Early metabolic response using FDG PET/CT and molecular phenotypes of breast cancer treated with neoadjuvant chemotherapy

Between July 2006 and September 2008, a total of 78 stage II or III breast cancer patients who received neoadjuvant docetaxel/doxorubicin chemotherapy were enrolled in this prospective study. The primary endpoint of this trial was evaluating pathologic complete response (pCR) rate, because pCR is repeatedly confirmed most relievable prognostic factor for neoadjuvant chemotherapy [17, 18], and the best predictor of improved outcome and prolonged survival [3]. Secondary endpoints were evaluating survival, toxicity, predictive factors, and early metabolic response. The eligibility criteria were described in our prior reports [14, 15]. In brief, the eligibility criteria were as follows: 1) pathologically-confirmed breast cancer by core needle biopsy, 2) initial clinical stage II or III, 3) objective measurable lesion, 4) ECOG performance 0~2, 5) previously untreated, and 6) adequate bone marrow, hepatic, cardiac, and renal functions. The patients received three cycles of neoadjuvant docetaxel/doxorubicin chemotherapy. The chemotherapeutic regimen consisted of docetaxel (75 mg/m²) and doxorubicin (50 mg/m²) by intravenous infusion every 3 weeks. After three cycles of neoadjuvant chemotherapy, the patients were re-evaluated for response and underwent curative surgery. Radiologic response was evaluated using breast magnetic resonance imaging (MRI) for the primary breast tumor and chest computed tomography (CT) for axillary, supraclavicular, internal mammary lymph nodes with RECIST criteria [19]. Both breast MRI and chest CT were performed in all the 78 patients. Subsequently, the patients received three more cycles of docetaxel and doxorubicin as an adjuvant chemotherapy, followed by hormonal or radiation therapy, if indicated [20]. Figure 1 shows the scheme for this study protocol. This regimen is known to be effective as neoadjuvant chemotherapy for stage II or III breast cancer [14, 15, 21]. We performed retrospective descriptive analysis for PET/CTs within this prospective study. This study protocol was conducted under the approval of the Institutional Review Board of Seoul National University Hospital (IRB approval number: 0510-506-159).

The molecular markers that we have examined included estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and Ki-67. We performed immunohistochemistry (IHC) using tissues obtained before treatment and evaluated the association with clinical outcomes. IHC was performed as previously described [15, 22]. The cut-off for ER and PR positivity was ≥ 10% positive tumor cells with nuclear staining. HER2 positivity was defined as either HER2 gene amplification by fluorescent in situ hybridization or scored as 3+ by IHC [23]. The molecular phenotypes of breast cancer were classified into luminal A, luminal B, HER2 and triple negative breast cancer (TNBC) [24]. TNBC was defined as ER(-), PR(-), and HER2(-). Luminal A phenotype was defined as ER(+) or PR(+) and HER2(-) tumor, and luminal B phenotype was defined as ER(+) or PR(+) and HER2(+). ER(-) and PR (-) and HER2+ tumor was classified as HER2 phenotype. Low expression of Ki-67 was defined as ≤5%. Histopathologic response was classified as proposed by Honkoop et al. [25] Specimens with no residual invasive carcinoma were classified as pCR. Residual ductal carcinoma in situ was included in the pCR category [26]. Specimens with only few scattered foci of microscopic residual invasive tumor were classified as minimal residual disease (MRD). Gross residual disease (GRD) comprised tumors showing macroscopic residual tumor or extensive residual tumor infiltration on microscopic examination. pCR and MRD were defined as histopathologic response and GRD as no histopathologic response.

Whole-body fluorine-18 FDG PET/CT scans were acquired before and after neoadjuvant chemotherapy for early metabolic response prediction. On the 15^th day of the first cycle of neoadjuvant chemotherapy, we obtained follow-up whole-body fluorine-18 FDG PET/CT scans. The CT scan protocol was as follows: tube voltage, 120 kV; current intensity, 50 mAs; scan time, 43.2 seconds; effective radiation dose, 4.8 mSv. FDG PET/CT was done using the same scanner (Gemini PET/CT system; Philips, Milpitas, CA, USA). Patients fasted for at least 6 hours before fluorine-18 FDG was injected intravenously. Patients were administered a weight-adjusted dosed of fluorine-18 FDG (5.18 MBq/kg), and images were acquired approximately 60 minutes (range: 50-75 minutes) after an intravenous injection of fluorine-18 FDG. Whole body emission scans were obtained for 2 minutes per bed position. PET/CT scanners automatically calculated the decay-corrected injected activity. We flushed the syringe and venous catheter three times, and residual activities in the syringe were less than 20 μCi and negligible. All patients were studied in the supine position with both arms raised above the head to pull their breast away.

Attenuation correction was based on the CT data, and PET data reconstruction was done by a 3-dimensional row action maximum likelihood algorithm. For quantitative assessment of tumor FDG uptake, region of interest were manually drawn on the slice with the highest radioactivity concentration of the primary breast tumor and in the adjacent slices.

