Use and Yield of Baseline Imaging and Laboratory Testing in Stage II Breast Cancer

This study suggests that liver function tests, tumor markers, and routine body imaging are unnecessary for the initial evaluation of asymptomatic women with stage II breast cancer, even for subgroups of young women and patients with human epidermal receptor 2-positive breast cancer, triple-negative breast cancer, or stage IIB disease.

Abstract

Background.

Despite guideline recommendations, baseline laboratory testing and advanced imaging are widely ordered in clinical practice to stage asymptomatic patients with clinical stage II breast cancer (BC).

Materials and Methods.

A retrospective study at two academic centers in Boston, Massachusetts, between 2006 and 2007 explored the use, results, and implications of laboratory tests, tumor markers, and imaging in patients with clinical stage II BC.

Results.

Among 411 patients, 233 (57%) had liver function testing, 134 (33%) had tumor marker tests, and 237 (58%) had computed tomography (CT) as part of their initial diagnostic workup. Median age was 52 (range, 23–90 years). On multivariable analysis, young age, more advanced stage, and tumor subtype (human epidermal growth receptor-positive [HER2+] and triple-negative breast cancer [TNBC]) were significantly associated with baseline CT. The rate of detection of true metastatic disease with use of baseline staging imaging was 2.1% (95% confidence interval, 0.7%–5%). It was 2.2% (3 of 135) for estrogen receptor/progesterone receptor-positive disease, 1.9% (1 of 54) for HER2+ disease, and 2.1% (1 of 48) for TNBC. At 5 years of follow-up, 46 of 406 patients were diagnosed with metastatic breast cancer. Thirty-four of 46 (73.9%) who developed recurrent disease had imaging at their initial diagnosis, and of these, five had abnormalities on their initial imaging that was correlated with where they developed metastatic disease.

Conclusion.

In this cohort of women with stage II BC, staging imaging at diagnosis had a low yield in detecting distant metastases (2.1%). The detection rate was not higher with HER2+ disease or TNBC, despite the trend that patients with these subtypes were more likely to undergo imaging.

Implications for Practice:

Despite guideline recommendations, asymptomatic patients with stage II breast cancer (BC) often undergo staging imaging with computed tomography, bone scanning, or positron emission tomography. Physicians have often reported that they order imaging despite recommendations because they believe that younger patients or patients with more aggressive BC phenotypes, such as human epidermal receptor 2-positive BC or triple-negative BC, benefit from staging imaging. In this cohort of women younger than those in prior studies, the yield of detecting distant metastatic disease in patients with clinical stage II BC was very low and the detection rate was not higher in the presence of HER2-positive or triple-negative BC.

Introduction

National and international guidelines discourage radiographic staging of asymptomatic patients with early breast cancer (BC) beyond routine mammography (and ultrasonography as necessary). For example, the National Comprehensive Cancer Network (NCCN) guidelines suggest that for patients presenting with clinical stage I–IIB disease, a complete blood count, alkaline phosphatase, and liver function testing (LFT) should be performed as part of the routine workup but that chest, abdominal/pelvis, and bone imaging be confined only to patients with concerning signs or symptoms or as directed by abnormalities in laboratory test results [1]. Tumor markers (TMs) are not recommended for initial staging.

Despite these recommendations, routine use of advanced imaging to rule out distant metastases is common and has increased over time [2, 3]. In the Medicare population (age ≥65 years), approximately 16% of patients with a diagnosis of stage II BC in 2005 underwent preoperative computed tomography (CT), a rise from 7% in 1992 [2]. At Washington University in St. Louis, Missouri, up to 46% of patients (570 of 1,247) underwent staging with CT, bone scanning, or positron emission tomography (PET) between 1998 and 2012 [4]. In a more recent study from Ontario, Canada, in 2012, approximately 35% of patients with stage II BC underwent chest CT for staging and 40% had abdominal/pelvis CT [3]. Furthermore, although LFT is recommended by the NCCN to screen for occult metastatic disease, limited data support this recommendation [5].

Both patients and their physicians believe that laboratory testing and advanced imaging are an important part of their initial workup [6]. In a survey of 245 women with early BC, the majority expressed an expectation that imaging be performed as part of their initial evaluation: Fifty-seven percent indicated they would want imaging done even if the chance of detecting metastases was ≤10%, and 70% noted they would be uncomfortable with their doctor if imaging was not ordered [6]. Although similar survey data from physicians are sparse, physicians frequently order staging tests despite the NCCN guidelines and have expressed concern that the guidelines may not be appropriate for certain subsets of patients, for example, those with node-positive (stage IIB) disease, particularly the human epidermal receptor type 2 (HER2)-positive (HER2+) or triple-negative BC (TNBC) subtype.

