Fax: (206) 598-4496
Residual tumor uptake of [99mTc]-sestamibi after neoadjuvant chemotherapy for locally advanced breast carcinoma predicts survival
Version of Record online: 6 JAN 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 4, pages 680–688, 15 February 2005
How to Cite
Dunnwald, L. K., Gralow, J. R., Ellis, G. K., Livingston, R. B., Linden, H. M., Lawton, T. J., Barlow, W. E., Schubert, E. K. and Mankoff, D. A. (2005), Residual tumor uptake of [99mTc]-sestamibi after neoadjuvant chemotherapy for locally advanced breast carcinoma predicts survival. Cancer, 103: 680–688. doi: 10.1002/cncr.20831
- Issue online: 3 FEB 2005
- Version of Record online: 6 JAN 2005
- Manuscript Accepted: 18 OCT 2004
- Manuscript Revised: 8 OCT 2004
- Manuscript Received: 13 AUG 2004
- National Institutes of Health. Grant Numbers: CA72064, CA42045
- breast carcinoma;
- locally advanced breast carcinoma;
- neoadjuvant chemotherapy
Studies utilizing serial [99mTc]-sestamibi (MIBI) scintimammography have reported accurate prediction of tumor response in patients with locally advanced breast carcinoma (LABC) undergoing neoadjuvant chemotherapy. The pathologic response of LABC to presurgical treatment regimens is a prognostic indicator of survival. The authors tested whether MIBI uptake posttherapy predicted survival.
Sixty-two patients with LABC underwent MIBI scintimammography just before chemotherapy and 2 months after treatment initiation. An additional MIBI scan was performed if treatment lasted > 3 months. The affected breast was imaged within 10 minutes after injection to reflect early uptake, which the authors have shown to be related to tumor blood flow. MIBI uptake was quantified using the lesion-to-normal breast (L:N) ratio. Most patients (93%) received weekly dose-intensive doxorubicin-based treatment. Disease-free survival (DFS) and overall survival (OS) were compared with posttherapy primary MIBI uptake and with other established prognostic factors for neoadjuvantly treated LABC, namely, primary tumor pathologic response and posttherapy axillary lymph node status.
Patients with high uptake on the last observed MIBI scan (i.e., the L:N ratio was greater than the median value) had poorer DFS and OS (P < 0.01 and P = 0.01, respectively). Residual MIBI uptake retained independent prognostic significance in preliminary multivariate analysis that included other established prognostic markers.
High primary breast tumor MIBI uptake after neoadjuvant chemotherapy predicted poor survival, suggesting serial MIBI imaging may provide a useful quantitative surrogate end point for neoadjuvant chemotherapy trials. Given the association between MIBI uptake and tumor blood flow, this prognostic capability may be related to retained tumor vascularity after treatment. Cancer 2005. © 2005 American Cancer Society.
Breast carcinoma is the second leading cause of cancer mortality for all women in the United States and is the leading cause of cancer mortality among women 20–59 years old.1 Of new cancer cases diagnosed among U.S. women in 2003, 32% are expected to be breast carcinoma.1 Although greater utilization of mammography screening has increased detection of early-stage breast carcinoma, up to 28% of patients present with primary tumors that are > 5 cm, with or without fixed axillary lymph nodes, or skin/chest wall invasion, defined as locally advanced breast carcinoma (LABC).1
Neoadjuvant or presurgical chemotherapy is often used to manage LABC with the aim of reducing tumor volume to improve surgical resectability, treating preexisting microscopic systemic disease, and assessing in vivo tumor response. Studies have shown that no residual macroscopic tumor in the breast and the presence and number of metastatic axillary lymph nodes after neoadjuvant treatment for LABC are prognostic of disease-free survival (DFS) and overall survival (OS).2–4
We and others have reported that [99mTc]-sestamibi (MIBI) scintimammography (SMM) accurately predicts response to neoadjuvant chemotherapy in patients with LABC.5–7 Changes in MIBI uptake measurements after 2 months of therapy and at the completion of therapy predicted the degree of pathologic response.6 We also investigated MIBI uptake mechanisms in breast tumors and reported uptake to be related closely to tumor blood flow, as measured by [15O]-water positron emission tomography scans (PET).8 We also found that subjects with increased tumor blood flow after neoadjuvant therapy, measured by [15O]-water PET, were more likely to have a poorer pathologic response and subjects with low tumor blood flow after therapy had a lower risk for disease recurrence.9, 10
We present follow-up data in a patient population with LABC treated by neoadjuvant therapy and studied by serial MIBI imaging studies. Our overall hypothesis is that quantitative in vivo measurements of MIBI uptake in breast tumors may provide useful predictive prognostic information, complementary to established prognostic indicators. The aim of the current study is to determine whether radiotracer imaging with MIBI posttherapy is predictive of survival among patients with LABC. As primary tumor response to presurgical chemotherapy is indicative of survival and a decline in MIBI uptake after serial SMM predicts primary tumor response, we hypothesize that serial MIBI uptake measurements are prognostic of survival.
