P3-14-16: Molecular Phenotype and the Use of HER-2 Targeted Agents Influence the Accuracy of Breast MRI after Neoadjuvant Chemotherapy

Background: Improved understanding of factors affecting the accuracy of breast magnetic resonance imaging (MRI) after neoadjuvant systemic therapy (NST) can lead to more tailored use of MRI in deciding surgical extent after NST.

Materials and Methods: We analyzed the imaging and clinicopathologic data of 463 patients who underwent NST. We aimed to investigate whether the molecular subtypes, as well as the use of targeted therapies, were associated with changes in the accuracy of MRI predicting residual tumor extent.

Results: The accuracy of MRI predicting the residual tumor extent was most accurate in triple negative breast cancer and was least accurate in Luminal A subtype (Pearson's correlation coefficient of 0.754 and 0.531, respectively). Multivariate analysis suggested estrogen receptor status as an independent factor influencing the MRI accuracy. In HER2−amplified tumors, the use of HER2−targeted agents was associated with less accurate MRI prediction. Dual HER2−blockade by using trastuzumab and pertuzumab resulted in lowest MRI accuracy among the patients treated with HER2−targeted agents.

Conclusion: The accuracy of MRI in predicting residual tumor extent was lowest in ER positive tumors treated with NST. In HER2 positive tumors, the use of HER2−targeted agents resulted in less accurate MRI after NST.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-14-16.

P2-08-12: Additional Lesion Found in Preoperative Breast MRI; Is Routine Use Justified?

Preoperative breast MRI has been increasingly performed in patients with newly diagnosed breast cancer due to its high sensitivity in assessing the extent and additional malignant foci. But due to it's low specificity, role of routine preoperative breast MRI has become an issue. In this study we aught to analyze the characteristics of the additional lesion found in preoperative breast MRI and to evaluate the clinicopathological factors in association with likelihood of having additional malignancy.

We retrospectively reviewed 2491 patients who undergone surgery due to breast cancer in Seoul National University Hospital(SNUH) between Jan 2006 and Dec 2010. Neoadjuvant chemotherapy cases, patients undergone initial sonography in other center or ones with prior excision were excluded and total 1068 patients were analyzed. The additional lesion was defined as the lesion not found in initial sonography and detected in preoperative breast MRI. The pathology of the additional lesion was reviewed and the association between the clinicopathologic factors and additional malignancy were evaluated. Accuracy of breast MRI was estimated regarding cancer yield, positive predictive value(PPV).

Mean age at diagnosis was 50.9 years (21 to 85 years). Overall detection rate of additional lesion was 26.2%(280 out of 1068). Mean size of the additional lesion was 9.8mm(3-51). Additional lesions consist of 99(35.4% of 280) C4 or higher, 174(62.1% of 280) below C4, 7(2.5% of 280) C0 lesions. Among them 100 patients undergone onstage surgery. 5(55% of 100) lesions were in ipsilateral breast and 45(45% of 100) in contralateral breast. Breast conserving surgery and mastectomy rate of the 100 onstage-operation group was 36% versus 64% and 64.6% versus 35.4% in total 1068 patients, showing no significant change of operation method of the primary cancer owing to additional lesion.

Among the 100 patients, 54(19.3% of 280) were benign, 3(1.1% of 280) were atypical ductal hyperplasia, 13(4.6% of 280) were in situ carcinoma, 19(6.8% of 280) were invasive carcinoma and 11(3.9% of 280) were unknown. Cancer yield was 2.99%(32 out of 1068) and PPV of preoperative breast MRI was 39.0%(31 out of 82).

