Kosiorek, Heidi

Background Endoscopic endonasal resection (EER) of sinonasal mucosal melanoma (SMM) is a newer surgical alternative to traditional external and/or open resection (OR). Studies on long-term outcomes are necessary to validate EER for this aggressive sinonasal malignancy. Objective To compare outcomes of EER versus OR in SMM. Methods A case series of patients who underwent surgical resection of SMM at a tertiary-care institution (2000–2015) was studied retrospectively. Demographics, tumor site and stage, surgical approach, surgical margin status, local control, and survival were compared between those who underwent EER and OR. Results Twenty-two patients met inclusion criteria. Nine underwent EER and 13 underwent OR. The mean age in the EER and OR groups was similar, 78.7 and 72.3 years, respectively. Two-thirds of patients were women (EER, 66.7%; OR, 61.5%). The nasal cavity was the most common primary tumor site (EER, 77.8%; OR, 84.6%). The local tumor stage in both groups was similar, with the majority of cases being T4 (EER, 55.6%; OR, 61.5%; p = 0.99). Negative margins were achieved in all EERs and in 69.2% of ORs. Median follow-up was 25.0 months for the overall group (range, 1.7–172.9 months), 32.6 months (range, 3.4–58.7 months) for EER and 14.1 months (range, 1.7–172.9 months) for OR cohorts. The 5-year overall survival was statistically similar in both groups (EER, 53.3%; OR, 22.7%; p = 0.214) as was disease-free survival (EER, 55.6%; OR, 22.8%; p = 0.178). Local control, however, was significantly higher in the EER cohort (EER, 85.7%; OR, 37.6%; p = 0.026). Conclusion In carefully selected patients with sinonasal melanoma, endoscopic surgery with an experienced team may offer comparable survival and improved local control over open surgery. Prospective, multicentered studies with larger cohorts are needed to validate these results.

101 Background: Diabetes mellitus (DM) has been associated with an increased risk of mortality among patients with many types of cancers. The aim of this case-control study was to determine the impact of DM on short-term overall survival in prostate cancer patients, and to examine the impact of prostate cancer on glycemic control in DM. Methods: 276 patients with newly diagnosed prostate cancer from 2007-2014 with DM were identified from the Institutional Cancer Registry and matched to 276 prostate cancer patients without DM according to age and year of prostate cancer diagnosis. The file was linked to the electronic medical record to obtain information on DM and cancer therapies, and laboratory results. Results: The mean age of the entire prostate cancer cohort was 72 years, most (93%) were white, the most common Gleason score (in 52%) was 7, and the majority (56%) were tumor stage 2. No differences in age, race/ethnicity, Gleason score, or tumor stage were detected between patients with and without DM, although DM patients had higher body mass index (P = .031). Among those with DM the mean hemoglobin A1c (HbA1c) was 6.7%. In Kaplan-Meier survival analysis (median follow-up time of 43.7 months), 5-year overall survival was estimated at 88% for DM patients versus 93% in non-DM patients. Hazard ratio (for matched pairs) was 1.64 (95% CI: 0.77-3.46, P = .20). Moreover, mean HbA1c and glucose values among DM cases did not significantly change over 1 year (P ≥ .13). Mean glucose among DM patients was significantly (p < 0.01) higher compared to non-DM patients. Conclusions: DM did not adversely impact survival in prostate cancer patients. In addition, prostate cancer or its treatment does not appear to affect glycemic control. Both patients and their providers can be reassured that the two concurrent diagnoses do not adversely interact to worsen short term outcomes.