MATERIALS AND METHODS
Patients presenting to the University of Washington Breast Cancer Specialty Center with histologically confirmed breast carcinoma undergoing induction chemotherapy were eligible for the study. Patients were clinically staged according to the TNM classification of malignant tumors and judged to have LABC.11 The enrollment period was October 1994 to March 2002. Inclusion and exclusion criteria have been described.6, 10 Exclusion criteria included patients with an excisional biopsy specimen > 50% of the original primary tumor, identification of unsuspected distant metastases, and nonsurgical candidates for tumor resection. Seventy-one patients were enrolled initially in the study. Nine patients were excluded from the analysis as follows: five had a substantial portion of their breast tumor excised in the diagnostic biopsy; one patient was also a heart transplant candidate who was unable to undergo chemotherapy treatment; one patient recently had ceased breast-feeding at the time of the baseline study and images were noninterpretable; no follow-up was available for one patient; and one noncompliant patient had a significant delay from the final MIBI scan to surgery (33 weeks). Thus, 62 subjects were included in this cohort for analyses. Signed informed consent for imaging and follow-up was obtained according to University of Washington human subjects committee guidelines.
Pretherapy Clinicopathologic Parameters
Patient age, menopausal status, tumor size, and clinical axillary lymph node status as assessed by the referring oncologist were recorded before the start of neoadjuvant chemotherapy. Tissue biopsy samples were obtained by fine-needle aspiration biopsy, incisional biopsy, or core needle biopsy. Measurements of tumor histologic grade and in vitro assays to assess estrogen receptor (ER) expression, c-erb-b2 (HER-2/neu) expression, p53 overexpression, and Ki-67 (MIB1) index of tumor proliferation were performed as previously reported.9
Pathologic response to therapy was determined by standard definition. A macroscopic complete response (mCR) was defined as no evidence of tumor in the surgical specimen after macroscopic inspection with confirmation by histologic examination.2, 3, 12 Patients with macroscopic abnormalities after pathologic evaluation and minimal evidence of invasive tumor after histologic analysis were considered to have a mCR. mCRs were assessed further for microscopic evidence of invasive carcinoma by microscopic examination of the specimen. Patients with no evidence of microscopic residual disease were classified as pathologic complete responders (pCR) and as pathologic nonresponders (pINV) otherwise.13
Serial MIBI SMM studies were performed before the start of chemotherapy, 2 months after therapy, and before surgery if treatment continued for > 3 months as previously described.6 Patients received 20–30 mCi (740–1110 Mbq) MIBI (Bristol-Meyers Squibb Imaging, North Billerica, MA) intravenously in the arm contralateral to the affected breast. The gamma cameras used for image acquisition included a single-head Geneal Electric XCT or dual-head General Electric MG gamma camera with low-energy, high-resolution collimation (General Electric, Waukesha, WI). Images were acquired into a 256 × 256 matrix and the energy photopeak was centered over 140 keV with a 13% window. Patients were imaged in the prone lateral position without breast compression, beginning with the lesion side first, 5–10 minutes after the injection was administered.14 All images were acquired for a preset time of 10 minutes. Images were analyzed according to methods reported previously utilizing a lesion-to-normal breast uptake ratio (L:N). The intraobserver and interobserver reproducibility of the L:N measurements are 11% and 18%, respectively.15
The primary end points considered in the study were breast carcinoma recurrence and mortality due to breast carcinoma. Local disease recurrence was defined as invasive disease limited to the ipsilateral breast, chest wall, or axillary lymph nodes, and distant disease recurrence was defined as metastases to other parts of the body. For disease-free analysis, a subject with no history of disease recurrence was censored at the time of death or at the time of last clinical follow-up. Subjects with competing risks were censored at the time of diagnosis of an unrelated carcinoma or at the time of death if due to a cause other than breast carcinoma. DFS was calculated from the date of surgery and OS was calculated from the date of breast carcinoma diagnosis. For the current analysis, chart review for patient clinical follow-up dates and disease status was completed as of October 15, 2003.