In univariate analysis, young age and premenopausal patients showed to have higher rate of additional cancer found in MRI(p=0.022, p=0.036). Breast density, size and LN status of the primary cancer didn't show significancy and neither the hormone receptor status with each p value 0.705, 0.381, 0.973, 0.375 respectively. Lobular carcinoma(ILC or mixed IDC with ILC) and low grade carcinoma showed significancy of having additional malignancy (p=0.019, 0.022). In multivariate analysis age, low grade carcinoma and lobular carcinoma showed independent association with p value 0.014, 0.039, 0.035 respectively(HR 0.95, 95%CI:0.94 o 0.99),(HR 0.39, 95%CI:0.16 to 0.96),(HR 5.66, 95%CI:1.13 to 28.39).

Routine preoperative breast MRI use can result in overtreatment also with delay in surgical management. In our data, younger age, low grade carcinoma, lobular carcinoma showed independent association having additional malignant foci in breast MRI. With the basis of mammography and sonography, preoperative breast MRI should only be done when additional gain is considered to overcome the flaws.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-08-12.

Chemotherapy-induced amennorrhea: Predictive markers of response to neoadjuvant chemotherapy in breast cancer

281

Background: There is controversy about the benefit of chemotherapy-induced amenorrhea (CIA) in breast cancer. We investigated significance of CIA after neoadjuvant chemotherapy for predicting response to neoadjuvant chemotherapy in breast cancer patients.

Methods: We reviewed the records of 198 premenopausal patients with breast cancer treated with neoadjuvant chemotherapy between January 2005 and December 2010. Chemotherapy-induced amenorrhea (CIA) was defined as serum FSH level ≥40 IU/L after completion of all scheduled neoadjuvant chemotherapy and prior to definitive surgery.

Results: Among 198 breast cancer patients, 132 pts (66.7%) developed CIA after neoadjuvant chemotherapy. 156 pts (78%) underwent DA chemotherapy. The age of CIA patients was older than non-CIA patients (41.55±5.55 vs. 38.27± 6.86 years, p=0.001). The incidence of CIA after neoadjuvant chemotherapy was significantly higher in responder group (responder vs. nonresponder: 87 pts (74.4%) vs. 45 pts (55.6%); p=0.006). Additionally, FSH level after all scheduled neoadjuvant chemotherapy was significantly higher in responder group (FSH 56.41±32.41 mIU/ml vs. 45.76±30.31 mIU/ml; p=0.021). In univariate analysis, CIA (p=0.006) and total number of chemotherapy cycle regardless of chemotherapy regimen (p=0.04) were significantly predictive of tumor response. CIA was only significant predictive factor for tumor response after neoadjuvant chemotherapy on multivariate analysis (p=0.012).

Conclusions: CIA is independent predictive markers of response to neoadjuvant chemotherapy in locally advanced breast cancer.

Positional effects of polymorphisms in probe-target sequences on genoplot images of oligonucleotide microarrays

Single nucleotide polymorphisms (SNPs) present in probe-target sequences (SPTS) have been shown to be associated with abnormal genoplot images. We explored the effects of SPTS positions on genoplot images using a data set from a genome-wide association study typed on an Illumina Human Hap300 platform. We screened the physical genomic positions of 308,330 autosomal probes to identify SPTS candidates deposited in dbSNP. The genoplot images across 293 individuals were inspected further in SNPs bearing an SPTS candidate. We identified 35,185 SNPs bearing a single SPTS candidate, including 264 SNPs showing abnormal genoplot images. The frequencies of SPTS at distances within 10 bases from the target SNP were significantly higher in the 264 SNPs showing abnormal genoplot images, than in the remaining 34,921 SNPs (49.62 vs 12.87%; Fisher exact test; P = 2.2 x 10(-16)). Of these 264 SNPs, we randomly selected 20 SNPs and resequenced them in 97 individuals. An SPTS within 10 bases of the target SNP was confirmed in all 20 SNPs, except for one SNP with a small deletion (7 bases) in the probe-target sequence. Taken together, these results suggest an association of a proximal SPTS with an abnormal genoplot image, which could result in spurious genotype detections, highlighting the importance of minimizing systematic errors in microarray experiments.