Abstract Introduction: The programmed cell death-1 (PD-1) immune checkpoint is a normal regulator of the autoimmune response to inflammatory stimuli. Several cancers have taken advantage of this mechanism by harboring its ligand (PD-L1), thus activating the checkpoint to evade immune destruction. Increased expression of PD-L1 on tumor cells has been linked with worse outcomes in several cancers, including breast cancer. In triple-negative breast cancer (TNBC), expression of PD-L1 is increased compared to luminal cancers and as such, has become a favorable target for anti-tumor therapies. However, few studies evaluate the impact of PD-1 and PD-L1 expression on overall survival (OS) in TNBC. The aim of this study was to assess OS based on PD-1 and PD-L1 positivity and distribution in the tumor microenvironment. Methods: All patients with newly diagnosed TNBC from 1999 to 2015 with adequate pathologic specimen were identified for evaluation. Surgical pathology specimens were stained by immunohistochemistry (IHC) for PD-1 and PD-L1. Those considered PD-1+ or PD-L1+ had a staining score of 3, 4, or 5. Patient and clinical characteristics of the adjuvant group were compared by PD-1 and PD-L1 status with Chi-Square and ANOVA, as appropriate; characteristics of the neoadjuvant population were described. OS was calculated using Kaplan-Meier method and compared by Log Rank test for PD-1, PD-L1 positivity and tumor microenvironment status. Results: Fifty-two patients met inclusion; 38 had surgery prior to chemotherapy while 14 had neoadjuvant chemotherapy. There were 15 PD-1+ patients (39%) and 21 PD-L1+ (55%) patients in the treatment naive cohort. No significant difference was found in the treatment naive group between age, race, postmenopausal status, final stage, or pathologic characteristics (e.g. tumor size, mitotic rate, and grade) for all PD-1+ and PD-L1+ patients. Of the 21 PD-L1+ patients, staining was present in the tumor, tumor/non-tumor (mixed), and non-tumor microenvironment 42.9%, 33.3%, and 23.8% of the time, respectively (p=0.008). No significant difference was found in OS between PD-1+ and PD-1- patients (p=0.509) or PD-L1+ and PD-L1- patients (p=0.240). However, distribution of PD-L1 staining significantly associated with OS. Those demonstrating PD-L1 staining in the tumor had a significantly higher OS than those with PD-L1 staining in the non-tumor microenvironment (p=0.008). In the subset of PD-L1+ patients, those with staining in the tumor show a 10-year OS of 100% versus 60% for those with staining in non-tumor (p=0.006). In the neoadjuvant cohort, 14.3% of patients expressed PD-1 positivity, while 42.9% had PD-L1 positivity; the distribution of the PD-L1 staining in tumor 14%, mixed 29%, and non-tumor 50%. No difference in OS was found based on PD-1 or PD-L1 (p=0.660) positivity in the neoadjuvant cohort. Conclusion: A large number of TNBC stain for PD-1 and/or PD-L1. The implications of this are yet to be determined. In this small cohort, pattern of staining had a significant impact on overall survival in treatment naïve patients. This finding will need to be validated on a larger cohort but may have important implications for treatment by defining patients who may benefit by checkpoint inhibitor therapy. Citation Format: Linnaus ME, Kosiorek H, Barrett MT, Dueck A, Anderson KS, Ocal IT, McCullough AE, Annamalai L, Yearley JH, Pockaj BA. Effect of PD-1 and PD-L1 in the tumor microenvironment on overall survival of triple-negative breast cancer patients [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr PD5-03.

