Assessing the Benefit of Radiation Therapy After Breast-Conserving Surgery for Ductal Carcinoma-In-Situ
http://www.100md.com
《临床肿瘤学》
the Departments of Radiation Oncology and Internal Medicine, and Division of General Medicine and Biostatistics, University of Michigan Health System, Ann Arbor
Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, MI
Division of Emergency Medicine, A.I. duPont Hospital for Children, Wilmington, DE
ABSTRACT
PURPOSE: To assess women's preferences regarding the trade-off between the risks and benefits of treatment with radiation therapy (RT) after breast-conserving surgery (BCS) for ductal carcinoma-in-situ (DCIS).
PATIENTS AND METHODS: Utilities were obtained from 120 patients and 210 nonpatients for eight relevant health states using standard gambles.
RESULTS: Differences in utilities obtained from patient and nonpatient participants between health states were relatively similar. Reduction in the likelihood of local recurrence associated with RT did not result in higher utilities. Utilities for noninvasive recurrence were only lower after initial treatment with RT. Patient and nonpatient participants had the lowest utilities for invasive local recurrence, regardless of initial treatment or manner of salvage therapy. When comparing patient and nonpatient utilities directly, patients had higher utility for being without recurrence after initial RT and lower utility for invasive recurrence salvaged by mastectomy after initial BCS alone. None of the clinical or sociodemographic factors examined explained more than 5% of the variability in the patients' or nonpatients' utilities or their differences.
CONCLUSION: The principal benefit associated with adding RT to BCS for DCIS seems to be its ability to reduce invasive recurrences.
INTRODUCTION
As the number of women diagnosed with ductal carcinoma-in-situ (DCIS) continues to grow,1 so has interest in how best to care for them. Although mastectomy remains an effective treatment, many women opt for breast-conserving therapy, which typically consists of lumpectomy followed by adjuvant radiation therapy (RT).2-4
As the use of RT has increased, questions about the need for RT have also increased.5 Results from the three randomized trials that address this issue are remarkably similar.6-8 Adding RT reduces the relative risk of both invasive and noninvasive local recurrences by 35% to 65%, but the absolute reduction in each type of recurrence is only in the range of 5% to 10%, and there is no evidence of a survival benefit.
The benefit associated with RT in this setting depends on the trade-off between the fear and consequences of both noninvasive and invasive local recurrence versus the inconvenience and potential toxicity of treatment. Because this trade-off revolves around the preferences women have for different outcomes, utilities are an appropriate means of assessing the potential value of RT in this setting. Utilities are global measures of quality of life that represent the strength of one's preference for a particular state of health and are measured on a scale ranging from 0 to 1, where 0 is equivalent to death and 1 is equivalent to optimal health.9 By comparing utilities for the range of probable outcomes after treatment with and without RT, it is possible to assess its incremental impact on quality of life. Utilities can also be used to calculate quality-adjusted life-years, which is the preferred measure of benefit in cost-effectiveness analyses.10
The issue of whose utilities should be measured is currently a matter of some debate. In older studies, utilities were commonly elicited from physicians and/or nurses.11 Concerns about their ability to accurately determine patients' preferences led many to argue that utilities should be elicited from patients. However, more recently, because cost-effectiveness analyses are primarily intended to inform decisions regarding the allocation of societal resources, it has been recommended that utilities be obtained from the general public.12
We previously collected patients' utilities to assess the benefit of RT after lumpectomy for invasive breast cancer and found that patients valued RT for its ability to decrease their fear of recurrence and the likelihood of developing an invasive recurrence requiring mastectomy.13 We hypothesized that similar benefits might also exist for DCIS. Accordingly, the goal of this study was to measure both patients' and nonpatients' utilities for the relevant health states after treatment of DCIS with lumpectomy, both with and without RT, to assess the impact of RT in this setting.
PATIENTS AND METHODS
Eligibility Criteria
Patient participants. Potential patient participants were identified from lists of women scheduled for routine follow-up in one of five radiation oncology departments in Southeast Michigan. To be eligible, patients had to be women with a history of DCIS who had undergone lumpectomy followed by RT, who had completed RT at least 1 month before their visit and had no history of recurrence or contralateral breast cancer.
Nonpatient participants. Potential nonpatient participants were identified from lists of women scheduled for a routine health maintenance visit in one of several general internal medicine clinics within the University of Michigan Health System. To be eligible, the women had to have no history of breast cancer.
Enrollment
After obtaining permission from their radiation oncologists, potential participants were sent a letter describing the study along with an opt-out card. Permission was not obtained before sending nonpatient participants similar letters. Potential subjects who returned the opt-out card were not contacted. Subjects who did not return the card were contacted by phone, and the study was described again. If a subject was not interested in participating, she was not contacted again. Interested subjects were interviewed after obtaining informed written consent, which generally occurred after the subjects saw their physician. Screening of potential participants ended when our accrual goals (see Statistical Methods) were met. This study was reviewed and approved by the University of Michigan Institutional Review Board and by the Institutional Review Boards at the other four participating centers.
Patient participants. Two hundred ninety-three potential patient participants were identified. At the request of their radiation oncologists, 26 women (8%) were not sent letters. Ninety-one women (31%) chose not to participate in the study after receiving the letter, and 56 (19%) could not be interviewed because of scheduling difficulties. One hundred twenty-four patients were successfully interviewed, but four were later found to be ineligible. Accordingly, the final patient sample consisted of 120 patients (42%).
Nonpatient participants. Eight hundred fifty-seven women were sent letters. Two hundred forty-four women (28%) actively chose not to participate in the study after receiving the letter, and 24 (3%) were found to be ineligible. Another 310 women (36%) could not be reached by telephone before their appointment, and 69 (8%) could not be interviewed because of scheduling difficulties, resulting in a final sample of 210 nonpatient participants (25%).
Interview
Participants were only interviewed once in the presence of a trained interviewer. The interview was based on approaches developed by Furlong et al14 and Sumner et al15 and was performed on a laptop using U-Titer (Washington University, St Louis, MO), a Macintosh-based program developed specifically for utility assessment.16 The 45-minute interview consisted of the following three parts (listed in order of occurrence): a ranking exercise, standard gambles, and a questionnaire.
