Search

Purpose Evidence supporting the clinical utility of predictive biomarkers of anthracycline activity is weak, with a recent meta-analysis failing to provide strong evidence for either HER2 or TOP2A. Having previously shown that duplication of chromosome 17 pericentromeric alpha satellite as measured with a centromere enumeration probe (CEP17) predicted sensitivity to anthracyclines, we report here an individual patient–level pooled analysis of data from five trials comparing anthracycline-based chemotherapy with CMF (cyclophosphamide, methotrexate, and fluorouracil) as adjuvant chemotherapy for early breast cancer. Patients and Methods Fluorescent in situ hybridization for CEP17, HER2, and TOP2A was performed in three laboratories on samples from 3,846 of 4,864 eligible patients from five trials evaluating anthracycline-containing chemotherapy versus CMF. Methodologic differences did not affect HER2-to-CEP17 ratios but necessitated different definitions for CEP17 duplication: > 1.86 observed copies per cell for BR9601, NEAT, Belgian, and DBCG89D trials and > 2.25 for the MA.5 trial. Results Fluorescent in situ hybridization data were available in 89.3% (HER2), 83.9% (CEP17), and 80.6% (TOP2A) of 3,846 patient cases with available tissue. Both CEP17and TOP2A treatment-by-marker interactions remained significant in adjusted analyses for recurrence-free and overall survival, whereas HER2 did not. A combined CEP17 and TOP2A–adjusted model predicted anthracycline benefit across all five trials for both recurrence-free (hazard ratio, 0.64; 95% CI, 0.51 to 0.82; P = .001) and overall survival (hazard ratio, 0.66; 95% CI, 0.51 to 0.85; P = .005). Conclusion This prospectively planned individual-patient pooled analysis of patient cases from five adjuvant trials confirms that patients whose tumors harbor either CEP17 duplication or TOP2A aberrations, but not HER2 amplification, benefit from adjuvant anthracycline chemotherapy.

Purpose Inhibiting angiogenesis is one of the most promising avenues for new therapies for ovarian cancer. We investigated the efficacy and safety of a novel agent, BIBF 1120, a triple angiokinase inhibitor, after chemotherapy for relapsed disease. Patients and Methods We conducted a randomized, double-blind, controlled phase II trial in 83 patients who had just completed chemotherapy for relapsed ovarian cancer, with evidence of response, but at high risk of further early recurrence. The patients were randomly assigned to receive maintenance therapy using BIBF 1120 250 mg or placebo, twice per day, continuously for 36 weeks. End points were progression-free survival (PFS), toxicity, and overall survival. Results Thirty-six–week PFS rates were 16.3% and 5.0% in the BIBF 1120 and placebo groups, respectively (hazard ratio, 0.65; 95% CI, 0.42 to 1.02; P = .06). Four patients continued on BIBF 1120, including two patients for another year or more. The proportion of patients with any grade 3 or 4 adverse events was similar between the groups (34.9% for BIBF 1120 v 27.5% for placebo; P = .49; mostly grade 3). However, more patients on BIBF 1120 experienced diarrhea, nausea, or vomiting (mainly grade 1 or 2 and no grade 4). There was a higher rate of grade 3 or 4 hepatotoxicity in patients on BIBF 1120 (51.2%) compared with patients on placebo (7.5%; P < .001), but this was rarely of clinical significance, and patients continued with the trial treatment. A single-level dose reduction to 150 mg was made in 15 patients, all on active drug. Conclusion BIBF 1120 is well tolerated and associated with a potential improvement in PFS. The observed treatment effect is sufficient to justify further study within a large phase III trial.

2565Background: Estimation of glomerular filtration rate (GFR) is essential for carboplatin chemotherapy dosing. However, the most accurate method to estimate GFR in patients with cancer is unknown...

e22101Background: PERSEPHONE is a randomised controlled non-inferiority trial comparing 6 months of trastuzumab to the standard 12 months in patients with HER2 positive early breast cancer. An impo...

