Abstract
BACKGROUND:
After primary surgical resection, breast cancer survivors regularly undergo surveillance using multiple modalities to detect recurrences.
OBJECTIVE:
The goal of this study was to determine how breast cancer recurrences were detected in our sample population of breast cancer survivors. We hypothesize that the majority of recurrences are patient-detected.
METHODS:
This was a retrospective observational study evaluating patients with a detected breast cancer recurrence between 2010 and 2015. Patients were analyzed by initial detection modality (patient versus clinician versus imaging). Other variables of interest included insurance status, ambulation capability, living situation, age at diagnosis and recurrence, and time to recurrence. Statistical methods included chi-square tests, log-rank tests, and analysis of variance.
RESULTS:
115 patients were identified with a recurrence (mean age 61 years). 88 (77%) recurrences were patient-detected, 14 (12%) were detected by imaging, and 13 (11%) were clinician-detected (chi-square goodness of fit test p < 0.001). Median time to recurrence was 4 years in the patient-detection group compared to 3 for the clinician group and 2 for imaging detection (log-rank test p = 0.01).
CONCLUSION:
While the majority of recurrences were patient-detected, these were detected later and at a more advanced clinical stage. This could represent an opportunity to enhance outcomes by empowering patient detection strategies.
Introduction
Breast cancer is the most common cancer amongst women. In 2013, the prevalence of women living with breast cancer was estimated to be over 3 million in the United States. In 2016, there is estimated to be an incidence of 246,600 breast cancer cases [1]. Based on SEER data from 2006 to 2012, the 5-year survival rate amongst women diagnosed with breast cancer is 89.7% [1]. With a rising prevalence of breast cancer-survivors, there is an increasing focus on detecting second primary breast cancers and locoregional recurrences in this patient population. Numerous studies have demonstrated that early detection of recurrence results in increased survival [2,3]. However, many have begun to question traditional post-treatment surveillance programs on the basis of modern patterns of recurrence detection, concern with overall healthcare cost-reduction, and prevention of unnecessary utilization of resources. Historically, intensive hospital/clinic-based surveillance regimens with frequent clinical exams and imaging tests were recommended in hopes of improving early detection of recurrence and therefore improving clinical outcomes and quality of life in women with a history of breast cancer. However, there is little evidence to support improvement within these parameters amongst breast cancer survivors [2–5].
Additional studies have demonstrated a consistent trend in local recurrence risk for stage 1–3 breast cancers - increasing in the first two years post-treatment, peaking during the third year, and declining thereafter. Furthermore, a large proportion of these recurrences were detected by mammography rather than clinical examination [5]. Furthermore, recent studies suggests no survival benefit to recurrences detected by clinical assessment, especially in comparison to recurrences detected through mammography or self-detected by patients [6–9]. Cumulatively, these findings raise several questions regarding post-treatment breast cancer surveillance, including which modality of surveillance is optimally suited to detect recurrences.
Our study was designed to address these questions and help inform a value-based model of breast cancer recurrence surveillance. Our goal was to determine how breast cancer recurrences were detected in our sample population of breast cancer survivors. We also aimed to evaluate the impact, if any, of patient socioeconomic factors on patterns of recurrence detection. We hypothesize that the majority of recurrences in our breast cancer population are detected by patients themselves, independent of routine scheduled clinical exams.
Methods
The study was a retrospective chart review that evaluated breast cancer patients at the University of Vermont Cancer Center with a detected breast cancer recurrence between 2010 and 2015. Of 1999 charts reviewed, 115 patients met criteria. The recurrence database was organized according to the Surveillance, Epidemiology, and End Results (SEER) Program Code Manual. Recurrences were characterized as in-situ (stage 0), loco-regional (stage 1–3), or distant (stage 4). Throughout the surveillance of the breast cancer patients, only the highest stage of the recurrence was recorded. Although the majority of the recurrences were biopsy proven, a small minority were clinically determined to be a recurrence. All patients’ charts were audited by a single reviewer to control for inter-reviewer variability. The categorical variables of interest were modality of detection (patient, clinician, routine imaging), insurance (Medicare, private), ambulation status (assistance, no assistance), and living situation (isolated, non-isolated). Our intention was to evaluate the additional variables potentially contributing to a patient’s ability to access care. The continuous variables of interest were age at diagnosis, age at recurrence, and time to recurrence detection.
