CMET-36. IMMUNOTHERAPY VERSUS STANDARD OF CARE IN MELANOMA BRAIN METASTASES WITH KNOWN BRAF STATUS

INTRO/OBJECTIVE

A mutation of the BRAF protein is seen in approximately 50% of melanoma patients. Immune checkpoint inhibitors (ICI) are standard therapy in melanoma patients independent of a patient’s BRAF status. The primary objective of this study is to investigate the impact of BRAF status in patients treated with ICI compared to non-ICI systemic therapy on overall survival (OS) in patients with melanoma brain metastasis (MBM).

METHODS

We reviewed 351 patients with MBM treated at our tertiary care center between 2000 and 2018. Of these, 144 had known BRAF status, 71 of which were BRAF mutant and 73 were BRAF wild-type. OS was calculated from the date of diagnosis of MBM to compare the efficacy of ICI to other systemic therapies. Many of these patients received multiple lines of treatment including targeted therapies at some point during their care. The log-rank test and Cox proportional hazard model was utilized to determine differences in OS.

RESULTS

Eighty-four percent of patients received local therapy that included either surgery, stereotactic radiosurgery, or whole brain radiation therapy. In BRAF wild-type patients, 40 received ICI and 33 underwent non-ICI systemic therapy with a median survival (5.6 vs 7.1 months) and 2-year survival (28% vs 32%), respectively (p=0.64). Of the BRAF mutant patients, 33 received ICI and 38 did not with a median survival (17.1 vs 9.0 months) and 2-year survival (36% and 19%), respectively (p=0.014). When controlling for age, KPS, ECM, and number of lesions, BRAF mutant MBM patients treated with ICI compared to non-ICI had an OS hazard ratio, HR=0.4 (95% CI=0.21 – 0.78, p=0.0069).

CONCLUSION

ICI therapy in BRAF mutant MBM patients results in improved OS compared to those with non-ICI systemic therapy. No such difference was observed in the BRAF wild-type cohort.

CMET-23. IMPACT OF SYSTEMIC THERAPY IN MELANOMA BRAIN METASTASIS

BACKGROUND

Melanoma is the third most common malignancy that results in brain metastasis and is associated with a median overall survival (OS) of approximately 9 months. In recent years, management of melanoma brain metastases (MBM) by surgery and radiation [stereotactic radiosurgery (SRS) and whole brain radiation therapy (WBRT)] has been bolstered by targeted therapy and immune checkpoint inhibitors (ICI).

METHODS

351 patients, treated for MBM at our tertiary care center from 2000–2018, were grouped into: received chemotherapy, ICI, or targeted therapy. 34% of patients treated with ICI had received other systemic therapies as well as part of their management. OS was calculated from the date of diagnosis of the brain metastases. The Kaplan Meier analysis was utilized to determine median OS and difference in OS was determined by utilizing the Cox proportional hazard model.

RESULTS

The median survival after the diagnosis of brain metastasis was 10.4, 11.96, and 7.06 months in patients who received ICI, chemotherapy and targeted therapy respectively. A multivariate model was developed including the type of systemic therapy, presence of extracranial metastases, age, KPS and number of intracranial lesions. 114 patients underwent SRS alone, 56 underwent SRS and WBRT, 43 underwent SRS and surgical removal, 28 had surgical removal, SRS and WBRT, and 78 had no intracranial therapy. Compared to patients who received chemotherapy, patients who received immunotherapy had a hazard ratio, HR = 0.628 (confidence interval = 0.396 – 0.994, p-value = 0.047). Presence of EC metastases (HR= 1.25, p-value < .001), lower KPS (HR = .97, p-value < .0001) and multiple brain lesions (HR = 1.117, p-value < .0001) were associated with significantly worse OS.

CONCLUSIONS

Addition of ICI significantly improves the OS in MBM compared to chemotherapy. Lower performance status, multiple brain metastases, and EC metastases are associated with poor OS.

RTHP-20. MULTI-INSTITUTIONAL ANALYSIS OF HIGH-RISK WHO GRADE II OLIGODENDROGLIOMA TREATED WITH ADJUVANT RADIOTHERAPY PLUS TEMOZOLOMIDE (TMZ) VERSUS PROCARBAZINE, LOMUSTINE, AND VINCRISTINE (PCV)

PURPOSE

To compare outcomes of adults with high-risk grade II oligodendroglioma treated with adjuvant radiotherapy plus adjunctive TMZ versus PCV.

