Neoadjuvant therapy guideline adherence for locally advanced rectal cancers in a region with sociodemographic disparities

71

Background: Quality and practice guidelines from the American College of Surgeons Commission on Cancer (CoC) and the National Comprehensive Cancer Network (NCCN) recommend neoadjuvant chemoradiation (NCR) for locally advanced rectal cancer (LARC). We examined guideline adherence in a healthcare system serving a region with significant socioeconomic disparities and poor cancer outcomes. Methods: We performed a retrospective analysis of patients from 2005-2014 with stage II and III rectal cancer in our local 5- hospital healthcare system in the Mid-South region of the US. We examined the associations between guideline adherence and patient demographic, socioeconomic, and clinicopathologic data. Results: Of 157 stage II/III RC patients, 96 (61.1%) received NCR. By univariate analysis, factors associated with receipt of NCR included white race versus non-white (OR = 2.14, p = 0.024), private insurance versus no or public insurance (OR = 2.70, p = 0.005), employed status versus unemployed (OR = 2.29, p = 0.031), age at diagnosis (OR = 0.74, p = 0.032), and appropriate local staging with EUS or MRI (OR = 6.67, p = < 0.0001). Those who were diagnosed and treated later in the study period were more likely to get NCR (OR per 1 year = 1.20, p = 0.006). In addition, receipt of NCR was protective against death at time of follow up (OR = 0.41, p = 0.009). By multivariate analysis, those with private insurance (OR = 2.51, p = 0.023), younger age at diagnosis (OR per 10 years = 1.39, p = 0.048) and with appropriate local staging (OR = 6.67, p < 0.0001) were more likely to have received NCR. Conclusions: Guideline adherence for LARC in our system, which serves a population with socioeconomic disparities, is relatively low. Sociodemographic factors such as employment and race influenced receipt of NCR, while insurance status and age were independent determinants. Despite some increase in compliance over time, adherence to NCR for LARC remains lower than expected and represents an important target for improvement efforts.

Molecular comparison between peritoneal metastases (PM) and primary gastric (GC) and gastroesophageal junction (GEJ) cancer

4053

Background: PM from GC or GEJ portend a poor prognosis and molecular differences are ill defined. Methods: We compared genomic profiles of primary (P) GC and GEJ with PM patients (pts) and other metastases (OM) sent to Caris Life Sciences. Testing comprised immunohistochemistry (IHC) including programmed death ligand 1 (PD-L1) combined positive score (CPS), copy number alterations (CNA), 592-gene next-generation sequencing (NGS), microsatellite instability (MSI) and tumor mutational burden (TMB). Results: 1366 cases were identified: 1041 GC (707 P, 98 PM, 236 OM) and 325 GEJ (248 P, 5 PM, 72 OM). PM were increased in GC versus GEJ (9% v. 2%, p < 0.0001). 91% GC and 93% GEJ were adenocarcinoma (AD); GC were more likely signet ring (SR) histology versus GEJ (11% v. 3%, p < 0.0001) and GC PM were more likely SR versus other OM or P (13% v. 12% v. 7%, p = 0.067). The mean age of PM pts (57 years) was younger than primary GC (63, p = 0.002) and OM (61; p = 0.044). More PM GC pts were female than P or OM (48% v. 35% v. 34%, p = 0.03). No molecular profiling differences were seen between GEJ and GC pts and they were combined for analysis; findings from 1246 AD pts are shown below (see Table). OM (9%, p = 0.041) had more CNA in CCNE1 than PM (2%, p = 0.041) or P (5%, p = 0.002). Conclusions: Compared to P and OM GC, PM pts were younger, more likely female and had a higher incidence of SR histology. PD-L1, HER2 IHC and ERBB2 CNA were reduced in PM versus P, suggesting novel therapeutic targets are needed. [Table: see text]

Molecular comparison of primary colorectal cancer (pCRC) with peritoneal metastases (PM)

