Bone marrow and peripheral blood samples from 60 patients with suspected bone sarcoma were examined for the presence and number of micrometastatic osteosarcoma cells by a sensitive immunomagnetic detection assay, using in parallel two osteosarcoma-associated antibodies. Forty-nine of the patients had osteosarcoma, and of these, as many as 31 (63%) had tumor cells in bone marrow, in many cases with a high number of cells. Only four (8%) were positive also in blood. None of 38 control bone marrow samples were positive, including 11 from patients with suspected bone sarcoma at time of sampling who later were found not to have osteosarcoma. Fifteen of 28 patients without overt metastases at primary diagnosis (54%) were positive, 12 of whom had localized high-grade primary tumors in the extremity. Four of these have relapsed compared with none of 10 negative patients. In the group of 22 patients with extremity localized nonmetastatic osteosarcoma, information was available on the histologic response to preoperative chemotherapy in 15 patients. None of the three patients in the bone marrow–negative group who had a poor response to chemotherapy have relapsed, whereas two of the four poor responders in the bone marrow–positive cohort are dead of disease. Among 12 patients with overt metastasis at primary diagnosis, 11 (92%) were positive in bone marrow with a very high number of osteosarcoma cells. The immunomagnetically isolated cells were further characterized by the use of fluorescent latex microparticles with surface-bound antibodies targeting different membrane markers. Moreover, in cases with numerous osteosarcoma cells in bone marrow attempts to grow the selected cells in vitro were successful in two of eight attempts, and in two of five cases, isolated cells produced tumors with osteosarcoma characteristics in nude mice. In conclusion, already at primary diagnosis, a very high fraction of osteosarcoma patients had malignant cells in bone marrow, and a correlation between the presence of tumor cells, clinical stage, and disease progression was found. The data show the clinical potential of this immunomagnetic method. Attempts to subgroup osteosarcoma patients for more individualized treatment based on the presence of micrometastatic cells should be studied in a larger cohort of patients.

During the last two decades, impressive progress in the treatment of patients with high-grade osteosarcoma has been achieved, clearly linked to the success of a multimodality approach aimed at treating micrometastatic disease present at time of diagnosis (1, 2). Historical evidence has revealed that up to 20% of osteosarcoma patients without overt metastasis at diagnosis could have been cured by surgery alone (36). It would be important to identify those patients to spare them from toxic post-operative adjuvant chemotherapy currently given to all. In the group of patients who would otherwise relapse, ∼50% have chemosensitive tumors and are cured by the adjuvant therapy (7). However, a survival plateau of 60% to 70% has apparently been reached, above which it seems difficult to advance with the available diagnostic and therapeutic armamentarium (712). Furthermore, patients presenting with overt metastases still have a dismal prognosis despite aggressive combination chemotherapy, and for this group, new treatment strategies are needed (1317).

Because osteosarcoma displays considerable heterogeneity in metastatic capacity and chemosensitivity (7, 18), it might be possible to improve the results further by more individualized treatment. Unfortunately, it has thus far not been possible at the time of primary treatment to identify patients belonging to the different risk groups, and improved methods are needed for initial staging of the disease, for early detection of relapse, and for better means to predict response to the effect of the highly toxic combination chemotherapy. One possible step in this direction would be to assess the presence of tumor cells in bone marrow and peripheral blood, before and after chemotherapy. Adding detection of such micrometastatic cells to the known prognostic factors might provide a new biological staging system for guiding “risk adapted therapy.”

For decades it has been a goal to identify micrometastases in osteosarcoma patients. In two theses from The Mayo Clinic, a tritiated thymidine labeling method was used (19, 20), and researchers at our institution have used a technique employing Milliporefilters in the vein draining the primary tumor (21, 22). More recently, several methods have been used to detect micrometastatic tumor cells in various types of cancer and to show relationships to established clinical variables (2330). We have used an immunomagnetic method (3133) permitting rapid isolation and detection of tumor cells present in samples of peripheral blood and bone marrow aspirates from cancer patients. Here we report the results obtained in a single institution study on 60 patients with suspected bone sarcomas, of whom 49 had osteosarcoma.

Patients. Sixty patients admitted to our hospital were included in the study following informed consent. They were either in a primary diagnostic work-up for suspected bone sarcoma, or presenting with recurrent osteosarcoma during follow-up, after having completed neoadjuvant combination chemotherapy and surgery (see http://www.ssg-org.net for clinical protocols SSG-VIII, ISG/SSG-I&II describing chemotherapy combinations given to all high-grade osteosarcoma). Forty-nine of the patients had osteosarcoma (Table 1), six patients had Ewing sarcoma, two had malignant fibrous histiocytomas in bone, and one had a chondrosarcoma. Two patients turned out to have benign lesions: one with aneurysmal bone cysts and one with localized osteomyelitis.

Table 1.

