George, Thomas J.

231 Background: The recent development of KRAS G12C inhibitors underscores the potential to target KRAS mutations. Right-sided and left-sided colon tumors (RT and LT) exhibit different molecular features. We characterize the prevalence of KRAS-variants, interrelation with primary tumor location, and association with immune biomarkers in CRC. Methods: We retrospectively reviewed CRC tumors of all stages (with known sidedness) that underwent NGS with the Tempus xT assay (DNA-seq of 648 genes at 500x coverage, full transcriptome RNA-seq). Bivariate analyses were performed to compare KRAS alterations, immune biomarkers, and co-mutations by tumor location. P-values comparing individual co-mutations between groups were adjusted for false discovery (FDR). Results: A total of 3,391 CRC were analyzed (RT: n = 442 [13%], transverse: n = 116 [3%], LT; n = 2,833 [84%]) of which 1486 (44%) tumors harbored KRAS mutations. Overall, KRAS mutations were more frequent in RT compared to transverse tumors and LT (52% vs 41% vs 43%, p<0.001, respectively). The most frequent KRAS mutation variants observed were G12D (29 %), G12V (22%), G13D (16%), and G12C (5.7%). There was no significant difference in the prevalence of KRAS variant types between LT and RT (p=0.5). Significant differences in genomic co-mutations with various KRAS variants were observed in the following genes: TP53, FBXW7, and NF1 (FDR- P<0.05). RT and transverse tumors were more likely have MSI-H and TMB-H (>10 mut/mb) status than LT (MSI-H: 18% vs 22% vs 2.2% and TMB-H (20% vs 22% vs 3%, P<0.001), respectively. CRC tumors harboring G13D variants were more likely to be associated with and MSI-H and TMB-H status (and 7.7% and 8.5%) compared to G12D (2.8 % and 3.9 %), G12V (1.8 % and 2.1%), and G12C (0% and 2.4%); P = 0.003 and 0.001. Conclusions: The most frequent KRAS mutation variants observed in CRC tumors were G12D, G12V, G13D, and G12C. There was no significant difference in the prevalence of KRAS variant types between tumors of the left vs right colon. CRC tumors that harbored G13D variants were significantly more likely to be associated with MSI-H and TMB-H status. [Table: see text]

PURPOSE This randomized, open-label trial compared the efficacy and safety of adjuvant nab-paclitaxel + gemcitabine with those of gemcitabine for resected pancreatic ductal adenocarcinoma (ClinicalTrials.gov identifier: NCT01964430 ). METHODS We assigned 866 treatment-naive patients with pancreatic ductal adenocarcinoma to nab-paclitaxel (125 mg/m 2 ) + gemcitabine (1,000 mg/m 2 ) or gemcitabine alone to one 30-40 infusion on days 1, 8, and 15 of six 28-day cycles. The primary end point was independently assessed disease-free survival (DFS). Additional end points included investigator-assessed DFS, overall survival (OS), and safety. RESULTS Two hundred eighty-seven of 432 patients and 310 of 434 patients completed nab-paclitaxel + gemcitabine and gemcitabine treatment, respectively. At primary data cutoff (December 31, 2018; median follow-up, 38.5 [interquartile range [IQR], 33.8-43 months), the median independently assessed DFS was 19.4 ( nab-paclitaxel + gemcitabine) versus 18.8 months (gemcitabine; hazard ratio [HR], 0.88; 95% CI, 0.729 to 1.063; P = .18). The median investigator-assessed DFS was 16.6 (IQR, 8.4-47.0) and 13.7 (IQR, 8.3-44.1) months, respectively (HR, 0.82; 95% CI, 0.694 to 0.965; P = .02). The median OS (427 events; 68% mature) was 40.5 (IQR, 20.7 to not reached) and 36.2 (IQR, 17.7-53.3) months, respectively (HR, 0.82; 95% CI, 0.680 to 0.996; P = .045). At a 16-month follow-up (cutoff, April 3, 2020; median follow-up, 51.4 months [IQR, 47.0-57.0]), the median OS (511 events; 81% mature) was 41.8 ( nab-paclitaxel + gemcitabine) versus 37.7 months (gemcitabine; HR, 0.82; 95% CI, 0.687 to 0.973; P = .0232). At the 5-year follow-up (cutoff, April 9, 2021; median follow-up, 63.2 months [IQR, 60.1-68.7]), the median OS (555 events; 88% mature) was 41.8 versus 37.7 months, respectively (HR, 0.80; 95% CI, 0.678 to 0.947; P = .0091). Eighty-six percent ( nab-paclitaxel + gemcitabine) and 68% (gemcitabine) of patients experienced grade ≥ 3 treatment-emergent adverse events. Two patients per study arm died of treatment-emergent adverse events. CONCLUSION The primary end point (independently assessed DFS) was not met despite favorable OS seen with nab-paclitaxel + gemcitabine.