506Background: Adjuvant trastuzumab has significantly improved outcomes for HER2+ EBC pts, using the 12m duration empirically adopted from the pivotal registration trials. A shorter duration could reduce toxicities and cost whilst providing similar efficacy. No reduced-duration trial to date has demonstrated non-inferiority. Methods: PERSEPHONE is a randomised phase 3 non-inferiority trial comparing 6 to 12m trastuzumab, the largest reduced-duration non-inferiority trial internationally. Mapping onto standard UK practice, all HER2+ EBC pts were eligible. Stratification is by ER status, chemotherapy (CT) type, and CT and trastuzumab timing. The primary endpoint is DFS from diagnosis (first relapse or death from any cause). Randomising 4000 pts (1:1) enabled the trial to assess the non-inferiority of 6m (5% 1-sided significance, 85% power), defining non-inferiority as ‘no worse than 3%’ below the 12m arm’s assumed 80% 4-yr DFS. The pre-planned definitive DFS analysis required 500 events. Results: 4089 pts w...

TPS591 Background: No specific targeted therapies are available for Triple Negative Breast Cancers (TNBC), an aggressive and diverse subgroup. The basal TNBC sub-group show some phenotypic and molecular similarities with germline BRCA (gBRCA). In gBRCA patients, and potentially other homologous recombination deficiencies, these already compromised pathways may allow drugs called PARP inhibitors (olaparib) to work more effectively. Aims: To establish if the addition of olaparib to neoadjuvant platinum based chemotherapy for basal TNBC and/or gBRCA breast cancer is safe and improves efficacy (pathological complete response (pCR)). Trial design: 3-stage open label randomised phase II/III trial of neoadjuvant paclitaxel and carboplatin +/- olaparib, followed by clinicians' choice of anthracycline regimen. Stage 1 and 2: Patients are randomised (1:1:1) to either control (3 weekly carboplatin AUC5/weekly paclitaxel 80mg/m2 for 4 cycles) or one of two research arms with the same chemotherapy regimen but with two different schedules of olaparib 150mg BD for 12 days. Stage 3: Patients are randomised (1:1) to either control arm or to the research arm selected in stage 2. Methods: Stage 1 - Safety: both research arms combined. Stage 2 - Schedule selection criteria: pCR rate and completion rate of olaparib protocol treatment. It is a “pick-the-winner” design with 53 patients in each research arm. This allows a 90% power, 5% one-sided significance level to test null hypothesis of pCR ≤35% versus an alternative hypothesis of pCR ≥55% in each of the research arms. Stage 3 - Efficacy: anticipated pCR ~55-60% for all trial patients and ~60-65% for gBRCA patients. The trial is powered to detect an absolute improvement of 15% (all patients) and 20% (gBRCA patients) by adding olaparib to chemotherapy (enriched design). TNBC patient recruitment will be capped, to ensure required gBRCA patients are enrolled. Enrichment design is applied with overall significance level 0.05(α) = 0.025(αall)+ 0.025(αgBRCA) and 80% power. Target accrual: 527 [gBRCA 220] Current accrual: 17 Sites activated: 12 [expected number of sites 30-50].

1014Background: ARTemis compared the addition of Bev to neoadjuvant D-FEC in HER2 negative breast cancer. Improved pathological complete response (pCR) with Bev has been reported previously. Method...

e23040Background: Novel biomarkers are required to accurately assess tumour burden and response in cancer. In patients with relapsed high-grade serous ovarian carcinoma (HGSOC) we tested (i)if TP53...

A weekly, intensive chemotherapy regimen has been used to treat 70 patients with small-cell lung cancer (SCLC). Forty-five patients had limited disease (LD) and 25 extensive disease (ED) with good prognostic features. The regimen consisted of cisplatin 50 mg/m2 intravenously (IV) day 1 and etoposide 75 mg/m2 IV days 1 and 2, alternating weekly with ifosfamide 2 g/m2 IV day 8 and doxorubicin 25 mg/m2 IV day 8, for a total of 12 weeks. Dose modifications were made according to defined hematologic criteria. Responding patients with limited disease subsequently received mediastinal radiotherapy. Overall response to chemotherapy was 91% with a complete response (CR) rate of 50%. Forty-five patients with limited disease (LD) achieved an overall response rate of 91% with a CR rate of 51%, and 25 patients with extensive disease (ED) achieved an overall response rate of 92% with a CR rate of 48%. Median survival for the whole group was 54 weeks (LD, 58 weeks; ED, 42 weeks). Hematologic toxicity was predictable, without the wide fluctuations in WBC count seen in conventional 3-weekly regimens. In all, one quarter of treatment courses were delayed, most frequently because of leukopenia. Dose reductions were required in 63% of cases. The average delivered dose intensity was calculated and shown to be 73% of projected. Nonhematologic toxicity was mild with nausea and vomiting being the most common. This weekly schedule of chemotherapy has proved to be active and well tolerated and is currently being compared with conventional 3-weekly chemotherapy in a randomized study.