METHODS

We queried a multi-institutional database from the Cleveland Clinic and Mayo Clinic of 158 adults with WHO grade II glioma who were high-risk (age > 40 and/or subtotal-resection [STR]) and underwent adjuvant chemoradiation (TMZ or PCV) between 1997–2017. We included patients with oligodendroglioma by 1p/19q-codeletion or traditional histopathology. Fisher’s exact test was used to associate factors with chemotherapy group. The log-rank test was used to compare OS and PFS by chemotherapy group and clinical characteristics.

RESULTS

52 patients were included with a median follow-up of 44 months (range, 3–259). Median age was 44 (range, 22–73), 49 (94%) underwent STR, and 28 (54%) were male. Presenting symptoms were seizure in 35 (67%) patients, sensory in 14 (27%), and motor in 8 (15%). Oligodendroglioma classification was by 1p/19q-codeletion in 27 (52%) and traditional histopathology in 25 (48%) patients. Median radiation dose was 54 Gy (range, 45–60). Chemotherapy was TMZ in 34 (65%) patients and PCV in 18 (35%). Patients who were older (P=0.003), lacked seizures (P=0.03), or had motor symptoms (P=0.04) were more likely to receive TMZ. Median OS was 223 months (95% CI, 181-not estimable) and median PFS was 118 months (95% CI, 69–223). Treatment with TMZ versus PCV was not associated with OS (median 186 vs. 223 months, respectively; P=0.71) or PFS (median 110 vs. 131 months, respectively; P=0.19). Age >40 (P=0.009) and motor symptoms (P=0.027) were associated with adverse OS. Presence of motor symptoms was associated with worse PFS (P=0.008).

CONCLUSION

There was no statistically significant difference in OS or PFS between adjunctive TMZ versus PCV for adult high-risk grade II oligodendroglioma. A larger cohort with longer follow-up will provide additional insight.

Absolute lymphocyte count in patients with glioblastoma treated with temozolomide chemoradiation

e13564

Background: Standard glioblastoma (GBM) management includes radiotherapy, chemotherapy, and steroids; all of which can result in immunosuppression and a low absolute lymphocyte count (ALC). Previous literature identified an association between low CD4 and worse progression free survival (PFS) and overall survival (OS). There remains a lack of research addressing predictors of immunosuppression in patients with GBM. The primary objective of this study is to identify the degree of immunosuppression, measured by ALC, in GBM patients receiving concurrent temozolomide chemoradiation (CRT). Secondary objectives include associations between ALC, PFS, and OS, and whether there are any predictors of immunosuppression in patients with GBM. Methods: We retrospectively reviewed 231 newly diagnosed GBM patients who underwent surgery followed by standard of care CRT. We also analyzed the association between ALC and age, sex, MGMT methylation status, and extent of surgical resection. ALC was collected at the time of surgery, CRT start date, and two, four, six, and ten weeks post-CRT start date. Common Terminology Criteria for Adverse Events (CTCAE) protocol version 5.0 was then used to grade low ALC as grade 0, 1, 2, 3, or 4. Results: Of the 231 patients analyzed, 139 were males, 74 underwent gross total resection of the tumor, 129 patients were less than 65 years, and 79 (42.5%) were MGMT methylated. 37 patients had grade 3-4 low ALC. In a univariate analysis, grade 3-4 low ALC at 4 weeks (±14 days) post-CRT start was associated with higher mortality (HR 1.54, P = 0.028) but had no significant association with PFS (HR 1.22, P = 0.29). Logistic regression analysis was used to identify risk factors for grade 3-4 low ALC and its association with survival. None of the risk factors that we tested such as age, gender, type of surgery, or molecular markers including MGMT, IDH, or EGFR were associated with low ALC. Conclusions: Our study demonstrated that patients with ALC grade 3 or 4 at 4 weeks (±14 days) of CRT had a significantly higher mortality (HR 1.54, P = 0.028) but had no significant association with PFS (HR 1.22, P = 0.29).

