Purpose:BRCA1/2 germline mutations are associated with impaired DNA double-strand break repair. We tested whether breast cancers in BRCA1/2 mutation carriers were more responsive to induction treatments than in noncarriers. Patients and Methods:The BRCA1 and BRCA2 genes were screened for germline mutation in a retrospective cohort of 90 patients (with 93 tumors) with a family history of breast and/or ovarian cancer, treated with induction anthracycline-containing chemotherapy and/or radiotherapy. Median tumor size was 40 mm. Clinical responses and breast preservation rates were correlated to BRCA1/2 mutation status, and to other clinical and pathologic factors. Results:A complete clinical response was achieved in 15/39 (46%) BRCA1/2-mutated tumors and in 7/54 (17%) nonmutated tumors (P = 0.008). Complete or major clinical response rate was observed in 55 of the 74 tumors treated with induction chemotherapy (74.3%). The overall complete or major clinical response rate in the tumors treated with induction radiotherapy was 68% (13/19 tumors). Following induction treatment by either chemotherapy or radiotherapy, more breast-conserving treatments could be performed in mutation carriers than in noncarriers: the rates of breast preservation were 82% in BRCA1/2-mutated tumors and 63% in nonmutated tumors, respectively (P = 0.045). BRCA1 mutation was the sole predictor of breast conservation. Conclusion:Breast conservation after induction treatment was higher in BRmut+ tumors, and clinical response was related to aggressive tumor features correlated with BRCA1/2 mutations. This suggests that impaired repair mechanisms related to the BRCA1/2 mutations increased the chemosensitivity and radiosensitivity of large breast cancers. Further studies will need to determine the long-term outcome in these patients.

BACKGROUNDAngiosarcomas are rare, heterogeneous and a retrospective study was conducted to describe their natural history.PATIENTS AND METHODSWe reviewed 161 files of angiosarcoma treated in three institutions of the French Sarcoma Group from 1980 to 2004. Survival and prognostic factors for survival were analyzed.RESULTSMedian age was 52 years. Primary sites were the breast (35%), skin (20%) and soft tissues (13%). At initial diagnosis, 31 (19%) had metastases. Surgery was the first treatment in 121 (75%) patients combined with chemotherapy or radiotherapy in 34 and 32, respectively. Ninety (74%) of these 121 patients relapsed, mostly locally (50). With an average time since initial diagnosis of 8.1 years, 123 (76%) patients progressed and 76 (47%) died. Median survival was 3.4 years [95% confidence interval (CI) 2.4-5.8], and the 5-year overall survival (OS) rate was 43% (95% CI 33-53). In multivariate analysis, liver primary site [relative risk (RR) = 12.62], performance status (PS) of two or more (RR = 3.83), presence of metastases at diagnosis (RR = 2.50), soft tissue tumor (RR = 0.31) were correlated to OS. PS, liver and soft tissue tumors were identified as independent prognostic factors for progression-free survival.CONCLUSIONSAngiosarcomas have an overall poor outcome, but with a clearly distinct prognosis depending on the primary site.

10773 Background: Inflammatory breast carcinoma is a distinct clinicopathologic entity with poor survival outcome. Early response to initial chemotherapy was reported as predictive for survival and complete pathological response is considered as a major prognostic factor. This study was conducted to determine the safety and to estimate the efficacy of sequential high dose chemotherapy with doxorubicine (D) + cyclophosphamide (CPM) and docetaxel (Doc), in terms of pathological response, disease-free and overall survival Methods: Fifty-four patients entered into the trial from July 1997 to March 1999; treatment consisted in four 3 weekly cycles of D75 mg/m2 + CPM 6 g/m2 in C1 and C2; Doc 100 mg/m2 q 2w in C3, C4 and C6; D75 mg/m2 + CPM3 g/m2 in C6 and C7. Mastectomy with axillary dissection was performed for responding patients, followed by radiotherapy. Results: An interim analysis showed the same efficacy reported in previous high dose chemotherapy trial Pegase 02, without Docetaxel, and unexpected toxicity with 2 toxic deaths that warranted an early closure of the trial The overall 35% [22.5%–47.9%] grade 1 or 2 pathological response according to Chevallier as compared to the 32% reported in the Pagase 02 trial, suggest no greater efficacy for additionel Docetaxel to high dose D + CPM Five-year disease-free survival and overall survival showed a better tendency in Pegase 05 trial, with respectively 42% and 62% as compared to Pegase 02 trial with 33% and 50% Conclusions: these results suggest some benefit from dose intensified CT in terms of complete pathological response rate with improvment of the disease- free and survival outcome. The positive impact of additional docetaxel will be assessed in the last randomized multicenter Pegase 07 trial. No significant financial relationships to disclose.