1014 Background: Bevacizumab (bev) has been used with NACT in breast cancer trials. Geparquinto reported benefit for bev in triple negative (neg) patients (pts) (pCR 36.4% v 27.8% p=0.02), as did C...

TPS656 Background: PERSEPHONE is a randomised controlled trial comparing six months of trastuzumab to the standard 12 months in patients with HER2 positive early breast cancer. Methods: 4,000 patie...

552 Background: PERSEPHONE is a randomised trial comparing 6month (6m) to 12month (12m) of trastuzumab in patients (pts) with HER2-positive early breast cancer. The trial has recruited 3,166 of 4,0...

e21503 Background: Historically, GISTs were often misdiagnosed as other tumours. However, with the increasing awareness of GISTs, one needs to be careful in confirming the diagnosis since other tumours can simulate GISTs on endoscopy/imaging/histology. We report our 8 year experience of GIST mimics from the CGSG database. Methods: 250 cases discussed in the specialist multidisciplinary meeting from 2004 – 2012 were reviewed. Tumours initially suspected to be GISTs but turned out to be other tumours were included in this analysis. Patient demographics, endoscopic, imaging and histological features were analysed in 41 GIST mimics. Central review by specialist histopathologist was undertaken with immunohistochemical and molecular analysis to differentiate these tumours from GISTs. Results: GIST mimics N= 41. Median age 56 years (range 11 – 82 years), male:female ratio 1:1. Median size 8 cm (range 1.9 – 30 cm). Site of origin of tumours: stomach 34%, oesophagus/junction 15%, mesentery 12%, small bowel 10%, colon 5%, retroperitoneum 5%, pancreas 5%, liver 2% and indeterminate site of origin 12%. Suspicion of GIST was raised from imaging (46%), endoscopy (41%) and histology (39%). The histological subtypes of GIST mimics are summarised in the Table. Conclusions: Many benign and malignant tumours do mimic GISTs on imaging/endoscopy or histology. The diagnosis of GISTs should be made by a specialist multidisciplinary team with experience in imaging and histopathology. Distinguishing GISTs from their mimics is of critical importance both for prognosis and management. [Table: see text]

1015 Background: Neo-tAnGo was an NCRI UK neoadjuvant breast cancer study testing the addition of gemcitabine to anthracycline and taxane-based treatment and also the sequencing of chemotherapy. In a translational substudy, sequential fresh tissue was analysed for PAM 50 subgroups. Methods: Neo-tAnGo recruited 831 patients. 162 patients were consented for Neo-tAnGo Science, for 3 additional fresh tissue samples to be taken at diagnosis (diag), mid-chemotherapy (CT) and end-CT. Standard methodology was used for PAM 50 subtyping. Results: Fresh tissue samples at diag were received from 123 patients (pts). 45 pts (37%) had 2 additional samples provided (at mid- and end-CT). 57 pts (46%) had 1 additional sample provided (34 at mid-CT, 23 at end-CT). PAM 50 subtyping at diag was as follows: 31 (25%) BASAL; 30 (24%) HER2; 39 (32%) LUM B; 13 (11%) LUM A; 10 (8%) NORMAL-like. Pathological complete response rates (pCR: no disease in breast or axillary nodes, n = 121 pts) differed between PAM 50 subtype at diag (p = 0.04): BASAL 11/31 (35%); HER2 8/30 (27%); LUM B 3/37 (8%); LUM A 1/13 (8%); NORMAL-like 3/10 (30%). Of the 94 pts who were not PAM50 NORMAL-like at diag and had 2+ samples, 42 (45%) ‘shifted to NORMAL-like’. This was associated with slightly higher pCR rates (24% vs 15% who didn’t ‘shift to NORMAL-like’, p = 0.44) and borderline significantly higher rates of pCR or minimal residual disease (MRD

