Differential impact of tumor suppressor gene (TP53, PTEN, RB1) alterations and treatment outcomes in metastatic, hormone-sensitive prostate cancer

BACKGROUND

Altered tumor suppressor genes (TSG-alt) in prostate cancer are associated with worse outcomes. The prognostic value of TSG-alt in metastatic, hormone-sensitive prostate cancer (M1-HSPC) is unknown. We evaluated the effects of TSG-alt on outcomes in M1-HSPC and their prognostic impact by first-line treatment.

METHODS

We retrospectively identified patients with M1-HSPC at our institution treated with first-line androgen deprivation therapy plus docetaxel (ADT + D) or abiraterone acetate (ADT + A). TSG-alt was defined as any alteration in one or more TSG. The main outcomes were Kaplan-Meier-estimated progression-free survival (PFS) and overall survival, analyzed with the log-rank test. Clinical characteristics were compared with the χ² test and Kruskal-Wallis rank sum test. Cox regression was used for univariate and multivariable analyses.

RESULTS

We identified 97 patients with M1-HSPC: 48 (49%) with ADT + A and 49 (51%) with ADT + D. Of 96 patients with data available, 33 (34%) had 1 TSG-alt, 16 (17%) had 2 TSG-alt, and 2 (2%) had 3 TSG-alt. The most common alterations were in TP53 (36%) and PTEN (31%); 6% had RB1 alterations. Median PFS was 13.1 (95% CI, 10.3-26.0) months for patients with normal TSGs (TSG-normal) vs. 7.8 (95% CI, 5.8-10.5) months for TSG-alt (P = 0.005). Median PFS was lower for patients with TSG-alt vs TSG-normal for those with ADT + A (TSG-alt: 8.0 [95% CI, 5.8-13.8] months vs. TSG-normal: 23.2 [95% CI, 13.1-not estimated] months), but not with ADT + D (TSG-alt: 7.8 [95% CI, 5.7-12.9] months vs. TSG-normal: 9.5 [95% CI, 4.8-24.7] months). On multivariable analysis, only TSG-alt predicted worse PFS (hazard ratio, 2.37; 95% CI, 1.42-3.96; P < 0.001).

CONCLUSIONS

The presence of TSG-alt outperforms clinical criteria for predicting early progression during first-line treatment of M1-HSPC. ADT + A was less effective in patients with than without TSG-alt. Confirmation of these findings may establish the need for inclusion of molecular stratification in treatment algorithms.

Comparison of care process for newly diagnosed breast cancer in insured versus uninsured populations: Opportunities for improving health equity

e18518

Background: Initiatives enhancing equitable oncologic care are an increasingly emphasized priority. Our study aims to identify aspects of breast cancer (BC) care in which differences exist based on insurance coverage status. Methods: We performed a retrospective, case control study consisting of 39 Hispanic ethnicity uninsured patients (UP) with newly diagnosed BC at federally qualified health centers and 119 insured patients (IP) diagnosed at Mayo Clinic Arizona (MCA). Patients were matched 3:1 for age, stage, year of diagnosis, ER and HER2 status. Demographic information, clinical variables, and zip code level specific socioeconomic information were compared. Continuous variables were compared by Wilcoxon rank-sum test and categorical variables by chi-square test. All patients ultimately received their cancer treatment at MCA. Results: Similar treatment patterns with chemotherapy, surgery, and radiation treatment were observed between groups. Primary language was Spanish for 94% of UP and English for 97.5% of IP. The majority of UP were of Hispanic ethnicity (97.4%); IP were 83.2% non-Hispanic White, 9.2% Hispanic, 3.4% African American. Zip code level information reflected more unemployment with a median of 10.6% versus 6.9% p ˂ 0.001, percent of high school or lower (53.0 % v 23.2 %, p ˂ 0.001), and lower income for UP (33733.5 v 64728.0 p values ˂ 0.001). UP BMI was significantly higher (30.6 V 24.7, p=0.005), with presence of more co-morbidities; diabetes (28.2% v 5.0%, p ˂ 0.001), hypertension (35.9 % v 20.2%, p= 0.046), dyslipidemia (28.2% v 12.6%, p = 0.023), metabolic syndrome (p 23.7% v 8.5, p= 0.013), and tobacco use (17.9% v 2.5%, p ˂ 0.001). IP had higher alcohol use (52.9% v 5.3%, p ˂ 0.001). Genetics consultation was performed for 62.2% IP versus 35.9% UP (p=0.004), lower acceptance of oncology nutrition consultation for UP (29.4% vs 7.4%, p= 0.024) Median time from abnormal mammogram to biopsy (25.5 days vs. 14 days, p=0.056), and interval from diagnosis to treatment (62 days vs. 39 days) (p=0.001) were less favorable for UP compared to IP. Conclusions: In comparing the status of UP (primarily Hispanic, Spanish-speaking) and IP (primarily non-Hispanic White, English-speaking) with newly diagnosed BC we identified greater prevalence of co-morbidities and adverse social determinants of health in the former group. We identified access to genetic counseling services, access to oncology nutrition consultation, and timeliness of diagnostic biopsy and initiation of treatment as disparate features in the care pathway. These observations can allow development of tailored interventions to achieve greater equity in delivery of BC care.

Tumor Frameshift Mutation Proportion Predicts Response to Immunotherapy in Mismatch Repair-Deficient Prostate Cancer

Background

Genomic biomarkers that predict response to anti‐PD1 therapy in prostate cancer are needed. Frameshift mutations are predicted to generate more neoantigens than missense mutations; therefore, we hypothesized that the number or proportion of tumor frameshift mutations would correlate with response to anti‐PD1 therapy in prostate cancer.

Methods

To enrich for response to anti‐PD1 therapy, we assembled a multicenter cohort of 65 men with mismatch repair‐deficient (dMMR) prostate cancer. Patient characteristics and outcomes were determined by retrospective chart review. Clinical somatic DNA sequencing was used to determine tumor mutational burden (TMB), frameshift mutation burden, and frameshift mutation proportion (FSP), which were correlated to outcomes on anti‐PD1 treatment. We subsequently used data from a clinical trial of pembrolizumab in patients with nonprostatic dMMR cancers of various histologies as a biomarker validation cohort.