Albumin levels and Prognostic Nutritional Index in glioblastoma

e13567

Background: Albumin levels are widely used to estimate patients’ nutritional status. Low perioperative albumin levels are associated with worse outcomes. Moreover, Prognostic Nutritional Index (PNI), which is calculated from serum albumin levels and peripheral blood lymphocyte count as follows: PNI = (Albumin x 10) + (0.005 x ALC), has been used to predict both short and long term outcomes in patients with wide variety of tumors. The primary objective of this study was to characterize perioperative albumin levels and PNI in newly diagnosed GBM patients. Secondary objectives included associations between albumin levels and PNI on progression free survival (PFS) and overall survival (OS). Methods: We retrospectively reviewed 568 newly diagnosed GBM patients who underwent surgery followed by standard of care chemoradiation. We analyzed the association between albumin and PNI on age, sex, MGMT methylation status, and extent of surgical resection on PFS and OS using a multivariate Cox proportional hazard model. Results: Of the 568 patients collected, 355 (62.5%) were males, 158 (27.8%) had gross total resection of the tumor, and 197(42.5%) were MGMT methylated. Both albumin and PNI were associated with OS but not PFS. The hazard ratio (HR) for OS among the top 2 quartiles of both albumin level and PNI were significantly higher than the bottom two quartiles. The median albumin level was 4.0 and the median PNI was 40. The point for significant high hazard ratio (HR) was around median value for both Albumin and PNI based on restricted cubic spine Cox regression models. The Kaplan-Meir (KM) estimated median OS was 15.2 months for albumin > 4, and 7.6 for albumin ≤4. The KM estimated median OS was 14.6 months for PNI > 40, and 5.7 for PNI≤40 (P logrank < 0.001 for both). While controlling for other factors that may also be associated with early death including age, gender, surgery type and MGMT status, HR = 1.9 (95% CI = 1.4- = 2.6) for Albumin < 4, and HR = 2.1 (95% CI = 1.5- 3.0) for PNI < 40 compared to their counterpart. Conclusions: Glioblastoma patients with perioperative albumin > 4 had a median OS of 15.2 months and 7.6 months for albumin ≤4, and a median OS of 14.6 months for PNI > 40 and 5.7 months for PNI≤40 (P logrank < 0.001 for both).

Impact of EGFR amplification status in newly diagnosed glioblastoma treated with Stupp protocol

e13569

Background: Standard post-surgical glioblastoma (GBM) treatment, per Stupp protocol, includes six-weeks of concurrent Temozolomide chemoradiation followed by at least six cycles of adjuvant-Temozolomide. Previous investigations into epidermal growth factor receptor (EGFR) amplification as a prognostic factor in GBM have yielded contradicting results, requiring further investigation. The primary aim of this study was to determine the degree to which EGFR amplification, in newly diagnosed GBM, impacted progression free survival (PFS) and overall survival (OS). Methods: Data from 582 patients who underwent surgical intervention for GBM at a tertiary care institution between 2012 and 2018 were analyzed. Only adult patients who underwent treatment per Stupp protocol and had pathological analysis on EGFR and CEP7 were included. Amplification and non-amplification status was calculated by a ratio of EGFR/CEP7 > 2 and < 2, respectively. PFS and OS outcomes were compared using Cox proportional hazard models stratified by surgery type and sex. Results: Of the original 582 patients, 122 were treated per Stupp protocol and had documented EGFR analysis. Of patients who were EGFR amplified, 41 (58.5%) were male and 25 (48.1%) were female (p = 0.38) and median amplification was 1.07 and 1.16 (p < 0.001), respectively. EGFR non-amplified patients had a PFS hazard ratio, HR = 0.70 (95% CI = 0.44 – 1.12, p = 0.14); and an OS HR = 0.60 (95% CI = 0.35 – 1.03, p = 0.065). When the EGFR/CEP7 ratio was stratified by quartile, it was found that Q4 compared to Q1 (Q4 > 6.50 vs 0 < Q1 ≤ 1.06) had a PFS HR = 2.1 (95% CI = 1.11 – 4.07, p = 0.024); and an OS HR = 2.48 (95% CI = 1.10 – 5.60, p = 0.028). Conclusions: There was no statistical difference in prevalence of EGFR amplification by sex. However, despite statistical significance, there was minimal difference in median degree of amplification by sex (0.09). Trends begin to show that patients who were EGFR non-amplified had better PFS and OS outcomes than patients who were EGFR amplified, although this was not statistically significant. Patients with very high EGFR amplification (Q4) had significantly poorer PFS and OS outcomes than patients with very low EGFR amplification (Q1).

Incidence and patterns of immune related adverse events (irAEs) during chemoimmunotherapy (ChemoIO) in patients with advanced nonsquamous NSCLC