5514 Background: Weekly paclitaxel (wPxl) has activity in platinum-resistant ovarian cancer (PROC). Upregulated Src kinase activity is seen in Pxl-resistant ovarian cancer models. This trial investigated the combination of wPxl and the oral Src inhibitor saracatinib (AZD0530) in PROC. Methods: Patients with PROC (defined as relapse within 6 months of prior platinum chemotherapy, confirmed either by CT scan or symptomatic CA125 rise) were randomised 2:1 to receive four 8 week cycles of wPxl (80mg/m2/week x6 with 2 week break) plus saracatinib (S; 175mg od) or placebo (P) continuously, starting 1 week prior to wPxl, until disease progression. Patients were stratified as 2 lines of prior chemotherapy; 78% (wPxL+S) vs 72% (wPxL+P) of patients received ≥1 cycle of wPxl; relative dose intensity was 96% vs 98% for wPxL+S and wPxL+P respectively. The 6-month PFS rate was 29% (wPxL+S) vs 35% (wPxL+P). Median PFS was 3.9 vs 5.3 months (HR 1.04; 95% CI 0.68, 1.59; p=0.86); median OS was 12.7 vs 12.8 months (HR 1.50, 95% CI 0.63, 3.56; p=0.36); RR were 0.0% vs 2.9% (CR) and 29% vs 38.9% (PR) for wPxL+S vs wPxL+P respectively. Median DoR was 5.6 vs 3.6 months; TTP was 3.9 vs 5.5 months (HR 1.10; 95% CI 0.71, 1.72;p=0.67). Grade 3+ Serious Adverse Events were 36.2% vs 30.6%; the most frequent toxicities (any grade) were abdominal pain (4.3%) and febrile neutropenia (4.3%) for wPxL+S, and vomiting (5.6%) for wPxL+P. Conclusions: In this randomised phase II trial, the addition of saracatinib to wPxl did not improve 6-month PFS in patients with PROC. Clinical trial information: NCT01196741.

10541 Background: Cambridge GIST study group was formed in 2003. GIST specialist multidisciplinary team (MDT) and GIST clinic were established with central review of histology/pathology and management plans. We present our 10 year experience of a prospective database which formed the template for the United Kingdom national GIST registry. Methods: All patients (pts) who have been referred to the GIST MDT were included in the study. The following data was prospectively collected: patient demographics, presenting symptoms, site, size, histology including mutational data (where available) risk stratification of the GISTS, surgical interventions, systemic therapies, other tumours occurring in GIST pts. Results: 260 patients (pts) were reviewed in the GIST MDT. 29 pts had endoscopic/imaging diagnosis of a GIST and were not included in the final analysis. 41 pts had other tumours diagnosed when GIST was initially suspected. Histologically confirmed GISTS N=190. Male: Female 52%:48%. Median age 64 years (range 14-94), 4%

TPS665 Background: “Multi-parameter” prognostic tests such as Oncotype DX are increasingly used to identify women with ER +ve HER2 -ve breast cancer treated with endocrine therapy who are unlikely to benefit meaningfully from adjuvant chemotherapy. The supporting evidence for predictive testing is retrospective and is strongest for women with negative nodes. Randomised trials to validate testing are in progress but more evidence is needed, especially for node positive disease. There is early evidence that other multi-parameter tests which are in development have a predictive utility. Methods: OPTIMA will assess the value of multi-parameter tests in a UK population. OPTIMA prelim, the feasibility phase, will recruit 300 patients from May 2012 (with a 200 patient bridging extension to the main study). Eligible patients will have ER +ve HER2 -ve tumours with involved nodes (pN1-2). Patients will be randomised to the standard arm of both chemotherapy and endocrine therapy, or to the “test-directed treatment” arm in which patients will be assigned either the same chemotherapy and endocrine therapy or endocrine therapy only according to the result of an Oncotype DX test. OPTIMA prelim has 3 objectives. (1) To establish whether randomisation to test-directed treatment is acceptable to patients and clinicians. This will be done through qualitative research with in-depth interviews with OPTIMA researchers and by recording and analysing consultations when the trial is discussed with potential participants. (2) Alternative multi-parameter tests will be evaluated on all patients’ tumours and their performance will be compared to Oncotype DX using statistical and cost-effectiveness analysis to select a candidate test(s) appropriate for NHS use to be evaluated in the main trial. (3) The success of the main trial depends on efficient tumour sample collection and analysis to avoid treatment delays. The obstacles to this will be analysed in detail using a sample tracking database. The main study is an efficacy trial of 2-3 arms with the same design but uses tests selected in OPTIMA prelim. It will compare 5-year relapse free survival of test-directed vs. conventional treatment with a non-inferiority hypothesis.

