Abstract P4-16-12: CARE: A pilot study of the effects of short-term fasting on tolerance to (neo) adjuvant chemotherapy in breast cancer patients

Background:

Preclinical evidence shows that short-term fasting (STS) protects normal cells and makes cancer cells more vulnerable to chemotherapy. This pilot study examines the feasibility and the effects of STS on tolerance to chemotherapy in patients with breast cancer.

Patients and methods:

Eligible patients had histologically confirmed, HER2-negative, early stage breast cancer and adequate bone marrow, liver and renal function. Women receiving (neo) adjuvant TAC courses (docetaxel/adriamycin/cyclophosphamide; day 1, q 3 weeks with G-CSF support at day 2) were randomized to fast 24 hours before and 24 hours after start of chemotherapy or to eat according to the guideline for healthy nutrition. The primary endpoint of the study was to compare neutrophil count after therapy. Secondary endpoints were side effects of chemotherapy, other hematologic counts and chemotherapy-induced DNA damage in leukocytes.

Results:

A total of 13 patients were included of which 7 patients fasted for 48 hours around the chemotherapy infusion (arm A) and 6 patients had a normal diet according to healthy nutrition guidelines (arm B). The median age was 52 years versus 53 years, BMI was 25.5 kg/m2 versus 22.9 kg/m2 and stage III was 43% versus 17% of patients in arms A and B, respectively. Patients were generally motivated to fast and the fasting was well tolerated. Plasma glucose levels were significant lower in fasting patients compared to controls. However, other metabolic parameters showed no significant difference. Fasting did not result in significant differences in neutrophil count or side effects of chemotherapy. Hemoglobin levels and erythrocyte counts after therapy were significantly higher in patients who fasted. Leukocytes of the patients which were isolated at various time points during therapy will soon be analysed for chemotherapy-induced DNA damage and presented at San Antonio.

Conclusion:

This is the first study evaluating the feasibility of 48 hours STS and its impact on side effects of chemotherapy in a homogeneous group of cancer patients. STS was well tolerated and had a beneficial effect on hemoglobin level, but not on experienced side effects. DNA analysis will follow. Larger studies are required to produce more insight into the possible benefits of STS during chemotherapy.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-16-12.

Abstract P1-08-19: Changes in circulating vitamin D levels as a predictor for pathological response to neoadjuvant chemotherapy (NAC) in breast cancer (BC): A Dutch breast cancer trialists group (BOOG) side-study

Background:

Vitamin D (vit D) status is suggested to be of prognostic value for treatment outcome in women with breast cancer. However, there are no data of the predictive value of vit D status and changes of vit D levels for response to neoadjuvant chemotherapy (NAC).

Methods:

A subset of patients (pts) from the NEOZOTAC trial in whom vit D data were available was evaluated. NEOZOTAC is a randomized phase III study comparing the efficacy of NCT with or without zoledronic acid (ZA) in pts with stage II/III, measurable, HER2-negative BC. Vit D deficiency and severe deficiency were defined as vit D levels of ≤ 50 and ≤25 nmol/L, respectively. Baseline vit D levels were available for correlation to pathological response of 165 pts (83 ZA-arm), while 67 pts (35 ZA arm) could be evaluated for changes in vit D levels between baseline and cycle 6. Pts who were allocated to the ZA arm should by protocol receive daily supplements of calcium/vit D 500/400 IU. Pathological response was assessed using the Miller and Payne scoring system; pathological complete response (pCR) was defined as absence of tumor cells in the tumor bed and good response was defined as ≥90% decrease of tumor cellularity.

Results:

Vit D was measured in 168 pts and was done in 75% of pre/perimenopausal pts and 51.3% of postmenopausal pts. There was no significant relation between baseline vit D deficiency (< 50 nmol/L) and pCR (pCR 25.8% for deficient pts vs. 14.1% for non-deficient pts, P = 0.06). Pts with severe vit D deficiency (<25 nmol/L) tended to respond less (pCR 10.5 vs 19.9%, p = 0.53). At the end of chemotherapy, good pathological responders seemed to have a slight increase in vit D levels compared to non-responders who rather showed a decrease (mean 1.11 vs. -9.71, P = 0.08). After multivariate analysis correcting for menopausal status and treatment arm, this result was significant (P = 0.03, 95% C.I. 1.004-1.055). When pts in the ZA arm were analyzed separately, again, good response was rather associated with an increase than a decrease (mean = 9.8 vs. -1.6, P = 0.12). From 17 out of 35 ZA treated pts who were vit D deficient at baseline, only 5 (29.4%) reached levels >50 nmol/L at the end of treatment.