Results

Nineteen of 65 patients with dMMR metastatic castration‐resistant prostate cancer were treated with anti‐PD1 therapy. The PSA₅₀ response rate was 65%, and the median progression‐free survival (PFS) was 24 (95% confidence interval 16–54) weeks. Tumor FSP, more than overall TMB, correlated most strongly with prolonged PFS and overall survival (OS) on anti‐PD1 treatment and with density of CD8+ tumor‐infiltrating lymphocytes. High FSP similarly identified patients with longer PFS as well as OS on anti‐PD1 therapy in a validation cohort.

Conclusion

Tumor FSP correlated with prolonged efficacy of anti‐PD1 treatment among patients with dMMR cancers and may represent a new biomarker of immune checkpoint inhibitor sensitivity.

Implications for Practice

Given the modest efficacy of immune checkpoint inhibition (ICI) in unselected patients with advanced prostate cancer, biomarkers of ICI sensitivity are needed. To facilitate biomarker discovery, a cohort of patients with DNA mismatch repair‐deficient (dMMR) prostate cancer was assembled, as these patients are enriched for responses to ICI. A high response rate to anti‐PD1 therapy in these patients was observed; however, these responses were not durable in most patients. Notably, tumor frameshift mutation proportion (FSP) was identified as a novel biomarker that was associated with prolonged response to anti‐PD1 therapy in this cohort. This finding was validated in a separate cohort of patients with nonprostatic dMMR cancers of various primary histologies. This works suggests that FSP predicts response to anti‐PD1 therapy in dMMR cancers, which should be validated prospectively in larger independent cohorts.

Biomarkers of immune checkpoint inhibition sensitivity are needed. This article reports on genomic biomarkers that may predict response to anti‐PD1 therapy in prostate cancer.

The effect that β-lactam antibiotics have on progression free and overall survival in multiple myeloma patients undergoing autologous stem cell transplantation

e20518

Background: Gut microbiome dysbiosis is correlated with graft-versus-host disease (GVHD) in allogeneic stem cell transplant (allo-SCT) patients. In the allo-SCT population, antibiotics have been associated with increased risk for GVHD mortality and relapse due to loss of gut obligate anaerobes . It has been shown that antibiotics may negatively impact the efficacy of checkpoint inhibitors for patients with solid tumors. Based on these studies, we performed a retrospective analysis to determine if antibiotic treatment affects outcomes of multiple myeloma (MM) patients after autologous SCT (ASCT). Methods: This is a single institution retrospective study at Hackensack University Medical Center. A list of consecutive MM patients treated from 1/2012 to 12/2015 was obtained and an electronic medical record review of the first 217 who received ASCT was performed. Baseline characteristics, treatment history, transplant course and antibiotic treatment (including β-lactams, fluoroquinolones, macrolides, metronidazole, and vancomycin) were reviewed. Prophylactic antibiotics were excluded. Response was defined using the IMWG criteria. Median progression free survival (PFS) and overall survival (OS) were estimated using the Kaplan-Meier method. Log rank tests were used to compare the difference in survival between stratified groups. The LASSO cox regression analysis method was used for multivariate analyses of PFS and OS. Results: Of the 217 patients, 205 patients were available for analysis. Median age at ASCT was 61. β-lactams were associated with decreased median PFS (1.95 vs 4.77 years (yrs), p < 0.01) and decreased median OS (7.51 vs 13.45 yrs, p = 0.01). Multivariate analysis adjusting for lasso-selected age, gender, complete remission (CR) after ASCT, and ISS demonstrated independent progression risk associated with β-lactam use (HR = 2.00, 95% CI, 1.28–3.12, p < 0.01). β-lactams were associated with worse OS in multivariate analysis adjusting for lasso-selected age, gender, CR after ASCT and high risk cytogenetics (HR = 1.89, 95% CI, 1.07–3.40, p = 0.03). Conclusions: In this preliminary study, β-lactams predicted for decreased PFS and OS compared to patients who did not receive β-lactams in MM patients undergoing ASCT. The study was limited by its retrospective nature but demonstrates one of the first evaluations of antibiotic effect on the ASCT population in MM. Prospective studies evaluating the impact of antimicrobials on patient outcomes and the gut microbiome are ongoing.

Patterns of Twitter use among trainees in hematology-oncology related areas

11041

Background: Twitter (TW) is an essential tool in the medical community. Few studies have examined the use of TW by medical trainees. We aimed to assess its utilization among trainees in hematology-oncology related areas (HORAs). Methods: 576 training programs in HORAs were obtained. We contacted a potential pool of 3,142 trainees. A 50 item survey was distributed between 2/2019-5/2019, focusing on: demographics, professional use, attitudes toward TW, and patient interactions. Responses were analyzed using parametric descriptive statistics. Results: 442 responses (14% response rate) were received; 203 (46%) used a TW account for professional activities. See table for demographics. The most common reasons for using TW were: continuous education (73%), dissemination of information (62%), and networking (62%). > 80% of users have positive views on using TW for: promoting academic discussions (92%), trainees education (89%), and conference networking (83%). 50 (25%) have used TW for collaborations, with abstracts and papers being the most common (20%). Most agreed or strongly agreed that TW: is a useful education tool (89%), and might be useful for career advancement (85%). The most common challenges were: the value of TW content can be decreased by irrelevant content and biased sources (64%), not peer-reviewed (51%), and difficulty searching information (50%). 68% considered that programs should promote the engagement of trainees, but only 21% reported instruction in the use of TW. 83% agree or strongly agree that TW is useful for physicians and patient interactions, but 82% have concerns about legal repercussions. Conclusions: Almost half of trainees in HORAs use TW for professional activities. The most common uses are for education, dissemination of information and networking. Most trainees that use TW have positive views regarding education and academic collaborations. Use of Twitter may strengthen trainees’ education, provide mentorship opportunities, and promote career advancement. Challenges on TW use should continue to be addressed. [Table: see text]

