This multidisciplinary meeting was convened on November 29-30, 2001, at the Bethesda Marriott (Bethesda, MD), to address issues concerning the biology of childhood osteogenic sarcoma and the potential targets for therapeutic development. Various members of the scientific community from different clinical and laboratory-based disciplines were invited to participate.

General Introduction - Barry D. Anderson, M.D., Ph.D., and Richard Gorlick, M.D.

Dr. Richard Gorlick opened the meeting by addressing the goals and general objectives of the meeting as follows:
•   To review the current state of knowledge and share information regarding osteosarcoma biology
•   To identify, prioritize, and support the development of biology studies of potential clinical relevance in osteosarcoma
•   To discuss appropriate methods for analysis of osteosarcoma samples
•   To discuss tissue resources that may be available for the conduct of osteosarcoma biology studies (e.g., P9851 – Osteosarcoma Biology Protocol)

In the past, most osteosarcoma biology studies have been single institution-based and have studied a single gene or pathway. Dr. Gorlick described P9851 as a single biology study that serves as a companion to all osteosarcoma clinical studies and encompasses the work of numerous laboratories. The major accomplishment of this protocol has been the accrual of specimens within the entire Cooperative Group setting; however, the assays being performed are based on the interests of the individual investigators. Finite amounts of tissue exist, so the inclusion of new, additional assays may require the elimination of current assays. As the biological investigation is broadened, the resources are likely to remain finite, so less tissue may be available for all studies. Dr. Gorlick asked whether analyzing tissue was the most relevant study due to the possible artifact of sample heterogeneity and stromal contamination. He also asked if the correct sample procurement techniques were being utilized and whether there were any alternatives to the current techniques.

Dr. Gorlick concluded his introductory talk by listing the general objectives of each session as follows:
•   The potential clinical relevance of the biological work presented (potential prognostic factor, therapeutic target, complements clinical trial directions, etc.)
•   Whether the current laboratory effort dedicated to a particular biological investigation thus far is sufficient, excessive, or insufficient
•   The most appropriate method(s) that should be utilized to pursue the biological investigation
•   Identify specific plans for the conduct of research related to the biological topic

Pathogenesis of Osteosarcoma

p53/SV40 – David Malkin, M.D.

Dr. Malkin described work with p53 and SV40. Dr. Malkin specializes in the role of germ-line p53 mutations in cancer predisposition and potential association of p53 and oncogenic viruses. He described the major functions of the p53 tumor suppressor gene as:

1. modulation of cell cycle arrest at G1/S
2. induction of apoptosis
3. assisting in DNA repair
4. inhibition of blood vessel formation (antiangiogenesis)

P53 is activated through different pathways including:

1. DNA damage
2. aberrant growth signal from oncogenes
3. chemotherapeutic agents, UV light, nucleoside depletion, or protein kinase inhibitors (e.g. ATR and casein kinase II)

There are different mechanisms of inactivating p53 in human cancers, but his focus has been on mutations in the p53 gene, and viral infection. Dr. Malkin gave an overview of adenoviral-mediated p53 gene therapy. The rationale for investigating p53 gene therapy in osteosarcoma includes the knowledge that metastatic osteosarcoma is refractory to conventional therapy and that this drug-resistance can be associated with p53 mutations. He has hypothesized, as tested in four osteosarcoma cell lines (SaOS-2, HOS, KHOS/NP, MNNG), that the introduction of wild-type p53 into osteosarcoma cell lines inhibits cell growth, induces apoptosis, and sensitizes cells to chemotherapeutic agents. The studies concluded that the efficiency of adenoviral infection varies among the four cell lines in a dose-dependent manner. Ad-wtp53 (wild type p53), but not Ad-mutp53 or Ad-lacZ, treatment causes a decrease in cell viability and suppresses cell growth in a dose-dependent manner, induces apoptosis, and sensitizes osteosarcoma cell lines to cisplatin and, to a lesser degree, adriamycin. Future studies will:

1. determine changes in levels of Bax, p21, and VEGF after treatment (using ELISA and Western blot)
2. measure cell cycle changes after treatment (using FACS)
3. test the p53-induction system with other sarcomas
4. consider development of an in vivo model to examine the therapeutic potential of adenoviral p53 introduction, especially in metastatic lung disease (osteosarcoma tends to metastasize to the lung, and the adenovirus appears to have an affinity for respiratory tissue)

Dr. Malkin then discussed SV40, a DNA polyomavirus that consists of three major non-structural proteins: the small t antigen (itself is not sufficient to transform cells, but enhances the ability of large T antigen to transform cells), the smaller t antigen (function unknown), and the large T antigen (assists in viral replication, interacts with p53 and Rb, and inhibits p53 function, thus promoting cell proliferation). A 1996 study concluded that 11 of 18 osteosarcoma samples showed SV40 (Carbone, et al, Oncogene 13:527, 1996); a 1998 study showed no correlation of presence of SV40 with p53 or Rb mutation status (Mendoza et al, Oncogene 17:2457-62, 1998); and a 1997 study showed 50% of osteosarcomas tested showed SV40 in each of four regions of the viral genome (Lednicky, et al, Int. J. Cancer 72:791-800, 1997). Although SV40 is not observed in all tumors, it may exist in select families. It is postulated that SV40 large T antigen may inactivate the wild-type p53 allele in some cell types of susceptible individuals, resulting in tumor development. In 2 families with Li-Fraumeni syndrome, in which family members with cancer harbored germline heterozygous p53 mutations, SV40 large T antigen was amplified and expressed in tumors in which the wild-type p53 allele was retained. These included osteosarcoma and choroid plexus carcinoma, but not in rhabdomyosarcoma (Malkin et al, Oncogene 20: 4441-4449, 2001). In conclusion, Dr. Malkin expressed the need for: prospective studies of presence of SV40 in newly diagnosed patients; serologic analysis using newly developed ELISA assays that can differentiate between closely related potential etiologic viruses; PCR analysis using multiple primer sets of both tumor material and peripheral blood (buffy coat); Southern blot of material with adequate DNA; and correlation of virus status and other gene markers (e.g. p53/Rb).

Other Suppressors - Marc Hansen, Ph.D.

Dr. Hansen examined 32 osteosarcoma specimens to discern predictors of favorable outcome. He described a highly altered genome as common in osteosarcoma, and a number of genes that show LoH (3q, 13q, 17p, 18q). RB1, p53, and 3q have been suggested as having some predictability for outcome.

He looked further at 3q and identified EDR3, which is homologous to genes in the polycomb family. EDR3 binds to condensing chromatin. The polycomb genes regulate pRB by regulating p16. In a cancer model, EDR3 shows a shift in reversible growth arrest and irreversible growth arrest, and it is hypothesized that EDR3 regulates this shift. If this function is lost, p16 cannot be regulated and a tumor develops.

Dr. Hansen’s recent work with chromosome 18q (also associated with Paget’s Disease) examined RANK (a TNF-alpha family member) in the small region, 18q21.3. Various tests have confirmed that a normal copy of RANK is overexpressed in the nucleus of osteosarcoma cells. RANK activates the NFKappa-B pathway. RANKL (RANK Ligand) was also expressed in the nucleus, but was not overexpressed. RANKL is in the same region as RANK, but is usually found in the cytoplasm of normal cells. OPG expression in osteosarcoma cell lines was also nuclearized without being overexpressed. The RANK/RANKL is normal for growth stimulation and the decoy is OPG. RANK/RANKL also co-localize in the primary tumor nucleus; no other TNF-alpha does that. TRAIL, another TNF-alpha family member, was not found in the nucleus of osteosarcoma cells.

RANK is displaced to the cytoplasm with Ohs50 hybrids (need to define Ohs50), thus there is an upregulation and a nuclearization of RANK and TRAIL in osteosarcoma. Tumorigenesis also seems to correlate with nuclearization. Dr. Hansen discussed a model that demonstrates an upregulation of RANK and TRAIL, and nuclearization of RANK, RANK1, and OPG during tumorigenesis. More research is needed to decipher these findings.

Cytogenetics of Osteosarcoma - Julia A. Bridge, M.D.

Dr. Bridge began by stating that fresh tissue is needed for proper culture. Karotyping is both diagnostically useful and provides direction for molecular studies. Although karotyping is very difficult, the advantages do outweigh the limitations. A University of Nebraska Medical Center study analyzed 128 osteosarcoma specimens, of which 70% were clonally abnormal. Most are complex karotypes and have pronounced cell-to-cell variation or heterogeneity within the same patient. There are marker chromosomes in the majority of osteosarcomas (58%), and ring chromosomes are common. There is also evidence of genomic amplification (hsr or dmin) in at least 1/3 of cases.

Dr. Bridge recently collaborated with Swedish researchers to examine chromosome 12. Amplification of MDM2 was found in all low-grade and some high-grade tumors. In contrast to the low grade osteosarcoma, however, there was also an overrepresentation of chromosome 12p sequences in high-grade tumors, which may represent a pathway to aggressive osteosarcoma.

Cytogenetic abnormalities can be numerical or structural. Common numerical abnormalities in osteosarcoma include: gain of chromosome 1, loss of chromosome 9, 10, 13, and/or 17, and partial or complete loss of long arm of chromosome 6. Structural abnormalities include: certain common chromosomal regions that show structural rearrangements on chromosomes 11 and 19.

Dr. Bridge concluded by stating that despite the karyotypic complexity observed in osteosarcoma specimens, it is evident that specific chromosomal bands and/or regions are nonrandomly involved in osteosarcoma and may provide useful clinical information. Additional molecular investigations are needed to address the involvement of these recurrently involved chromosomes and chromosomal breakpoints.

Q/A Session

Q to Dr. Bridge: Were the high-grade tumors she studied originally low-grade tumors?

A: The tumors presented as high-grade tumors.

Q to Dr. Bridge: Were the karotypes representative of osteosarcoma in general?

A: They were representative of osteosarcoma in general.

Q to Dr. Bridge: Could a karotype be derived from every osteosarcoma sample?

A: She responded that when there is a lot of necrosis, the chance of successfully obtaining a result is less likely. The elapsed time between when the specimen is surgically removed and then received in the laboratory can also influence the ability to successfully obtain results. Notably, the ability to obtain cytogenetic results does not appear to be dependent on whether or not it is a high-grade or low-grade tumor.

Molecular Pathology of Human Osteosarcoma - Marc Ladanyi, M.D.

Dr. Ladanyi presented a synthesis of available data on the molecular pathology of human osteosarcoma. A model of the pathogenesis of osteosarcoma from 1999 proposed that most osteosarcomas have some type of combined inactivation of the RB and p53 tumor suppressor pathways, but the timing or sequence of these alterations remains unclear. Amplification of the 12q13 region (containing MDM2 and CDK4) or INK4A deletion can affect both pathways, and indeed these alterations seldom coexist with RB or P53 alterations. Because these tumors almost universally show evidence of different types of genetic alterations inactivating the RB and p53 tumor suppressor pathways, it would be expected that individual alteration types would not be efficient at identifying prognostic subsets. Indeed, new data show low prognostic significance of p53 mutations in sporadic osteosarcoma. In a recent study (Wunder, et al; CTOS meeting 11/01), 22% showed p53 mutations, but there was no impact on distant recurrence. Another interesting finding was that p53 status was concordant in all paired samples of primary and distant metastases, further supporting p53 pathway alteration as an early, pre-clinical event in osteosarcoma pathogenesis.

It may be useful to place osteosarcoma in the context of other sarcomas. There are two major classes of sarcomas:

1. sarcomas, such as osteosarcoma, with complex unbalanced karotypes and alterations of p53 pathway in most cases (karotypic complexity indicates major telomere dysfunction in pathogenesis)
2. sarcomas with balanced karotypes (reciprocal translocations) and alterations of p53 pathway in relatively few cases, such as Ewing's Sarcoma (suggesting a lack of a major role for telomere dysfunction in their biology)

Dr. Ladanyi discussed telomere function, which is just beginning to be investigated in osteosarcoma. Telomere shortening limits the growth and reproductive capacity of human cells. Telomere erosion has a proposed dual role in tumorigenesis including: limiting tumorigenesis when telomere erosion acts as a tumor suppressor; and promoting tumorigenesis, when impaired p53 function allows telomere erosion to progress to a late crisis (a.k.a. "genetic catastrophe").

There appears to be an interrelationship between telomere function, p53, and karotype instability. A premalignant stage includes proliferation of cells and shortening of telomeres, and only cells that lack p53 function are able to survive this stage. In the absence of p53 function, chromosomal instability develops. Therefore, either p14ARF loss or p53 mutation may allow the pre-neoplastic cell to proceed to late crisis.

