) in RX (62 days) when compared with BX (85 days; Figure 2a; distributions shown in Figures 2b and c). When paired samples (ie, BX and RX of the same patient) were compared, a trend (P<0.13 and 0.21, respectively) towards similar tendencies was observed (Figure 2a). Furthermore, paired analyses gave significant (P<0.03) lower times to freezing in RX compared with BX in so-called fibrohistiocytic tumors and tumors of uncertain differentiation (Figure 2a; individual sample pairs shown in Figures 2d and e).

Dependency of Success Rates on Freezing Samples Before Explant Culture

From 23 samples (21 osteosarcoma and 2 chondrosarcoma) obtained frozen from The Denver Clinic for Extremities at Risk, 20 (87%) were viable after thawing, with outgrowing cells in the explant culture. The success rate, defined as percent of samples that yielded enough cells to be frozen, was 76%. Biopsy- and resection-derived patient material was also immediately frozen from patients in our hospital, and 20 randomly selected samples (5 nerve sheet, 4 chondrogenic, 4 adipocytic, and 7 other sarcoma entities) were thawed and tested for viability. Of all, 17 (85%) showed viability also in this cohort. Two of the three frozen samples that failed were resections that also showed no success in direct explants without prior freezing. Therefore, high viability after freezing and thawing was observed independent of the tumor origin.

Characterization of Cells by Karyotyping and FISH

A tumor in situ consists of a mixture of tumor cells and supporting stromal cells (fibroblasts and endothelial cells). Because the explant technique does not discriminate between the outgrowth of both cell types, the resulting cell cultures may also consist of a mixture of tumor and stromal cells. To estimate the percentage of tumor cells, conventional karyotyping was carried out in samples (both BX and RX) from three sarcoma entities representing the malignant forms of adipocytic and osteogenic tumors and in Ewing sarcoma. Thus, eight osteosarcoma, one Ewing patient and one patient with myxoid liposarcoma were analyzed (Figure 3). Karyotyping was performed after thawing passage 1 cells at passages 2–43 (median 3; 25–75% percentile 2–8). All seven cell lines karyotyped at passages ≥5 showed a high percentage of abnormal cells, whereas only 21% (3 of 14) of cell lines karyotyped at passages <5 exhibited a high percentage of abnormal cells (P<0.002). All osteosarcoma samples were from conventional osteosarcoma, five having osteoblastic and three chondroblastic features. High degrees of complex karyotypes were found in 7 of 17 osteosarcoma samples (41%) derived from eight patients. From five osteosarcoma patients (63%) at least one cell line was obtained with a high degree of abnormal karyotypes. Two samples (BX and RC) from a Ewing patient (named ‘Ewing patient 1’, also see below) with fast progressive and deadly disease showed a highly complex karyotype, whereas the RX derived cells were considered as normal, since they only contained a pericentric inversion of chromosome 10, which most likely represents a constitutional aberration. One cell line from a myxoid liposarcoma also had a highly complex rearranged karyotype at late passage 43. Taken together, in 48% of samples evaluated with chromosome banding analysis the majority of cells were identified as tumor cells with a tendency of higher abnormality rates at later passages (Table 3).

Figure 3
figure 3

Abnormal karyotypes in osteosarcoma (OS), Ewing sarcoma and myxoid liposarcoma- (LS) derived cells. (a) BX of osteoblastic OS6 and (b) BX of the Ewing patient 1. (c) Percentage of cells with abnormal karyotypes, cell morphology and survival. Values in black indicate high degree of abnormality. †, cells derived from an enzymatic digest, not by the explant technique; RC, recurrence; M, metastases; n.p., not present; -, not applicable; ob, osteoblastic; cb, chondroblastic; LFU, lost follow up; DOD, died of disease; grade; Salzer-Kuntschik.40

Table 3 Summary of sample success rates with a high degree of chromosomal abnormalities as assessed by karyotyping or FISH analysis

