Abstract
Background/Aim: Positron emission tomography/computed tomography (PET/CT) with 18F-fluorodeoxyglucose (18F-FDG) has recently been used to investigate lymph node (LN) metastases and several predictive features in patients with breast cancer (BC). The aim of this study was to assess the value of this non-invasive imaging procedure for axillary staging. Patients and Methods: Fifty patients with early primary unilateral, locally advanced, or recurrent invasive operable BC were enrolled. All patients underwent preoperative 18F-FDG PET/CT, and the results were compared with the histopathology of dissected axillary LNs and their biological and immunohistochemical characteristics. The diagnostic performance of 18F-FDG PET/CT in detecting LN metastases from primary or recurrent BC was analyzed. The mean values of the initial PET/CT parameters, including the primary tumour (SUV T) and ipsilateral axillary LNs (SUV LN), were compared with the clinicopathological features of patients to determine their usefulness for predicting clinical interactions. Results: The sensitivity, specificity, overall accuracy, positive predictive value, and negative predictive value of 18F-FDG PET/CT for axillary LN staging were 87%, 90%, 88%, 93%, and 82%, respectively. Bivariate analyses showed strong interactions of nuclear grade (p=0.05), progesterone receptor expression (p=0.001), Ki-67 index (0.027), and local relapse with the SUV T. A high SUV LN value was significantly correlated with a higher nuclear grade score (p=0.05), oestrogen receptor negativity (p=0.001), progesterone receptor negativity (p=0.014), a high Ki-67 index (>20%; p=0.048), LN metastasis (p<0.001), a basal tumour (p=0.04), and locoregional recurrence (p<0.001). Conclusion: PET/CT is a reproducible, non-invasive imaging modality that is useful for evaluating a primary BC mass and its relationship with metastatic axillary LNs, thereby predicting tumour behaviour and guiding clinical practice.
Traditionally, pathologic determination of tumour size, histological grade, axillary lymph node (LN) involvement, and hormone receptor and human epidermal growth factor receptor-2 (HER2) status have been used as prognostic factors in patients with breast cancer (BC). Determination of the Ki-67 index is also strongly recommended at the time of planning targeted therapies, given that its positivity is associated with a higher probability of relapse and worse overall survival (1, 2).
It has since long been recognized that accurate evaluation of locoregional nodal status is essential because of its implications in treatment and prognosis. The 10-year survival rate in patients with axillary LN metastases depends on the number of nodes involved and ranges from 30% (>10 nodes) to 70% (1-3 nodes), compared with 90% in those without these metastases (3). However, management of the axilla in patients with operable BC is still one of the most controversial areas in clinical oncology. For a long time, the standard procedure used for the assessment of locoregional progression has been axillary LN dissection (ALND), which is associated with significant morbidity, including arm and shoulder pain, lymphedema, nerve injury, hematoma, and limitation of shoulder movement (4). Moreover, ALND does not provide a survival advantage when the LNs are tumour-negative (5). In this context, sentinel LN biopsy (SLNB) is a significant advancement in clinical practice. Nevertheless, false-negative (FN) axillary results can occur in up to 15% of patients, and there is also concern that the disease may be upstaged when SLNB is coupled with very thorough pathological examination (immunohistochemical analysis and multistep sectioning of sentinel nodes) that may detect micrometastases, the clinical relevance of which has still to be demonstrated (6).
Based on these considerations, in order to make progress in the treatment of operable BC, new strategies should be developed that permit a less invasive method of axillary sampling and do not impair patient's quality of life. Positron emission tomography/computed tomography (PET/CT) with 18F-fluorodeoxyglucose (18F-FDG) is a recently developed imaging modality that has been used for the investigation of nodal metastases and several predictive features in patients with cancer. FDG-PET/CT imaging reflects the higher glycolytic rate of cancer cells compared to normal cells, allowing discrimination based on metabolic rate. This ability offers potential advantages for detection of subclinical metastases when compared with radiological criteria, which are based on anatomical findings (7).
Unlike conventional morphologic diagnostic imaging, FDG-PET is a functional diagnostic imaging technique that quantifies and evaluates the activity of cancer cells by measuring the maximum standardized uptake value (SUVmax) before and after chemotherapy, and is a new tool for assessment of the aggressiveness of the disease (8, 9). However, a definitive conclusion remains to be reached concerning the prognostic relevance of the level of 18F-FDG uptake in BC, and the cut-off value for the SUV reduction rate with regard to the diagnosis of responders and non-responders varies in different reports (10, 11). Future studies on the effects of neoadjuvant chemotherapy using FDG-PET scans will need to standardize the histologic evaluation criteria and include the consequences for axillary LNs (12). In this context, a correlation of the level of the SUV in BC with clinicopathological parameters may have prognostic implications in terms of predicting biological characteristics and the baseline risk pre-therapeutically. Indeed, the distribution of FDG uptake has been associated with underlying pathophysiological characteristics, including vascularization, perfusion, tumour aggressiveness, necrosis, hypoxia, and gene expression (13, 14). Therefore, we hypothesized that significantly different values for PET image-derived heterogeneity may be observed between the different histological profiles and molecular subgroups of BC, assuming that heterogeneous parameters such as tumour phenotype, nuclear pleomorphism, and steroid receptor and HER2 status, together with the Ki-67 index, have different underlying pathophysiological properties. Parallel to the development of techniques to identify favourable tumour biology and the ability to select patients who are expected to have a more protracted disease course or tumours that are responsive to systemic therapies, more patients will be considered for metastasectomy or cytoreductive surgical techniques (15, 16). PET imaging, with further refinements in technical performance, might prove to be a very useful tool for functional and biological localization of malignant lesions, with all the recognized limitations (17).