9685 Background: Age ≤35 years is considered as a bad prognostic factor in women with breast cancer (BC). Methods: Clinical and pathological data of 926 women ≤35 years with invasive non metastatic BC, consecutively treated between 1990–99 in 8 french institutions were analysed. Clinical and pathological size, node involvement, histological type, SBR histological grading, hormone receptors (HR), and treatment characteristics were analysed for all patients. Distant Recurrence Free Survival (DRFS) and overall survival (OS) were analysed using Kaplan Meier estimates and comparisons were made using the logrank test. Multivariate analyses were carried out using Cox hazards model. Results: With a median FU of 87 months,239 deaths, 177 loco-regional and 323 distant recurrences were observed. Five and 10 years survival rates were 82.2% and 66.5%. Five and 10 years DRFS rates were 67.2% and 50.8%. No difference was observed according to age (≤30 years old, 31–33 and 34–35, p = 0.74) neither in DRFS nor in OS (p=0.99 and p = 0.74). Compared to T1, T2 was associated with a RR of 1.65 [1.23–2.22], and T3/T4 with a RR of 2.73 [1.92–3.88] for the OS (p<10-4) and T2 was associated with a RR of 1.57 [1.23–2.02], and T3/T4 with a RR of 2.25 [1.65–3.07] for the DRFS (p<10-4). Compared to negative node involvement, pN1 status was associated with a RR of 2.22 [1.60–3.09], and pN2 with a RR of 4.35 [3.14–6.03] in OS (p<10-4) and pN1 nodes involved was associated with a RR of 2.64 [2.01–3.47], and 4 and more with a RR of 4.22 [3.17–5.62] for the DRFS (p<10-4). Compared to positive HR, negative HR were respectively associated with a RR of 1.37 [1.05–1.82] (p=0.019), and 1.49 [1.14–1.96] (p<10-2) in OS. But no difference was observed according to HR in DRFS. Compared to grade I in SBR, grade II and III were respectively associated with a RR of 2.40 [1.35–4.27], and 2.61 [1.46–4.65] for the OS (p<10-3) and a RR of 1.86 [1.20–2.89], and 1.75 [1.12–2.72] in DRFS (p<10-3). Multivariate analysis results will be presented during the meeting. Conclusions: In these retrospective analyses, the most important prognostic factor for YA with breast cancer is tumour size followed by pathological grade (SBR), pN and HR status. No significant financial relationships to disclose.

Abstract Completion of the working draft of the human genome has made it possible to analyze the expression of genes according to their position on the chromosomes. Here, we used a transcriptome data analysis approach involving for each gene the calculation of the correlation between its expression profile and those of its neighbors. We used the U133 Affymetrix transcriptome data set for a series of 130 invasive ductal breast carcinomas to construct chromosomal maps of gene expression correlation (transcriptome correlation map). This highlighted nonrandom clusters of genes along the genome with correlated expression in tumors. Some of the gene clusters identified by this method probably arose because of genetic alterations, as most of the chromosomes with the highest percentage of correlated genes (1q, 8p, 8q, 16p, 16q, 17q, and 20q) were also the most frequent sites of genomic alterations in breast cancer. Our analysis showed that several known breast tumor amplicons (at 8p11-p12, 11q13, and 17q12) are located within clusters of genes with correlated expression. Using hierarchical clustering on samples and a Treeview representation of whole chromosome arms, we observed a higher-order organization of correlated genes, sometimes involving very large chromosomal domains that could extend to a whole chromosome arm. Transcription correlation maps are a new way of visualizing transcriptome data. They will help to identify new genes involved in tumor progression and new mechanisms of gene regulation in tumors.

