P1-11-02: Racial/Ethnic Differences in Adjuvant Trastuzumab Receipt for Women with Breast Cancer within the National Comprehensive Cancer Network

Background Racial/ethnic disparities in breast cancer care are well documented. Although adjuvant trastuzumab has been shown to improve disease outcomes for women with Human Epidermal Growth Factor Receptor 2 (HER2)-positive cancers, the ‘real world’ utilization and toxicity of adjuvant trastuzumab are unknown. Because therapy involves one year of treatment and the costs of treatment are high, a risk for treatment disparity exists. We examined differences in receipt and completion of adjuvant trastuzumab by race/ethnicity, education, employment, and insurance for women diagnosed with HER2−positive breast cancer.

Methods Using the National Comprehensive Cancer Network (NCCN) Breast Cancer Outcomes Database, we identified 1,146 women with stage I-III HER-2 positive breast cancer who presented to participating NCCN institutions during 2005–2008. In multivariable logistic analyses, we assessed the effect of race/ethnicity on the likelihood of trastuzumab therapy, and among women who initiated trastuzumab, the likelihood of completing ≥ 270 days of therapy, adjusting for center, diagnosis year, age, insurance, comorbidity, education, employment, and tumor characteristics. We also examined reasons for discontinuation of therapy among those who stopped treatment prematurely.

Results Among patients eligible for this analysis, 75% women were Caucasian, 9% were African-American, and 9% were Hispanic. Most women had managed care insurance (71%) and were employed/student (52%). About one-third (36%) had a college degree and 39% had a high school education or less. Overall, most women (82%) received neo/adjuvant trastuzumab and there were no racial/ethnic differences in receipt of therapy (adjusted odds ratio [OR] 1.11, 95% confidence interval [CI] .72-1.71 for African-American and OR 1.39, 95% CI .76-2.54 for Hispanic, versus Caucasian women). Among the 769 women who initiated neo/adjuvant trastuzumab and had ≥ 365 days of follow-up, 84% completed ≥ 270 days of trastuzumab. Rates of completion were lower for African-American (72%) and Hispanic (82%) women than Caucasian women (85%). In adjusted analyses, African-American women but not Hispanic women had lower odds of completing therapy compared with Caucasian women (OR .45, 95% CI .29-.70, p=0.0003). Indemnity insurance (versus managed care) was associated with lower odds of trastuzumab completion, as was having a high school education or less (versus college education). Among the 123 women who did not complete trastuzumab, 26% stopped early for toxicity, and this occurred more frequently for African-American women than Caucasian women (50% vs. 21%), but small sample precluded a meaningful test for statistical significance.

Conclusion: Compared with Caucasian women, African-American women had similar rates of initiation of adjuvant trastuzumab but much lower rates of completion that were not explained by differences in education, employment, or insurance. Because of the significant benefits conferred by adjuvant trastuzumab therapy for HER2−positive breast cancer, interventions to assure completion of therapy could lead to improved outcomes. Further exploration of racial differences in toxicity and tolerance of therapy are also warranted.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P1-11-02.

P1-08-05: Age and Survival in Women with Early Stage Breast Cancer: An Analysis Controlling for Tumor Subtype

Background: Previous research has suggested that young age at diagnosis is an independent risk factor for breast cancer recurrence and death in women with early stage breast cancer. However, young women are more likely to have aggressive subtypes of breast cancer. No prior studies have adequately controlled for tumor phenotype, including HER-2/neu (HER2) status, in particular. Recent evidence has suggested that the prognostic effect of young age varies by tumor subtype.

Methods: We examined data from women with newly diagnosed Stage 1–3 breast cancer presenting to one of 8 NCCN centers between January 2000 and December 2007. Multivariate Cox proportional hazards models were used to assess the relationship between age and breast cancer specific survival, controlling for known prognostic factors and treatment. In addition, we conducted stratified analyses by estrogen receptor (ER) and HER2 status.

Results: 19,633 women with Stage 1–3 breast cancer eligible for analysis including 2,177 (11%) who were age 40 years or younger at diagnosis. Younger women were more likely to be non-white or Hispanic, more educated, employed, and to have higher stage, high grade, ER-negative, progesterone receptor (PR) negative, and HER2−positive disease, and treated with chemotherapy and trastuzumab (all variables P< 0.0001 by Chi-Square test). 5-year survival among younger women was 94.1 (95% Confidence Interval [CI] 92.9−95.3) and 96.3 (95% CI 95.9−96.6) for older women. In a multivariate Cox proportional hazards model controlling for sociodemographic, disease, and treatment characteristics, women age < 40 or younger at diagnosis had increased mortality compared to older women (Hazard Ratio [HR] 1.26, 95% CI 1.02−1.56). In stratified analyses, age 40 or less was associated with increased mortality among women with ER-positive disease (HR 1.44, 95% CI 1.01−2.05), but was not among those with ER-negative disease (HR 1.15, 95% CI 0.85−1.55). Younger age was associated with a statistically significant increase in mortality among women with HER2−negative disease (HR 1.29, 95% CI 1.00−1.68), but this difference did not reach statistical significance among those with HER2−positive disease (HR 1.30, 95% CI 0.82−2.09). Conclusions: The effect of age on short-term survival of women with early breast cancer appears to vary by breast cancer subtype, particularly ER status. Further research to elucidate differences in breast cancer biology and efficacy of therapy within tumor types by age is warranted.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P1-08-05.

Abstract P2-15-01: Long-Term Risk Perceptions and Quality of Life of Women with Ductal Carcinoma In Situ

Background: Research has demonstrated that many women with ductal carcinoma in situ (DCIS) overestimate their risk of future breast cancer events. However, no prior studies have evaluated risk perceptions in long-term follow-up.

