Invasive lobular carcinoma (ILC) makes up 10-15% of all breast cancer. While neoadjuvant therapy (NAC) is increasingly used in locally advanced invasive cancers, some evidence suggests that ILC does not respond to NAC. The I-SPY trial has shown an association between expression of high risk gene signatures and increased pathologic complete response (pCR) to NAC. We hypothesized that compared to invasive ductal carcinoma (IDC), ILC would have decreased expression of high risk gene signatures, and decreased rates of pCR. Methods
We analyzed data from 221 patients in the I-SPY trial, a prospective, multicenter neoadjuvant breast cancer trial with serial pathology, imaging, and expression arrays. A dedicated breast pathologist at each site reviewed histology, and e-cadherin staining was performed centrally on cases identified as ILC. We compared rates of wound healing signature activation, NKI (Netherlands Cancer Institute) profile status, and proportion of pCR in ILC compared to IDC. Other comparisons were the presence of p53 mutations, molecular subtype, and MRI phenotype. We used the chi-squared test, Fisher's exact test, and Student's t-test for comparisons. Results
Of 221 patients, histology on core biopsy was ductal in 195 (88%), lobular in 18 (8%), and mixed ducto-lobular in 8 (3.6%); the last cases were included in the IDC group. ILC and IDC cases were similar by tumor size, age at diagnosis, and race. Expression array data were available on 12 of 18 (67%) ILC and 149 of 203 (73%) IDC. The proportion of p53 mutations, basal subtype and NKI high risk did not differ significantly. However, ILCs were significantly less likely to express the activated wound healing signature compared to IDCs (50% vs 80%, p = 0.03) and had different imaging phenotypes, with most ILCs appearing multinodular or diffuse on MRI (p = 0.028). Three of 18 (17%) ILCs had a pCR, compared to 55/197 (28%) IDCs. Of the 18 ILC, 11 of 15 tested were e-cadherin negative (73%). Of 18 ILCs, 16 (89%) were hormone receptor (HR) positive, and comparison to HR+ IDC showed similar rates of pCR in the two groups (19% pCR in HR+ ILC vs 14% pCR in HR+ IDC). There was no significant difference in expression of the activated wound healing signature when comparing ILCs to HR+ IDC. Conclusions
There is significant heterogeneity within ILC, such that no one molecular subtype or genetic expression signature identifies all ILCs. The rates of pCR among ILC are similar to those seen in HR+ cancers. Although larger studies are needed to confirm these findings, we saw no evidence that ILC differs from IDC in response to NAC.