498

Background: PM from CRC are associated with poor outcomes; however, molecular differences are not defined. Methods: We compared tumor profiles of pCRC and PM patients (pts) from Caris Life Sciences. Testing included next-generation sequencing (NGS) of 592 genes, immunohistochemistry (IHC), copy number variants (CNV), microsatellite instability (MSI) and tumor mutational burden (TMB). Mutations were termed pathogenic (PATH) or variants of undetermined significance (VUS). TMB in mutations/Mb (MMB) was compared. Results: 617 pCRC and 348 PM pts had similar gender (55% male) and age (median 59). 232 pCRC were left-sided (LS), 189 right-sided (RS), 147 rectum (R) and 49 not otherwise specified (NOS); PM were 45 RS, 29 LS, 22 R and 252 NOS. For pts with IHC testing, expression was increased in PM in TOPO1 (62% v. 52%, p < 0.01), ERCC1 (27% v. 18%, p < 0.01) and MLH1 (96% v. 92%, p < 0.05) and decreased in PD-1 (36% v. 65%, p < 0.01), TOP2A (76% v. 100%, p < 0.01) and PTEN (64% v. 72%, p < 0.05). By sidedness, LS PM were more frequently TOP01 and PD-L1 and less commonly MGMT positive compared to pCRC. PTEN IHC was higher in R pCRC than PM. 7 CNVs were increased in PM ( ADGR2A2, CCND1, ELL, FGF3, FGF4, JAK3 and PDGFRB) and FLT3 CNVs were decreased. MYC CNVs were more common in RS PM compared to pCRC. No difference was seen in PM and pCRC PATHs in KRAS, BRAF, SMAD2, SMAD4, PTEN. PM had more PATHs in GNAS (8% v. 1%, p < 0.01) while pCRC PATHs were increased in APC (76% v. 48%, p < 0.01), TP53 (72% v. 53%, p < 0.01), ARID1A (29% v. 12%, p < 0.05), PIK3CA (22% v. 15%, p < 0.05) and FBXW7 (13% v. 7%, p < 0.01). LS PM had increased FLCN PATHs (12% v. 2%, p < 0.01); R PM had more PATHs in KMT2D (20% v. 1%, p < 0.01) and RNF43 (13% vs. 3%, p < 0.05). VUS were increased in 39/592 (7%) genes for PM compared to pCRC. No MSI or fusion difference was seen. 53% pCRC (median = 8) pts had TMB ≥8 MMB compared to 43% PMs (median = 7; p = 0.03); no TMB difference was seen for LS, RS or R subgroups. Conclusions: Compared to pCRC, PM had more PATHs in GNAS and less in classic CRC markers APC and TP53. While TMB was generally lower in PM, differences in IHC expression, CNV and VUSs may serve as biomarkers for PM requiring further study.

Utility of restaging patients with stage II/III rectal cancer following neoadjuvant chemo/XRT: A systematic review

495

Background: In the US, patients with clinical stage II/III rectal cancer typically receive neoadjuvant chemoradiation (chemo/XRT) over 5-6 weeks followed by a 6-10 week break before proctectomy. As this chemotherapy is delivered at radio-sensitizing doses, there is essentially a 3-month window during which potential systemic disease is untreated. Evidence regarding the utility of restaging patients prior to proctectomy is limited. Methods: PubMed, Scopus, Web of Science, and the Cochrane Library were searched for studies evaluating the utility of restaging patients with locally advanced rectal cancer after completion of long course chemo/XRT, and reporting changes in management after restaging. Studies that were non-English, included < 50 patients, or examining the diagnostic accuracy of specific imaging modalities were excluded. Study quality was evaluated using the modified Newcastle Ottawa Scale. Results: Eight studies were identified including a total of 1251 patients restaged between completion of chemo/XRT and proctectomy. All studies were retrospective (6 single institution, 2 multi-institution). Restaging identified new metastatic disease in 72 (6.0%) patients, with 4 studies reporting specific sites: liver (n = 28), lung (n = 8), adrenal (n = 1), bone (n = 1), and multiple sites (n = 7). Overall, progression (distant or local) was detected in 85 (6.8%) patients and resulted in a reported change in management in 71 (5.7%) patients. One study identified an association of high-grade tumors with progression (p ≤ 0.05), however, this was not reported in any other study. Moreover, tumor-related prognostic characteristics were inconsistently reported among studies, precluding meta-analysis. Conclusions: Although restaging between completion of neoadjuvant chemo/XRT and proctectomy detects disease progression in only a small percentage of patients, findings may alter the treatment plan. A multi-institutional collaboration with analysis of well-defined prognostic variables may better identify a group of patients most likely to benefit from restaging.

