Results Summary
What was the research about?
About 1 million women in the United States have been diagnosed with a condition called ductal carcinoma in situ, or DCIS. In DCIS, abnormal cells are found inside milk ducts in the breast. DCIS may or may not turn into cancer. Having DCIS increases a woman’s risk of getting a second diagnosis of DCIS or cancer in either breast.
One way doctors treat DCIS is to remove the abnormal cells and some of the tissue around them. This surgery is called a lumpectomy. Sometimes a patient and her doctor decide to have a mastectomy to remove the entire breast. To kill any abnormal cells that surgery may have missed, doctors may also add radiation. Choosing what kind of surgery to have is personal. Patients must decide if keeping their breast is worth the higher risk of getting DCIS again or cancer. Having information about the risks and benefits of each treatment could help patients and doctors choose a treatment.
In this study, the research team looked at records from national databases to learn
- What patient traits, such as age or race, may affect a woman’s risk of getting another DCIS or breast tumor in the other breast
- How likely it is that a woman who had treatment for DCIS will have a mastectomy if she has a second breast cancer
Next, the research team updated a computer model to predict the chances of survival, getting another DCIS, or getting breast cancer in the other breast. They compared the predicted outcomes from the model with data from other sources to see if the outcomes were like real patient experiences. The team used the model to create an online resource for women diagnosed with DCIS. This website allows women to see what’s likely to happen after different treatments.
What were the results?
Patient traits. Traits linked to getting another DCIS or cancer in the other breast included
- Age
- Year of DCIS diagnosis
- Race
- Tumor size
- Type of estrogen receptors, which are proteins on cells that tell them to grow
Women treated in parts of the country where doctors often use radiation to treat DCIS were more likely to have a mastectomy for a second DCIS, even if they had not been treated with radiation for the first one.
Predicting future outcomes. The computer model could accurately predict the chance that a woman would get DCIS or breast cancer within 10 years. The model successfully estimated how long women ages 45 and 60 would live after their first DCIS diagnosis. But the model overestimated how long women aged 70 years would live.
What did the research team do?
To build the computer model, the research team used data from other studies and databases. Next, the team used different data to check the computer model’s results. They then developed the website.
What were the limits of the study?
Some information may be missing from the databases. For example, the databases may not list all the patients who got a second breast tumor. Also, the databases didn’t have information about patient preferences for different treatments. The results might be different if the databases had this missing information.
Future research could look at how a patient’s actual choice for the first DCIS treatment may affect future outcomes.
How can people use the results?
The model can help predict future outcomes based on the initial treatment choice for DCIS. Doctors and patients could use the information from the model to help choose a DCIS treatment.
Professional Abstract
Objective
To determine the incidence of and risk factors for developing contralateral breast cancer after treatment for ductal carcinoma in situ (DCIS), examine treatment of second breast cancers after DCIS, and create a patient-facing decision support tool
Study Design
| Design Elements | Description |
|---|---|
| Design | Retrospective and simulation study |
| Data Sources and Data Sets | Data from the National Surgical Adjuvant Breast and Bowel Project B17 (1985–1992) and B24 trials (1991–1998); the UK, Australia, and New Zealand DCIS trial (1990–2008); an observational study of newly diagnosed patients with DCIS treated in British Columbia (1985–1997); Surveillance, Epidemiology, and End Results (SEER) databases (1990-2008); SEER-Medicare (1990-2010); European Organization for Research and Treatment of Cancer (EORTC) randomized trial 10853 |
| Analytic Approach | Multivariable competing risk regression, multivariable logistic regression modeling, and discrete event simulation modelling for outcome calculation |
| Outcomes | Contralateral breast cancer incidence, disease-free survival, invasive disease-free survival, overall survival, and likelihood of breast preservation |
To examine the risks and benefits of treatment options for DCIS, researchers began with an initial analysis to identify patient risk factors for developing a new contralateral breast tumor after receiving treatment for a first DCIS. A second analysis investigated the effect of regional patterns of radiation treatment on the likelihood of receiving a mastectomy at the time of a second breast tumor among women who did not receive radiation for the first DCIS. Analyses used SEER and SEER-Medicare databases.