Prognostic value of neutrophil-lymphocyte ratio and platelet-lymphocyte ratio in patients with advanced solid tumors treated with PD-1 inhibitors

e14132

Background: The use of PD-1 inhibitors (PD1) has been limited due to the lack of prognostic markers of response. Previous studies have suggested that indirect inflammatory markers such as neutrophil lymphocyte ratio (NLR) and platelet lymphocyte ratio (PLR) may correlate with response to treatment and improved survival. We sought to evaluate the impact NLR and PLR in a real world cohort of patients (pts) with advanced solid tumors treated with PD1. Methods: Records of all pts treated with PD1 between 2011-2017 at the John Theurer Cancer Center were reviewed. NLR and PLR were registered at baseline and longitudinally. We dichotomized as low (lNLR/lPLR), and high (hNLR ≥5/hPLR ≥160). Univariate logistic regression and Cox proportional hazards were used for progression free survival (PFS), and overall survival (OS). Landmark analyses were performed at cycles 2 (C2), and 5 (C5). Results: 178 pts received 1131 cycles of PD1. Median age was 65 (range 24-93); 43% were female. ECOG was ≥1 for 85%. 142 pts (80%), and 89 (50%) received ≥2C and ≥5C respectively. In univariate analyses, baseline hNLR and hPLR were independently associated with inferior OS (median 7.8 vs 18.3 months; HR 1.77, 95% CI 1.21-2.43; p = 0.004), ) and (median 6.4 vs 15.6 months; HR 1.42, 95% CI 1.09-2; p = 0.04) respectively. On landmark analyses, lNLR and lPLR were associated with longer PFS and OS at 2C, and 5C. Longitudinally, lNLR and lPLR correlated with response rate; NLR decreased by 0.08 (95% CI: -0.19 to -0.04; p = 0.03) and PLR by 17 (95% CI: -29 to -14; p = 0.07) per month in responders compared with non-responders. hNLR or hPLR did not correlate with increase in autoimmune toxicities. Conclusions: hNLR, hPLR are adversely prognostic markers in pts receiving PD1 inhibitors in a “real world cohort”. These markers correlate with a longitudinal response and may help predict response. Further prospective studies are needed to determine their utility in decision making during treatment.

Characterization of Comorbidities Limiting the Recruitment of Patients in Early Phase Clinical Trials

BACKGROUND

Early phase clinical trials evaluate the safety and efficacy of new treatments. The exclusion/inclusion criteria in these trials are usually rigorous and may exclude many patients seen in clinical practice. Our objective was to study the comorbidities limiting the participation of patients with breast, colorectal, or lung cancer in clinical trials.

MATERIALS AND METHODS

We queried ClinicalTrials.gov on December 31, 2016. We reviewed the eligibility criteria of 1,103 trials. Logistic regression analyses were completed, and exclusion was studied as a binary variable.

RESULTS

Out of 1,103 trials, 70 trials (6%) excluded patients >75 years of age, and 45% made no reference to age. Eighty-six percent of trials placed restrictions on patients with history of prior malignancies. Regarding central nervous system (CNS) metastasis, 416 trials (38%) excluded all patients with CNS metastasis, and 373 (34%) only allowed asymptomatic CNS metastasis. Regarding chronic viral infections, 347 trials (31%) excluded all patients with human immunodeficiency virus, and 228 trials (21%) excluded all patients with hepatitis B or C infection. On univariate analysis, chemotherapy trials were more likely to exclude patients with CNS metastasis and history of other malignancies than targeted therapy trials. Multivariate analysis demonstrated that industry-sponsored trials had higher odds of excluding patients with compromised liver function.

CONCLUSION

Many clinical trials excluded large segments of the population of patients with cancer. Frequent exclusion criteria included patients with CNS metastasis, history of prior malignancies, and chronic viral infections. The criteria for participation in some clinical trials may be overly restrictive and limit enrollment.

IMPLICATIONS FOR PRACTICE

The results of this study revealed that most early phase clinic trials contain strict exclusion criteria, potentially excluding the patients who may be more likely to represent the population treated in clinical settings, leaving patients susceptible to unintended harm from inappropriate generalization of trial results. Careful liberalization of the inclusion/exclusion criteria in clinical trials will allow investigators to understand the benefits and drawbacks of the experimental drug for a broader population, and possibly improve recruitment of patients with cancer into clinical trials.

Abstract A27: Diversity in multiple myeloma clinical trials

Background: Multiple myeloma (MM) accounts for approximately 1% of all cancers and 10% of hematologic malignancies in the United States (U.S.). MM occurs in all races, but the incidence in African Americans is two to three times higher than in non-Hispanic whites. MM is also slightly more frequent in men than women (1.4:1). Many clinical trials lack appropriate representation of specific patient populations, limiting the generalizability of the evidence obtained. Therefore, we determined the representation of ethnic minorities, the elderly, and women in MM clinical trials.

Methods: Enrollment data from all therapeutic trials reported as completed in clinicaltrial.gov from 2000 to 2016 were analyzed. Clinical trials including other hematologic malignancies and with recruitment outside of the U.S. were excluded. Enrollment fraction (EF) was defined as the number of enrollees divided by the 2013 SEER database MM complete prevalence. Chi-square test was used to estimate differences in categorical data.

Results: Out of 177 MM clinical trials (CT), 78 (44%) reported ethnicity with a total of 12,055 enrollees. Regarding enrollees' ethnic composition, 84% were non-Hispanic White (NHW), 8.6% African American (AA), 2.8% Asian, 1.8% Hispanic, and 0.1% Native American/Alaskan Indian. Out of those 78 CT, 52 (66%) were phase II, 15 (19%) phase III, and 11 (14%) phase I. Most of the results were published from 2012 to 2016 (74%). Forty-six (59%) trials were sponsored by industry, 7 (9%) by NCI, and 25 (32%) were investigator initiated. Participation in CT varied significantly across ethnic groups, NHW were more likely to be enrolled in CT (EF of 0.23) than AA (EF of 0.08, p &lt; 0.0001) and Hispanics (EF of 0.05, p&lt; 0.0001). Males had a higher recruitment rate than females (58% vs. 42%), but this could be explained by the higher incidence of MM in males. Enrollees' median age was 62 years. Younger patients (&lt; 65 years) were more likely to be enrolled in CT than the elderly (66% vs. 34%, p&lt;0.0001). Industry-sponsored trials were less likely to recruit AA compared with investigator-initiated trials (7.6% vs. 12%, p&lt;0.01).