TPS1144 Background: Bevacizumab is a new humanised monoclonal antibody which targets vascular endothelial growth factor (VEGF) and thereby the neoangiogenic process in cancer. Bevacizumab has shown promising anti-tumour effect when given concurrently with taxane-based chemotherapy in breast cancer. However two neoadjuvant breast cancer trials have produced conflicting results. The NSABP-B40 study showed significant benefit for bevacizumab only in the ER+ve patients (pathological complete response (pCR) in ER+ve population 15.2% vs 23.3%, p=0.008). Whereas the GEPARQUINTO trial reported significant benefit only in the triple negative patients (pCR 27.8% vs 36.4% p=0.021). Methods: ARTemis is a phase III randomised trial to determine whether the addition to neo-adjuvant chemotherapy of bevacizumab is more effective than standard chemotherapy alone in patients with HER2-negative early breast cancer. A total of 400 patients will be randomised into each of the two treatment arms which will allow an absolute difference in the pCR rates in excess of 10% to be detected at the 5% (2-sided) level of significance with an 85% power. Primary outcome is pathological complete response rates after neo-adjuvant chemotherapy, i.e. no residual invasive carcinoma in the breast, and no evidence of metastatic disease within the lymph nodes. Secondary outcomes are disease free survival, overall survival, complete pathological response rates in breast alone, radiological response after 3 and 6 cycles of chemotherapy and rate of breast conservation and toxicities. Results: Recruitment into ARTemis began in April 2009 and as of January 2012 had recruited 469 (59%). ARTemis is due to complete recruitment by Dec 2012 and the first planned interim analysis of the primary outcome will be Dec 2013. Conclusion: The IDSMC reviewed the trial in June 2011 and recommended continuation of recruitment to this important trial given there were no safety concerns, and results from the other neoadjuvant studies (NSABP-B40 and GEPARQUINTO) are conflicting.

574 Background: FFPE is a valuable and widely available resource for translational research which to date has been under-used due to technical limitations. Improvement in technology has enabled genome-wide analysis of FFPE samples. We have assessed gene expression and copy number changes in the same cohort of breast cancers to identify markers or pathways important in prediction of treatment response. Methods: FFPE tissues from patients treated with neoadjuvant adriamycin/cyclophosphamide followed by taxanes in a clinical study were used. Gene expression profiling was assessed using the cDNA mediated annealing selection and ligation assay using the cancer panel which assess 502 genes (DASL assay, Illumina). Data was analysed using BeadStudio software. Copy number changes were assessed using the Molecular inversion probe assay with the 50K SNP panel (Affymetrix, California) and analysed using Nexus software (Biodiscovery). Results: Gene expression profiling was carried out on 44 samples. 12/44 (27%) patients had a pathological complete response (pCR) following chemotherapy. Significant differential expression of genes between pCR and non-pCR cancers were shown. TNFRSF5, CTSD, BCL3, ARNT, BIRC3, TGFBR1, MLLT6, and EVI2A were over-expressed and COL18A1, FGF12, IGFBP1 and NOTCH4 which were down-regulated in cancers that have a pCR (p ≤ 0.01). Copy number changes were assessed in 33 samples and comparison of copy number changes in pCR vs. non-pCR showed gains in regions 6q22, 21q21, 4p14, 4q21, 4p14, and loss at 11q11 (p ≤ 0.01). Three regions containing microRNA coding sequences, mir130a (11q11) mir142 (17q23) and mir21 (17q23) showed significant loss among pCR tumours (p < 0.05). Conclusions: This feasibility study shows that FFPE can be used for gene expression and copy number analysis which is a useful tool for the discovery of predictive markers for treatment response in neoadjuvant treatment trials. The role of TNFRSF5, microRNA 21/130a/142, and 11q11 loss should be further investigated as predictive markers of response to chemotherapy. [Table: see text]