Current recommendations do not incorporate the routine use of 18F-FGD PET/CT for staging primary BC, and direct scientific evidence to support the ability of this procedure to impact the management procedure is limited (18). First, the true incidence of primary BC in its various stages that is detectable by 18F-FDG PET/CT needs to be assessed systematically. FDG uptake may then provide quantitative information on both the primary tumour and metastatic LNs additional to that obtained by clinicopathological methods. How such detection aids targeted individualized treatment and alters survival and quality of life should then be assessed. Finally, the cost to the health service provider of performing this procedure needs to be accurately calculated.
In this study, the value of full diagnostic FDG-PET/CT for axillary LN staging in patients with BC was assessed, evaluating at the same time the possible influence of biological markers on the glycolytic pathways. The aims of the study were to determine the sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) of FDG-PET/CT and to assess the possible role of this imaging modality in preselection of candidates for SLNB instead of immediate ALND. However, more detailed knowledge of the initial clinicopathological implications for optimal use and interpretation of this non-invasive imaging technique are needed before a definitive conclusion can be reached concerning the prognostic relevance of the preoperative level of 18F-FDG uptake in BC. Further studies need to be performed in large numbers of patients to verify these results and to further investigate the role of PET/CT in staging of BC, particularly in the detection of axillary nodal involvement and inherently aggressive biological criteria with poorer clinical outcomes, for an optimal disease-tailored approach to management of locoregional and systemic disease.
Materials and Methods
Study design and conduct. A total of 50 consecutive eligible patients (48 women, two men, aged 34-88 years) with early primary unilateral, locally advanced, or recurrent invasive BC were prospectively recruited for preoperative staging by 18F-FDG PET/CT between November 2015 and July 2018. All patients had undergone triple assessment including mammography, ultrasonography, and core biopsy, prior to recruitment. The indication for surgery, the surgical strategy, and study eligibility were determined on the basis of the findings of preoperative PET/CT imaging and no evidence of distant metastases. The functional imaging evaluation of the presence or absence of regional nodal spread of malignancy (quantitative and semiquantitative analysis, see below) was compared with the reference standard for assessment of the axillary nodes, i.e., a histological examination of resected axillary LNs obtained by SLNB alone, by SLNB and ALND, or by ALND alone. Forty-four patients (88%) with histologically proven primary BC were required to proceed to SLNB or ALND after completion of PET/CT and to undergo a level I/II ALND after SLNB, if the SLNB result was positive, or if the PET/CT scan showed increased uptake in the ipsilateral axilla. Nine (20%) of these 44 patients had received neoadjuvant chemotherapy before surgery. The remaining six patients (12%) enrolled in the study had locoregional recurrence that was presumed to be resectable, i.e., one focus or a limited number of hot foci of uptake demonstrated on PET/CT images. The patients were divided into two groups, one with locally advanced BC (LABC) that included 30 patients (60%) with ipsilateral axillary LN metastases and the other without LABC that included 20 patients (40%) without LN metastases. A positive PET/CT examination confirmed by pathology was considered as true-positive (TP) and otherwise as false-positive (FP). A negative imaging examination was considered to be true-negative (TN) if there was no histological nodal involvement and to be FN if there was metastatic invasion. Patients with uncontrolled diabetes mellitus, evidence of systemic metastatic disease at presentation, active infection, serious organ dysfunction, or distant metastases at the time of diagnosis of BC and without operation were excluded. Eligible patients were enrolled during the preoperative surgical consultation and provided informed, written consent (ethical statement n. 52.12).
Maximum intensity projection of positron-emission tomography images showing the 18F-fluorodeoxyglucose distribution in patient 1. Coronal (a-b) and sagittal (c) positron-emission tomography / computed tomographic images demonstrating hypermetabolic lesions in the upper-outer quadrant of right breast (SUV T=9.1) and in the right axilla (SUV LN=10.3).