Complex unbalanced karotypes are of significance in osteosarcoma. Karotypic complexity reflects chromosomal fusion-bridge-breakage cycles that occur due to advanced telomere erosion. This etiologic scenario may be similar to that found in epithelial cancers. Thus any model of the molecular pathology of human osteosarcoma needs to integrate the emerging data on telomere dysfunction.

The 2001 model of osteosarcoma pathogenesis suggests there must be an inappropriate proliferative signal that affects the osteoblast, or precursor cells, followed by loss of cell cycle control leading to early crisis. If this early crisis is followed by telomere dysfunction, karotype instability, and selection of p53 inactivation, then a late crisis occurs. Finally, selection of telomerase activation and stabilization of an abnormal karotype leads to fully malignant osteosarcoma.

Future needs and directions include: moving from lists of genetic alterations to functionally related groups of genetic alterations (hyperproliferative, cell cycle control, apoptosis, DNA damage response); a better understanding of the timing and relationship of common oncogenetic events in osteosarcoma (alterations of p53 and RB as early preclinical events, p53 pathway alterations in osteosarcoma in familial retinoblastoma; RB pathway alterations in osteosarcoma in Li-Fraumeni syndrome); developing a comprehensive analysis of the p53 and RB pathways in a single large set of osteosarcoma by new high-throughput approaches; a better understanding of different “equivalent” common oncogeneic events (preferential 12q13 amplification in low-grade/surface osteosarcoma, preferential p53 missense mutation in adult osteosarcoma); a better understanding of the paradox of carcinoma-type cytogenetics in the setting of a younger age range; and defining the biologic/genetic subsets according to telomerase status and karotypic complexity (e.g. approx. ½ of osteosarcoma are telomerase-positive and the other ½ are telomerase-negative and show alternative lengthening of telomeres).

Telomerase - Jeffrey S. Dome, M.D.

Dr. Dome discussed telomeres and the functions they serve, which include:

1. preventing chromosomes from being recognized as damaged DNA
2. preventing end-to-end fusions and recombinations
3. accommodating the loss of DNA that occurs with each round of replication

It is hypothesized that normal somatic cells have finite proliferative capacity, and that telomere length is one of the checkpoints that determines whether a cell stops dividing. As cells divide, the telomere length gradually decreases to a critical size, at which point senescence is triggered. Rare cells bypass this checkpoint by inactivating the Rb or p53 pathways, in which case they continue to divide until telomeres become very short and apoptosis is triggered. These two checkpoints can be overcome by activating mechanisms to lengthen telomeres; a central feature of cancer cells is that they must come up with a way to maintain telomeres. About 85% of all cancers activate an enzyme called telomerase to maintain telomeres and the other 15% use a recombination-based method called ALT (alternative telomere lengthening). Unlike most cancers, at least 50% of osteosarcomas are dependent upon the ALT mechanism to maintain telomeres.

The ALT vs. telomerase mechanisms are a means to the same end, but they are not equivalent. This is evident in the length of the telomere itself. Telomerase-dependent osteosarcoma cell lines have short telomeres and a minor range of length, whereas ALT-dependent osteosarcoma cell lines have long telomeres and great variability in telomere length. The ALT cell lines have greater genetic instability and more translocations than the telomerase-positive cell lines. These findings are based on only 3 cell lines, but the information is suggestive. It has been hypothesized that ALT-dependent osteosarcomas have different clinical behavior than telomerase-dependent osteosarcomas.

Telomerase inhibition may represent a promising adjunct to conventional therapy. To test the theory that telomerase inactivation leads to arrest of cellular division and apoptosis, mutant forms of telomerase were developed. These mutant forms of telomerase resulted in shortening of the telomeres and eventual apoptosis. The cells with regular telomerase activity continued to grow. This suggests that a telomerase inhibitor could inhibit cellular proliferation and telomerase may prove to be a therapeutic target. However, it can take numerous cell doublings for growth arrest and apoptosis to occur, so a telomerase-inhibitory approach may be most efficacious in the setting of minimal residual disease (MRD). Emerging data indicates that telomere shortening is associated with increased radiosensitivity. Therefore, telomerase inhibition may be used with other treatments to increase the effectiveness of therapy.

Limitations of telomerase inhibition in osteosarcoma include:

1. 50% of osteosarcomas are telomerase-negative
2. there is unclear potential for telomerase-dependent tumors to "convert" to ALT dependent tumors
3. substantial tumor growth may occur before telomerase inhibition takes effect

Future directions include: determining the frequency of ALT versus telomerase- dependent osteosarcomas; comparing clinical features (stage, response rate, RFS) between ALT- and telomerase-dependent osteosarcomas; comparing molecular phenotype between ALT- and telomerase-dependent osteosarcomas; and assessing telomerase inhibitors in telomerase-positive osteosarcomas.

Discussion - David H. Ebb, M.D.

Q to Dr. Dome: What is the doubling time of the cell lines that he investigated?

A: HELA cells have rapid doubling time. Cells died in about a month, but it varies depending on cell type.

Q to Dr. Hansen: Was phosphorylation modified?

A: Phosphorylation does get modified. The working hypothesis is that there are a variety of targets such as SKb1, n-myc, etc. He is currently investigating osteosarcoma cell line microarrays.

Q to Dr. Dome: Regarding the ALT pathway, how much is it used in the development of stem cells?

A: We know stem cells have telomerase activity as well as malignant cells. Telomerase inhibition may not be as selective as originally believed, but stem cells aren’t always dividing, so non-dividing stem cells would not be affected by telomerase inhibition.

Q to Dr. Malkin: Is there a pharmaceutical application for adenoviral p53 introduction on the horizon ?

A: Yes, adenoviral-based p53 vectors have been developed, and some are in early clinical trials, as well as in preclinical testing.

Q to Dr. Hansen: Questions about RANK and its potential clinical significance. Is this related to tumor aggressiveness?

A: There hasn’t been a large enough survey to say anything about this clinically and about outcome. RANK is being studied in breast cancer that has metastasized to bone.

Q to Dr. Hansen: Where would RANK be placed in the proposed model?

A: Based on frequency, mechanisms of p53 and RB are first, and RANK occurs soon after.

Q to Dr. Hansen: Is there a correlation between overexpression of RANK and 18q?

A: He does not think so. All of the tumors have had this expression pattern, but he does not think this relationship is mandatory.

Cytokine/Death Receptor Pathways in Osteosarcoma

Fas Pathway - Eugenie S. Kleinerman, M.D.

Dr. Kleinerman began by stating that the morning’s discussion reviewed why osteosarcoma develops, but there is a difference between why it arises and why it metastasized. As one looks at tumors, tumor metastases have a propensity to grow in certain areas (i.e., osteosarcoma tends to grow in the lung).

In murine models, SAOS-LM6 cells were selected to have a more metastatic phenotype. Lung metastases with different propensity to metastasize to the lung were harvested. Fas expression was examined to determine its role in osteosarcoma metastases. Using Northern analysis, it was found that LM6 expressed a lower level of Fas. Flow cytometry showed that LM6 expressed a lower level of surface Fas.

To test whether Fas affects metastatic potential of the cells, Fas-transfected cells and control cells were injected into mice. The mice injected with control cells developed metastases, but the mice with Fas-transfected cells did not. Therefore, Fas may be very important to metastatic potential of osteosarcoma.

She then investigated whether Fas could be altered to impact metastatic potential. The adenovirus vector Ad.mIL-12 was used to transfect osteosarcoma cells to express IL-12. Cells that had been transfected had a reduced metastatic potential. IL-12 had an impact on Fas expression. IL-12 is composed of p40 and p35; it is only able to increase Fas expression when both of these subunits were available. Therefore, Dr. Kleinerman hypothesized that high Fas expression leads to tumor cell death and low Fas expression leads to increased growth.

Cells can also be transfected with the adenoviral vector through aerosol therapy. A study was conducted in which mice were injected with tumor cells, then aerosolized IL-12 adenovirus was administered. In the majority of mice receiving the adenovirus, pulmonary tumor cells were eliminated. This aerosol approach using liposomal 9-nitrocamptothecin is currently being tested in a Phase I trial for patients 10 years and older at MD Anderson.

Dr. Kleinerman also discussed INT-0133-Phase III Trial of Doxorubicin, Cisplatin and Methotrexate with and without Ifosfamide, with and without Muramyl Tripeptide Phosphatidylethanolamine (MTP-PE) for Treatment of Osteogenic Sarcoma. She considered whether Fas showed any relevance to this study. MTP can induce IL-12 in patients, because it activates macrophages which produce IL-12. 4-HC, the active metabolite of cyclophosphamide, upregulated Fas ligand expression in LM6 and LM6-IL-12 transfected osteosarcoma cells. The osteosarcoma cells were more sensitive in the presence of IL-12. She hypothesized that ifosfamide and IL-12 upregulate FasL and Fas respectively and induce osteosarcoma cell death.

Q/A Session

Q to Dr. Kleinerman: Could she predict the effects of ifosfamide alone?

A: There is data that shows if you upregulate Fas ligand, you can impair the immune system because CTL cells have Fas and thus the FasL on the tumor cell induces CTL death and immunosuppression.

Q to Dr. Kleinerman: What is the effector cell population in the nude mouse model?

A: The lung epithelial cells express Fas ligand. T cells don’t look like they play a role, but an increase in the influx of NK cells was seen in the nude mice that received IL-12.

Cytokine Apoptosis & Inhibition of Angiogenesis in Murine Solid Tumor Models - Jon M. Wigginton, M.D.

Dr. Wigginton discussed cytokine apoptosis and inhibition of angiogenesis in murine solid tumor models.

His rationale for expecting complementary antitumor effects by IL-12 and IL-2 is as follows: different signaling pathways are used by IL-12 vs. IL-2; IL-2 upregulates IL-12 receptor expression on T and NK cells; and the combination has enhanced immune stimulating effects for T and/or NK cells (cytotoxicity, cytokine production, and proliferation).

Evaluation has been done mostly in mouse-based renal cell carcinoma models, so the data presented focused on this tumor type. A pulse regimen which administered IL-2 over a series of days was conducted. The pulse regimen seemed to be well-tolerated and achieved tumor regression. Systemic IL-12/Pulse IL-2 inhibits tumor neovascularization and induces tumor regression via mechanisms which are dependent on CD8+ T cells, IFN-gamma, and Fas/Fas-L. The following statements can be made from the data:

1. most mice cured of the renal cancer are resistant to rechallenge with the tumor
2. the regimen is ineffective in SCID mice
3. the tumor becomes infiltrated with CD8+ T cells, and depletion of these effector cells ablates the anti-tumor response

The role of IFN-gamma in the antitumor activity of this regimen is critical. Dr. Wigginton investigated whether the noted antiangiogenic effect played an early role in tumor regression. A recent study examined tumor angiogenesis samples 5 and 12 days after treatment ended. A trend of reduced vascularity to tumors was seen as early as 5 days of treatment. Therefore, IFN-gamma appears to play a central role in the significant reduction in tumor vascularization .

The impacts of IL-12/IL-2 on Fas/Fas-L gene expression may be as follows: RENCA (renal cell carcinoma) does not express Fas-L; IL-12/IL-2 potently enhances the production of IFN-gamma and TNF-alpha; and IFN-gamma/TNF-alpha enhances Fas expression on RENCA and EOMA (hemangioendothelioma).

The working hypothesis is that administration of IL-12/Pulse IL-2 induces Fas expression on tumor or endothelial cells which leads to apoptosis. The role of Fas in vascular endothelial injury induced by treatment with IL-12/Pulse IL-2 appears to be Fas ligand dependent, and antitumor activity against metastatic RENCA appears to be Fas ligand dependent.

A Phase I study of the IL-12/Pulse IL-2 schedule in adults has begun.

Fas & TRAIL-induced cell death in ES & NB – Maria Tsokos, M.D.

Dr. Tsokos discussed Fas and TRAIL-induced cell death in Ewing’s Sarcoma and neuroblastoma models. Fas has been implicated in a variety of tumors. If Fas is expressed on the tumor cell surface then it will die, unless it has developed a mechanism to prevent Fas triggered cell death. This could be done by downregulatation of Fas, Fas ligand, or Pro-Caspase 8 (FLICE), or upregulation of apoptosis inhibitory proteins, specifically bcl-2, or FLICE-inhibitory protein (FLIP). The Fas “counterattack” model refers to the Fas ligand expressed on tumor cell surface and its use to avoid activated T-cells, allowing tumor cells to escape surveillance.

