If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Breast cancer (BC) patients undergoing surveillance often fear recurrence. Given that routine imaging is not recommended, recognizing metastatic disease early requires a knowledge of recurrence patterns. The aim of this study was to analyze the most common presentations of metastatic disease.
Patients and Methods
A retrospective review was conducted of patients who were initially diagnosed with early-stage BC and who later developed metastatic disease. Data collected included method of metastatic disease diagnosis, types of symptoms at diagnosis, and survival. Chi-square tests as well as logistic and Cox regression models were used.
Metastatic diagnoses were made from reported symptoms in 77.6% of patients, clinical examination in 3.2%, and 7.8% incidentally on imaging. Among those with symptoms, musculoskeletal pain was the most common (33.7%) and was more frequently noted at scheduled (48.9%) compared to acute-care visits (26.0%, P < .01). Receptor status was associated with nervous system symptoms at metastasis (P = .01), with higher odds of nervous system symptoms in triple-negative (odds ratio = 3.02) compared to estrogen receptor/progesterone receptor–positive, HER2- cases. On multivariable analysis, initial stage (P = .03), receptor status (P < .01), age (P < .01), and time to recurrence (P < .01) were significantly associated with 10-year survival after diagnosis of metastasis, whereas the presence of symptoms was not (P = .27). Providers of BC patients undergoing surveillance should modify their threshold of suspicion for recurrence depending on the characteristics of the initial diagnosis and the symptoms subsequently reported.
In this retrospective study, patients who presented with symptoms did not have shorter survival compared to those who were diagnosed in other ways.
Great strides have made in adjuvant therapy for early-stage breast cancer (BC) in the last few decades, fortunately leading to significantly fewer cases of metastatic disease. However, 25% to 40% of patients still develop metastatic BC, even after receiving standard multimodal therapy.
A common question encountered in practice is, “How do I know if the cancer is back?” This can be a difficult question to answer, given that patients can present with metastatic BC in several ways, and routine surveillance imaging of the body (other than the breasts) is not recommended.
Moreover, survivorship care is increasingly being transferred to advanced-practice providers in dedicated survivorship clinics or to primary-care providers. Thus, the ability to recognize metastatic disease early requires adequate knowledge of recurrence patterns.
The aims of this study were to analyze the most common ways in which patients with previously treated BC present with metastatic disease and to investigate if there are differences in overall survival (OS) by manner of presentation.
Patients and Methods
Any patient at the Holden Comprehensive Cancer Center who had been initially diagnosed with stage I-III BC from January 1, 2000, through December 31, 2017, and then later developed metastatic disease was included. The patient cohort was created using TriNetX, a global health research network that provides access to a network of in-house clinical data. Three separate database searches were conducted, given the difficulty of creating a single search that would capture all eligibility criteria. These searches, or groups, eventually comprised the final cohort:
Any female patient, 18 years or older, with a diagnosis of BC who received denosumab or zoledronic acid.
Any female patient, 18 years or older, with a diagnosis of BC who had a computed tomographic or positron emission tomographic scan followed by chemotherapy.
All male patients diagnosed with BC in the above-stated time frame, given the rarity of male BC.
The above groups were merged, and overlap was taken into account. After obtaining institutional review board approval, each patient’s chart was retrospectively reviewed for eligibility. Finally, additional eligible patients identified from the BC clinic were added to the cohort if not initially listed. Patients with incomplete information regarding their metastatic presentation were excluded. Data on vital status were obtained from Iowa Cancer Registry. Figure 1 provides a flow diagram illustrating subject selection.
Active data collection took place from January 26, 2018, through July 15, 2018. Using a standardized review form, data were collected on demographic variables; also collected were the details of the initial and subsequent metastatic BC diagnoses. To look for potential differences and inconsistencies in data collection, an initial set of 20 patient charts was reviewed by 3 of the investigators (S.P., N.I., N.G.), and modifications to the methodology were implemented as appropriate. This procedure was repeated with another 20 charts to ensure that the modifications were successful. Throughout the data collection period, answers to any data item that were considered debatable or uncertain were reviewed by the primary investigator (S.P.).
We were particularly interested in the following: how the metastatic recurrence was discovered (symptoms, finding at examination, or other method of detection) and whether this differed among receptor subtypes; if patients had symptoms, where and when they sought care (to an oncology clinic at a prescheduled follow-up or acute-care visit, an emergency department, urgent care, or primary care); and OS, stratified by manner of metastatic presentation.
The association between characteristics at initial diagnosis and presence of symptoms at metastatic presentation was evaluated by Firth-penalized logistic regression models. Cox regression models were utilized to assess factors associated with time to recurrence (TTR) and OS. Time was calculated from initial diagnosis to metastatic recurrence for TTR and from metastatic recurrence to death due to any cause for OS. Estimated effects of predictors are reported as odds ratios (OR) or hazard ratios (HR) along with 95% confidence intervals (CIs). All statistical testing was 2 sided and was assessed for significance at the 5% level using SAS 9.4 (SAS Institute, Cary, NC).
