Abstract

PURPOSE

This study aimed to compare the value of differences (∆) in parameters obtained via both dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) and flourine-18 fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) between baseline and post-neoadjuvant therapy in predicting the pathological response to neoadjuvant therapy in breast cancer.

METHODS

A total of 109 patients who underwent both baseline and post-neoadjuvant therapy DCE-MRI and ¹⁸F-FDG PET/CT examinations were retrospectively analyzed. The DCE-MRI parameters and ¹⁸F-FDG PET/CT parameters [metabolic tumor volume (MTV), standardized uptake value (SUV)_max, SUV_mean, and total lesion glycolysis] were recorded at both time points. Additionally, the ∆s between these parameters were calculated. Postsurgical pathology reports were documented, and the patients were subsequently categorized into two groups: those exhibiting pathologic complete response (pCR) and those exhibiting partial response. Parameters from DCE-MRI and ¹⁸F-FDG PET/CT were compared to determine which predicted pathological response to neoadjuvant therapy more effectively.

RESULTS

Patients with partial response demonstrated a higher rate of histologic grade 3 than those with pCR (P = 0.030). The only DCE-MRI parameter to indicate a significant difference between the two groups (P = 0.024) was the ∆(%)wash-out rate. Among the baseline parameters, only MTV successfully predicted pathological response (P = 0.033). The only post-neoadjuvant therapy parameter to be predictive of pathological response (P = 0.003) was SUV_mean. In receiver operating characteristic analysis, ∆SUV_mean emerged as the most significant parameter for predicting pathological response, followed by post-neoadjuvant SUV_mean [area under the curve: 0.724 (95% confidence interval: 0.630–0.805) and 0.673 (0.577–0.760), respectively].

CONCLUSION

The ∆¹⁸F-FDG PET/CT parameters are better than ∆DCE-MRI in predicting pathologic response to neoadjuvant therapy. Among these parameters, ∆SUV_mean is the most successful.

CLINICAL SIGNIFICANCE

Neoadjuvant chemotherapy (NAC) response is one of the most important criteria in breast cancer prognosis. The two most important imaging modalities in breast cancer diagnosis and follow-up protocols are MRI and ¹⁸F-FDG PET/CT. However, it is not clear which of these two modalities is more successful in predicting the difference in treatment response between baseline and post-NAC.

Breast cancer is a significant public health concern, and women in particular face high breast cancer incidence and mortality rates.¹ Prognostic factors are critical for estimating recurrence risk and providing increasingly personalized treatment for patients.² Current therapeutic approaches in breast cancer focus on neoadjuvant chemotherapy (NAC)–also referred to as primary or induction chemotherapy–which is administered prior to surgical intervention.³ Histological response following NAC remains the most important factor associated with recurrence risk.^2-4 In addition, early prediction of NAC response can reduce patient costs and protect patients from unnecessary side effects and time loss, leading to the use of alternative methods. Therefore, additional and more precise methods of predicting NAC response and estimating prognosis are necessary.

Although novel imaging parameters have been developed to characterize the biological features of breast tumors, factors affecting the quality of the therapeutic response remain uncertain.^{5, 6} Dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) shows tissue vascularity and perfusion due to blood flow by analyzing signal intensity.⁷ Semi-quantitative parameters derived from DCE-MRI can be calculated based on the signal intensity curve.⁸ Although studies investigating the relationship between these semi-quantitative parameters and therapeutic responses in various tumor tissues are not uncommon in the literature,^7-9 research examining the relationship of the differences (∆) in these parameters between diagnosis and post-neoadjuvant therapy with NAC response remains limited.

Flourine-18 fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) is a valuable tool in the staging of breast cancer at diagnosis and in evaluating therapeutic response and suspicion of recurrence using standardized uptake value (SUV) and its associations with various predictive and prognostic factors.¹⁰ The SUV_max parameter corresponds to the point with the highest SUV value measured within a region of interest (ROI); in other words, it defines the most active part of the tumor and does not provide insight into the overall tumor status. Metabolic tumor volume (MTV) and total lesion glycolysis (TLG) have been introduced as semi-quantitative measures of ¹⁸F-FDG uptake within a heterogeneous tumor mass. The MTV represents the volume of tumoral tissue demonstrating a specific ¹⁸F-FDG uptake activity beyond the intensity of FDG uptake by tumor tissues.¹¹ Several studies have also compared the efficacy of TLG with that of MTV in predicting therapeutic responses.^{2, 10, 11}

This study aims to evaluate whether semi-quantitative ∆DCE-MRI parameters and ∆¹⁸F-FDG PET/CT indices (∆MTV, ∆SUV_max, ∆SUV_mean, and ∆TLG) could predict pathological responses to NAC in patients with breast cancer during initial staging. These parameters are also compared with other predictive and prognostic factors [age, cellular proliferation index, histologic grade, hormone receptor expression, and human epidermal growth factor receptor 2 (HER2) status] in terms of the pathological response after NAC.