522 Background: Neo-tAnGo used a 2-by-2 factorial design, addressing: (i) gemcitabine (G) in a sequential neoadjuvant chemotherapy (CT) regimen of epirubicin/cyclophosphamide (EC) and paclitaxel (T); and (ii) the sequencing of these treatment components (EC then T ± G versus T ± G then EC). Methods: Patients (Pts) with early breast cancer (T2 tumours or above) were randomised to EC then T, T then EC, EC then TG or TG then EC. All components were given x 4 cycles. (E= 90 mg/m2 day (d)1 every (q) 21d; C = 600 mg/m2 d1 q21d; T = 175 mg/m2 d1 q14d; G = 2,000 mg/m2 d1 q14d.) The primary endpoint was pCR, defined as absence of invasive disease in the breast and axillary lymph nodes. 800 pts were required to detect 10% differences in the primary endpoint pCR rates, at the 5% (2-sided) significance level with 85% power. Stratification was by age, inflammatory/locally advanced disease, tumour size, clinical involvement of axillary nodes and oestrogen receptor (ER) status. Results: Between January 2005 and September 2007, 831 pts were randomised by 88 consultants from 57 UK centres. Characteristics were balanced across groups: 63% [Table: see text]

Background: In ER+ early breast cancer, 5 years of tamoxifen substantially reduces the annual recurrence rate throughout the first decade (years 0-9). Despite this, appreciable risks of recurrence remain, with similar annual recurrence risks for ER+ disease during years 0-4, 5-9 and 10-14 following surgery. It is not reliably known how 10 years duration of adjuvant tamoxifen compares with the current standard of 5 years. Methods: Between 1991 and 2005, 6,934 women with ER+ (39%), or ER untested (61%) invasive breast cancer (53% node negative, 31% node positive, 16% node status unspecified) from 176 UK centers, who had completed 4+ years of adjuvant tamoxifen, were randomized between continuing for another 5 years and stopping. Information on compliance, hospital admission(s), breast cancer recurrence (including new contralateral disease), other new primary cancer, death and cause of death was sought annually. Results: 78% of those allocated to continue and 3% of those allocated to stop were taking adjuvant tamoxifen at 3 years following randomization; fewer than 1% switched to any other adjuvant hormone treatment. With a median follow-up of 4.2 years, there were fewer recurrences (415 vs 442; RR=0.94, 95% CI 0.81-1.09; p=0.4) among those allocated to 10yrs than 5yrs tamoxifen. Although breast cancer mortality was lower among those allocated 10yrs, with just 374 deaths from breast cancer recorded, data remain immature. Despite a doubling in the risk of endometrial cancer (76 vs 35) with 10yrs tamoxifen, there was no increase in deaths from endometrial cancer (10 vs 12) or from any other non-breast cancer cause. Conclusions: Although no significant reduction in recurrence has yet been seen in aTTom, the results are consistent with preliminary findings from the ATLAS trial, which reported a DFS but not overall survival advantage to longer tamoxifen (Peto et al SABCS 2007). Combining results from these two large studies indicate that continuation of tamoxifen beyond the first 5 years reduces recurrence over the next few years, but further follow-up is needed to assess reliably the longer-term effects on recurrence and the net effects, if any, on mortality.

10019 Background: Imatinib Mesylate (IM) induces an overall response rate of 84% in patients (pts) with metastatic GISTs. Standard RA is by RECIST criteria. This may underestimate the true benefit of IM in such pts. Choi et al observed ↓ in size (10%) and or density (D) (15%) correlated well with disease specific survival. We report our experience using both RECIST and Choi criteria for RA in GIST pts on IM. Methods: Between 2003 and 06, 24 pts with metastatic or inoperable GISTs were treated with IM. 50 lesions (Liver 20, Omentum 21, Primary 9) were identified at presentation on contrast CT (Siemens multi slice CT, IV Niopam, GE PACS system). 3 monthly CT was used for RA until tumour progression (PD) or death. Parameters recorded were: 1.Largest dimension 2. Lesion density (Mean Hounsfield units, HFU) 3.New lesions 4.Any new features. RECIST and Choi criteria were both applied for RA. Results: N=50. At 3 months (m) only 16% of the lesions achieved partial response (PR) by RECIST and 24% increased in size by a mean of 22.5% (Range 3–150%). RA as measured by Choi criteria revealed 88% of lesions had either a 10% ↓ in size or a 15% ↓ in D at 3 m. Mean ↓ in D was 29% (95% CI 24.6–33.4). At 18 m 50% of the lesions achieved PR and 20% had PD by RECIST criteria, and with Choi criteria 68% RR was recorded. Calcification, haemorrhage and perforation within the lesions spuriously altered the mean D value in 15% of the lesions which made assessment by Choi criteria difficult. In 5 pts new hepatic cystic lesions (range 2–22) appeared with a mean D of 16 HF. None were detected in extra-hepatic sites. 4 pts showed new calcifications within lesions 3–18 m on IM. Enhancing nodules within lesions were seen in 4 pts predating clinical PD by 6–18 m. Conclusions: Choi criteria were found to be reproducible in our pts. RRs at 3 months using Choi criteria (88%) were far greater than by RECIST criteria (16%). RRs were more similar at 18 m (68% Choi vs 50% RECIST).Using density for RA can become unreliable in the presence of perforation, calcification and haemorrhage. These criteria should be evaluated by a radiologist experienced in assessing the response to IM in GISTs with awareness of their limitations. No significant financial relationships to disclose.

