[210] BRST2, ER, e-Cadherin, CDX2, and NKX3.1 Utility in Distinguishing Diffuse Gastric from Lobular Carcinoma

Gloria H Lewis, Robert West, Reetesh K Pai, Teri A Longacre, Kristin C Jensen. Stanford University, Stanford, CA; University of Pittsburgh, Pittsburgh, PA; VA Palo Alto HCS, Palo Alto, CA

Background: Distinguishing diffuse gastric (DGC) from lobular breast carcinoma in metastases is morphologically challenging. Few reports on the utility of immunohistochemistry (IHC) are published. We studied a series of gastric (GC) and breast carcinomas (BC) by IHC, including NKX3.1, a prostate marker reportedly expressed in ductal (9%) and lobular (27%) carcinomas.
Design: BRST2, ER, e-cadherin (ECD), CDX2, and NKX3.1 IHC performed on tissue microarrays containing 103 GC and 260 BC were divided based on infiltration pattern: diffuse (single file) vs clustered (tubules/clusters). Mixed cases were considered diffuse. Any IHC labeling was considered positive and correlated by subtype: GC vs BC and diffuse GC (DGC) vs diffuse BC (DBC). P-values were calculated by 2-tailed Fisher's exact test.
Results: IHC are summarized in table 1; sensitivity, specificity, positive (PPV) and negative predictive values (NPV) of DGC vs DBC are in table 2. There were 37 DGC and 99 DBC. ER positivity was seen in tumor cells of 3 GC all of which were DGC (8.6%); however, ER positivity was present in stromal cells in 44.6% (45/101) of GC with negative tumor. Other than GC, the 3 patients with ER positive tumors (2 females and 1 male) had no cancer history. Two cases of BC infiltrating in tubules/clusters (BCTC) negative for NKX3.1, BRST2, and ER were positive for CDX2. The 2 patients had no cancer history other than BC. NKX3.1 was present in comparable proportion of DBC (9.8%) vs BCTC (9.6%) (p=1.0) and in ECD positive (9.9%) vs ECD negative (5.9%) cases (p=0.7).

Table 1 IHC
 GC%BC%pDGC%DBC%p
BRST20 (0/103)74 (168/227)0.00010 (0/37)81.5 (66/81)0.0001
ER2.3 (3/101)75.9 (173/228)0.00018.6 (3/37)89.9 (89/99)0.0001
ECD96.1 (99/103)83.4 (176/211)0.000991.9 (34/37)67.5 (52/77)0.0049
CDX287.2 (89/102)0.9 (2/221)0.000191.9 (34/37)0 (0/79)0.0001
NKX3.10 (0/103)9.2 (21/228)0.00040 (0/37)9.8 (8/82)0.0564




Table 2 DGC vs DBC
 BRST2ERECDCDX2NKX3.1
Sensitivity %81.589.991.987.29.8
Specificity %10091.432.599.1100
PPV %10096.739.597.8100
NPV %71.276.289.394.433.3



Conclusions: BRST2, ER, and NKX3.1 are highly specific for BC, while CDX2 is highly specific and sensitive for GC. A panel including these markers is most useful in distinguishing metastatic GC from BC. However, ER should be interpreted carefully because ER can be positive in stromal cells in GC when tumor cells are negative. NKX3.1 may be useful, but is seen in less than 10% of BC. We found no correlation of NKX3.1 with either DBC or ECD negative BC.
Category: Breast

Wednesday, March 6, 2013 1:00 PM

Poster Session VI # 16, Wednesday Afternoon

 

Close Window