e20705

Background: Based on superior outcomes noted in KEYNOTE-189 with the addition of pembrolizumab to platinum doublet chemotherapy, a majority of pts with nonsquamous NSCLC are now treated with chemoIO. irAEs can cause significant morbidity and occasionally mortality in pts treated with immunotherapy. Since chemotherapy is immunosuppressive, it is plausible that incidence of irAEs would be lower in pts receiving chemoIO compared to immunotherapy. The incidence of irAEs in KEYNOTE-189 was 22.7% vs 29.2% in KEYNOTE-024 with pembrolizumab monotherapy. We sought to evaluate the incidence and patterns of irAEs in pts with advanced nonsquamous NSCLC treated with chemoIO at our institution and to determine if the rate varied with the intensity of chemotherapy [carboplatin/ pemetrexed/ pembrolizumab (CPP) vs pemetrexed/ pembrolizumab (PP)]. Methods: We retrospectively reviewed the charts of 73 pts with advanced nonsquamous NSCLC who received CPP followed by maintenance PP. In addition to clinicopathologic variables, we noted the date, incidence, type and grade of irAEs. Descriptive statistics were used to report the incidence and patterns of irAEs. McNemar’s test was used to determine if there was a significant difference in rate of irAEs during CPP vs PP. Results: Of the 73 pts, 52% were females, 67% former smokers, 49% had previously received radiation and 11% had a history of autoimmune disease. 35 pts received 4 cycles of CPP. 21 patients (28%) developed irAEs, of which 6 had irAEs during CPP and 15 developed irAEs on PP. The rate of irAEs was significantly higher after stopping carboplatin (p = 0.049). 10 pts (13%) stopped pembrolizumab due to irAEs. Organ systems involved and grade of irAEs are listed below. Conclusions: The immunosuppressive effects of intensive chemotherapy may be protective against irAEs in pts receiving chemoIO. [Table: see text]

Real-world treatment patterns and outcomes in ER+/PR+/HER2+ metastatic breast cancer (MBC) patients: A National Cancer Database analysis

1033

Background: Treatment patterns and clinical outcomes are unclear for MBC patients diagnosed with estrogen receptor-positive (ER+) and/or progesterone receptor-positive (PR+) human epidermal growth factor 2 positive (HER2+) disease (i.e. triple positive). This study aimed to: 1) examine the utilization of first-line therapy among ER+/PR+/HER2+ MBC patients and 2) compare overall survival (OS) between the identified regimens. Methods: We analyzed triple positive MBC patients included in the National Cancer Database who were treated with endocrine therapy (ET) or chemotherapy (chemo) between 2010 and 2015. Kaplan-Meier method was used to estimate distributions of OS, which were compared among patient cohorts using the log-rank test. Results: A total of 6,234 patients were included in the analysis; of these, 3770 (60%) received ET and 2464 (40%) received chemo. Of those with complete survival data, there was no difference in median OS between patients treated with chemo vs. ET; however, those who received anti-HER2 therapy had significantly better OS than those who did not (median OS 49.4 vs. 41.0 months, p < 0.0001). Median OS stratified by ET or chemo with and without anti-HER2 further supported these findings, revealing the addition of anit-HER2 therapy to chemotherapy and ET resulted in superior median OS (Table). Conclusions: This is the first study to evaluate treatment utilization and OS among real-world triple positive MBC patients treated with ET or chemo. Our results suggest the majority of patients in the United States are treated with first-line ET; furthermore, the reported OS outcomes support the consideration of ET plus anti-HER2 therapy as a first-line treatment option for ER+/PR+/HER2+ MBC. Prospective trials evaluating de-escalation of systemic therapy in this subgroup of patients and future research to identify biomarkers to determine which patients can avoid chemotherapy are warranted. [Table: see text]

Low platelet counts in patients with glioblastoma

e13565

Background: Radiotherapy and concurrent chemotherapy with Temozolomide (TMZ) have myelosuppressive effect, and thrombocytopenia is commonly seen in this patient population seen in 5-10% of glioblastoma (GBM) patients. There is a lack of data analyzing the thrombocytopenia and it’s on the progression free survival (PFS) or overall survival (OS) of these patients. The primary objective of this study was to identify the degree of thrombocytopenia in newly diagnosed GBM patients receiving concurrent TMZ based chemoradiation (CRT). Secondary objectives included associations between thrombocytopenia PFS, and OS. Methods: We retrospectively reviewed 484 newly diagnosed GBM patients who underwent surgery followed by standard of care CRT. We also analyzed the association between platelet counts and age, sex, MGMT methylation status, and extent of surgical resection. Platelet count was collected at the time of surgery, CRT start date, and two, four, six, and ten weeks post-CRT start date. Patients were grouped into quartiles according to their platelets count. Results: Of the 484 patients collected, 308 were males, 139 had gross total resection of the tumor, 229 patients were older than 65 years, and 171 (42.1%) were MGMT methylated. In a univariate analysis, a platelet count less than 180,000 (lowest quartile) was associated with higher mortality (HR 1.63, P < 0.001) but had no significant association with PFS (HR 1.16, P = 0.48). Among the 118 patients who had platelet count lower than 180,000, 4 had platelets count less than 100,000 necessitating their TMZ to be stopped during CRT. In a multivariate analysis model adjusting for age, gender, MGMT status, and type of surgery, platelet counts less than 180,000 was also associated with significantly higher mortality (HR 1.60, P < 0.001). Conclusions: Our study concluded that patients who had platelet counts less than 180,000 at the time of surgery or CRT with TMZ had significantly higher mortality (HR 1.60, P < 0.001) but had no association with PFS (HR 1.16, P = 0.48).[Table: see text]