In order to meet the criteria for a patient-detected recurrence, a patient must have presented with symptoms (e.g. a patient-detected lump) that led to subsequent imaging, clinician evaluation, or biopsy that confirmed a recurrence. The criteria for clinician-detected recurrence was met if the recurrence was detected during a scheduled/routine clinician appointment. The criteria for image-detected recurrence was met if a recurrence was detected during scheduled, or routine, breast imaging. Patients were divided into multiple cohorts based on specified criteria, including manner of recurrence detection (patient, clinician, or image-detected), time from initial diagnosis to recurrence detection (less than 1 year, 1–3 years, and greater than 3 years). All patients were documented as being involved with regular surveillance including mammography and clinical exam.
All patients considered in this study were adherent to the American Society of Clinical Oncology guidelines followed at the breast center. The practice at University of Vermont, Medical Center at the time of the study was adherent to the ACO guidelines as of 2013. A history, physical examination, and mammography are recommended for breast cancer follow-up. Physical examinations should be performed every 3–6 months for the first 3 years, every 6–12 months for years 4 and 5, and annually thereafter. For women who have undergone breast-conserving surgery, a post-treatment mammogram should be obtained 1 year after the initial mammogram and at least 6 months after completion of radiation therapy. Thereafter, unless otherwise indicated, a yearly mammographic evaluation should be performed. The use of complete blood counts, chemistry panels, bone scans, chest radiographs, liver ultrasounds, pelvic ultrasounds, computed tomography scans, [18F]fluorodeoxyglucose–positron emission tomography scans, magnetic resonance imaging, and/or tumor markers (carcinoembryonic antigen, CA 15-3, and CA 27.29) is not recommended for routine follow-up in an otherwise asymptomatic patient with no specific findings on clinical examination.
Key outcome variables analyzed included Cohorts were analyzed and compared on outcomes – statistical significance was determined for categorical variables via chi squared analysis or Fisher’s exact test, and for continuous variables via analysis of variance. Time to recurrence was analyzed using Kaplan–Meier survival analysis.
Results
115 patients in the designated study period were identified with a documented breast cancer recurrence. Baseline characteristics were assessed within the patient recurrence group (See Fig. 1). Mean age at time of recurrence was 61 years old. At initial diagnosis (mean age – 55 years old), 12 (10%) patients had stage 0 breast cancer, 40 (35%) patients had stage 1 breast cancer, 37 (32%) patients had stage 2 breast cancer, 24 (21%) patients had stage 3 breast cancer, and 2 (2%) patients had stage 4 breast cancer. At the time of initial treatment, 43 (37%) patients underwent a mastectomy and 72 (63%) patients underwent lumpectomy.
88 (77%) of 115 breast cancer recurrences were detected by the patient, 14 (12%) were detected by scheduled imaging, and 13 (11%) were detected by the clinician during a routine surveillance visit (chi-square goodness of fit test p < 0.001, see Fig. 2). There was a significant difference in time of recurrence detection with respect to mode of detection. Median time to recurrence was 4 years in the patient detection group compared to 3 years for the clinician group and 2 years for imaging detection (log-rank test p = 0.01, see Fig. 3).
Recurrences detected by the patient were distant recurrences (stage 4, 85.2%), followed by local recurrences (stages 1–3, 12.5%) and in-situ disease (stage 0, 2.3%). Recurrences detected by the clinician were distant recurrences (stage 4, 69.2%), followed by local recurrences (stages 1–3, 23.1%) and in-situ disease (stage 0, 7.7%). Lastly, Recurrences detected by imaging were distant recurrences (stage 4, 42.9%), followed by local recurrences (stages 1–3, 42.9%) and in-situ disease (stage 0, 14.3%). These differences were statistically significant (p = 0.004, Fisher’s exact test, see Table 1).
The overall 5-year survival rates for patient-detected, clinician-detected, and imaging-detected recurrences were 42%, 23%, and 7%, respectively. The Kaplan–Meier survival curves for the three modalities of detection can be viewed in Fig. 5. Lastly, there were no significant differences in these observations among the socioeconomic and demographic dimensions evaluated.