PET/CT scan protocol. All patients fasted for at least 5 hours before injection of intravenous 18F-FDG. The serum glucose level was normal (≤107 mg/ml) in all cases. The patients then received an intravenous injection containing 370-450 MBq of 18F-FDG and were hydrated with 500 ml of intravenous saline (0.9% sodium chloride). Physical activity was kept to a minimum, with a rest period of 60 min post-injection. Images were obtained 60 minutes after administration of FDG. All examinations were performed using the PET/CT Discovery Iq system (GE Medical Systems, Fairfield, CT, USA). According to the National Electrical Manufacturers Association (NEMA), this system the highest sensitivity in the industry, up to 22 cps/kBq. The whole system is optimized for oncology practices administering 18F, which accounts for nearly 94 percent of all PET procedures with a sensitivity up to 75 kcps at 2.4 kBq/ml, largest axial field-of-view coverage in the industry up to 26 cm. For the PET/low-dose computed tomography (ldCT), a low-amperage CT scan was acquired for attenuation correction of the PET images (80 mA, 140 kV, field of view about 420-500 mm, and a CT slice thickness of 3.75 mm). The CT dose index for ldCT was 4.0175 (±0.84) mGy and the dose-length product was 473.296 (±161.09) mGy-cm. After non-enhanced CT, a total-body PET examination in the caudocranial direction from the upper thighs to the vertex was performed (3.5 min per bed). Reconstruction was performed using the novel Q Clear algorithm. The images were reviewed by a nuclear physician (A.C., B.T) and a surgical oncologist (O.C.B, P.O). The SUV and metabolic tumour volume of the target lesion(s) were measured. The PET/CT images were then reviewed by two nuclear medicine physicians on a dedicated PET/CT workstation (Advantage 4.47 GE Healthcare, Fairfield, CT, USA) that allowed visualization of the PET and CT images separately or in fusion mode in the axial, coronal, and sagittal planes. Uptake was considered pathological when an area of focal tracer uptake higher than the background was visually detected. The maximum SUV (SUVmax) values were also determined and considered, but no absolute cut-off value was used for the diagnosis. In cases of disagreement between the readers, the patients were re-examined and a consensus was reached. The FDG uptake in the primary tumour (SUV T) and LNs (SUV LN) was semi-quantitatively analyzed using the SUVmax (g/ml), which was calculated based on the measured activity, decay-correction of the administered dose, and patient weight (Figure 1). When calculating the SUV LN, the axillary LN showing the highest SUV within the whole axillary space was selected. The NT ratio was calculated by dividing the SUV LN by the SUV T (19). In case of mild-low 18F FDG uptake, the co-registered CT was used for the creation and contour of the regions of interest.
Classification of groups and staging. The size of the breast tumour was measured, and the tumours were classified according to the American Joint Committee on Cancer (8th edition) staging criteria, i.e., T1a-b (<10 mm), T1c (11-20 mm), T2 (21-50 mm), or T3 (>50 mm) (20). The number, maximum size, and nuclear grade of the involved invasive BC and axillary LN metastases were also examined by histopathology. Microscopic examination of the resected BCs and LNs was used as a reference to evaluate the ability of 18F-FDG PET/CT to detect the primary tumour and axillary metastases.
The intrinsic BC subtypes were identified according to the clinicopathological criteria recommended by the 2013 St. Gallen International Expert Consensus Report (21). The patients were categorized based on the receptor status of their primary tumour as follows: luminal A (oestrogen receptor-positive [OR+] or progesterone receptor-positive [PR+] and HER2−); luminal B HER2− (OR+, HER2−, and at least one of Ki-67 “high” or PR “negative or low”); luminal B HER2+ (OR+, HER2 overexpressed or amplified, any Ki-67 value, any PR); HER2 (OR− or PR− and HER2+), or basal (OR− or PR− and HER2−). Tumours were considered HER2-positive only if they were scored as 3+ by immunohistochemistry (IHC; strong, complete membrane-staining in >10% of cancer cells) or showed HER2 amplification (ratio >2) using fluorescence in situ hybridization (FISH). In the absence of positive FISH data, tumours scored as 2+ on IHC were considered negative for HER2. Tumours were also classified as luminal or non-luminal according to hormone receptor expression. Furthermore, a useful surrogate definition was developed to distinguish the pathological characterization of steroid hormone receptor status and the threshold value for “high Ki-67” based on a combination of OR ≥50%, PR ≥20%, and Ki-67 status ≥20%, in order to determine whether some variables could be considered independently associated with 18F-FDG uptake without a requirement for molecular diagnostics (22). The frequencies of the patients' stages and those of short-term development of remote metastasis in the non-LABC and LABC groups are given in percentages. Furthermore, the prognostic impact of the level of SUV in the tumours was explored for various cut-off values to compare the biological characteristics and baseline risk of BC pre-therapeutically.
Endpoints and statistical analysis. The diagnostic performance of 18F-FDG PET/CT in detecting primary or recurrent LN metastases was analyzed based on the calculation of sensitivity (TP/[TP+FN]), specificity (TN/[TN+FP]), PPV (TP/[TP+FP]), and NPV (TN/[TN+FN]). The overall accuracy was calculated as the percentage of all TP and TN cases out of the total number of cases (23). All statistical analyses were performed using IBM SPSS version 23 software (IBM Corp., Armonk, NY, USA). The independent t-test, chi-square test, analysis of variance, and Spearman's rank correlation coefficient were used to establish the correlation between diagnostic performance defined by all metabolic parameters (SUV T, SUV LN, NT ratio) and clinicopathological variables. The sensitivity, specificity, PPV, and NPV of 18F-FDG PET/CT imaging for primary invasive BC and axillary LN staging were analyzed using standard statistical analyses. Means, standard deviations (SDs), and 95% confidence intervals were calculated for all the quantitative variables. PET/CT parameters, stage I-III disease, and tumour characteristics across biological profiles were compared between the non-LABC and LABC groups using the t-test, and p-values <0.05 were considered statistically significant. Numeric data are expressed as the mean±SD. Receiver-operating characteristic (ROC) curve analysis was performed to identify an optimal cut-off for the primary tumour (SUV T) and nodal SUVmax (SUV LN) for prediction of axillary metastasis and to determine the diagnostic utility of the NT ratio in all patients. The area under the curve ranges from 0.5 (indicating a test with no information) to 1.0 (indicating a perfect test).