Dr. Tsokos investigated the reason for Fas upregulation and how this occurs. Fas ligand exists in transmembrane and soluble forms in ESFT. She hypothesizes that ESFT cells may avoid cellular suicide by downregulating their surface FasL (by cleavage) and Fas (by internalization). When cells were treated with synthetic MMPIs, there were increased levels of surface FasL and Fas (in the absence of increased transcription), and cytoplasmic Fas is decreased. Exposure to MMPIs sensitized ESFT cells to a Fas-activating antibody (CH11) and sensitized cells to doxorubicin. Another study using TRAIL (an apoptosis-inducing ligand) was conducted. Receptor 5 (chemokine receptor 5, or CCR5) is needed to allow TRAIL to work and induce apoptosis. Lines lacking this receptor were the ones that were unable to respond.

Caspase 8 activation was found in Ewing’s Sarcoma cell lines, but not in the neuroblastoma cell lines. The lack of measurable expression of Caspase 8 in neuroblastoma does not translate to a lack of Caspase 8 presence, because FLIP is expressed in neuroblastoma. Neuroblastoma cells do not release cytochrome c from mitochondria during Fas activation, in contrast to ESFT cells. Caspase 3 is activated with antisense bcl-2; and Caspase 8 is important to antisense bcl-2 function (treatment of neuroblastoma with antisense bcl-2 caused cell death but only where Caspase 8 was present).

N-myc induces transcriptional upregulation of FasL, and in the presence of a functional Fas pathway, sensitizes neuroblastoma cells to apoptosis.

Interferon Gamma - Janet Houghton, Ph.D.

Dr. Houghton began by discussing the preclinical and clinical development behind modulation of the Fas signaling pathway in therapy for colon cancer. Fas has the following properties:

1. type I transmembrane protein
2. regulates apoptosis in cells for immune system
3. expressed in many different cells and tissues
4. expressed in colonic epithelial cells
5. functional in cells outside of immune system
6. Fas ligation induces apoptosis

Her studies confirmed that agents such as doxorubicin, topotecan, and VP-16 induced cytotoxicity without involvement from Fas. In observing Caspase 3 and Caspase 8 activity, she realized that Fas expression best correlated to apoptosis. Certain cytokines can elevate the level of Fas expression (specifically IFN-gamma). In a study investigating Fas expression in HT29 colon carcinoma cells after 2-hour exposure to IFN-gamma, the level of Fas was elevated, implying that IFN-gamma could sensitize cells to apoptosis. Potentiation of apoptosis by IFN-gamma is dependent on fluorouracil-induced DNA damage and the Fas death receptor, but is independent of p53.

The design of a Phase I clinical trial was to determine the maximum tolerated dose of IFN-gamma added to a standard regimen of 5-FU and leukovorin for metastatic GI cancers (19 patients enrolled; all colorectal cancer patients had prior exposure to 5-FU/LV regimens). Patient outcome data showed that 5FU disposition is not altered by IFN-gamma administration and there is a delayed absorption of IFN-gamma.

IFN-gamma upregulates Fas expression in HT29 cells at pharmacologically relevant concentrations, as determined in the plasma of patients treated in the phase 1 study. Exposure to IFN-gamma for 1 minute to 2 hours sensitizes HT29 cells to CH11 in clonogeneic assay. Finally, 5FU/LV combined with IFN-gamma induces cytotoxicity in HT29 at pharmacologically relevant concentrations. A Phase II trial is being planned.

Discussion - Holcombe E. Grier, M.D.

Q to Dr. Wigginton: How is, and how much is, Wyeth supplying IL-12?

A: They have an approved LOI for study in neuroblastoma in IL-12. IL-12 is unlikely to go forward as a single agent, and the drug’s future is dependent on results of the current clinical trials. There will be a company decision in about a year or so to see if cancer applications of IL-12 will continue.

Q to Dr. Hougton: She was asked about the great variability in the biology of osteosarcoma vs. colon cancer, and the fact that the data in the past panel was not directly related to osteosarcoma. Can this really be translated to osteosarcoma? Is there enough of a biological link between the previous presentations and osteosarcoma?

A: She responded that models of all sorts are needed.

A: Dr. Tsokos added that what is missing in all models is investigation of different pathways. We have great models, but don’t seem to generalize beyond a single tumor line.

Q: Dr. Gorlick asked what missing background is needed?

A: Dr. Tsokos responded that more cell lines and more pathways are needed.

Q: Any data on 5- FU and LV in osteosarcoma?

A: Data is from the 1970s saying it is inactive in osteosarcoma.

Q: It was mentioned that there are missing background studies. What background studies are needed?

A: It would be beneficial to have several cell lines that show response, and among the cell lines with no response, to determine why there is a lack of response by focusing on the upregulation or downregulation of certain proteins; very few people are doing lab research in osteosarcoma because of lack of funding and lack of interest in sarcomas.

Q: How would you look at the Intergroup trial in relation to the "counterattack" model? Is the B-minus result a significant result?

A: It could be real because you see the same results in two different patient populations, the non-metastatic and metastatic populations.

Q: We heard the hypothesis that MTP functions through IL-12…then IL-12 models and using knockouts of INF-gamma and effector knockouts…the question is, should we pursue immunologically-based trials in osteosarcoma? What we hear is to put together a hypothesis in a group of models to make it logically worth pursuing. Should we look at IL-12 if that is the major mediator? Should we do it in many different models to effectively dissect it?

Q: What’s the hypothesis?

A: Effects of IL-12, effected Fas, Fas ligand, and apoptosis. But now this needs to be further dissected and pursued in backgrounds where you knock-out various components to make it all fit.

Q: Any idea if GMCSF can perform similarly?

A: Aerosol GMCSF has been looked at in melanoma.

Q: What is the theory behind it being IFN-gamma independent?

A: You can upregulate Fas but not measure any IFN-gamma; this shows it is independent.

Q: Seems like IL-12 is fairly difficult to examine, but IFN-gamma is easier to look at. Any way to look only at IFN-gamma to see if that alone can do IFN-gamma?

A: There has been one trial with IFN-gamma (a Phase III ovarian cancer trial) that showed an advantage. Also, IFN-gamma is commercially available. But you need a model that is immunocompetent. The dog model could answer the GMCSF, IFN-gamma, and possibly IL-12 questions.

Drug Resistance/Growth Factor Pathways in Osteosarcoma

p-Glycoprotein (PGP) - Mark C. Gebhardt, M.D.

Although dramatic improvement in osteosarcoma patient survival has been seen with adjuvant chemotherapy, 20-40% of patients ultimately have tumor relapse, probably due to drug resistance. P-glycoprotein has been investigated as a means to identify these "non-responders" early, allow treatment modification, and potentially improve their survival. Similarly, PGP status may identify patients who could receive less toxic therapy.

A study was conducted in which PGP levels were viewed as a predictor of poor outcome in osteosarcoma. A variety of statistical tests determined that PGP+ patients were more likely to develop metastases and die (true for stage II B patients). Histologic necrosis was also investigated. Many other studies on PGP can be found in the literature, including Takeshita 1996, Baldini 1992, Wunder 1993, Baldini 1995, Baldini 1999, Stein 1993, Posl, 1997, Chan 1997, and Bodey 1995. Dr. Gebhardt described several possible optimal methods to assess PGP, which include: immunohistochemistry (but which antibody?); mRNA (RT-PCR), and in situ hybridization.

The possible mechanisms of drug resistance:

1. multidrug Resistance Protein (MRP)
2. topoisomerase II
3. glutathione S-transferases
4. enhanced DNA repair
5. altered drug metabolism or inactivation
6. reduced intracellular influx

MRP-positive cells have been detected in osteosarcoma. Caveats of the MDR theory in osteosarcoma include: the demonstration of a lesser survival in MDR-negative tumors may not be a reflection of decreased drug accumulation in PGP-positive tumors, the presence of PGP does not necessarily confer resistance, and its absence does not imply sensitivity.

The current literature concludes that there is no correlation between PGP status and percentage of necrosis. PGP status may predict disease outcome, but not histologic necrosis. The literature also concludes that PGP may be a sign of aggressiveness as well as a marker of drug resistance, and may be useful in identifying high-risk osteosarcoma.

The conclusions of Dr. Gebhardt’s study included:

1. PPG expression is associated with a poorer metastasis-free survival and overall survival
2. expression of PGP and its level were significantly predictive of time to death
3. the monthly risk of death was four times higher among patients with PGP expressing tumors as compared to those with PGP negative tumors

A larger study, along with confirming the best assay of PGP, is needed.

Future needs include: determining whether PGP status is a predictor of outcome in larger groups of patients; how best to detect drug resistance (PGP and other mechanisms); risk-based therapy; other mechanisms of drug resistance; and newer chemotherapy and reversal agents.

p-Glycoprotein (PGP) - Irene L. Andrulis, Ph.D.

Dr. Andrulis continued the discussion of p-glycoproteins.

The study of molecular alterations in osteosarcoma include: identification and characterization of alterations in tumors; determining the clinical significance (prognostic or predictive value) of the alteration; and designing new therapeutic agents based on the molecular alteration.

Dr. Andrulis presented the same questions as Dr. Gebhardt: how are the "non-responders" identified, can they be treated differently and improve their survival, and are there patients who need less (or less toxic) therapy?

Improvements in therapy could result from investigation in the following areas: MDR1 expression in osteosarcoma; intrinsic resistance determined in biopsy specimens; resection specimen and relapse specimen-acquired resistance; low-level of drug resistance and subsequent detection in RNA; and looking at MDR1 expression in different sites within the same case (to determine whether the level of expression remains the same in different tumor sites within the same patient).

Drug resistance can be intrinsic (in the absence of treatment) or acquired (after treatment with chemotherapeutic agents). In a pilot study of 15 patients, the expression of MDR1 was correlated with poorer overall survival. A larger prospective study was then done. MDR1 expression was examined using RT-PCR and then correlated with DFS. Between 1989-1994, 123 newly diagnosed patients with high-grade, non-metastatic extremity osteosarcomas all received neoadjuvant chemotherapy that included doxorubicin. Although it was found that tumor size was predictive of outcome, the degree of tumor necrosis was not significant in the multivariate analysis (although it was in the univariate analysis). Patients with high levels of MDR1 did poorly, but patients with low levels of MDR1 also did poorly. Therefore, no association between survival and MDR1 could be made.

Dr. Andrulis inquired as to why there is a lack of correlation with MDR1 expression level and DFS in the prospective survival. The reasons include: conflicting results with IHC due to sensitivity of technique, antibodies detecting other proteins, and expression of MDR2 in osteosarcoma; other mechanisms of drug resistance; and p53 and drug resistance. Many osteosarcoma specimens have very high levels of expression of MDR1 and MDR2.

p53 is a tumor suppressor gene that is interesting because wild-type p53 can inhibit MDR1. p53 gene mutations in osteosarcoma fall in the standard genetic region, and in looking at other exons (including exon 4 and 10), some non-missense mutations are seen. When progressive sets of tumor biopsies from the same patient were examined at diagnosis and definitive surgery, it was seen that if the biopsy had a p53 mutation, then the resection had a mutation. When correlating type of p53 mutation and MDR1 expression, it was found that those who had missense had higher MDR1 expression, and those who had non-missense had lower levels of MDR1.

Ongoing studies are examining genetic alterations associated with progression and microarray data to identify novel genes/pathways.

INT-0133 MDR/PGP Study (Additional Presentation) - Cindy Schwartz, M.D.

Dr. Schwartz discussed INT-0133-Phase III Trial of Doxorubicin, Cisplatin and Methotrexate with and without Ifosfamide, with and without Muramyl Tripeptide Phosphatidyl Ethanolamine (MTP-PE) for Treatment of Osteogenic Sarcoma.

The objectives of the study were to determine whether multiple drug resistance encoded by PGP is useful to determine prognosis or to assign therapy, and to determine whether histologic response correlates with PGP positivity. The rationales behind the study included that PGP is an ATP-dependent efflux pump for hydrophobic substances and the presence of PGP has been associated with poor outcome in retrospective studies of patients with osteosarcoma. This study was intended to be the first large-scale prospective study of the effect of PGP expression on outcome. 679 patients with non-metastatic disease from CCG and POG were enrolled on this trial between November 1993 and November 1997. Antibodies used included C-494 and JSB-1. Results showed that 107 and 104 patients were evaluated with C-494 and JSB-1, respectively, and discordance was only 14% between C-494 and JSB1. EFS, osteosarcoma, and histological response were identical in those evaluated vs. those not evaluated. EFS and increased risk of death did not differ significantly in patients with tumors that demonstrated antibody positivity at study entry compared to antibody negative patients. The patients on regimen A (no ifosfamide) with antibody positivity at study entry were not at significantly increased risk for adverse events or death. The patients on Regimen B (+ ifosfamide) with antibody positivity at study entry were not associated with increased risk of adverse events, but there was evidence of a reduced risk of death in these patients. Each regimen had a chemotherapeutic agent with a PGP substrate in them.