Conclusions:

Baseline vit D status was not predictive for pCR. However, increase in vit D levels during therapy tended to be associated with better pathological response. Therefore, achieving higher vit D levels can be important. Daily suppletion with calcium/ vitamin D 500/400 might be inadequate for achieving sufficient levels after NAC.

Contact information:

Dr. J.R. Kroep, M.D., Ph.D., Department of Medical Oncology, email:j.r.kroep@lumc.nl or A. Charehbili, BSc. Department of Surgery and Medical Oncology, email: a.charehbili@lumc.nl or LUMC datacenter, Department of Surgery, phone +31(0)71-5263500, fax +31(0)71-5266744, email: datacenter@lumc.nl, Leiden University Medical Center (LUMC), Leiden, The Netherlands.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P1-08-19.

Abstract P1-06-04: The predictive value of tumor-stroma ratio for radiological and pathological response to neoadjuvant chemotherapy in breast cancer (BC): A Dutch breast cancer trialists’ group (BOOG) side-study

Background

Intra-tumoral stroma interacts with tumor cells and has a profound effect on tumor behavior. The tumor-stroma ratio (TSR) is of prognostic value in BC and other types of solid tumors. However, the predictive value of this parameter for achieving pathological complete response (pCR) after neoadjuvant chemotherapy is unknown.

Methods

We evaluated the relation between TSR and neoadjuvant treatment response in a retrospective cohort of 69 patients (pts) treated with various regimens of neoadjuvant chemotherapy at our institution who were diagnosed with BC between 1991 and 2007 and of whom radiological response was recorded. The percentage of intra-tumoral stroma was visually estimated on diagnostic sections from primary tumor tissue by two observers. The cut-off point between stroma-rich and stroma-poor tumors was set to 50% (as determined in previous investigations). These results were validated in a cohort from the NEOZOTAC trial: a national, multicenter, randomized study comparing the efficacy of TAC (docetaxel, adriamycin and cyclophosphamide i.v. day 1) chemotherapy with or without zoledronic acid 4 mg i.v., q 3 weeks, 6 times in 250 pts with stage II/III, measurable, HER2-negative BC. Radiological response (complete or partial) was evaluated following RECIST 1.1 criteria. pCR was centrally revised and defined as absence of residual tumor cells in the original tumor bed.

Results

In the retrospective cohort (n = 69) 62.3% of the specimens were classified as stroma-rich. In univariate analysis TSR was significantly associated with radiological response (76.0% stroma-poor vs. 48.8% stroma-rich, P = 0.03). This finding persisted after multivariate analysis for T-status, N-status and ER-status (Odds Ratio [OR] 0.17, 95% C.I.: 0.04-0.78). In the validation set, in which 47.9% of the specimens were stroma-rich (211 cases evaluated), TSR did not predict for radiological response (79.5% stroma-poor vs. 79.2%, P = 0.96). However, when validation data were split on basis of ER-status, TSR was a significant and independent predictor for radiological response in ER-negative pts. (89.5% vs. 50%, P = 0.048, 95% C.I.: 0.01 - 0.98). In the validation set, TSR predicted for pCR with greater pCR rates in stroma-poor tumors (P = 0.03, 22.7% vs 10.3%). Final response results of the pilot and the enlarged sample size of all 250 pts of the validation set will be presented.

Conclusions

TSR might be a marker for radiological and pathological response to neoadjuvant chemotherapy, especially for the ER- tumor subgroup. Considering the simplicity and low cost of TSR assessment, it should be further evaluated and will be prospectively studied in the next neoadjuvant chemotherapy trial of the BOOG.

Contact information:

Dr. J.R. Kroep, M.D., Ph.D., Department of Medical Oncology, email:j.r.kroep@lumc.nl or T.J.A. Dekker, MSc. Department of Surgery and Medical Oncology, email: t.j.a.dekker@lumc.nl or LUMC datacenter, Department of Surgery, phone +31(0)71-5263500, fax +31(0)71-5266744, email: datacenter@lumc.nl, Leiden University Medical Center (LUMC), Leiden, The Netherlands.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P1-06-04.