Abstract Introduction: Homologous recombination (HR) is essential for repairing double-stranded DNA breaks. Tumors deficient in DNA HR display high levels of copy number aberrations including amplifications, deletions, and breakpoints. Studies in BRCAmut cancers have shown that HR deficiency (HRD) renders triple negative breast carcinoma (TNBC) sensitive to carboplatin or Poly(ADP-ribose) polymerase (PARP) inhibitors via a synthetic lethal interaction, as previously described. Clinical observations in breast and other solid tumors suggest that the HRD phenotype may arise independently of BRCA1/2 mutation. The implications of HRD enrichment in PD-L1, PD-L2, Jak2 (PDJ) 9p24.1 amplified TNBC are not known. Our goal was to determine the HRD status in PDJ enriched tumors, and the potential association of the PDJ amplicon with a BRCA like HRD phenotype. Methods: Archival fresh frozen and formalin-fixed paraffin embedded (FFPE) tumor samples from 41 patients with TNBC were obtained from 1998-2015. Tumor populations (diploid, tetraploid, and aneuploidy) were sorted using DNA content flow cytometry and interrogated with oligonucleotide array comparative genomic hybridization (aCGH). Tumor homologous recombination deficiency (HRD) scores were calculated by number of interstitial aberrations (IA) in their genomes (HRD-IA). A cut-off of &lt; 40 (low), &gt; 40 (high) or &gt; 50 (very high) was assigned based on our studies of known BRCAmut tumors. Data were compared between HRD groups using chi-square, ANOVA F-test or log-rank test. Results: 14/41 (34%) TNBC patients had high HRD-IA score and 6/41 (14.6%) had very high HRD-IA score. The median age was 61.8 years (SD = 12.8), the average tumor size was 3.1cm (SD = 2.6), 42% were node positive, 24% were pre-menopausal, 19% were Black, Hispanic, or Asian/Pacific Islander. Genetic testing was negative for BRCA1/2 mutations in 9/41 patients, including 3 high and very high HRD-IA score patients. There was no difference in the OS at 5-years [74% versus 59% (p = 0.907)] in the high HRD-AI group. 12/41 (29%) patients were PDJ+. In the PDJ+ patients, 4/12 (33.3%) had HRD &gt; 40 and 1/12 (8.3%) had HRD &gt; 50 compared to the PDJ- patients where 10/29 (34%) had high HRD-IA score and 5/29 (17%) had a very high HRD-IA score, p=0.463. Patients with inflammatory TNBC were significantly more likely to have high HRD-IA score 4/6 (66%) compared to non-inflammatory TNBC 8/31 (26%), p=0.05. Conclusions: We describe a unique group of TNBC patients with similar patient characteristics, tumor size, nodal status, and stage when compared by low, high and very high HRD-IA score. For inflammatory TNBC, there was statistically significant percentage of patients with a high HRD-IA score. We did not see a statistically significant correlation with 9p24.1 amplification based on HRD-IA score which has been suggested in other data sets. The HRD-IA score did not statistically impact OS but there was a suggestion that those with a higher score may have an improved OS potentially due to chemotherapy sensitivity. Further study with a greater number of patients to adequately power the study is needed to determine whether our initial findings can be confirmed and whether the HRD-IA score can be used to define a specific subset of TNBC patients. Citation Format: Gawryletz CD, Anderson KS, Northfelt DW, Kosiorek HE, Linnaus ME, Ocal IT, McCullough AE, Pockaj BA, Barrett MT. Homologous recombination deficiency in PD-L1, PD-L2, Jak2 (PDJ) amplified triple negative breast carcinoma [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-05-19.

Abstract Introduction: We have previously identified a 9p24.1 amplicon targeting PD-L1/JAK2 (PDJ) in subset (26.7%) of triple negative breast cancer (TNBC) patients. The PDJ amplicon is common in newly diagnosed and chemotherapy treated TNBCs and is associated with a worse prognosis. Our goal in this study was to determine whether immunohistochemistry (IHC) evaluation could identify those patients who harbor the PDJ amplicon. Methods: TNBC patients from 1999 to 2015 whose tumors were flow-sorted and evaluated by array-based comparative genomic hybridization (CGH) were identified; paraffin slides were obtained for IHC staining evaluation of JAK-2, phosphorylated STAT3 (pSTAT3), and PD-L1. Pathologic analysis consisted of scoring the stains for intensity (0-3+) and the relative percent of tumor cells with positive staining; positive score was defined as 3+ for any percent staining. Statistical analysis of IHC staining was performed to determine association with the PDJ amplicon defined by focal 9p24.1 copy number gain with CGH log2ratios of &gt;1.0 for each sorted TNBC