3606 Background: Standard adjuvant CT for stage III CC are FOLFOX and CAPOX. Recently, IDEA study separated stage III patients (pts) into low risk (T1 to 3, N1) and high risk (T4 or N2). We recently confirmed benefit of OX in both risk groups. However, we observed a difference in the two high-risk subgroups, with benefit in N2 but not in T4 (Margalit O et al; Clin Colorectal Cancer 2021). This prompted us to compare outcomes (OS/TTR) between treatment with OX vs. without OX within sub-stage III CC groups defined by T and N. Methods: We pooled 4941 stage III CC pts from the three studies evaluating 6 months of CT with fluoropyrimidine (FP) ± OX: MOSAIC, C-07 and XELOXA. Baseline characteristics were compared using χ2 and t-test. OS was compared between OX and no OX in T and N subgroups. Kaplan-Meier analyses, adjusted and unadjusted Cox models stratified by study were used. Sub-groups classification was done according to OX benefit and verified by interaction test (Int) considered as significant with a P<0.1. We considered for recommendation of using OX-based adjuvant CT, 1) significant benefit in OS, 2) significant Int between substage and adjuvant therapy, and 3) the three individual trials showing similar results (benefit or non-benefit of OX). Results: In stage III population, T3 pts were 74.9%, T1-2 12.4%, T4 13.1%, while N stage was N1 64.7% and N2 35.3%. Population was well balanced according to treatment allocation in most subgroups. A significant benefit of OX was only observed in T3N1 and T3N2 (OS HR 0.76). Whatever N stage, there was no significant benefit of OX in the T1-2 and T4 subgroups. The effect of OX+FP vs FP alone in OS of the three studies differed between T3 and T1-2 subgroups (P = 0.047). Interaction was borderline between T3 and T4 subgroups (P = 0.10) but there was no interaction between T1-2 and T4 subgroups (P = 0.429). A benefit of OX in TTR remained in the T4 population. Discrepancy between advantage in time to relapse (TTR) and no advantage in OS was not explained by survival post relapse. Conclusions: Our analysis suggested that pts with T1-2N1-2 and T4N1-2 disease had no OS benefit of addition of OX to FP. The good survival achieved with FP alone in T1-2N1-2 pts (5-yr OS 89%) question the addition of OX. In the T4 population our results suggested that benefit of OX was limited and that further studies should assess this issue or at least stratify pts on T stage in the future adjuvant trials in CC. [Table: see text]