The study concludes that MDR-encoded PGP is not useful to determine prognosis in patients with localized osteogenic sarcoma treated with cisplatin, methotrexate, doxorubicin, + MTP-PE, + ifosfamide. Furthermore, histologic response is not correlated with PGP positivity. It is unclear whether further evaluation is needed to determine the usefulness of MDR-encoded PGP to assign therapy.

Discussion - Leonard H. Wexler, M.D.

Q: Given the seeming discrepancy of results that show no difference in the percentage of necrosis but does demonstrate a difference in survival, should this be looked at in 500 patients? If you do decide to look at it, how do you look at it?

A: Most of the studies have done it on biopsy materials, and not put the tumor into culture. A lot of care should go into a proper design.

Q: So is MDR an area that the groups should look at?

A: No consensus of the group was reached on this topic.

Q: What is the status of MDR reversal agents in other tumors?

A: Dr. Smith responded that overall results have been disappointing, and there have been few successes. This should not be a high priority, but it can be looked at some. Other participants said that this should not be investigated much further.

A: If doxorubicin is the major agent, then there may be some merit to it. In the Toronto study, RT-PCR and IHC were done and very few were positive.

Dr. Andrulis commented that PGP does not have to be constantly measured, but wouldn’t discount this yet. In their study patients with very, very low MDR1 as well as those with very high levels of expression were those patients with a poor prognosis. Microarrays should be used to see what is happening in these pathways.

Q: Should we continue to do MDR related studies?

A. No consensus of the group was reached on this topic.

Antibody Based Strategies for Metastatic Solid Tumors - Nai-Kong V. Cheung, M.D., Ph.D.

Dr. Cheung discussed potential therapy targets using mouse or chimeric antibodies, anti-idiotypes as vaccines, pretargeting strategies, and retargeted T-cells.

Potential targets are GD2 (monoclonal antibodies 3F8 and 5F11) and gp58 (antibody 8H9). 3F8 targets GD2 (disialoganglioside) on neuroblastoma. GD2 is abundant on neuroblastoma (5-10x10⁶) and shows persistence on the cell surface. gp58 is seen in many pediatric tumors, including osteosarcoma.

Mouse or chimeric antibodies constitute active therapy alone or with BRM (Beta-glucan). Beta-glucan enhances the anti-tumor effect of monoclonal antibodies. Used alone, Beta-glucan has no anti-tumor effect, but with MoAb the effect is significant and in mouse tumor models, Beta-glucan + MoAb prolongs progression-free survival.

A pretargeting strategy uses scFv-streptavidin (radioisotopes, biologics). Following tumor targeting of scFv-streptavidin, a clearing agent such as biotin-LC-NM-(Gal-Nac)₁₆ removes residual fusion proteins from the blood, significantly improving the tumor:blood AUC ratio. The whole antibody has good uptake but not a good ratio, while fragments give a better ratio but poor uptake. Using GD2 as a target, whole antibody or antibody-fragment alone results in a tumor:blood AUC ratio of <5:1 , while the pretargeting approach gives a >100:1 ratio. This system allows a high-level of uptake in the tumor and is very robust.

Retargeting T-cells involves modification of T-cells and development of chimeric immune receptors (T-bodies). All T-cells would recognize the same antigen and would proliferate when the anti-idiotype (8H9 scFv-CD28-symbol chimeric immune receptor) is added. When used in rhabdomyosarcoma, tumor suppression is seen.

Growth Hormone - Lee J. Helman, M.D.

Dr. Helman discussed Insulin-like Growth Factor -1 (IGF-1) and osteosarcoma.

Characteristics described for IGF-1 and osteosarcoma are as follows: peak incidence of osteosarcoma occurs during peak GH and IGF-1 concentrations; prevalence of osteosarcoma is highest in dog breeds with the highest IGF-1 concentrations; hypophysectomy is associated with anti-tumor activity in rodent osteosarcoma models (anti-tumor effect is associated with IGF-1 inhibition); IGF-1 inhibition by growth hormone blockade results in decreased osteosarcoma growth; and osteosarcoma cell lines are dependent on IGF-1 for in vitro growth and survival.

Osteosarcoma cells express IGF-1 receptors, are stimulated to proliferate by IGF-1, and are more chemoresistant after IGF-1 addition. There is interest in taking advantage of this because there are many inhibitors available for this pathway.

SMS 201-995pa LAR (OncoLar - Sustained Release Sandostatin) is a long acting formulation of octreotide acetate. Octreotide acetate can effectively inhibit IGF-1 through the blockade of growth hormone.

A Phase I study (95-C-0119) of OncoLar with/without Tamoxifen in relapsed osteosarcoma administered 60 mg to 6 patients, 90 mg to 6 patients, 60mg plus tamoxifen to 6 patients, and 90 mg plus tamoxifen to 3 patients. OncoLar was administered every 4 weeks, and growth hormones were measured; the only sustained decrease was in circulating IGF-1. This was not dose-dependent and addition of tamoxifen had no influence.

The study concluded that OncoLar treatment leads to a sustained decrease in IGF-1 levels (40-50%) that was dose independent between 60 mg and 90 mg. The study also concluded that no other component of the IGF pathway is altered, and there was no additional decrease with tamoxifen. At the level of IGF-1 reduction achieved, however, no measurable tumor response occurred. growth hormone receptor

Future studies to consider include:

1. combination with cytotoxic therapy
2. development of a Pegvisomant- growth hormone receptor analogue that prevents growth hormone receptor dimerization and is a stronger inhibitor of circulating IGF-1
3. combinations of growth hormone receptor antagonists with Somatostatin analogues

Dr. Helman posed two hypotheses. First is that the inhibition of the IGF-1 growth signal, combined with DNA damage from conventional chemotherapy, may act in an additive or potentially synergistic manner to induce tumor apoptosis. Also, enhanced tumor apoptosis may significantly improve survival outcomes in patients with osteosarcoma without increasing treatment related toxicity.

Q/A Session

Q: Would a 50% reduction in IGF-1 be expected to cause significant anti-tumor activity?

A: No.

IGF - Jeffrey A. Toretsky, M.D.

Dr. Toretsky asked whether IGF-1 contributes to tumorigenesis, tumor growth and progression, or chemotherapy resistance; and can modulation of IGF-1 reduce tumorigenesis, reduce tumor growth and progression, or modulate chemotherapy resistance? Epidemiologic considerations include: 1) does growth predispose to osteosarcoma; 2) is increased size/rate of growth adequate to implicate IGF-1 level as a risk factor; 3) and do elevated levels of IGF-1 predispose to osteosarcoma? Multiple studies have addressed whether increased growth (both weight and height) identify persons more likely to develop osteosarcoma. The three most recent studies are case-controlled, however all utilized different control groups. When either a normalized growth curve from National Center for Health Statistics or geographic neighbors were used, no difference in final height or growth velocity were observed between cases and controls. The most recent study utilized controls matched from birth data and evaluated the height and weight one year prior to diagnosis. This study did show that size one year prior to diagnosis was a risk factor for developing osteosarcoma. The authors did suggest selection bias might play a role in their results. I believe overall it is unclear whether "big" people are more likely to develop osteosarcoma.

IGF-1 activity is activated through its cognate transmembrane receptor; which can also be activated by IGF-2. The IGF-IR is required by many oncogenes for transformation (IGF-1R null mouse embryonic fibroblasts do not transform independently with SV40 lg T, Ha or Ki ras, or c-myc); and the signaling pathway activated by IGF-1R leads to cell survival.

IGF-1R studies in osteosarcoma have determined the following: human tumors express IGF-1R; cell line signaling/cell survival is inhibited by aIR3-antisense, or suramin; and p53, occasionally mutated in osteosarcoma regulates IGF-1R expression.

A recent study by Rodriguez-Galindo, et al, 2001, determined that circulating concentrations of IGF-1 and IGFBP-3 are not predictive of incidence or clinical behavior of osteosarcoma. However, the study was limited by the small number of patients and lack of statistical power. Further investigation regarding the role of the IGF-1 pathway in the development and progression of osteosarcoma is warranted, and suggested by the authors, and could conceivably provide new potential targets for therapeutic intervention.

Dr. Toretsky investigated the relationship of IGF-1 levels to EFS in the Ewing’s Sarcoma Family of Tumors (ESFT). Using data from 111 patient samples available for analysis, it was shown that serum IGF-1 and IGFBP-3 proteins survive prolonged freezing. Preliminary studies also showed that IGF-1 and IGFBP-3 levels in ESFT patients can identify patients with the most widespread disease, but are not independent predictors of prognosis. A high IGFBP-3 to IGF-1 ratio suggested a better prognosis. Improved survival among metastatic patients was also associated with IGFBP-3 to IGF-1 ratios that fell between the 75^th and 90^th percentile of the samples measured. This "Mama Bear Effect" is seen in those patients with a mid-range level of free IGF-1 that ultimately show the highest survival rate. Patients with very high ratios had extremely low IGF-I levels and had severe systemic disease. Those with low ratios might have had high levels of IGF-I and thus potentially did poorly in part due to IGF-1R modulated chemoresistance in a small population of cells that ultimately regrow. While there is in vitro data to support this hypothesis, a prospective trial is underway to determine if this result is reproducible in patients currently undergoing therapy.

Current plans for osteosarcoma include: investigating collected serum from 214 patients as part of protocols (P9851); measuring IGF-1 and IGFBP-3; and evaluating these results as a function of DFS in the next several years. We are considering measuring receptor levels as well, however, it is challenging to attempt quantitation of IGF-1R utilizing either immunohistochemistry or ELISA. There is a potential for new therapies which target block ligand binding sites on the receptor, inhibit the receptor kinase activity, or by targeting downstream signaling.

Dr. Toretsky concluded that IGF is likely a part of a multifactorial pathogenesis of osteosarcoma and needs to be better characterized in a wider panel of osteosarcoma cell lines and patient samples. Current studies will identify whether IGF-1 levels play a role in chemosensitivity/relapse. Inhibition of IGF signaling may provide a future therapy against osteosarcoma, but more background and better agents are needed.

Discussion - Neyssa Marina, M.D.

Q: What are IGF-1 levels in normal patients? How do your results relate to normal?

A: That is very difficult to get to because IGF-1 normal values have a large range which varies with age. It is better to look at it in regards to puberty, but the data just isn’t there. There isn’t a good control group.

Q: Is there any evidence that the receptor is abnormal? Maybe not abnormal levels, but rather during growth a receptor malfunctioned and doesn’t know how to shut off signal?

A: There is not much information suggesting a mutated receptor. Currently there are no known activation mutations occurring in patients. No one that I am aware of has done systemic searches into receptor mutation for the IGF-I receptor. With regard to the epidemiology, the thinking has changed over time. There are epidemiological data in epithelial cancers that says very high IGF are associated with a higher risk. But a predominance of evidence says that IGF has some role. It is a necessary but possibly insufficient for tumorigenesis. Once a tumor is there, components of pathways can be blocked and cells can be pushed to make a death decision.

Q: Are there pharmaceutical companies looking into receptor inhibitors for osteosarcoma?

A (Dr. Helman): As of last May, four companies. When you inhibit IGF-1, insulin is inhibited which is obviously not a good thing to do. This must be targeted without harming the insulin receptor. These unique compounds should be available in the next few years.

A (Dr Toretsky): There are compounds being developed but people are very tight-lipped about it.

Q: Do cells still die with dominant negative AKT? The survival curve is different, the dominant AKT cells die at same point that the other cells don’t. Can you rescue the AKT cells?

A: This hasn’t been researched yet.

Q: Can you utilize the fact that osteosarcoma is right outside the growth curve? And possibly see whether or not there is a difference between a young patient and a patient in the middle of a growth spurt?

A: A study has already been done on this, and didn’t see a difference between these two groups. No cytogenetics were available; it was strictly a RNA/DNA study.

Q: It seems that for a lot of these agents they have to be combined with chemotherapy. What preclinical data do you need to see and what data can we gather for 3F8 to convince you to go ahead with a Phase III trial since you may never get a Phase II trial?

A: There are several. But compelling animal data (specifically in dogs) that shows blockage of pathways is key. A caveat is that the route would have to be relatively non-toxic since it may be needed for a lengthy period of time.

Q: Maybe look at rhabdomyosarcoma situations. There is a dwindling but still sizable number of patients with osteosarcoma who relapse. Maybe these patients could be used to answer some of these questions, and give them a biologic agent to show efficacy. There may be ethical issues which could be covered with crossover design.

A: A possible trial would be patients with A plus A+B, but this would be a Phase III question because it may be more toxic and costly and in need of more power. There should be two things at the same point in development and compare them in Phase II to see which one moves forward to Phase III. An example is aerosol GMCSF vs. an AKT inhibitor.

Q: How to prioritize? Which drugs are most important to move forward? Should there be excitement about the antibodies mentioned by Dr. Cheung?