Conclusions: Despite the higher incidence of MM in African Americans and the elderly, the former only represented 8.6% of the study participants and 66% of these were less than 65 years of age; therefore, we are lacking data in the tolerability of these new agents in our aging MM population. We also observed that industry studies were less likely to recruit AA patients. Collaborations between investigators, sponsors, and the community are necessary to increase our minority and elderly patients' access to clinical trials.

Citation Format: Narjust Duma, Miguel Gonzalez Velez, Jesus Vera-Aguilera, Richardo Parrondo, Veronica Mariotti, Jonas Paludo, Yucai Wang, Ronald Go, Alex Adjei. Diversity in multiple myeloma clinical trials [abstract]. In: Proceedings of the Tenth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2017 Sep 25-28; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2018;27(7 Suppl):Abstract nr A27.

Comprehensive analysis of tumor mutational load, genomic alterations, and PD-L1 status in gastrointestinal cancers using a multiplatform molecular profiling tool

93

Background: Tumor mutational load (TML) has been proposed as a biomarker of responsiveness to novel immune checkpoint inhibitors (ICIs) due its association with increased neoantigen formation and tumor immunogenicity. Gastrointestinal cancers (GICs) are generally insensitive to ICIs; therefore the aim of this study was to evaluate the TML, clinically relevant genomic alterations (CRGAs), and PDL-1 status of GIC patients (pts). Methods: We analyzed samples of pts with GICs using a multiplatform profiling tool (Caris Life Sciences, Phoenix, AZ). TML and CRGAs were calculated based on next generation sequencing and mutational load was stratified as high (≥17 mut/Mb), intermediate (8-16 mut/Mb), and low (≤ 7 mut/Mb). PD-L1 status was determined by IHC. Descriptive statistics and simple linear regression were used for analysis. Results: A total of 85 pts with GICs were analyzed. The median pt age was 65. 53% were males and 47% were females. Tumors were colorectal 51%, pancreatic 13%, esophageal 13%, gastric 9%, hepatobiliary 11%, anal canal 4%, and small intestine 1%. 80 cases (94%) had an adequate sample for profiling and TML was available in 59 cases (74%). Cases were 5% TML-high, 56% TML-intermediate, and 39% TML-low. The median TML was 9 mut/Mb and median CRGAs was 3. The most commonly detected CRGAs were TP53 (56%), APC (44%), KRAS (40%), ATM (18%) and PTEN (6%). A positive but very weak correlation was found between TML and number of CRGAs (R2 = 0.3). PD-L1 status was available in 74 cases (93%) with 86% being PD-L1 negative and 14% PD-L1 positive. PD-L1 expression was most commonly seen in esophageal (50%) and anal canal (30%) cancers. Among 57 tumors tested for both biomarkers, no correlation was found between a high TML and PD-L1 expression. Conclusions: TML varies among GICs, and higher TML scores were not associated with PD-L1 expression, but there was a weak correlation between TML and the number of CRGAs. Further analysis by stratification of GIC type and prognostic analysis to assess the correlation between TMB and response to ICIs is warranted.

Characterization of genomic alterations and biomarker expression patterns of gastrointestinal cancers using a multiplatform molecular profiling tool

594

Background: Given the heterogeneity of gastrointestinal cancers (GICs), molecular profiling is becoming part of the standard of care for many solid tumors. The aim of this study was to evaluate the molecular profile of patients (pts) with GICs using a multiplatform profiling tool and to assess how the pattern of detected molecular alterations could guide clinical decision-making in these tumors. Methods: We retrospectively analyzed samples of 85 pts with GICs via a multiplatform profiling service (Caris Life Sciences, Phoenix, AZ) in order to evaluate clinically relevant genomic alterations (CRGAs) and clinically relevant biomarker expression (CRBs) in GICs. Results: A total of 85 pts with GICs were analyzed. 45 (53 %) were males and 40 (47%) were females. Tumors were colorectal 43 (51%), pancreatic 11 (13%) esophageal 11 (13%) gastric 8 (9%), hepatobiliary 8 (11%), anal canal 3 (4%), and small intestine 1 (1%). 80 cases (94%) had adequate sample for profiling. CRGAs were identified in 71 cases (89%) with a median of 3 CRGAs in the cohort. The most commonly detected CRGAs were TP53 45 (56%), APC 35 (44%), KRAS 32 (40%), ATM 14 (18%) and PTEN 5 (6%). The median number of CRBs was 9 and high expression levels were seen of mismatch repair biomarkers (MLH1, MSH2, MSH6, PMS2) in 45 (56%), TOPO1 in 44 (55%), PTEN in 27 (34%), TOP2A in 20 (25%), ERCC1 in 19 (24%), and both TS and TUBB3 in 17 (21%). Based on actionable CRGAs and CRBs, 94% of pts matched at least one FDA approved treatment with proven clinical benefit, with a median of 4 available therapies per pt. In addition, there was a median of 189 chemotherapy and 64 targeted therapy clinical trial opportunities available for these pts given the molecular blueprint of their GICs. Conclusions: Multiplatform molecular profiling identified a high frequency of actionable CRGAs and CRBs in GIC pts. This approach has the potential to aid in clinical decision-making by providing a stratification of beneficial therapeutic alternatives individualized to the molecular framework of tumor. Larger prospective studies are warranted to further investigate the impact of profiling guided treatment decisions on patient outcomes.

Representation of Minorities and Women in Oncology Clinical Trials: Review of the Past 14 Years

Purpose:

Many cancer clinical trials lack appropriate representation of specific patient populations, limiting their generalizability. Therefore, we determined the representation of ethnic minorities and women in cancer clinical trials.