We undertook a study to describe the use, results, and implications of routine laboratory and radiographic testing in patients presenting with clinical stage II BC at two academic practices in Boston, Massachusetts, between 2006 and 2007. We focused on clinical stage II disease because our pilot work showed that this is the group of patients with the most variation in practice. By restricting our cohort to patients diagnosed between 2006 and 2007, we were able to collect data on those treated with modern-era adjuvant therapy and who had at least 5 years of follow-up to accurately determine whether any initial equivocal findings later evolved to metastatic disease. Furthermore, we evaluated factors associated with the decision to order staging studies and whether such factors were associated with true differences in the presence or absence of metastatic involvement at diagnosis.

Materials and Methods

Study Description

This retrospective, descriptive study was conducted at two academic centers: Beth Israel Deaconess Medical Center (BIDMC) and Dana-Farber Cancer Institute (DFCI) in Boston, Massachusetts. The study was approved by the institutional review board at the Dana-Farber/Harvard Cancer Center.

Patients were included if they were diagnosed with stage II BC, defined by the American Joint Committee on Cancer, seventh edition, between January 1, 2006, and December 31, 2007. Eligible patients were identified by the BIDMC cancer registry and using the Clinical Operations and Research Information System at DFCI. Patients needed to have their long-term cancer follow-up at BIDMC or DFCI. For all identified cases, charts were abstracted by using each institution’s electronic medical records. For complete details, see the supplemental online Appendix.

Statistical Analysis

Logistic regression models were used to assess the association between patients’ baseline characteristics and physicians’ use of staging imaging and laboratories. Univariate logistic regression models were used on each patient and tumor characteristic variable. Variables were used in multivariate logistic regression model if the p value from the univariate logistic regression model was <.2. A backward selection based on Akaike information criterion was used to build the final multivariate logistic regression model. Because brain CT/magnetic resonance imaging (MRI) or PET was ordered for only a few patients, their association with baseline characteristics was tabulated and odds ratios (ORs) with 95% confidence intervals (CIs) from univariate logistic regression were given.

Associations between treatment assignment and staging imaging/laboratories were assessed among patients who were not restaged and had complete information on treatment assignment by using the Fisher exact test. If patients had confirmed metastases or second primary cancers at initial diagnoses, patients were excluded from the analysis. Decision of ordering advanced staging imaging/laboratory tests was cross-tabulated with medical subspecialties and oncology center locations. Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, http://www.sas.com).

Results

Use of Staging Laboratories and Imaging

A total of 547 charts were reviewed; among these, 411 patients met our eligibility criteria and were included in the dataset. In the cohort, 190 of 411 (46.2%) women were ≤50 years of age at the time of their diagnosis. Two hundred thirty-three (57%) patients underwent LFT. Of those who had LFT, 20.2% (47 of 233) did not receive chemotherapy; among those who did not undergo LFT, 25% (44 of 176) did not receive chemotherapy. One hundred thirty-four (33%) patients had TM tests, 237 (58%) had body CT (torso, chest, or abdomen), 17 (4%) had brain CT or MRI, and 38 (9%) had PET (Table 1). Table 2 shows the number of patients undergoing LFT, TM testing, and advanced imaging by modality for each institution.

Table 1.

Number of patients with stage II breast cancer undergoing liver function testing, tumor marker testing, or advanced imaging

Table 2.

Number of patients at BIDMC vs. DFCI with stage II breast cancer undergoing liver function testing, tumor marker testing, or advanced imaging, by modality

Patient and Tumor Factors Associated With Staging Evaluations

Tumor subtype was significantly associated with the decision to order LFT (p = .04) (supplemental online Table 1). Patients with TNBC were more likely to have LFT compared with patients with estrogen receptor (ER)/progesterone receptor (PR)-positive BC (OR, 2.2; 95% CI, 1.2–4.2). Stage and histology also tended to be associated with the decision to order LFT, although the associations were not statistically significant (p = .05 for each).