Patient and tumor variables and the presence of bone marrow micrometastases

Osteosarcoma subtypePatient no.GenderGradePhenotypeStatusImmunomagnetic detection
TP-39.2.27
Extremity nonmetastatic ob CR − − 
 ob CR 2+ 2+ 
 mixed CR − − 
 10 ob CR 3+ 3+ 
 15 ob CR − − 
 19 ob DOD 
 22 ob CR − 
 23 ob CR 2+ 
 24 ob CR 
 26 ob CR − − 
 27 chondro CR − − 
 33 mixed CR 
 34 ob CR 2+ 2+ 
 35 mixed CR − − 
 41 ob CR − − 
 44 ob CR − − 
 45 ob CR2 2+ − 
 57 chondro CR − 2+ 
 58 ob CR − − 
 05 ob CR − − 
 08 telangio DOD 2+ 2+ 
 013 ob CR 2+ 2+ 
 012 ob DOD 2+ 2+ 
Axial nonmetastatic chond/fibro DOD − − 
 mixed DOD Not tested 
 18 ob DOD 2+ 
 43 chondro CR − − 
 47 ob DOD* − − 
Primary metastatic 11 ob CR 4+ 3+ 
 13 ob DOD 3+ 2+ 
 25 ob DOD 4+ 2+ 
 29 mixed DOD 4+ 2+ 
 30 mixed CR 2+ 2+ 
 37 telangio DOD 2+ 2+ 
 38 chondro DOD 3+ 
 39 fibro DOD − − 
 40 mixed DOD 2+ 2+ 
 50 chondro DOD 4+ 
 01 ob DOD 3+ 3+ 
 06 ob DOD 3+ 2+ 
Recurrent disease ob DOD 2+ 2+ 
 chondro DOD 2+ 
 ob DOD 4+ 3+ 
 12 ob DOD* − − 
 17 ob DOD 
 20 fibro CR2 − − 
 09 fibro DOD − − 
 010 ob CR2 − − 
 011 ob DOD 3+ 3+ 
Osteosarcoma subtypePatient no.GenderGradePhenotypeStatusImmunomagnetic detection
TP-39.2.27
Extremity nonmetastatic ob CR − − 
 ob CR 2+ 2+ 
 mixed CR − − 
 10 ob CR 3+ 3+ 
 15 ob CR − − 
 19 ob DOD 
 22 ob CR − 
 23 ob CR 2+ 
 24 ob CR 
 26 ob CR − − 
 27 chondro CR − − 
 33 mixed CR 
 34 ob CR 2+ 2+ 
 35 mixed CR − − 
 41 ob CR − − 
 44 ob CR − − 
 45 ob CR2 2+ − 
 57 chondro CR − 2+ 
 58 ob CR − − 
 05 ob CR − − 
 08 telangio DOD 2+ 2+ 
 013 ob CR 2+ 2+ 
 012 ob DOD 2+ 2+ 
Axial nonmetastatic chond/fibro DOD − − 
 mixed DOD Not tested 
 18 ob DOD 2+ 
 43 chondro CR − − 
 47 ob DOD* − − 
Primary metastatic 11 ob CR 4+ 3+ 
 13 ob DOD 3+ 2+ 
 25 ob DOD 4+ 2+ 
 29 mixed DOD 4+ 2+ 
 30 mixed CR 2+ 2+ 
 37 telangio DOD 2+ 2+ 
 38 chondro DOD 3+ 
 39 fibro DOD − − 
 40 mixed DOD 2+ 2+ 
 50 chondro DOD 4+ 
 01 ob DOD 3+ 3+ 
 06 ob DOD 3+ 2+ 
Recurrent disease ob DOD 2+ 2+ 
 chondro DOD 2+ 
 ob DOD 4+ 3+ 
 12 ob DOD* − − 
 17 ob DOD 
 20 fibro CR2 − − 
 09 fibro DOD − − 
 010 ob CR2 − − 
 011 ob DOD 3+ 3+ 

NOTE: Histologic grades 1 and 2, osteosarcoma of low malignancy; grades 3 and 4, high-grade osteosarcoma. The dominating phenotypes of differentiation are as follows: ob, osteoblastic/sclerotic; chondro, chondroblastic; fibro, fibroblastic; telangio, telangioectatic; mixed, several of the elements present.

Abbreviations: CR, complete remission; CR2, complete remission after post-relapse chemotherapy.

*

Dead of loco-regional tumor growth, without metastases.

The mean age of the 49 patients; 23 females and 26 males with osteosarcoma was 23 years (range, 6-51 years). The histologic osteosarcoma subtypes were osteoblastic/sclerotic in 30 cases, chondroblastic in seven, fibroblastic in three, telangiectatic in two, and a mixed pattern in the remaining seven cases (Table 1). One of the patients had a low-grade osteosarcoma (histologic grade 2 on a four-grade scale). Among the 28 patients without overt metastases, 23 had their primary tumor localized in an extremity, the three most common sites being distal femur (9), proximal humerus (5), and proximal tibia (4).

In 15 of the 22 patients with nonmetastatic high-grade extremity localized osteosarcoma, who received preoperative chemotherapy, the histologic response to preoperative chemotherapy was assessed. The grading was done according to Huvos.

One patient had a pelvic tumor and four had craniofacial localization. Twelve patients had overt metastases at primary diagnosis, nine with lung metastases only, whereas one patient also had multiple skeletal lesions. One patient had both lymph node manifestations and lung metastases, whereas one had bony metastases only.

The remaining nine samples were obtained from patients with relapsed osteosarcoma. Five of these had lung metastases only; three had both pulmonary and skeletal metastases, whereas one had a local relapse, lymph node metastases and also malignant ascites.

Serving as negative controls, samples were obtained from 18 patients with nonsarcoma malignancies and from nine healthy donors.

Sample collection. Both peripheral blood and bone marrow samples were drawn from all patients. Bone marrow (10-15 mL) was aspirated from a single site of the posterior iliac crest and 20-mL peripheral blood was obtained by antecubital venopuncture. The bone marrow sampling was done either in local anesthesia or immediately before biopsy when the patients were under general anesthesia.

The collected cell samples were separated by density gradient centrifugation (Lymphoprep, Medinor, Oslo, Norway). Mononuclear cells from the interphase layer were washed in PBS with 1% human serum albumin and the number of mononuclear cells was counted in an automated cell counter. Typically, 1 to 2 × 108 mononuclear cells were recovered from bone marrow and 5 to 9 × 107 from peripheral blood. The mononuclear cell fraction was used for immunomagnetic cell isolation and in 22 of the cases for preparation of cytospins subsequently studied by immunocytochemistry.