3067 Background: The KEYNOTE-177 trial demonstrated pembrolizumab’s superiority over first-line chemotherapy in patients with MSI-H/dMMR mCRC. However, in a subgroup analysis, patients with KRAS or NRAS mutations did not show the same favorable PFS benefit with PD-1 blockade therapy (HR 1,19; CI 0.68-2,07). The impact of RAS mutations on the immunologic characteristics of the TME of MSI-H/dMMR CRC has not been well characterized. Methods: A retrospective review of deidentified records of patients with MSI-H/dMMR CRC tumors was conducted using next-generation sequencing data (Tempus |xT assay, DNA-seq of 595-648 genes at 500x coverage, and full transcriptome RNA-seq). MSI-H determined by assessment of 239 loci by NGS. Several immune markers were assessed, including tumor mutational burden (TMB), neoantigen tumor burden (NTB, ScanNeo), PD-L1 expression, immune infiltration, and canonical immune pathways (82 geneset signatures). Results: A total of 463 MSI-H/dMMR CRCs were analyzed, of which 110 (24%) tumors harbored RAS mutations ( RASmut) [ KRAS: 93%, NRAS 6% and HRAS 1%}, while 353 were RAS-wild-type ( RASWT). Compared to MSI-H/dMMR RASWT, MSI-H/dMMR RASmut tumors were more frequently identified in males (53% vs. 38%; P= 0.005), and younger patients (median age: 57 yrs vs. 71 yrs, P< 0.001). Although there were no significant differences in median TMB (40 mut/MB for both, p = 0.9) or frequency of TMB-high status (≥10 mut/MB) between the two groups, RASmuttumors tended to have a lower tumor NTB (16 vs. 12 neoAg/Mb, P <0.001) and lower % CD8 T cell but higher % CD4 T cell infiltration (P < 0.05). Significant differences were observed in genomic alterations co-occurring with RASmut compared to RASWT (e.g., MLH1 (23% vs. 8.8%, P < 0.001), MSH6 (36% vs. 24%, P = 0.017), APC (60% vs. 20%, P< 0.001), ARID1A (54% vs 30%, P< 0.001), PIK3CA (36% vs 19%, P< 0.001), and TP53 (32% vs. 19%, P = 0.014). Pathway enrichment analysis identified 14 differentially expressed pathways among RASmut tumors. Four pathways showed significant upregulation, including Hedgehog, Wnt, TGFβ, and cancer stem cell pathways. Ten pathways of interest showed significant downregulation among RASmut tumors. The majority (9/10) were immune-related, including cytokine signaling [ JAK-STAT, TGFβ, TH1], innate immune [ NK cells], and adaptive immune events (CD8 T cell, Tregs)]. Conclusions: MSI-H/dMMR CRCs harboring RASmut exhibited overall upregulated WNT/SHH pathway activity, coupled with reduced NTB, cytokine signaling, and innate and adaptive immune events. TGFβ is pleiotropic, and different members were associated with variable modulation. These data suggest that MSI-H/dMMR CRCs harboring RAS mutations are less immunogenic and appeared to contain a TME that is less sensitive to immune checkpoint blockade than MSI-H/dMMR RAS wt CRCs.

3066 Background: The BRAFV600E mutation is associated with the hypermethylator phenotype CIMP, which can also lead to the MSI-H phenotype. BRAFV600E mutation and MSI-H/dMMR status seem to be biologically intertwined; however, the impact of coexisting BRAFV600E mutations on the TME and immunometabolomic features of MSI-H/dMMR CRC tumors is not well characterized. Methods: A retrospective review of deidentified records of patients with MSI-H/dMMR CRC tumors was conducted using next-generation sequencing data (Tempus |xT assay: DNA-seq of 595-648 genes at 500x coverage, and full transcriptome RNA-seq). Several immune markers of tumor immunogenicity in BRAF wild-type ( BRAFwt) vs. V600E-mutated (BRAFV600E) tumors were assessed, including tumor mutational burden (TMB), neoantigen tumor burden (NTB, ScanNeo), PD-L1 expression, immune infiltration, and canonical immuno-metabolomic pathways (82 geneset signatures). Results: A total of 459 MSI-H/dMMR CRC tumors were analyzed, of which 123 (27%) tumors harbored BRAFV600Emutations, and 336 (73%) were BRAFwt. MSI-H/dMMR BRAFV600E tumors were more frequently identified in females (69% vs. 55%; P= 0.01), non-Hispanic or non-Latino (100% vs. 73%; P =0.001), and older patients (median age: 76 yrs vs. 62 yrs; P< 0.001). Compared to BRAFWT, BRAFV600E tumors exhibited significantly higher TMB (Median: 49 mut/MB vs. 36 mut/MB; P < 0.001) and were more frequently associated with TMB-High status (> 10 mut/MB; 100% vs. 95%; P = 0.008); however, no significant differences were observed with tumor NTB, immune score, or T cell infiltration (CD4 or CD8). NK cell infiltration was higher in the BRAFV600Ecohort (< 0.001). When compared to BRAFWT tumors , BRAFV600E tumors harbored a greater frequency of mutations in MSH6(42% vs. 20%), B2M (33% vs. 16%), BRCA2 (31% vs. 12%), ATM (23% vs. 12%), and TP53 (30% vs. 19%) but a lower frequency of MSH2 (3.3% vs. 11%), all P< 0.05. Pathway enrichment analysis identified 10 significantly altered signaling pathways, most of which related to stromal/immune cell signaling and metabolism. Five were upregulated among BRAFV600E tumors: glycerophospholipid, galactose, cyclin-dependent cell signaling; Nucleotide, and TH1 inflammation. Five pathways were downregulated (Wnt, Notch, TH17 inflammation, amino sugar, and cancer stem cell signaling). Conclusions: MSI-H/dMMR BRAFV600E CRCs undergo broad metabolic reprogramming (e.g., glycerophospholipidgalactose, nucleotide). A rise in lipid metabolism, particularly with NK inflammation, suggests that BRAFV600Emutated tumors may be associated with a non-classical immune component. BRAFV600E and BRAFwt CRCs exhibited similar NTB and T cell infiltration, suggesting that both are likely to benefit from immune checkpoint inhibitors.