9024 Background: Angiosarcoma is a heterogeneous group of rare sarcoma types. Its natural history and outcome remains poorly understood. In order to enhance our knowledge and to determine prognostic factors, a retrospective study of angiosarcomas was planned. Methods: Descriptive data of patients with angiosarcomas treated in three institutions between 1980/01/01 and 2004/04/30 were obtained retrospectively. The overall survival and prognostic factors were estimated using the Kaplan-Meier method and Cox’s proportional-hazards models. Hazard ratios (HR) are given with their confidence intervals (CI) at 95%. Results: Median age at the diagnosis was 52 on 161 patients, 35% males and 65% females. Angiosarcomas occurred most frequently in the breast (37.8%), skin (21.2%) and superficial soft tissues (13.2%). Grade 1, 2, and 3 tumors represented 20, 38.9 and 41.1% of documented cases respectively. Tumors were metastatic at the diagnosis in 26,9% of the cases, mostly in lung (25.5%) and 67.3% were multifocal. Surgery was used as first treatment in 80%, chemotherapy and radiotherapy were used in 38.5 and 30% of these patients. With a median follow up of 1.8 years, 56% of patients relapsed, mostly locally and 47.2% died. The median survival was 3.4 years (CI95% 2.4 - 5.8). The overall survival is significantly different according whether patients have or not metastasis at the diagnosis (p=0.0001), their 5-year survival is respectively 23% and 48%, and according to their grade (p=0.03). HR calculated for each angiosarcoma’s localisation (ever/never), after adjusting on age, initial metastatic status and grade, showed that bone (HR 8.7 CI95% 1.6 - 48.3), heart (HR 2.9 CI95% 0.9 - 9.7), and liver (HR 10.8 CI95% 2.7 - 43.4) localisations are bad prognostic factors. Conclusions: Angiosarcomas have a specific pattern of occurrence and relapse, a high risk of relapse and a better prognostic than usually admitted mostly in the metastatic situation. Significant adverse prognostic factors are bone and liver localizations, metastases and histological grade 3. No significant financial relationships to disclose.

702 Background: Level of circulating Tumoral Cells (TC) have been recently shown to predict survival in MBC (Cristofanilli, M, NEJM, 2004, 351, 781). We have previously shown that the detection of TC in BM defined moreaccurately the outcome than TC in in the Peripheral Blood (PB) in non metastatic patients (Pierga, JY, Clin Cancer Res, 2004). Material and Methods: We have retrospectively analysed prognostic value of immunocytochemical (ICC) detection of TC in BM of 122 MBC patients. Cytospun Ficoll fractions of BM aspirations were obtained from 11/98 to 12/01 before treatment with chemotherapy. The pancytokeratin monoclonal antibody A45-B/B3 was used. Immunostained cells were detected by a manual analysis and an automated cellular imaging system (ACIS). 3.106 mononuclear cells were screened per patient. Results: Median follow-up was 38 months. Median overall survival was 22 months. Median age was 53 years (34–78). Eighty percent of the patients (97) received first line chemotherapy for MBC. TC were detected in BM of 12 patients (10%) by conventional cytology and in 86 pts (70.5%) by ICC. Median number of TC was 9 (1–3000). Incidence of TC was non significantly higher in lobular (19 pts) compared to ductal carcinomas (84% vs 66%, p=0.1). TC detection in BM was highly associated with clinical or radiological bone metastasis (p=0.0001). Adverse prognostic factors for overall survival were hormonal receptor negativity (p=0.005) and more than one line of chemotherapy (p=0.005). Presence of TC in BM was not associated with a poorer survival rate (median 20 vs 28 months, logrank test, p= 0.39), neither was presence of TC in cluster. Conclusion: We did not find a prognostic value of TC detection by ICC in BM of stage IV breast cancer patients. This could be related to bone metastasis contamination. Circulating TC could be a more efficient tool to evaluate prognostic and monitor tumor response at this stage of the disease. No significant financial relationships to disclose.