15018 Methods: CTCR-OV01 is a randomised phase II neoadjuvant study in stage III-IV epithelial ovarian cancer (EOC). Patients (pts) were consented for diagnostic and fresh tissue research biopsies, and after diagnosis was confirmed, were randomised (1:1) to: Arm A: Carboplatin AUC 7 every 3 weeks (C) × 3 cycles (cy); then interval debulking surgery (IDS); then C AUC 6 with paclitaxel 175 mg/m2 q 3w (CT) × 3 cy; and finally paclitaxel 175 mg/m2 q 3w (T) × 3 cy. Arm B: T × 3 cy; IDS; CT × 3 cy; C × 3 cy. At IDS further fresh tissue samples were collected. Clinical Endpoints: (i) Partial response rate after three cycles of chemotherapy based on CA125 (CA125 RR) and CT scan (CT RR). (ii) Median progression-free survival (PFS). Translational endpoints: Candidate genes and molecular profiles as predictive markers of response/resistance to C and T. Results: Jan ‘02-Dec ‘04, 48 pts were randomised in a single centre study. 4 patients were excluded from analysis (3pts severe toxicity; 1pt incorrect histopathology), leaving 44 patients (21pts A/23 pts B). Median age 60 yr (range 36–75 yrs); 42/44 serous papillary histology; stage III/IV = 77%/23%; grade I/II/III = 2%/34%/64%. All prognostic factors were well balanced between treatment arms. Median follow-up is 22 months (IQR 17–33.5). CA125 RR to pre-operative chemotherapy was A/B = 65%/60% (p = 1). CT RR was A/B = 47%/35% (p = 0.15). CA125 RR, was significantly associated with improvement in progression-free survival (PFS) compared to non-responders (17.3 v 12.4 months; log rank test p = 0.027). Optimal debulking surgery was possible in A/B = 53%/63%. Median PFS was 14 months (m) (A/B = 13 m/14 m (p = 0.53)). Supervised analysis of Affymetrix expression data showed significant enrichment for differential extracellular gene expression in paclitaxel resistant patients. Conclusions: Prospective controlled randomized trials using neoadjuvant treatment are ideally suited for translational research in EOC and provide unique sample sets molecular analysis. No significant financial relationships to disclose.

We commenced a study in September 1981 to investigate the role of high-dose combination chemotherapy in the management of patients with non-Hodgkin's lymphomas who had failed conventional therapy. Fifty patients with diffuse intermediate- and high-grade non-Hodgkin's lymphomas were treated with high-dose combination chemotherapy with autologous bone marrow rescue (ABMT) and have a minimum follow-up of 1 year. Twenty patients had disease that was still responsive to conventional-dose chemotherapy, 15 had achieved a partial response (PR) to first-line therapy, and five were showing PR to salvage therapy after relapse. Twelve of these patients (60%) achieved complete remission (CR) (two following boost radiotherapy) and three patients have nonprogressive masses on computed tomographic (CT) scan as the only abnormality. None of these patients died during the procedure. Twenty-nine patients had disease not responsive to chemotherapy at conventional dosages: 19 had no response to first-line therapy and 10 showed no response to salvage therapy given after relapse. Only three of these patients achieved CR, all of short duration only. Only two patients in this group remain alive more than 2 years after the procedure and both have nonprogressive abnormalities on CT scan. Nine patients (31%) died of sepsis during the procedure. In those patients with disease not responsive to conventional-dose therapy, dose escalation is associated with a high procedure-related mortality and a low response rate. In those patients who still have chemotherapy-responsive disease the response rate is high and mortality is low.(ABSTRACT TRUNCATED AT 250 WORDS)