Abstract OT3-1-03: DIRECT: A phase II/III randomized trial with dietary restriction as an adjunct to neoadjuvant chemotherapy for HER2-negative breast cancer

Background:

Preclinical evidence shows that short-term fasting protects normal cells, while cancer cells are sensitized to chemotherapy. Furthermore, a specifically designed very low calorie, low amino acid substitution diet (“Fasting Mimicking Diet”, FMD) has similar effects on chemotherapy as short-term fasting. This trial evaluates the impact of FMD on tolerance to and efficacy of neoadjuvant chemotherapy in women with HER2-negative early breast cancer.

Trial design:

DIRECT is a Dutch, randomized, open-label multicenter phase II/III trial. Women receiving neoadjuvant TAC courses (docetaxel/adriamycin/cyclophosphamide; day 1, q 3 weeks with G-CSF support at day 2) will be randomized with or without FMD for 3 days prior to and the day of chemotherapy and 3 days prior to surgery.

Eligibility criteria:

Eligible women are WHO 0-2, age ≥18 years, HER2-negative, stage II or III breast cancer and adequate bone marrow, liver and renal function, BMI > 19kg/m2 and absence of diabetes mellitus.

Study endpoints:

The primary endpoints are grade III/IV toxicity (phase II) and the pathologic complete response rate (pCR) (phase III). Secondary endpoints are grade I/II toxicity, metabolic and inflammatory response to chemotherapy, DNA damage, apoptosis, immunology and nutrient sensing pathways in the tumor, biomarkers as single nucleotide polymorphisms, Ki67 and tumor stroma/ratio, patient's quality of life and (disease free) survival. Optional side studies include chemotherapy-induced DNA damage and nutrient sensing pathways in leukocytes and proteomics.

Statistical Methods:

Using a 5% significance level based on the two-sided Fisher's exact test with a power of 80%, 128 patients (64/arm) will be enrolled to show a 50% decrease of grade III/IV adverse events in the experimental arm (phase II) and 250 patients (125/arm) are needed to show an improvement of the pCR rate from 18% to 36% (phase III).

Target accrual:

Recruitment will start in September 2013. The expected end of accrual of 250 patients from multiple centers in the Netherlands will be the last quarter of 2015.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr OT3-1-03.

Abstract P1-12-05: First-line chemotherapy with pegylated liposomal doxorubicin versus capecitabine in elderly patients with metastatic breast cancer: results of the phase III OMEGA study of the Dutch Breast Cancer Trialists' Group (BOOG)

Background The efficacy and feasibility of chemotherapy in elderly metastatic breast cancer (MBC) patients (pts) have been studied in various phase II studies. However, results of prospective randomized studies in elderly MBC pts are scarce.

Methods In this phase III multicenter study, MBC pts ≥ 65 years eligible for first-line chemotherapy were randomized between pegylated liposomal doxorubicin (PEGdoxo) (45mg/m2, IV, q 4 wks) or capecitabine (Cape) (1000 mg/m2 PO bid, days 1–14, q 3 wks). Other eligibility criteria were ECOG performance status (PS) ≤ 2 (3 allowed if due to pain or pre existing comorbidity), adequate bone marrow and organ functions. Stratification factors were PS (0–1 vs 2–3), HER2 status, visceral/non-visceral disease, adjuvant hormonal therapy (HTx), and HTx for MBC. Baseline geriatric assessment (GA) included functional status, instrumental activities of daily living, cognition, mood, comorbidity, polypharmacy and nutritional status. Chemotherapy was continued for 24 wks in the absence of progressive disease (PD) or unacceptable toxicity. Primary endpoint was progression-free survival (PFS), secondary endpoints were response rate, overall survival (OS), toxicity (CTC criteria) and compliance.