The standard uptake values (SUVs) were calculated from the amount of FDG injected, body weight, and soft tissue uptake in the attenuation-corrected regional images as follows: SUV = (activity/unit volume)/(injected dose/body weight). Maximal SUV was defined as the SUV value on one pixel with the highest counts within the region of interest. The lesion chosen for analysis was the primary breast lesion with the highest SUV. The pre-treatment baseline maximal SUV of the tumor (pre-SUV) and the SUV of the tumor after the first cycle of chemotherapy (post-SUV) were compared with clinico-pathologic parameters. Calculation of the uptake index was as follows: ΔSUV% = 100 × (pre-SUV - post-SUV)/pre-SUV.

This is a subproject of a phase II trial. The primary endpoint of the main project was pCR, and we designed this trial to test null hypothesis that the pCR rate was at most 10% against the alternative hypothesis that it was at least 20%. A sample size of 78 patients provided 80% power to test this hypothesis and α 0.05, using Simon's two stage minimax design [27]. The statistical consideration was performed according to the main project. The subproject has investigated the early metabolic change as primary endpoint. This was an exploratory analysis only, without an own sample size calculation for subproject. The primary objective of this subproject was to find a correlation between early metabolic response and pathological response. The association between histopathologic and metabolic response were assessed by a chi-square test. The cut-off of the SUV change for metabolic response was not prospectively defined. Receiver-operating characteristics (ROC) analysis was performed to determine a threshold for the prediction of metabolic response. The Mann-Whitney U test or Kruskal-Wallis test were used to compare SUVs according to different molecular phenotypes. All statistical tests were two-sided, with the level of significance established at P < 0.05. SPSS software (SPSS, Inc., Chicago, IL, USA) was used for all statistical analyses.

The pretreatment characteristics are given in Table 1. Twelve patients (15.7%) were initially staged clinical stage II and the others (84.3%) were initial staged clinical stage III. The median follow-up duration was 18.7 months (range, 4.9-30.9 months). The tumor size was a median of 4.5 cm in the greatest dimension (range, 2.0-11.0 cm). At the end of follow-up, eight patients had developed recurrent disease, and one patient had died. The overall histopathologic response rate was 17.9%. Table 2 shows the results to neoadjuvant chemotherapy.

ROC analysis were performed to determine optimal cut-off values of Δ SUV% to differentiate metabolic responder and non-responder, based on histopathologic response (Figure 2). The cut-off to achieve the highest positive predictive value was Δ SUV% of 50% (sensitivity of 85.7%, specificity of 60.9%). Among the all 78 patients, 37 patients (47.4%) achieved metabolic response, while 41 patients (52.6%) did not.

Δ SUV% was significantly greater in patients with histopathologic response than patients without histopathologic response (63% vs. 34%, P = 0.001, Table 3). Pre-SUVs were not different between histopathologic responder and non-responder, however post-SUVs were significantly differed, and this led to a difference of Δ SUV%. Histopathologic response was higher in early metabolic responders determined by FDG PET/CT after one cycle of neoadjuvant chemotherapy (P = 0.002) (Table 4). Among 14 histopathologic responder, 12 (85.7%) showed metabolic response. Importantly, 39 patients among the 41 metabolic non-responders (95.1%) were eventually revealed as histopathologic non-responder (Table 4).

The FDG PET/CT results were different according to the molecular phenotype of the breast cancer. The ER negative phenotype showed a higher pre-SUV (8.6 vs. 6.4, P = 0.047) and Δ SUV% (48% vs. 30%, P = 0.038) than ER positive phenotype. In patients with high Ki-67 expression, a similar phenomenon was observed in that the pre-SUV and Δ SUV% were higher than in the low Ki-67 expression group (8.5 vs. 6.2, P = 0.018; 49% vs. 27%, P = 0.052, respectively). In the TNBC group, the pre-chemotherapy SUV was higher than in the non-TNBC group (9.8 vs. 6.4, P = 0.008); however, the Δ SUV% was not different (P = 0.799). Table 5 and Figure 3 show the serial FDG PET/CT results according to molecular phenotypes. Histopathologic response rate was higher in ER negative patients than ER positive patients (27.5% vs. 7.9%, P = 0.037) and early metabolic response rate tend to higher in ER negative patients (57.5% vs. 36.8%, P = 0.068). In this context, histopathologic response rate was significantly different according to molecular phenotype (P = 0.018) and metabolic response tend to higher in HER2+ and TNBC subtypes than luminal type (66.7% vs. 50.0% vs. 40.0%, P = 0.255). Sensitivity and specificity of metabolic response for histopathologic response were higher in ER positive phenotype (100.0% and 68.6%, respectively) than those of ER negative phenotype (81.8% and 51.7%, respectively).