Results
Patient demographics and tumour characteristics. Data were available for 44 patients with newly diagnosed BC and for 6 with locoregional recurrence after previous breast-conserving treatment or mastectomy, all of whom underwent 18F-FDG PET/CT for preoperative staging. The median patient age was 57.4 (range=34-88 years). The median tumour size was 2.6 cm (range=0.5-8 cm). Classifying the patients according to histological type, 46 (92%) had invasive ductal carcinoma and 4 (8%) had invasive lobular carcinoma. Thirty percent of the tumours were scored as grade 1, 48% as grade 2, and 22% as grade 3. The incidence of luminal and non-luminal tumours was 90% and 10%, respectively. The majority of patients (n=21, 42%) had luminal A tumours, and the remaining patients had luminal B (n=22, 44%), HER2-positive luminal B (n=2, 4%), basal (n=4, 8%), or non-luminal HER2-positive (n=1, 2%) tumours. OR ≥50% was positive in 42 (84%) of 50 tumours and PR ≥20% was positive in 29 (58%) of 50 tumours. Overexpression of c-erbB-2 was found in 3 tumours (6%). Twenty-eight patients (56%) had high Ki-67 expression (≥20%). Twenty (40%) of the 50 patients were staged as pN0, and 30 (60%) had ipsilateral axillary LN metastases according to the gold standard. N stage was categorized by the staging system of the American Joint Committee on Cancer (8th edition): 15 patients (30%) were in N stage 1, 6 (12%) in N stage 2, and 9 (18%) in N stage 3. There were 14 patients (28%) in stage I, 18 (36%) in stage II, and 18 (36%) in stage III. At the time of analysis, one patient had died during a median follow-up of 13.5 (range=4-29 months). Forty-two (92%) of the 50 patients were disease-free and 4 (8%) had recurrence of systemic disease (lung metastases in 2 and bone metastases in 2), of which 3 (75%) were in the LABC group. The characteristics of the study participants are listed in Table I.
Analysis of PET features and clinicopathological factors. Axillary LN involvement was detected on pathologic examination in 30 patients (60%) who had total axillary dissection. FDG-PET was positive in the axilla in 28 cases (56%) and negative in 22 (44%). Twenty-six (92.8%) of the 28 axillary LN-positive results were TP and 2 (7.2%) were FP. Four (18.1%) of the 22 axillary LN-negative results were FN and 18 (81.8%) were TN. The FN rate was 13% (4/30) and the FP rate was 10% (2/20). The sensitivity, specificity, PPV, and NPV of FDG PET/CT for detection of axillary LN metastases overall was 87%, 90%, 93%, and 82%, respectively. The diagnostic accuracy was 88% (Table II).
Patients and tumour characteristics.
18F-FDG uptake was lower for all parameters in invasive lobular lesions (SUV T, 4.1±2.8; SUV LN, 1.8±2.6) than in invasive ductal lesions (SUV T, 4.4±2.9; SUV LN, 2.4±3.5), but no statistical analysis was performed because of the small number of cases.
Diagnostic performance of 18 F-FDG PET/CT in axillary lymph node staging.
The average SUV T in the 50 patients was 4.3±2.9 (range=0.9-13.5) and the mean SUV LN was 2.3±3.4 (range=0.1-14.6). There were no significant differences in either volumetric parameter (SUV T, 5.0±3.3 vs. 3.7±2.2; SUV LN, 2.5±3.2 vs. 2.2±3.7) for tumours measuring more or less than 2 cm or according to T stage (p>0.05).
With regard to the tumour nuclear grade and its components, the mean SUV T was significantly different between the grade 1 (3.3±1.8), grade 2 (4.5±2.9), and grade 3 (5.8±3.3) tumour groups (p=0.05; correlation coefficient, R2=0.3). Moreover, the mean SUV LN was significantly higher in the grade 3 lesions (4.3±5.1) than in the grade 1-2 lesions (1.8±2.6; p<0.001), as were the number of positive LNs (7.1±9.7 vs. 2.6±3.8; p<0.001) and the Ki-67 index (38.8±21.7 vs. 17.6±10.0; p<0.001). Specifically, the mean quantitative expression of OR (54.0±44.2 vs. 82.6±25.3; p=0.001) and PR (15.0±32.4 vs. 58.8±38.0; p=0.001) tended to be lower in the grade 3 group.
In relation to the above findings, considering the mean quantitative expression of OR (76.3±32.2), PR (49.2±40.9), and Ki67 (22.3±15.8), the bivariate analysis did not show a significant interaction between SUV T and OR (p>0.05; correlation coefficient, R2=−0.09) but showed that the SUV T was inversely proportional to PR (p=0.036; correlation coefficient, R2=−0.3) and directly proportional to Ki-67 (p=0.027; correlation coefficient, R2=0.3). Furthermore, the bivariate analysis for SUV LN and the same variables showed an inverse interaction with OR (p=0.004; correlation coefficient, R2=−0.4) and PR (p=0.05; correlation coefficient, R2=−0.3), and a directly proportional interaction with the number of metastatic LNs (p=0.01; correlation coefficient, R2=0.4; (Table III).