Results Between April 2007 and August 2011, 78 pts were randomized to PEGdoxo (n = 40) or Cape (n = 38). The study was prematurely closed due to slow accrual and supply problems with PEGdoxo. Mean age was 74 years (range 65–86; 75+ 54%; 80+ 13%). Pt characteristics were balanced between the two arms: PS 0–1 77%, ER+ 68%, HER2+ 5%, visceral/non-visceral disease 76%/24%, adjuvant HTx 46%, HTx for MBC 56%, ≥ 3 metastatic sites 50%. Only 22 out of 75 pts with a baseline GA had no geriatric condition (29%), while 32 pts (43%) and 21 pts (28%) had one or ≥ 2 geriatric conditions, respectively. Chemotherapy was given for 6 months in 38%, with a mean dose intensity of 84% in both arms. Reasons for early treatment discontinuation were: PD (31%), toxicity (28%), pt withdrawal (3%). After a median follow up of 32 months, 74 pts had PD and 56 pts had died. The median PFS was 5.7 and 7.7 months with PEGdoxo and Cape (HR 0.68, 95% CI: 0.42–1.11, p = 0.12) and the median OS was 13.8 and 16.8 months, respectively (HR 0.84, 95% CI: 0.49–1.42, p = 0.51). Response was evaluable in 64 pts, with a partial response (PR) in 7 (21%) and 6 pts (19%), and stable disease in 21 (64%) and 17 pts (55%) for PEGdoxo and Cape, respectively. Toxicity was acceptable, mainly being grade 1–2, with for PEGdoxo/Cape grade 1 alopecia in 14/4 pts (grade 2 in 1 PEGdoxo pt), grade 3 fatigue in 5/5 pts, grade 3 HFS in 4/6 pts and grade 3 mucositis in 4/1 pts, respectively. Pts with ≥ 1 geriatric condition more frequently experienced grade 3–4 toxicity, after correcting for type of chemotherapy, age and PS (HR 2.24, 95% CI: 1.21–4.16). Pts aged 75+ had a twofold higher risk of dying, irrespective of treatment arm (HR 2.31, 95% CI: 1.31–4.07).

Conclusions First-line chemotherapy with either PEGdoxo or Cape was feasible in elderly MBC pts, with adequate dose intensity and acceptable toxicity, even in non-fit pts or pts aged 75+. Baseline GA correlated with toxicity.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-12-05.

P5-20-08: Multicenter Results of Scalp Cooling To Prevent Chemotherapy-Induced Alopecia in 1500 Breast Cancer Patients

Background Chemotherapy-induced alopecia (CIA) is a frequent occurring side effect of cancer treatment that has high psychological impact on many patients and their relatives. CIA may be prevented by scalp cooling.

Methods Breast cancer patients who received scalp cooling could participate in this registration study from 2006 and onwards. Nurses and patients completed questionnaires. Scalp cooling was performed using the Paxman PSC1 or PSC2 system and was considered satisfying when patients did not wear a wig or head cover. Logistic regression analyses will be used to examine determinants of the scalp cooling result, including age, type/length/thickness/chemical manipulation (dyeing, waving, colouring) of hair, type/dose of chemotherapy, cytostatic infusion time, post-infusion cooling time, dampening hair or use of conditioner before scalp cooling and previous treatment with chemotherapy.

Results The use of scalp cooling increased from 4 hospitals in 2005 to 60 out of 100 Dutch hospitals in 2011. The registration comprised about 1500 scalp cooled breast cancer patients of whom 79% were treated in the adjuvant setting. Overall, patients’ satisfaction with the result of scalp cooling was 50%, but varied for different chemotherapy schemes and dosages from 8% in docetaxel 75 mg/m2, doxorubicin 50 mg/m2, cyclophosphamide 500 mg/m2 (TAC) to 81% in paclitaxel 70–90 mg/m2. The results of the regression analyses will be presented at the conference.

Discussion Scalp cooling is effective in most commonly used chemotherapy regimens for breast cancer patients. Therefore, scalp cooling should be offered more often. Efficacy depends on the type and dose of chemotherapy. Other factors influencing the result have never been studied in multivariate analyses before. Scalp cooling is ineffective in patients treated with a combination of a taxane and an anthracyclin. The frequent use of scalp cooling in the adjuvant setting indicates minimal fear for an increased incidence of scalp skin metastases after scalp cooling in breast cancer patients. Multicenter registration of results improves information for medical professionals and patients and will lead to more frequent use of scalp cooling and improvement of the method.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-20-08.

Abstract P3-15-01: Primary G-CSF Prophylaxis during the First Two Cycles Only or Throughout All Chemotherapy Cycles in Breast Cancer Patients at Risk of Febrile Neutropenia: Final Results from a Phase III Trial of the Dutch Breast Cancer Trialists Group (BOOG)

Background: In patients at risk of febrile neutropenia (FN), the highest incidence of FN is seen during the first chemotherapy cycles. This observation questions the effectiveness of continued G-CSF prophylaxis during later chemotherapy cycles.