Disseminated tumor cells (DTC) in bone marrow are independently related to poor outcome in patients with breast cancer. Phenotypic characterization of DTC may be useful to improve evaluation of the metastasizing potential of DTC and also to more accurately target aggressive tumor cells. DTC were screened in bone marrow aspirates from breast cancer patients by immunocytochemistry with an anticytokeratin (anti‐CK) antibody (A45B/B3). Because the cell permeabilization and fixation required for intracellular CK staining is deleterious for mRNA, we used microaspiration to isolate single tumor cells stained with a monoclonal antibody directed against a membrane epitope, epithelial cell adhesion molecule (EpCAM), in CK‐positive cases. Urokinase‐type plasminogen activator receptor (uPAR) was quantified by real‐time quantitative RT‐PCR. The SKBR3 human breast cancer cell line was used to calibrate RT‐PCR. A linear relationship was observed between the cycle threshold (Ct) of uPAR and 18S gene expression and SKBR3 cells spiked (1, 3, 7, 10 and 20) in control patient bone marrow. EpCAM‐positive cells were aspirated in 21 out of 25 bone marrow specimens from breast cancer patients with CK‐positive cells and uPAR mRNA expression was determined in 16 cases. A high level of uPAR mRNA in DTC was detected in 8 out of 16 patients (50%) and was associated with a more aggressive primary tumor phenotype (estrogen receptor [ER]‐negative, progesterone receptor [PR]‐negative or HER2‐positive) ( p = 0.01). We demonstrated that real‐time quantitative RT‐PCR was reliably adapted to phenotype analysis of isolated micrometastatic cells. A larger study would be useful to confirm the importance of uPAR to define higher risk subgroups of breast cancer patients with micrometastatic disease. © 2004 Wiley‐Liss, Inc.

STI571, or imatinib, selectively inhibits BCR/ABL, PDGFR and c‐kit kinase activity. It has been reported that a large proportion of small cell lung cancer (SCLC) cell lines and tumors express c‐kit and that STI571 inhibits tumor cell growth. We therefore investigated the therapeutic efficacy of STI571, alone or combined with chemotherapy, in human SCLC cells or tumors xenografted into nude mice. The level of c‐kit mRNA expression was variable in SCLC tumors (positive for 2 of 4 xenografts), and c‐kit protein was not detected by immunohistochemistry. On the 4 xenografted tumors, PDGFRα and PDGFRβ were not detected by immunohistochemistry. STI571 induced inhibition of proliferation of the SCLC6 cell line without inducing apoptosis; in contrast, in combination with etoposide or topotecan, the growth inhibition of SCLC6 cells induced by STI571 was increased, with apoptotic DNA fragmentation. Four human SCLC xenografts (SCLC6, SCLC61, SCLC74 and SCLC108) were transplanted into mice. After intraperitoneal injection of STI571, we observed 80%, 40% and 78% growth inhibition of SCLC6, SCLC61 and SCLC108 tumors, respectively, without any significant inhibition of SCLC74 tumor growth. In mice bearing responsive SCLC tumors, we observed an increase of growth inhibition induced by chemotherapy (etoposide + ifosfamide or topotecan) by concomitant and continuous administration of STI571, associated with an increase of toxic deaths. In SCLC6‐bearing mice receiving sequential treatments, we observed a reduction of toxic deaths but a decrease of synergistic antitumor efficacy. In conclusion, the efficacy of STI571 alone in SCLC xenografted tumors was variable and did not depend on c‐kit expression. Moreover, a significant increase of chemotherapy‐induced growth inhibition was obtained by concomitant administration of STI571 that should be carefully investigated in SCLC patients. © 2004 Wiley‐Liss, Inc.