Methods: As part of a multicenter longitudinal cohort study, we mailed long-term follow-up surveys to 315 of 392 women who had previously responded to a survey 18 months after they were diagnosed with DCIS (33 of the 392 were excluded because they were receiving follow-up care at outside centers, and 44 because of recurrence/death). We evaluated psychosocial distress by Hospitalized Anxiety and Depression Scale (HADS) and Revised Impact of Events Scale (RIES), quality of life (QOL) by Short Form Health Survey (SF-36), and risk perceptions with items used previously in the cohort.

Results: One hundred ninety-three women (61%) responded. They were a median age of 53 yrs (range 31-89) and a median of 5.8 yrs from diagnosis (range 4.3-7.0). Twelve were excluded due to report of recurrence. Of the 181 remaining, 32% perceived at least a moderate risk of developing DCIS again within 5 yrs; 43% perceived at least a moderate lifetime risk of developing DCIS again; 27% perceived at least a moderate risk of developing invasive breast cancer within 5 yrs; 38% perceived at least a moderate lifetime risk of developing invasive breast cancer; 24% perceived at least a moderate risk of DCIS spreading to other parts of their bodies. For qualitative responses of ≥ moderate perceived risk, median quantitative risk perception was 10% regarding 5-yr risk of DCIS or invasive cancer, 20% regarding lifetime risk, and 50% pertaining to risk of DCIS spreading. The proportion of women reporting ≥ moderate perceived risk of DCIS spreading remained stable over time (25% at baseline; 26% at 18 mos), but this proportion decreased for other perceptions of risk (e.g., at baseline and 18 mos, 51% and 50% of these same women perceived ≥ moderate risk of developing DCIS again within 5 yrs). In a multivariable model, worse financial status (OR 2.6, 95% CI 1.3-5.3) and higher perceived risk on earlier surveys (OR 24.4, 95% CI 12.7-47.2) were the only predictors of ≥ moderate perception of risk of DCIS spreading to other parts of the body in long-term follow-up. Non-significant covariates included age, race, education, grade of DCIS, comedonecrosis, mastectomy, radiation, marital status, employment status, comorbidity, and HADS and RIES scores. This contrasted with earlier survey data, which demonstrated an association between higher risk perceptions and anxiety by HADS and RIES. In long-term follow-up, only 7% were found to be anxious (HADS Anxiety ≥11) and 0.6% were depressed (HADS Depression ≥11). Median SF-36 scores were 48.9 (range 15.7-58.2) and 47.1 (range 23.6-60.3) for the physical and mental components, respectively, consistent with overall good QOL.

Conclusions: Women with a history of DCIS continue to harbor inaccurate perceptions of risk of future breast cancer events even at 6 yrs follow-up. Future research should evaluate how these excessive risk perceptions impact health behaviors. Interventions to improve risk perceptions are warranted.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P2-15-01.

Clinicopathological features and sites of recurrence according to breast cancer subtype in the National Comprehensive Cancer Network (NCCN)

543

Background: Gene expression profiling has defined multiple breast cancer subtypes which can approximated using standard immunohistochemical markers. Methods: We assessed clinicopathological features and sites of recurrence for patients (pts) presenting to NCCN sites with stage I-III breast cancer from Jan 2000 to Dec 2006 where estrogen receptor (ER), progesterone receptor (PR), and HER2 status were known. Tumors were grouped as luminal A (ER+ and/or PR+, and HER2-), HER2+ (any ER or PR, and HER2+), or triple-negative (ER-, PR-, and HER2-). Chi-square compared proportions across tumors; univariate logistic regression estimated risk of first site of recurrence. Results: 12,858 pts met inclusion criteria. Median follow-up from NCCN presentation was 3.2 years. Subtype distribution was: triple-negative (TN) 17%; HER2+ 18%; luminal A 66%. Compared to pts with luminal A and HER2+ tumors, TN were younger (p<0.0001), more likely African-American (p<0.0001) and overweight (p=0.0006). TN and HER2+ tumors were less often detected by screening mammography (TN, 28.9%; HER2+, 33.6%; luminal A, 48.4%) and less likely to present as T1 (TN, 46.5%; HER2+, 50.5%; luminal A, 67.0%) or diagnosed as stage I (TN, 32.6%; HER2+ 33.2%; luminal A, 49.4%) than luminal A (all p<0.0001). Rate of node positivity was lowest in TN (TN, 37.1%; HER2+, 44.9%; luminal A, 38.1%; p<0.0001). 83% of TN tumors were high grade; 93% were invasive ductal histology. Extensive intraductal component and lymphovascular invasion were more often associated with HER2+, compared to TN or luminal A (p<0.0001). Recurrences were recorded for 1,235 pts. Relative to luminal A, TN and HER2+ were more likely to experience lung (TN, odds ratio [OR] 2.27, 95% confidence interval [CI] 1.50, 3.43; p=0.0001; HER2+, OR 1.65, 95% CI 1.05, 2.60; p=0.03) and brain (TN, OR 5.32, 95% CI 2.85, 9.91; p<0.0001; HER2+, OR 5.53, 95% CI 2.93, 10.43; p<0.0001) as first site of recurrence; bone was less likely (TN, OR 0.23, 95% CI 0.16, 0.33; p<0.0001; HER2+, OR 0.38, 95% CI 0.28, 0.53; p<0.0001). Conclusions: Clinicopathological features and patterns of recurrence differed significantly by subtype and may inform the design of future clinical trials.

No significant financial relationships to disclose.