Impact of invasive regional lymph node examination (IRLNE) and time to radiation on overall survival for patients with stage I NSCLC treated with SBRT- a NCDB analysis

e20041

Background: SBRT is the standard of care for patients (pts) with early stage NSCLC who are not surgical candidates due to co-morbidities. Current NCCN guidelines state that IRLNE prior to definitive treatment for stage IA NSCLC is optional. Performing SBRT without IRLNE may increase the risk for occult nodal metastasis that could lead to regional recurrences and impact survival, however small retrospective studies have failed to demonstrate any benefit of IRLNE. Prior studies have also shown that delay in time to surgery impacts survival in stage 1 NSCLC pts, and due to co-morbidities SBRT pts are more likely to have a delayed time to radiation. However, the impact of time to radiation on OS in such pts is not well defined. Methods: We queried the NCDB for pts with stage 1 NSCLC treated with SBRT diagnosed from 2004 to 2014. We identified whether regional lymph nodes were examined for these pts. In addition, we obtained data regarding other clinicopathologic and treatment associated variables. Kaplan Meier method was used for survival analysis. Cox proportional hazards model was used to assess the impact of the co-variates on OS. Results: We identified 15,338 pts with stage 1 NSCLC treated with SBRT. Of these, 943 pts had IRLNE. Median age was 75 years (range 26-90) and Charlson-Deyo score was ≥ 1 in 6,427 pts. Median OS was 35.4 months (95% CI 34.6-36.2 months). The median time to radiation from diagnosis was 57 days. In a multivariate analysis, there was no impact of IRLNE on OS. Other clinicopathologic variables that impacted OS are listed in the table. A delay in initiation of radiation was associated with significantly worse OS. Conclusions: Invasive staging prior to SBRT for stage I NSCLC does not impact OS. A delay in initiating definitive radiation adversely affects survival in this population and the risk is incremental per week delay. [Table: see text]

Association of cell-free DNA (cfDNA) levels with myeloid-derived suppressor cells (MDSC) levels in blood of patients (pts) with muscle invasive (MI) and metastatic (met) bladder cancer (BC)

4528

Background: cfDNA can be detected in healthy individuals but higher concentrations are present in pts with cancer. MDSC are immature immunosuppressive cells that can be mobilized from bone marrow by tumor-related factors. Higher blood MDSC levels have been associated with worse outcomes in pts with solid tumors including BC. We assessed correlations between cfDNA and MDSC levels in pts with MIBC and met BC. Methods: Peripheral blood from pts with MIBC and met BC was collected in Streck BCT tubes and processed to obtain cf nucleic acid extracts. Total cfDNA was determined by fluorimetry. Cell-free DNA fragment size was measured by Bioanalyzer DNA analysis; 100-400 bp fragments (mono- and di-nucleosomal fragments linked to granulocytic processing of apoptotic and necrotic tumor cells) were designated low molecular weight (LMW-frags). The % of MDSC (CD33+/HLADR-) and subtypes were measured. MDSC subtypes were defined as polymorphonuclear (PMN-MDSC: CD15+/CD14-), monocytic (M-MDSC: CD15-/CD14+) and uncommitted (UNC-MDSC: CD15-/CD14-). Spearman’s correlation test was used for analysis. Results: Blood from 37 pts (19% women, 40% never smokers) with MIBC or met BC was collected: 15 (41%) with MIBC and 22 (59%) with met BC at time of collection. There was a significantly positive correlation between total MDSC and cfDNA levels (r = 0.57, P = 0.0003). Among MDSC subtypes, there was a significantly positive correlation between PMN-MDSC and cfDNA levels (r = 0.61, P < 0.0001). The higher level of LMW-frags was significantly but moderately associated with higher total MDSC (r = 0.43, P 0.008) and PMN-MDSC (r = 0.41, P 0.01) levels. There was no significant correlation between cfDNA level and other MDSC subtypes. Conclusions: There was a positive correlation between total and PMN-MDSC with cfDNA levels in blood from pts with MIBC and met BC. That may suggest a putative role for MDSC in mediating cfDNA release into the circulation, consistent with prior reports of granulocyte-mediated ctDNA processing. Further studies need to identify mechanisms and implications of our findings and potential correlation with clinical outcomes.

