Gabriel, Emmanuel M

Abstract Background: DNIV accounts for 6%–10% of newly diagnosed breast cancer cases. Despite widespread mammography screening, its incidence is increasing in the United States and survival of this disease has only modestly improved since the late 1970s. As patient data accumulates in EHR, it’s promising to generate practice-based evidence through utilization of observational data sources. However, assembly of a DNIV cohort based on EHR data is challenging, as key pathologic and staging information are stored in unstructured clinical narratives and not available as structured data. In this study, we developed a rule-based algorithm to phenotype DNIV using natural language processing (NLP) techniques, and implemented the algorithm on our institutional EHR to extract potential DNIV cases. Methods and Results: We defined DNIV as those with either (1) M1 disease identified at time of initial presentation or M1 disease identified within 4 months after definitive surgery. We first developed a reference case list of DNIV verified by physician chart review of the EHR. We next refined the algorithm on a training dataset containing 51 positive and 38 negative reference cases. Next we tested the performance on the testing data containing 23 positive and 55 negative cases. The phenotyping algorithm identified key data elements using NLP, i.e., stage IV breast cancer, definitive surgery, stage 0-III, recurrent breast cancer and associated dates. To identify DNIV cases, phenotyping algorithm integrated temporal relations among the key data elements. The following steps were conducted in the following sequential order: (1) Identification of patients with breast cancer diagnosis using ICD-9 and ICD-10 codes. (2) Patients are positive cases if there are explicit mentions that delineate DNIV from recurrent metastatic breast cancer, such as “de novo stage IV” or “primary intact” detected by NLP at time of diagnosis. (3) Otherwise, patients with definitive surgery are selected if stage IV was within 5 years before definitive surgery or within 4 months after definitive surgery, along with patients with stage IV but without definitive surgery. (4) We further excluded patients with stage 0-III detected within 5 years after stage IV. (5) We further excluded patients with recurrent breast cancer detected before or at time of detection of stage IV. The remaining patients were left as positive cases. Precision of the algorithm was 70%, recall was 87% and F1, the weighted average of precision and recall, was 77%. We implemented our algorithm to interrogate 10 million clinical documents in a cohort of 56,548 patients with breast cancer diagnosis codes who presented to our institution between 2004 and 2018 and had research authorization. We identified 1918 potential DNIV cases. Conclusion: Our future focus is on algorithm refinement. An algorithm-generated cohort could serve as a data source for further study on outcomes related to DNIV and ideally as automated data abstractor and staging process. Citation Format: Liwei Wang, Karthik Giridhar, Kimberly Corbin, Brenda Ernst, Sadia Choudhery, Emmanuel Gabriel, Feichen Shen, Hongfang Liu, James Jakub. Identifying de novo stage IV breast cancer (DNIV) cases in Electronic Health Records (EHR) using natural language processing [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-050.

Background. Acral melanoma (AM) is a rare subtype of melanoma, which is one of the least common in Caucasian patients but is a common subtype of melanoma in Chinese patients. It is unclear if prognosis differs between Chinese and Caucasian patients diagnosed with AM. The aim of our study is to investigate patient characteristics and survival differences between Chinese and Caucasian AM patients. Methods. Two large institutional melanoma databases from Fudan University Shanghai Cancer Center (FUSCC) and Mayo Clinic enterprise were retrospectively reviewed from 2009 to 2015. Clinicopathologic and survival data were collected and analyzed between the two groups. The primary outcome was disease-specific survival (DSS) and was calculated using the Kaplan Meier (KM) method. Results. The Chinese group presented with more advanced disease compared with Caucasians: thicker Breslow depth (median 3.0 mm vs. 1.2 mm, p = 0.003 ), more ulcerated disease (66.1% vs. 29%; p < 0.001 ), and advanced stages (stage II/III 84.3% vs. 37.1%; p < 0.001 ). No significant difference was identified in terms of age at diagnosis, location, histologic subtypes, or node positive rate. The 5-year DSS rate was 68.4% and 73% ( p = 0.56 ) in Chinese and Caucasians AM patients, respectively. Male gender, Breslow thickness, ulceration, and positive sentinel lymph nodes were independent poor prognostic factors on multivariate analysis. Conclusions. There appears to be no difference in stage-stratified survival between Chinese and Caucasians, supporting the implementation of clinical trials for AM that could include both Chinese and Caucasian patients.

Background Recent studies have shown an increase in the incidence of gastric cancer (GC) among young adults in Asia and Latin America. However, it is unknown if a similar trend is happening in the United States (US). Methods A retrospective review of the National Cancer Data Base was conducted to identify patients diagnosed with gastric adenocarcinoma between the years of 2004 and 2013. Results A total of 93 734 patients were included. The 2 age groups below 40 did not see a change in GC incidence; however, age groups above 40 had increasing incidence. Patients aged 18-25 had the largest proportion of stage 4 disease and a poor survival (median 11.5 months), compared to older patients. Conclusion Despite the increasing trend of GC among individuals, the incidence of GC among young adults is not increasing. However, this subpopulation presents at more advanced stages (clinical stage 4) and thus has worse survival.