After an in-depth review of the issues regarding the diagnosis and treatment of DCIS, participants were asked to read, in random order, eight hypothetical health states and to rank them from most to least desirable. The health states represented the range of possible local outcomes after treatment with breast-conserving surgery (BCS) with or without RT. These scenarios were based on those used previously.13 Briefly, the eight health states were as follows: (A) BCS followed by RT without recurrence; (B) a noninvasive recurrence salvaged with mastectomy after initial treatment with BCS and RT; (C) an invasive recurrence salvaged with mastectomy after initial treatment with BCS and RT; (D) BCS alone without recurrence; (E) a noninvasive recurrence salvaged with BCS and RT after initial treatment with BCS alone; (F) a noninvasive recurrence salvaged with mastectomy after initial treatment with BCS alone; (G) an invasive recurrence salvaged with BCS and RT after initial treatment with BCS alone; and (H) an invasive recurrence salvaged with mastectomy after initial treatment with BCS alone. Where relevant, it was emphasized that treatment with adjuvant RT has been shown to decrease the risk of DCIS and invasive recurrences but not to increase survival. The recurrence rates cited were based on data from the National Surgical Adjuvant Breast and Bowel Project B-17 trial available at the initiation of this study (8% and 4% and 13% and 13% for noninvasive and invasive recurrence with and without RT, respectively).17 Where relevant, participants were also informed that RT has not been proven to improve survival, commonly causes temporary acute side effects (eg, fatigue and skin irritation) and minor chronic effects (eg, changes in the appearance of the breast), and rarely can also cause more serious long-term complications (eg, pneumonitis, cardiac toxicity, and RT-associated malignancy). Participants were also informed that an invasive recurrence would lead to the need for axillary staging and that the likelihood of cure after salvage therapy for either a DCIS or invasive recurrence is high.
A variation of the classic standard gamble technique for measuring utilities was used during the second part of the interview. In the classic standard gamble, the participant is asked to choose between the following two alternatives: choice A is a gamble between optimal health and immediate death; and choice B is certain life in an intermediate health state. In an iterative fashion, the probabilities of optimal health and immediate death in choice A are varied systematically, and the participant is asked repeatedly to choose between the two alternatives. The process continues until the participant is indifferent; at that point, the probability of optimal health in choice A is the respondent's utility for the intermediate health state.18 To avoid having to constantly confront the participant with death as the worst possible outcome and to improve precision, we used a modification known as partial cascading.14 Using the rankings, the seven highest ranked health states were first assessed in seven standard gambles, using optimal health as the best possible outcome and the lowest ranked health state as the worst possible outcome. These seven standard gambles were conducted in random order. The lowest ranked health state was then assessed relative to optimal health and death. Using the results of these eight standard gambles, utilities for the eight intermediate health states could then be derived.
The questionnaire differed slightly for patients and nonpatients. Both questionnaires included items dealing with participants' age, marital status, number of dependents at home, race, education, and income. Patients were then asked whether they had taken tamoxifen, experienced RT toxicity, were satisfied with their cosmetic outcome, were fearful of recurrence, and were satisfied with their choice of breast-conserving therapy. Nonpatients were asked additional questions that allowed us to calculate their 5-year and lifetime breast cancer risk using the Gail model.19
Statistical Methods
The expected standard deviations of standard gamble utilities obtained from experienced patient participants and nonexperienced participants are 0.2 and 0.3, respectively.14 Therefore, a sample size of 120 patient participants was chosen so that the mean utilities for the eight health states could at least be estimated with a 95% CI of ± 0.04. A sample size of 210 nonpatient participants was chosen to achieve similar CIs. Differences in baseline characteristics between nonparticipants and participants from both patient and nonpatient groups and between participating patients and nonpatients were evaluated using t tests and Pearson 2 tests where appropriate. Because the utility data were not normally distributed (see Results), nonparametric statistical procedures were used.20 The Friedman rank sum test was first used to test whether participants' utilities for the eight health states within each of the two groups differed from one another overall. The Wilcoxon-Nemenyi-McDonald-Thompson procedure was then used to perform all pairwise comparisons, while adjusting for multiple comparisons. The Wilcoxon rank sum test was used to determine whether the patients' and nonpatients' utilities for the eight health states differed between the two groups. Finally, stepwise regression was used to examine whether any of the baseline covariates were significant predictors of participants' utilities for the health states themselves or the differences in utility between any two health states. Because of the skewed distribution of the utility scores, the logit transformation was used to make the data approximately normal before performing the regression analysis. Covariates examined included the following: age, race, marital status, number of dependents, education, and income for all participants; months since completion of RT, toxicity, treatment with tamoxifen, satisfaction with local outcome, fear of recurrence, and satisfaction with treatment decision for patient participants; and 5-year and lifetime risks of breast cancer for nonpatient participants.
RESULTS
Study Population Characteristics
The characteristics of the 120 DCIS patients and 240 nonpatient participants who were interviewed are listed in Tables 1 and 2. Nonpatients, compared with patient participants, were younger (mean age, 50 v 61 years, respectively; P < .001), more likely not to be white (16% v 7%, respectively; P = .03), less likely to have dependents at home (mean, 0.65 v 1.1 dependent, respectively; P < .001), and better educated (71% v 50% college graduates, respectively; P < .001). We found no evidence that our nonpatient participants were at greater risk of developing breast cancer than women in the general population. As calculated by the Gail model, the nonpatients' mean 5-year and lifetime risks of being diagnosed with breast cancer were 1.3% and 10.2%, respectively, compared with 1.2% and 10.1% for all women, respectively.
Interviews
All enrolled participants successfully completed the interview. The mean time ± standard deviation to complete the interview for patient and nonpatient participants was 42 ± 13 minutes and 43 ± 11 minutes, respectively.
Standard Gamble Utilities
Descriptive statistics. Patients' and nonpatients' standard gamble utilities are listed in Tables 3 and 4, respectively. For both groups, the median utilities for all health states are higher than the means, reflecting the fact that their distributions are skewed upward. In addition, as expected, there is considerable variability in the utilities, which are greatest for the recurrent invasive health states.