Monoclonal antibodies. Two monoclonal antibodies (mAb) were used for immunomagnetic isolation. TP-3 binds to an epitope on an osteosarcoma-associated cell surface antigen with homology to the bone izoenzyme of alkaline phosphatase (3437). The high affinity mAb 9.2.27 (obtained from Dr. R. Reisfeld, Scripps Research Institute, La Jolla, CA), originally developed against melanoma (38), recognizes an epitope on the high molecular weight melanoma-associated antigen. 9.2.27 has been shown to bind also some subgroups of sarcoma, including osteosarcoma (39). Both mAbs have previously been shown nonreactive with mononuclear cells in peripheral blood and bone marrow from normal donors (34, 39) and react poorly with fibroblastically differentiated osteosarcoma.4

4

Unpublished results.

In addition, the MRK-16 mAb (Alexis Corp., Kamiya Biomedical Co., Seattle, WA) binding the multidrug resistance-associated p-glycoprotein 170 (MDR1), and the anti-melanoma mAb EP-1 (kindly provided by Dr. P.G. Natali, Regina Elena Cancer Institute, Rome, Italy), were chemically conjugated to 2-μm fluorescent latex microspheres (see below). Immunocytochemistry was done on cytospins from mononuclear cells suspensions using APAAP technique (DAKO, Copenhagen, Denmark). The cytospins, each containing 5 × 105 mononuclear cells, were air-dried and kept frozen at 70°C before incubation with 9.2.27 and TP-3.

Immunomagnetic isolation and detection of tumor cells. We have previously presented a rapid and simple procedure for immunomagnetic detection of cancer cells in samples of peripheral blood lymphocytes and bone marrow from patients with various forms of cancer (31, 33). The sensitivity of this method is approximately two target cells in 2 × 107 mononuclear cells (32), depending on the affinity of the monoclonal antibody used and the number of antigen epitopes expressed in a particular target cell population. Superparamagnetic monodisperse particles with a diameter of 4.5 μm, coated with polyclonal sheep anti-mouse IgG (Dynabeads SAM450, Dynal, Oslo, Norway), were preincubated with one of the tumor-associated mAbs and washed before the isolation procedure (Fig. 1). Typically, 60 μg of purified mAb was added to 30 mg (4 × 108 beads) of Dynabeads. SAM450 Dynabeads alone were used as negative control. Briefly, 2 × 107 isolated mononuclear cells were resuspended in 1 mL PBS with 1% human serum albumin in a plastic tube, and immunobeads were added in a concentration of 0.5:1 to the total number of cells. The mixture was incubated under rotation for 30 minutes at 4°C (Figs. 1 and 2A). After incubation, the cells were diluted with PBS + 1% human serum albumin and the tube was put in a magnet holder (Dynal). Cells reactive with the mAbs bind the beads as rosettes, and cell-bead rosettes are trapped on the wall of the test tube. The supernatant, containing unbound cells, was decanted. The remaining positive fraction in a volume of ∼200 μL was placed on ice for microscopic detection of rosetted cells (Fig. 2B). Twenty-microliter aliquots of the positive fraction were transferred to a counting chamber and the microscopic evaluation was done using a Zeiss Axioscope (Carl Zeiss, Jena, Germany) equipped with appropriate illumination and filters for the subsequent detection of fluorescent microspheres. A sample containing at least two cells with five or more TP-3 or 9.2.27 beads attached as a rosette was regarded as positive.

Fig. 1.

Schematic illustration of the immunorosetting technique Immunomagnetic beads coated with an antibody targeting tumor-associated antigens are added to mononuclear cells isolated from bone marrow or peripheral blood samples. After incubation at 4°C for 30 minutes, rosetted cells with iron-containing beads attached to the cell surface are isolated by the use of a magnet and 20 μL fractions of an enriched cell suspension of ∼200 μL are examined by microscopy.

Fig. 1.

Schematic illustration of the immunorosetting technique Immunomagnetic beads coated with an antibody targeting tumor-associated antigens are added to mononuclear cells isolated from bone marrow or peripheral blood samples. After incubation at 4°C for 30 minutes, rosetted cells with iron-containing beads attached to the cell surface are isolated by the use of a magnet and 20 μL fractions of an enriched cell suspension of ∼200 μL are examined by microscopy.

Close modal
Fig. 2.

Example of rosetted tumor cells from bone marrow before and after magnetic enrichment. A, a microscopic picture of a mononuclear cell suspension prepared from a bone marrow sample after incubation with antibody-coated Dynabeads but before magnetic enrichment. Free beads and a majority of normal mononuclear cells are seen together with a forming cell-bead rosette. B, similar picture examined after magnetic isolation. Only free Dynabeads and two cell-bead rosettes are seen.

Fig. 2.

Example of rosetted tumor cells from bone marrow before and after magnetic enrichment. A, a microscopic picture of a mononuclear cell suspension prepared from a bone marrow sample after incubation with antibody-coated Dynabeads but before magnetic enrichment. Free beads and a majority of normal mononuclear cells are seen together with a forming cell-bead rosette. B, similar picture examined after magnetic isolation. Only free Dynabeads and two cell-bead rosettes are seen.