The researchers used data from multiple studies and databases to update a DCIS simulation model to predict outcomes for a woman who was disease-free after receiving treatment for the first DCIS. The model simulated treatment options and nine possible outcomes related to survival, breast cancer recurrence, or new breast cancer. Simulation inputs included the patient’s age and recurrence risks for DCIS and invasive disease in the same breast. If the simulated individual had a recurrence, the model selected an appropriate treatment. Treatment strategies included breast conservation surgery (BCS) alone, BCS with radiation, BCS with tamoxifen, BCS with radiation and tamoxifen, mastectomy with breast reconstruction, and mastectomy without reconstruction.
To validate the model, researchers compared program predictions with data from the European Organisation for Research and Treatment of Cancer and SEER.
Based on results from the simulation model, researchers created an online decision-support resource that enabled women to see predicted outcomes based on their treatment choices. Four patient stakeholders provided feedback throughout the study and helped design the decision-support tool.
Results
Retrospective analyses. Age, year of diagnosis, race, tumor size, and tumor estrogen receptor status were significant predictors of developing a contralateral recurrence after treatment for the first DCIS (p<0.05).
Researchers also found a relationship between regional radiation use for DCIS and type of surgery for second breast cancer. Patients who had their second breast cancer treated in geographic regions where doctors often used radiation had increased odds of receiving a mastectomy. This relationship was true even among patients who had not previously received radiation and who were candidates for a second lumpectomy.
Predictive model. The model successfully predicted 10-year local recurrence rates and breast cancer-specific survival for each treatment. Although the predicted effect of radiation on local recurrence was larger in the model than in the validation data (16% versus 11% difference, respectively), the model ranked the outcomes of treatment options consistently. The model predictions closely matched real SEER data for cancer-specific survival in all age groups and overall survival in 45- and 60-year-old women, but in 70-year-old women, the overall survival rate predicted by the model was 8% higher than that reported in validation data.
Limitations
Shortcomings of the SEER data used in this study, such as missing information, as well as underdiagnoses of contralateral breast cancer and limited data about radiation use may make outcome estimates less accurate.
Conclusions and Relevance
Predictors of outcomes following treatment for DCIS include age, race, date of diagnosis, tumor size, tumor estrogen receptor status, and geographic region where patients receive treatment. The model accurately predicted recurrence and survival in most women. When incorporated into the online decision-support resource, this information may help physicians and patients with DCIS make informed treatment decisions.
Future Research Needs
Future research could also further refine the model and the decision-support resource.
Final Research Report
View this project's final research report.
Journal Citations
Related Journal Citations
Peer-Review Summary
Peer review of PCORI-funded research helps make sure the report presents complete, balanced, and useful information about the research. It also assesses how the project addressed PCORI’s Methodology Standards. During peer review, experts read a draft report of the research and provide comments about the report. These experts may include a scientist focused on the research topic, a specialist in research methods, a patient or caregiver, and a healthcare professional. These reviewers cannot have conflicts of interest with the study.
The peer reviewers point out where the draft report may need revision. For example, they may suggest ways to improve descriptions of the conduct of the study or to clarify the connection between results and conclusions. Sometimes, awardees revise their draft reports twice or more to address all of the reviewers’ comments.
Peer reviewers commented, and the researchers made changes or provided responses. Those comments and responses included the following:
- The reviewers questioned the adequacy of the multivariable models of the choice of breast cancer treatments, breast conserving surgery versus radiation therapy. The researchers acknowledged that there were some unmeasured factors, such as previous hormone use and availability of radiation oncology services, that could influence women’s decision making. They noted in their limitations that the study did not include these factors as potential confounders.
- The reviewers expressed concern about the low rate of reporting of estrogen receptor (ER) status, 14 percent, in the data set, given the results showing the association of ER status with the occurrence of contralateral breast cancer. The reviewers noted the possibility that this association was invalid given the amount of missing data for ER status. They also noted the possibility of this association being invalid given the potential for a difference in the women for whom ER status was known and the women whose ER status was unknown. The researchers acknowledged that the lack of additional information on ER status limited the conclusions that could be made from these results. However, they also noted that although testing for ER status varied systematically based on the sophistication of specific clinics or regions, there was no reason to believe that the consequent systematic selection of patients for ER testing would bias the clinical characteristics of ER positive versus ER negative cases.