To test whether MIBI uptake measurements were associated with DFS and OS, Kaplan–Meier plots were constructed with continuous L:N ratio variables dichotomized above and below the median. The log-rank test was employed to determine differences between survival curves. DFS and OS were compared with established primary prognostic factors of mCR and residual axillary lymph node metastases. Axillary lymph node status was dichotomized to 0–3 positive lymph nodes versus ≥ 4 positive lymph nodes based on a previously published analysis.4 Secondary established prognostic factors for breast carcinoma included ER status, HER-2/neu gene overexpression, and the Ki-67 proliferative index.16–18 Univariate analyses also were performed using the Cox proportional hazards model. Relative risks (hazard ratios [HR]) were estimated, 95% confidence intervals (95% CI) were calculated, and significance was assessed by the Wald test.19 Based on univariate analysis and a priori considerations of potential confounding or precision variables, a model was derived for preliminary multivariate analyses and examined for best fit by likelihood ratio (LR) tests. All statistical tests were two-sided and P ≤ 0.05 was significant. A global test of the proportional hazards assumption was performed to ensure constancy of relative risks over time. Stata statistical software (Version 7.0) was used for all analyses (Stata, College Station, TX).
Sixty-two patients diagnosed with LABC were included in the study. Of these patients, 52 had infiltrating ductal carcinoma, 4 had infiltrating ductal carcinoma with lobular features, and 6 had infiltrating lobular carcinoma. The mean age of the patients at the time of breast carcinoma diagnosis was 48 years (range, 30–85 years) and the average tumor size at presentation was 5.2 cm (range, 1.9–11.0 cm). TN status was as follows: 1 of 62 (2%) patients had T1N2 disease, 5 of 62 (8%) patients had T2N0 disease, 3 of 62 (5%) patients had T2N1 disease, 2 of 62 (3%) patients had T2N2 disease, 7 of 62 (11%) patients had T3N0 disease, 24 of 62 (39%) patients had T3N1 disease, 1 of 62 (2%) patients had T3N2 disease, 1 of 62 (2%) patients had T4N0 disease, 15 of 62 (24%) patients had T4N1 disease, and 3 of 62 (5%) patients had T4N2 disease. The 9 patients with T1–2N0–1 presented with tumors that involved a significant portion of a small breast. Eight of 19 (42%) T4 carcinomas were inflammatory. Approximately 60% of all patients had clinically palpable lymph nodes (Table 1). A weekly dose-intensive, doxorubicin-based chemotherapy regimen was administered to 57 of 62 (93%) patients. Three of 62 (5%) patients received cyclophosphamide, methotrexate, and 5-fluorouracil (5-FU), 1 of 62 (2%) patients received paclitaxel/vinerelbine, and 1 of 62 (2%) patients received 5-FU only. Treatment regimens were described in greater detail in an earlier report.10 Surgical management after chemotherapy comprised total mastectomy and axillary lymph node dissection (ALND) in 51 of 62 (82%) patients and 11 of 62 (18%) patients received lumpectomy and ALND.
|Characteristics||(%) Percent of patients|
|Age at diagnosis (yrs)|
|Clinical tumor classification|
|Clinical lymph node classification|
|Clinically palpable lymph nodes||60|
Twenty-four of 62 (39%) patients had a pathologic mCR to neoadjuvant chemotherapy and 38 of 62 (61%) patients had a response other than mCR. Of the 24 patients with an mCR, 13 (21% overall) had microscopic evidence of residual invasive disease (pINV) and 11 (18% overall) had pCR. Eighteen of 62 patients (29%) had negative axillary lymph nodes after neoadjuvant chemotherapy. The median number of positive axillary lymph nodes in the remaining 44 patients was 2 (range, 1–32 lymph nodes).
