Borad, Mitesh J.

Abstract Biliary tract cancer (BTC) remains among the most challenging malignancies with a poor prognosis and limited treatment options, particularly in pretreated patients. As most patients experience disease progression after first-line treatment, effective second-line and subsequent treatments are required. Although the addition of modified FOLFOX (fluorouracil, leucovorin, and oxaliplatin) to active symptom control improved the overall survival of patients with progressing advanced BTC despite gemcitabine plus cisplatin treatment, its efficacy was modest. Moreover, most clinical trials demonstrated modest efficacy of molecular-targeted agents for molecularly unselected pretreated advanced BTC. Patients with advanced BTC carry a relatively high druggable genetic alteration rate and have shown promising responses to molecular-matched therapies targeting gene alterations such as FGFR2 fusions/rearrangements, IDH1 mutation, and HER2 overexpression/amplification. Additionally, tumor-agnostic approaches, including BRAF V600E, NTRK fusion, and RET fusion, have expanded the treatment options for some patients. Immune checkpoint inhibitors have shown limited efficacy as mono- or combination therapy in patients with pretreated advanced BTC. Therefore, developmental efforts have shifted to immune checkpoint inhibitor and other combinations such as vascular endothelial growth factor receptor inhibitors or radiation. In addition to refining combination strategies to enhance the therapeutic potential of immune checkpoint inhibitor, future research should focus on elucidating the tumor microenvironment. This review delineates the evolution of systemic therapies in patients with pretreated advanced BTC. By examining past developments and recent advances through prospective trials, it highlights novel approaches that may improve outcomes in this challenging disease.

Abstract Purpose: Monitoring disease progression in patients with high-grade gliomas (HGG) is challenging due to treatment-related changes in imaging and the requirement for neurosurgical intervention to obtain diagnostic tissue. DNA junctions in HGG often amplify oncogenes, making these DNA fragments potentially more abundant in blood than monoallelic mutations. In this study, we piloted a cell-free DNA approach for disease detection in the plasma of patients with HGG by leveraging patient-specific DNA junctions associated with oncogene amplifications. Experimental Design: Whole-genome sequencing of grade 3 or 4 isocitrate dehydrogenase–mutant or wild-type astrocytomas was utilized to identify amplified junctions. Individualized qPCR assays were developed using patient-specific primers designed for the amplified junction. ctDNA levels containing these junctions were measured in patient plasma samples. Results: Unique amplified junctions were evaluated by individualized semi-qPCR assays in presurgical plasma of 18 patients, 15 with tumor-associated focal amplifications and three without tumor-associated focal amplifications. high copy-number junctions were robustly detected in the plasma of 14 of 15 (93.3%) patients with amplified junctions and none of the controls. Changes in junction abundance correlated with disease trajectory in serial plasma samples from five patients, including increased abundance of amplified junctions preceding radiographic disease progression. Conclusions: In patients with grade 3 or 4 astrocytomas who had tumor-associated amplifications, patient-specific amplified junctions were successfully detected in assayed plasma from most patients. Longitudinal analysis of plasma samples correlated with disease trajectory, including cytoreduction and progression.

243 Background: HER2-amplified aCRC is associated with poor prognosis despite new anti-HER2 regimens. Insight on biomarkers that predict response to these therapies is limited. This study aims to evaluate whether baseline alterations in oncogenic drivers along the HER2 pathway predict response to anti-HER2 therapy in HER2-amplified aCRC. Methods: In this international multi-center study involving Mayo Clinic, Duke Cancer Institute and National Cancer Center Hospital East (Japan), we collected baseline genomic alterations from tissue or blood samples using next-generation sequencing (NGS) prior to anti-HER2 therapy initiation. We defined PIK3CA, RAS, RAF, MAP2K, and MET as oncogenic drivers of resistance within the HER2 pathway. Overall survival (OS) was evaluated with the Kaplan-Meier method and a log-rank test was used to compare median OS (mOS) between pts with and without alterations in the above genes. A chi-square test (significance level 0.05) was performed to compare overall response rate (ORR) and disease control rate (DCR). Results: We identified 39 pts with HER2-amplified aCRC and baseline NGS. Median age was 57 (range 30-77), 54% were male. Baseline alterations in HER2 pathway oncogenic drivers were identified in 17 pts (43.5%). 18 (46%) received monoclonal antibodies (mAbs) plus tyrosine kinase inhibitors, 12 (30%) received dual mAbs and 7 (18%) received trastuzumab deruxtecan. The mOS was 41.9 months (mos) (95%CI 23.7-60.1) for other alterations vs 24.2 mos (95%CI 14.1-34.3) in pts with HER2 pathway alterations (p = 0.1). ORR was 47% vs 41% (p = 0.7) and DCR 82% vs 73% (p = 0.4), for pts with HER2 pathway alterations and without HER2 pathway alterations, respectively. 4 pts receiving trastuzumab/tucatinib had complete response, none of which had HER2 pathway alterations. Conclusions: Pts with baseline alterations in oncogenic drivers of the HER2 pathway had clinically worse outcomes compared to pts without such alterations despite receiving HER2-targeted therapy, though results lacked statistical significance. Larger studies are needed to confirm the lack of benefit from anti-HER2 therapy in this subgroup.