Within-group comparisons. The results of the pairwise comparisons between health states within the patient and nonpatient groups are listed in Table 5. Overall, the results are surprisingly similar. For the health states after treatment with RT, both groups had the strongest preference for being without recurrence, followed first by a DCIS recurrence and then by an invasive recurrence (ie, A > B > C). For the health states after treatment with lumpectomy alone, patient and nonpatient participants alike were indifferent to a DCIS recurrence but not to an invasive recurrence (ie, D = E = F > G and H). Both groups had similar utilities for the nonrecurrent health states, whether they had or had not received initial treatment with RT (ie, A = D). Patients and nonpatients also had similar utilities for a DCIS recurrence requiring mastectomy whether or not they had received upfront RT (ie, B = F). In contrast, patients and nonpatients not only had different utilities for an invasive recurrence requiring mastectomy after initial treatment with RT compared with no RT (ie, C v H), but the direction of the difference was also reversed. Patient participants preferred having received RT (ie, C > H), whereas nonpatient participants preferred not to have had received RT (ie, C < H). The only other relevant difference between the two groups was that patient participants preferred to preserve their breast after an invasive recurrence after initial treatment with BCS alone (ie, G > H), whereas nonpatient participants were indifferent (ie, G = H). Neither group expressed a preference for breast preservation when confronted with a DCIS recurrence after initial treatment with BCS alone (ie, E = F).
Between-group comparisons. Overall, there were few differences that were statistically significant when comparing patient and nonpatient participants' utilities directly (Table 6). Patients had higher utilities than nonpatients for being without recurrence after BCS and RT (A) and lower utilities for having an invasive recurrence salvaged by mastectomy after BCS alone (H). Differences in utilities for being without recurrence after BCS alone (D) and for having had an invasive recurrence salvaged by mastectomy after BCS and RT (C) were of borderline significance (P = .09 and P = .06, respectively).
Factors associated with participants' utilities. Although several of the clinical and sociodemographic factors examined were statistically associated with patient and nonpatient participants' utilities, none of the factors explained more than 5% of the variability in the utilities themselves or in the differences between health states.
DISCUSSION
Our results suggest that the principal benefit associated with adding RT to BCS for the treatment of DCIS is the ability of RT to reduce actual invasive recurrences, which is an event perceived both by patients and nonpatients as having a significant negative impact. The potential quality-of-life benefit associated with a reduction in the risk of recurrence after treatment with RT seems to be negated in both groups by concerns regarding the inconvenience and potential toxicities of treatment. In addition, compared with nonpatients' predictions, patients perceive their current quality of life (ie, disease free after lumpectomy and RT) to be higher and predict that it would decrease to a lower level after an invasive recurrence salvaged by mastectomy. Finally, both patient and nonpatient participants have considerable variation in their preferences that are not explained by the sociodemographic and/or clinical factors examined.
On the basis of our previous experience with invasive breast cancer,13 we expected women to prefer the nonrecurrent health state after initial treatment with RT (A), with its lower risk of recurrence, versus without RT (D; ie, expected A > D). However, in the present study, we found that the utilities for these two states were not significantly different, implying that women do not feel that the reduction in the fear of recurrence associated with 5% and 9% absolute decreases in the likelihood of noninvasive and invasive recurrences, respectively, outweighs the potential toxicity and inconvenience of upfront RT. A likely explanation for this finding is that the absolute reduction in the likelihood of an invasive recurrence is only 9% compared with 30% in our prior invasive study. It is also possible that a small difference exists between the two nonrecurrent health states but that we lacked the statistical power to detect it.
The negative impact associated with an invasive recurrence was one of the most consistent findings across both groups of participants. Given the potential need for axillary dissection and/or chemotherapy and the possibility of developing incurable metastatic disease after such an event, the finding that health states C, G, and H were the least preferred states comes as no surprise. It is also not surprising that, after initial treatment with adjuvant RT, both groups of women first favored no recurrence, followed by a DCIS recurrence and then an invasive recurrence. However, it was surprising that women do not seem to be bothered by a noninvasive recurrence without or with the loss of their breast after initial treatment with surgery alone (ie, D = E = F). Perhaps after gambling and losing, women are more accepting of a local recurrence, whether or not it results in loss of the breast, as long as it is not invasive. As for their feelings regarding the recurrent invasive health states, patient and nonpatient participants differed somewhat. Probably because of cognitive dissonance, patients previously treated with RT preferred having received initial treatment with RT if they went on to experience an invasive recurrence salvaged with mastectomy (ie, C > H). In contrast, nonpatients preferred not to have received upfront RT (ie, H > C), which is understandable given the inconvenience and potential toxicity of RT if the ultimate outcome was the same anyway. Lastly, patients preferred a second lumpectomy and RT versus mastectomy as salvage therapy of an invasive recurrence after initial treatment with lumpectomy alone (ie, G > H), whereas nonpatients were indifferent (ie, G = H). One potential explanation for this finding is that patients were more predisposed towards breast conservation given that they had all chosen lumpectomy when diagnosed with DCIS.
As is true in our study, patients often value their current state of health (A) more highly than nonpatients predict they would. Numerous theories exist regarding why this might be the case.21,22 The most compelling explanation in this case is that nonpatients often fail to account for their ability to adapt to adverse circumstances over time. As for patients' lower utility for an invasive recurrence salvaged with mastectomy after initial treatment without RT, the most likely explanation has to do with a difference in their vantage point.22 Given their prior decision to opt both for breast conservation and RT, it is not surprising that patients perceive an invasive recurrence resulting in the loss of their breast after foregoing RT more negatively than nonpatients.
As has been true in our previous studies,13,23 we found that participants vary considerably in their preferences for various outcomes and that these differences are not explained by the sociodemographic and/or clinical factors examined, implying once again that participants' feelings about these issues are complex.