Methods

Patient selection and study design

The present study was conducted in full accordance with the guidelines of the Declaration of Helsinki, revised in 2000 in Edinburgh. Approvals for the study were obtained from the Ethics Committee of University of Health Sciences Turkey, Adana City Training and Research Hospital and the Turkish Ministry of Health (28.03.2024/3232). The requirement for informed consent from the patients was waived due to the retrospective nature of the study. Of the 262 cases diagnosed during this period, 109 female patients who underwent both preoperative DCE breast MRI and ¹⁸F-FDG PET/CT at both diagnosis and after the completion of NAC were included in the study, following the application of the exclusion criteria (Figure 1). All patients underwent clinical tumor staging via baseline clinical examination, mammography, and ultrasonography along with a tru-cut biopsy. Both ¹⁸F-FDG PET/CT and cranial MRI examinations were performed for tumor–node–metastasis staging. To avoid the partial volume effect during measurement of ¹⁸F-FDG PET/CT uptake, all lesions in the study had a long axis >15 mm.

Semi-quantitative parameters from DCE-MRI at the time of diagnosis were recorded. Subsequently, semi-quantitative DCE-MRI parameters obtained after NAC were documented, including (1) maximum enhancement, (2) maximum relative enhancement, (3) T0 (s), (4) time to peak (s), (5) wash-in rate (s^-¹), and (6) wash-out rate (s^-¹). Differences between the measurements obtained at diagnosis and those recorded after NAC were calculated. Additionally, the SUV_max, SUV_mean, TLG, and MTV values obtained from ¹⁸F-FDG PET/CT at diagnosis and after NAC were recorded, and the Δs between these parameters were calculated. The percentage changes (Δ%) for all ¹⁸F-FDG PET/CT and MRI parameters were calculated using the following formula:

Percentage change (Δ%) = (delayed parameter − baseline parameter) / baseline parameter × 100.

Magnetic resonance imaging acquisition

The MRI protocols for this study adhered to the standards established by the breast MRI accreditation program of the American College of Radiology.¹² A three-dimensional T1-weighted fast gradient echo-based DCE series with one pre-contrast and three sequential post-gadolinium contrast-enhanced sequences, each with a scan duration of approximately 3 minutes, was imaged. The examinations were conducted using a 3-Tesla scanner (Ingenuity; Philips Healthcare, the Netherlands) with a 16-channel breast coil. The DCE-MRI was performed using a fat-saturated three-dimensional fast gradient echo sequence with the following parameters: repetition time/echo time, 5.8/3 ms; flip angle, 15°; spatial resolution, 0.5 × 0.6 × 1.4 mm; voxel size, 0.8 × 0.8 × 2; temporal resolution, 60 s; number of sections, 200; and field of view, 360 × 360 mm. Fat suppression was achieved via the short-scar periareolar inferior pedicle reduction technique. All images were acquired in axial orientation. The contrast agent used was gadobutrol (Gadovist, Bayer Schering Pharmaceuticals, Mississauga, Canada) administered at a dose of 0.1 mmol/kg body weight and a rate of 2 cc/s, followed by a 15 cc saline flush.

Magnetic resonance imaging evaluation

All obtained DCE-MRI were retrospectively reviewed based on consensus by two experienced radiologists with 9 and 7 years of experience in breast MRI. The radiologists were blinded to clinicopathological findings and clinical follow-up data except for data on the presence of breast cancer. All images obtained using the software were analyzed. Regions of interest were defined as areas demonstrating abnormal signals on the MRI and were manually delineated using the oval-shaped function (Figure 2). Time–intensity curves and time-to-peak values were automatically generated based on these ROIs.

In accordance with previously published methodologies, multiple circular ROIs measuring 25 mm² (approximately 5.6 mm in diameter) were carefully placed within the solid portion of the tumor, as indicated by DCE-MRI and T2-weighted imaging.^{2, 7, 9} The ROIs were meticulously positioned within the tumor to avoid cystic regions or visual artifacts, and average values within each ROI were recorded. These analyses were performed using a dedicated workstation (IntelliSpace, Philips, Netherlands).

