Presented in part at the 47th Annual Meeting of the American Society of Clinical Oncology; June 3-7, 2011; Chicago, IL and at the 2011 American Society of Clinical Oncology Breast Cancer Symposium; September 8-10, 2011; San Francisco, CA.
In this retrospective, single-institution study, the authors examine the maximum standardized uptake value (SUVmax) on positron emission tomography/computed tomography (PET/CT) images as a prognostic variable in patients with newly diagnosed metastatic breast cancer (MBC).
Patients with ≥1 metastatic lesion on PET/CT images that were obtained within 60 days of their MBC diagnosis between January 1, 2001 and December 31, 2008 were included. Patients were excluded if they had received chemotherapy ≤30 days before the PET/CT images were obtained. Electronic medical reports were reviewed to determine the SUVmax and overall survival. Because of intraindividual variation in the SUV by body site, separate analyses were conducted by metastatic site. Relationships between site-specific PET/CT variable tertiles and overall survival were assessed using Cox regression; hazard ratios for the highest tertile versus the lowest tertile were reported.
In total, 253 patients were identified, and their median age was 57 years (range, 27-90 years). Of these, 152 patients (60%) died, and the median follow-up was 40 months. On univariate analysis, SUVmax tertile was strongly associated with overall survival in patients who had bone metastases (N = 141; hazard ratio, 3.13; 95% confidence interval, 1.79-5.48; P < .001). This effect was maintained on multivariate analysis (HR = 3.19; 95% confidence interval, 1.64-6.20, P = .002) after correcting for known prognostic variables. A greater risk of death was associated with SUVmax tertile in patients who had metastases to the liver (N = 46; hazard ratio, 2.07; 95% confidence interval, 0.90-4.76), lymph nodes (N = 149; hazard ratio, 1.1; 95% confidence interval, 0.69-1.88), and lung (N = 62; hazard ratio, 2.2; 95% confidence interval, 0.97-4.95), although these results were not significant (P = .18, P = .31, and P = .095, respectively).
Metastatic breast cancer (MBC) represents a significant public health problem, because approximately 33% of the 1 million women who are diagnosed annually worldwide with early stage disease eventually will experience relapse, and a smaller number of patients present de novo with this disease.1 After diagnosing MBC, clinicians and patients can use prognostic information to guide the selection of appropriate therapies to improve quality of life and maximize survival. Currently available prognostic tools focus on several clinicopathologic features, such as age and time from diagnosis of primary breast cancer, as well as tumor-related factors, like expression levels of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). Although useful, this approach is limited because of the extensive heterogeneity in breast cancer biology and variable responses to treatments like endocrine therapy, chemotherapy, and novel targeted agents.2 Therefore, a more refined prognostic system is desirable, because this could guide the selection of more effective therapies and allow greater individualization of treatment.
Positron emission tomography/computed tomography (PET/CT) is a widely used diagnostic tool that combines anatomic imaging with functional imaging using [18F]-2-fluoro-2-deoxy-D-glucose (FDG), a biomarker of cellular metabolism. This yields quantitative information about tumor activity by assigning a standardized uptake value (SUV) as a measurement of the relative uptake of FDG in a given lesion. The improved diagnostic performance of PET/CT imaging over conventional imaging (such as ultrasound, bone scan, and CT alone) has been investigated in a variety of settings, including for the staging of high-risk patients with early breast cancer3, 4 and in the detection of bone metastases in patients with MBC.5
Although studies have examined PET/CT imaging as a predictor of treatment response in the primary tumor,6, 7 significantly less is known about how PET/CT imaging can be used as a prognostic tool by quantifying radiotracer accumulation in metastases. Limitations in published series include small numbers, lack of histologic correlates, and the intraindividual variation in SUV by body site, which occurs because of a variety of technical reasons, including motion artifact. Therefore, in the current retrospective, single-institution study, we examined baseline FDG avidity on PET/CT images assessed by the maximum SUV (SUVmax), by body site, as a predictor of overall survival (OS) in patients with MBC.