Quality of breast cancer care in NCCN centers as assessed by the ASCO/NCCN quality measures: Overall performance and reasons for nonconcordance

6506

Background: To assess the quality of breast cancer care in the National Comprehensive Cancer Network (NCCN), we compared practice against the ASCO/NCCN quality measures (QM). Methods: Using the NCCN Outcomes Database, we studied the care of women with newly diagnosed stage I-III breast cancer treated at 8 NCCN centers in 2003–6 to determine the proportion whose care was consistent with the 3 QMs (tamoxifen or anastrozole within 1 year of diagnosis for HR+ >1 cm tumors [HT]; post- lumpectomy radiation within 1 year of diagnosis for women <70 yo [RT]; and chemotherapy within 120 days of diagnosis of HR- >1 cm tumors for women <70 yo). Based on chart review, reasons for non-concordant care were classified as: 1) treatment recommended, but declined; 2) MD recommended against treatment; 3) non-recommended treatment administered; or 4) system problem (referral to the relevant specialist not documented or specialist not seen, delayed treatment initiation, or no relevant documentation in the chart). Results: 5,175 women were analyzable on at least one QM. Overall, treatment was consistent with the QM for 6,628/7,265 (91%) of the unique patient-recommendation pairs. Non-concordant care was received by 11% of patients eligible for the HT QM, 5% for the RT QM, and 13% for the chemotherapy QM. The reasons for non-concordance are shown in the table. Conclusions: Overall, breast cancer care in the NCCN is highly concordant with the ASCO/NCCN quality measures. System problems rather than patient refusal or physician recommendations account for the majority of care that deviates from the QM recommendations. This suggests that quality improvement targeting systems rather than physicians would be especially efficient and effective.

[Table: see text]

No significant financial relationships to disclose.

Identifying high-priority quality measures for breast cancer quality improvement using data from a nationally representative sample

6507

Background: We previously developed a simple, data-based strategy for selecting high-priority quality measures, and applied it to practice performance data from eight National Comprehensive Cancer Network (NCCN) centers to identify high-priority measures for breast cancer care. We now seek to evaluate the generalizability of these results by identifying which measures rank highly when practice performance data from a nationally representative sample, rather than a select group of cancer centers, are used. Methods: Using SEER-Medicare data, we assessed practice performance for 9933 women 65–70 years old diagnosed 1998–2002 with non-metastatic invasive breast cancer. For each quality measure, we determined the number of eligible patients, the number who did not receive recommended therapy, and the overall and target (goal) concordance values. We entered these data into a previously described algorithm to derive a numerical prioritization score for each measure. Spearman correlations were used to analyze comparisons between datasets. Results: We were able to evaluate practice performance relative to most (18/30; 60%) of the measures included in our original NCCN analysis. The absence of oral hormone therapy and DCIS pathology data in SEER-Medicare precluded assessments of quality relative to the remaining measures. The highest-ranking measures recommended chemotherapy for hormone-receptor positive tumors >1 cm, axillary node surgery and radiation after lumpectomy for stage I-II cancer, and no radiation after mastectomy for node-negative cancer <5 cm. Excluding hormone therapy and DCIS measures, the 6 highest-ranking measures in the SEER-Medicare and NCCN analyses were identical. Correlation between the prioritization scores for the 18 measures common to both datasets was 0.74. Conclusions: While SEER-Medicare data only include patients >65 and NCCN data only represent select institutions, information from both can be used to identify high-priority quality measures. That the same measures rank highly in both analyses suggests these are comprehensive, generalizable, policy ready national standards, and that this is a valid method for identifying high-priority quality measures.

No significant financial relationships to disclose.

The impact of obesity on adherence to the National Comprehensive Cancer Network (NCCN) guidelines recommending chemotherapy for patients with operable breast cancer

6517

Background: Obesity (measured using body mass index [BMI]) is regarded as a prognostic factor for worse breast cancer survival. We hypothesized that obesity may influence adherence to NCCN guidelines recommending chemotherapy for patients with operable breast cancer at NCCN centers. Methods: We identified women < 70 years diagnosed with stage I, II, or III breast cancer from 1997 to 2006 at 8 NCCN centers for whom guidelines recommended chemotherapy. Body mass index was assessed in categories (<25 kg/m2 [normal], 25 to <30 kg/m2 [overweight], 30 to 39kg/m2 [obese], ≥40 kg/m2 [morbidly obese]) and in 5 kg/m2 increments. Multivariable logistic regression analysis adjusting for center, age at diagnosis, ethnicity, comorbidity score, and guideline was used to examine the association between BMI at diagnosis and non-receipt of chemotherapy. Results: 9,389 women were eligible for the study: 38% normal weight; 30% overweight; 23% obese; 5% morbidly obese; and 4% unknown. In multivariable analysis with BMI as a categorical variable, there was no association between weight status and non-receipt of chemotherapy (p = 0.35). When BMI was assessed in 5kg/m2 increments, weight status was a statistically significant predictor of non-receipt of chemotherapy (p = 0.02), but the odd ratios exceeded 1.0 only for BMIs ≥42.6kg/m2. Other patient-related factors associated with non-receipt of chemotherapy included older age at diagnosis (p < 0.01), presence of comorbidities (p < 0.01) and center (p < 0.01). Conclusions: Overall, the quality of breast cancer care as measured by adherence to NCCN guidelines recommending chemotherapy was not affected by patient overweight or obese status. Chemotherapy use was lower among patients with extreme morbid obesity, which may represent appropriate clinical decision-making. Evaluating factors that may contribute to worse prognosis among obese patients is essential for individualizing care and improving breast cancer outcome.

No significant financial relationships to disclose.

Direct-to-consumer advertising for bleeding disorders: a content analysis and expert evaluation of advertising claims

OBJECTIVE

In the United States, the Food and Drug Administration (FDA) requires that all direct-to-consumer advertising (DTCA) contain both an accurate statement of a medication's effects ('truth') and an even-handed discussion of its benefits and risks/adverse effects ('fair balance'). DTCA for medications to treat rare diseases such as bleeding disorders is unlikely to be given high priority for FDA review.