Next to karyotyping, FISH analysis was performed on six cell lines of patients with dedifferentiated liposarcoma (DDLS) and with pathologically confirmed MDM2 amplification to estimate the number of cells that exhibit MDM2 amplification (Figure 4). FISH analysis was performed after thawing passage 1 cells at passages 2–6 (median 3; 25–75% percentile 3–4). In only one cell line (17%) the MDM2 amplification was found in 100% of the analyzed nuclei, and in two additional samples the MDM2 amplification was found in a low percentage (2%) of nuclei. From seven patients with Ewing sarcoma and pathologically confirmed EWSR1 rearrangements, the rearrangement was again detected in three cell lines (43%) with high percentages (>90% of analyzed nuclei). Taken together, 29% (two of seven samples; Table 3) of liposarcomas showed high degree, and two showed a low degree of abnormality. In osteosarcoma and Ewing sarcoma, 41% (7 of 17) and 44% (4 of 9), respectively, of samples showed a high degree of abnormality. In conclusion, independent on the investigated entities and the method used for assessing the degree of abnormality, 14 (41%) of the 34 investigated cell lines were confirmed to be composed mostly of abnormal tumor cells (Table 3).

Figure 4
figure 4

Fluorescence in situ hybridization analysis in cells derived from dedifferentiated liposarcoma and Ewing sarcoma. (a) Percent of abnormal cells. Representative pictures showing MDM2 amplification (b) in liposarcoma (LS) patient 4 (LS4) and rearrangement in the EWSR1 gene in Ewing patient 1 (Ewing1) (c).

Ewing Patient 1

From Ewing patient 1, who died 52 weeks following diagnosis, cells were obtained from naive BX (week 0), chemotherapy-treated RX (week 21) and from a recurrence (RC) at week 32. BX-derived cells showed a high degree of abnormality by karyotyping and FISH (Figures 3 and 4) and RC-derived cells were also highly abnormal by karyotyping. The short survival of this patient and the highly complex karyotype is consistent with short survival of a subgroup of Ewing patients with complex karyotypes and/or trisomy 20.28 Therefore it was surprising to find an inversion of chromosome 10 as sole aberration in the RX-derived cells 10 (Figure 3). The failure to detect the highly complex karyotype seen in BX- and RC-derived cells is most likely consistent with a cytogenetic remission, as the patient underwent chemotherapy. However, it is not excluded that sarcoma cells were suppressed by non-tumor cells such as fibroblasts as often observed in long-term cultures. We therefore investigated the morphology and CD99 expression29 in these cells (Figure 5). BX-derived cells grew as loosely attached patches that did not reach confluence. RX-derived cells were firmly attached and looked fibroblast like, whereas RC-derived cells were almost unattached or became loosely attached after several days of seeding (Figure 5a). BX and RC derived strongly stained for CD99, whereas staining of RX-derived cells was faint, although above isotype control (Figure 5b). Staining in suspension showed high degree of CD99-positive cells in BX- and RC-derived cells (Figure 5c). RX-derived cells, lacking the complex karyotype, having low CD99 expression and a fibroblast-like morphology, should therefore be considered as stromal cells rather than tumor cells.

Figure 5
figure 5

CD99 expression in biopsy- (BX), resection- (RX), and recurrence- (RC) derived cells from Ewing sarcoma patient 1. (a) Bright field images. (b) Staining with CD99 antibody or isotype control of cells grown on gelatin-coated glass slides. (c) Staining with CD99 antibody or isotype control and Hoechst nuclear dye of BX and RC derived cells in suspension. After staining cells were transferred to a Neubauer chamber and pictures were analyzed with ImageJ to calculate the percentage of CD99-positive cells. Scale bars, 200 μm.

DISCUSSION

Overall, the success rates, defined as percent of samples yielding enough cells to be frozen, were high (50–100%) in a total of 734 investigated samples, with bone sarcomas having a significant higher success rate than soft tissue sarcomas (Table 2). When low times to freezing were considered as an additional parameter for overall success, bone sarcomas had significant lower times to freezing than soft tissue sarcoma. With 75–91% success rates and times to freezing ranging from 58 to 68 days, the chondrogenic and osteogenic sarcomas and Ewing sarcoma seem the most promising entities to obtain patient-derived tumor cells, also with respect to malignancy and/or incidence. Unfortunately, the most common malignant soft tissue tumors, liposarcoma, had a low success rate and high times to freezing. Here, we speculate that the high lipid content might keep the tissues floating, thereby preventing them from adherence to the plastic dishes and ultimately cellular outgrowth.