Abstract Background Disease-free survival (DFS) with a 3-year median follow-up (3-year DFS) was validated as a surrogate for overall survival (OS) with a 5-year median follow-up (5-year OS) in adjuvant chemotherapy colon cancer (CC) trials. Recent data show further improvements in OS and survival after recurrence in patients who received adjuvant FOLFOX. Hence, reevaluation of the association between DFS and OS and determination of the optimal follow-up duration of OS to aid its utility in future adjuvant trials are needed. Methods Individual patient data from 9 randomized studies conducted between 1998 and 2009 were included; 3 trials tested biologics. Trial-level surrogacy examining the correlation of treatment effect estimates of 3-year DFS with 5 to 6.5-year OS was evaluated using both linear regression (RWLS2) and Copula bivariate (RCopula2) models and reported with 95% confidence intervals (CIs). For R2, a value closer to 1 indicates a stronger correlation. Results Data from a total of 18 396 patients were analyzed (median age = 59 years; 54.0% male), with 54.1% having low-risk tumors (T1-3 and N1), 31.6% KRAS mutated, 12.3% BRAF mutated, and 12.4% microsatellite instability high or deficient mismatch repair tumors. Trial-level correlation between 3-year DFS and 5-year OS remained strong (RWLS2 = 0.82, 95% CI = 0.67 to 0.98; RCopula2 = 0.92, 95% CI = 0.83 to 1.00) and increased as the median follow-up of OS extended. Analyses limited to trials that tested biologics showed consistent results. Conclusions Three-year DFS remains a validated surrogate endpoint for 5-year OS in adjuvant CC trials. The correlation was likely strengthened with 6 years of follow-up for OS.

Abstract Objective Social determinants of health (SDoH) are nonclinical dispositions that impact patient health risks and clinical outcomes. Leveraging SDoH in clinical decision-making can potentially improve diagnosis, treatment planning, and patient outcomes. Despite increased interest in capturing SDoH in electronic health records (EHRs), such information is typically locked in unstructured clinical notes. Natural language processing (NLP) is the key technology to extract SDoH information from clinical text and expand its utility in patient care and research. This article presents a systematic review of the state-of-the-art NLP approaches and tools that focus on identifying and extracting SDoH data from unstructured clinical text in EHRs. Materials and Methods A broad literature search was conducted in February 2021 using 3 scholarly databases (ACL Anthology, PubMed, and Scopus) following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A total of 6402 publications were initially identified, and after applying the study inclusion criteria, 82 publications were selected for the final review. Results Smoking status (n = 27), substance use (n = 21), homelessness (n = 20), and alcohol use (n = 15) are the most frequently studied SDoH categories. Homelessness (n = 7) and other less-studied SDoH (eg, education, financial problems, social isolation and support, family problems) are mostly identified using rule-based approaches. In contrast, machine learning approaches are popular for identifying smoking status (n = 13), substance use (n = 9), and alcohol use (n = 9). Conclusion NLP offers significant potential to extract SDoH data from narrative clinical notes, which in turn can aid in the development of screening tools, risk prediction models, and clinical decision support systems.

Abstract Background Colon cancer (CC) incidence in young adults (age 20-49 years), termed early-onset CC (EO-CC), is increasing. Methods Individual patient data on 35 713 subjects with stage III colon cancer from 25 randomized studies in the Adjuvant Colon Cancer ENdpoint database were pooled. The distributions of demographics, clinicopathological features, biomarker status, and outcome data were summarized by age group. Overall survival, disease-free survival, time to recurrence, and survival after recurrence were assessed by Kaplan-Meier curves and Cox models stratified by treatment arms within studies, adjusting for sex, race, body mass index, performance status, disease stage, grade, risk group, number of lymph nodes examined, disease sidedness, and molecular markers. All statistical tests were 2-sided. Results Using a 5% difference between age groups as the clinically meaningful cutoff, patients with stage III EO-CC had similar sex, race, performance status, risk group, tumor sidedness, and T stage compared with patients with late-onset CC (age 50 years and older). EO-CC patients were less likely to be overweight (30.2% vs 36.2%) and more commonly had 12 or more lymph nodes resected (69.5% vs 58.7%). EO-CC tumors were more frequently mismatch repair deficient (16.4% vs 11.5%) and less likely to have BRAFV600E (5.6% vs 14.0%), suggesting a higher rate of Lynch syndrome in EO-CC. Patients with EO-CC had statistically significantly better overall survival (hazard ratio [HR] = 0.81, 95% confidence interval [CI] = 0.74 to 0.89; P &lt; .001), disease-free survival (HR = 0.91, 95% CI = 0.84 to 0.98; P = .01), and survival after recurrence (HR = 0.88, 95% CI = 0.80 to 0.97; P = .008) in the analysis without molecular markers; however, age at onset of CC lost its prognostic value when outcome was adjusted for molecular markers. Conclusion Tumor biology was found to be a more important prognostic factor than age of onset among stage III colon cancer patients in the Adjuvant Colon Cancer ENdpoint database.