Thirty-one patients were imaged with MIBI at baseline before chemotherapy treatment and 2 months after treatment initiation (median, 2.3 months). Thirty-one patients underwent additional MIBI imaging near the completion of chemotherapy or just before surgery (i.e., a median of 3.5 months from baseline study). MIBI image examples are depicted in Figure 1. The median baseline MIBI L:N ratio was 4.4 (range, 2.4–15.4), the median 2-month MIBI L:N ratio was 3.0 (range, 1.0–10.7), and the presurgical (final) L:N ratio was 2.8 (range, 1.0–9.3). There was a significant difference in the mean percent change in the MIBI L:N uptake ratio (baseline vs. 2 months, ± the standard error of the mean) for mCR (−50% ± 6%) versus other than mCR (−18% ± 5%; P < 0.001; Fig. 2), consistent with previous results.6
The median follow-up time for DFS was 33 months (range, 0–106 months) with 19 tumor recurrences. Of the tumor recurrences, 6 were local and 13 were distant metastases. The OS median follow-up time was 50 months (range, 11–110 months) with 19 deaths recorded. Fifteen of 19 deaths were confirmed due to breast carcinoma, 2 deaths were probable, and 2 deaths were due to other causes. Deaths due to other causes included complications from heart disease and metastatic head and neck squamous cell carcinoma. For the entire cohort, the 75% DFS probability time point was 32.1 months and the median DFS was > 106 months. The 75% OS probability time point for the cohort was 55 months and the median OS was > 110 months.
A poor primary tumor response to treatment (other than mCR) was associated with a 3-fold increase in disease recurrence risk (HR = 3.1, 95% CI = 0.9–10.8). Patients who achieved an mCR were more likely to remain free of disease compared with patients with macroscopic residual disease after therapy (Fig. 3). However, the difference was of marginal significance (P = 0.08). Although patients with ≥ 4 positive lymph nodes were nearly twice as likely to have disease recurrence or die of breast carcinoma compared with patients with 0–3 positive lymph nodes, the findings were nonsignificant in this small cohort (P = 0.20; Tables 2 and 3). There was a trend for poorer OS among patients with a response other than mCR compared with mCR pathologic partial or nonresponse compared with patients with an mCR (HR = 2.6, 95% CI = 0.7–9.4, P = 0.11; Fig. 4). There was no significant survival difference (DFS or OS) among patients who had a pCR to neoadjuvant therapy (referent group, n = 11) versus patients with a pINV (n = 51): DFS (HR = 1.4, 95% CI, = 0.3–6.2; P = 0.63) and OS (HR = 1.6, 95% CI = 0.3–7.0; P = 0.53).