Although a number of investigators have reported utilities collected from patients for local outcomes associated with the treatment of invasive breast cancer,11,24 to the best of our knowledge, we are the first to report utilities for DCIS. In general, after accounting for differences in health state descriptions, study populations, and utility elicitation techniques, our results are similar to those reported previously. Our results add to this literature by directly assessing preferences for outcomes associated with DCIS from both patients and nonpatients.
One potential limitation of our study is that the health state scenarios may have inadvertently biased participants for or against RT. Although the content of the scenarios was determined by experienced clinicians, decisions still had to be made regarding what information to include in the scenarios. When constructing the scenarios, a conscious effort was made to present the information in a fair and neutral fashion and, when faced with uncertainty, to bias the scenarios against RT. Although more recent results from the National Surgical Adjuvant Breast and Bowel Project B-17 trial have been reported, the absolute differences between the rates of noninvasive and invasive recurrences without and with RT have not changed significantly with longer follow-up.6 Given our participants' high education and income levels, our study can also be criticized on the grounds that our study populations are not representative. However, it should be noted that two thirds of our patient participants were recruited from community-based radiation oncology departments, and we found no evidence that our nonpatient participants were at greater risk of developing breast cancer than women in the general population. We also acknowledge that our decision to only include women treated with lumpectomy and RT among our patient participants may also have biased our results. However, we wanted experienced patient participants who were knowledgeable about both lumpectomy and RT. Finally, our decision to require permission from their physician to contact patients and our inability to schedule interviews with all interested participants may also have introduced bias.
In conclusion, it does not seem to be an increased risk of recurrence, mastectomy, or even a noninvasive recurrence, but rather an invasive recurrence, that women fear most after breast-conserving therapy for DCIS. This implies that invasive recurrence should probably be the primary focus when assessing the benefit of any treatment for DCIS.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all of the women who agreed to participate in this study and acknowledge the assistance of Elizabeth Oldread, Kerry Paisley, and Glenn Wright in collecting and helping to analyze the data presented in this article.
NOTES
Supported by Clinical Research Training Grant for Junior Faculty No. CRTG-99-248-01-CCE from the American Cancer Society.
Presented in part at the 37th Annual Meeting of the American Society of Clinical Oncology, San Francisco, CA, May 12-15, 2001.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
American Cancer Society: Cancer Facts and Figures 2003. Atlanta, GA, American Cancer Society, 2003
Ernster VL, Ballard-Barbash R, Barlow WE, et al: Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 94: 1546-1554, 2002
Innos K, Horn-Ross PL: Recent trends and racial/ethnic differences in the incidence and treatment of ductal carcinoma in situ of the breast in California women. Cancer 97: 1099-1106, 2003
Winchester DJ, Menck HR, Winchester DP: National treatment trends for ductal carcinoma in situ of the breast. Arch Surg 132: 660-665, 1997
Silverstein MJ, Lagios MD: Benefits of irradiation for DCIS: A Pyrrhic victory. Lancet 355: 510-511, 2000
Fisher B, Land S, Mamounas E, et al: Prevention of invasive breast cancer in women with ductal carcinoma in situ: An update of the national surgical adjuvant breast and bowel project experience. Semin Oncol 28: 400-418, 2001
Houghton J, George WD, Cuzick J, et al: Radiotherapy and tamoxifen in women with completely excised ductal carcinoma in situ of the breast in the UK, Australia, and New Zealand: Randomised controlled trial. Lancet 362: 95-102, 2003
Julien JP, Bijker N, Fentiman IS, et al: Radiotherapy in breast-conserving treatment for ductal carcinoma in situ: First results of the EORTC randomised phase III trial 10853—EORTC Breast Cancer Cooperative Group and EORTC Radiotherapy Group. Lancet 355: 528-533, 2000
Torrance GW: Measurement of health state utilities for economic appraisal. J Health Econ 5: 1-29, 1986
Weinstein MC, Siegel JE, Gold MR, et al: Recommendations of the Panel on Cost-effectiveness in Health and Medicine. JAMA 276: 1253-1258, 1996
Earle CC, Chapman RH, Baker CS, et al: Systematic overview of cost-utility assessments in oncology. J Clin Oncol 18: 3302-3317, 2000
Russell LB, Gold MR, Siegel JE, et al: The role of cost-effectiveness analysis in health and medicine. JAMA 276: 1172-1177, 1996
Hayman JA, Fairclough DL, Harris JR, et al: Patient preferences concerning the trade-off between the risks and benefits of routine radiation therapy after conservative surgery for early-stage breast cancer. J Clin Oncol 15: 1252-1260, 1997
Furlong W, Feeny D, Torrance G: Guide to the Design and Development of Health-State Utility Instrumentation: Centre for Health Economics and Policy Analysis Working Paper. Hamilton, Ontario, Canada, McMaster University, 1990
Sumner W, Nease R, Littenberg B: U-titer: A utility assessment tool. Proc Annu Symp Comput Appl Med Care 701-705, 1991
Sumner W, Nease R, Littenberg B: U-titer: A utility assessment tool. Proc Annu Symp Comput Appl Med Care 701-705, 1991
Fisher B, Dignam J, Wolmark N, et al: Lumpectomy and radiation therapy for the treatment of intraductal breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-17. J Clin Oncol 16: 441-452, 1998
Weinstein MC, Feinberg HV: Clinical Decision Analysis. Philadelphia, PA, W.B. Saunders, 1982
Gail MH, Brinton LA, Byar DP, et al: Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 81: 1879-1886, 1989
Hollander M, Wolfe DA: Nonparametric Statistical Methods (ed 2). New York, NY, John Wiley & Sons, Inc, 1999
Dolan P: Whose preferences count? Med Decis Making 19: 482-486, 1999
Ubel PA, Loewenstein G, Jepson C: Whose quality of life? A commentary exploring discrepancies between health state evaluations of patients and the general public. Qual Life Res 12: 599-607, 2003
Chan JL, Kabeto MU, Oldread AE, et al: The use of preferences to measure the benefit of adjuvant radiation therapy for stage I seminoma. Int J Radiat Oncol Biol Phys 53: 934-941, 2002
Tengs TO, Wallace A: One thousand health-related quality-of-life estimates. Med Care 38: 583-637, 2000(James A. Hayman, Mohammed)
Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, MI
Division of Emergency Medicine, A.I. duPont Hospital for Children, Wilmington, DE
ABSTRACT
PURPOSE: To assess women's preferences regarding the trade-off between the risks and benefits of treatment with radiation therapy (RT) after breast-conserving surgery (BCS) for ductal carcinoma-in-situ (DCIS).