Methods:

Enrollment data from all therapeutic trials reported as completed in ClinicalTrials.gov from 2003 to 2016 were analyzed. We calculated enrollment fractions (EFs) for each group, defined as the number of enrollees divided by the 2013 Surveillance, Epidemiology, and End Results (SEER) database cancer prevalence.

Results:

Of 1,012 clinical trials, 310 (31%) reported ethnicity with a total of 55,689 enrollees. Participation varied significantly across ethnic groups. Non-Hispanic whites were more likely to be enrolled in clinical trials (EF, 1.2%) than African Americans (EF, 0.7%; P < .001) and Hispanics (EF, 0.4%; P < .001). A decrease in African American (6% v 9.2%) and Hispanic (2.6% v 3.1%) enrollment was observed when compared with historical data from 1996 to 2002. Younger patients (age younger than 65 years) were more likely to be enrolled in clinical trials than the elderly (64% v 36%; P < .001). Low recruitment of female patients was observed in clinical trials for melanoma (35%), lung cancer (39%), and pancreatic cancer (40%).

Conclusion:

We observed a decrease in recruitment of minorities over the past 14 years compared with historical data. African Americans, Hispanics, and women were less likely to be enrolled in cancer clinical trials. Future trials should take extra measures to recruit participants that adequately represent the US cancer population.

Representation of Minorities and Women in Oncology Clinical Trials: Review of the Past 14 Years

PURPOSE

Many cancer clinical trials lack appropriate representation of specific patient populations, limiting their generalizability. Therefore, we determined the representation of ethnic minorities and women in cancer clinical trials.

METHODS

Enrollment data from all therapeutic trials reported as completed in ClinicalTrials.gov from 2003 to 2016 were analyzed. We calculated enrollment fractions (EFs) for each group, defined as the number of enrollees divided by the 2013 Surveillance, Epidemiology, and End Results (SEER) database cancer prevalence.

RESULTS

Of 1,012 clinical trials, 310 (31%) reported ethnicity with a total of 55,689 enrollees. Participation varied significantly across ethnic groups. Non-Hispanic whites were more likely to be enrolled in clinical trials (EF, 1.2%) than African Americans (EF, 0.7%; P < .001) and Hispanics (EF, 0.4%; P < .001). A decrease in African American (6% v 9.2%) and Hispanic (2.6% v 3.1%) enrollment was observed when compared with historical data from 1996 to 2002. Younger patients (age younger than 65 years) were more likely to be enrolled in clinical trials than the elderly (64% v 36%; P < .001). Low recruitment of female patients was observed in clinical trials for melanoma (35%), lung cancer (39%), and pancreatic cancer (40%).

CONCLUSION

We observed a decrease in recruitment of minorities over the past 14 years compared with historical data. African Americans, Hispanics, and women were less likely to be enrolled in cancer clinical trials. Future trials should take extra measures to recruit participants that adequately represent the US cancer population.

Clinical implications of next-generation sequencing in the treatment of brain cancer at a large academic institution

e13521

Background: The use of next-generation sequencing (NGS) in clinical practice has increased the treatment (tx) options for cancer patients (pts). The expansion of genomic libraries used by NGS databases has resulted in increased identification of targetable genomic alterations (GAs). The aim of this study was to identify the clinical implications of genomic library expansion in the detection of GAs in pts with brain cancer at a large academic institution. Methods: We retrospectively analyzed 71 consecutive pts with brain cancer at the John Theurer Cancer Center that had NGS performed between 02/2014 and 09/2016. GAs were identified using the FoundationOne assay (Foundation Medicine, Cambridge, MA). GAs, number (n) of available genomic-directed tx and n of clinical trials were reviewed. The NGS assay interrogated 236 genes and introns of 19 genes until 09/2014, and subsequently was expanded to include 315 genes and introns of 28 genes. We compared median survival, n of GAs found, n of available trials, and n of tx available in pts who received NGS until 09/2014 (G1, n = 33) with pts who received NGS after 9/2014 (G2, n = 38). Results: Median survival was 30 months (range 19.9-40.1), median age was 62 years (range 26-82), the median n of GAs/sample was 5 (range 1-11). There was a significant positive correlation between n of GAs/sample and n of available trials and tx (r = .5, p = .00 and r = .3, p = .00, respectively). There was a negative correlation between survival and n of GAs (r = -.3, p = .02). G1 harbored 142 GAs with a median n of 4 GAs/sample (range 1-10), while G2 harbored 170 GAs with a median n of 5.5 GAs/sample (range 0-11). There was an absolute increase of 19.7% in GAs in G2 compared to G1. There was no difference in median overall survival. Conclusions: The expansion of genomic libraries increased the detection of GAs, and was positively correlated with the n of tx and clinical trials available for brain cancer pts. Survival was not affected by the expansion of the genomic library, but higher n of GAs was correlated with shorter survival. Expansions of NGS databases lead to increased n of potential tx options for brain cancer pts. Further studies are needed to investigate the impact of NGS targeted tx on survival.