METHODS

We reviewed all DTCA for bleeding disorder products appearing in the patient-directed magazine HemeAware from January 2004 to June 2006. We categorized the information presented in each advertisement as benefit, risk/adverse effect, or neither, and assessed the amount of text and type size devoted to each. We also assessed the readability of each type of text using the Flesch Reading Ease Score (FRES, where a score of >or=65 is considered of average readability), and assessed the accuracy of the advertising claims utilizing a panel of five bleeding disorder experts.

RESULTS

A total of 39 unique advertisements for 12 products were found. On average, approximately twice the amount of text was devoted to benefits as compared with risks/adverse effects, and the latter was more difficult to read [FRES of 32.0 for benefits vs. 20.5 for risks/adverse effects, a difference of 11.5 (95% CI: 4.5-18.5)]. Only about two-thirds of the advertising claims were considered by a majority of the experts to be based on at least low-quality evidence.

CONCLUSION

As measured by our methods, print DTCA for bleeding disorders may not reach the FDA's standards of truth and fair balance.

Hope and prognostic disclosure

6510

Background: Physicians sometimes selectively convey prognostic information to support patients’ hopes. However, the relationship between prognostic disclosure and hope is not known. Methods: We surveyed 194 parents of children with cancer (overall response rate 70%) in their first year of cancer treatment at the Dana-Farber Cancer Institute and Children’s Hospital, Boston, Mass, and the children’s physicians. We evaluated relationships between parents’ recall of prognostic disclosure by the physician and the extent to which physician communications “always” made them feel hopeful. A five-item index of prognostic disclosure assessed whether prognostic information was provided in any form, in quantitative terms, and in written form, whether the physician gave prognostic information before the parent asked, and whether parents wanted additional prognostic information beyond what they had already received. Results: Parents were less likely to report hopeful communication when the child’s likelihood of cure was low (OR .70 per category of decreasing likelihood of cure, P=.0003). However, parents who reported having received more extensive prognostic information were more likely to report hopeful communication, even when the prognosis was poor. In a multivariable model, parental report that physician communication “always” made them feel hopeful was associated with increased prognostic disclosure (OR 1.67 per element of disclosure, P=.009) and higher perceived communication quality (OR 5.39, P<.0001). In contrast, communication-related hope was inversely associated with the child’s likelihood of cure (OR .67, P=.006). Conclusions: Although physicians sometimes selectively convey prognostic information to preserve hope, we found no evidence that prognostic disclosure makes parents less hopeful. Instead, disclosure of prognosis by the physician can support hope for parents of children with cancer, even when the child’s prognosis is poor.

No significant financial relationships to disclose.

[Table: see text]

Communication and understanding of prognosis among stage IV lung and GI cancer patients

9024

Background: Patients' understanding of prognosis is essential to informed decision making about treatment options. There is little data on what prognostic information oncologists communicate to patients and what patients actually understand about their prognosis. Methods: We audio taped consultations between 90 newly diagnosed stage IV non small cell lung cancer patients and stage IV GI cancer patients and medical oncologists at academic teaching hospitals. Interactions were audio taped until patients arrived at a treatment decision. Audiotapes were digitized, coded and analyzed using the RIAS system. All conversation relating to prognosis was blocked, coded and analyzed. After meeting with an oncologist patients were surveyed about their understanding of prognosis and desire for information about life expectancy. Results: 276 patients were contacted of whom 190 consented to participate. 87 patients were ineligible, 13 patients withdrew from the study, and complete data were obtained on 90 patients. The mean age of participants was 62 years (±11), 65% were women. Analysis of the audiotapes revealed that in 78% of cases oncologists told patients that their cancer was not curable. When patients were surveyed, however, only 32% of those who were told that their cancer was not curable actually understood that there was no cure for their cancer (CI 0.19–0.45). Patients’ age, gender, education, and marital status were not associated with an understanding of prognosis. Patient factors associated with understanding of prognosis included a desire for good and bad medical information (CI 0.75–0.99; p<0.0001), an interest in knowing the likelihood that treatment would cure their disease (CI 0.54–0.94; p<0.01), and a belief that information about life expectancy would influence their medical decisions (CI 0.56–0.94; p<0.01). Conclusion: In a majority of cases oncologists actually told patients that their cancer is not curable. Most patients, however, did not understand what was communicated. Patients’ understanding of prognosis is determined not only by what oncologists communicate, but also by patients’ desire for prognostic information.

No significant financial relationships to disclose.

Measuring clinically significant chemotherapy-related toxicities using Medicare claims from CALGB breast and lung cancer trial participants

6595

Background: Because the elderly are numerically underrepresented in cancer clinical trials, the benefits and toxicities of cancer therapies in the general population of elderly patients is not known. Nevertheless, clinicians need such information. A solution may be found in analyses of observational data. Specific Aim: We performed a criterion validation study to determine the accuracy with which observational Medicare claims data measure clinically significant chemotherapy-related toxicities in elderly Medicare beneficiaries with cancer. Methods: We created a cohort of 175 elderly clinical trial patients treated on two Cancer and Leukemia Group B (CALGB) trials (i.e., 9,344 adjuvant breast and 9,730 advanced lung cancer trials) and merged participants’ CALGB data with their Centers for Medicare and Medicaid Services (CMS) data. From CALGB data, we identified all grade III/IV toxicities with a frequency of ≥3%. We reviewed diagnostic and procedure codes from CMS coding manuals, developed initial algorithms to measure the toxicities and then finalized the algorithms after empiric review of individual patients’ actual CMS codes incurred during the observation period (i.e., date of first trial treatment through 90 days following last trial treatment). We compared results of each of our CMS toxicity algorithms to gold-standard of CALGB grade III/IV toxicity information in order to calculate the CMS algorithms’ test characteristics. Results: The following 15 grade III /IV chemotherapy-related toxicities occurred in ≥3% of the 175 patients: white blood cell, hemoglobin, platelets, anorexia, nausea, vomiting, diarrhea, stomatitis, sensory neuropathy, motor neuropathy, motor or sensory neuropathy, dyspnea, hyperglycemia, infection, and malaise. Of these, only the CMS-based algorithm measuring ‘grade III/IV vomiting‘ had a sensitivity, specificity, and area under the ROC of ≥ 80%. Conclusions: The results of this preliminary study suggest that CMS claims data may be of only limited value in measuring clinically significant chemotherapy- related toxicities in elderly Medicare beneficiaries with cancer. Future research will focus on confirming these findings in a larger and more diverse patient sample.