PATIENTS AND METHODS: Utilities were obtained from 120 patients and 210 nonpatients for eight relevant health states using standard gambles.
RESULTS: Differences in utilities obtained from patient and nonpatient participants between health states were relatively similar. Reduction in the likelihood of local recurrence associated with RT did not result in higher utilities. Utilities for noninvasive recurrence were only lower after initial treatment with RT. Patient and nonpatient participants had the lowest utilities for invasive local recurrence, regardless of initial treatment or manner of salvage therapy. When comparing patient and nonpatient utilities directly, patients had higher utility for being without recurrence after initial RT and lower utility for invasive recurrence salvaged by mastectomy after initial BCS alone. None of the clinical or sociodemographic factors examined explained more than 5% of the variability in the patients' or nonpatients' utilities or their differences.
CONCLUSION: The principal benefit associated with adding RT to BCS for DCIS seems to be its ability to reduce invasive recurrences.
INTRODUCTION
As the number of women diagnosed with ductal carcinoma-in-situ (DCIS) continues to grow,1 so has interest in how best to care for them. Although mastectomy remains an effective treatment, many women opt for breast-conserving therapy, which typically consists of lumpectomy followed by adjuvant radiation therapy (RT).2-4
As the use of RT has increased, questions about the need for RT have also increased.5 Results from the three randomized trials that address this issue are remarkably similar.6-8 Adding RT reduces the relative risk of both invasive and noninvasive local recurrences by 35% to 65%, but the absolute reduction in each type of recurrence is only in the range of 5% to 10%, and there is no evidence of a survival benefit.
The benefit associated with RT in this setting depends on the trade-off between the fear and consequences of both noninvasive and invasive local recurrence versus the inconvenience and potential toxicity of treatment. Because this trade-off revolves around the preferences women have for different outcomes, utilities are an appropriate means of assessing the potential value of RT in this setting. Utilities are global measures of quality of life that represent the strength of one's preference for a particular state of health and are measured on a scale ranging from 0 to 1, where 0 is equivalent to death and 1 is equivalent to optimal health.9 By comparing utilities for the range of probable outcomes after treatment with and without RT, it is possible to assess its incremental impact on quality of life. Utilities can also be used to calculate quality-adjusted life-years, which is the preferred measure of benefit in cost-effectiveness analyses.10
The issue of whose utilities should be measured is currently a matter of some debate. In older studies, utilities were commonly elicited from physicians and/or nurses.11 Concerns about their ability to accurately determine patients' preferences led many to argue that utilities should be elicited from patients. However, more recently, because cost-effectiveness analyses are primarily intended to inform decisions regarding the allocation of societal resources, it has been recommended that utilities be obtained from the general public.12
We previously collected patients' utilities to assess the benefit of RT after lumpectomy for invasive breast cancer and found that patients valued RT for its ability to decrease their fear of recurrence and the likelihood of developing an invasive recurrence requiring mastectomy.13 We hypothesized that similar benefits might also exist for DCIS. Accordingly, the goal of this study was to measure both patients' and nonpatients' utilities for the relevant health states after treatment of DCIS with lumpectomy, both with and without RT, to assess the impact of RT in this setting.
PATIENTS AND METHODS
Eligibility Criteria
Patient participants. Potential patient participants were identified from lists of women scheduled for routine follow-up in one of five radiation oncology departments in Southeast Michigan. To be eligible, patients had to be women with a history of DCIS who had undergone lumpectomy followed by RT, who had completed RT at least 1 month before their visit and had no history of recurrence or contralateral breast cancer.
Nonpatient participants. Potential nonpatient participants were identified from lists of women scheduled for a routine health maintenance visit in one of several general internal medicine clinics within the University of Michigan Health System. To be eligible, the women had to have no history of breast cancer.
Enrollment
After obtaining permission from their radiation oncologists, potential participants were sent a letter describing the study along with an opt-out card. Permission was not obtained before sending nonpatient participants similar letters. Potential subjects who returned the opt-out card were not contacted. Subjects who did not return the card were contacted by phone, and the study was described again. If a subject was not interested in participating, she was not contacted again. Interested subjects were interviewed after obtaining informed written consent, which generally occurred after the subjects saw their physician. Screening of potential participants ended when our accrual goals (see Statistical Methods) were met. This study was reviewed and approved by the University of Michigan Institutional Review Board and by the Institutional Review Boards at the other four participating centers.
Patient participants. Two hundred ninety-three potential patient participants were identified. At the request of their radiation oncologists, 26 women (8%) were not sent letters. Ninety-one women (31%) chose not to participate in the study after receiving the letter, and 56 (19%) could not be interviewed because of scheduling difficulties. One hundred twenty-four patients were successfully interviewed, but four were later found to be ineligible. Accordingly, the final patient sample consisted of 120 patients (42%).
Nonpatient participants. Eight hundred fifty-seven women were sent letters. Two hundred forty-four women (28%) actively chose not to participate in the study after receiving the letter, and 24 (3%) were found to be ineligible. Another 310 women (36%) could not be reached by telephone before their appointment, and 69 (8%) could not be interviewed because of scheduling difficulties, resulting in a final sample of 210 nonpatient participants (25%).
Interview
Participants were only interviewed once in the presence of a trained interviewer. The interview was based on approaches developed by Furlong et al14 and Sumner et al15 and was performed on a laptop using U-Titer (Washington University, St Louis, MO), a Macintosh-based program developed specifically for utility assessment.16 The 45-minute interview consisted of the following three parts (listed in order of occurrence): a ranking exercise, standard gambles, and a questionnaire.