A total of 2028 patients were identified via the database search method, and 365 patients were subsequently deemed eligible for analysis on chart review. An additional 6 patients were added from the BC clinic who had not been identified via the database search. Therefore, a total of 371 patients comprised the final cohort.
Female patients comprised 96.5% of the cohort. With regard to the initial cancer diagnosis, the majority of patients had stage II or III disease (80.6%, Table 1). Patients with triple-negative (TN) and hormone-positive disease accounted for 19.0% and 74.1% of the cohort, respectively. While most metastatic diagnoses were made as a result of reported symptoms (77.6%), 3.2% were made with clinical examination findings and 4.1% by abnormal laboratory results. Diagnoses made incidentally via imaging done for other reasons accounted for 7.8% of the cohort. A total of 6 patients (1.6%) had a locoregional recurrence diagnosed initially, with metastatic findings discovered on subsequent staging scans.
Table 1Diagnostic Characteristics of 371 Initial and Metastatic Diagnoses
Initial Receptor Status
How Diagnosis of Metastasis Was First Suspected
Clinical exam by physician
Screening breast imaging in which local recurrence was found, prompting systemic imaging
Screening chest X-ray
Incidental on imaging
Laboratory results only
Abbreviations: ER = estrogen receptor; PR = progesterone receptor.
A similar proportion of patients presenting with symptoms was noted at scheduled versus acute-care visits (49.1% and 50.9%, respectively). Among those with symptoms, 40.4% had a clinical encounter with a nononcologic provider. Musculoskeletal pain was the most common symptom (33.7%); it was more frequently noted at scheduled visits (48.9%) compared to acute-care visits (26.0%, P < .01). Figure 2 depicts the subtypes of symptoms that were analyzed and the frequency with which they occurred.
Neither initial stage nor receptor status was associated with the presence of symptoms at diagnosis of metastasis (P = .58 and .68, respectively). Upon further analysis of specific types of symptoms, metastatic bone pain was not associated with any particular receptor subtype (P = .12). However, receptor status was associated with nervous system (NS) symptoms at diagnosis of metastasis (P = .01), with higher odds of NS symptoms in TN (OR = 3.02; 95% CI, 1.05-8.44) compared to estrogen receptor (ER)/progesterone receptor (PR)-positive, HER2− cases. HER2+ cases were not found to have a higher odds of NS symptoms compared to ER/PR+, HER2− cases (ER/PR+ HER2+: OR = 3.52 and 95% CI, 0.88-12.29; ER/PR− HER2+: OR = 3.82 and 95% CI, 0.79-14.97).
Initial stage and receptor status were associated with TTR (both P < .01). Those with stage III disease (vs. stage I, HR = 1.69 and 95% CI, 1.24-2.31) and TN BC (vs. ER/PR+, HER2−, HR = 2.52 and 95% CI, 1.90-3.35) had the shortest TTR. On multivariable analysis, initial stage, receptor status, age, and TTR were significantly associated with 10-year survival after diagnosis of metastasis (Table 2, Figure 3) whereas the presence of symptoms at diagnosis of metastasis was not (HR = 1.21, 95% CI, 0.86-1.70).
Table 2Multivariable Analysis for 10-Year Overall Survival
The aim of this study was to analyze the most common ways in which patients with previously treated BC present with metastatic disease in order to better understand the patterns of recurrence and the impact on survival. Although several studies have explored how the various molecular subtypes of BC recur as well as the risk factors that predict recurrence, to our knowledge, this is the first study to specifically explore the different presentations of symptoms at metastasis.
which likely reflects the increased recurrence rates that occur in patients with both receptor subtypes. As expected, most metastatic recurrences occurred in those with disease of a higher stage and grade at initial diagnosis, with patients with stage II and III cancers accounting for 80.6% of the cohort.
With regard to the diagnosis of metastatic recurrence, patients most commonly presented with a symptom that triggered further evaluation (77.6%). Pandya et al
drew a similar conclusion in their retrospective study of the earliest indicators of BC recurrence, finding that 54.3% of patients who experienced relapse presented with symptoms or with an abnormality detected by self-examination. Of note, a patient-detected breast abnormality was considered a symptom in our data analysis instead of a separate means by which the recurrence was diagnosed. Additionally, Pandya et al included patients with both metastatic presentations and locoregional recurrences. As a result, a significant proportion of recurrences were found by physical examination by the clinician (19.4%), which could account for the differences in cumulative percentages compared to our study. Analogous findings were noted in a retrospective review by Dewar and Kerr,
whereby 92 metastatic recurrences were reported, only one of which was asymptomatic, with the remainder of patients experiencing symptoms at relapse. In addition, more patients with metastatic symptoms presented at acute-care visits. Although we found that the overall number of patients with symptoms presenting at acute-care visits and scheduled visits was comparable, those with nonmusculoskeletal complaints were more frequently reported with the former. This may reflect the severity of nonmusculoskeletal symptoms that involve the cardiac, pulmonary, and NSs.