High-income is a stronger predictor than race in understanding disparate survival outcomes in PDAC

241

Background: Patients with pancreatic ductal adenocarcinoma (PDAC) from low income and minority racial groups have a lower reported long-term survival rates. It is unknown whether this is related to access to care, variations in genetic polymorphisms, or income status. We hypothesized that income status predicts survival better than race in PDAC. Methods: The pancreatic cancer data set of the National Cancer Database (NCDB) was studied for years 2011-15. Income groups were divided into top quartile (high income) or bottom three quartiles (non-high income) while racial groups were classified as Caucasian or non-Caucasian. Kaplan-Meier survival analysis and Cox proportional hazard models (CoxHR) were utilized. Analysis was controlled for established risk factors such as stage, grade, lymphovascular invasion, resection, and margin status. Results: Of the 164,631 patients meeting criteria, the average age was 68 ± 12 years, 51% were male, and 84% were Caucasian (the remaining patients were predominantly African American, 11%). Of patients with stage I or II PDAC who underwent resection, Caucasian patients had worse survival (CoxHR = 1.21, P<0.001) while high-income patients had better survival (CoxHR= 0.86, P<0.001). To investigate only differences due to income or race, a survival model of highly selected, low risk patients (stage I or II lymph node negative well-differentiated tumors without lymphovascular invasion and in patients who underwent margin negative resection who received chemotherapy) found that high-income predicted survival similar to race (CoxHR =0.89, P<0.001 vs. CoxHR = 1.12, P=0.005, respectively). On multivariable analysis, high-income was more impactful on survival (CoxHR = 0.83, P<0.001) than Caucasian race (CoxHR = 1.16, P <0.001). High-income, non-Caucasian patients had a median survival of 64 months while all other groups had a median survival of 40 months (P=0.005). Conclusions: Caucasian PDAC patients have worse survival compared to non-Caucasian patients after selecting for patients with favorable tumor biology who received adequate therapy. The data suggests that high-income is slightly more important than race in understanding disparate outcomes in PDAC.

Is rising BMI associated with an increased rate of clinically relevant pancreatic fistula after distal pancreatectomy for pancreatic adenocarcinoma?

360

Background: Clinically relevant pancreatic fistula (CR-POPF), following distal pancreatectomy (DP) remains a clinical challenge. Prior studies investigating the relationship between body mass index (BMI) and CR-POPF have yielded conflicting results. We examined this relationship utilizing our institutional database and hypothesized that BMI is associated with CR-POPF in patients having DP for pancreatic ductal adenocarcinoma (PDAC). Methods: Patients who underwent DP for PDAC at a single institution from 2007 to 2018 were retrospectively reviewed. A CR-POPF was defined as ISGPS grade B or C fistula. Uni- and multi-variable logistic regression analysis to assess factors associated with CR-POPF following DP was performed, controlling for factors such as gland texture, operative drain placement, gender, and smoking status. Results: 78 patients met the inclusion criteria. 51% were female, 51% were Caucasian, and the average age was 59 ± 15 years. The median BMI was 26 (interquartile range 24 to 29). Overall, 19% (n = 15) of patients had a CR-POPF. With a mean follow up 2.8 ± 2.5 years, the presence of a CR-POPF was not associated with long-term survival (P = 0.17). On univariable logistic regression, older age was associated with a decreased risk of CR-POPF (OR = 0.95, P = 0.015) while increasing BMI was associated with an increased risk of CR-POPF (OR = 1.1, P = 0.044). After controlling for multiple factors on multivariable logistic regression analysis, BMI (OR = 1.12, P = 0.035) was the only factor associated with development of a CR-POPF while older age (OR = 0.94, P < 0.001) was slightly protective of CR-POPF development. Conclusions: For patients undergoing DP for PDAC, increasing BMI is associated with an increased risk of CR-POPF, independent of other factors. These findings should be considered during preoperative counseling. Although there is no specific cut-off for the association between BMI and CR-POPF, efforts to diminish the risk of CR-POPF should be focused on patients with higher BMI based on this data.