2113 Background: Elderly patients (>70 years) receive chemotherapy less frequently than younger and usually are excluded from clinical trials. The taxanes are among the most active agents available for the therapy of a variety of solid tumors, particularly breast, non-small cell lung cancer (NSCLC) and prostate cancer. Methods: In order to determine the Dose Limiting Toxicity (DLT) and the Maximum Tolerated Dose (MTD) of Docetaxel in elderly patients with advanced disease, to characterize the pertinent pharmacokinetic parameters in this population and to evaluate the response rate, two strata were defined: (A) 70 to 75 years and (B) older than 75 years. Patients received Docetaxel from 50 mg/m2 (50/ 60/ 70/ 75/ 80 mg/m2), administrated as 1 hour infusion every 3 weeks. A geriatric evaluation was performed with the Mini Nutritional Assessment (MNA> or = 12), the Activity Daily Life scale (Katz scale > or = 5) and the mini-Mental Score (MMS) before inclusion. Results: The median number of cycles was 5 in 30 patients. Patients characteristics were: 16 pts >75, 14 pts 70–75, Tumor types : Metastatic Breast Cancer (21), prostate (7), NSCLC (2). EGOG PS 0–1 (27), 2 (3). The main toxicity was neutropenia grade 3–4 (12/150 cycles); fatigue, fluid retention, nausea-vomiting were common but rarely severe (grade 1–2). DLT: 1 febrile neutropenia in stratum A at dose level 70 mg/m2, 2 in stratum B at dose level 60 mg/m2 (1 febrile neutropenia and 1 sepsis). Conclusion: MTD was not reached, Docetaxel is well tolerated in elderly patients. Accrual at dose level 80 mg/m2 is ongoing in both strata. PK and response rate results will be presented. No significant financial relationships to disclose.

PURPOSE:The presence of tumor cells in bone marrow has been reported to represent an important prognostic indicator in breast cancer, but the clinical significance of circulating cells in peripheral blood is less well known. The aim of this study was to evaluate the feasibility of identifying cytokeratin (CK)-expressing cells in peripheral blood with an automat-assisted immunohistochemical detection system and to compare it with detection of tumor cells in bone marrow samples.EXPERIMENTAL DESIGN:Cytospun Ficoll fractions of peripheral blood and bone marrow were obtained simultaneously in 114 breast cancer patients at different stages of the disease (I to IV) before treatment with chemotherapy. The pancytokeratin (CK) monoclonal antibody A45-B/B3 (anti-CKs 8, 18, and 19) was used for epithelial cell detection. Immunostained cells were detected by an automated cellular imaging system (ChromaVision Medical System).RESULTS:CK+ cells were detected in 28 (24.5%) patients in blood and in 67 (59%) patients in bone marrow. Twenty-six (93%) patients with CK-positive cells in blood also had positive bone marrow (P < 0.001). Positive cells were detected in peripheral blood in 3/39 (7.5%) operable breast cancers (stage I/II), 9 of 36 (25%) locally advanced breast cancers (stage III), and 16 of 39 (41%) patients with metastatic disease (stage IV; P = 0.017). In the subgroup of nonmetastatic patients (n = 75), prognostic factors for poor disease-free survival were: absence of estrogen receptor; presence of CK+ cells in bone marrow (P = 0.012); clinical nodal involvement; large tumor size (T4); and presence of tumor emboli. Presence of circulating CK+ cells in the peripheral blood was not statistically correlated with disease-free survival. On multivariate analysis, independent indicators for disease-free survival were: absence of estrogen receptor (P = 0.043) and presence of CK+ cells in bone marrow (P = 0.076).CONCLUSIONS:The clinical relevance of circulating epithelial cells as a prognostic factor is not supported by the present data, especially in comparison with tumor cells in the bone marrow. However, this method of detection may be useful to monitor the efficacy of treatment in advanced or metastatic breast cancer.