Patterns of systemic treatment utilization in ER+/PgR+/HER2+, early-stage breast cancer (BC): An analysis of the National Cancer Database

547

Background: The preferences and trends of treatment utilization of adjuvant endocrine therapy (ET) versus chemotherapy (CH) for small node-negative triple positive (TP) BC are unclear. We sought to determine these preferences and assess the impact on outcome. Methods: This is a retrospective study from the National Cancer Database including patients with TP stage I BC, 2004-2015. Treatment selection was evaluated for association with patient clinical and demographic characteristics using logistic regression. Overall survival (OS) was estimated using the Kaplan-Meier method and compared among patient and treatment cohorts by log-rank test and Cox regression. Results: Of 37,777 patients analyzed, 79% were White (Non-Hispanics), 10% African Americans, and 5% Hispanic/Latinos. 57% were 50-70 years old. 86% received adjuvant endocrine therapy versus 14% CH first. Around 40 % of all patients received anti-Her2 therapy. Patients younger than 70 years, with male BC, diagnosed with poorly differentiated BC, African Americans and Hispanics were more likely to be treated with chemotherapy. OS rate at 5-year was 92.3% (95% CI: 0.918-0.928). In multivariate analysis for patients with survival data, an increased rate of death was associated with: treatment in community versus academic/research centers, CH first versus ET, no treatment with anti-Her2 therapy, government versus private /no insurance, Native American ethnicity. A slight but statistically significant reduction in the in the risk of death at 5 years was evident for patients receiving anti-Her2 therapy plus ET therapy, 5-year OS 93.5% (CI: 89.2-98%), when compared to patients receiving anti-Her2 therapy plus CH 92.7 % (CI: 89.4-96). Conclusions: This study provides real world data of common practices in the US . The majority of patients with node negative Stage I, ER+/PR+/Her2+ BC received adjuvant ET and anti-Her2 therapy, not chemotherapy. These patients had a similar to slightly improved 5 year- survival when compared to anti-Her2 therapy plus CH, supporting the use ET plus anti-Her2 therapy in this setting. Future studies should focus on better selecting patients with hormone receptor positive and Her 2 + early stage BC who would benefit from adjuvant CH. Disparity in outcome also warrants further evaluation. [Table: see text]

Disparities in treatment patterns and overall survival (OS) in hormone receptor-positive HER2+ metastatic breast cancer (MBC): A National Cancer Database Analysis

1032

Background: Determinants of variation in therapy utilization and OS are unclear for patients diagnosed with hormone receptor+, human epidermal growth factor 2 positive (HER2+) MBC. This study aimed to identify if there are disparities in first-line treatment patterns and outcomes for this subpopulation of MBC patients. Methods: We analyzed MBC patients included in the National Cancer Database diagnosed with estrogen receptor-positive (ER+) and/or progesterone receptor-positive (PR+) and HER2+ disease (i.e. triple positive) treated with endocrine therapy or chemotherapy between 2010 and 2015. Analyses describe the distribution of treatments administered in the first-line setting. Kaplan-Meier method was used to estimate distributions of OS, which were compared among patient cohorts using the log-rank test. Results: Of the 6215 patients diagnosed with triple positive MBC, the majority were 50-70 years old (n=3414 [55%]), female (n= 6122 [98%]), and white (n=4478 [72%]). Four distinct treatment patters were identified; hormonal therapy was the most common (n= 2289 [37%]), followed by hormonal therapy + anit-HER2 (n=1471 [24%]), chemotherapy (1280 [20%]), and chemotherapy + anit-HER2 (n=1175 (19%)). Significant differences in demographic, socioeconomic, and disease characteristics were identified across groups. Disparities in OS were also observed; the unadjusted 5-year OS was substantially lower among older patients, African Americans (AA), those with government insurance, and lower income (Table). Conclusions: This is the first study to report disparities in treatment patterns and OS among real-world triple positive MBC patients. Further investigation is required to determine if there are independent causal associations between poor prognosis and the identified demographic and socioeconomic characteristics. [Table: see text]

Circulating cell-free DNA (cfDNA) levels and fragmentation pattern can distinguish nonmuscle invasive (NMI) from muscle-invasive (MI) and metastatic (met) bladder cancer (BC)