MATERIALS AND METHODS
After we obtained approval from the Institutional Review Board, institutional databases from January 1, 2001 to December 31, 2008 and patients were included if, by report, they had evidence of ≥1 FDG-avid lesion on a PET/CT image that was obtained within 60 days of their MBC diagnosis (Fig. 1). Lesions at any of the following common MBC sites were eligible for inclusion; bone, liver, lymph node (LN), and lung. According to standard staging criteria, patients with axillary LN metastases were included only if they had additional sites of MBC. The date of MBC diagnosis was defined as the date a patient underwent biopsy of a metastatic site (when available) or the date of first radiologic imaging consistent with metastases as assessed by the treating physician. Because of possible confounding effects on image acquisition, patients who had received chemotherapy within 30 days before PET/CT imaging were excluded (Fig. 1). Patients were permitted to be actively receiving endocrine therapy (for early stage or newly diagnosed MBC). Electronic medical records were reviewed to determine known prognostic variables, including age, histology, grade, tumor phenotype (ER, PR, and HER2 expression), first-line treatment administered, and time from early stage disease to MBC as well as OS, which was calculated from the date of MBC diagnosis to the date of either death or last follow-up.
According to institutional standards PET/CT images from the middle skull to the upper thighs were obtained 60 minutes after administration of 12 to 15 millicuries FDG on 1 of 4 hybrid PET/CT systems; and, using standard formulae, the SUVmax by site (bone, liver, LN and lung) was calculated. For patients who had multiple metastases by site, the single lesion with the highest SUVmax was used for calculations. In a secondary analysis, all PET/CT scans were reviewed by a single investigator who had 7 years of experience in the interpretation of PET/CT (G.A.U.), blinded to survival data, and the SUVmax was recalculated. This demonstrated a close correlation between the SUVmax by report and by secondary review, as measured by the Spearman rank correlation. Therefore, in an attempt to recreate real-world practice, subsequent analyses were based on the SUVmax by original report. At this point, lesions that were deemed unmeasurable by G.A.U. were excluded from analysis, resulting in the exclusion of 21 patients (Fig. 1). OS was estimated for the entire cohort using the Kaplan-Meier method. Cox models were used to correlate OS with standard prognostic variables, irrespective of anatomic site.
All PET/CT images were analyzed by anatomic site, and patients with a lesion at that site were classified based on tertiles of SUVmax. Tertiles (as opposed to quartiles, quintiles, etc) were chosen to balance the flexibility gained by adding more groups with the need to keep group sizes sufficiently large for subgroup analyses. Some patients who had lesions at multiple sites contributed to several analyses. Relations between SUVmax tertiles and OS were assessed using Cox regression. Hazard ratios (HRs) for the highest versus the lowest tertile are reported in the text, and HRs for the middle versus the lowest tertile also are reported in the figures and tables. When a PET/CT variable was identified as significant (P < .1) on site-specific, univariate analysis, multivariate models were built to adjust for that variable. All statistical analyses were performed using the statistical software packages SAS (version 9.2; SAS Institute Inc., Cary, NC) and R (version 2.11.1; R Foundation for Statistical Computing, Vienna, Austria). P values < .05 were considered significant.
On initial review, we identified 285 patients who underwent PET/CT imaging within 60 days of their MBC diagnosis; and, after a review by the dedicated nuclear radiologist (G.A.U.), 32 patients were excluded (Fig. 1). Therefore, the final analysis included 253 patients (median age, 57 years; range, 27-90 years) who underwent PET/CT imaging a median of 9 days before MBC diagnosis (range, from 58 days before to 59 days after MBC diagnosis). The median time from diagnosis of early stage disease to MBC diagnosis was 2.2 years (range, 0-24.9 years). Other baseline characteristics are provided in Table 1.
Table 1. Baseline Characteristics of Evaluable Patients (N = 253)
No. of Patients
Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; MBC, metastatic breast cancer; PET/CT, positron emission tomography/computed tomography; PR, progesterone receptor.
Triple-negative tumors are defined as tumors without expression of ER, PR, and HER2.
Other histologies include papillary, colloid, and mixed ductal and lobular carcinomas.
These include abnormal liver blood tests or complete blood counts.
Stage IV was defined as metastatic disease diagnosed within 60 days of the primary breast tumor diagnosis.
These patients had stage I through III disease, but the details were unknown.