Tumor size, histologic type, and race were significantly associated with the decision to order TM tests (supplemental online Table 2). Compared with patients who had tumors ≤2 cm, those with tumors 2–5 cm or with >5 cm were more likely to have TMs checked (OR, 1.3 [95% CI, 0.8–2.0] and 4.2 [95% CI, 1.3–12.8], respectively; p = .04). Compared with patients with ductal-only BC, patients with both ductal and lobular histologic types were less likely to receive tumor makers tests (OR, 0.3; 95% CI, 0.1–0.7), and patients with nonductal/lobular BC were more likely to receive TM tests (OR, 4.0; 95% CI, 1.6–10.5) (p < .001).

Age, stage, and tumor subtype were significantly associated with the decision to order body CT (Table 3). Physicians were more likely to order body CT for younger patients than for patients aged 71 years or older (p = .005). In addition, physicians were 2.9 times (95% CI, 1.8–4.7) more likely to order body CT for patients with stage IIB disease than for patients with stage IIA disease (p < .001). Physicians were also more likely to order body CT for HER2+ disease and TNBC than for ER/PR-positive disease (OR, 3.0 [95% CI, 1.5–5.8] and 2.1 [95% CI, 1.1–3.9], respectively; p = .002).

Table 3.

Univariate and multivariate logistic regression of baseline characteristics and body computed tomography

Only 20 of 411 (4.9%) patients in our cohort reported systemic symptoms at diagnosis, and 19 of these underwent some type of advanced imaging. Patient report of systemic symptoms was not associated with the decision to order LFT, TM tests, or CT (Table 3, supplemental online Tables 1 and 2).

Only 38 (9%) patients had PET, and 17 (4%) patients had head CT or brain MRI. Of the PET CT examinations performed, 18 of 38 (47.4%) were performed to better elucidate an abnormal finding seen on prior imaging, such as bone scanning or CT, whereas 20 of 38 (52.6%) of patients who underwent PET did so upfront as part of their staging evaluation. Among the 17 patients who underwent brain imaging, only 4 had symptoms that may have prompted these imaging evaluations. ORs and 95% CIs of having PET and head imaging by baseline characteristics were derived from univariate logistic regression models (supplemental online Tables 3 and 4, respectively). Because of the small number of events, no formal test was conducted and a multivariate model was not built.

Provider and Treatment Factors Associated With Staging Evaluations

Because of the referral nature of the two academic practices and the multidisciplinary nature of BC care, we recorded the medical specialty and institution of the ordering provider. We found that most laboratory and imaging tests were ordered by medical oncologists at the Harvard Cancer Center or affiliated institutions (Table 4). We did observe variation between providers on who ordered laboratory and imaging tests.

Table 4.

Laboratory and imaging testing by medical subspecialty and oncology center locations

We also evaluated the effect of receipt of neoadjuvant chemotherapy on the likelihood of routine staging evaluation (supplemental online Table 5). A total of 406 patients were tabulated in supplemental online Table 5 after exclusion of 5 patients diagnosed with metastatic BC (MBC; see discussion later in text). Patients who planned to receive neoadjuvant chemotherapy were more likely to have LFT (p = .04), body CT (p < .001), head CT or brain MRI (p < .001), and PET (p = .02) than patients who underwent surgery without first receiving preoperative chemotherapy. Clinical trial participation was also associated with increased odds (OR, 2.0 [95% CI, 1.2–3.5]; p = .01) of having CT.

Results of Staging Evaluations

Overall, approximately a third of torso CT examinations identified lesions for which additional imaging with another modality or interval imaging was recommended. Only 5%–10% of bone scans triggered a recommendation for additional testing (Table 5).

Table 5.

Number of torso computed tomographic and bone scan findings that prompted evaluation with additional imaging or interval imaging

Although CTs and bone scans identified many lesions for which distant metastases could not be excluded, only 12 patients underwent diagnostic biopsy. Among the patients who underwent biopsy, 4 were diagnosed with MBC, 1 was diagnosed with a second primary cancer (lung), and 7 had normal biopsy results. A fifth patient was diagnosed with MBC on the basis of imaging but could not undergo biopsy. In summary, 5 patients who presented with clinical stage II disease were diagnosed with de novo metastatic BC on the basis of staging imaging and laboratory work. The patient, clinical, and tumor characteristics of these patients are presented in Table 6. Among the 5 patients, 3 were ER/PR-positive, 1 was HER2+, and 1 had TNBC.

Table 6.