L'activation de HER2 est observée dans 15 à 30 % des carcinomes mammaires. La surexpression de HER2 au niveau des micrométastases de la moelle osseuse a été rapporté chez 60 % à 100 % des patientes. Or, dans notre expérience, la surexpression de HER2 est identique entre tumeurs mammaires primaires et métastases viscérales. Le but de ce travail a été d'évaluer le statut de HER2 de micrométastases de la moelle osseuse de patientes atteintes de carcinomes mammaires et de le comparer à celui de leurs tumeurs mammaires primaires. La détection de micrométastases de la moelle osseuse est réalisée dans le cadre d'une étude prospective, par immunocytochimie sur produit d'aspiration de moelle osseuse, à l'aide d'un anticorps anti cytokératines 8,18,19 (A45-B/B3). La détection est faite par screening manuel et par un système d'analyse d'image assisté par ordinateur (ACIS Chromavision). Les carcinomes mammaires primaires et les micrométastases de la moelle osseuse de 27 patientes (3 stades II, 1 stade III et 23 stades IV) ont été analysés. Le statut de HER2 des carcinomes mammaires primaires a été déterminé par immunohistochimie (Anticorps anti CB11, Novocastra) et confirmé par FISH pour les cas 2+, celui des micrométastases de la moelle osseuse par FISH. Une surexpression de HER2 était observée pour 5/27 (18,5 %) des carcinomes mammaires primaires et une amplification de HER2 dans 4/27 (15 %) des micrométastases de la moelle osseuse. La concordance du statut de HER2 entre les micrométastases et les carcinomes mammaires primaires était de 90 %. Dans 2 cas, HER2 était surexprimé dans la tumeur primaire mais pas dans les micrométastases et dans un cas, HER2 était amplifié dans les micrométastases (6 copies) et pas dans la tumeur primaire. Dans 4 cas, sans amplification de HER2, une trisomie du chromosome 17 était observée dans les micrométastases de la moelle osseuse. Le statut de HER2 des micrométastases de la moelle osseuse est identique à celui de la tumeur mammaire primaire. L'amplification de HER2 est une altération génétique qui précède l'extension micrométastatique.

9053 Background: Expression of the MAGE-A family genes has been found in many types of tumors but not in normal adult tissues with the exception of testicular germ cells. A frequency of 20% MAGE mRNA positive sarcomas has been reported previously (Van den Eynde, Curr Opin Immunol 1997). NA17 is another tumor-associated antigen encoded by an intronal sequence from the N-acetylglucosaminyltransferase-V gene, which is significantly expressed in melanomas. Because NA17 and MAGE gene products are presented by HLA class I molecules and recognized by CD8+ cytolytic T lymphocytes in vivo, they have become promising target antigens for tumor vaccination in melanoma patients. Both are evaluated in ongoing melanoma vaccination trials at the Institut Curie. Methods: In 20 surgically resected bone and soft tissue sarcomas, expression of NA17 and Mage 1, 2, 3, 4, 6 and 10 was measured by semi-quantitative RT-PCR. After RNA extraction with CsCl gradient and reverse transcription, expression was quantified by PCR with comparison to dilutions of reference cell lines. All samples were histologically confirmed to contain tumoral tissue, corresponding to bone sarcomas in 8 and soft tissue sarcomas in 12 cases. Results: Analysis by the multimarker RT-PCR revealed an expression of at least one MAGE gene in 6 of 20 tumors, corresponding to a positivity rate of 30%. Among these 6 cases, MAGE 3 was expressed in 6, MAGE 6 in 4, MAGE 1 and MAGE 2 in 3, MAGE 4 and MAGE 10 in 2. For NA17, 5 of 20 samples exhibited positivity (25%). Overall, 8 of 20 tumors (40%) expressed at least one gene, and 4/20 (20%) expressed 3 or more genes simultaneously. Conclusions: This short series shows that NA17 and MAGE genes are expressed in 4/8 bone sarcomas and 4/12 soft tissue sarcomas. Sarcoma patients showing disease progression after conventional treatment might represent eligible candidates for HLA typing and future sarcoma vaccination trial combining NA17 and MAGE antigen immunization. No significant financial relationships to disclose.