Treatment was unknown in 2 patients.
Endocrine therapy included aromatase inhibitors, antiestrogen alone, etc; chemotherapy, cytotoxic chemotherapy; targeted therapy, anti-HER2 therapy, angiogenesis inhibitors, etc, possibly in combination with either endocrine therapy or chemotherapy.
Most patients (n = 202; 80%) had a history of early stage breast cancer and subsequently developed metastatic disease. In the remaining 51 patients (20%), MBC was diagnosed within 60 days of primary breast tumor diagnosis. Almost all patients (n = 214; 85%) had invasive ductal carcinoma and ER/PR-positive, HER2-negative was the largest subgroup (n = 142; 56%). Most patients (n = 116; 46%) received endocrine therapy as their first-line treatment. Seventy-four patients (29%) received chemotherapy, and 49 patients (19%) received targeted therapy, possibly combined with chemotherapy or endocrine therapy (Table 1).
Table 2. Maximum Standardized Uptake Value by Disease Site As Prognostic Variables
Overall, 103 patients (41%) had evidence of visceral metastases (defined as lung or liver disease) on PET/CT images. According to anatomic site, the numbers of patients with metastases on PET/CT images were as follows; bone, 141 patients (56%); liver, 46 patients (18%); LN, 149 patients (59%); and lung, 62 patients (25%). In total, 228 patients (90%) had at least 1 biopsy result that confirmed the MBC diagnosis. Among the patients with FDG-avid lesions, according to anatomic site, the numbers with positive biopsies were as follows: bone, 71 of 141 patients (50%); liver, 30 of 46 patients (65%); LN, 44 of 149 patients (30%); and lung, 37 of 62 patients (60%). At median follow-up of 40 months, 152 patients (60%) had died, and the median OS was 40 months (95% confidence interval, 33-47 months).
A review of all available PET/CT scans demonstrated a close correlation between the SUVmax by original report and by independent review (G.A.U.). The Spearman correlation for these analyses by site was as follows; bone, 0.90; liver, 0.90; LN, 0.88; and lung, 0.85. The definition of tertiles translated into broadly similar SUVmax values in bone, liver, and LN (approximately <6, 6-10, and >10). However, lower cutoffs were defined in the lung tertiles (<3.4, 3.4-7.4, and >7.4) (Table 2).
Standard Prognostic Variables
We first examined known prognostic variables for the whole cohort irrespective of disease site and demonstrated the inferior OS of patients with triple-negative disease (negative for ER, PR, and HER2; HR, 3.0) compared with ER/PR-positive and HER2-negative disease (P < .01). Similarly, patients who had visceral metastases (N = 103) had inferior survival (HR, 1.4; P = .03) compared with patients who did not. Conversely, A longer time from primary breast cancer to MBC diagnosis was associated with improved OS (P < .01; HR, 0.9 and 0.5 for 3-5 years and >5 years, respectively, vs <3 years). Patients who received targeted therapy (including with endocrine therapy or chemotherapy) or chemotherapy alone in the first-line setting had significantly decreased survival (P < .001; HR, 1.6 and 3.5, respectively) compared with patients who received endocrine therapy.
Maximum Standard Uptake Value As a Prognostic Variable
A strong correlation between the SUVmax in bone and OS was observed (P < .001) (Table 3). By using the tertile with the lowest SUVmax as the reference group (median, 4.7; range, 2.1-5.8), patients in the highest tertile of SUVmax (median, 11.2, range, 9.3-29.6) had the shortest survival (HR, 3.13) (Fig. 2A). The magnitude of this effect was greater than that for some of the other prognostic variables that were assessed in the bone cohort, such as the presence of visceral metastases (HR, 2.7; P < .001), but it was smaller than the effect of other variables, such as triple-negative histology (HR, 4.16; P < .001). Receipt of targeted therapy and chemotherapy in the first-line setting were associated with inferior OS (HR, 1.4 and 3.5, respectively; P < .001).
Table 3. Standardized Uptake Value in Bone As a Prognostic Variable: Cox Model for Overall Survival
Abbreviations: CI, confidence interval; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hazard ratio; MBC, metastatic breast cancer; PR, progesterone receptor; Ref, reference category; SUVmax, maximum standardized uptake value.
