Abstract P3-10-11: Clinical implications of 63-gene signature associated with response to PARP inhibition in triple-negative and luminal B breast cancer

Background: Two recent randomized phase 3 trials have demonstrated that treatment with PARP inhibitors results in an improvement in progression-free survival (PFS) in metastatic, BRCA-mutant, HER2-negative breast cancer patients. We have previously identified a pathway-enriched 63-gene expression signature predictive of response to olaparib in seven patient-derived xenograft breast tumors, with a high overall accuracy of 86%. We found that the prevalence of our gene signature was 45% in a cohort of triple-negative breast cancer (TNBC) patients. We wanted to better understand if there were correlations between our 63-gene signature and other known prognostic markers and to determine the prognostic significance of our mutational gene signature in different PAM50 breast cancer subtypes.

Methods: We used a publicly available dataset from the NCI GDC Data Portal of TNBC patients (n = 82) to undertake clinico-pathological correlations with our 63-gene expression signature. We correlated the presence or absence of the signature with age, tumor size, lymph node status, and stage using chi2 analysis, in addition to overall survival (OS) and PFS with STATA SE. We also correlated our gene signature with known TNBC subtypes from TNBCtype. Using the METABRIC cohort (n = 2509), we looked at the mutational frequency of our gene set in cBioPortal in different breast cancer subtypes and determined the prognostic value in each subtype.

Results: We did not find any statistically significant correlations between the 63-gene expression signature and age, tumor size, lymph node status, or stage amongst the 82 TNBC patients. All TNBC subtypes including 2 basal-like, immunomodulatory, low androgen receptor, 2 mesenchymal-based, and unspecified were identified in both gene-signature predicted sensitive and resistant groups, but there were no statistically significant differences between groups. The median follow-up of the TNBC cohort was 24 months, and no statistically significant associations were identified with OS or PFS. In the METABRIC cohort, the mutational frequency of any of the 63 genes for the following subgroups was identified: basal (n = 209), 85.2%; HER2+ (n = 224), 68.8%; claudin-low (n = 218), 48.2%; luminal B (n=475), 25.1%; and luminal A (n=700), 12.7%. The median follow-up of the METABRIC cohort was 127 months. We found that patients with a mutation in any of the 63 genes demonstrated a poorer overall survival, 122.8 months, in comparison to patients without any mutation, 164.6 months (P = 0.0002). In particular, luminal B patients with a mutation in any of these genes demonstrated a poorer overall survival, 90.0 months, in comparison to patients without a mutation, 132.1 months (P = 0.009). No statistically significant difference in overall survival was observed for patients with or without any mutation amongst the luminal A subtype (P = 0.26).

Conclusion: We found that patients with a mutation from our 63-gene set demonstrated a worse prognosis in comparison to patients without a mutation amongst the luminal B subtype. This is suggestive that there may be a role for our 63-gene signature to select patients amongst the luminal B subtype who may benefit from PARP inhibition.

Citation Format: Beniey M, Marois F, Haque T, Hassan S. Clinical implications of 63-gene signature associated with response to PARP inhibition in triple-negative and luminal B breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-10-11.

Occult gastrointestinal bleeding in high-risk intensive care unit patients receiving antacid prophylaxis: frequency and significance

Gastroccult reagent was used every 4 h to detect blood in gastric juice in 41 ICU patients at risk of GI bleeding (GB) and receiving antacid prophylaxis (gastric pH greater than 3.5). Of the present patients, 27% (11/41) had at least one episode of occult GB (three consecutive positive determinations; a total of 14 episodes). Endoscopy identified acute gastroduodenal mucosal lesions (stress ulcers) as the most frequent lesion in this group (eight patients). Sepsis was the most frequent underlying condition associated with occult GB due to stress ulcer. Hematemesis occurred in 36% (4/11) of patients with occult GB and was due to stress ulcer in three patients and to benign gastric tumor in one. No overt GB occurred in the absence of previous occult GB. We conclude that: a) risk of GB persists in critically ill ICU patients in spite of antacid prophylaxis (gastric pH greater than 3.5); b) high-risk patients can be identified through periodic testing for the presence of blood in gastric juice using the reagent; c) when occult GB occurs, treatment should be based on the endoscopy results. In the absence of acute gastroduodenal mucosal lesions, antacid prophylaxis should not be modified, and specific treatment of the identified lesion(s) should be initiated. In the presence of stress lesions, antacid prophylaxis should be reinforced if the pH of the gastric content is less than 3.5 and a septic complication should be actively sought if the pH is greater than 3.5.