sample. Results: Eleven of 43 TNBC patients evaluable by IHC had the PDJ amplicon. There was no association between PDJ amplification and IHC staining for JAK-2, pSTAT3, or PD-L1 regardless of staining intensity or percentage tumor cells positive. Table 1 describes PDJ amplicon status and positivity for JAK2, pSTAT3 and PD-L1. Of PDJ-positive samples, 64%, stained positive for JAK2, 27% positive for pSTAT3, but 0% for PD-L1 in tumor cells. However, in the PDJ-negative group, 69% still stained positive for JAK2, 19% positive for pSTAT3, and 9% for PD-L1. Table 1: PDJ amplicon status and IHC staining JAK2 (3+) (n;%)pSTAT3 (3+) (n;%)PD-L1 (3+) (n;%)PDJ - (n=32)22;69%6;19%3;9%PDJ + (n=11)7;64%3;27%0;0% JAK2: Most PDJ-positive samples (10/11, 91%) showed some JAK-2 staining. Although one had only 1+ staining at 5% of the cells, the majority (7/11, 64%) had staining of 3+ and &gt;10% of the cells with 2 patients demonstrating 3+ staining of 75% and 90% of the cells. However, those patients without the PDJ amplicon also exhibited JAK2 staining, with 66% (n= 22) of PDJ-negative cases staining strongly for JAK-2 (3+) and &gt; 10% of cells. pSTAT3: All but one PDJ-positive case demonstrated some staining (1+ or greater) for pSTAT3; the staining intensity and percent positivity were much less than the JAK-2 staining with only 2 cases (18%) having 3+ staining of &gt; 10%. Similar staining was seen in the PDJ-negative cohort; 75% of PDJ-negative patients had some pSTAT3 staining (1+ or greater) and 1 patient (5%) exhibited 3+ staining &gt; 10%. PD-L1: Finally, the PD-L1 staining was low overall with only 18% of PDJ-positive cases demonstrating some staining for PD-L1. Notably, only one case stained 3+ with 20% positivity while the majority of the samples had a 2+ intensity encompassing 5-50% of the cells. Conclusions: IHC staining for JAK2, pSTAT3, or PD-L1 was not associated with the presence of the PDJ amplicon. Notably positive IHC staining for JAK2 was observed for both PDJ-positive and PDJ-negative tumor cells, thereby nullifying its application as a screening tool for PDJ amplification. Alternative methods, such as fluorescence in situ hybridization, are needed to identify PDJ-positive patients for further study. Citation Format: Linnaus ME, Kosiorek H, Ocal IT, Dixon L, Barrett MT, Gawryletz CD, Anderson KS, McCullough AE, McEachron TA, Pockaj BA. Immunohistochemistry cannot be used to detect PD-L1/JAK-2 amplification [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-03-13.

248 Background: Improved understanding of TME in CCA may allow for development of novel therapeutics targeting stromal components. Hyaluronic acid (HA) is a polysaccharide that has been found in the TME of a number of solid tumors characterized by desmoplastic stroma, such as pancreatic cancer. Elevated HA levels were prognostic in these malignancies. HA prevalence and association with clinic-pathological features in CCA is unknown. Methods: A retrospective study was conducted using clinically annotated formalin-fixed paraffin embedded tissue from resected intrahepatic CCA samples. Specimens were analyzed for HA, scored using a positive pixel count algorithm (Aperio Imagescope) and positivity was calculated as % of total pixel count. Clinical variables were examined by chi-square or ANOVA F-test. Overall survival (OS) and progression-free survival (PFS) were estimated using Kaplan-Meier method and compared via log-rank test. Multivariate analysis was conducted using Cox proportional hazards regression models. Results: 142 patient samples were stained for HA. HA was identified in all samples (100%), with a range of HA positivity of 32.8 – 91.8%. Median HA value was 67.1% and 95% of patients had HA positivity ≥ 45%. Factors associated with higher than median HA ( > 67%) included perineural invasion (PNI) (40% vs. 21%, p = 0.011), necrosis (67% vs. 41%, p = 0.003), lymph node positivity (32% vs. 14%, p = 0.027), and small tumor size ( < 6.8 cm vs. > 8.4cm, p = 0.014). Stratification of HA staining by median positivity value was not an independent predictor of survival based on multivariate analysis. OS for patients with HA > 67% was 40.1 months (mo) vs 48.5 mo with HA ≤ 67%, p = 0.537. PFS for patients with HA > 67% was 14.4 mo vs. 25.7 mo with HA ≤ 67%, p = 0.177. Conclusions: Prevalence of HA staining in resected CCA is higher than seen in many solid tumors and is ubiquitous in our study. Absence of statistically significant association with OS and PFS might be attributable to ubiquitous prevalence, narrow signal range, and association with known negative prognostic factors. Further pre-clinical and clinical evaluation would be warranted.