Close modal

Characterization and growth of isolated cells. Fluorescent latex microparticles (Molecular Probes Europe, Leiden, the Netherlands) were conjugated with different antibodies and used in a double staining procedure to show that cells magnetically selected with SAM450 beads coated with either TP-3 or 9.2.27 also bound latex particles with antibodies targeting one of the other epitopes/antigens, thus providing additional evidence that the rosetted cells were indeed tumor cells (Fig. 3A-B). In eight cases, from which a high number of rosetted cells were recovered from bone marrow, attempts were made to cultivate the cells in vitro under standard conditions using RPMI 1640 containing 10% FCS and l-glutamine (Fig. 4A). Moreover, in five of these cases isolated bead-rosetted cells were used in attempts to establish in vivo growth by injecting the cells s.c. into the flanks of BALB/c nude mice (Fig. 4B-C).

Fig. 3.

Double staining of rosettes with fluorescent latex beads. A, picture of an osteosarcoma cell isolated from a bone marrow sample, coexpressing the osteosarcoma-associated antigen, shown by the binding of TP-3/Dynabeads (beads with yellow center), and the HMW proteoglycan, visualized by the bound 9.2.27 latex particles (blue fluorescence). B, the expression on a micrometastatic tumor cell of three different antigen epitopes. 9.2.27-coated magnetic Dynabeads (yellow center), and MRK-16 (green fluorescence), and EP-1 (red fluorescence) latex particles are all bound to the cell surface.

Fig. 3.

Double staining of rosettes with fluorescent latex beads. A, picture of an osteosarcoma cell isolated from a bone marrow sample, coexpressing the osteosarcoma-associated antigen, shown by the binding of TP-3/Dynabeads (beads with yellow center), and the HMW proteoglycan, visualized by the bound 9.2.27 latex particles (blue fluorescence). B, the expression on a micrometastatic tumor cell of three different antigen epitopes. 9.2.27-coated magnetic Dynabeads (yellow center), and MRK-16 (green fluorescence), and EP-1 (red fluorescence) latex particles are all bound to the cell surface.

Close modal
Fig. 4.

Verification of the malignant phenotype. A, microscopy at passage 20 of an in vitro culture of cells initially seeded as cell-bead rosettes. B and C, Similarly rosetted cells obtained from a patient with a high number of osteosarcoma cells in bone marrow were inoculated into the flank of a nude mouse. The pictures show histologic sections of a developing tumor xenograft: (B) H&E staining illustrates malignant growth and invasion of skeletal musculature. Magnification, × 100. (C), higher magnification of the central part of the tumor shows abundant osteoid formation and mitoses. Magnification, × 400.

Fig. 4.

Verification of the malignant phenotype. A, microscopy at passage 20 of an in vitro culture of cells initially seeded as cell-bead rosettes. B and C, Similarly rosetted cells obtained from a patient with a high number of osteosarcoma cells in bone marrow were inoculated into the flank of a nude mouse. The pictures show histologic sections of a developing tumor xenograft: (B) H&E staining illustrates malignant growth and invasion of skeletal musculature. Magnification, × 100. (C), higher magnification of the central part of the tumor shows abundant osteoid formation and mitoses. Magnification, × 400.

Close modal

Immunomagnetic enrichment. To illustrate the advantage of this method for detecting tumor cells rosetted with antibody-coated magnetic beads, a suspension of mononuclear cells from a patient sample was admixed with TP-3-coated beads and studied before and after the use of the magnetic device. As can be seen in Fig. 2, the high number of normal cells present initially (Fig. 2A) was in this case reduced to almost none after enrichment (Fig. 2B).

Presence of micrometastatic cells: relation to clinical data. Altogether, 31 of 49 (63%) osteosarcoma patients were found to have tumor cells in bone marrow (Table 1), of which only four (8%) were positive in peripheral blood. The bone marrow–positive samples included 13 of 23 (57%) patients without overt metastases with the primary tumor in one of the extremities (only one of these was positive in peripheral blood) and two of five of those with axial tumors. As many as 11 of 12 (92%) of the patients with overt metastases at primary diagnosis had positive bone marrow samples. Five of nine patients (56%) with metastatic relapse were positive at the time of examination.

The total number of tumor cells present in the immunomagnetically selected aliquot (final volume of 200 μL, isolated from ∼2 × 107 mononuclear cells) in each positive sample varied between 5 and 104 (Table 1). Less than 10 rosettes (+) were found in six samples, 14 contained between 10 and 100 cells (2+), six had in the range of 101 to 500 cells (3+), and five had >500 cell-bead rosettes (4+).

Four of the 12 (33%) positive patients with high-grade nonmetastatic extremity tumors have thus far relapsed, compared with none of the 10 bone marrow–negative patients (Fig. 5). One patient with a low-grade osteosarcoma (patient 7 in Table 1) did not receive adjuvant chemotherapy and was excluded from this analysis. With a mean follow-up of 75 months (range, 27-177 months), the P value comparing the survival curves for bone marrow–positive and –negative patients nearly reached significance (log-rank, P = 0.06) with metastasis-free survival of 65% and 100%, respectively. Three of the positive patients (patients 19, 08, and 012 in Table 1) relapsed after, 10, 27, and 41 months respectively, and are all dead of disease (DOD), whereas one patient (patient 45) relapsed after 30 months but is in second complete remission after complete surgical removal of lung metastases. Histology response data was available for 15 patients with extremity localized, nonmetastatic disease and who had received neoadjuvant chemotherapy. Eight showed a good histologic response, and only one (patient 45) has relapsed. Two of the seven poor responders, both bone marrow positive, are DOD, whereas none of the three bone marrow–negative patients with poor histologic response have relapsed.

Fig. 5.

Metastasis-free survival of high-grade nonmetastatic osteosarcoma patients with extremity-localized tumors.

Fig. 5.

Metastasis-free survival of high-grade nonmetastatic osteosarcoma patients with extremity-localized tumors.