No significant financial relationships to disclose.

Could treatment with tamoxifen be superior to aromatase inhibitors in early-stage breast cancer after pharmacogenomic testing? A modeling analysis

502

Background: Adjuvant endocrine treatment with upfront aromatase inhibitors yields improved disease-free survival compared to tamoxifen in unselected women with estrogen receptor-positive breast cancer. However, levels of endoxifen, the active tamoxifen metabolite, are known to vary with the number of mutant alleles of the cytochrome P450 CYP2D6 enzyme. We created a Markov model to examine whether the optimal treatment strategy for patients with wild type genetics (wt/wt) might differ from that for patients with homozygous (*4/*4) or heterozygous (wt/*4) mutations. Methods: Annual recurrence risks with aromatase inhibitors and tamoxifen in unselected women were taken from the BIG 1–98 trial. Based on data from Goetz et al., we assumed that the percent of patients who are homozygous and heterozygous for mutations are 6.8% and 21.1%, respectively, and that the hazard ratio (HR) for increased cancer recurrence on tamoxifen among *4/*4 carriers is 1.86 relative to patients with wt/wt or wt/*4 genotypes. Because the efficacy of tamoxifen among wt/*4 patients is not known, we tested the full possible range, from efficacy same as wt/wt patients (Ehet=0) to that of *4/*4 mutation carriers (Ehet=1). Results: In the unselected group, 5-year disease-free survival (5-DFS) with aromatase inhibitors and tamoxifen was 0.840 and 0.813, respectively. Under baseline assumptions, tamoxifen is more effective than aromatase inhibitors among wt/wt patients as long as the effect in heterozygotes is at least 54% (Ehet=0.54) of that in *4/*4 patients. With increasing HR for *4/*4 patients, tamoxifen estimates exceed those of aromatase inhibitors in the wt/wt cohort even at lower assumed Ehet ratios (see table ). Conclusion: In patients without CYP2D6 mutations, modeling suggests that initial treatment with tamoxifen could be superior to treatment with aromatase inhibitors, supporting the use of genetic testing to determine the optimal treatment strategy.

[Table: see text]

No significant financial relationships to disclose.

Reliability of data collection for the National Comprehensive Cancer Network (NCCN) non-Hodgkin lymphoma (NHL) multi-center outcomes database: A methodological approach

6525

Background: Clinical trials and outcomes studies often rely on non-physicians to abstract complex data from medical records. We assessed the reliability of chart abstraction among personnel groups in a multi-center outcomes study of indolent/aggressive NHL treated in NCCN centers. Methods: We developed 20 standardized charts of patients with newly-diagnosed NHL. Raters included 6 Clinical Research Associates from participating sites (CRAs), 3 project staff who conduct CRA training, and 3 medical oncologists. Raters each received a set of standardized charts, detailed instructions and training on a sample chart and abstracted all charts independently. We assessed reliability on 5 variables: MD-reported and rater-determined disease stage; International Prognostic Index (IPI- low-low intermediate, intermediate-high, high); Charlson comorbidity index score; and presence of any item from the Charlson index. We used intraclass correlation coefficients (ICCs) to calculate reliability. We considered coefficients from 0–0.20 ‘slight’, 0.21–0.40 ‘fair’, 0.41–0.60 ‘moderate’, 0.61–0.80 ‘substantial’ and >0.80 ‘almost perfect’(1). Results: Overall reliability was “almost perfect/substantial” for MD-reported stage, rater-determined stage, and IPI, but only “moderate” for the 2 Charlson-based comorbidity measures (Table). Reliability varied by rater group; no rater group was consistently more reliable than others. Conclusions: Trained CRAs abstracted key clinical variables with a very high degree of reliability, and performed at a level similar to study trainers and oncologists. Elements of the Charlson index were less reliable than other data types, possibly due to inherent ambiguity in the index itself. Use of trained CRA staff is reasonable to collect stage and IPI scores in a multi-center outcomes study, however abstracted Charlson scores should be interpreted with caution. (1)Biometrics. 1977. 33:159–74.

No significant financial relationships to disclose.

[Table: see text]

Chemotherapy use for hormone receptor-positive, node-negative breast cancer

11004

Background: High quality evidence has shown that adjuvant chemotherapy (chemo) improves survival for hormone receptor- positive, node negative breast cancer measuring >1 cm (H+/N-/T>1cm BC). Since 2002 National Comprehensive Cancer Network (NCCN) guidelines have recommended chemo for this group. Because the benefit from chemo in this situation is modest there is debate regarding which women, if any, can forego chemo. We describe trends in chemo use and identify factors associated with not receiving chemo for H+/N-/T>1cm BC. Methods: We identified women less than 70 years old with H+/N-/T>1cm BC diagnosed 1997–2004 and treated at 8 NCCN institutions. We analyzed the frequency of chemo use on a yearly basis (chi square). A multivariable logistic regression model assessed the relationship between likelihood of not receiving chemo and year of diagnosis, institution, tumor size, histologic characteristics and socio- demographic variables. We incorporated interaction terms to explore how chemo use varied over time and between institutions. Results: Of 3190 women with H+/N-/T>1cm BC, 56% received chemo. Chemo use was less common for 1.1–2 cm than for >2 cm tumors (48 vs. 83%; p<.01). There was no significant change in chemo use over time, but there was significant variability among institutions (range 47–65%; p<.01). On multivariable analysis, predictors independently associated with not receiving chemo included lower histologic/nuclear grade, non-ductal/lobular histology, absence of lymphovascular invasion or HER2 overexpression, and older age (p<.01 for all). The interaction between institution and year of diagnosis was also significant (p<.01). Over the 2 years following the NCCN guideline’s definitive recommendation for chemo in 2002, the relative change in institutional chemo use ranged from a 25% decrease to a 28% increase. Conclusions: Many women did not receive chemo despite clinical trial evidence. Providers considered molecular features, tumor size and patient age when deciding who should receive chemo. While overall trends in chemo use did not change, use appeared to increase at some institutions and decreased at others after the 2002 guideline recommendation for chemo. This suggests a lack of consensus at NCCN institutions regarding when to administer chemo for H+/N-/T>1cm BC.