Acquisition of flourine-18 fluorodeoxyglucose positron emission tomography/computed tomography

Patients were placed in the supine position during imaging, and CT transmission scanning was performed without intravenous contrast enhancement. The scan utilized a low tube current (135 kVp, 46–79 mAs), a slice thickness of 4.0 mm, a gantry rotation of 0.53 s, and a collimator width of 5 × 3 mm. Following a minimum fasting period of 6 hours, an intravenous injection of 370–555 MBq ¹⁸F-FDG was administered. Whole-body images were acquired using a PET/CT scanner (Ingenuity; Philips Healthcare, the Netherlands) 40–60 minutes after injection. Visual analysis employed a four-point certainty scoring system (definitely negative, equivocal: probably negative, equivocal: probably positive, and definitely positive) and included assessments of anatomical site and lesion size. The emission data were acquired for 2.5 min per bed (six to seven beds), which were later attenuation-corrected with the digital CT data. Image reconstruction used the ordered subsets expectation maximization algorithm of two iterations and eight subsets.

Flourine-18 fluorodeoxyglucose positron emission tomography/computed tomography evaluation

All ¹⁸F-FDG PET/CT images were retrospectively evaluated by a nuclear medicine specialist with 17 years of experience in PET-CT. This specialist was blinded to clinicopathological and clinical follow-up data that did not cover the presence of breast cancer. The foci of FDG uptake were identified if FDG accumulation exceeded that of comparable normal contralateral or surrounding tissues. The workstation automatically generated SUV_max, SUV_mean, and MTV (cm³) from the volume of interest (VOI) (Figure 3). The target lesion margins inside the VOI were also automatically defined, and voxels exceeding 41% of the SUV_max within the VOI were included in the MTV and SUV_mean calculations. Among the relative thresholds of 40% SUV_max for tumor delineation and contouring on PET/CT in patients with breast cancer, it is the most appropriate and popular cut-off value and prognostic marker.^1-4 The TLG was calculated by multiplying MTV by SUV_mean. Due to the study’s retrospective nature, semi-quantitative measurements were retrospectively reproduced by the nuclear medicine specialist taking part in the study, who was blinded to all other data.

Chemotherapy regimen and histological evaluation

The NAC regimens included combinations such as doxorubicin and cyclophosphamide, doxorubicin and cyclophosphamide followed by taxane, doxorubicin and docetaxel, and taxane combined with anti-HER2 agents.

The pathological characteristics of the tumors examined in this work, including their histological type, histological grade, Ki67 proliferation index, hormone receptor expression, and HER2 status, were analyzed based on biopsies obtained at diagnosis prior to NAC. Immunohistochemical (IHC) analysis was used to determine whether the tumor tissue was positive or negative for HER2 and hormone receptors [estrogen receptor (ER) and progesterone receptor (PR)]. For hormone receptor expression positivity, ≥10% was accepted. The HER2 expression results were grouped as negative, 1+, 2+, or 3+. Then, the fluorescence in situ hybridization (FISH) method was used for gene amplification in cases with HER2 IHC staining score 2+. Cases with an IHC staining score of 3+ were defined as HER2 positive, or in the case of an IHC staging score of 2+, FISH positive.The Elston and Ellis¹³ criteria (grades 1, 2, and 3) were used to determine histological grades. The IHC study of the tumor tissue obtained positive or negative results for ER and PR and stratified these results according to the modified Allred scale.¹⁴ Histological responses to antibodies in the breast were assessed using Miller–Payne grading.¹⁵ Pathologic complete response (pCR) is defined as the complete disappearance of cancer cells on histopathologic examination of mastectomy material after neoadjuvant treatment. Since the presence of residual ductal carcinoma in situ (DCIS) has prognostic significance and affects the comparability of the results with other studies, in this study, the pCR group was considered to be the absence of any tumor cells, including DCIS. The patients were categorized into five subgroups based on the recommendations of the 12^th International Breast Conference: (1) luminal A; (2) luminal B-HER2(−); (3) luminal B-HER2(+); (4) HER2(+); (5) triple-negative.¹⁶ The threshold for Ki67 positivity was 14%.