Discussion
The 2015 Breast Cancer Survivorship Care Guidelines by the American Cancer Society (ACS) and American Society of Clinical Oncology (ASCO) recommend annual routine mammography (of breast treated with breast conserving surgery and intact contralateral breast), and clinical breast exam for the surveillance of breast cancer survivors [10]. Our study found that the vast majority of breast cancer recurrences were detected by the patient, followed by routine imaging, and lastly by routine clinical exam. This finding was consistent across all socioeconomic-demographic groups. Not surprisingly, our study also demonstrated that surveillance imaging, although only the second most common means of detecting a breast cancer recurrence, identified cancers in the earliest stages. Prior studies have demonstrated that women, who self-detected recurrences or had a mammogram-detected local recurrence, had a better survival rate compared to those detected by clinical examination [10,16,17,27]. In our study, mammogram-detected and clinician-detected recurrences were associated with lower stages of detection, compared to patient-detected recurrences which were associated with later stages of disease. Given that the vast majority of breast cancer recurrences were self-detected by patients, the finding of late-stage disease may indicate an opportunity to better prepare patients to take on their crucial role in surveillance, through education and training of breast cancer survivors in self-assessment.
Our study, and others like it, also re-evaluates clinician-directed surveillance strategies. Historically, clinical exam played a significant role in breast cancer recurrence detection, however, the utility of clinical breast exam for surveillance has diminished, with recurrences detected by clinical exam estimated at only 15% in recent studies [18–20]. A systematic review of the literature demonstrated that studies published prior to 2000 reported that 15% of recurrences were detected by mammogram compared to 46% detected by routine clinical exam. However, studies published after 2000 reported 40% of recurrences detected by mammogram and only 15% detected by routine clinical examination [20]. Similarly, in a European study of 110 patients with recurrent disease following treatment with breast conserving surgery and radiation between 1991 and 1998, only 13.5% of recurrences were clinically detected. Survival was also lower for this cohort of patients [19]. This finding is particularly relevant as data have demonstrated the trend toward attrition in patient adherence to guideline-concordant post-treatment imaging surveillance for breast cancer survivors [19].
The cumulative data suggest that the utility of routine clinical assessment in the surveillance of breast cancer recurrence has decreased as expertise with breast imaging has increased. Furthermore, there is no evidence to suggest that clinical examination confers a survival advantage compared with other methods of detection. Such findings reinforce the critical role that imaging plays in surveillance, underscores the importance of patient education in self-detection of breast cancer recurrence, and points to the potential diminishing returns of routine clinical follow-up in surveillance. Current recommendations in breast cancer surveillance guidelines should be revisited and reconsidered in light of the emerging evidence. Placing greater emphasis on patient education and strategies to improve adherence to recommended breast imaging surveillance could have a significant impact on improving early detection of recurrence in breast cancer. This distribution of service has the potential to improve outcomes for patients while reducing clinical inefficiencies. To date, few studies of alternative models of surveillance have been conducted, and those that have are generally underpowered. The studies do indicate that patient willingness to undergo alternative models of surveillance and satisfaction with these models is high [21].
Of course, these findings do not take into account the potential therapeutic benefit of the social interaction and counseling that a clinician may provide to a patient. This type of interaction with a trusted clinician could be beneficial to patients on an emotional, spiritual or psychological level. The findings also do not challenge the critical needs for timely clinical evaluation for symptomatic patients, or clinical follow-up of abnormal imaging results.
Limitations of our findings include the retrospective nature of our study, and the relatively small sample size within a single institution. Despite these limitations, the study was adequately powered for statistical significance and demonstrated positive findings. A future prospective and/or multi-institutional study on patterns of breast cancer recurrence detection may provide further evidence to inform clinical policy and guidelines in this area of care.
Conclusion
Cancer recurrence remains an ever-present risk for patients following treatment for breast cancer. In our series, the majority of recurrences, particularly late recurrences, are detected by the patients themselves. Despite a large volume of literature on breast cancer recurrence, patient self-detection of recurrence remains an understudied diagnostic modality. Our findings suggest the potential benefit of shifting more of the focus of breast cancer surveillance to patient detection and compliance with breast imaging surveillance, while optimizing efficiency in routine clinical exams. The number of breast cancer recurrences that were identified by patient detection and routine breast imaging were significantly greater than the number of recurrences identified through scheduled clinical exams. While routine evaluation and assessment by a clinician remains a cornerstone for quality breast cancer patient care, improvement in patient education and self-assessment may be an important and underexplored factor in post-treatment care.