No significant financial relationships to disclose.

Preferences for and expectations about experimental therapy among participants in randomized trials

6013

Background: Equipoise is often viewed as the ethical precondition for randomized controlled trials (RCTs). Equipoise is usually defined as a state of uncertainty or disagreement among expert clinicians, but some ethicists argue instead that individual participants should be indifferent between the study arms. The degree to which cancer RCT participants have preferences for either arm is not known. Methods: Adults who had enrolled in a 2-arm RCT at a Dana-Farber/Harvard Cancer Center institution within 21 days were eligible. A mailed questionnaire asked about treatment assignment, treatment preference, and expectations about the relative efficacy and toxicity of the 2 arms. The primary analysis was a goodness-of-fit chi-square test of the null hypothesis that equal proportions of respondents would prefer the standard vs. investigational arm. Results: 82 of 126 RCT participants (65%) responded. 31 (38%) were assigned to the standard arm, 26 (32%) were assigned to the experimental arm, and 25 (30%) were enrolled in blinded trials. Respondents tended to prefer assignment to the experimental rather than the standard arm (p = 0.006), and expected the investigational arm to be more effective than the standard arm (p = 0.0004). Respondents did not consistently expect that either arm would be more toxic ( Table ). Conclusions: Cancer RCT participants tend to prefer the experimental arm. Thus either the design of cancer RCTs must be altered to accommodate participant preference, or individual-participant equipoise must be abandoned.

[Table: see text]

No significant financial relationships to disclose.

Chemotherapy-related amenorrhea in women with early breast cancer: The effect of paclitaxel or dose density

10506

Background: The effect of the addition of paclitaxel (T) to adriamycin and cyclophosphamide (AC), or dose density (DD, q2wk) on menstrual functioning in women with early breast cancer remains uncertain. Methods: We reviewed data from premenopausal women treated at Dana-Farber Cancer Institute and Brigham and Women’s Hospital who received adjuvant chemotherapy with either AC or AC-T from 1997–2005 for whom pre- and post-treatment menstrual data were available. A multivariable logistic regression model evaluating amenorrhea at 6 months or greater follow-up was conducted. Results: Of the 403 patients eligible for analysis, average age at diagnosis was 41 years (range 25–50); average weight was 67kg; 10% were smokers, 53% had never smoked; 24% were nulliparous. 60% of women received AC, 40% AC-T; 20% of women received DD therapy (either AC or AC-T); 72% received tamoxifen. The multivariable model adjusting for age, weight, gravity, parity, smoking, alcohol use, and tamoxifen use, revealed that risk of amenorrhea was not statistically different in patients receiving AC-T versus AC (OR = 1.45, 95% CI 0.78–2.69) or in patients receiving DD therapy versus standard (OR = 1.61, 95% CI 0.74–3.49). However, a planned future analysis including data from an additional 200+ patients may yield significant differences. Age directly correlated with risk of amenorrhea, OR = 1.40, 95% CI 1.30–1.50) for every year increase in age, and women receiving tamoxifen were more likely to be amenorrheic (OR = 1.99, 95% CI 1.05–3.79). Conclusions: The addition of T to AC or use of DD therapy does not appear to increase the likelihood of amenorrhea in follow-up. These findings may assist premenopausal women with breast cancer when making decisions at diagnosis and in follow-up.

No significant financial relationships to disclose.

Impact of patient and physician factors on oncologists’ recommendations for adjuvant chemotherapy in stage III colon cancer

6051

Background: Population-based studies indicate that elderly patients with colon cancer are less likely to receive adjuvant therapy (tx). Lack of tx may reflect patient preference, appropriate consideration of comorbid illness, or physician bias. Methods: To study physician preference for tx of elderly patients with colon cancer, a vignette-based survey was developed then mailed to a nationally representative sample of 1,000 oncologists. Patient age (61/72/83 yrs), comorbidity level (none/mild/severe CHF with symptoms on minimal exertion) and preference were varied across 8 vignettes. Physician preference for recommending tx was measured using a 7-pt Likert scale; mixed effects linear regression was used to evaluate the results. Results: 485 oncologists returned the survey (RR = 49%); 363 that had seen patients with colon cancer during the previous year were included in further analyses. Median age of the respondents was 52 (range 30–77), 73 (20%) were women. Median time since graduation from medical school was 25 yrs (range 5–53); 17% were employed in academic centers, 8% in HMOs and the remainder in the community. Patient age and comorbid illness interacted to significantly influence physician recommendations. Among patients with mild comorbidity, physician preference for tx of an 83 yr old patient was 1.1 lower than for a 72 yr old (3.8 vs 4.9 on a 7-pt Likert scale) and 1.7 lower than for a 61 yr old (each p<0.0001). This age effect was similar among patients with no comorbidity (p=0.30), with physicians’ preferences consistently higher by about 1.5 than for patients with mild comorbidity (p<0.0001). Among patients with severe comorbidity, preference for tx of an 83 yr old was only 0.9 lower than for a 61 yr old patient (p<0.0001), whereas the decrease was 2.6 for a 61 yr old with severe vs mild comorbidity. Patient preference against tx resulted in 0.3 (p<.0001) decrease in physicians’ recommendations for tx. Among physician factors, only type of employment was associated with tx recommendations with a 0.3 decrease in preference for tx (p=0.0014) among academic oncologists. Conclusions: Patient rather than physician factors have the greatest effect on oncologists’ preferences for adjuvant tx with age and severe comorbidity having the strongest effect.