A: It is a reasonable target and an imaging trial will be starting soon which will give some more information. No imaging studies are planned with the GD2 antibody, but it is a reasonable system. All sarcomas are eligible for upcoming trial. As for the anti-idiotypic antibody for GD2, a study has been on for 4-5 years with slow accrual. Some antibody responses are being seen, but it is early.

New Treatment Approaches

Isotopes: Samarium - Peter M. Anderson, M.D., Ph.D.

Dr. Anderson presented data regarding Samarium (¹⁵³Sm-EDTMP, ethylene-diamine-tetra-methylene-phosphonate), which is a bone-seeking radiopharmaceutical that has been FDA-approved since April 1997. Samarium emits a short-range therapeutic beta particle and also has a gamma photon that allows routine gamma camera imaging. It has a short half-life of 47 hours. The dose-limiting toxicity has been hematologic toxicity because it binds to bone causing the tissue to become radioactive. Thus, bone marrow is also affected. However, the short-half life of samarium allows for rescue of the marrow with stem cells.

A study conducted at the Mayo Clinic administered high-dose Samarium to patients with bone metastases (JCO 2002,20:189-196). Accrual was difficult and the study was expanded to include myeloma patients. Patients were all high-risk and ranged in age from 12-43 years, with a mean age of 20.5. To date, the results show: flat dosing; no "scout dose"; and, if the dose is > 3 mCi/kg, the patient is required to get peripheral blood stem cell (PBSC) rescue. The uptake into the bone is linear with dose. A technetium-avid bone scan predicts tumor uptake of the samarium. "Bone specific" palliative RT for resistant disease was permitted. All patients have neutropenia and require transfusions and GMCSF.

Administration is by 30-minute central line infusion. Dosimetry at Day +1, +2, and +5 allows estimates of residual radioactivity at time of PBSC infusion and dose to tumor. The Day 13 radioactivity estimate was <3.6 mCi. Infusion of PBSC was done on Day +14. In terms of toxicity, there was minimal non-hematologic toxicity, hypocalcemia, and hematologic toxicity. Hypocalcemia was manageable, but the hematologic toxicity resulted in a poor recovery if there was a low number of PBSC (<2 x 10⁶ CD34+/kg). Future studies should not give high-dose samarium if there is inadequate PBSC supply stored. A "flare" reaction of bone pain was common, but improved pain control was seen in 100% of patients. Osteosarcoma recurrence (both bone and lung) remains a major problem after Samarium treatment in this high-risk population of patients presenting with bone metastases.

In conclusion, Dr. Anderson stated that samarium is a bone-specific therapy that may provide more effective radiotherapy than external beam radiation alone. It appears safe, effective, and well-tolerated, except for temporary pancytopenia. A future clinical trial concept would be to compare high-dose vs. standard dose samarium (if there are no PBSC for harvesting) with the primary endpoint of radiation dose to indicator lesion and secondary endpoints of PET responses, time to progression, pattern of disease progression, and ability to provide additional therapy. Use of a radiation sensitizer (low dose gemcitabine) for several days after the samarium may be another means to increase radiobiologic effectiveness.

Q: Was chemotherapy administered to any patients after Samarium?

A: Yes, some patients received ifosfamide/etoposide and some received Gemcitabine. This depended on prior therapy and dose of stem cells (generally complete recovery if >5 x 10⁶ CD34+/kg)

Inhaled GMCSF - Carola A.S. Arndt, M.D.

Dr. Arndt discussed aerosolized GM-CSF for patients with osteosarcoma and lung metastases. She presented background information from Dranoff (1993) who studied GM-CSF in the B-16 melanoma model. The B-16 melanoma cells were transduced with various immunomodulatory proteins; mice were vaccinated with irradiated non-transduced or transduced cells, and challenged with live non-transduced B-16 cells. Irradiated tumor cells expressing GM-CSF reliably stimulated potent, specific, long-lasting immunity in three different tumor models; GM-CSF was the best of >10 cytokines tested. GM-CSF resulted not only in anti-tumor response, but also protected mice from subsequent tumor challenge.

Dr. Arndt also described work by Kumar (1999) who studied the localized effects of GM-CSF and found that GM-CSF transduced melanoma cells had lower tumorigenicity than control cells. Tumors transduced with GM-CSF had dense macrophage infiltration, and local GM-CSF produced by transduced cells facilitated killing of non-transduced bystander tumor cells (co-culture experiments). GM-CSF also enhances marrow recovery post-chemotherapy, modulates the host defense versus bacterial fungi, and acts as a vaccine.

One of the principles of aerosol delivery for inhaled GM-CSF is that small particles take a long time to settle. Nebulizers provide >95% particles of 0.4-3.0 u, of which about 15-20% are deposited in lungs. Aerosol cytokines have the following characteristics: lymphatic absorption; receptor bearing cells in lymphatics, bronchial, and pulmonary lymph nodes; cells provide receptor bearing "gauntlet"; and the effect is dose dependent.

The Phase I dose escalation study (Mayo 97-02-01) allowed intra-patient dose escalation if the patient tolerated the initial dose well. There were two osteosarcoma patients out of seven patients that completed all three dose levels. Of the two, one patient had "too numerous to count" metastases and progressed, and one patient was stable for 11 months. Five of six patients had stable disease and continued aerosolized GM-CSF for an additional two to eight months. No toxicity was seen in any patient.

A Phase II study with 45 patients was completed with eight osteosarcoma patients. Two of eight patients with gross disease progressed. The remaining six achieved second CR, with two patients disease-free at 2.5 and four years.

Dr. Arndt’s proposal for a follow-up study begins with the administration of aerosolized GM-CSF to patients with isolated pulmonary recurrences of resectable osteosarcoma. A thoracotomy will be performed one week later, and a descriptive analysis of histologic findings will be conducted. The time to next recurrence will be analyzed, and patients with bilateral metastases will have initial thoracotomy. Upon recovery, inhaled GM-CSF will be administered for one week, followed by second thoracotomy and continued GM-CSF. Histologic evaluation would include: nodule evaluation for evidence of expression of Fas/FasL; nodules evaluation by routine H & E for evidence of macrophage and neutrophil infiltration; and nodule evaluation for the presence of dendritic cells by immunostains. A similar retrospective tissue study of resected lung metastases at first recurrence prior to chemotherapy is being planned to obtain baseline information.

In the future, Dr. Arndt proposes a pilot study testing administration of inhaled GM-CSF to patients with localized, non-metastatic osteosarcoma beginning right after surgery and continuing for six months. She also proposed inhaled GM-CSF use for patients with pulmonary metastases at diagnosis, post-surgery for lung metastases, and post-therapy in patients at high-risk for future lung metastases.

Antifolate Resistance - Richard Gorlick, M.D.

Dr. Gorlick presented information on antifolate resistance, specifically looking at methotrexate. Most soft tissue sarcomas are intrinsically resistant to methotrexate. The literature suggests it is the peak methotrexate level, and not the AUC, that correlates with therapeutic response in osteosarcoma - the peak level is predominantly determined by dose. High-dose methotrexate is a major component of therapy for osteosarcoma, and tumor resistance to methotrexate is known to occur. The reasons for methotrexate resistance may include intrinsic or acquired methotrexate resistance. The use of neoadjuvant chemotherapy only allows the study of viable/resistant tumors after methotrexate has been administered and all at the time of initial biopsy. Examination of tumors also obtained from the 30% of patients who relapse can be used to study acquired methotrexate resistance.

Trimetrexate has been investigated as an alternative approach for antifolate therapy. A Phase II trial of simultaneous trimetrexate and leucovorin given orally for 21 days was conducted in relapsed osteosarcoma patients. Toxicity was acceptable with myelosuppression being the major side effect, and objective responses were observed in 8%, which may be good in this population (n=39, CR=1, PR=2, MR=1, SD=8).

A phase II window study proposal is under development by the COG with the rationale that the response rate to trimetrexate and simultaneous leucovorin would be predicted to be higher in the window setting than in a conventional phase II trial because dihydrofolate reductase overexpression is not as frequent at diagnosis as compared to relapse. Furthermore, trimetrexate with leucovorin would be predicted to be at least as efficacious as high-dose methotrexate with less toxicity. Also, the lack of renal toxicity may allow chemotherapy dose intensification in subsequent clinical trials, as it may be possible to administer trimetrexate simultaneously with other agents.

Viral Approaches - Thomas A. Gardner, M.D.

Dr. Gardner discussed an osteocalcin promoter-based cancer gene therapy for metastatic disease. There are three components of gene therapy: vectors (naked DNA; liposomal; virus: retroviral, adenoviral, AAV, etc.), transcriptional regulators (universal: LTR, CMV, RSV, SV40, etc.; tumor/tissue-specific: PSA, CEA, OC, AFP, etc.), and therapeutic genes (replacement: p53, CFTR, etc.; toxic: TK, CD, P450, DT, etc.).The positive attributes of using adenoviruses as a viral vector for gene therapy include: high transduction efficiency, large insert capability, non-integration into host DNA, and broad natural tropism. Negative attributes include: moderate immune response and broad natural tropism.

Osteocalcin is a non-collagenous Gla protein produced by osteoblasts. It is synthesized and secreted during bone mineralization, and its immunohistochemical staining is positive in osteoblastic osteosarcoma (100%), chondroblastic osteosarcoma (100%), and fibroblastic osteosarcoma (71%).

Recent osteocalcin promoter-based gene therapy studies have shown in vitro and in vivo growth inhibition of rat osteosarcoma with ad-OC-TK/ACV (adenoviral osteocalcin gene vector). These studies have also shown enhanced in vitro and in vivo growth inhibition of rat and human osteosarcoma with the addition of 5-FU chemotherapy to Ad-OC-TK/ACV and in vivo growth inhibition of metastatic rat osteosarcoma with systemic Ad-OC-TK/ACV. In vitro and in vivo growth inhibition of metastatic human prostate cancer with Ad-OC-TK/ACV has also be demonstrated in preclinical models.

A Phase I trial in prostate cancer (OBA Protocol #9812-276) evaluated safety, gene transfer, and activity of the gene therapy approach. This study showed that repeated intratumoral injections were well-tolerated. No viral shedding was observed after Day 3 and no unexpected toxicities occurred.

Dr. Gardner then discussed Ad-OC-TK / ACV gene therapy for osteosarcoma, the development of chemo-gene therapy for osteosarcoma, the potential suppression of osteosarcoma pulmonary metastases , and the use of isolated lung perfusion to enhance gene therapy delivery.

A new Phase I trial (OBA #0010-426 - A Phase I Study of Intratumoral Injections of OCaP1 into Osseous Metastatic or Post-Surgical Locally Recurrent Prostate Cancer) was reviewed and a possible Phase I Trial of Ad-OC-E1a Pulmonary Metastatic Osteosarcoma was discussed.

Dr. Gardner summarized his presentation as follows: Ad-OC-TK/ACV can inhibit ROS (reactive oxygen species) and MG-63 (p53 gene mutated) cell growth in vitro; intralesional Ad-OC-TK/ACV can inhibit ROS and MG-63 growth in vivo; chemotherapy can enhance cancer gene therapy; systemic Ad-OC-TK/ACV can inhibit ROS pulmonary metastases; isolated lung perfusion can concentrate adenoviral delivery to further inhibit pulmonary metastases; and Ad-OC-TK/ACV maintains potency in osteocalcin+ cell lines and can be administered systemically.

Q/A Session

Q: Does this localize to growing bone?

A: The answer is yes. We need to address this more, possibly in adolescent rodents.

Q: Any data on presence of osteocalcin levels in patients with osteosarcoma?

A: Yes, high levels are seen in a series of metastatic and primary osteosarcomas. There are also cases where they are not so typical.

Slow Release Platinum – Stephen J. Withrow, D.V.M.

Dr. Withrow discussed slow release platinum (open cell polylactic acid and cisplatin - OPLA Pt), and presented a dog model of osteosarcoma. OPLA Pt is administered via a sponge-like device that can be sized and shaped to necessary specifications. An immediate drug dose is given upon insertion, but the small holes of the sponge device allow for slower drug release. The sponge delivery device is biodegradable, like an absorbable suture.

The rationale for drug delivery by such a device includes: 1) the problem of local recurrence in many solid tumors; 2) some sarcoma patients are not salvaged because of marginal resections; 3) most marginal resections leave tumor close to the wound bed; and 4) intracavitary treatment increases the local dose while decreasing the systemic dose. Preliminary work includes in vitro elution in serum, mouse-transplantable carcinoma, and dogs-cortical allografts (tumor bearing dogs). If intravenous platinum is given, it is taken in quickly and depleted quickly. If the implant device is used, the release of platinum is slow.