Characteristics of patients diagnosed with de novo metastatic disease after staging imaging and laboratory tests

Regarding the laboratory tests, approximately 58% of patients had alanine aminotransferase (ALT), aspartate aminotransferase, alkaline phosphatase, and/or total bilirubin testing (Table 7). The results of only a small fraction of these tests were positive, and despite an abnormal ALT result for one of the patients diagnosed with de novo metastatic disease, she did not have metastatic liver disease; this abnormal result was unrelated to her cancer diagnosis because it normalized at a repeat evaluation and was not initially correlated with disease progression. Approximately 30% of patients in our cohort had carcinoembryonic antigen (CEA) or CA27.29 testing. Of the 19 CEA and CA27.29 tests with abnormal results, one was correlated with MBC in a patient, but this abnormal result normalized during the patient’s clinical course despite disease progression; this suggests that the abnormal finding was a false-positive result.

Table 7.

Laboratory test results

Overall, the rate of detection of true metastatic disease at the time of initial diagnosis based on baseline staging imaging was 2.1% (5 of 237; 95% CI, 0.7%–5.0%). It was 2.2% (3 of 135; 95% CI, 0.5%–6.4%) among ER/PR-positive patients, 1.9% (1 of 54; 95% CI, 0%–9.9%) among HER2+ patients, and 2.1% (1 of 48; 95% CI, 0.1%–11.1%) among patients with TNBC.

At approximately 5 years of follow-up, 46 of 406 patients were diagnosed with MBC, including 8 patients with HER2+ disease, 13 with TNBC, and 25 with ER/PR-positive, HER2-negative disease. Twelve of the 46 patients did not have any type of staging imaging at their initial diagnosis, whereas 73.9% (34 of 46) did. Of the 34 patients who had baseline staging imaging and developed recurrent BC, 5 had abnormalities on their initial staging that corresponded to where they later developed metastatic disease. The disease-free interval ranged from 5 months to 7.5 years. The characteristics of these patients are presented in supplemental online Table 6.

Discussion

Despite studies reporting low yield of laboratory testing and imaging for patients with early BC, the use of these tests remain high [4, 7–11]. We evaluated practice patterns for use of laboratory testing and imaging in patients with stage II breast cancer within two academic practices in the U.S. We found that more than half of patients had laboratory tests and/or advanced imaging ordered; however, the rate of detection of true metastatic disease was low across all breast cancer subtypes and in both node-negative and node-positive patients.

The NCCN continues to recommend liver function tests as part of the routine workup of early breast cancer patients. In our dataset of more than 400 patients, LFT yielded no true-positive results. Our results are consistent with those of a recent study in more than 1,000 women with early BC, of whom 24.7% had elevated LFT results; however, this had only an 8.0% positive predictive value for detecting metastatic disease. Of note, LFT contributes to approximately $25.7 million in health expenditure annually [7]. Although baseline LFT has value for chemotherapy dosing, our data strongly suggest that these tests should not be used for staging purposes. In our cohort, 20.2% of patients who had LFT did not receive neoadjuvant or adjuvant chemotherapy.

With respect to imaging, our data contribute to previous work indicating that the yield of CT to detect metastases in early BC is approximately 1% [4, 8–10]. Despite these data, many physicians have continued to order imaging; our data demonstrate that in practice, factors such as nodal status and tumor subtype (e.g., HER2 or TNBC subtype) influence the likelihood of baseline testing. Some of this may relate to perceived limitations of prior studies, including patient characteristics (e.g., studies from the Surveillance, Epidemiology, and End Results-Medicare database have included only patients age ≥65 years), exclusion of patients receiving neoadjuvant chemotherapy, unavailability of tumor subtype, and lack of long-term follow-up to allow understanding of the natural history of indeterminate lesions [2–4, 8–11].

Our patient population is younger than that seen in the average cancer center in the United States: In our cohort, nearly half of patients were ≤50 years of age. Unlike prior work [4], our study also included patients who received neoadjuvant chemotherapy. We also collected information regarding histological confirmation of metastatic disease. Finally, we followed patients out to 5 years, which allowed us to assess the natural history of indeterminate lesions and lesions that did not prompt an initial biopsy.

Although HER2+ BC and TNBC are known to be more aggressive BC subtypes, we found no evidence that patients with these subtypes should have different staging evaluations than those with ER/PR-positive BC. In fact, although a greater portion of patients with HER2+ BC and TNBC underwent extra testing as part of their staging evaluation, the percentage identified to have distant metastases did not differ by subtype; 2.2% (95% CI, 0.5%–6.4%) identified to have de novo metastatic disease were ER/PR-positive and HER2-negative, 1.9% (95% CI, 0%–9.9%) to have HER2+ disease, and 2.1% (95% CI, 0.1%–11.1%) to have TNBC. Although the overall rate is higher in this stage II cohort than what had previously been reported in the literature (1.8% for Washington University [4] and 0% for the Korean study [9]), this rate may be explained by our younger population and/or inclusion of patients who received neoadjuvant therapy, but it nonetheless remains low and does not justify the use of CT in this population.