No significant financial relationships to disclose.

Direct-to-consumer advertising in oncology: A content analysis of print media

6034

Background: Direct-to-consumer advertising (DTCA) is an increasingly prevalent and contentious medium of medical communication. Content analysis of cancer-related DTCA, with a focus on how benefit and risk/side effect information is presented, is essential to understanding its potential effect on oncology outcomes. Methods: We reviewed all product-specific DTCA for oncology-related medications appearing in the patient-focused cancer magazines CURE, Coping with Cancer, and MAMM: Women, Cancer and Community during 2005. We assessed the Flesch reading ease score (FRES), calculated with the formula [206.835 - (1.015 x average sentence length) - (84.6 x average syllables per word)], for the benefit and risk/side effect information presented in each advertisement. Ranging from 0 to 100, a FRES of 65 or higher is considered understandable for the average person. We also assessed the largest type size (in mm) and the placement (first, middle or final third of text) of benefits and risks/side effects, as well as the ads’ use of appeals to clinical trial data, cost, celebrity endorsement, physician endorsement, and implied patient testimonials (visual and written). Results: Of 75 ads reviewed, many were repeated, such that 15 unique ads were analyzed. The mean FRES for benefit information was 40.64, while the mean FRES for risks/side effects was 32.30, a difference of 8.34 [95% CI: 1.16, 15.52]. The mean largest benefit type size was 7.10 mm, while the mean largest risk/side effect type size was 2.33 mm, a difference of 4.77 mm [95% CI: 2.35, 7.17]. No ads mentioned cost, and none contained a celebrity or physician endorsement. 53% referenced clinical trials, and 60% offered an implied patient testimonial. In their first third of text, 93% presented benefits; 13% presented risks/side effects. In their final third of text, 27% presented benefits; 93% presented risks/side effects. Conclusion: Oncology print DTCA is difficult to read, as assessed by a standard readability measure. Moreover, compared to risks/side effects, benefits are presented in a larger typeface, earlier in the text, and in language that is modestly easier to read. Oncology providers should be aware of such differences, as they may influence patients’ perceptions of cancer-related medications.

No significant financial relationships to disclose.

Understanding of prognosis among parents of children with cancer: Parental optimism and the role of the parent-physician interaction

6033

Background: Patients often overestimate their chances of surviving cancer. Factors that contribute to accurate understanding of prognosis among parents of children with cancer are not known. Methods: We conducted a cross-sectional survey of 194 parents of children with cancer (overall response rate 70%), treated at the Dana-Farber Cancer Institute and Children’s Hospital, Boston, Mass, and the children’s physicians. Our main outcome measure was agreement between parent and physician ratings of the child’s likelihood of cure. Results: The majority of parents (61%) were more optimistic about the likelihood of their child’s cure than their child’s physician was. Most parents, however, believed that their opinions about prognosis either matched (70%) or were more pessimistic (26%) than those of their physicians. When physicians were confident in their knowledge of the child’s prognosis, parent and physician perceptions of prognosis were more likely to agree (OR 2.55, P = .004). Parents whose role in decision-making matched their ideal role were more likely to give prognostic estimates that agreed with physician perceptions of prognosis (OR 1.89, P = .019). In contrast, parent confidence in knowledge (OR .07, P < .0001), coping strategies (disengagement, OR .31, P = .007; reliance on emotional support from others, OR .31, P = .029), and use of intuition to understand prognosis (OR .51, P = .012), were associated with overestimation of likelihood of cure. Conclusions: Many parents overestimate their children’s chances of being cured of cancer. Neither physician nor parent attributes alone account for this finding; rather, successful communication about prognosis requires that physicians relate to parents’ individual communication and decision-making needs.

No significant financial relationships to disclose.

Effect of distance to radiation treatment facility on use of radiation therapy after mastectomy in elderly women

6030

Background: Post-mastectomy radiation therapy (PMRT) has been shown to improve survival in patients at high risk for local or regional recurrence. However, radiation therapy necessitates daily travel to treatment centers for several weeks. We sought to study the effect of distance to the nearest radiation treatment facility on receipt of PMRT in elderly women. Methods: Using the linked Surveillance, Epidemiology, and End Results-Medicare (SEER-Medicare) database, we analyzed 19,787 women with Stage I or II breast cancer who received mastectomy as definitive surgery between 1991 and 1999. Multivariable logistic regression was used to investigate the association of travel distance with receipt of PMRT after adjusting for clinical and sociodemographic factors. Results: Overall, 2,075 (10.5%) patients treated with mastectomy received PMRT. In addition to cancer and patient characteristics, increasing distance from the nearest radiation treatment facility was independently associated with a decreased likelihood of receiving PMRT, (OR = 0.996 per additional mile, p = 0.01). The decline in PMRT use appeared at distances greater than 25 miles and was statistically significant for those patients living greater than 75 miles from the nearest radiation facility (odds of receiving PMRT of 0.58 [95% CI: 0.34–0.99] versus those living within 25 miles of such a facility). In secondary analyses, the effect of distance was only significant in women aged 75–80 years (OR = 0.992 per additional mile, p = 0.03), and those above the age of 80 (OR = 0.989, p = 0.02). When analyses were conducted separately by geographic region, the effect of distance was only significant in the Midwest (OR = 0.992, p = 0.014). The effect of distance was not significant (OR = 1.00, p = 0.87) among women with positive nodes, but was significant in women with no positive nodes (OR = 0.992, p = 0.013). Conclusions: Oncologists must be cognizant of the potential barrier to quality care that is posed by travel distance, especially for elderly patients. Mechanisms to ameliorate the effect of distance on receipt of radiation therapy by assisting individuals with transportation limitations or policies to decrease the centralization of RT services, may help to remove barriers to potentially life-saving treatment.