Close modal

The two positive patients with craniofacial localization of their primary tumor are both DOD. One of the negative patients is alive with no evidence of disease, one died of uncontrollable local relapse that invaded the brain, whereas the third, a patient with pelvic osteosarcoma and a chondroblastic/fibroblastic differentiation, is dead of lung metastases.

Among the 12 patients with overt metastases at the time of initial osteosarcoma diagnosis, 11 (92%) had tumor cells in bone marrow. The samples in this group contained the highest number of tumor cells (range, 200-10,000/2 × 107 mononuclear cells). Two bone marrow–positive patients, both negative in peripheral blood, are still alive with no evidence of disease (following complete surgical removal of all palpable lung metastases and intensified post-operative chemotherapy; in one patient with PBSC support). The one patient negative for bone marrow tumor cells is also DOD. She had a fibroblastic differentiated osteosarcoma, most probably nonreactive with the antibodies used for rosetting. The remaining nine bone marrow–positive patients in this group are all DOD, including the three patients that had rosettes also in peripheral blood.

Five of the nine (56%) patients with relapsed metastatic disease at inclusion had positive bone marrow samples are all DOD. Interestingly, two of the four bone marrow–negative patients had several years of disease-free interval between primary treatment and the time of hospital admittance for surgical removal of lung metastases, at which time the test was done. Both are alive with no evidence of disease at 6 and 8 years of follow-up. A third negative patient also had a solitary lung metastasis at the time of bone marrow sampling. He later developed a local relapse in the posterior thoracic wall and subsequently died due to meningitis caused by an uncontrollable local tumor invasion into the spine and mediastinum. The last bone marrow–negative patient is DOD. He had a recurrent osteosarcoma with a purely differentiated fibroblastic phenotype, presumably with tumor cells not expressing the epitopes recognized by the mAbs here used.

None of six patients with Ewing sarcoma, two with primary MFH in bone and one with chondrosarcoma, had any tumor cells detected in their bone marrow and peripheral blood samples with the two mAbs here used. In addition, one patient diagnosed with an aneurysmal bone cyst and another with a chronic osteomyelitis were both negative. In all these 11 cases, osteosarcoma was suspected at inclusion and the correct diagnosis was not known at the time of immunomagnetic detection.

Control bone marrow and peripheral blood samples from nine healthy donors and 18 patients with nonsarcoma malignancies (10 with breast or prostate carcinomas, five with malignant melanoma, and three with non-Hodgkin's lymphoma) were in all cases negative, further showing the specificity of the method used. The samples from nonsarcoma patients were obtained before start of chemotherapy.

Verification of the nature of immunomagnetically isolated cells. To confirm the osteosarcoma nature of the immunomagnetically rosetted cells, we used chromophore-containing latex microparticles with surface-bound antibodies relevant to the malignant phenotype but different from the one used for magnetic isolation. As shown in Fig. 3A, such particles (2 μm, blue fluorescence) conjugated with 9.2.27 antibody bound to the rosetted cells isolated from a patient sample with TP-3-coated SAM450 Dynabeads (beads with yellow center). Figure 3B shows a tumor cell found in a sample from another patient that bound two different antibody-latex microparticle complexes (MRK16 on green fluorescent and EP-1 on red fluorescent 2-μm beads) in addition to the 9.2.27 antibody-Dynabeads used for enrichment. The findings confirm the specificity of our method and show that this approach can be used to study the expression of clinically relevant antigens on the surface of micrometastatic tumor cells.

A final proof of the malignant phenotype of isolated micrometastatic cells is in most cases impossible. However, among eight attempts to cultivate isolated rosetted cells in vitro, a long-term primary culture was established in two cases with morphology compatible with tumor cells. In one of these, the cells could be subcultured for about 4 months before they died, whereas in the other case the cells could be established as a permanent cell line, OS25 (Fig. 4A). The bone marrow sample from which these cells were isolated was obtained from a patient with primary metastatic disease who later died of osteosarcoma.

In five cases where numerous rosetted cells were isolated from bone marrow, bead/cell complexes were inoculated s.c. into the flank of BALB/c nude mice. In two of these cases, a tumor developed, but only one of them was successfully established as a xenograft line, presently in passage 36. The xenograft originates from the same patient as the cells used to establish the in vitro OS25 cell line. Histologic examination of the xenograft revealed that the tumor had retained its osteosarcoma morphology (Fig. 4B-C).

Comparison with immunocytochemistry. Cytospins of bone marrow mononuclear cells from altogether 22 patients were examined. In 13 cases immunomagnetically positive with the 9.2.27 antibody, only three were immunocytochemistry positive using the same antibody (data not shown). Two patients with Ewing sarcoma were negative with both methods. Of eight IMD-positive cases using the TP-3 antibody, six stained positive for tumor cells with immunocytochemistry. Notably, with IMD, the calculated number of detected tumor cells was in the range of 5 to 10,000 cells/2 × 107 mononuclear cells, whereas in three immunocytochemistry-positive cases, only one single cell was in each case found after screening four cytospin each with 5 × 105 mononuclear cells. With the TP-3 antibody, about the same number of IMD-detected cells was found as with 9.2.27, but with immunocytochemistry the results differed somewhat, with from 2 to >100 immunocytochemistry positively stained cells.

High-grade osteosarcoma is considered a systemic disease characterized by subdetectable tumor spread at primary diagnosis in a majority of the patients, and adjuvant chemotherapy together with aggressive surgery is today considered standard management of the disease. Polydrug chemotherapy in osteosarcoma is very intense and entails considerable risk of acute and long-term toxicity. One of our aims is to identify groups of osteosarcoma patients with different likelihood of benefiting from therapeutic interventions with systemic therapy (7), with the objective to be able to better tailor the chemotherapy given to this heterogeneous group of patients. In this disease, adjuvant chemotherapy is given with the intention of either eradicating occult, micrometastatic tumor cells, thereby preventing subclinical disease to develop into overt metastases, or to delay the time to relapse (812). The objective of the present study was to explore the validity and putative clinical usefulness of an immunomagnetic method to detect the presence of micrometastatic tumor cells in peripheral blood and bone marrow of patients with osteosarcoma.