At 5 years of follow-up, 11.3% (46 of 406; 95% CI, 8.4%–14.8%) of the patients diagnosed with stage II disease developed recurrent metastatic BC. This occurred even although most patients (34 of 46) had imaging performed at diagnosis. Only a small subset of this group (5 patients total [1.2%]) had evidence of distant metastases on their baseline imaging when reviewed retrospectively. It is not at all clear whether earlier recognition of metastatic disease would have altered their overall outcomes.

Finally, although our study was not formally designed to compare practice patterns between individual providers, qualitatively we did note substantial variation between providers. These variations point to an opportunity to formulate institutional practices to reduce the use of unnecessary testing overall.

This study was limited to two institutions in Boston, Massachusetts, and results may not be generalizable to other centers nationally. The study included patients diagnosed with stage II BC in 2006 and 2007 and therefore may not reflect current practice patterns. The American Society of Clinical Oncology's “Choosing Wisely” recommendations were released in 2012 and discouraged the use of imaging for staging purposes in early BC, so it is possible that practice patterns have changed at these institutions [12]. However, this study does provide data that additional testing, including LFT, TM testing, and imaging for patients with newly diagnosed stage II BC will rarely identify metastatic disease and change clinical management, and this is true even among young patients and those with aggressive BC subtypes, such as HER2+ disease and TNBC.

One argument for ordering staging imaging is to detect distant metastases and prevent overtreatment with neoadjuvant/adjuvant therapies. Yet we do not know whether in a small subset of patients (e.g., in HER2+ breast cancer) highly effective systemic approaches might lead to long-term disease-free survival, even if they have a small volume of macroscopic disease visible on CT or bone scanning. Furthermore, the neoadjuvant/adjuvant regimens given to HER2+ patients in 2015 are similar (although not identical) to what these patients would receive if they were diagnosed with metastatic disease. It is only in the ER/PR-positive patients with clinical stage II BC that reclassifying them as stage IV may spare them upfront treatment with chemotherapy—even then, the number of clinical stage II ER/PR-positive patients who truly have metastatic disease and may be overtreated with chemotherapy if staging imaging is not performed is relatively small, particularly because we are also treating fewer patients with chemotherapy (including those with node-positive disease), due in part to the introduction of OncotypeDx (Genomic Health, Redwood City, CA, http://www.genomichealth.com/) and other predictive markers.

Conclusion

We provide compelling evidence that LFT, tumor markers, and routine body imaging are unnecessary for the initial evaluation of asymptomatic women with stage II BC, even for subgroups of young women and patients with HER2+ breast cancer, triple-negative breast cancer, or stage IIB disease.

See http://www.TheOncologist.com for supplemental material available online.

This article is available for continuing medical education credit at CME.TheOncologist.com.

Author Contributions

Conception/Design: Brittany L. Bychkovsky, Chiara Battelli, Mary Jane Houlihan, Steven E. Come, Nancy U. Lin

Provision of study material or patients: Brittany L. Bychkovsky, Tsai-Chu Yeh, Nancy U. Lin

Collection and/or assembly of data: Brittany L. Bychkovsky, Jazmine Sutton, Laura Spring, Jennifer Faig, Ibiayi Dagogo-Jack, Tsai-Chu Yeh, Nancy U. Lin

Data analysis and interpretation: Brittany L. Bychkovsky, Hao Guo, Jazmine Sutton, Laura Spring, Jennifer Faig, Ibiayi Dagogo-Jack, Mary Jane Houlihan, Steven E. Come, Nancy U. Lin

Manuscript writing: Brittany L. Bychkovsky, Hao Guo, Jazmine Sutton, Laura Spring, Jennifer Faig, Ibiayi Dagogo-Jack, Chiara Battelli, Mary Jane Houlihan, Steven E. Come, Nancy U. Lin

Final approval of manuscript: Brittany L. Bychkovsky, Hao Guo, Jazmine Sutton, Laura Spring, Jennifer Faig, Ibiayi Dagogo-Jack, Chiara Battelli, Mary Jane Houlihan, Tsai-Chu Yeh, Steven E. Come, Nancy U. Lin

Disclosures

The authors indicated no financial relationships.