Statistical analysis

Statistical analysis was conducted using SPSS version 23.0 statistical software (SPSS, Chicago, IL, USA). Categorical variables were summarized as counts and percentages, and continuous variables were presented as means and standard deviations or medians (Q1–Q3) where applicable. Comparisons of categorical variables were conducted using the chi-squared test. The Kolmogorov–Smirnov test was applied to assess the normality of parameter distributions. For non-normally distributed parameters, the Mann–Whitney U test was employed. The sensitivity and specificity of the relevant parameters in predicting patient response to treatment were calculated. Moreover, the area under the receiver operating characteristic curve (AUC) was analyzed to determine the cut-off values. A P value of <0.05 was considered statistically significant in all tests.

Results

The rate of patients with histological grade 3 was found to be higher in those exhibiting a partial response than those achieving pCR (P = 0.030). No significant differences were observed between the treatment groups in relation to the remaining parameters (P > 0.05) (Table 1).

In patients with pCR, baseline MTV, ∆(%)MTV, ∆(%)SUV_mean, and ∆(%)TLG values were significantly higher (P = 0.033; P = 0.014; P < 0.001; and P = 0.014, respectively), whereas post-NAC SUV_mean values were significantly lower than those with a partial response (P = 0.003) (Table 2).

Tumor size and/or changes in tumor size were not associated with response to NAC. However, patients who achieved pCR exhibited significantly higher ∆(%)wash-out values (P = 0.024). No significant differences were observed between the two groups in terms of the remaining DCE-MRI parameters (Table 3).

Baseline MTV (AUC: 62.5%), post-NAC SUV_mean (AUC: 67.3%), ∆MTV (AUC: 64.5%), ∆SUV_mean (AUC: 72.4%), ∆TLG (AUC: 64.4%), and the DCE-MRI parameter ∆wash-out (AUC: 63.5%) were found to be statistically significant predictors of treatment response (P < 0.05). The ∆SUV_mean parameter demonstrated the best diagnostic test performance at predicting treatment response among these parameters, with a cut-off value of >30.5 and an AUC of 72.4 (Table 4).

The MTV, SUV_mean, ∆MTV, ∆SUV_mean, ∆SUV_mean, ∆TLG, and ∆wash-out values after neoadjuvant treatment and the distribution between biological subtypes were similar (P = 0.837; P = 0.325; P = 0.897; P = 0.664; P = 0.782; P = 0.425, respectively) (Figure 4).

Discussion

The response to NAC is the most significant parameter for predicting breast cancer prognosis.^{17, 18} Therefore, the ability to predict the response to NAC can be considered equivalent to anticipating the course and prognosis of the disease.¹⁹ DCE breast MRI and ¹⁸F-FDG PET/CT are the two most utilized imaging modalities for patients with breast cancer worldwide. Therefore, the current study aimed to identify the most effective parameter among these two imaging techniques for predicting the response to NAC.

Although SUV is a widely accepted parameter in clinical practice, SUV_max remains controversial.²⁰ This is because SUV_max reflects the single-pixel value of the maximum intensity of ¹⁸F-FDG activity rather than the entire tumor mass. Furthermore, SUV_max is affected by body weight.²¹ In the current study, no SUV_max value (baseline, post-NAC, or ∆) successfully predicted the response to NAC. On the other hand, SUV_mean, as an average measurement, avoids these limitations. Accordingly, ∆SUV_mean and post-NAC SUV_mean were determined to be the most valuable ¹⁸F-FDG PET/CT parameters for predicting histopathological response in this study.

By adjusting for lean body mass, MTV reflects metabolic activity across the entire tumor mass.^{22, 23} In addition, MTV appears to be correlated with total tumor burden and tumor aggressiveness in various solid tumors, thus providing useful prognostic information.²⁴ Both TLG and MTV are volumetric parameters used to estimate total radioactivity within a tumor above a set threshold.²⁵ They have both proven to be useful in assessing the therapeutic effects and prognosis of chemotherapy.²⁶ The size of the tumor is also included in the calculation of TLG and MTV. It is not surprising that TLG and MTV increase with increasing tumor size, as found in some recent studies. In addition, some studies have shown that poorly differentiated tumors have significantly higher TLG and larger MTV than well-differentiated tumors.^{27, 28} In this study, both ∆MTV and baseline MTV predicted response to NAC, whereas MTV after NAC was not significantly associated with treatment response. In this study, the superior value of SUV_mean compared with MTV and TLG likely stems from its independence from tumor size, as tumor size was found to be unrelated to NAC response. The study’s independent evaluation of ¹⁸F-FDG PET/CT parameters and overall analysis determined ∆SUV_mean to be the most effective parameter for predicting NAC response.