According to the threshold values applied in this study, there was no significant difference in the SUV T between the OR-positive (≤50%) group and the OR-negative (≥50%) group (4.2±2.8 vs. 5.1±3.5; p>0.05) or between the PR-positive (≤20%) group and PR-negative (≥20%) group (3.7±2.5 vs. 5.3±3.1; p>0.05). However, the mean SUV LN was significantly higher in OR-negative tumours (5.5±5.3 vs. 1.7±2.7; p=0.001) and PR-negative tumours (3.3±4.0 vs. 1.6±2.8; p=0.014). Furthermore, the NT ratio differed considerably between OR-positive and OR-negative cases (0.38±0.45 vs. 1.05±1.54; p=0.003), while PR negativity was associated with a larger tumour size (3.3±2.1 vs. 2.2±1.4 cm; p=0.0016) and involvement of more nodes (5±7.8 vs. 2.5±3.6; p=0.006).
Correlations between mean values of clinicopathological parameters and SUV T/SUV N states.
When categorized as ≤20% and >20%, the Ki-67 index showed no correlation with SUV T (3.3±2.0 vs. 5.2±3.3; p>0.05), but in the event of positivity (>20), the mean SUV LN was significantly higher (2.9±4.0 vs. 1.6±2.4; p=0.048) and the tumour size was larger (3.1±1.9 vs. 2.0±1.4 cm; p=0.045) than in pathological lesions with a Ki-67 index ≤20. The mean quantitative expression of OR− (70.5±35.3 vs. 83.7±26.8; p=0.038), and PR− (35.5±39.7 vs. 66.6±36.11; p=0.006) also differed significantly according to Ki-67 status (>20%).
With regard to c-erbB-2 status, the mean SUV T (4.7±3.0 vs. 4.3±2.9; p>0.05) and SUV LN (1.4±0.7 vs. 2.4±3.5; p>0.05) were not significantly different between the tumour groups with scores of 2+ and 3+ or between those with scores 0 and 1+. However, the mean quantitative PR- expression tended to be lower in the HER2+ group (1.0±1.7 vs. 52.2±40.2; p=0.001).
There were no statistically significant differences in the SUV T between the tumour groups with and without metastatic LNs (4.8±2.8 vs. 3.7±3.0; p>0.05). Furthermore, the mean SUV LN (3.6±4 vs. 0.5±0.7; p<0.001) and NT ratios (0.69±0.88 vs. 0.20±0.31; p=0.05) were higher in the node-positive group than in the node-negative group.
There was no significant difference in the SUV T for stage I-III tumours (3.2±1.9 vs. 4.2±3.2 vs. 5.3±3.0; p>0.05) or between the SUV LN (1.2±3.8 vs. 2.6±3.4 vs. 2.9±3.2; p>0.05) and the respective categories. Nevertheless, the mean NT ratios were significantly different in stages I-III (0.07±0.09 vs. 0.56±0.55 vs. 0.72±1.02; p=0.048).
There were no critical differences in the SUV T between the five molecular subtypes of BC (p>0.05), but the mean SUV LN was significantly higher in triple-negative (OR-/PR-/HER2-) tumours than in the others (5.9±6.1 vs. 2.0±3.1; p=0.008). At the same time, the SUV LN was significantly lower in luminal A tumours (1.7±2.4 vs. 2.9±4.0; p=0.04) than in the other subtypes. Moreover, the probability of a higher SUV LN appeared to be stronger in patients with non-luminal tumours than in those with luminal tumours (4.9±5.7 vs. 2.0±3.1); however, the difference was not statistically significant (p=0.08).
In summary, the mean SUV T and SUV LN were significantly higher in the group with locoregional recurrence than in the group without (6.1±6.0 vs. 4.2±2.7, p=0.01 and 7.0±5.7 vs. 1.7±2.5, p<0.001, respectively). There was also a marked difference in the mean NT ratio between cases with and without local relapse (0.77±0.04 vs. 0.46±0.75; p=0.01) and in the number of positive nodes involved (7.5±11.4 vs. 3.0±4.6; p=0.007). Correlations between the clinicopathological parameters and the SUV T, SUV LN, and NT ratio are shown in Table IV. The results on further pathways and crosstalk between the biological features are summarized in Table V and Figure 2.
Patients' prognosis and cut-off values for metabolic parameters. A ROC curve analysis was also performed to identify which metabolic parameters (SUV T, SUV LN, NT ratio) better reflected the probability of axillary metastasis on final pathology. Our hypothesis was that nodal status would still be a potential signature of the intrinsic biological properties of a primary tumour.
The results indicated that a SUV T of 3.1 was the optimal cut-off value for predicting progression of lymphatic disease (sensitivity, 70%; specificity, 60%; area under the curve, 0.639; standard error, 0.0844); however, this finding was not statistically significant (p=0.14; Figure 3). Furthermore, the analysis revealed that a SUV LN of 0.99 (sensitivity, 77%; specificity, 84%; area under the curve, 0.869; standard error, 0.052; p<0.001; Figure 4) and an NT ratio of 0.12 (sensitivity, 80%; specificity, 70%; area under the curve, 0.814; standard error, 0.068; p<0.001) were the best threshold values for detecting nodal malignancy (Figure 5).