Evolution of pancreatic cancer survival over the past two decades

e15722

Background: Pancreatic cancer (PaCa) is a highly lethal disease, with a 5-year overall survival (OS) rate of approximately 6%, and a median OS of only 3–6 months (m). Despite recent improvements in surgical techniques and increased use of combination chemotherapy (CT), OS remains poor. This study aims to examine the factors that led to increased OS in PaCa patients (pts) over the past two decades in a single academic institution. Methods: All medical records of pts diagnosed with PaCa at the John Theurer Cancer Center from 1990 to 2012 were reviewed, and 916 PaCa pts were included in this analysis. We compared one group of pts diagnosed from 1990 to 2003 (G1, n = 482), with a group of pts diagnosed from 2004 to 2012 (G2, n = 434) in terms of OS, demographics, tumor features and treatment (tx). Results: Median age at diagnosis was 70.5 years (range 26-96). There was no significant difference between G1 and G2 in terms of age at diagnosis, stage of disease and number of pts who received surgery. A significantly higher percentage of pts received CT in G2 compared to G1 (66.5% vs 51.0%, p = .00). Tumors of the pancreatic head were more common in G1 compared to G2 (51.8% vs 44.4% p = .02). More pts in G2 received two or more CT agents compared to G1 (49.0% vs 34.1%, p = .00). Median OS was significantly longer in G2 compared to G1 (9m vs 5m, p = .00), in pts who received CT compared to pts who did not (3m vs 9m, p = .00) and in pts who received surgery compared to pts who did not (5m vs 19m, p = .00). Pancreatic head location was associated with improved OS compared to other locations (9m vs 5m, p = .00). No OS difference was found between pts who received combination with two or more agents vs single agent CT. Conclusions: In line with multiple studies, analysis of PaCa data from our institution showed an increase OS in pts diagnosed with PaCa in more recent years, and in those who received surgery and CT. CT was administered in a larger number of pts in G2, which might account for the better OS in this group. Pts diagnosed with tumors of the pancreatic head had better survival, which could be explained by earlier presentation leading to earlier diagnosis and tx. Further research in PaCa therapeutics is needed, as long-term OS in PaCa pts remains poor despite recent advances.

Prognostic significance of t(11;14) expression by FISH in patients with newly diagnosed multiple myeloma in the era of novel therapies

e19525

Background: Rearrangements of the immunoglobulin heavy chain (IGH) on chromosome 14 are identified by FISH in about 15-20% of patients (pts) with newly diagnosed multiple myeloma (MM). Historically there is variation on the significance on prognosis of these rearrangements: typically, t(4;14), t(14;16) and t(14;20) have high risk (HR), and t(11;14) have standard risk (SR). A recent study (Kaufman et al, Leukemia. 2016 30:633-9) suggests that t(11;14) may confer a worse prognosis We report the prognostic significance of t(11;14) in a single-institution MM cohort. Methods: 87 pts with t(11;14) by CD 138 selected FISH at diagnosis were identified, pts without symptomatic MM were excluded. Cox regression was used for statistical analysis. Progression free survival (PFS), and overall survivals (OS) from diagnosis and post autologous stem cell transplant (ASCT) were analyzed by Kaplan-Meier. Results: Median age at diagnosis was 62 years, 45 pts (52%) were male, and 24 pts (27%) had ISS 3. All pts received either a proteasome inhibitor or an immunomodulatory agent, and 42 (48%) received triplet treatment as induction. Sixty-nine (79%) pts had ASCT, and overall response rate (ORR, partial response or better) post ASCT was 73%. For pts with HR FISH (defined as t(14;16), p53 del, 1q21 gain or 1p del) compared to SR FISH, the ORR post ASCT was 70% vs 77% (p = 0.67). OS from diagnosis was 93% at 3 years, 74% at 4 years and 51% at 5 years. Seven patients (8%) developed plasma cell leukemia, and there was no association between HR and SR FISH (p = 0.66). In multivariate analysis, ISS stage was an independent risk factor for mortality; pts with stage 3 had 7.3 times (CI: 1.16-36.4) and 5.7 times (CI: 1.63-20.0) the risk of mortality than pts with stage 1 and 2. Having an ASCT reduced mortality by 87% (CI: 0.04-0.41). Conclusions: Despite the use of novel therapies the OS at 5 years of our pts with MM was not significantly improved compared to SEER data from 1992-2013 (51% vs 48.5%). Pts with t(11;14) who had ASCT had increased survival compared to those who did not. Our results suggest that t(11;14) may confer a worse prognosis. Further prospective studies evaluating the risk of t(11;14) are warranted.

Clinical implications of genomic-directed therapies by comprehensive genomic profiling in breast cancer patients at a large academic cancer center

e12037

Background: Increased use of comprehensive genomic profiling (CGP) has recently led to improved genomic characterization of tumors, increased access to individualized therapies and increased availability of clinical trials in breast cancer patients. The aim of this study was to evaluate the clinical impact of genomic profiling in breast cancer patients with the use of a CGP assay at a large cancer center. Methods: We retrospectively analyzed 101 consecutive breast cancer patients who received CGP at the John Theurer Cancer Center between 12/2011 and 08/2016. Genomic alterations (GAs) were identified using the FoundationOne assay (Foundation Medicine, Cambridge, MA). GAs, number of available genomic-directed therapies and number of available clinical trials were reviewed. The CGP interrogated up to 315 genes and introns of 28 genes. Results: Median age at diagnosis was 58 years (range: 35-83 years). With a median follow-up of 189 months (range 1-189), median survival was 163 months (range 142-184). A total of 560 GAs were found in our population, with a median of 5.0 GAs/sample (range 0-16), a median of 2.0 therapies/patient (range 0-11), and a median of 11.0 clinical trials/patient (range 0-36). The most frequent GAs found were TP53 (47.5%, n = 48), PIK3CA (34.7%, n = 35), MYC (22.8%, n = 23), CCND1 (19.8%, n = 20), FGF3 (16.8%, n = 17), FGF4 (15.8%, n = 16), and ZNF703 (14.9%, n = 15). A significant positive correlation was found between number of GAs and the number of available targeted therapies and clinical trials (r = 0.5 and r = 0.7, p = 0.00, respectively). Increasing age is a predictor of having a PIK3CA mutation (OR = 1.05; CI:1.01-1.09, p = 0.00) while decreasing age is a predictor of having a MYC mutation by logistic regression (OR = 0.95; CI:0.91-0.95, p = 0.03). Conclusions: The systematic use of CGP led to the identification of a high number of GAs, which correlated with a median of 2.0 individualized therapies and a median of 11.0 clinical trials available for breast cancer patients. The clinical impact of genomic-directed individualized therapies needs to be further investigated in prospective, randomized studies.