Malignant tumor cells are expected to evade and migrate faster than less malignant ones and should therefore give rise to cells with shorter times to freezing. We therefore investigated time to freezing in those entities that include benign, intermediate, and malignant tumors. Indeed, malignant chondrogenic tumors had shorter times to freezing than benign tumors (Figure 2). Higher success rates were also related to malignancy in other tumors listed in Table 4.18, 30, 31, 32, 33 A similar trend was observed in osteogenic tumors and tumors of uncertain differentiation. In comparison, cells derived from metastatic colon cancer could not be established by the explant technique and instead required patient-derived xenograft (PDX) transplantation, indicating an inverse relationship between success and malignancy.34 In melanoma, success rates were lower in slow cycling cells, which needed further enrichment by selective trypsinization. A slow cycling phenotype is correlated with a mesenchymal expression profile resembling de-differentiated melanoma and MAPK pathway inhibitor resistance (personal communication).35

Table 4 Success rates in different cancer entities by enzymatic digestion and/or the explant technique

Success rates and time to freezing may also depend on the amount of material that can be used for explantation. The amount of material is usually limited in initial BX, but the tissue is naive, whereas in RX larger amount of material is usually obtained but the material may be necrotic due to the size of the tumor and/or intervention by chemo- and/or radiotherapy. In chondrogenic tumors that usually are not treated before RX, success rates were higher and time to freezing lower than those of BX probably due to the fact that more material can be explanted from RX than BX (Figure 3). On the other hand, in adipocytic tumors that are often treated by chemo- and/or radiotherapy, success rates were lower and times to freezing higher in RX than in BX. A similar trend was observed in paired samples of the same patients from fibrohistiocytic tumors and tumors of uncertain differentiation. This may indicate that indeed the viability of RX may be compromised, and not be compensated by higher amounts of tissue, in some sarcoma entities.

Following a biopsy or surgery, the cultivation of fresh explants may not be feasible in every hospital. Here we have shown that viable freezing of tissue in DMSO-containing medium yields a high success rate in later explants even after shipping the samples on dry ice several months after tissue withdrawal. Therefore, sharing of rare patient material between institutions and later expansion by the here described explant technique is possible.

A major issue inherent to all procedures for the isolation of cancer cells is contamination by stromal cells. We therefore estimated the amount of (abnormal) cancer cells in some entities known to express specific chromosomal aberrations using conventional karyotyping and/or FISH analysis for gene rearrangement or amplification. Here, we aimed to characterize the cells at early passages to avoid in vitro acquisition of genomic instability by passaging.3 Even under these stringent conditions our results for osteosarcoma closely match the 40% obtained after enzymatic digest (Table 4).19 Overall, 41% of samples in these three entities yielded cancer cells with none or minor stromal contamination. This is higher than the median of 14% in explants from several epithelial cancers listed in Table 4. Because we cannot exclude that success rates may be lower in the here not further characterized sarcoma entities, further characterization of explant-derived cells in other sarcomas is warranted. Of note, a considerable amount of samples (59%) had an apparent normal morphology and probably are to be considered as tumor-associated stromal cells. Such stromal cells could be useful to study tumor stroma interaction in two-dimensional (2D) or three-dimensional (3D) co-culture systems in vitro36 or in vivo.37

The time to freezing was <1 month for 10% of the samples (not shown) and <2 months for 40% of samples (Figure 1). Therefore, in 57% of bone tumors and 26% of soft tissue tumors a considerable amount of cells could be obtained in <2 months after diagnostic biopsy (eg, for Ewing patient 1, enough BX-derived cells were obtained two months before resection). This opens the possibility to use the explant technique to assess postoperative treatment options (eg, drug screening) as described in a dog patient.17 To this end, we have shown that drug testing using patient-derived cells is feasible in 2D cultures and in 3D spheroid culture.38

There are several ways to improve further the success or to modify the here described method. First, the abnormality rates assessed here were in most cases derived from early passages, when stromal contamination is expected to be higher compared with late passages. Under the assumption that cancer cells will overgrow stromal cells that will eventually become senescent, cultures with initially low percentage of abnormality should enrich in cancer cells over time. Indeed, we found higher percentage of abnormal cells by karyotyping in later passages compared with early passages (not shown). Second, selective enrichment techniques could be applied as described (Table 4).35 Third, patient material could initially be expanded as PDX in mice that markedly increased the subsequent success in explant culture of colon cancer and hepatocellular carcinoma samples (Table 4).34, 39

In conclusion, the here described explant technique proved that patient-derived cell lines from different kinds of sarcoma entities can be established for use in preclinical studies and in personalized medicine.