Discussion
The role of 18F-FDG PET/CT in the planning of management of patients with BC is increasing (24). The prognosis and therapeutic decision-making are based primarily on precise data on the real extent of the disease, which is directly related to several biological parameters and metastatic involvement of locoregional LNs. Involvement of these nodes is accepted as the most reliable predictor of disease recurrence and survival (25). However, despite being a non-invasive imaging tool, the efficiency of PET/CT in evaluating locoregional nodal status has not yet been well-defined in clinical practice, and its sensitivity for detection of axillary metastases remains low in the range of 37-85% depending on the axillary tumour burden and affinity of the BC for 18F-FDG (26-28).
Correlations between clinicopathological parameters and the levels of SUV T, SUV N, and NT ratio.
This functional diagnostic procedure may help identify patients with a high axillary LN burden who could move directly to ALND and not require the additional step of SLNB (29). However, an FN scan in the evaluation of axillary disease is mainly secondary to smaller and fewer tumour-positive LNs than TP cases (28). Some recent studies in which IHC and multistep sectioning were used to increase the rate of detection of axillary micrometastases with ALND/SLNB have cast doubt on the role of 18F-FDG PET/CT in the axillary staging of BC. In these studies, PET proved to have poor sensitivity, ranging from 20-43% (23, 30). However, any discussion concerning this oncological indication should take into consideration the fact that there is presently no alternative strategy to routine ALND that does not entail an intrinsic low risk of downstaging the axillary status, given that even SLNB has had a non-negligible FN rate in almost all studies reported to date (31). This limitation of PET/CT should be analyzed in relation to the importance of axillary metastases for the outcome in patients with BC. Limited data suggest that only a few axillary micrometastases become clinically evident during follow-up, and further analysis of these patients has demonstrated that these locoregional relapses have no major impact on overall survival (32-34).
Biological pathways and cross-talks according to patients' characteristics.
ROC curve analysis of the discriminative power of SUV T for the prediction of axillary node metastasis (SUV T=3.1; sensitivity (70%); specificity (60%); area under the curve (0.639); SE (0.0844); p=0.14).
Moreover, full-diagnostic 18F-FDG PET/CT has good overall accuracy and high specificity, and has the potential to be able to triage patients to SLNB or ALND. In patients with a high risk of axillary LN metastases and being a priori not candidates for SLNB, an unremarkable scan can still identify a subgroup of patients who can safely undergo SLNB instead of immediate ALND. In this context, high risk is defined as a greater than 40% probability of axillary node metastases, corresponding to primary tumours with a diameter greater than 2-3 cm. Such an approach may extend the indication for SLNB to patients with an up to 60% risk of lymphatic metastases, corresponding to a tumour diameter of 4-5 cm, and thus avoid the comorbidity of unnecessary ALND in a substantial proportion of patients (35).
ROC curve analysis of the discriminative power of SUV LN for the prediction of axillary node metastasis (SUV N=0.99; sensitivity (77%); specificity (84%); area under the curve (0.869); SE (0.052); p<0.001).
ROC curve analysis of the discriminative power of NT ratio for the prediction of axillary node metastasis (NT ratio 0.12; sensitivity (80%); specificity (70%); area under the curve (0.814); SE (0.068); p<0.001).
This study shows that the specificity (90%) and PPV (93%) of PET/CT for detection of axillary metastases in patients with primary BC are high. Furthermore, in 13 patients with negative procedures who underwent axillary dissection based on clinical indications, the nodes were negative in 11 cases and positive in 2. This translates into a potential 25% (11/43) reduction in unnecessary ALND procedures and associated morbidities if perioperative PET/CT imaging is used. However, the suboptimal sensitivity and NPV and an SLNB when 18F-FDG-avid axillary nodes are absent remain important.
Correlations between clinicopathological parameters and tumor features (Student's t-test).
It seems reasonable that these strategies, when supported by the clinical use of more sophisticated biological and molecular markers available from the primary tumour, might one day abolish the need for histopathological evaluation of axillary LNs with related mapping techniques to establish the best treatment approach following surgery (36, 37). In this regard, the ongoing evolution of metabolic imaging represents a promising development, given that tumours with aggressive characteristics and an unfavourable prognosis show a higher degree of 18F-FDG uptake (38, 39). However, it is also known that breast carcinomas have a variable metabolic rate and that characterization of small lesions (<0.8 mm) is less reliable, probably because of the partial volume effect and low metabolic activity, which may also indicate inflammatory lesions (40).
Despite these limitations, some studies have found a positive relationship between the SUVmax and tumour size (41). Nevertheless, in our study, the non-significant correlation between tumour size and SUV T/LN led us to consider that the size and metabolism of the tumour are not closely related. This discrepancy in the literature may be explained by differences in histology (necrotic or fibrotic tissue) and biology (OR and PR expression) between tumours that may influence the SUV by affecting the tumoral microenvironment. Moreover, although tumour size is an obvious marker of deep changes in cell biology and a strong predictor of survival in patients with BC, this well-established relationship does not appear to hold in a small subset of cancers with more aggressive phenotypes and pathways with higher risk of invasion, as shown by other researchers (42).
Several reports, including ours, have found a significant relationship between FDG uptake and pathological grade, which is attributed to the finding that less differentiated tumours with higher tumour growth rates have a higher metabolic rate (43). This finding might reflect high Glut-1 levels at the membrane and an increased level of hexokinase in the cytoplasm, which are encountered in high-grade human cancers, including BC (11, 44, 45).