AbstractBackgroundRelatively few cutaneous head and neck melanoma (CHNM) patients with were included in the multicenter selective lymphadenectomy trial II (MSLT‐II). Our objective was to investigate whether immediate completion lymph node dissection completion of lymph node dissection (CLND) was associated with survival benefit for sentinel lymph node (SLN) positive CHNM using the National Cancer Database.MethodsSLN positive patients with CHNM from 2012 to 2014 were retrospectively analyzed. Patients were divided into two groups: those who underwent SLN biopsy (SLNB) only versus those who underwent SLNB followed by CLND (SLNB + CLND). The primary outcome was 5‐year overall survival (OS).ResultsAmong 530 SLNB + patients, 342 patients underwent SLNB followed by CLND (SLNB + CLND). The SLNB only group had fewer positive SLN, less advanced pathologic stage, and a lower rate of adjuvant immunotherapy. There was no significant difference in 5‐year OS between the two groups (51.0% vs 67%; P = .56). After adjusting for pathologic stage, there remained no difference in 5‐year OS among patients with stage IIIA (63.0% vs. 73.6%, P = 0.22) or IIIB/IIIC disease (39.1% vs 57.8%; P = .52).ConclusionsUsing a large nationwide database, CLND was not shown to be associated with improved OS for patients with SLNB positive CHNM, validating the results of MSLT‐II.

Background Recent studies have shown an increase in the incidence of gastric cancer (GC) among young adults in Asia and Latin America. However, it is unknown if a similar trend is happening in the United States. Methods A retrospective review of the National Cancer Database was conducted to identify patients diagnosed with gastric adenocarcinoma between the years of 2004 and 2013. Results A total of 93 734 patients were included. The two age groups below 40 did not see a change in GC incidence; however, age groups above 40 had increasing incidence. Patients aged 18 to 25 had the largest proportion of stage 4 disease and a poor survival (median 11.5 months), compared to older patients. Conclusion Despite the increasing trend of GC among individuals, the incidence of GC among young adults is not increasing. However, this subpopulation presents at more advanced stages (clinical stage 4) and thus has worse survival.

Background: Gastric cancer is one of the leading causes of cancer-related mortality worldwide, accounting for 8.2% of cancer-related deaths. The purpose of this study was to investigate the geographic and sociodemographic disparities in gastric adenocarcinoma patients. Methods: We conducted a retrospective study in gastric adenocarcinoma patients between 2004 and 2013. Data were obtained from the National Cancer Data Base (NCDB). Univariate and multivariable analyses were performed to evaluate overall survival (OS). Socio-demographic factors, including the location of residence [metro area (MA) or rural area (RA)], gender, race, insurance status, and marital status, were analyzed. Results: A total of 88,246 [RA, N = 12,365; MA, N = 75,881] patients were included. Univariate and multivariable analysis showed that RA had worse OS (univariate HR = 1.08, p < 0.01; multivariate HR = 1.04; p < 0.01) compared to MA. When comparing different racial backgrounds, Native American and African American populations had poorer OS when compared to the white population; however, Asian patients had a better OS (multivariable HR = 0.68, p < 0.01). From a quality of care standpoint, MA patients had fewer median days to surgery (28 vs. 33; p < 0.01) with fewer positive margins (6.3% vs. 6.9%; p < 0.01) when compared to RA patients. When comparing the extent of lymph node dissection, 19.6% of MA patients underwent an extensive dissection (more than or equal to 15 lymph nodes) in comparison to 18.7% patients in RA (p = 0.03). Discussion: This study identifies socio-demographic disparities in gastric adenocarcinoma. Future health policy initiatives should focus on equitable allocation of resources to improve the outcomes.