HRs shown are for a 1-unit increase in the covariate (age) or compared with the reference category. Multivariate models were built to adjust for tumor phenotype and variables with P < 0.1 on univariate analysis.
All patients had grade 2 or 3 disease; the comparison here is for grade 3 versus grade 2.
Endocrine therapy included aromatase inhibitor, antiestrogen alone, etc; chemotherapy, cytotoxic chemotherapy; targeted therapy, anti-HER2 therapy, angiogenesis inhibitor, etc, in combination with either endocrine therapy or chemotherapy.
Targeted with or without chemotherapy/endocrine therapy
It is noteworthy that grade (P = .09), age (P = .45), histologic subtype (P = .95), and time from primary breast cancer to MBC diagnosis (P = .14) had no significant effect on prognosis (Table 3). We also examined the impact of endocrine therapy at the time of PET/CT imaging and observed no effect on prognosis (P = .62; N = 50 patients on PET images; results not shown).
A greater risk of death was also observed among patients with higher SUVmax in other metastatic sites (liver, LN, and lung), but this achieved statistical significance only in bone (Table 2, Fig. 2B-D). Multivariate models were constructed by site, as described above. In multivariate analyses, the prognostic effect of SUVmax in bone was maintained (P = .002) after correcting for known prognostic variables (Table 3).
To the best of our knowledge, this retrospective study represents the first large series to correlate FDG avidity on PET/CT images and OS in patients with newly diagnosed MBC, and it has several key findings. First, the SUVmax in bone was strongly correlated with OS, and this effect was maintained in multivariate analyses after correction for tumor phenotype, grade, and the presence of visceral metastases—the prognostic variables that had P values < .1 on univariate analysis. Second, a greater risk of death was also observed among patients who had a higher SUVmax in other metastatic sites (liver, LN, and lung), but the difference did not reach statistical significance. Third, an independent review of PET/CT images demonstrated that the SUVmax was highly reproducible by the reader and could be applied outside of a specialist center. Taken together, these findings suggest a possible role for PET/CT imaging as an additional prognostic tool for patients with newly diagnosed MBC.
Because cancers typically demonstrate greater than physiologic metabolic activity with high glucose uptake, higher FDG uptake on a PET/CT image may be indicative of a more aggressive phenotype. In a variety of malignancies, a higher SUVmax at diagnosis has been associated with inferior survival.8-18 In breast cancer, there are significant challenges to the use of PET/CT imaging as a prognostic tool, including extensive tumor heterogeneity, variable uptake of FDG, and differential response to subsequent treatment. Higher SUV correlates with a variety of adverse prognostic variables, including tumor grade, HER2 expression, ER negativity, and greater numbers of circulating tumor cells.17, 19-22 The use of PET/CT imaging in a prognostic model is compounded further by differential site-specific metastases for biologic subgroups of breast cancer. For example, strong expression of ER and PR, like what occurs in lobular carcinoma, is associated with a relative increased risk of bone metastases and lower SUV values on PET/CT images.22 Therefore, in an attempt to take into account variable tumor biology as well as intraindividual variation in FDG uptake, we examined SUVmax at individual anatomic sites and corrected for known prognostic variables that were identified as significant on univariate analysis.
In our study, patients who had bone metastases (N = 141) and the highest SUVmax tertile (9.3-29.6) had inferior survival compared with patients in the lowest tertile (2.1-5.8; HR, 3.13; P < .001). This effect was maintained in multivariate analyses (HR, 3.19; P = .002) after correcting for known prognostic variables. Similar findings were observed in patients who had metastases in liver (N = 46; HR, 2.07), LN (N = 149; HR,1.1), and lung (N = 62; HR, 2.2), although these results did not reach statistical significance (P = .18, P = .31, and P = .095, respectively).
