Abstract 48P
Background
Atypical ductal hyperplasia (ADH) histologically has the same cytological and architectural features as low-grade ductal carcinoma in situ (LGDCIS). ADH itself is a benign condition but it is a known risk factor for invasive breast cancer (IDC), patients diagnosed with it are at a 4-5-fold increased risk of developing breast cancer in the next 5 years. ADH alone does not require any treatment, however, the standard of care is for the patient to proceed to surgery with partial mastectomy. The rationale for this is that studies have shown that between 5 to 50% of cases are upgraded to high grade DCIS or IDC with analysis of a larger tissue sample. Studies have looked at factors that may be associated with a risk of an upgrade without success resulting in a large number of women undergoing a surgical procedure. Our study aims to measure Large-scale DNA organization (LDO) in normal and LGDICS ducts between specimen with or without adjacent IDC.
Methods
Paraffin blocks from 21 patients were retrieved, 11 from pure DCIS and 10 from DCIS later upgrade to IDC. Tissue sections were stained with H&E and reviewed by a pathologist to identify areas. Another adjacent section was stained with Feulgen-thionin, a stochiometric stain for DNA. These 2 slides were scanned with the 3DHISTECH scanner. Regions of interests identified on the H&E images by our expert pathologist were identified on the Feulgen images. Each area was imaged at high resolution and exported onto our software. We imaged DCIS duct areas AND normal duct areas and randomly selected a few hundred nuclei per specimen LDO analysis was performed by calculating over 150 nuclear features form about 100 nuclei from normal and LGDCIS. Random Forest algorithm was built to discriminate 1) Normal duct nuclei between PURE specimens and MIXED specimens and 2) LGDCIS duct nuclei between PURE and MIXED specimens.
Results
We obtained an accuracy of respectively 67% and 72% to correctly classify 64 494 normal nuclei and 20025 DCIS nuclei using LDO features.
Conclusions
This preliminary study confirmed the potential of large-scale DNA organisation to identify DCIS cases surrounded by IDC. If validated, it could be used to predict on breast biopsies the likelihood of an upgrade on excisions, and therefore assist in appropriate clinical management.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.