In light of the current literature, we believe that a breast mass with a high SUV may indicate the aggressiveness of BC at the time of initial diagnosis, as evidenced by the greater number of involved nodes, as well as the higher Ki-67 proliferation index in the grade 3 group in our study.
Most of the recently published studies have shown a correlation between hormone receptor status and SUV (38, 41, 46). The results of our study confirmed a higher 18F-FDG uptake in OR-negative and PR-negative tumours than in OR-positive and PR-negative tumours. The correlations were statistically significant, especially for nodal SUVmax, highlighting the prognostic significance of this combination of marked biochemical alterations of LNs to discriminate high-risk from low-risk subjects (47, 48). Microarray analysis has confirmed these results and identified genes associated with increased glucose use as measured by PET, suggesting that increased glucose consumption can be linked to a specific genetic signature (41). Furthermore, it is possible that improved predictive models based on PET/CT findings in patients with axillary LN involvement for preoperative evaluation of the chances of disease recurrence will be useful when choosing therapeutic strategies (49). In this context, adequate knowledge of the initial clinical stage of a breast tumour is also necessary to ensure local disease control and make decisions regarding appropriate systemic interventions, given that in the present series we were able to document a significantly higher pT size and positive LN involvement in the PR-negative subgroup. This is particularly important because evaluation of PR expression levels ≥20% can impact the molecular classification of BC, maximizing the number of patients classifiable as having a luminal A-like profile. In these patients, the use of cytotoxic drugs could largely be avoided, in the light of the more indolent course and better prognosis of cancer (22).
Although the Ki-67 cut-off value is controversial, it is also one of the most important prognostic markers that, when integrated with other predictive factors, aids in better assessment of patients who are at risk of adverse outcomes and in selection of optimal treatment for BC (2, 50). A correlation between all volumetric parameters and the Ki-67 index was found when our cohort was evaluated as a whole, with the strongest interactions occurring with the SUV LN and the mean quantitative hormonal status expression being lower in the “high Ki-67” cases. A high Ki-67 index indicates a high rate of mitosis and proliferation, and it has also been reported to be associated with the response to treatment, local recurrence, and metastasis (51, 52). According to our results, there is also a reverse association between Ki-67 and hormone receptor status, such that if the Ki-67 index is positive, OR/PR expression will likely be lower in the malignant tissue, in conformity of such a biological association with SUVmax measurements. This finding would suggest that glucose metabolism is accelerated in these poorly differentiated and aggressive tumours to meet the energy demand of rapid growth. To date, no imaging modality or parameter that can predict the metastatic potential of BC has been identified, and 18F-FDG PET/CT may provide a non-invasive mean of assessing tumour characteristics and facilitating individualized management of cancer.
Overexpression of the c-erbB-2 oncoprotein is a well-known contributor to tumour aggressiveness and poor prognosis, and the advent of trastuzumab, an antibody treatment targeting this receptor has been a breakthrough (53). Some studies have found a significant relationship between 18F-FDG uptake and expression of the c-erbB-2 oncogene (54, 55). However, in those studies, 46% of tumours were <2 cm in size, which might have biased their results. No significant difference in c-erbB-2 overexpression on SUV was found in our present series or in several other studies (38, 56-58), suggesting that HER2 has no major influence on glycolytic pathways. However, although many pathways are regulated by these receptors and the cross-talk between these receptors are still incompletely understood, it is possible that the c-erbB2 status inversely interacts with both OR and PR in terms of 18F-FDG uptake, and may characterize their metabolic effects (58).
An increased SUV LN has been suggested to reflect intense proliferation of nodal BC cells (59). Similarly, in the LABC patients participated in this study, who were classified most commonly as stage II-III, a relatively high number of metastatic LNs was demonstrated (6.0±6.5; p<0.001), with a significantly high SUV LN (3.6±4.0) (p<0.001) and NT ratio (0.69±0.88; p=0.05). However, it was also reported that an FN scan is necessary in the evaluation of the axillary involvement particularly because of the small size of LNs and due to limitations in the anatomical resolution of this imaging procedure which was reported to be 5-6 mm (26). Given that micrometastases cannot be excluded by a negative PET/CT, it is considered that the metabolic features of the primary mass should be evaluated carefully, as in our study, in order to predict an occult axillary LN metastasis. Recently, the NT ratio was introduced as a new PET/CT parameter and found to be useful in evaluating metastatic LNs in non-small cell lung cancer and BC (19, 60). Similarly, in our study, the mean NT ratio in the LABC group was significantly higher than that in the non-LABC group, and it is concluded that it can be used as a promising quantitative parameter for the preoperative evaluation of axillary LNs suspected to be metastatic, as well as for characterizing the different pathological stages of BC. In addition, combining the NT ratio and the SUV N, on the basis of the set cut-off values identified in the current study to predict the burden of axillary disease, may be important for minimizing FP cases. However, tumours with a low SUV, including low-grade malignancies and benign inflammatory lesions, also warrant attention, and whether the NT ratio is efficacious in patients receiving chemotherapy, endocrine therapy, and molecular targeted therapy is unclear.