|Characteristics||No. (%)||Log–rank P value||Hazard ratio||95% CI||Univariate Cox P value|
|Age at diagnosis||0.04||0.9||0.9–1.0||0.07|
|Estrogen receptor status||0.12|
|HER-2/neu gene status||0.66|
|Ki-67 proliferative indexa||0.02|
|Other than high||24 (45)||1.0||—||—|
|Primary tumor pathologic response||0.05|
|< mCR||38 (61)||3.1||0.9–10.8||0.08|
|Residual positive lymph nodes||0.19|
|≥ 4||25 (40)||1.8||0.7–4.8||0.20|
|Two-month MIBI L:N uptake ratio†||0.05||1.2||1.0–1.4||0.03|
|Final MIBI L:N uptake ratio†||0.001||1.3||1.1–1.6||0.001|
|Characteristics||Log–Rank P value||Hazard ratio||95% CI||Univariate Cox P value|
|Age at diagnosis||0.82||1||0.9–1.0||0.51|
|Estrogen receptor status||0.14|
|HER-2/neu gene status||0.23|
|Ki-67 proliferative indexa||0.29|
|Other than high||1.0||—||—|
|Primary tumor pathologic response||0.10|
|Residual positive lymph nodes||0.16|
|Two-month MIBI L:N uptake ratiob||0.03||1.1||0.9–1.4||0.09|
|Final MIBI L:N uptake ratiob||<.01||1.2||1.0–1.5||0.01|
Kaplan–Meier survival curves for MIBI L:N ratios were estimated by dichotomizing above and below the median values. Patients with high MIBI uptake (L:N ≥ 3.0, n = 31) after 2 months of treatment had poorer DFS (HR = 1.2, 95% CI, = 1.0–1.4; P = 0.03) and poorer OS (HR = 1.1, 95% CI, = 0.9–1.4; P = 0.09). MIBI uptake from studies performed before surgery (final MIBI) was also prognostic of DFS and OS. Patients with a presurgical L:N ratio greater than the median (L:N ≥ 2.8, n = 32) had a 35% increased risk of disease recurrence (HR = 1.3, 95% CI, = 1.1–1.6; P < 0.01) and a 25% increased risk of mortality (HR = 1.2, 95% CI, = 1.0–1.5; P = 0.01).
There was not a statistically significant difference in DFS or OS for subjects with primary tumors that were ER negative or HER-2/neu positive (DFS: P = 0.13 and P = 0.66, respectively; OS: P = 0.15 and P = 0.24, respectively). Individuals with a high Ki-67 proliferative index pre-therapy were 3 times as likely to experience disease recurrence (HR = 3.0, 95% CI, = 1.0–8.9; P = 0.03). However, data for this index marker were missing for 14% of patients.
To address whether the inclusion of the two probable deaths due to breast carcinoma as events had an effect on the observed association between MIBI uptake and OS, we performed a secondary analysis and censored these two patients at the time of death. The secondary analysis did not alter the association between the variables examined. 95% CIs were slightly wider, however, and the HR values were similar.
To determine whether MIBI uptake measurements significantly contributed to survival outcomes, we performed preliminary multivariate analysis including the established prognostic indicators of survival in the model (e.g., age at time of breast carcinoma diagnosis, ≤ 3 positive axillary lymph nodes vs. ≥ 4 lymph nodes, and mCR vs. other than mCR). The final (presurgical) L:N ratio, used as a continuous input variable, independently remained a significant predictor of disease recurrence likelihood for DFS (HR = 1.4, 95% CI, = 1.0–1.8; P = 0.01). Model comparison demonstrated MIBI to be a significant addition to the model (LR test, P < 0.01). The final L:N ratio did not remain a significant independent predictor of OS in the multivariate analysis (HR = 1.1, 95% CI, = 0.9–1.5; P = 0.24) in our small series with a limited number of patient deaths.
In our study utilizing MIBI SMM to monitor and predict response to presurgical chemotherapy, we observed that subjects with high residual MIBI uptake in breast tumors after 2 months of chemotherapy and posttreatment had a significantly increased risk of breast tumor recurrence and an increased risk of death due to breast carcinoma. In accordance with other studies, we found that patients with ≥ 4 metastatic axillary lymph nodes were more likely to have disease recurrence or die compared with patients with no or 1–3 residual metastatic axillary lymph nodes after neoadjuvant chemotherapy. In addition, patients with a pathologic response other than mCR had a risk of disease recurrence or mortality that was nearly three times the risk of patients with a pathologic mCR. The differences observed in axillary lymph node status and pathologic response, however, were not significant in this small cohort. MIBI uptake retained independent significance in multivariate analysis. The final L:N ratio was a significant independent predictor of DFS. On average, the HR for disease recurrence increased by 27% for every doubling in the L:N ratio. This suggests that MIBI uptake ratios may provide prognostic information that is distinct from standard markers.