Racial differences in abnormalities by FISH in minorities with multiple myeloma: A single-center experience

8044

Background: Racial disparities of FISH abnormalities in multiple myeloma (MM) have been well described in whites (W) but partially described in minorities (M) (Paulus et al, ASH 2016, 4432). We aimed to explore racial-based differences of FISH abnormalities using the largest cohort of m to date. Methods: CD-138 selected FISH was done on 799 consecutive patients (pts). Pts without symptomatic MM, and biopsy >6 months after diagnosis were excluded. The abnormalities evaluated included standard and intermediate risk: IGH rearrangements (IGH r), t(4;14), t(11;14), and high risk: t(14;20), t(14;16), del13q, del 17p, 1q21. Chi-square was used for statistical analysis. Due to smaller numbers, all m (Hispanic (H), Black (B), Asian (A) and Other (O)) were included into the same group for statistical analysis. Results: 482 pts were eligible, 343 (71%) were W, 52 (10%) H, 50 (10%) B, 19 (3%) A, and 18 (3%) O. Median age was 65 years, 54% were male, and 26% ISS stage 3. There were no were no statistically significant differences in FISH abnormalities between the m (Table1). Overall W had more abnormalities in IGH r, t(4;14), t(11;14), t(14:20), 1q21 gain compared to M. Most notably W had more IGH r (39% vs 28%; p=0.019) and t(11;14) (20% vs 12%; p=0.024). There were no statistically significant differences between W and m in the high risk FISH abnormalities. Conclusions: We had significant differences in FISH compared to M. W had more IGH r and t(11;14) than M, and there was no difference in high risk FISH abnormalities between W and M. This study confirms the biological racial disparities that exist in minorities with MM. Further studies with more inclusion of minorities are needed to elucidate these disparities and its effects on risk stratification and outcomes. [Table: see text]

Genomic-directed therapy of gastrointestinal cancers by comprehensive genomic profile (CGP): Clinical and genomic characteristics at the John Theurer Cancer Center (JTCC)

803

Background: The routine use of comprehensive genomic profiling (CGP) has led to better genomic characterization, more personalized treatments and higher enrollment in clinical trials for many solid cancers. The aim of this study was to evaluate the clinical and genomic prolife of patients (pts) with gastrointestinal cancers (GIC) with the use of a CGP assay and their potential therapeutic benefit at a large comprehensive cancer center. Methods: We retrospectively analyzed 83 consecutive pts with GIC that had CGP at the JTCC between 01/2014-09/2016. Demographics, CGP results, and clinical characteristics were studied. Clinically relevant genomic alterations (CRGAs) were identified using the FoundationOne assay (Foundation Medicine, Cambridge, MA), and were defined as alterations linked to approved therapies and those under evaluation in genotype-driven clinical trials. Results: 83 pts with GIC were analyzed. 46.9% males and 53.0% females. The median age at diagnosis was 58.5 years. 73 (88%) patients were stage III/IV at diagnosis. Tumors consisted of colorectal (36%), gastric (24%), pancreatic (12%) and cholangiocarcinoma (6%). All the pts were found to have at least one genomic alteration (GA). CRGAs were identified in 63 (76%) of cases, with an average of 2 per pt. The most frequently identified GAs were TP53 43 (29.9%), APC 24 (16.7%), KRAS 31 (21.5%), ARID1A 12 (8.3%), PIK3CA 11 (7.6%), and SMAD4 11 (7.6%). Of the 63 pts with CRGAs, 23 (36%) received one of the therapies suggested by the NGS assay, and 9 (14%) were enrolled in a clinical trial. Conclusions: The use of CGP assay identified a high frequency of CRGAs in GIC pts. The assay provided information for genomic-directed therapy in 36% of the patients. The routine use of CGP in GIC may lead to pts receive more personalized treatments. The potential benefit of selected targeted therapies in individual cases suggests that larger studies of treatment guided by routine CGP in gastrointestinal malignancies are warranted.

Comorbidities limiting recruitment of colorectal cancer (CRC) patients in early-phase trials

791

Background: With the surge of drug development in the past decade, early phase clinical trials (EPCT) have gained value evaluating the potential benefits of new therapies. The inclusion/exclusion criteria in EPCT are usually rigorous and may exclude many patients (pts) commonly seen in clinical practice. Our objective was to identify the most common comorbidities excluded in EPCT for CRC. Methods: ClinicalTrials.gov was queried on December 1stof 2015. We reviewed the characteristics and eligibility criteria of 369 phase I/II interventional drug trials including: experimental arm therapy, location, and exclusion/inclusion criteria. Logistic regressions were completed and exclusion was studied as a binary variable. Results: Of the 369 trials, 68% were phase II and 32% phase I. 46% were conducted in the United States, 30% in Europe, 15% in Asia and 9% in other locations. 74 (20%) trials excluded pts > 70 years of age. 142 (39%) trials required creatinine levels < 1.5 mg/dl, liver enzymes (AST/ALT) < 2.5 and bilirubin < 1.5 of the upper limit of normal. Cytopenia was a significant exclusion factor: 147 (47%) trials required Hgb > 9 g/dl and 218 (59%) excluded pts with platelets < 100,000/dl. In terms of comorbidities, 98 (27%) trials excluded pts with heart failure (NYHA class 3/4), 74 (20%) with atrial fibrillation, 112 (31%) with any anticoagulation therapy and 155 (42%) with positive HIV. Trials located in the US were more likely to exclude pts with Hgb < 9g/dl (OR: 1.5, 95%CI: 1.1-2.3, p < 0.05), immunotherapy trials were more likely to exclude pts on any anticoagulation (OR:1.8, 95%CI: 1.2-2.8, p < 0.007) and targeted therapy trials were more likely to exclude pts with history of DVT/PE or cardiovascular diseases (OR: 3.4, 95%CI: 1.9-5.8, p < 0.0001; OR: 2.3, 95%CI: 1.3-3.8, p < 0.002, respectively). Conclusions: 20% of EPCTs on CRC excluded pts with advanced age, organ dysfunction and common comorbidities. Many of the EPCT reviewed were not inclusive of our aging oncology population who are more likely to have multiple comorbidities. Investigators should review whether sufficient justification exists for every exclusion criterion before their incorporation in future trial protocols.