In several other tumor types, evidence for a relationship between the presence of micrometastatic cells in peripheral blood or bone marrow and outcome has been reported (2630, 33, 40), but the biological and clinical implications of such data are not yet clear (30, 31). In some other types of sarcoma genetic markers exist that can be used in micrometastasis detection, and in patients with Ewing tumor the presence of such cells was associated with poor prognosis (4042). In osteosarcoma, however, no report has to our knowledge been published using any of the most commonly applied methods, and there is no known genetic marker consistently expressed in osteosarcoma that could serve as a target for PCR-based techniques (43).

Antibodies that bind to osteosarcoma but not to normal hematopoietic cells are available and can be applied in immunologic methods (3439). The immunomagnetic, positive selection approach used here has several advantages. It allows rapid (2-3 hours per test) routine screening of as much as 2 × 107 mononuclear cells (3032). This implies high sensitivity compared with other methods, which is supported by the finding of a much lower number of positive samples and number of tumor cells when assessed by immunocytochemistry. Such sensitivity differences have also been seen in other studies where both methods were systematically studied (3133). The specificity of the method is here shown by the finding that all 27 control nonosteosarcoma samples tested were negative, as were all bone marrow samples from patients suspected to have osteosarcoma at the time of examination but who were later found to suffer from other bone tumors or diseases. The method has also the advantage that live tumor cells are isolated and can be used in further in vitro and in vivo characterization studies.

The frequency of positive bone marrow samples found in this study was high, with altogether 31 of 49 (63%) of the patients having micrometastatic tumor cells in bone marrow compared with only four (8%) in peripheral blood. The data presented are remarkable in that a high fraction of bone marrow samples were positive already at diagnosis, both in patients without (54%) and with (92%) metastatic disease detectable with conventional diagnostic procedures. The difference between these two groups indicates that in osteosarcoma micrometastatic bone marrow disease reflects both the aggressiveness and dissemination of the disease.

Most importantly, a strong relationship between the bone marrow findings and the clinical data was found with regard to disease progression. A higher fraction of patients with micrometastatic cells in bone marrow have relapsed compared with those with negative samples, particularly evident in patients with high-grade extremity osteosarcoma and no detectable metastases at diagnosis. Thus, after a mean follow-up time of 75 months, 4 of 12 positive patients have relapsed compared with 0 of 10 negative patients (Fig. 5). It should also be noted that some of the positive patients still alive have, after bone marrow examination, been successfully treated with surgery of lung metastases and chemotherapy. Moreover, some bone marrow–negative patients had a tumor phenotype possibly not expressing the antigens recognized by the antibodies used. The results suggest that micrometastasis detection with our method can potentially be useful in prognostication and in clinical decision-making in osteosarcoma. This possibility will be examined in prospective studies in a larger cohort of patients.

That patients examined at the time of recurrence had a relatively low fraction of positive bone marrow (56%) suggests a possibility not previously discussed; that is, that the presence of tumor cells in bone marrow may depend not only the disease stage and clinical aggressiveness of the tumor but also on the site of growth of metastases. Clinically, this observation fits with the fact that patients with relapsed osteosarcoma can obtain long-time survival and even cure with radical surgery alone. Moreover, the effect of systemic treatment at relapse is not proven in well-designed trials and micrometastasis status could become a parameter for selecting patients for systemic second-line chemotherapy.

In publications on the clinical relevance of micrometastatic cells in different tumor types, associations to clinical variables have been based on small numbers of detected cells with immunocytochemistry and/or on the level of mRNA transcripts of tumor-associated genes expressed, calculated from the intensity of signals obtained with reverse transcription-PCR methods. In this study, we found a high number of live cells rosetted with magnetic beads coated with osteosarcoma-associated antibodies. In several samples the nature of the selected cells could also be further confirmed by binding of additional, fluorescent particles with antibodies recognizing other tumor-associated antigens, and in individual cases it was possible to cultivate the bead-rosetted cells in vitro and even as a xenograft in nude mice. These results provide strong evidence in support of the validity of our method. The use of fluorescent antibody-particle complexes can be further extended to examine the expression of clinically and biologically relevant proteins expressed on the surface of micrometastatic tumor cells, such as antigens used as targets for antibody-based therapies and vaccines, or prognostic markers such as HER-2/neu and MDR (4146).

MDR expression on cells in the primary tumor was initially reported to be a biological marker and an independent indicator for poor prognosis in osteosarcoma (47), but later studies have not confirmed this (43, 44). It may be more meaningful to study MDR expression directly on the micrometastatic cells. Two of seven of our samples studied for MDR expression on isolated cells were positive with the fluorescent bead technique (Fig. 3B), and both patients who are dead of disease.

Molecular characterization of micrometastatic tumor cells may provide important biological and clinical information. In most immunocytochemical studies, however, commonly only one to two tumor cells are detected (2325), making molecular studies difficult and without obvious representativity of the total population of micrometastatic cells. Our method yields a much higher number of isolated live cancer cells, facilitating molecular studies. In tumor types with a limited number of target cells found in bone marrow, molecular studies may, however, require amplification procedures (41) to be used for microarray studies. In osteosarcoma the high number of tumor cells found in many samples makes such amplification unnecessary.