Abstract Background: Immune checkpoint inhibitors (ICPIs) have led to dramatic improvement in outcome of several cancers. Program death ligand1 (PD-L1) staining and tumor mutational burden (TMB) have emerged as independent predictive biomarkers of ICPIs in lung cancer. Here we examine the landscape of TMB and PD-L1 expression in breast cancer and present a case of patient with high TMB and PD-L1 negative breast cancer with exceptional response to ICPIs. Methods: Hybrid-capture based comprehensive genomic profiling of 395 cancer related genes using the FoundationOne assay was performed on 14,867 breast carcinomas sequenced in the course of routine clinical care. Ventana (SP-263) PD-L1 status (n=1425) and hormone receptor status was available for a subset of patients. Subgroup analyses were performed based on histological type [invasive lobular carcinoma (ILC, n=740)], molecular subtypes [ER-positive (ER+; n= 1371), HER2-amplified (HER2+; n=1522), and TNBC (n=917)], patient age (≤45, 46-60, ≥61), and local vs. metastatic disease (n=5241 and 6710). Results: Consistent with previous reports, the rates of positive PD-L1 staining are highest in TNBC (14%) and lowest in HER2+, ILC, and ER+ disease (6.0%, 5.1%, 2.3%). Interestingly, the rate of PD-L1 positivity, defined as ≥1% tumor staining, was significantly lower in metastatic disease vs. local disease (6.3% vs. 11.1%; p = 0.005). The frequency of high TMB, defined as &gt;10 mutations/mb, was greatest in ILC and HER2+ disease (13.6% and 9.9%) and lowest in TNBC and ER+ disease (7.0% and 6.9%). Rates of high TMB were associated with increased patient age (3.7%, 9.3%, and 12.8% frequency in patients ≤45, 46-60, and ≥61) and were significantly higher in metastatic vs. local disease (11.1% vs 5.3%; p&lt;2E-23). PD-L1 positive and TMB high populations were not significantly co-occurrent (OR = 1.02, p = 0.87). Similar percentages of PD-L1 positivity were observed in both TMB low (9.3%) and TMB high (9.5%). However, among patients with very high TMB (&gt;20 mut/mb), there was a significant association between TMB and PD-L1 positivity (OR = 2.6, p = 0.023). Nevertheless, even with high cutoff, 79% of the TMB high samples were PD-L1 negative. We also report on a stage IIIb (T4, N2, M0) ER+ HER2- breast cancer patient with high TMB (40muts/mb) and negative PD-L1 IHC who previously progressed on aromatase inhibitor with CDK4/6 inhibitor and chemotherapy but achieved durable complete response of &gt; 1 year with nivolumab in combination with capecitabine. Conclusions: Predictive biomarkers for ICPIs are critical to identify a subset of breast cancer patients who may respond to immunotherapy. TMB high and PD-L1 positivity do not significantly co-occur and the majority of TMB high cases were PD-L1 negative. Nevertheless, this group of patients may still benefit with ICPIs. Further studies are needed to evaluate this subset of patients. Citation Format: Ethan Sokol, Lee Albacker, Aixa Soyano, Ricardo Parrondo, Brenda Ernst, Emmanuel Gabriel, Garrett Frampton, Jeffrey Ross, Siraj Ali, Jon Chung, Saranya Chumsri. Incidence of high tumor mutation burden (TMB) and PD-L1 positivity in breast cancers and potential response to immune checkpoint inhibitors (ICPIs) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4894.

Background: External ear melanoma (EEM) is a rare condition with controversies in the literature. We analyzed patients with EEM in the United States compared to other head and neck melanomas (OHNMs). Methods: The National Cancer Database (NCDB) was used to select patients with head and neck melanoma from January 1, 2004 to December 31, 2015. Mann–Whitney and χ 2 tests were used to estimate statistical significance, and multivariate logistic regression to identify independent associations adjusted for confounders. Results: A total of 137,233 patients met the study criteria. Among them, 16,991 (12.4%) had EEM and 120,242 (87.6%) had OHNM. For patients with EEM, the mean (standard deviation) age was 66.26 (15.798) years. Most of the patients with EEM were men (85.5%), insured by Medicare (52.4%), and treated in Academic/Research Programs (47.7%) or Comprehensive Community Cancer Programs (32.3%). Most of the EEM tumors had invasive behavior (68.0%) were Stages 0 (30.3%) or I (40.3%), and were without ulceration (76.9%). Mean time to receive any treatment was 14.1 days for EEM compared with 14.6 days for OHNM (P < 0.001). We noticed a greater proportion of EEM in men (14.8%; adjusted odds ratio [aOR] 2.72 [2.605–2.852]; P < 0.001) compared to women (6.22%; reference). EEM was an independent factor for tumor Stage I (14.47%; aOR 1.61 [1.101–1.224], P < 0.001) and invasive behavior (13.86%; aOR 1.268 [1.15–1.389]; P < 0.001) compared to OHNM. Conclusion: EEM was associated with higher odds of invasive behavior compared to OHNM. Furthermore, men were found to have a higher likelihood to develop EEM compared to women.

Background: Eyelid melanoma (EM) is a rare condition that accounts for <1% of all cutaneous melanomas. In this analysis, patients diagnosed with EM in the United States were compared to those with other melanomas in the head and neck. Methods: The National Cancer Database was used to select patients with head and neck melanoma from 2004 to 2015. Results: A total of 137,233 patients met the criteria of the study. Among them, 2694 were diagnosed with EM (2%) and 134,539 in other head and neck locations (98%). Regarding the patients with EM, the mean age was 68.56 years (standard deviation 14.04). A greater number of them were treated in Academic/Research Programs (53.6%), insured by Medicare (55.7%), tumor in situ (52.2%), stage 0 (49%), and without ulceration (75.9%). The mean time to treatment commencement was 18.96 days for EM compared to 14.4 days for other melanoma in the head and neck regions (P < 0.001). There was a greater proportion of EM in female patients (odds ratio [OR] 2.371 [2.191–2.566], P < 0.001), compared to male; age greater than 80 years (OR 1.622 [1.360–1.934], P < 0.001) compared to 0 to 49 years. Moreover, EM was an independent predictor for treatment in Academic/Research Programs (OR 1.258 [1.160–1.365], P < 0.001) and diagnosis at autopsy (OR 1.414 [1.083–1.847], P = 0.011) compared to other melanoma in the head and neck region. Conclusion: These novel findings have the potential to guide clinical decision making in head and neck melanoma management.