Anthracyclines remain an important group of chemotherapeutic agents, despite their inherent cardiotoxicity. This cardiotoxicity may be even more of a concern in the future, as combination therapies of anthracyclines with newer agents become routine. Such combinations may be highly effective, but cardiotoxicity may also be increased. Dexrazoxane reduces the incidence of cardiotoxicity, as demonstrated in numerous clinical trials in both adults and children. Evidence from the literature suggests no effect of dexrazoxane on the antitumour efficacy of anthracyclines, and there is no adverse effect on survival. Dexrazoxane is therefore a valuable tool for oncologists using anthracycline-based regimens.

Despite high-dose chemotherapy with autologous stem cell transplantation, most patients with multiple myeloma relapse and eventually die. Previous studies have suggested that multiple myeloma cells express estrogen receptors and that antiestrogens induce in vitro growth inhibition and apoptosis of tumor cells (1–3). We therefore conducted a phase II feasibility trial of highdose tamoxifen in patients with relapsed or refractory multiple myeloma who were previously treated by two or three chemotherapy regimens including thalidomide. As in a previous phase I trial (4), tamoxifen was administered twice daily at doses of 400 mg/m per day on day 1 and 300 mg/m per day on days 2–6. Thereafter, doses were adjusted according to pharmacokinetic measurements to target serum tamoxifen concentrations between 4 M and 5 M for a total treatment duration of 39 days. Concentrations of serum tamoxifen and its main metabolites were measured using a high-pressure liquid chromatography assay (5) on days 2 and 3, 9, 16, 23, and 30 to adjust the doses administered on days 6–13, 13–20, 20–27, 27–34, and 34–39, respectively (ranges overlap because treatment doses were adjusted after morning tamoxifen administration). We based the tamoxifen dose adjustment on the estimated clearance from the analysis of concentration–time data using a nonlinear regression model (MicroPharm, Newcastle Emlyn, Camarthenshire, U.K., and INSERM, Paris, France). Estrogen receptor , estrogen receptor , and progesterone receptor transcripts were quantified by quantitative real-time reverse transcription–polymerase chain reaction (TaqMan; Applied Biosystems, Foster City, CA) (6) on CD38-positive bone marrow cells. After inclusion of the sixth patient, the overall lack of response and neurologic toxicity led us to prematurely stop patient inclusion. Tamoxifen was administered for 39 days to two patients but was discontinued at day 10, 22, 29, or 30 for the other four patients because one patient experienced vertebral complication and three patients experienced multiple myeloma progression. Neurotoxicity was observed in three patients, with dizziness, vertigo, light-headedness, instability on tandem walk, dysmetria, and tremor. Evaluation of response showed disease progression after treatment in all patients. Adjusted tamoxifen doses ranged from 212 to 436 mg/m per day (median 330 mg/m per day). After day 9, the median concentration was 4.0 M (range 2.0– 6.0 M), and four patients reached the target serum tamoxifen concentrations. There was no relationship between tamoxifen and/or tamoxifen metabolite concentrations (mean or maximum) and the severity of side effects (Table 1). Estrogen receptor , estrogen receptor , and progesterone receptor expression was analyzed in a series of 12 patients, including the above six patients and six consecutive new patients to enlarge our multiple myeloma population. Before any treatment, only four of 12 tumor samples expressed estrogen receptor : one tumor sample expressed substantial levels and the expression levels in those of the remaining three patients (including that from one tamoxifen protocol patient) were close to the positive cutoff value. Tumors from three patients (including one tamoxifen protocol patient) expressed high levels of estrogen receptor transcripts. None expressed progesterone receptor transcripts. No correlation was observed between estrogen receptor and estrogen receptor expression (r .46). Analyses of tumors from two patients performed before and after 4 months of tamoxifen treatment indicated that expression of estrogen receptor was not modified by treatment. In conclusion, despite high serum tamoxifen concentrations, our feasibility trial showed the absence of substantial estrogen receptor expression in all plasmacytosis tumor cells except one tumor sample and the absence of activity against multiple myeloma by tamoxifen. Thus, our data do not support the use of high-dose tamoxifen given orally as a treatment for patients with refractory or relapsed multiple myeloma.