The applications for OPLA include its use at the local tumor site for limb sparing-procedures or for soft tissue sarcomas. This delivery of platinum also acts as a radiation sensitizer.

A protocol design for canines was presented. Over a period of 10 years, 80 dogs had surgical resection of osteosarcoma with allograft placement. Half of the dogs received the platinum implant in the wound, and all dogs received I.V. chemotherapy. In the implant group there was a 15% recurrence rate, while the control group had a 55% recurrence rate. Among the dogs who had incomplete resection of their tumor, there was a statistically significant decrease in the tumor recurrence rate among the group that received the platinum implant .

Problems associated with the use of the implant can include infection and slow degradation of the polymer. Issues that remain unresolved include: 1) the best carrier and the best release profile; 2) the influence of a surgical drain and the radiation-heat effect on the drug activity; 3) whether this delivery technique will work with other drugs; and 4) whether this approach will be active in both local disease and metastatic disease.

Antiangiogenic Approaches - Mark W. Kieran, M.D.

Dr. Kieran described angiogenesis and the agents available to inhibit it. He then discussed potential surrogate markers of response, and reviewed issues of trial design. He defined anti-angiogenesis as the inhibition of the induction, stabilization or progression of neo-vascularization. The vascular endothelium, pericytes, vascular basement matrix, and bone marrow cells are all components of angiogenesis that can also be targeted. Regulatory molecules include: Ang1, Tie2, Ang2, VEGF, EphB2, and EphB4.

Available anti-angiogenesis agents include SU5416, squalamine, thalidomide, celecoxib, oral cytoxan, and oral VP-16. SU5416 is a small molecule inhibitor of VEGFR that is currently in a Phase I PBTC trial. Squalamine is a NHE3 pump inhibitor in a new Phase I trial in COG with carboplatin. Thalidomide is currently in phase II trials with other agents; it inhibits bFGF and VEGF signals. Celecoxib inhibits COX-2 activity of endothelial cells and is in Phase II trials in combination with other agents (including celecoxib, oral cytoxan and oral VP-16 above). There are a variety of oncogenes that relate to angiogenesis including K-ras, H-ras, c-myb, N-myc, c-myc, HER-2, EGFR, and PyMT for which inhibitors are now available. STI571 and ZD1839 are now being studied for their antiangiogenic effects.

There are currently no good surrogate markers of angiogenic response; even standard radiographic markers of response are inadequate for efficacy. Treatment endpoints including microvessel density, bFGF, VEGF, PDGF, etc levels, endothelial proliferation, migration or tube formation, bone marrow derived endothelial precursors are being studied as surrogate markers of angiogenesis inhibition.

The standard phase II trials identify a radiographic treatment response as a complete response, partial response, or stable disease. Progressive disease is defined as a greater than 25% increase in tumor size. In clinical trials of antiangiogenic agents, patients would remain on study unless the increase in tumor size was greater than 100% . Patients would undergo response evaluation at 8 weeks, and patients with progression (if clinically appropriate) could remain on therapy for up to 16 weeks. The reason for this stems from the over-abundance of tumor induced blood vessels. Unlike standard chemotherapy which shrinks the tumor within a few weeks if effective, antiangiogenic agents that kill part of the tumor vasculature can leave enough vessels behind so that the tumor can keep growing. Only after prolonged therapy (depending on the treatment and tumor) will tumor stabilization and eventually tumor regression be possible.

Issues to be considered in the design of antiangiogenic agent clinical study include the length of time a patient should remain on therapy and the timeframe within which a tumor response should be expected. In addition, if a patient demonstrates tumor progression on an antiangiogenic agent, one could add additional agents while continuing the initial one since endothelial cell resistance is unexpected. Tumors that initially respond and then progress on an antiangiogenic agent may have induced other stimulators of neovascularization. As such, adding in agents to cover the new stimulator may require that the original stimulator continue to be suppressed.

In summary, Dr. Kieran stated that a limited number of antiangiogenic agents are available in pediatrics. The issues related to study endpoints and the lack of good surrogate markers of response make evaluation of agent activity difficult. The potential application of new radiologic technologies is limited by expense.

Bisphosphonates - John H. Healey, M.D.

Dr. Healey discussed bisphosphonates, which were first identified in 1965. The structure is based on pyrophosphate, and there are two families of bisphosphonates. There are three essential elements to their activity: direct effects on mature osteoclasts; effects on osteoclast precursor cells; and effects on osteoblasts. The bisphosphonates are negatively charged and the means of cell entry is still unknown. These chemicals bind to the bone for the duration of the half-life of the bone, which is 10-12 years in humans, three years in dogs, and 200 days in rats.

The activities of bisphosphonates in bone cells include apoptosis, inhibiting cytokines, inhibiting metalloproteinases, disrupting adhesion molecules, immunomodulation against myeloma, and antiangiogenesis mechanisms. Specifically, bisphosphonates inhibit IL–6, a cytokine which promotes myeloma and osteosarcoma cell growth and encourages angiogenesis. Inhibition of metalloproteinases is important because metalloproteinases are present in bone resorption patients without osteoclasts.

Research in bisphosphonates and fracture healing has shown the following: etidronate causes ostomalacia and retards fracture healing; alendronate at 2 mg/kg/day caused larger, less mineralized callus that had normal strength; incadronate (YM-175) at 10–100 µg/kg for long-term (25-49 weeks) in rats delayed fracture healing, but it didn’t impair the mechanical strength.

Bisphosphonates prevent bone loss and fractures, and also have the ability to prevent osseous and non-osseous metastases. This is well established in patients with breast cancer and microscopic bone marrow "metastases" on staging studies. (Diel, IJ NEJM 98) Potential roles for bisphosphonates in sarcoma patients include preventing metastases, preventing osteoporosis, preventing periprosthetic osteolysis from particulate debris, and preventing bone loss from the stress bypass effect. It is clinically important to look at osteoporosis in children with sarcoma. Nine of 108 lower extremity osteosarcoma and Ewing's Sarcoma patients treated ‘90-’99 suffered a lower extremity fracture due to treatment-related osteoporosis.

In preclinical studies, Walker 256B carcinoma was implanted intraosseously into Wistar-Lewis rats to view experimental metastases. Pretreatment with clodronate inhibited the development of bone metastases compared with controls. Shorter intervals between the bisphosphonate therapy and the inoculation of tumor cells gave the best results. This data suggest, that in the human setting, early therapy might give better results.

Toxicities associated with bisphosphonates administration include fever, uveitis, leukopenia (transient, mild), and myalgia (GI upset rare). There have been no particular problems reported in children who have been treated for osteogenesis imperfecta, hypercalcemia of malignancy, and fibrous dysplasia.

In conclusion, bisphosphonates are active against osteoporosis, in prevention of metastatic tumor development, and possibly against primary tumor progression. These agents are safe in adults, but the data with children is limited.

Q/A Session

Q: Dr. Kieran was asked what should be done now to look at angiogenesis in osteosarcoma?

A: Right now thalidomide, Celebrex, and other agents are being used. A trial started two months ago at DFCC has already completed half of its accrual.

Guest Presentation: Safety and Pharmacokinetics of Celecoxib when Combined with Metronomic Administration of Low-Dose Chemotherapy of Pediatric Recurrent Solid Tumors - David Malkin, M.D., on behalf of Sylvain Baruchel

Dr. Malkin discussed the combination of metronomic chemotherapy and celecoxib as an antiangiogenic treatment. Metronomic dosing of chemotherapy uses lower doses administered more frequently, blocks endothelial cell proliferation, appears more tolerable, and there is no acquired drug resistance (endothelial cells are normal, homogeneous, and genetically stable). COX-2 is expressed within tumor neo-vasculature, neoplastic cells, and stroma cells (quiescent vasculature only expresses COX-1). Celecoxib is antiangiogenic at pharmacologic doses. Vinblastine is also being used.

COX-2 is upregulated in most human tumors. COX-2 -/- mice have decreased VEGF expression, reduced tumor angiogenesis, and tumor growth. COX-2 inhibitors decrease VEGF production and prevent VEGF-induced MAPK activation in the endothelial cell compartment. COX-2 inhibitor suppresses activation of small GTPase Cdc42/Rac and alphaVbeta3 integrin in vitro and in vivo.

The objectives of this study are to evaluate the safety and toxicity of chronically administered low-dose chemotherapy plus celecoxib. Pharmacokinetics are also examined, and a response rate is estimated. Other objectives are to validate angiogenic growth factors (VEGF, bFGF, VCAM-1) as surrogate markers of angiogenesis, and to evaluate dynamic MRI (dMRI) as a tool to assess specific antiangiogenic tumor response. The study design is open label, non-randomized, Canadian multicenter Phase I/II pilot study, for patients >21 with histologically or cytologically confirmed solid tumor (excluding brain stem tumors) measurable by radiographic imaging. dMRI parameters are being used to provide tissue perfusion, blood volume, capillary permeability, and extravascular volume.

Thus far, 17 patients have been enrolled, with five having osteosarcomas. Celecoxib has been well tolerated. There have only been three adverse events reported, although they are most likely due to cyclophosphamide and not celecoxib, except for a rash.

Surrogate markers of tumor angiogenesis have a large interpatient variability (especially with VEGF and bFGF), and more patients are needed to validate these biomarkers. However, VEGF and bFGF showed a trend of decreasing during the course of treatment despite observing tumor progression.

Challenges include: defining the optimum biological dose of COX-2 inhibitors in combination with chemotherapy based on clinical and biological, validated surrogate markers.

Discussion – Mark Bernstein, M.D.

Comments: The best administration was when the drug was given every couple of days. More an issue of vincristine not being given often enough to show the improvement.

There is a lot of work suggesting that stroma within tumors are secreting a variety of factors that help with regulation. Therefore, these agents may be affecting the epithelial cells but not affecting the stromal cells that are also helping the proliferation of the tumor.

Q to Dr. Kieran: Is there a chance that the tumor is being depleted of cells that it needs to keep going?

A: There is no question that when any cell type is inhibited, a multitude of effects can be had. There is no single mechanism for all things.

Q to Dr. Healy: Should bisphosphonates be good in lung metastases because it worked in breast, even though those breast patients already had bone marrow metastases; would it be worrisome that it was going after the bone?

A: Yes, but it still reduced both types of metastases. In the orthopedic world these agents are going to be used, but the animal models must be reviewed to see what the data show.

New Target Identification in Osteosarcoma and Model Systems for Testing

cDNA Expression Arrays & Oligonucleotide Expression Arrays - Paul S. Meltzer, M.D., Ph.D.

Dr. Meltzer discussed gene expression profiling for osteosarcoma with cDNA microarrays. With a growing catalog of over 30,000 human genes there is a need to recognize this resource as a catalog of candidate drug targets.

Thus far, from the expression profiling of cancers, it has been learned that distinct cancers have distinct patterns of gene expression. Using expression data, it is possible to develop robust formal diagnostic classifications. Novel subgroups can be recognized which lack previously defined clinico-pathologic correlates. Also, clinical application of array data does not necessarily require arrays; small groups of genes may carry most of the useful information.

Future directions include improving the diagnostic categorization of tumors, identifying useful predictive markers for outcome and therapeutic response (array or conventional), and identifying points for intervention (critical pathways, novel drug targets).

To use microarrays appropriately in a clinical study, proper steps need to be taken when designing trials. The question under investigation should be defined with an appropriate patient sample. There is a need to develop an appropriate and rigorous statistical analysis of array data in order to develop a formal classifier. The design should strive for a result using genes which carry information relevant to the question posed, and allow for validation of the data.

In a pilot study in osteosarcoma, one of the initial goals was to demonstrate whether high-quality microarray data could be generated from osteosarcoma. There were 14 primary osteosarcomas, four Ewing’s Sarcomas, eight osteosarcoma pulmonary metastases, one Ewing’s Sarcoma metastases, and two normal lung samples. There was a near 100% success rate of obtaining evaluable data from this sample set.

The data analysis for bone tumors in the pilot study, after being filtered for quality, showed on cluster analysis a strong cluster of Ewing’s Sarcoma -specific genes and a larger group of osteosarcoma-specific genes. By hierarchical clustering of samples there was significant variability seen with osteosarcoma samples, but not as much variability was seen with normal lung. When viewing the weighted gene list, genes most significantly associated with disease were looked for (these are most highly specific) and ranked by a weighted discriminator method, and a random permutation test was performed.

The results showed there is a very strong difference in signal between osteosarcoma and Ewing’s Sarcoma. There was also a very strong signal difference between osteosarcoma metastases and normal lung. The osteosarcoma metastases are very similar to what is seen in a primary osteosarcoma. Therefore, viewing osteosarcoma metastases vs. osteosarcoma primary may be a good way to see what is upregulated or downregulated during this process. The pilot study sample is not big enough to give good data, but this area should be looked at further.