Methods: We conducted a multicenter phase III study in breast cancer. Patients were randomized to primary G-CSF prophylaxis during the first two chemotherapy cycles only (experimental arm) or to primary G-CSF prophylaxis throughout all chemotherapy cycles (standard arm). Patients were eligible if treated with 3-weekly chemotherapy, being at risk of FN according to criteria of international guidelines. This was an equivalence study, aimed to include 230 patients, with FN as primary endpoint. Results: An independent Data Safety Monitoring Board recommended premature closure of the study because of an unacceptable increase in FN rate in the experimental arm. In total, 85 patients enrolled the experimental arm and 84 patients the standard arm till 15 th December 2009. Baseline characteristics were well belanced: age ≥65years in 93% versus 95%, ECOG PS of zero in 82% versus 88%, treated with docetaxel containing chemotherapy in (neo)-adjuvant setting in 97% versus 100%, respectively. At least one episode of FN was seen in 27 (32%) patients of the experimental arm compared with 4 (5%) patients in the standard arm (P<0.0001). Notably, in the experimental arm FN occurred predominantly in the third cycle, whereas in the standard arm FN was seen throughout all cycles. Fever, infection and/or mucositis led to serious adverse events in 31 (36%) patients in the experimental arm versus in 11 (13%) patients in the standard arm (P<0.001). In both arms, 5% of patients stopped or changed treatment because of toxicity. Most patients experienced only one FN episode. In these, chemotherapy was mostly given as planned, but in 61 % of patients with secondary G-CSF prophylaxis and in 21% of patients with secondary antibiotic prophylaxis (in the experimental arm). Conclusion: We conclude that in patients at risk of FN, primary G-CSF prophylaxis cannot be limited to the first chemotherapy cycles, despite the well known increased FN incidence in the first cycles without prophylaxis. With only 2 cycles protected, FN risk is increased in the third cycle. Support: The Netherlands organization for health research and development (ZonMw) and sanofi-aventis Netherlands B.V.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P3-15-01.

Abstract P5-11-08: Effects of Exemestane or Tamoxifen on Bone Health within the Tamoxifen Exemestane Adjuvant Multinational (TEAM) Trial: A Meta-Analysis

Background: TEAM is the largest AI phase III trial comparing exemestane with tamoxifen followed by exemestane as adjuvant breast cancer therapy in postmenopausal women. We performed a meta-analysis of three randomized sub-studies of the TEAM trial conducted in Germany, the Netherlands/Belgium and the United States to determine the effects on bone health.

Methods: Patients were randomised to exemestane or tamoxifen as adjuvant therapy for hormone receptor-positive breast cancer. Bone mineral density (BMD) was assessed by dual-energy X-ray absorptiometry at baseline and after 6, 12 and 24 months’ treatment. Bone turnover markers were also measured.

Results: 412 patients were evaluable. Patients in the tamoxifen group showed a mean increase in lumbar spine BMD of 1.2% from baseline to month 12 and 0.2% to month 24. Patients in the exemestane group showed a mean decrease from baseline of 2.6% after 12 months and 3.5% after 24 months. There were significant differences in the changes in BMD at the lumbar spine between treatment groups (P<0.0001 at both timepoints). In the tamoxifen group, a mean increase in total hip BMD of 0.8% from baseline to month 12 and a mean decrease from baseline of 0.4% after 24 months was observed, compared with a mean decrease of 1.3% after 12 months and 3.3% after 24 months in the exemestane group. Changes in BMD from baseline at the total hip were also significantly different between exemestane and tamoxifen (P<0.05 at both timepoints). Bone turnover markers decreased from baseline with tamoxifen and increased with exemestane.

Conclusions: After 24 months, exemestane treatment resulted in decreases in BMD and increases in bone turnover markers. In contrast, BMD was increased and bone turnover markers were decreased with tamoxifen. BMD and bone turnover changes appeared to stabilise after initial treatment.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P5-11-08.