4523

Background: Occult MI and met BC may be under-staged. Circulating cfDNA may be a dynamic, low-cost and minimally invasive biomarker. We evaluated correlations between total circulating cfDNA and presence of MIBC and met BC. We hypothesized that the relative abundance of circulating low molecular weight cfDNA would correlate with BC stage. Methods: Peripheral blood from pts with BC was collected in Streck BCT tubes and processed to obtain cf nucleic acid extracts. Total cfDNA quantity (ng/ml) was assessed by fluorimetry. cfDNA fragment size was measured by Bioanalyzer DNA analysis. Wilcoxon rank sum test and Fisher’s Exact test were used to compare cfDNA quantity and fragmentation pattern among pts with NMIBC, MIBC, met BC. Results: Blood was obtained from 58 pts with BC (20% women, 34% never smokers, median age 71 (29-89). There was no significant difference in cfDNA between MIBC and met BC, however, it was significantly lower in pts with NMIBC vs MIBC and met BC (table). The concentration of low molecular weight fragments (LMW-frags) (100 - 400) base pairs and the ratio of LMW-Frag to cfDNA were significantly different between pts with NMIBC and pts with MIBC or met BC (table). Using median values as the cutoff, there was a significantly higher proportion of pts with cfDNA > 7 ng/ml and LMW-frags > 1.6 ng/mL, in MIBC & met BC vs NMIBC (p < 0.001). The % of pts with LMW-frags to cfDNA > 30%, was significantly different among NMIBC, MIBC and met BC groups: 16%, 53%, 78%, respectively (p < 0.001). Conclusions: This exploratory study suggests that cfDNA levels may correlate with BC stage. Measuring the relative abundance of LMW-frags with the expected size of cf DNA can enhance the specificity of cfDNA analysis for distinction between MIBC and met BC. Further studies are needed to confirm findings and define the optimal cut-points for optimal BC staging. [Table: see text]

Blood myeloid derived suppressor cells (MDSC) in metastatic urothelial carcinoma (mUC) are correlated with neutrophil-to-lymphocyte ratio (NLR) and overall survival (OS)

436

Background: MDSC have been linked to the chronic inflammatory microenvironment of tumor cells and pathologic outcomes in UC patients (pts) undergoing cystectomy. NLR is an established inflammatory biomarker with prognostic properties in mUC. We hypothesized that MDSCs correlate with NLR and OS in mUC. Methods: MDSCs were measured in blood samples from mUC patients by fresh unfractionated whole blood (WB) and peripheral blood mononuclear cells (PBMC). MDSCs were identified by flow cytometry in WB and defined as LinloCD33+/HLADR- (Total MDSC). MDSC subsets were defined as polymorphonuclear (PMN-MDSC: CD15+/CD14-), monocytic (M-MDSC: CD15-/CD14+), and uncommitted (UC-MDSC: CD15-/CD14-). MDSC populations were presented as % of live nucleated blood cells from PB and absolute numbers from WB. Spearman’s correlation assessed correlations between MDSC & NLR. Kaplan Meier curves and log rank test estimated OS from the time of MDSC collection to last follow up or date of death. Results: Of 79 pts, 77% were men and 42% were never smokers with a median age of 69 (31-83). Overall, 71% had pure UC and 81% had lower tract UC. Prior therapies include intravesical therapy (22%), neoadjuvant chemotherapy (31%), and cystectomy/nephroureterectomy (61%). Median follow up was 12 months (range: 0.6-36.5). PMN-MDSC was the predominant subset in WB and PBMC. There was significant correlation between individual MDSC subsets in WB and PBMC (p≤0.001). Negative correlation was noted between NLR and WB UC-MDSC:PMN-MDSC ratios (rho = -0.27, p = 0.03), as well as NLR and PB UC-MDSC:PMN-MDSC (rho = -0.28, p = 0.02). Median survival was 17.7 months (95% CI: 11.0-NA months). Overall 1-yr and 3-yr survival were 0.60 (95% CI: 0.49-0.73) and 0.15 (95% CI: 0.03-0.67), respectively. Higher WB UC-MDSC levels were associated with shorter OS (HR 2.85, 95% CI: 1.43-5.65, p = 0.003). Conclusions: Specific MDSC subsets correlate with NLR. Higher WB UC-MDSC levels have negative prognostic roles for OS. Given the feasibility of serial blood draws, dynamic assessment of MDSC over time and further validation with longer follow up are needed.