Dr. Meltzer concluded that high-quality microarray data can be obtained from bone tumors despite their content of extracellular matrix, and clinical correlations should be supported in large studies.

Q/A Session

Q: Why wasn’t osteosarcoma compared to normal bone?

A: No normal bone was available. Regeneration tissue needs to be looked at since bone is a regenerating tissue. With microarrays, it is always better to look at more data from additional samples and relevant tissues.

Q: Have you looked at any small-cell osteosarcomas?

A: No, there were none available to study.

cDNA Expression Arrays & Oligonucleotide Expression Arrays - Deborah E. Schofield, M.D.

Dr. Schofield discussed the analysis of osteosarcoma using oligonucleotide expression arrays. In a study she presented, samples of frozen tissue from 34 primary, childhood or adolescent osteosarcomas, all treated on CCG or German protocols, were analyzed. Follow-up was 3-10 years. There were 18 non-responders and 16 responders. Sample processing was as follows: tissue section and selection, RNA extraction +/- amplification, cDNA synthesis, in vitro transcription and cRNA labeling, and fragment cRNA mix.

Many investigators have been trying to amplify the tissue since there is a lack of samples. Good linear amplification has been demonstrated so perhaps the needle biopsies may supply enough material for these evaluations.

Affymetrix U95Av2 arrays were used. A series of oligonucleotides were synthesized in situ by photolithography, representing known exonic sequences from 3’ to 5’ of ~12,500 genes of interest. Bioinformatics has been key to deciphering significant information from the data output.

Molecular targeting for cancer therapeutics involves the identification of specific gene targets with known biologic function for which therapeutic interventions may be available. The Genetrix tool was used to look at clusters and specific genes. p53 and telomerase expression levels were examined for their relationship to survival and the data showed this to not be a very good predictor of outcome. Her2 expression levels were also looked at, but not as a predictor of outcomes.

For rhabdomyosarcoma, the most expressed gene was FGFR-4 (fibroblast growth factor receptor 4). This gene showed some prognostic significance depending on where the delineation for outcome is defined. Other genes associated with death include: sodium channel 2 (hBNaC2) mRNA, VIP2 receptor KIAA0192 gene (partial cds), PEG3 mRNA (partial cds), RNA polymerase II, 14.5 Kda subunit, autoantigen mRNA, serine/threonine protein kinase EMK, CD44 gene (cell surface hyaluronate receptor gene), and EPH-related receptor tyrosine kinase.

For osteosarcoma, the top gene with the strongest association for survival was Growth Arrest Specific 1 (GAS-1). Other genes associated with survival include: protein tyrosine phosphatase non-receptor type 1, hematopoietic cell-specific Lyn substrate 1, folate receptor 2 (fetal), 47h11 Homo sapiens cDNA, Ta08106 k1 Homo sapiens cDNA, Syndecan T, paired mesoderm homeobox T, ELAV (embryonic lethal abnormal vision), and constitutive endothelial nitric oxide synthase gene. The top gene associated with death was Bone Morphogenetic Protein 4 (BMP-4). Other genes associated with death include: guanine nucleotide binding protein (G protein), sperm associated antigen B, KlAAO751 gene product, Galanin, G protein-coupled receptor 39, translocase of inner mitochondrial membrane 8 (yeast) homolog A, nitric oxide synthase 2A (inducible, hepatocytes), 32g4 Homo sapiens cDNA, and Wnt 10B. CD-31 didn’t show up on the list of genes and has the opposite effect of what would be predicted. Very high levels of CD-31 are associated with survival but only if the cutoff is very high. Tumor suppressing subtransferable candidate 3 showed the strongest association with histologic response to therapy, and tyrosine kinase 2 showed the strongest association with lack of histologic response to therapy.

It is hoped that gene expression profiling may be used to predict prognosis and identify potential novel therapeutic targets in these tumors.

Q/A Session

Comment: Care should be taken. A microarray does not allow viewing of a single gene. Pathways should be looked at. The technique is not designed to look at single genes, rather to look at the pathway.

Comment: The really important issue that biostatisticians have been looking at is sample size requirements for discrimination. The sample size is specific to the array being produced and the hypothesis being testing.

Comment: Yes, that is true. But it is really critical to look at outcome events and first define homogenous sets.

Q: Research has to look beyond the p53 pathway. With Dr. Meltzer’s algorithm, it is not going to be the p53 pathway; it will be much more complicated in osteosarcoma. Technical question: what are the benefits to two different types of arrays being used?

A: The short answer is cost is a big factor with different techniques. It is difficult but not impossible to make a comparison.

Q: What is difference/benefit between Affymetrix and spotted arrays?

A: Very complicated, but Affymetrix appears to give better data.

Comment: Go back to the previous question. A separate set of data for validation is needed. As criteria are developed for how many samples to look at, they should be doubled so they can be validated. Sample size should essentially be made 2x larger so enough samples will be collected to properly to test and validate.

Q: Most of our patients have pre- and post-therapy samples. Looking at histologic response is pretty much subjective, but has what is left as predictive been looked at? Do you have pre- and post- samples?

A: Only a handful, but also have primary and metastases on a handful of patients. Much of what is frozen is necrotic tissue, and it is not very viable tissue.

Q: Is the data ready to give clues as to what should be researched in osteosarcoma?

A: Probably not yet. This is pilot testing data, and it needs to be done on larger groups. Participants will await new data with interest.

Comment: Not every gene is on the chips or is not measured well. So just because the gene isn’t in the results, it may not be on the chip.

Proteomics/CGH - Ching C. Lau, M.D., Ph.D.

Dr. Lau discussed proteomics and CGH.

There are three levels of comprehensive screening targets: genome, transcriptome, and the proteome. The genome is huge and only a small portion actually codes for proteins. The transcriptome (mRNA) is a smaller subset to look at but still problematic because only a subset is expressed in any one cell type at any one time. However, high-throughput technology is available to screen the transcriptome. Finally, purists would argue to view only the proteome because this is a smaller subset and contains only proteins that determine the biological behavior of a cell. Furthermore, there is ample evidence that mRNA expression level may not necessarily correlate with protein expression level or activity due to post-translational modifications and sub-cellular localization. There are many examples where expression level of mRNA is constant but the protein level is variable. Therefore, protein screening theoretically gives you more relevant data. However, proteomics is difficult to carry out, because high-throughput technology is not readily available and existing technology is not very good for quantitation. It is very labor-intensive with low output.

Therefore, RNA screening is currently the most popular and feasible. Dr. Lau briefly discussed their efforts in using cDNA microarrays to generate expression profiles for molecular classification of osteosarcoma. The objectives included: discovering classes of gene expression profiles in samples of primary osteosarcoma; comparing gene expression of primary osteosarcoma to expression of normal osteoblasts; comparing gene expression of osteosarcoma before and after chemotherapy; comparing gene expression of primary osteosarcoma to expression of lung metastases for the same patient, for those patients who present with lung metastases, and in patients who later develop lung metastases; predict response to neoadjuvant therapy; and predict long-term outcome. One of the technical hurdles in analyzing osteosarcoma samples is the small quantity of biopsy specimens and low number of viable cells in the surgical specimens of good responders to neo-adjuvant chemotherapy. Thus RNA amplification is required to minimize sample size requirements but it has to be done carefully in order not to introduce bias into the profile. The T7 amplification schema is favorable for this. Summary of T7 studies: optimized T7 procedure yields >500 fold amplification; 10-20 % are false positives, but very few false negatives (more worried about false negatives so this is acceptable); most of the DE genes from amplified RNA have higher DE ratios than from unamplified RNA (perhaps some of the false positives are actually true positives that become detectable because of increased sensitivity after T7 amplification and can’t be picked up in unamplified studies); and over 90% of the top 100 significant genes were also present in unamplified RNA. Clustering results showed minimal difference between amplified and unamplified samples.

mRNA is not the only way to get data. The CGH method detects gains/losses/amplifications in a single hybridization and maps to a chromosomal region. No cell culture is needed, and archival materials can be used. Limitations include: its limited resolution, its dependence on chromosomal morphology, the labor intensity for data generation, and the need for confirmation by FISH. A significant improvement is to use array CGH which replaces normal metaphase chromosomes with a carefully chosen series of BACs as a platform for detecting chromosomal abnormalities. There is good correlation between the chromosomal CGH vs. array CGH, but the array has higher resolution.

CGH can only detect net gain of genetic materials and should be complemented by spectral karyotyping (SKY). SKY is a 24-color FISH technique that can be used to detect subtle translocation. If you couple SKY with CGH, additional data is given and a better picture of what is going on. 8q24, 12p11, and 14q24 are some of the interesting recurrent regions of abnormalities in osteosarcoma worth looking at.

Finally, Dr. Lau summarized some proteomics data generated using the SELDI ProteinChip Arrays. It can be used for capturing protein in a solid phase and create protein profiles that can be identified by MS-TOF.

Genes Related to Metastases - G. Peter Beardsley, M.D., Ph.D.

Dr. Beardsley discussed genes that are related to metastases. The basic premise of his presentation was that metastasis is a major clinical feature of osteosarcoma and is the proximate cause of death. Therefore, the expression of genes associated with the "metastatic phenotype" in tumor tissue is likely to be predictive of outcome in osteosarcoma. A caution on tumor heterogeneity: a small proportion of tumor cells actually undergo metastasis, and the properties of the gross tumor many not reflect the primary tumor’s true metastatic potential.

The focus of his current project is on the c-met proto-oncogene and matrix metalloproteinases. Cellular responses to c-met ligand bindings activate growth cycle (mitogenesis), start membrane motors (motogenesis), and basement membrane lysis (morphogenesis). c-met immunostaining shows that high c-met expression is a strong negative prognostic indicator in breast cancer (Rimm, et al). Dr. Beardsley and his colleagues at Yale are examining c-met expression levels in osteosarcoma samples using tissue microarray methodology. This technique requires very small amounts of paraffin embedded material, which overcomes a significant problem in tissue acquisition. They are also working on c-met as a therapeutic target).

Matrix metalloproteinases exhibit the following family characteristics: high sequence homology, are enodpeptidases, are Ca++ and/or Zn++ dependent, are involved in the lysis/remodeling of the extracellular matrix, and number more than a dozen. MMP expression shows varied correlations with tumor growth and invasion (good: MMP-2, MMP-9, MMP-14; poor: Matrilysin, stromelysin-3). MMPs as therapeutic targets: Several inhibitors of MMPs are in clinical trials. Although overall results have been disappointing, some activity has been seen in established tumors. They may work best early in therapy to decrease metastasis, and this kind of activity would not be detected in standard Phase I or II clinical trials.

Canine Models - Chand Khanna, D.V.M., Ph.D./Stephen J. Withrow, D.V.M.

Dr. Stephen Withrow presented information on the canine model for bone cancer research. The etiology of canine bone cancer is largely unknown. Large body size and the purity of breed may be contributing factors.

In dogs, more males are affected by osteosarcoma, and 75% of the time it affects the long bones (front leg – 60%, rear leg – 40%). Stage IIB is the most common presentation, and low-grade histology is rarely seen. At presentation, 15% are Stage III. Pulmonary metastases are very common (leading cause of death). Local control is done through NSAIDs, palliative radiation, full-course radiation, amputation, and limb sparing surgery.

There is scientific support of the dog as a model. Osteosarcoma develops spontaneously in dogs, and dogs are large animals that have similar physiology to humans. There is also a rapid case accrual, and autopsy compliance is common. The metastatic rate of osteosarcoma in dogs is similar to humans (80-90%, lung>bone), but time to local and systemic recurrence is quicker in dogs.

Eligibility for treatment on protocol at Colorado State University calls for "localized disease," <50% of bone length involved, a projected one-year survival, and necropsy. Surgery includes "marginal-marginal" excision, allograft replacement, or plate arthrodesis. A treatment regimen model involves cisplatin at Day 1 and 21 , radiation (10fxns, Monday, Wednesday, and Friday), surgery at Day 42, and monthly follow-up. Percent of tumor necrosis and local recurrence mimics the human presentation of the disease.

Prognostic variables that appear to be indifferent for survival are breed, sex, site (except mandible), whether or not limb is spared vs. amputated, and its histologic variant. Poor prognostic variables include young age, metastases at diagnosis, large volume, and high bone ALP. New studies underway include use of chemotherapy (gemcitabine), antiangiogenesis agents, gene therapy, immunotherapy, or bisphosphonates.

The dog model shows potential in causation and epidemiology. Trials can be done in institutions, but need specific resources to make them successful. The limitations of the dog model include: human proteins and immune response to these reagents and therapy; poor canine genome definition; drug/trial costs (good and bad); and toxicity and owner compliance.