In osteosarcoma, the search for better and additional markers to identify subgroups of patients may be particularly useful. The ultimate aim is to help individualize therapy for this group of mostly adolescent patients, giving early aggressive therapy to patients who may benefit from it, and avoid toxic adjuvant therapies to those that do not. The approach here described may represent a new, valuable tool in this direction.

Grant support: Norwegian Cancer Society.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Ø. Fodstad is presently at the Cancer Research Institute, University of South Alabama, Mobile, Alabama. Phone: 251-460-6995; Fax: 251-460-6994; E-mail: [email protected].

We thank Professor Jahn Nesland (Department of Pathology, The Norwegian Radium Hospital) for performing and scoring the samples studied by immunocytochemistry.

1
Malawer MM, Link MP, Donaldson SS. Sarcomas of bone. In: Devita VT, Hellman S, Rosenberg SA, editors. Cancer: principles and practice of oncology. 5th ed. Philadelphia: Lippincott-Raven Publishers; 1997. p. 1789–852.
2
Souhami R, Cannon SR. Osteosarcoma. Oxford textbook of oncology. In: Peckham M, Pinedo HM, Veronesi U, editors. Oxford: Oxford University Press; 1995. p. 1969–76.
3
Cade S. Osteogenic sarcoma: a study based on 133 patients.
J R Coll Surg Edinb
1955
;
1
:
79
–111.
4
Dahlin D, Unni K. Osteogenic sarcoma of bone and its important recognizable varieties.
Am J Surg Pathol
1977
;
1
:
61
–72.
5
Huvos AG. Osteogenic sarcoma. In: Huvos AG, editor. Bone tumors. Diagnosis, treatment and prognosis. Philadelphia: WB Saunders; 1991. p. 85–155.
6
Harter KW. Osteosarcoma and less common sarcomas of childhood. In: Cassady JR, editor. Radiation therapy in pediatric oncology. Springer Verlag, Berlin 1994. p. 305–17.
7
Bruland ØS, Pihl A. On the current management of osteosarcoma. A critical evaluation and a proposal for a modified treatment strategy.
Eur J Cancer
1997
;
33
:
1725
–31.
8
Rosen G, Marcove RC, Caparros B. Primary osteogenic sarcoma: the rationale for preoperative chemotherapy and delayed surgery.
Cancer
1979
;
43
:
2163
–77.
9
Meyers PA, Heller G, Healey J, et al. Chemotherapy for nonmetastatic osteogenic sarcoma: the Memorial Sloan-Kettering experience.
J Clin Oncol
1992
;
10
:
5
–15.
10
Sæter G, Alvegård TA, Elomaa I, et al. Treatment of osteosarcoma of the extremities with the T-10 protocol, with emphasis on the effects of preoperative chemotherapy with single agent high-dose methotrexate. A Scandinavian Sarcoma Group Study.
J Clin Oncol
1991
;
9
:
1766
–75.
11
Bacci G, Picci P, Ferrari S, et al. Primary chemotherapy and delayed surgery for nonmetastatic osteosarcoma of the extremities: results in 164 patients preoperatively treated with high doses of methotrexate followed by cisplatin and doxorubicin.
Cancer
1993
;
72
:
3227
–38.
12
Winkler K, Bielack S, Delling G, et al. Effect of intraarterial versus intravenous cisplatin in addition to systemic doxorubicin, high-dose methotrexate, and ifosfamide on histologic tumor response in osteosarcoma (study COSS-86).
Cancer
1990
;
66
:
1703
–10.
13
Roth JA, Putnam JBJ, Wesley MN, Rosenberg SA. Differing determinants of prognosis following resection of pulmonary metastases from osteogenic and soft tissue sarcoma patients.
Cancer
1985
;
55
:
1361
–6.
14
Beattie EJ, Harvey JC, Marcove R, Martini N. Results of multiple pulmonary resections for metastatic osteogenic sarcoma after two decades.
J Surg Oncol
1991
;
46
:
154
–5.
15
van Rijk Zwikker GL, Nooy MA, Taminiau A, Kappetein AP, Huysmans HA. Pulmonary metastasectomy in patients with osteosarcoma.
Eur J Cardiothorac Surg
1991
;
5
:
406
–9.
16
Meyers PA, Heller G, Healey JH, et al. Osteogenic sarcoma with clinically detectable metastasis at initial presentation.
J Clin Oncol
1993
;
11
:
449
–53.
17
Bacci G, Briccoli A, Picci P, et al. Osteosarcoma of the extremities metastatic at presentation: results obtained with primary chemotherapy followed by simultaneous resection of the primary and metastatic lesions.
Cancer J
1990
;
3
:
213
–8.
18
Sæter G, Bruland ØS, Follerås G, Boysen M, Høie J. Extremity and non-extremity high grade osteosarcoma. The Norwegian Radium Hospital experience during the modern chemotherapy era.
Acta Oncol
1996
;
35
:
129
–34.
19
Carlson MJ. Circulating sarcoma cells: The incidence of 3Thymidine labeling in the peripheral blood of normal and sarcoma patients. Thesis: University of Minnesota, 1978.
20
Kaiser TE. The detection of tritiated thymidine labeled cells in the peripheral blood of sarcoma patients and the nature of these cells bearing prognostic significance for sarcoma patients. Thesis: University of Minnesota, 1985.
21
Foss PO, Messelt OT, Efskind J. Isolation of cancer cells from blood and thoracic duct lymph by filtration.
Surgery
1963