The finding that ∆SUV_mean is the most significant parameter, both within the ¹⁸F-FDG PET/CT parameters and overall, demonstrates that changes in tumor biology (i.e., ∆ values) are the most effective predictors of response to NAC. These variations are better represented by the differences between baseline assessments and post-NAC evaluations, particularly in ∆SUV_mean and ∆MTV, which can predict pCR and, consequently, influence surgical procedures.

Invasive lobular carcinoma (ILC) also differs from invasive ductal carcinoma in terms of receptor expression.²⁹ In particular, ILC is almost always (95%) ER-positive, raising the possibility of increased utility of ER-targeting PET tracers for patients with ILC.²⁹ Sixteen alpha ¹⁸F-fluoroestradiol is an ER-targeting PET tracer with high sensitivity and specificity for the detection of ER-positive tumors. It can be used as a predictive biomarker to demonstrate ER heterogeneity, evaluate the pharmacokinetics of ER-targeted agents, quantify residual ER during endocrine therapy, and determine the biological optimal dose of novel ER-targeted drugs.

One factor that may influence the response to systemic chemotherapy is tumor perfusion. Tumor tissue with relatively poor perfusion is believed to be less exposed to chemotherapeutic agents and therefore may not respond adequately to NAC.^{29, 30} This deficiency could be a factor contributing to poor responses to intravenous chemotherapy. Prior studies on large breast tumors using various techniques, including histologic analysis of vessel density, PET with ¹⁵O-water, DCE-MRI, single-photon imaging with 99mTc-sestamibi, and Doppler sonography, all suggest that blood flow in LABC is highly variable.^31-33 Quantitative DCE-MRI, which assumes two chambers representing the extravascular extracellular space and blood plasma, provides absolute and thus objective measures of perfusion.^{3, 20, 21} However, among the DCE-MRI parameters evaluated both at baseline and post-NAC in this study, only ∆wash-out was able to predict the treatment response. Nevertheless, the specificity of ∆wash-out was found to be significantly lower than the parameters of ¹⁸F-FDG PET/CT.

Many factors may influence ¹⁸F-FDG uptake into tumor cells, including the up-regulation of glucose transporters and hexokinase enzymes and neo-angiogenesis, which in turn are related to the number of viable tumor cells, their aggressiveness, and proliferative activity.^{1, 7, 15} Angiogenesis is one of the most important of these, and one of the biggest effects of NAC drugs is that they affect tumor perfusion; that is, angiogenesis. However, as shown in this study, the disruption of angiogenesis is only one of the factors affecting the tumor’s pathological response. For this reason, DCE MRI parameters, whose main mechanism of action is to show tumor perfusion, show considerable variability due to the disruption of angiogenesis after neoadjuvant treatment, but do not show this pathological response. However, the main mechanism of action of PET-CT parameters is metabolic activity. Metabolic activity is higher in tumors with high perfusion/angiogenesis; however, its effect on metabolic activity, which is the main indicator of PET-CT parameters, is limited. Therefore, the rate of being affected by perfusion disorders caused by NAC drugs is limited.

Previous studies have suggested that advanced T and N stages and ER-negative status are associated with high SUV_max values in patients with breast cancer. Although lymph node response was not examined in this study, these parameters, which are effective in predicting the response of mass to neoadjuvant treatment, were found to be unaffected by biological subtypes.

This study had several limitations. First, the current literature describes DCE-MRI parameters using k_trans, v_e, and k_ep calculations. The authors were unable to use these parameters, since no application capable of calculating these values is available in the hospital; this can be regarded as a limitation concerning the integration of the study with existing literature. Second, FDG PET/CT was performed in the supine position, and no additional imaging was performed specifically for the position of the breasts. This may have been a misleading factor, especially in parameters such as MTV and TLG, where tumor volumes are also calculated. Some recent studies have examined the effect of breast position on PET-CT results.³⁴ Furthermore, the small sample size due to stringent exclusion criteria and the retrospective nature of the study were additional limitations.

In conclusion, in patients with LABC, ¹⁸F-FDG PET/CT performed both at baseline and post-NAC demonstrated the metabolic changes induced by treatment. The ∆SUV_mean parameter is the most successful predictor of pathological response to NAC among all PET/CT parameters.