Abstract S2-3: The Impact of Body Mass Index (BMI) on the Efficacy of Adjuvant Endocrine Therapy in Postmenopausal Hormone Sensitive Breast Cancer (BC) Patients; Exploratory Analysis from the TEAM Study

Background: Obesity is associated with an increased risk of breast cancer (BC) recurrence and decreased survival, also in case of adjuvant endocrine therapy. It is still not clear whether the activity of aromatase inhibitors and tamoxifen (T) given as adjuvant therapy is affected by body mass index (BMI), although both drugs are widely prescribed. In this analysis, we explored the outcome of TEAM patients (pts) treated with exemestane (E) versus T (2.75 yrs), and with E versus T followed by E (T/E) (5 yrs) in relation to BMI.

Patients and Methods: The TEAM trial is a randomized, international phase III study in postmenopausal hormone sensitive early BC pts comparing the activity and safety of adjuvant E (25 mg daily) or the sequence of T (20 mg daily) followed by E (T/E), both regimens given for five years. WHO BMI definitions were used: normal 18.5-24.9 kg/m2, overweight 25-30 kg/m2, obese >30 kg/m2. Disease-free survival (DFS) and overall survival (OS) were calculated by Kaplan-Meier method; results were compared by using the log-rank test and Cox proportional hazard modelling adjusted for country.

Results: Weight and height was known in 4741 pts. Mean BMI was 26.9 kg/m2 (SD 4.9); 39% had a normal BMI, 36.9% overweight, and 23.3% of pts was obese. Underweight pts (n=41, 0.9%) were excluded from further analysis. At 2.75 yrs (E vs T) disease relapse in normal weight, overweight and obese pts using E was observed in 8.1%, 6.8% and 7.5% respectively (p=0.57), and in 9.1%, 8.8%, and 12.5%, respectively (p=0.06) of pts using T. The hazard ratio (HR for risk of relapse on E vs T) in the three subgroups was 0.91 (95%CI 0.66-1.24), 0.78 (95%CI 0.55-1.089), and 0.57 (95%CI 0.39-0.84, p=0.004), respectively. At a median follow-up of 5.1 years, disease relapse in normal weight, overweight and obese pts using E occurred in 14.8%, 15.1% and 15.1%, respectively; and in pts using T in 17.0%, 16.9%, and 18.3%, respectively. Regarding DFS, the HR in normal weight, overweight, and obese pts was 0.87 (95%CI 0.69-1.10), 0.88 (95%CI 0.70-1.11), and 0.75 (95%CI 0.56-1.01, p=0.058), respectively, and with respect to OS 0.87 (95%CI 0.65-1.15, p= 0.32), 0.89 (95%CI 0.67-1.18, p= 0.43), and 0.71 (95% CI 0.51-1.01, p= 0.053), respectively.

Conclusions: After 2.75 years more disease events were observed in obese women using tamoxifen, which was not seen in obese exemestane users, whereas at 5 years these differences in disease recurrences disappeared in this group. In contrast to recent reports, there seems to be a difference regarding the influence of a high BMI on recurrence rate between tamoxifen and the aromatase inhibitor exemestane. Further research on this topic is warranted.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr S2-3.

Stability of preferences with regard to adjuvant chemotherapy: impact of treatment decision, experience and the passing of time

Research has shown that patients' preferences for adjuvant chemotherapy do not change as a result of experience. However, the preferences of experienced patients are usually more favourable than those of inexperienced patients. These results indicate a shift in preferences after the decision to proceed with adjuvant chemotherapy has been made, but before actual experience. We tested this assumption in early-stage breast and colorectal cancer patients. We asked patients to provide their preferences for chemotherapy before surgery and thus before they knew whether chemotherapy would be advised (T(1)), after surgery but before the start of chemotherapy (T(2)) and about 1 month after chemotherapy (T(3)). Patients who did not undergo chemotherapy co-operated at similar points in time. Preferences were measured on a nine-point scale, ranging from (1) 'very strong preference for no chemotherapy' to (9) 'very strong preference for chemotherapy'. As hypothesized, the preferences of patients who would be treated with chemotherapy became more favourable after the treatment decision had been made (n = 7, P = 0.06). The preferences of patients for whom chemotherapy was not part of the treatment plan showed the opposite effect (n = 38, P = 0.03). We did not find any effect of experiencing treatment (n = 22, P = 0.62) or the passing of time (n = 81, P = 0.25) on the stability of preferences. We conclude that the frequently observed discrepancy in treatment preferences between experienced and inexperienced patients seems to be an effect of the treatment decision and not of experience of the treatment.