Analysis of clinical and research implications of expanding next-generation sequencing (NGS) libraries in the treatment options of gastrointestinal cancers in a large cancer center

774

Background: The use of next-generation sequencing (NGS) in clinical practice has increased the therapeutic options for many cancers. Multiple NGS assays are now commercially used. The genomic libraries used by these assays are continuously being expanded, resulting in increased detections of genomic alterations (GAs) leading to potential new treatments. The aim of this study was to identify the clinical and research implications of a database expansion in the detection of GAs in patients with gastrointestinal cancers (GIC). Methods: We retrospectively analyzed 83 consecutive patients with GIC that had NGS at the John Theurer Cancer Center between 01/2014 and 08/2016. GAs were identified using the FoundationOne assay (Foundation Medicine, Cambridge, MA). GAs, number of genomic-directed therapies and number of clinical trials were reviewed. Results: Period 1 (P1) comprised 01/2014-09/2014, period 2 (P2) comprised 10/2014-08/2016. The NGS assay interrogated 236 genes and introns of 19 genes during P1, and was expanded to 315 genes and introns of 28 genes during P2. The 21 samples analyzed during P1 harbored a total of 82 GAs with an average of 3.9 GAs/sample (range 1-7). The 62 samples analyzed during P2 harbored a total of 342 GAs with an average of 5.5 GAs/sample (range 1-20); representing an increase of 42% in GAs from P1 to P2. 41 GAs in 29 genes were detected in P2 that were not interrogated during P1. The average of genomic-directed therapies with potential clinical benefit increased from 1.8 during P1 vs 2.6 during P2 (44.4% increase in potential therapies). Based on new genomic findings, more clinical trials were made available during P2; an average of 8 vs 4 clinical trials (100% increase in available clinical trials). Conclusions: Periodical updates on NGS and the expansion of genomic libraries increase the detection of GAs, potential new genomic-directed therapies and available clinical trials. Continued expansions of NGS are needed to improve genomic characterization, and increase in the personalized therapeutic options for our patients with GIC.

Predictors of recurrence in stage I/II colorectal cancer: Analysis of a single institution’s experience for over 24 years

534

Background: Surgical resection remains a mainstay of treatment for localized colorectal cancer (CRC). However, up to 35% of patients (pts) develop recurrence with a significant decrease in overall survival. The aim of this study was to evaluate clinical characteristics and specific predictors of recurrence for stage I/II CRC Methods: We performed a retrospective analysis of 2310 pts diagnosed with stage I/II CRC at our institution between 1990 and 2013, with a minimum follow up of 24 months. Tumor characteristics and recurrence data were studied. Cox regression was used for statistical analysis. Results: Of the 2310 pts, recurrence was identified in 276 (12%) pts, of which 88 (32%) pts had early recurrence ( < 1 year). Median time of recurrence was 625 days. 210 (76%) pts had local recurrence while 66 (24%) pts had distal recurrence, with liver being the main site of distant metastasis. Males had a higher recurrence rate than females (16% vs. 7%, p < 0.0001). When comparing recurrence pts (R) with non-recurrence (NR) pts, Hispanics were more prevalent in the R group (13% vs 8%, p < 0.01). R pts were more likely to have rectal and poorly differentiated tumors (33% vs. 13%, p < 0.0001; 23% vs. 14%, p < 0.0001, respectively). Mucinous adenocarcinomas were more prevalent in the R group (16% vs. 9%, p < 0.0005). In multivariate analysis: male sex (OR: 1.24, 95% CI: 1.21-1.56, p < 0.001), rectal tumors (OR: 4.2, 95% CI: 3.14-5.06, p < 0.001), tumor size (OR: 2.55, 95% CI: 1.31-4-51, p < 0.01), and positive margins (OR: 3.6, 95% CI: 2.13-5.21, p < 0.04) were independent predictors for recurrence. K-ras (+) status was a predictor of recurrence by univariate analysis. Age, race, and histologic grade were not predictors of recurrence. Pts with early recurrence had a lower median overall survival compared to pts that developed recurrence after 1 year, 37 months (95% CI: 26.6-49.6) vs. 59 months (95% CI: 42.4-74.3), p < 0.04. Conclusions: Pts with recurrence were more likely to have rectal and poorly differentiated tumors. Poorer survival was observed in pts with early recurrence. These findings could serve as important prognostic information and may help identify the pts that would benefit from aggressive surveillance after treatment.

An additional phosphate-binding element in arrestin molecule. Implications for the mechanism of arrestin activation

Arrestins quench the signaling of a wide variety of G protein-coupled receptors by virtue of high-affinity binding to phosphorylated activated receptors. The high selectivity of arrestins for this particular functional form of receptor ensures their timely binding and dissociation. In a continuing effort to elucidate the molecular mechanisms responsible for arrestin's selectivity, we used the visual arrestin model to probe the functions of its N-terminal beta-strand I comprising the highly conserved hydrophobic element Val-Ile-Phe (residues 11-13) and the adjacent positively charged Lys(14) and Lys(15). Charge elimination and reversal in positions 14 and 15 dramatically reduce arrestin binding to phosphorylated light-activated rhodopsin (P-Rh*). The same mutations in the context of various constitutively active arrestin mutants (which bind to P-Rh*, dark phosphorylated rhodopsin (P-Rh), and unphosphorylated light-activated rhodopsin (Rh*)) have minimum impact on P-Rh* and Rh* binding and virtually eliminate P-Rh binding. These results suggest that the two lysines "guide" receptor-attached phosphates toward the phosphorylation-sensitive trigger Arg(175) and participate in phosphate binding in the active state of arrestin. The elimination of the hydrophobic side chains of residues 11-13 (triple mutation V11A, I12A, and F13A) moderately enhances arrestin binding to P-Rh and Rh*. The effects of triple mutation V11A, I12A, and F13A in the context of phosphorylation-independent mutants suggest that residues 11-13 play a dual role. They stabilize arrestin's basal conformation via interaction with hydrophobic elements in arrestin's C-tail and alpha-helix I as well as its active state by interactions with alternative partners. In the context of the recently solved crystal structure of arrestin's basal state, these findings allow us to propose a model of initial phosphate-driven structural rearrangements in arrestin that ultimately result in its transition into the active receptor-binding state.