Historically, studies have shown that hormone-positive BC most commonly metastasizes to bone, and brain metastasis is more common in TN and HER2+ cancers.
Despite this, we found that metastatic bone pain was not associated with any particular receptor subtype. This may reflect the common finding of asymptomatic bony metastasis. In other words, although the site of metastasis may be associated with a specific receptor subtype, this does not always translate into the presence of symptoms. In contrast, however, receptor subtype was associated with NS symptoms at diagnosis of metastasis, with higher odds of symptoms in TN relative to ER/PR+, HER2− disease. This suggests that it may be relatively more common to have symptoms associated with NS metastasis versus bone metastasis.
Only a small percentage of metastatic diagnoses in our study were found incidentally on imaging performed for other reasons (7.8%), and even fewer by clinical examination (3.2%). However, in our study, 80.8% of patients were found to have evidence of metastatic disease during the clinical encounter itself (history/review of systems or physical examination). Thus, theoretically, the clinical encounter would suffice to detect a metastatic recurrence in most patients. Interestingly, a significant portion of patients presenting with symptoms (at scheduled or acute-care visits) had a clinical encounter with a nononcologic provider. Similarly, in a study performed in England, Grunfeld et al
found that almost half of patients with recurrences first presented to their general practitioners. While oncologists may be well aware of symptoms that should prompt further evaluation of recurrent cancer, other types of providers may not. We hope our study offers such providers guidance, so that they might modify their threshold of suspicion for recurrence depending on the characteristics of the initial diagnosis and symptoms subsequently reported. This is particularly important because one of the aims of survivorship care is to identify disease relapse early, before the patient experiences significant morbidity.
showed that the presence of visceral metastasis and/or brain metastasis was negatively associated with long-term survival of patients with metastatic BC, as was TN receptor status. In these studies, the presence or absence of symptoms in relation to location of metastasis was not examined. Our data showed that in the event that a patient experiences symptoms at metastatic recurrence, OS is not affected. The location of metastatic disease appears to be a more significant predictor of survival than symptoms.
Our study had several limitations that were potentially unavoidable. The determination of how a patient’s BC recurrence was first found (eg, via symptoms or incidentally on imaging) has not been described or standardized in the literature. We were therefore obliged to deduce such information by reviewing the clinicians’ notes in depth in addition to the reason documented for ordering particular radiologic images. At this time, no cancer surveillance mechanism in the United States systemically collects data on cancer recurrence, so some creativity was required to identify data elements that could be collected from TriNetX that were reflective of those patients who had experienced recurrence. Consequently, some patients who would have fit the criteria may have been missed. Last, we acknowledge that this study may be underpowered to detect a statistically significant difference in survival, irrespective of whether this difference would prove to be clinically significant.
Overall, we expect that our results will ease the anxiety of BC patients who are undergoing surveillance, as most recurrences will be diagnosed by reporting symptoms, and patients can be proactively involved in that regard. Moreover, they can be reassured by our findings, which showed that patients did not have a shorter survival as a result of presenting with metastatic symptoms. Further research on BC recurrence could potentially be performed using the database we have created to inform guidelines on improving survivorship care.
Clinical Practice Points
Current clinical guidelines do not support the use of routine surveillance imaging in follow-up care of BC patients, who often experience anxiety related to the possibility of cancer recurrence.
Previous studies have shown that BC recurrence is usually discovered as a result of symptoms.
Our study aimed to investigate methods of diagnosing recurrence, whether the method of diagnosis of metastasis resulted in a difference in survival, and whether certain BC subtypes were associated with differences in symptomatology.
Consistent with previous literature, we found that most patients present with symptoms at diagnosis of metastasis. We also found that having symptoms at diagnosis is not associated with 10-year survival.
Patients with TN BC were more likely to have NS symptoms at the time of diagnosis of metastasis, consistent with the fact that these patients have a higher chance of brain metastasis.
Our findings will, we hope, ease the minds of patients who are anxious about waiting until the onset of symptoms to obtain imaging.
Our results support the use of symptom-directed imaging and not surveillance imaging in the follow-care of BC patients. Additionally, our results provide guidance for providers for when to have a lower threshold to obtain symptom-directed imaging, based on baseline clinicopathologic factors.
The authors have stated that they have no conflict of interest.
Supported in part by the Holden Comprehensive Cancer Center Population Research and Biostatistics Cores ( P30 CA086862 ).
Metastatic breast cancer: therapeutic options according to molecular subtypes and prior adjuvant therapy.