There are potential advantages in utilizing MIBI for monitoring neoadjuvant treatment response to breast carcinoma. Quantitative in vivo imaging methods provide accurate assessment of residual viable tumor over the course of therapy. Changes in quantitative measurements obtained from serial MIBI imaging studies may help to guide therapy and may add information to traditional qualitative methods of measuring breast tumor response, such as a physical examination and mammography. Quantitative measurements could offer an earlier opportunity to change regimens if no improvement occurs or optimal surgical intervention may be determined by treating to maximal response. The ability to treat to maximal response may be important, as pervious studies have shown prolonged DFS and OS for patients with LABC who have no residual macroscopic tumor within the breast or axillary lymph nodes after presurgical chemotherapy.2–4
Investigations of MIBI uptake in tumors in other sites than the breast have been reported. Several small studies found quantitative measures of MIBI uptake in lung tumors to be predictive of survival.20–22 Breast carcinoma studies that examined MIBI localization in breast tumors compared with normal breast tissue reported its application in breast carcinoma diagnosis,23–26 assessment of response to therapy,5, 6 and identification of multidrug resistance.27–29 To our knowledge, the current study is the first to examine the relation between MIBI uptake in breast tumors undergoing neoadjuvant chemotherapy with DFS and OS. Our results indicate that high primary breast tumor uptake after neoadjuvant treatment predicts poorer DFS and OS, suggesting that serial imaging with MIBI may provide a useful quantitative surrogate end point for trials of neoadjuvant chemotherapy.
The rate of MIBI washout has been reported to correlate with breast tumor P-glycoprotein expression and to be a predictor of response to neoadjuvant chemotherapy.28, 30 In our studies (preliminary data not shown), MIBI washout from breast tumors pretherapy was not predictive of response or outcome. However, washout images were obtained only at 1 hour postinjection with no additional delayed washout views, as opposed to more detailed measurements in other reported studies.27, 28, 30 A previously published analysis of MIBI uptake and washout suggested that both blood flow and retention affected washout rates.8 More detailed kinetic analysis may be necessary to infer membrane pump efflux from MIBI images.
The most frequently used clinical application of MIBI is myocardial perfusion imaging. Studies investigating MIBI kinetics in myocardial tissue demonstrated high first-pass extraction and high cellular retention. Therefore, MIBI uptake reflects blood flow.31, 32 Another study also showed that MIBI uptake early after injection is correlated to breast carcinoma blood flow, as measured by [15O]-water PET.8 In that study, serial [15O]-water PET to measure tumor blood flow in patients with LABC treated with neoadjuvant therapy showed that blood flow is a strong predictor of response. Preliminary data also suggested that residual blood flow posttherapy is a predictor of DFS.9, 10 The findings of the current study parallel our results using [15O]-water PET. Other imaging modalities that measure tumor blood flow and tumor vascularity, such as Doppler ultrasound and contrast-enhanced magnetic resonance imaging scans, demonstrated similar findings in tumor response assessment and survival prediction among patients with breast carcinoma.33–37 Our results may provide additional insight into tumor biologic factors, such as retained vascularity, which may be associated with more resistant, aggressive tumors. Further discernment of breast tumor biologic characteristics may be derived from more detailed studies of MIBI kinetics, which are ongoing.
A small cohort with a finite number of disease recurrences and mortality events is a limitation to our study. Limited follow-up of out-of-state subjects who transferred care to their local primary physicians after completion of neoadjuvant and adjuvant chemotherapy regimens is an additional limitation. Continued efforts to obtain recent follow-up in these patients and further analyses are ongoing.
Posttherapy MIBI uptake in the primary breast tumor predicts DFS and OS as well or better than established markers and appears to provide independent prognostic information. More detailed analyses of MIBI kinetics in breast tumors and the biologic factors underlying these findings are ongoing.
The authors thank the technologist staff of the Division of Nuclear Medicine at the University of Washington for its technical assistance with the imaging studies and the University of Washington Breast Cancer Specialty Clinic for its assistance with patient referrals.
- 14Technical aspects of prone dependent-breast scintimammography. J Nucl Med Technol. 1994; 22: 165–170., , .
- 17Estrogen receptor immunocytochemistry in paraffin embedded tissues with ER1D5 predicts breast cancer endocrine response more accurately than H222Sp gamma in frozen sections or cytosol-based ligand-binding assays. Cancer. 1996; 77: 2514–2519., , , et al.
- 19The statistical analysis of failure time Data. 2nd edition. New York: John Wiley & Sons, 2002., .