;
53
:
241
–6.
22
Foss PO, Brennhovd IO, Messelt OT, Efskind J, Liverud K. Invasion of tumor cells into the bloodstream caused by palpation or biopsy of the tumor.
Surgery
1966
;
59
:
691
–5.
23
Beiske K, Myklebust AT, Aamdal S, et al. Detection of bone marrow metastases in small cell lung cancer patients. Comparison of immunological methods.
Am J Pathology
1992
;
141
:
531
–8.
24
Mansi JL, Gogas H, Bliss JM, et al. Outcome of primary-breast-cancer patients with micrometastases: a long-term follow-up study.
Lancet
1999
;
354
:
197
–202.
25
Pantel K, Cote RJ, Fodstad Ø. Detection and clinical importance of micrometastatic disease.
J Natl Cancer Inst
1999
;
91
:
1113
–24.
26
Wiedswang G, Borgen E, Kaaresen R, et al. Detection of isolated tumor cells in bone marrow is an independent prognostic factor in breast cancer.
J Clin Oncol
2003
;
21
:
3469
–78.
27
Pantel K, Muller V, Auer M, et al. Detection and clinical implications of early systemic tumor cell dissemination in breast cancer.
Clin Cancer Res
2003
;
9
:
6326
–34.
28
Ghossein RA, Bhattacharya S. Molecular detection and characterisation of circulating tumour cells and micrometastases in solid tumours.
Eur J Cancer
2000
;
36
:
1681
–94.
29
Taback B, Chan AD, Kuo CT, et al. Detection of occult metastatic breast cancer cells in blood by a multimolecular marker assay: correlation with clinical stage of disease.
Cancer Res
2001
;
61
:
8845
–50.
30
Funke I, Schraut W. Meta-analyses of studies on bone marrow micrometastases: an independent prognostic impact remains to be substantiated.
J Clin Oncol
1998
;
16
:
557
–66.
31
Marth C, Kisic J, Kærn J, Tropé C, Fodstad Ø. Circulating tumor cells in the peripheral blood and bone marrow of patients with ovarian cancer do not predict prognosis.
Cancer
2002
;
94
:
846
–52.
32
Flatmark K, Bjørnland K, Johannessen HO, et al. Immunomagnetic detection of micrometastatic cells in bone marrow of colorectal cancer patients.
Clin Cancer Res
2002
;
8
:
444
–9.
33
Faye R, Aamdal S, Høifødt HK, et al. Immunomagnetic detection and clinical significance of micrometastatic tumor cells in malignant melanoma patients.
Clin Cancer Res
2004
;
15
:
4134
–49.
34
Bruland ØS, Fodstad Ø, Funderud S, Pihl A. New monoclonal antibodies specific for human sarcomas.
Int J Cancer
1986
;
38
:
27
–31.
35
Bruland ØS, Fodstad Ø, Stenwig E, Pihl A. Expression and characteristics of a novel human osteosarcoma-associated cell surface antigen.
Cancer Res
1988
;
48
:
5302
–9.
36
Bruland ØS, Fodstad Ø, Skretting A, Pihl A. Selective radiolocalization of two radiolabelled anti-sarcoma monoclonal antibodies in human osteosarcoma xenografts.
Br J Cancer
1987
;
56
:
21
–5.
37
Bruland ØS, Aas M, Fodstad Ø, et al. Immunoscintigraphy of bone sarcomas. Results in five patients.
Eur J Cancer
1994
;
30
:
1484
–9.
38
Morgan AC Jr, Galloway DR, Reisfeld RA. Production and characterization of a monoclonal antibody to a melanoma specific glycoprotein.
Hybridoma
1981
;
1
:
27
–36.
39
Godal A, Bruland OS, Haug E, Aas M, Fodstad O. Unexpected expression of the 250 kD melanoma-associated antigen in human sarcoma cells.
Br J Cancer
1986
;
53
:
839
–41.
40
Fagnou C, Michon J, Peter M, et al. Presence of tumor cells in bone marrow but not in blood is associated with adverse prognosis in patients with Ewing's tumor.
J Clin Oncol
1998
;
16
:
1707
–11.
41
Athale UH, Shurtleff SA, Jenkins JJ, et al. Use of reverse transcriptase polymerase chain reaction for diagnosis and staging of alveolar rhabdomyosarcoma, Ewing sarcoma family of tumors, and desmoplastic small round cell tumor.
J Pediatr Hematol Oncol
2001
;
23
:
99
–104.
42
Schleiermacher G, Peter M, Oberlin O, et al. Increased risk of systemic relapse associated with bone marrow micrometastasis and circulating tumor cells in localized Ewing tumor.
J Clin Oncol
2003
;
21
:
85
–91.
43
Gorlick R, Anderson P, Andrulis I, et al. Biology of childhood osteogenic sarcoma and potential targets for therapeutic development: meeting summary.
Clin Cancer Res
2003
;
9
:
5442
–53.
44
Serra M, Scotlandi K, Reverter-Branchat G, et al. Value of P-glycoprotein and clinicopathologic factors as the basis for new treatment strategies in high-grade osteosarcoma of the extremities.
J Clin Oncol
2003
;
21
:
536
–42.
45
Valabrega G, Fagioloi F, Corso S, et al. ErbB2 and bone sialoprotein as markers for metastatic osteosarcoma cells.
Br J Cancer
2003
;
88
:
396
–400.
46
Serra M, Reverter-Branchat G, Maurici D, et al. Analysis of dihydrofolate reductase and reduced folate carrier gene status in relation to methotrexate resistance in osteosarcoma cells.
Ann Oncol
2004
;
15
:
151
–60.
47
Baldini N, Scotlandi K, Barbanti-Brodano G, et al. Expression of P-glycoprotein in high-grade osteosarcoma in relation to clinical outcome.
N Engl J Med
1995
;
333
:
1380
–5.