3597 Background: Colon cancer (CC) incidence and mortality have decreased since the 1970s, but the incidence in young adults (20-49 years) is increasing. There are limited data suggesting that, as a group, patients with early onset CRC (eoCC) may have different phenotypic characteristics compared to those with late onset CRC (loCC, age ≥ 50 years). Methods: Individual patient data on 35,713 subjects with stage III CC from 25 randomized studies (recruiting between 1987 and 2009) in the ACCENT database were pooled. The distributions of demographics, clinicopathological features, biomarkers, and outcome data were summarized by age group. Overall survival (OS), disease-free survival (DFS), recurrence free rate (RFR), and survival after recurrence (SAR) were assessed by Kaplan-Meier curves and Cox models stratified by treatment arms within studies, adjusting for gender, race, body mass index, performance status, disease stage, grade, risk group, number of lymph nodes examined, disease sidedness and molecular markers. Results: Using a 5% difference between age groups as the clinically meaningful cutoff, patients with stage III eoCC (n = 6246) had similar distributions according to gender, race, PS, risk group, tumor sidedness and T/N stage compared to those with loCC (n = 29467). Patients with eoCC were significantly less likely to be overweight (30.2% vs 36.2%) but more commonly had ≥ 12 lymph nodes resected (69.5% vs 58.7%). The eoCC tumors were more frequently mismatch repair deficient (16.4% vs 11.5%), and less likely to have BRAFV600E (5.6% vs 14.0%), suggesting a higher frequency of Lynch syndrome in eoCC. In univariate analysis, patients with stage III eoCC had significantly better OS, DFS, and SAR; the difference between 3-year DFS and RFR strongly suggests the OS/DFS difference between these the eoCC and loCC may be due to increased competing risks and comorbidities in patients with loCC. In multivariate analysis, age at onset lost its prognostic value when outcome was adjusted for molecular markers. The clear relation between age of onset and KRAS/BRAF status was confirmed in the interaction analysis. Conclusions: Tumor biology was an important determinant of prognosis regardless of patient age. In multivariate analysis age of onset was not a statistically significant determinant of outcome.[Table: see text]

Background: Advance directives are legal documents that include living wills and durable health care power of attorney documents. They are critical components of care for seriously ill patients which are designed to be implemented when a patient is terminally ill and incapacitated. We sought to evaluate potential reasons for why advance directives were not appropriately implemented, by reviewing the electronic health record (EHR) in patients with terminal cancer. Methods: A retrospective analysis of the EHR of 500 cancer patients from 1/1/2013 to 12/31/2016 was performed. Data points were manually collected and entered in a central database. Results: Of the 500 patients, 160 (32%) had an advance directive (AD). The most common clinical terminology used by physicians indicating a terminal diagnosis was progressive (36.6%) and palliative (31%). The most common clinical terminology indicating incapacity was altered mental status (25.6%), and not oriented (14%). 34 (6.8%) patients met all criteria of having a terminal diagnosis, a documented AD, and were deemed incapacitated. Of these patients who met all of these data points, their ADs were implemented on average 1.7 days (SD: 4.4 days) after which they should have been. This resulted in a total of 58 days of additional care provided. Discussion: This study provided insight on to how ADs are managed in day to day practice in the hospital. From our analysis it appears that physicians are able to identify when a patient is terminal, however, it is typically later than it should have been recognized. Further studies should be performed focusing on harnessing the power of the EHR and providing physicians formative and evaluative feedback of practice patterns to ensure that ADs are honored when appropriate.