204 Background: Allogeneic hematopoietic cell transplantation (HCT) is a medically complicated treatment modality used for various hematologic malignancies. Patterns and predictors of distress in post- transplant setting in a contemporary cohort of patients are not well-understood. Methods: 67 patients transplanted between 12/5/12- 4/21/15 completed National Comprehensive Cancer Network Distress Thermometer as part of their long term follow up clinic evaluation on day+100 after HCT. Demographic, socioeconomic and clinical outcome data were analyzed. A score ≥ 4 was considered distressed. Results: Mean age of patients was 50.8 years (range 20-72). Most patients were married or living with their significant other (83%), half were college graduates (58%), and half were employed full time (50%). 21 patients (31%) reported a distress score ≥ 4. Distressed patients reported increased problems with treatment decisions (15% vs 2%, p<0.05) and ability to have children (16% vs 0%, p<0.05). Physically, distressed patients were more likely to report fatigue (85% vs 44%, p <0.05), pain (60% vs 24%, p<0.05) and difficulty getting around (25% vs 7% p<0.05) Distressed patients were likely to report at least 1 emotional problem (N=16, 76%), and nearly a quarter reported ≥ 4 emotional problems (N=5, 24%). Specifically, sadness (25% vs 7%, p<0.05) and worry (64% vs 26%, p <0.01) were more common in distressed patients. Distressed patients had a longer length of hospitalization for transplant (30.8 days vs 24.9 p<0.001) and more hospitalized days during the first 100 days (34.9 vs 28.7 p<0.01). Conclusions: Approximately 30% of allogeneic HCT patients suffer significant distress at Day100 post-transplant. Distressed patients report problems with a variety of factors, including emotional, physical, family and practical issues that may impact function, compliance and quality of life. Higher number of hospitalized days is associated with increased distress and could be used to identify patients who may benefit from stress monitoring. Periodic screening of distress in HCT patients can help identify interventions to help improve patient reported outcomes.

6 Background: Traditional screening mammography (MG) enables early detection of breast cancer and therefore results in decreased mortality. Mammography, however, is an imperfect tool and limited by an overall sensitivity of 75-85%. Specificity is limited as well resulting in many false-positive examinations and unnecessary follow-up examinations. The recent development of contrast-enhanced digital mammography (CEDM) has made the usage of IV contrast enhancement with mammography a possibility. CEDM is an angiographic technique that uses digital subtraction, which removes the normal mammary gland on images in order to detect breast tumors and visualize their vascularity. The literature is promising with regard to CEDM as studies have found that CEDM achieves increased sensitivity without reductions in specificity in comparison to MG, US and/or MRI. Methods: In this multi-reader study we compared the diagnostic accuracy of CEDM as an adjunct to MG/US in patients with suspicious findings (BI-RADS 4). The study population included 31 patients who were imaged with CEDM prior to their scheduled US or stereotactic-guided biopsy. We compared the relative performance of MG/US alone versus MG/US/CEDM in determining the probability of malignancy. Histology served as the gold standard. Results: See table below. Conclusions: The addition of CEDM to the workup of suspected lesions on MG and/or US preliminarily decreases the number of false-positive MG thereby with potential to reduce cost and unnecessary patient harm. [Table: see text]