Furthermore, our study confirms that the 18F-FDG uptake in tumours is associated with distinct immunohistochemically defined biological subtypes. According to the results of previous studies, the SUV LN is significantly higher in basal BCs than in non-basal BCs, and with a greater difference in isolated semiquantitative values observed for the luminal A group (46, 61, 62). Basal-like tumours certainly have distinct epidemiological phenotypic and molecular features with distinct patterns of relapse and a poor prognosis, whereas among the OR-positive tumours, the luminal A type tends to show higher expression of pathways related to OR signalling and lower expression of proliferative (HER-2-related) genes, with specific consequences for the clinical outcome (63). Therefore, different oncogenes might influence the glycolytic pathway, and the resulting increase in metabolic activity is a hallmark of cancer that can be imaged using 18F-FDG PET/CT (64, 65). Consistent with the most recent results reported by Garcia et al, the present investigation shows a significant correlation between variability in nodal SUV and the more biologically aggressive molecular portraits, supporting the view that there is a relationship with glycolytic phenotypes (66).
Finally, our findings suggest that PET/CT is not only effective for detection of locoregional recurrence in association with conventional imaging findings, but may also allow better characterization of lesions in patients in whom there is a strong suspicion of relapse, all correlated with a higher 18F-FDG uptake compared to those of the non-recurrent group. In this clinical scenario, diagnostic functional imaging can add significant prognostic information to the restaging capabilities, given that survival of patients who develop an isolated locoregional recurrence differs significantly from that of patients who have a distant relapse. Consequently, determination of both the location and extent of recurrent disease is essential to guide therapeutic decisions and estimate the prognosis. Some authors have suggested that FDG-PET/CT can have a significant impact on subsequent therapeutic management; however, information concerning its utility for long-term prognostic stratification, when compared with conventional imaging, is limited. Therefore, accurate selection of lesions in patients who could actually benefit from use of this imaging approach appears to be important (67, 68).
The main limitation of our study is the relatively low number of patients in the subgroups and that all patients had undergone previous diagnostic core biopsies that removed a portion of the cancer tissue. Therefore, the overall amount of 18F-FDG uptake may have been reduced, resulting in lowering of the overall contrast of the lesion on PET/CT because of the partial volume effect. Moreover, classification based on steroid receptors and HER2 status is only an approximation of the underlying genotype-based BC subtype. Furthermore, the greatest drawback presently is the lack of standard methods and definitive cut-off values for calculating volumetric parameters in the literature (69-72). Finally, our findings could not be correlated with long-term disease-free survival in our patients, so the long-term prognostic value of our results is unclear.
Conclusion
PET/CT is a reproducible, non-invasive imaging modality that is useful for evaluating a primary BC mass and its relationship with metastatic axillary LNs, allowing for a selective approach to either ALND or SLNB. The findings of this study suggest that the nodal SUVmax and NT ratio before initial treatment could be an independent prognostic factor for LN involvement, thus predicting tumour behaviour and guiding clinical practice. Furthermore, the systemic nature of this powerful tool enables one-stage evaluation of the whole body and subsequent treatment, and future endeavours to improve its sensitivity and specificity will enhance its diagnostic utility in the management of BC.
Our findings suggest that histological and biological criteria might be taken into account, in addition to clinical stage, to select candidates for a basal 18FDG-PET/CT study. Our SUV analysis in 50 patients showed that high tumour grade (grade 3), hormone receptor negativity, high Ki-67 labelling index, triple negativity, and axillary involvement are associated with increased 18F-FDG uptake. The present evidence indicates that it might be difficult to predict the potential for metastasis to the axillary LNs from the SUV in primary BC; however, the values of parameters indicating the proliferative activity and aggressiveness of cancer cells were strongly correlated with high SUV in axillary LNs with metastases. Failure to detect nodal diffusion may result from the low metabolic rate of the tumour, which is an intrinsic disease characteristic, like type and grade. In conclusion, our preliminary findings indicate that the status of molecular markers, as determined using core needle biopsies and preoperative 18F-FDG PET/CT parameters, might be used to identify high-risk BC patients. These pathological and biological factors are also known to confer a poor prognosis and probably a greater incidence of occult distant metastases. When the results of longer clinical follow-up become available for this large prospective series of patients with BC, we will investigate whether the initial SUV has a prognostic value in addition to that of the classic clinicopathological factors.
Acknowledgements
This research project was supported by the Italian Ministry of Health.
Footnotes
Conflicts of Interest
No benefits, in any form, have been received or will be received from any commercial party related directly or indirectly to the subject of this article.
- Received October 20, 2018.
- Revision received November 10, 2018.
- Accepted November 15, 2018.
- Copyright© 2018, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- Pre-pectoral Breast Reconstruction Does Not Affect Early Immunological Response: A BIAL 2.20 Study Subanalysis
- Impact of COVID-19 Pandemic on Surgical Breast Cancer Patients Undergoing Neoadjuvant Therapy: A Multicentric Study
- Feasibility and Oncological Safety of Axillary Reverse Mapping in Patients With Locally Advanced Breast Cancer and Partial Response After Neoadjuvant Chemotherapy
- Clinical Outcomes Among Major Breast Cancer Subtypes After Neoadjuvant Chemotherapy: Impact on Breast Cancer Recurrence and Survival
- Evaluation of Concordance Between Histopathological, Radiological and Biomolecular Variables in Breast Cancer Neoadjuvant Treatment