After an in-depth review of the issues regarding the diagnosis and treatment of DCIS, participants were asked to read, in random order, eight hypothetical health states and to rank them from most to least desirable. The health states represented the range of possible local outcomes after treatment with breast-conserving surgery (BCS) with or without RT. These scenarios were based on those used previously.13 Briefly, the eight health states were as follows: (A) BCS followed by RT without recurrence; (B) a noninvasive recurrence salvaged with mastectomy after initial treatment with BCS and RT; (C) an invasive recurrence salvaged with mastectomy after initial treatment with BCS and RT; (D) BCS alone without recurrence; (E) a noninvasive recurrence salvaged with BCS and RT after initial treatment with BCS alone; (F) a noninvasive recurrence salvaged with mastectomy after initial treatment with BCS alone; (G) an invasive recurrence salvaged with BCS and RT after initial treatment with BCS alone; and (H) an invasive recurrence salvaged with mastectomy after initial treatment with BCS alone. Where relevant, it was emphasized that treatment with adjuvant RT has been shown to decrease the risk of DCIS and invasive recurrences but not to increase survival. The recurrence rates cited were based on data from the National Surgical Adjuvant Breast and Bowel Project B-17 trial available at the initiation of this study (8% and 4% and 13% and 13% for noninvasive and invasive recurrence with and without RT, respectively).17 Where relevant, participants were also informed that RT has not been proven to improve survival, commonly causes temporary acute side effects (eg, fatigue and skin irritation) and minor chronic effects (eg, changes in the appearance of the breast), and rarely can also cause more serious long-term complications (eg, pneumonitis, cardiac toxicity, and RT-associated malignancy). Participants were also informed that an invasive recurrence would lead to the need for axillary staging and that the likelihood of cure after salvage therapy for either a DCIS or invasive recurrence is high.
A variation of the classic standard gamble technique for measuring utilities was used during the second part of the interview. In the classic standard gamble, the participant is asked to choose between the following two alternatives: choice A is a gamble between optimal health and immediate death; and choice B is certain life in an intermediate health state. In an iterative fashion, the probabilities of optimal health and immediate death in choice A are varied systematically, and the participant is asked repeatedly to choose between the two alternatives. The process continues until the participant is indifferent; at that point, the probability of optimal health in choice A is the respondent's utility for the intermediate health state.18 To avoid having to constantly confront the participant with death as the worst possible outcome and to improve precision, we used a modification known as partial cascading.14 Using the rankings, the seven highest ranked health states were first assessed in seven standard gambles, using optimal health as the best possible outcome and the lowest ranked health state as the worst possible outcome. These seven standard gambles were conducted in random order. The lowest ranked health state was then assessed relative to optimal health and death. Using the results of these eight standard gambles, utilities for the eight intermediate health states could then be derived.
The questionnaire differed slightly for patients and nonpatients. Both questionnaires included items dealing with participants' age, marital status, number of dependents at home, race, education, and income. Patients were then asked whether they had taken tamoxifen, experienced RT toxicity, were satisfied with their cosmetic outcome, were fearful of recurrence, and were satisfied with their choice of breast-conserving therapy. Nonpatients were asked additional questions that allowed us to calculate their 5-year and lifetime breast cancer risk using the Gail model.19
Statistical Methods
The expected standard deviations of standard gamble utilities obtained from experienced patient participants and nonexperienced participants are 0.2 and 0.3, respectively.14 Therefore, a sample size of 120 patient participants was chosen so that the mean utilities for the eight health states could at least be estimated with a 95% CI of ± 0.04. A sample size of 210 nonpatient participants was chosen to achieve similar CIs. Differences in baseline characteristics between nonparticipants and participants from both patient and nonpatient groups and between participating patients and nonpatients were evaluated using t tests and Pearson 2 tests where appropriate. Because the utility data were not normally distributed (see Results), nonparametric statistical procedures were used.20 The Friedman rank sum test was first used to test whether participants' utilities for the eight health states within each of the two groups differed from one another overall. The Wilcoxon-Nemenyi-McDonald-Thompson procedure was then used to perform all pairwise comparisons, while adjusting for multiple comparisons. The Wilcoxon rank sum test was used to determine whether the patients' and nonpatients' utilities for the eight health states differed between the two groups. Finally, stepwise regression was used to examine whether any of the baseline covariates were significant predictors of participants' utilities for the health states themselves or the differences in utility between any two health states. Because of the skewed distribution of the utility scores, the logit transformation was used to make the data approximately normal before performing the regression analysis. Covariates examined included the following: age, race, marital status, number of dependents, education, and income for all participants; months since completion of RT, toxicity, treatment with tamoxifen, satisfaction with local outcome, fear of recurrence, and satisfaction with treatment decision for patient participants; and 5-year and lifetime risks of breast cancer for nonpatient participants.
RESULTS
Study Population Characteristics
The characteristics of the 120 DCIS patients and 240 nonpatient participants who were interviewed are listed in Tables 1 and 2. Nonpatients, compared with patient participants, were younger (mean age, 50 v 61 years, respectively; P < .001), more likely not to be white (16% v 7%, respectively; P = .03), less likely to have dependents at home (mean, 0.65 v 1.1 dependent, respectively; P < .001), and better educated (71% v 50% college graduates, respectively; P < .001). We found no evidence that our nonpatient participants were at greater risk of developing breast cancer than women in the general population. As calculated by the Gail model, the nonpatients' mean 5-year and lifetime risks of being diagnosed with breast cancer were 1.3% and 10.2%, respectively, compared with 1.2% and 10.1% for all women, respectively.
Interviews
All enrolled participants successfully completed the interview. The mean time ± standard deviation to complete the interview for patient and nonpatient participants was 42 ± 13 minutes and 43 ± 11 minutes, respectively.
Standard Gamble Utilities
Descriptive statistics. Patients' and nonpatients' standard gamble utilities are listed in Tables 3 and 4, respectively. For both groups, the median utilities for all health states are higher than the means, reflecting the fact that their distributions are skewed upward. In addition, as expected, there is considerable variability in the utilities, which are greatest for the recurrent invasive health states.