The current study has several strengths. First, this was a large series that included a broad representation of various subgroups of breast cancer, including 54 patients (21%) with HER2-positive disease and 50 patients (20%) with triple-negative breast cancer. Second, 90% of patients had at least 1 biopsy confirming the diagnosis of MBC (the gold standard), which contrasts to some other series in which the diagnostic performance of PET/CT imaging was compared with other imaging modalities and the results were confounded by the variable performance of the comparator. Third, we correlated FDG uptake (SUVmax) with OS, which is a clean endpoint, as this considers both variable tumor biology and treatment administered. Because the median OS of patients with MBC is 2 to 3 years,2 our results are relatively mature (median follow-up, 40 months). Finally, SUVmax values, which were rereviewed by a single investigator, were highly correlated with the values abstracted from the electronic medical record. We opted to use SUVmax by report in this study, because this reflects real-world practice, in which clinical decisions are made based on PET/CT reports, and selected cases are reviewed on an individual basis.
There are limitations to the current study, which was retrospective, did not assess tumor:background ratios, and included a heterogeneous population both in terms of variable follow-up imaging (timing and modality) and treatment regimens administered. Because not all patients at our institution with possible MBC underwent PET/CT imaging or had FDG-avid disease, selection bias may have played a role. Although 90% of patients underwent a biopsy of at least 1 site, we cannot be absolutely sure that all of the FDG-avid lesions observed on PET/CT images truly represented MBC. For example, although 44 patients in the LN analysis had biopsy-proven MBC in LN, this does not preclude the possibility that many false-positive, high SUVmax values (eg, reactive LNs) were included incorrectly in the correlation between SUVmax and OS. Furthermore, we examined PET/CT imaging from only 1 time-point and, thus, are unable to comment on the predictive effect of PET/CT imaging (with regard to treatment effect). Finally, because this was a retrospective study, the cost-effectiveness of PET/CT imaging could not be assessed.
Several studies have investigated the possible correlation between changes in SUVmax and response (or lack thereof).6, 7, 23-26 In both the preoperative and metastatic settings, a decline in the SUVmax has been associated with tumor response to endocrine therapy and chemotherapy and may be useful to identify a group of patients who are resistant to therapy.6, 7, 23, 27 Furthermore, it has been suggested that PET/CT imaging may be a superior predictor of progression-free survival in patients with bone metastases compared with other novel approaches, such as enumerating circulating tumor cells.26 In fact, emerging data indicate that PET/CT imaging is superior to conventional imaging, such as radionuclide bone scans, for detecting bone metastases from breast cancer.5, 28-30 However, caution is advised, because most data are from retrospective studies, variations in FDG response have been reported by body site, and the choice of therapy and sequence themselves may have an impact on changes in the SUVmax.27, 31
Significant challenges remain in defining the optimum use of PET/CT imaging in MBC. Some investigators have attempted to define specific cutoffs for SUV to define subgroups that may have inferior survival or may benefit most from selected therapies.19 However, given the variation in SUV and differential impact on survival by site, this may not be reproducible; and, if cutoffs are appropriate, then site-specific cutoffs may be advised. In this regard, the SUVmax may be a suboptimal PET/CT parameter in isolation, because it does not factor the size or the number of metastatic lesions into the analysis. Alternative PET/CT prognostic variables may include mean SUV, metabolically active volume, functional lesion volume, or total lesion glycolysis, all of which can be calculated using modern software.32 The composite measurement, total lesion glycolysis, is an attractive PET/CT parameter, because it attempts to overcome some of these limitations by incorporating metabolic volume into the assessment of FDG avidity.33 Therefore, currently, we are investigating total lesion glycolysis as a prognostic variable in patients with newly diagnosed MBC. An alternative approach is the use of novel radiotracers for subgroups of breast cancer, such as patients with HER2-positive disease, designed to offer greater specificity; however, to date, no dominant tracer has demonstrated superiority to FDG.34-37 Ultimately, prospective studies will be needed to individualize parameters and tracers for PET/CT imaging. Nonetheless, we hope that PET/CT imaging has a future role in novel drug development, in individualizing therapeutic approaches by selecting appropriate initial therapies for patients, and as an early signal to change therapy in patients who are not benefiting from a treatment.
In conclusion, in this large, retrospective study of patients who had chemotherapy-naive, newly diagnosed MBC, an increased SUVmax in bone was correlated with inferior survival. Similar effects were noted at other metastatic sites but did not reach statistical significance. These findings suggest that PET/CT imaging is a potential prognostic tool for patients who have newly diagnosed MBC.