Effect of macrophages on pancreatic cancer

324

Background: Pancreatic Ductal Adenocarcinoma (PDAC) is expected to be the second leading cause of cancer related deaths by 2030. TGF-β is a well-studied PDAC mediator with a context dependent role as initially a tumor suppressor with potential to convert to a tumor promoter in later stages. Tumor associated macrophages and interleukins, such as the pro-inflammatory interleukin, IL23, are not well studied regarding PDAC. We hypothesized PDAC treated with TGF-β and macrophages would induce a more aggressive phenotype. Methods: We investigated aggressive behavior with a primary PDAC cell line in vivo and a metastatic PDAC cell line in vitro. A primary pancreatic cell line, Panc-1 cells, were pre-treated with PBS, IL23, macrophages (10:1 ratio of Panc-1 cells to macrophages), IL23 + macrophages, TGF-ß, TGF-ß + macrophages, or TGF-ß + macrophages + IL23. After treatment, cells were orthotopically implanted into the pancreas of NOD SCID gamma mice with 5 mice per group. Mice weights were recorded twice weekly for 4-weeks. Primary lesions and metastasis were investigated with ANOVA. AsPC-1 cells, a metastatic pancreatic cell line, were pre-treated with the same seven treatments. We investigated pSTAT3 expression and the streak closure in vitro. Results: Panc-1 cells treated with macrophages had the largest pancreatic tumor weight and diameter compared to PBS control, IL23 alone, and TGF-β alone (P < 0.001). When macrophages treatment included TGF-β, pancreatic tumor weights and diameters decreased as compared to macrophages alone and macrophages + IL23 (P < 0.001). Macrophage treatment induced higher liver weights and higher number of surface liver metastatic lesions suggesting higher metastatic disease burden (P < 0.03). AsPC-1 cells treated with combinations of macrophages and TGF-β increased pSTAT3 expression compared to PBS control. AsPC-1 cells treated with macrophages closed the gap in the scratch assay faster than PBS control 24 hours after treatment (P < 0.001). Conclusions: We demonstrated macrophages have a key role in converting primary pancreatic cancer into a more aggressive phenotype in vivo whereas they have less effect on metastatic pancreatic cancer in vitro.

Survival outcomes for various treatment modalities in early-stage grade 3 follicular lymphoma (FL3): a National Cancer Database (NCDB) study

7551

Background: The prognosis, response to therapy and curability of FL3 is controversial. 5-year Overall Survival (OS) in the literature ranges from 35-72% (Ganti 2006). The aim of this study was to compare the OS for patients with early-stage FL3 managed with single- and multi-agent chemotherapy (CT) with and without radiotherapy (RT). Methods: We identified patients (pts) diagnosed with stage I & II FL3 between 2004 – 2012 from the NCDB and categorized into 3 groups based on therapy – pts given single agent CT with or without RT were combined due to small sample sizes (SA±RT), multi-agent CT without RT (MA-RT), and multi-agent CT with RT (MA+RT). We calculated OS for each group using Kaplan-Meier method and compared the results using Log Rank test. Cox regression model was used to identify factors which had significant impact on OS. Results: 1,563 pts were identified – 827 (53%) with stage I and 736 (47%) with stage II FL3. Median age was 61 yrs (range 18-90yrs); 750 (48%) males, 813 (52%) females; 1423 (91%) whites, 76 (5%) blacks. 112 (7%) received SA±RT, 886 (57%) MA-RT and 565 (37%) MA+RT. 5-year OS for MA+RT (95%) was significantly more than MA-RT (87%; HR 0.33, P<0.001) or SA±RT (88%; HR 0.38, P=0.007). Cox regression indicated that age (HR 1.05, P<0.001), sex (HR 0.66 for females, P=0.02), comorbidities (HR 1.60 for Charlson Deyo Score 1, P=0.04; HR 3.07 for Score 2, P=0.001), stage (HR 1.79, P=0.001), insurance status (HR 0.22 for insured, P<0.001) and increasing year of diagnosis (HR 0.92, P=0.03) also had significant impact on OS. Median radiation dose for the MA+RT was 36Gy (interquartile range 30.6 – 36Gy), and the proportion of patients who received greater than 36Gy decreased from 55% in 2004 to 38% in 2012 and at the same time, the proportion of patients who received intensity modulated RT increased from 5% in 2004 to 15% in 2012. Use of MA CT declined (2004 95% v 2012 89%, P=0.02) but there was no significant trend in use of RT (2004 39% v 2012 34%) during the periods studied. Conclusions: For pts with early-stage FL3, there was an association of improved survival with the use of MA+RT over other treatment strategies and appear to have outcomes superior to what has been previously reported.