Within-group comparisons. The results of the pairwise comparisons between health states within the patient and nonpatient groups are listed in Table 5. Overall, the results are surprisingly similar. For the health states after treatment with RT, both groups had the strongest preference for being without recurrence, followed first by a DCIS recurrence and then by an invasive recurrence (ie, A > B > C). For the health states after treatment with lumpectomy alone, patient and nonpatient participants alike were indifferent to a DCIS recurrence but not to an invasive recurrence (ie, D = E = F > G and H). Both groups had similar utilities for the nonrecurrent health states, whether they had or had not received initial treatment with RT (ie, A = D). Patients and nonpatients also had similar utilities for a DCIS recurrence requiring mastectomy whether or not they had received upfront RT (ie, B = F). In contrast, patients and nonpatients not only had different utilities for an invasive recurrence requiring mastectomy after initial treatment with RT compared with no RT (ie, C v H), but the direction of the difference was also reversed. Patient participants preferred having received RT (ie, C > H), whereas nonpatient participants preferred not to have had received RT (ie, C < H). The only other relevant difference between the two groups was that patient participants preferred to preserve their breast after an invasive recurrence after initial treatment with BCS alone (ie, G > H), whereas nonpatient participants were indifferent (ie, G = H). Neither group expressed a preference for breast preservation when confronted with a DCIS recurrence after initial treatment with BCS alone (ie, E = F).
Between-group comparisons. Overall, there were few differences that were statistically significant when comparing patient and nonpatient participants' utilities directly (Table 6). Patients had higher utilities than nonpatients for being without recurrence after BCS and RT (A) and lower utilities for having an invasive recurrence salvaged by mastectomy after BCS alone (H). Differences in utilities for being without recurrence after BCS alone (D) and for having had an invasive recurrence salvaged by mastectomy after BCS and RT (C) were of borderline significance (P = .09 and P = .06, respectively).
Factors associated with participants' utilities. Although several of the clinical and sociodemographic factors examined were statistically associated with patient and nonpatient participants' utilities, none of the factors explained more than 5% of the variability in the utilities themselves or in the differences between health states.
DISCUSSION
Our results suggest that the principal benefit associated with adding RT to BCS for the treatment of DCIS is the ability of RT to reduce actual invasive recurrences, which is an event perceived both by patients and nonpatients as having a significant negative impact. The potential quality-of-life benefit associated with a reduction in the risk of recurrence after treatment with RT seems to be negated in both groups by concerns regarding the inconvenience and potential toxicities of treatment. In addition, compared with nonpatients' predictions, patients perceive their current quality of life (ie, disease free after lumpectomy and RT) to be higher and predict that it would decrease to a lower level after an invasive recurrence salvaged by mastectomy. Finally, both patient and nonpatient participants have considerable variation in their preferences that are not explained by the sociodemographic and/or clinical factors examined.
On the basis of our previous experience with invasive breast cancer,13 we expected women to prefer the nonrecurrent health state after initial treatment with RT (A), with its lower risk of recurrence, versus without RT (D; ie, expected A > D). However, in the present study, we found that the utilities for these two states were not significantly different, implying that women do not feel that the reduction in the fear of recurrence associated with 5% and 9% absolute decreases in the likelihood of noninvasive and invasive recurrences, respectively, outweighs the potential toxicity and inconvenience of upfront RT. A likely explanation for this finding is that the absolute reduction in the likelihood of an invasive recurrence is only 9% compared with 30% in our prior invasive study. It is also possible that a small difference exists between the two nonrecurrent health states but that we lacked the statistical power to detect it.
The negative impact associated with an invasive recurrence was one of the most consistent findings across both groups of participants. Given the potential need for axillary dissection and/or chemotherapy and the possibility of developing incurable metastatic disease after such an event, the finding that health states C, G, and H were the least preferred states comes as no surprise. It is also not surprising that, after initial treatment with adjuvant RT, both groups of women first favored no recurrence, followed by a DCIS recurrence and then an invasive recurrence. However, it was surprising that women do not seem to be bothered by a noninvasive recurrence without or with the loss of their breast after initial treatment with surgery alone (ie, D = E = F). Perhaps after gambling and losing, women are more accepting of a local recurrence, whether or not it results in loss of the breast, as long as it is not invasive. As for their feelings regarding the recurrent invasive health states, patient and nonpatient participants differed somewhat. Probably because of cognitive dissonance, patients previously treated with RT preferred having received initial treatment with RT if they went on to experience an invasive recurrence salvaged with mastectomy (ie, C > H). In contrast, nonpatients preferred not to have received upfront RT (ie, H > C), which is understandable given the inconvenience and potential toxicity of RT if the ultimate outcome was the same anyway. Lastly, patients preferred a second lumpectomy and RT versus mastectomy as salvage therapy of an invasive recurrence after initial treatment with lumpectomy alone (ie, G > H), whereas nonpatients were indifferent (ie, G = H). One potential explanation for this finding is that patients were more predisposed towards breast conservation given that they had all chosen lumpectomy when diagnosed with DCIS.
As is true in our study, patients often value their current state of health (A) more highly than nonpatients predict they would. Numerous theories exist regarding why this might be the case.21,22 The most compelling explanation in this case is that nonpatients often fail to account for their ability to adapt to adverse circumstances over time. As for patients' lower utility for an invasive recurrence salvaged with mastectomy after initial treatment without RT, the most likely explanation has to do with a difference in their vantage point.22 Given their prior decision to opt both for breast conservation and RT, it is not surprising that patients perceive an invasive recurrence resulting in the loss of their breast after foregoing RT more negatively than nonpatients.
As has been true in our previous studies,13,23 we found that participants vary considerably in their preferences for various outcomes and that these differences are not explained by the sociodemographic and/or clinical factors examined, implying once again that participants' feelings about these issues are complex.
Although a number of investigators have reported utilities collected from patients for local outcomes associated with the treatment of invasive breast cancer,11,24 to the best of our knowledge, we are the first to report utilities for DCIS. In general, after accounting for differences in health state descriptions, study populations, and utility elicitation techniques, our results are similar to those reported previously. Our results add to this literature by directly assessing preferences for outcomes associated with DCIS from both patients and nonpatients.