Dr. Chand Khanna continued with the biology of canine osteosarcoma. p53 is altered in approximately 50% of tumors. Canine osteosarcoma has RB dysregulations, cytogenetic aberrations, and cDNA microarray cross-reactivity. There will eventually be a dog cDNA microarray, and there is continuing work on the Dog Genome Project. IGF-I inhibition plus chemotherapy in osteosarcoma was investigated. OncoLar was used to see if it could be moved to pediatric patients in combination with chemotherapy. "In vivo canine with OncoLar and carboplatin" used OncoLar + chemo + amputation in one arm, and chemo + amputation in the other arm (a great source of tissue is from amputation). Among 64 cases in 8 months, with survival as the clinical endpoint, there was no difference in necrosis as a primary tumor response, no significant difference in apoptosis between OncoLar and placebo group, and survival showed no improvement in dogs receiving OncoLar. This combination may not represent the best treatment approach with OncoLar.

Ongoing and future projects include: examining the influence of 70% serum IGF-I suppression on osteosarcoma primary tumor growth and metastasis; examining local regulation and production of IGF-I and GH in osteosarcoma; and evaluation of GH antagonists and GH-RH antagonists (murine K-series osteosarcoma model and canine osteosarcoma).

The dog model is quite valuable and hopefully can become part of the standard preclinical approach to therapy development.

Metastases Associated Genes - Chand Khanna, D.V.M., Ph.D.

Dr. Khanna presented additional information on the genetic determinants of osteosarcoma metastases.

In the murine osteosarcoma model, K7M2 has a more aggressive and K12 has a less aggressive metastatic potential. The murine microarray analysis, a 4k array, looked at 59 differentially expressed genes from K7M2 primary tumor (34 genes overexpressed) and K12 primary tumor (25 genes overexpressed). K7M2 shows greater genetic expression of genes related to cytoskeletal motility, adherence, and angiogenesis than K12.

Ezrin protein is expressed at the cell membrane in the more aggressive cells. These proteins are phosphorylated at the membrane where they are activated, and are important for signaling and interaction. A tissue array using Ezrin in dogs with osteosarcoma showed that metastases have a statistically higher level of Ezrin expression compared to primary tumor. Dogs with any level of Ezrin expression have poorer survival than dogs not expressing the protein, but this survival difference is not statistically significant. As for Ezrin in human osteosarcoma, there also may be an association with metastases. Exploratory results were somewhat disappointing, as Ezrin expression was detected in human osteosarcoma, but expression was not clinically distinctive.

For ongoing and future work, the murine osteosarcoma model will be utilized for the examination of the biology of Ezrin mutants in vitro/in vivo and Ezrin regulation during metastasis. For the spontaneous canine osteosarcoma, development of larger tissue arrays and reevaluation of phosphorylated Ezrin expression will be explored. For pediatric osteosarcoma, increasing sample size (metastases at diagnosis), reevaluating phosphorylated Ezrin expression, and standardizing tissue collection will be investigated.

Evaluation of Camptothecins, Epothilones and Signaling Inhibitors in Osteosarcoma Models - Peter Houghton, Ph.D.

Dr. Houghton began his talk by stating he believed that the best types of models would be useful for chemotoxic as well as molecularly targeted agents. Animal models of osteosarcoma can be spontaneous, transplanted (syngeneic), transplanted (xenogeneic), or orthotopic. A perfect model does not exist, but one needs to be found that parallels sensitivity to human cancer, parallels the genetic basis for human cancer, parallels the biology of human cancer, and allows for high throughput for drug screening.

Xenograft models of childhood cancer include rhabdomyosarcoma, neuroblastoma, Ewing’s Sarcoma, brain tumors, Wilms’ Tumor, and osteosarcoma. Animal models of cancer can help in understanding the genetic basis of cancer, developing cancer prevention and cancer treatment techniques, identifying carcinogens, and in drug development/PK/safety.

Preclinical development/validation issues with osteosarcoma models include the following: does the model identify clinically active agents prospectively; does the model encompass known clinical heterogeneity; what is the basis for false prediction; and can we develop approaches that minimize false predictors.

Clinically, there seems to be relative agreement with xenografts. Irinotecan (CPT-11) is somewhat active in osteosarcoma xenografts, but these xenografts are much less sensitive than rhabdomyosarcoma, neuroblastoma, or Wilms’ Tumor models. Signaling inhibitors (Gleevec, Iressa) do not have significant single agent activity; and Epothilone B (BMS247550) has some activity. Dr. Houghton concluded that osteosarcoma xenografts may be useful models for identifying new agents and combinations. They are generally less sensitive than other childhood tumors to camptothecins. Furthermore, orthotopic models may have some value, specifically to lung metastases.

Guest Presentation: Activation of a Redundant Caspase Pathway in Bones of p53 Null Mice - Janet M. Hock, B.D.S., Ph.D.

Osteosarcoma is thought to be a tumor of osteoprogenitor cells: multipotential, hormone-responsive stromal cells in the periosteum and marrow, which are capable of differentiating into many lineages depending on their environmental cytokines. As in other tumors, the p53 gene has been linked to increased susceptibility to osteosarcoma. Published studies suggest that p53 may have a role in the normal development and physiology of bone. Failures in skull growth and delayed longitudinal bone growth have been described in utero in p53 null mice. We undertook an extensive characterization of bone, including hormonal responsive to parathyroid hormone, in p53 null mice aged 6 and 12 weeks. Once daily parathyroid hormone, which has anabolic effects on bone mass and strength, promoted osteosarcoma in rats after 18 months of treatment in a 2-year toxicology study. Although the relevance of PTH-induced rat osteosarcoma to humans is not known, it may provide an animal model for studies of bone sarcomagenesis.

Our data showed that p53 null mice had smaller, thinner bones than their wild-type counterparts. The smaller bones were associated with decreased surface bone formation. Premature closure of sutures at the base of the spine was associated with loss of symmetry in the maxilla, suggesting a key role for p53 in normal development of the craniofacial complex. When challenged with PTH, expression of the AP-1 complex of genes, specifically c-fos and fra-2, was blunted in p53 null mice while in vitro induction of osteoclast differentiation was enhanced. As p53 is a key regulatory gene determining apoptosis and PTH inhibits apoptosis, we investigated bone cell caspase activity as a surrogate marker of apoptosis. PTH inhibited caspase activity in both p53 intact and null mice. However, p53 null mice had switched to an alternate caspase pathway as activity was seen on YVAD substrate, which recognizes caspase 1 (interleukin-converting enzyme-beta) rather than on DEVD substrate that recognizes the apoptotic caspases 2, 3 and 7. The implication of this change in caspase pathways on skeletal function and osteosarcoma susceptibility is unknown.

Q: How hard is it to get osteoblasts?

A: It depends. Early progenitors are in bone marrow, and they also can come out of other bone that is more advanced, but those aren’t as good in culture.

Discussion - Paul A. Meyers, M.D.

Q to Dr. Grier: The dog model is very important. The question is how to, or what organization, can gather more dogs. Is there further structure that is needed or more money? Is the money there from NCI?

A: The dog is a good model, but funding from an R01 mechanism is very difficult. People don’t see the relevance; osteosarcoma is a rare disease, and the research is in dogs.

A by Dr. Khanna: The way to move forward is to make the dog model part of a larger program. Defining questions that are important in this type of group could lead to a P01.

A by Dr. Toretsky: The model is very interesting, although, it is unknown how many of the other drugs correlate with human response. Is it going to teach about biology and replicate what would be seen in humans? Probably a directed RFA would be important.

Comment: Whether the model is predictive or not is dependent on the agent. Some agents are predictive and others may not be. It looks like the dog model is part of a component about how to get better ideas of where to be going. As for money, it would be helpful to describe what research should be done, and so the funds can be properly allocated.

Comment: The canine model seems great. It can be further enhanced, though, to enhance PK studies that go along with the model. Some of our plans should focus on this.

Concluding Remarks - Barry D. Anderson, M.D., Ph.D., and Richard Gorlick, M.D.

Pathogenesis of Osteosarcoma

Lessons:

•   Osteosarcoma likely results from a variety of genetic alterations far more complicated than many other pediatric malignancies
•   No signature translocations have been identified
•   Complex karotypes indicate tremendous genetic instability
•   Subtypes of osteosarcoma may be distinguished by viral or suppressor gene alterations
•   SV40 sequence in some tumors – T-Ag expression
•   Blockade of p53/RB function by T-Ag
•   Osteosarcoma pathobiology may evolve through a series of genetic alterations or cell regulatory modifications
•   A correlation between telomerase and aggressiveness remains unclear

Future Directions:
•   Biology samples to be collected or processed?
Discussion:

Should we look at mismatch repair genes? People expressed concern that we not look at one gene at a time. Massive genomic instability is the common genetic finding in osteosarcoma and this is a biologic issue that has not been explored. There are fairly straightforward ways to screen for mismatch repair. Should we incorporate people who have experience with DNA repair in this future work? This approach was questioned because an association of repair deficit syndrome was not known.

What type of samples need to be collected? Is all paraffin embedded the same? How is it processed? Are you trying to look at which studies you can do with tissue you already have or what new studies can do and how to get more tissue? Both. Tissue samples are becoming available and we need to know what to study with them and how to best collect them. Need to figure out best use of tissue since it is not an unlimited resource.

In terms of paraffin blocks, do you collect this? Maybe paraffin can be used for validation studies rather than primary studies. The COG 9851 protocol requirements include paraffin embedded tissue. The type of paraffin fixation is not specified, as this is viewed as the domain of the pathologist to specify this. The other issue is storage of the slides.

Cytokine/Death Receptor Pathways in Osteosarcoma

Lessons:
•   Influence of Fas/FasL in osteosarcoma pulmonary metastases (with possible explanation of INT-0133)
•   Multiple (and complicated) pathways to tumor control in the IL-12/pulse IL-2 therapy
•   Knowledge gained by looking at multiple cell systems and fresh tumor in Ewing’s Sarcoma and neuroblastoma
•   Ability to move toward cytokine/chemotherapy combination trials in colon cancer

Forward Movement:
•   Further work in osteosarcoma animal models
•   Further characterization of cytokine/cytokine receptors in the SAOS-LM6 model
•   Confirmation in other osteosarcoma lines
•   Fas/FasL in patients at presentation
•   Cytokine changes with aerosolized GMCSF study
•   Clinical background for possible future rx with aerosolized IL-12

Drug Resistance/Growth Factor Pathways in Osteosarcoma

Growth Factor Pathways:
•   Antibody-based strategies for targeted therapy are interesting, but too preliminary for osteosarcoma studies
•   IGF-1/GH – preliminary data suggest it might have a role in tumor etiology
•   P9851 might help better define the role of IGF-1 in osteosarcoma
•   Further preclinical evaluation is important to attempt to define its role as a therapeutic target

Discussion:

A participant added that before things are dismissed, you must understand how to measure success. Survival may incorporate too many things, and you may not exactly see the potential of the agent. Future symposium should have a section on the evaluation of endpoints to properly measure effectiveness of drug. But survival still has to be the gold standard. Others can be looked at, but survival is the most important.

Small molecule inhibitors for IGF-I may be available within 2-5 years.

P-glycoprotein is a low priority.

Comment on prioritization and use of samples. Use NWTSG as an example. They also have a reserve bank that no one touches unless it is high priority. Should also use their ideas to correlate markers with lower numbers.

New Treatment Approaches

New Treatment Approaches:
•   Concepts and Protocols in Development
•   Samarium: for patients with bone recurrence
•   Inhaled GMCSF: for patients with pulmonary only recurrence
•   Trimetrexate: window study for patients with disease metastatic at initial diagnosis, not eligible for Herceptin study
•   Viral approaches: needs further development work - MSK study
•   Slow release platinum: potentially interesting idea for incorporation: efficacy data in dogs, need safety and preliminary data in humans, discussion on priority
•   Antiangiogenic approach: in development for Ewing’s Sarcoma metastatic at diagnosis (vinblastine 1 mg/m2 3X/week with celecoxib 250 mg/m2 BID), as a backbone for standard cytotoxic treatment; squalamine could be incorporated with cisplatin
•   Bisphosphonates: data collection

Slow release platinum: it is currently unavailable; maybe Peter Anderson can look at it in xenograft model to get more pharmacology information on it. Combination with radiotherapy may work to get it into a phase I trial...possibly head and neck. The anti-angiogenic approach is not ready to advance. Bisphosphonates still need data. Someone suggested using a canine model to test all of these as a single agent and then add bisphosphonates.

New Target Identification in Osteosarcoma and Model Systems for Testing

Discussion:

All meeting participants were supporters of array technology. A further look at canine models was encouraged. If Peter Houghton can identify highly valuable compounds, then there is a need to pursue the dog model and increase testing in xenograft models.

Back to Scientific Conferences Home