No significant financial relationships to disclose.

Factors associated with fatigue in children with cancer at the end of life

8574

Background: Fatigue is prevalent in children with cancer but is poorly understood. Improved understanding of this symptom will inform development of interventions aimed at alleviating it. Methods: Cross-sectional survey conducted between 1997–2001 of 144 parents of children who died of cancer cared for at Dana-Farber Cancer Institute/Children’s Hospital Boston and Children’s Hospital and Clinics, St Paul and Minneapolis (response rate 65%). Parents reported the child’s experience of symptoms in the last month of life and associated suffering, and patient characteristics. Determinants of suffering from, and treatment of fatigue were explored with univariate analyses adjusting for physician and time since death. A logistic regression model was built to determine factors associated with suffering from fatigue, with a threshold for entry of P<.05, and accounting for physician clustering. Results: Almost all children experienced fatigue and 50% suffered significantly from it (a great deal/a lot versus some/a little/none at all). Only 15% received treatment, and of them, 79% were not successfully treated. Univariate analyses revealed significant associations between suffering from fatigue and suffering from pain, dyspnea, anorexia, diarrhea, and nausea/vomiting, as well sadness, anemia and successful treatment of pain (P<.05). Treatment of fatigue was associated with side effects from treatment of pain (OR 3.9, P=.008) and treatment of dyspnea (OR 3.9, P=.02). In the multivariate analysis of suffering from fatigue, successful treatment of pain remained significant (OR 4.2, P=.009). Conclusions: Suffering from fatigue is common in children with cancer at the end of life and efforts to palliate it are limited. Significant fatigue is highly associated with other symptoms and their treatments. Increased attention to treatment-related fatigue and evaluation of interventions such as stimulants may be effective in ameliorating fatigue in children with advanced cancer.

No significant financial relationships to disclose.

What do patients value in their hospital care? An empirical perspective on autonomy centred bioethics

OBJECTIVE

Contemporary ethical accounts of the patient-provider relationship emphasise respect for patient autonomy and shared decision making. We sought to examine the relative influence of involvement in decisions, confidence and trust in providers, and treatment with respect and dignity on patients' evaluations of their hospital care.

DESIGN

Cross-sectional survey.

SETTING

Fifty one hospitals in Massachusetts.

PARTICIPANTS

Stratified random sample of adults (N=27 414) discharged from a medical, surgical, or maternity hospitalisation between January and March, 1998. Twelve thousand six hundred and eighty survey recipients responded.

MAIN OUTCOME MEASURE

Respondent would definitely be willing to recommend the hospital to family and friends.

RESULTS

In a logistic regression analysis, treatment with respect and dignity (odds ratio (OR) 3.4, 99% confidence interval (CI) 2.8 to 4.2) and confidence and trust in providers (OR 2.5, CI 2.1 to 3.0) were more strongly associated with willingness to recommend than having enough involvement in decisions (OR 1.4, CI 1.1 to 1.6). Courtesy and availability of staff (OR 2.5, CI 2.1 to 3.1), continuity and transition (OR 1.9, CI 1.5 to 2.2), attention to physical comfort (OR 1.8, CI 1.5 to 2.2), and coordination of care (OR 1.5, CI 1.3 to 1.8) were also significantly associated with willingness to recommend.

CONCLUSIONS

Confidence and trust in providers and treatment with respect and dignity are more closely associated with patients' overall evaluations of their hospitals than adequate involvement in decisions. These findings challenge a narrow emphasis on patient autonomy and shared decision making, while arguing for increased attention to trust and respect in ethical models of health care.

Cell culture of mechanoreceptor neurons innervating proleg sensory hairs in Manduca sexta larvae, and co-culture with target motoneurons

The tip of each proleg in Manduca sexta larvae bears a dense array of mechanosensory hairs termed planta hairs (PHs), each innervated by a single sensory neuron (termed a PH-SN) located in the underlying epidermis. In the CNS, axon terminals of PH-SNs make direct, excitatory, nicotinic cholinergic synapses with proleg retractor motoneurons including the accessory planta retractor (APR). These synapses mediate a proleg withdrawal reflex, exhibit multiple forms of activity-dependent plasticity and weaken during the prepupal peak of ecdysteroids. In the present study we developed methods to dissociate PH-SNs from the epidermis and culture them alone or with APRs. The PH-SNs were fluorescently labeled in situ by introducing dye through the cut hair shaft or by retrograde axonal staining. Alternatively, unlabeled PH-SNs were utilized. The epidermis beneath the planta hair array was separated from the cuticle, enzymatically treated and mechanically dissociated into single cells. PH-SNs were cultured on glass coverslips coated with concanavalin A and laminin, in modified Leibovitz's IL-15 medium. Supplementation with medium conditioned by an insect cell line produced the best results. Dissociated PH-SNs had somatic diameters of ~10 micro m and typically bore a stout dendrite consisting of the inner and, occasionally, the outer dendritic segment. An axonal stump was sometimes retained. Viable PH-SNs typically extended new processes and often survived for 2-4 weeks. When co-cultured, PH-SNs and APRs exhibited robust growth and made close anatomical appositions. This culture system provides convenient experimental access to PH-SNs and may potentially permit sensorimotor synapses to be investigated in vitro.