Survival outcomes for various treatment modalities in advanced-stage grade 3 follicular lymphoma (FL3): A National Cancer Database (NCDB) study

7554

Background: The prognosis, response to therapy and curability of FL3 is controversial. 5-year Overall Survival (OS) in the literature ranges from 35-72% (Ganti 2006). The aim of this study was to compare the OS for patients with advanced-stage FL3 managed with various treatment modalities. Methods: We identified patients (pts) diagnosed with stage III & IV FL3 between 2004 – 2012 from the NCDB and categorized them into 3 groups based on therapy – pts given single agent chemotherapy with or without radiotherapy were combined due to small sample sizes (SA±RT), multi agent chemotherapy without radiotherapy (MA-RT), and multi agent chemotherapy with radiotherapy (MA+RT). We calculated OS using Kaplan-Meier method and compared the results using Log Rank test. Cox regression model was used to identify other factors which had significant impact on OS. Results: 2,808 pts were identified – 1,508 (54%) with stage III and 1,300 (46%) with stage IV disease. Median age was 60 yrs (range 21-90yrs); 1,331 (47%) males, 1,477 (53%) females; 2,559 (91%) whites, 142 (5%) blacks. 170 cases (6%) were treated with SA±RT, 2,508 (89%) with MA-RT and 130 (5%) with MA+RT. There was no significant difference in 5-year OS between MA-RT (83%) and MA+RT (82%; HR 1.07, P=0.76). There was no difference between SA±RT (73%) and MA+RT (82%; HR 0.62, P=0.069) likely due to small sample sizes, but survival for MA-RT (83%) was significantly higher than SA±RT (73%; HR 1.78, p<0.001). Cox regression indicated that age (HR 1.04, P<0.001), sex (HR 0.77 for females, P=0.008), comorbidities (HR 1.48 for Charlson Deyo Score 1, P=0.001; HR 2.59 for Score 2, P<0.001), stage (HR 1.29, P=0.007), insurance status (HR 0.65 for insured, P=0.048) and increasing year of diagnosis (HR 0.91, P<0.001) also had significant impact on OS. Use of MA chemotherapy declined (2004 96% v 2012 91%, P=0.008) but there was no significant trend in use of radiotherapy (2004 5% v 2012 3%) during the periods studied. Conclusions: MA chemotherapy in pts with advanced-stage FL3 was associated with improved survival compared to SA therapy, and radiation does not appear to influence outcomes. Outcomes were superior to what has been previously reported.

Prognostic significance of lymphadenectomy in patients with esophageal cancer receiving neoadjuvant chemoradiation

87

Background: The optimal number of lymph nodes that should be harvested in esophageal cancer patients remains to be defined, particularly in patients that receive neoadjuvant therapies. We investigated the impact of nodal resection and survival in esophageal cancer patients treated with neoadjuvant chemoradiation (NT).