One potential limitation of our study is that the health state scenarios may have inadvertently biased participants for or against RT. Although the content of the scenarios was determined by experienced clinicians, decisions still had to be made regarding what information to include in the scenarios. When constructing the scenarios, a conscious effort was made to present the information in a fair and neutral fashion and, when faced with uncertainty, to bias the scenarios against RT. Although more recent results from the National Surgical Adjuvant Breast and Bowel Project B-17 trial have been reported, the absolute differences between the rates of noninvasive and invasive recurrences without and with RT have not changed significantly with longer follow-up.6 Given our participants' high education and income levels, our study can also be criticized on the grounds that our study populations are not representative. However, it should be noted that two thirds of our patient participants were recruited from community-based radiation oncology departments, and we found no evidence that our nonpatient participants were at greater risk of developing breast cancer than women in the general population. We also acknowledge that our decision to only include women treated with lumpectomy and RT among our patient participants may also have biased our results. However, we wanted experienced patient participants who were knowledgeable about both lumpectomy and RT. Finally, our decision to require permission from their physician to contact patients and our inability to schedule interviews with all interested participants may also have introduced bias.
In conclusion, it does not seem to be an increased risk of recurrence, mastectomy, or even a noninvasive recurrence, but rather an invasive recurrence, that women fear most after breast-conserving therapy for DCIS. This implies that invasive recurrence should probably be the primary focus when assessing the benefit of any treatment for DCIS.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all of the women who agreed to participate in this study and acknowledge the assistance of Elizabeth Oldread, Kerry Paisley, and Glenn Wright in collecting and helping to analyze the data presented in this article.
NOTES
Supported by Clinical Research Training Grant for Junior Faculty No. CRTG-99-248-01-CCE from the American Cancer Society.
Presented in part at the 37th Annual Meeting of the American Society of Clinical Oncology, San Francisco, CA, May 12-15, 2001.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
American Cancer Society: Cancer Facts and Figures 2003. Atlanta, GA, American Cancer Society, 2003
Ernster VL, Ballard-Barbash R, Barlow WE, et al: Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst 94: 1546-1554, 2002
Innos K, Horn-Ross PL: Recent trends and racial/ethnic differences in the incidence and treatment of ductal carcinoma in situ of the breast in California women. Cancer 97: 1099-1106, 2003
Winchester DJ, Menck HR, Winchester DP: National treatment trends for ductal carcinoma in situ of the breast. Arch Surg 132: 660-665, 1997
Silverstein MJ, Lagios MD: Benefits of irradiation for DCIS: A Pyrrhic victory. Lancet 355: 510-511, 2000
Fisher B, Land S, Mamounas E, et al: Prevention of invasive breast cancer in women with ductal carcinoma in situ: An update of the national surgical adjuvant breast and bowel project experience. Semin Oncol 28: 400-418, 2001
Houghton J, George WD, Cuzick J, et al: Radiotherapy and tamoxifen in women with completely excised ductal carcinoma in situ of the breast in the UK, Australia, and New Zealand: Randomised controlled trial. Lancet 362: 95-102, 2003
Julien JP, Bijker N, Fentiman IS, et al: Radiotherapy in breast-conserving treatment for ductal carcinoma in situ: First results of the EORTC randomised phase III trial 10853—EORTC Breast Cancer Cooperative Group and EORTC Radiotherapy Group. Lancet 355: 528-533, 2000
Torrance GW: Measurement of health state utilities for economic appraisal. J Health Econ 5: 1-29, 1986
Weinstein MC, Siegel JE, Gold MR, et al: Recommendations of the Panel on Cost-effectiveness in Health and Medicine. JAMA 276: 1253-1258, 1996
Earle CC, Chapman RH, Baker CS, et al: Systematic overview of cost-utility assessments in oncology. J Clin Oncol 18: 3302-3317, 2000
Russell LB, Gold MR, Siegel JE, et al: The role of cost-effectiveness analysis in health and medicine. JAMA 276: 1172-1177, 1996
Hayman JA, Fairclough DL, Harris JR, et al: Patient preferences concerning the trade-off between the risks and benefits of routine radiation therapy after conservative surgery for early-stage breast cancer. J Clin Oncol 15: 1252-1260, 1997
Furlong W, Feeny D, Torrance G: Guide to the Design and Development of Health-State Utility Instrumentation: Centre for Health Economics and Policy Analysis Working Paper. Hamilton, Ontario, Canada, McMaster University, 1990
Sumner W, Nease R, Littenberg B: U-titer: A utility assessment tool. Proc Annu Symp Comput Appl Med Care 701-705, 1991
Sumner W, Nease R, Littenberg B: U-titer: A utility assessment tool. Proc Annu Symp Comput Appl Med Care 701-705, 1991
Fisher B, Dignam J, Wolmark N, et al: Lumpectomy and radiation therapy for the treatment of intraductal breast cancer: Findings from National Surgical Adjuvant Breast and Bowel Project B-17. J Clin Oncol 16: 441-452, 1998
Weinstein MC, Feinberg HV: Clinical Decision Analysis. Philadelphia, PA, W.B. Saunders, 1982
Gail MH, Brinton LA, Byar DP, et al: Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 81: 1879-1886, 1989
Hollander M, Wolfe DA: Nonparametric Statistical Methods (ed 2). New York, NY, John Wiley & Sons, Inc, 1999
Dolan P: Whose preferences count? Med Decis Making 19: 482-486, 1999
Ubel PA, Loewenstein G, Jepson C: Whose quality of life? A commentary exploring discrepancies between health state evaluations of patients and the general public. Qual Life Res 12: 599-607, 2003
Chan JL, Kabeto MU, Oldread AE, et al: The use of preferences to measure the benefit of adjuvant radiation therapy for stage I seminoma. Int J Radiat Oncol Biol Phys 53: 934-941, 2002
Tengs TO, Wallace A: One thousand health-related quality-of-life estimates. Med Care 38: 583-637, 2000(James A. Hayman, Mohammed)