Neoadjuvant durvalumab +/- tremelimumab affects the expression of immune checkpoint (IC) molecules on myeloid derived suppressor cells (MDSC) in patients (pts) with locally advanced renal cell carcinoma (RCC)

587

Background: In a single arm, open label phase 1b clinical trial the safety of neoadjuvant durvalumab +/- tremelimumab was studied in pts with (w) locally advanced RCC. Expression of IC molecules on immunomodulatory cells in peripheral blood (PB) and tumor (T) and the association w treatment (tx) was investigated. Methods: Pts with ≥ T2bN0-1M0 RCC received either durvalumab or combination durvalumab + tremelimumab prior to surgery. Blood samples were drawn prior to neoadjuvant tx, prior to surgery, and approximately 30 days after surgery before adjuvant tx. The percentage of MDSC (CD33+/HLADR-) and subtypes in PB and T and expression of PD1, PD-L1, and V-domain Ig suppressor of T cell activation (VISTA) were measured. MDSC subtypes included polymorphonuclear (PMN; CD15+/CD14-), monocytic (M; CD15-/CD14+) and uncommitted (UC; CD15-/CD14-). Linear mixed model was used for each MDSC subtype to estimate and compare cohorts over time. Results: Eighteen pts were enrolled: 4 women and 14 men, median age 62, 17 pts had T3-4 and 4 pts had N1 disease. Six pts received 1 dose of durvalumab and 12 pts received 1 dose of durvalumab + tremelimumab before surgery. Tx-related grade 3 adverse events (per CTCAE, v5.0) included thrombocytopenia, bilateral lower extremity weakness, hyperglycemia, chest pain, and diabetic ketoacidosis.One pt had grade 4 elevated lipase. One pt had sudden death from a non-drug related cardiac event 9 days after receiving combination therapy prior to surgery. PB and T samples from 17 pts were available. Expression of VISTA on M-MDSC and UC-MDSC were positively correlated in PB and T (Spearman’s rho = 0.61; P=0.03 for both). VISTA expression on UC-MDSC in PB was significantly higher in pts who received durvalumab monotherapy compared to those treated w durvalumab + tremelimumab (P=0.04). Frequencies of PD-L1 expression on M-MDSC and UC-MDSC in PB decreased significantly from pre- to post-neoadjuvant tx (P < 0.01). Conclusions: Neoadjuvant durvalumab + tremelimumab in pts w locally advanced RCC is feasible and affects the expression of IC molecules (PD-L1 and VISTA) on M-MDSC and UC-MDSC. Clinical trial information: NCT02762006.

Correlation of myeloid-derived suppressor cells (MDSC) with pathologic complete response (pCR), recurrence free survival (RFS), and overall survival (OS) in patients with urothelial carcinoma (UC) undergoing cystectomy

437

Background: MDSCs play an important role in maintaining a tumor immunosuppressive microenvironment. The association of circulating levels of MDSCs with pCR (pT0N0) and outcomes was investigated in patients (pts) with non-metastatic UC undergoing cystectomy. Methods: Peripheral blood samples from pts with non-metastatic UC was collected. MDSCs were measured in freshly purified peripheral blood mononuclear cells, using flow cytometry. Total (T) MDSC was defined as CD33+/HLADR-. T-MDSC subtypes were polymorphonuclear (PMN-MDSC: CD15+/CD14-), monocytic (M-MDSC: CD15-/CD14+], and uncommitted (UC-MDSC: CD15-/CD14-]. MDSC populations were presented as % of live nucleated blood cells. Wilcoxon rank sum test was used to compare MDSCs between pCR groups. Kaplan-Meier and log-rank test were used to analyze RFS and OS. Results: MDSC data were available for 124 pts (106 male, 18 female), median age 68, 28 (23%) never smokers, 93 (75%) pure UC. Thirty four pts (27%) received intravesical BCG; 49 (39%) received neoadjuvant chemotherapy (NAC); 22 (19%) had pCR (pT0N0) following surgery. PMN-MDSC was the dominant subtype (42%) and frequency of UC-MDSC and M-MDSC was 40% and 17%, respectively. Circulating levels of T-MDSC and PMN-MDSC were significantly lower in pCR patients than those in non-pCR patients (Table). Sixteen deaths were observed and 21 pts recurred after surgery. The median follow-up time of patients alive was 18.7 months (range 0.3-42.4). The median OS or RFS of all patients was not reached. One-year and two-year OS rates were 94% and 83%, respectively. One-year and two-year RFS rates were 82% and 69%, respectively. There was no association between MDSC subtypes with OS or RFS. Conclusions: Total- and PMN-MDSC subtypes in blood were significantly correlated with pCR in pts with non-metastatic UC who undergo cystectomy. The relatively short follow-up may impact the association with RFS and OS; additional follow-up is needed. [Table: see text]