Methods: Using our comprehensive esophageal cancer database we identified patients treated with NT followed by esophagectomy between 2000-2011. Clinical and pathologic data were compared using Fisher’s exact and chi-square while, Kaplan Meier estimates were used for survival analysis. Overall (OS) and disease-free survival (DFS) were compared with varying numbers of lymph nodes resected <10 and ≥10 (ST-1), <12 and ≥12 (ST-2), and <15 and ≥15 (ST-3). Multivariate analysis was analyzed by the Cox proportional hazard model.

Results: We identified 358 patients treated with NT and esophagectomy with a median follow-up of 18.5 months (range, 0-116 months). There was no survival benefit demonstrated for patients with increased lymph nodes removed during their surgery (ST-1 OS p=0.400, DFS p=0.8727; ST-2 OS p=0.6833, DFS p=0.6092; ST-3 OS p=0.1798, DFS p=0.4028). Patients were further stratified by pathologic response to NT and nodal harvest. There were no differences in OS or DFS in patients with increased nodal harvest when analyzed by complete (pCR) (ST-1 OS p=0.7278, DFS p=0.3602; ST-2 OS p=0.6182, DFS p=0.3592; ST-3 OS p=0.4489, DFS p=0.6976), partial (pPR) (ST-1 OS p=0.3762, DFS p=0.5061; ST-2 OS p=0.8036, DFS p=0.6497; ST-3 OS p=0.0890, DFS p=0.3364), or non response (pNR) (ST-1 OS p=0.6825, DFS p=0.7161; ST-2 OS p=0.7084, DFS p=0.8351; ST-3 OS p=0.5002, DFS p=0.7314) to NT. Multivariate analysis demonstrated that age (p=0.028), t-stage (p=0.006), pPR (p=0.025), and pNR (p<0.0005) to NT were all independent predictors of mortality.

Conclusions: In our experience, the number of lymph nodes resected was not predictive for overall or disease free survival in esophageal cancer patients treated with NT. In addition, extended lymph node resection did not improve survival for those with residual disease.

Effect of lymph node harvest for squamous cell cancer of the esophagus on survival

102

Background: The optimal number of lymph nodes harvested remains controversial in patients with esophageal cancer. Pathologic response to neoadjuvant therapy (NT) has demonstrated improved survival. However, little is known regarding the impact of NT or nodal harvest in patients with squamous cell carcinoma (SCC) of the esophagus. We examined the extent of LN harvest and outcome in patients who underwent esophagectomy for SCC.

Methods: After IRB approval, using a comprehensive esophageal cancer database we identified patients who underwent esophagectomy between 1994-2011. Clinical and pathologic data were compared using Fisher’s exact and chi-square when appropriate while Kaplan-Meier estimates were utilized for survival analysis. Nodal strata were set at 12 (ST-1), 15 (ST-2), and 20 nodes (ST-3). Pathologic response to NT was defined as complete (pCR), partial (pPR), or non-response (pNR).

Results: We identified 76 patients who underwent esophagectomy for SCC between 1994-2011. The median age was 62.5 years (40-85 months) with median follow up of 18.5 months (1-157 months). 48 (63%) were male and 28 (37%) were female. Twenty-eight patients (37%) underwent primary esophagectomy alone (PE) while 48 (63%) patients were treated with NT. Extent of lymphadenectomy had no significant impact on overall survival (OS) or disease free survival (DFS) for the entire cohort ST-1 p=0.8 and p=0.9, ST-2 p=0.5 and p=0.4, and ST-3 p=0.5 and 0.4, respectively. Among the patients who received NT, pCR was observed in 28 (58%), pPR in 14 (29)%, and pNR in 6 (13)%. When examining the degree of pathologic response to treatment, extent of LN harvest had no significant impact on OS or DFS for patients who underwent esophagectomy after NT (p=ns across all strata).

Conclusions: The extent of LN harvest failed to demonstrate an overall or disease free survival benefit in patients with squamous cell carcinoma of the esophagus. Moreover, patients treated with NT also did not benefit from increased nodal resection irrespective of their pathologic response.