451 Background: Perioperative chemotherapy with surgery is the standard of care for patients (pts) with resectable gastric or gastroesophageal junction adenocarcinoma (GEA). However, recent studies have shown more favorable responses after neoadjuvant immunotherapy (NIO) in patients with mismatch repair deficient (dMMR) cancers. Herein, we present the outcomes of 10 cases with dMMR GEA. Methods: We retrospectively reviewed patients with GEA at Mayo Clinic Arizona who met the following criteria: have dMMR disease by IHC, nonmetastatic operable disease, and received NIO +/- chemotherapy. Patient demographics, disease characteristics, and clinical outcomes were extracted. The primary endpoint was clinical complete response (cCR) in patients treated with non-operative management (NOM) defined as complete radiological response and no residual disease on endoscopic ultrasound (EUS) and biopsy. Results: 10 pts with dMMR GEA with a median age of 70 (43-85) were identified. 6 pts had T2N0, 1 had T4N0, 3 had N+ disease. 8 pts received NIO monotherapy upfront, 1 received NIO after progressing on chemotherapy, and 1 concurrently with chemotherapy. No grade 3/4 toxicities were reported, however, 2 pts discontinued therapy due to arthralgia and colitis. After NIO, 9/10 pts had significant or complete resolution of FGD avidity on PET/CT, while 1/10 had stable disease. 3/10 pts underwent surgical resection, with 2 achieving pathological complete response (pCR) and the third showing T1N0 disease with no recurrence 1 year after surgery. Notably, one of the pts with pCR passed away due to surgical complications. 7/10 patients had NOM, of whom, 4 had cCR and 3 had radiological response only, as no EUS data was available. 4/7 NOM pts had follow up data at 6 months with no evidence of progression, of whom one pt remained progression-free at 33 months. ctDNA testing was performed in 6/10 pts: 2 had undetectable ctDNA at baseline, while 3 became MRD-negative after NIO, consistent with radiological and/or pathological response, and 1 had persistent ctDNA both before and after NIO, despite cCR. Conclusions: This study demonstrates exceptional and prolonged response, along with good tolerability to NIO in pts with dMMR GEA, supporting the feasibility of non-operative management in such patients.

634 Background: CAT is the second leading cause of cancer-related death, yet the genetic underpinnings of VTE risk in BTC remain underexplored. The evaluation of CAT risk is based on Khorana score (KS), which is limited to clinical parameters. Incorporation of a genomic signature in CAT risk assessment may allow for individualized risk assessment. Methods: We collected data from consecutive patients (pts) with BTC treated at Mayo Clinic Hospitals (Phoenix, Rochester, Jacksonville) between 1998-2024. Genomic profiling was conducted on FFPEs and blood samples through next-generation sequencing (NGS) (NextSeq, Tempus xT CDx, FoundationOne CDx). Association between CAT and clinic-pathologic features (age, gender, race, ethnicity, BMI, tumor site (TS), surgery, stage, radiotherapy, systemic therapy, history of thromboembolism (hVTE)) were assessed with multivariate Cox regression (p<0.05). A separate Cox regression adjusting for age, hVTE, radiotherapy, systemic therapy, metastatic disease stratified for TS was used for gene associations. Benjamini–Hochberg method was applied (q<0.1) based on previous publications. Median overall survival (mOS) was calculated from the date of diagnosis to the date of death or last follow-up. Results: 740 adult pts were included, median age 62 years (IQR: 56-71), male/female (51%/49%), median BMI 26.6 (IQR: 22.8-31.1) with intrahepatic (496, 67%) and extrahepatic BTC (173, 23%), gallbladder cancer (71, 10%). 244 (33%) had metastatic disease. 216 (29%) experienced CAT. For those with CAT, 31 (14%) scored 1, 9 (4%) scored 2 and 1 (<1%) scored 3 on KS. mOS was 33 months (mos) (95% CI 28.9-38.4) vs. 55 mos (95% CI 25.1-39.1) (p=0.2) for pts with and without CAT, respectively. hVTE was associated with CAT (HR 3.14, 95% CI 1.98-4.98, p<0.001) along with systemic therapy (HR 1.19, 95% CI 1.11-2.89, p<0.01) and metastatic disease (HR 1.65, 95% CI 1.11-2.44, p<0.01). 305 pts (41%) underwent NGS. The most common altered genes were TP53 (126, 41%), KRAS (64, 21%), CDKN2A (63, 21%). The most common altered currently targetable genes were FGFR2 (43, 14%), IDH1 (30, 10%) and HER2/ERBB2 (16, 5%). MYC (4, 1%) (HR 11.60, 95% CI 3.04-44.25, p<0.001, q=0.058) and HER2/ERBB2 (HR 2.45, 95% CI 1.26-4.77, p=0.008, q=0.087) alterations were significantly associated with an increased risk of CAT. MYC alterations were all copy number gains (100%). The most frequent HER2/ERBB2 alterations were copy number gains (10, 62%). FGFR2 fusions (31, 10%) were linked to a lower CAT risk (HR 0.46, 95%CI 0.22-0.93, p=0.032, q=0.075) with FGFR2-BICC1 being the most frequent (5,16%). Conclusions: MYC and HER2/ERBB2 alterations are novel predictive biomarkers for an increased risk of CAT in BTC, while FGFR2 fusions confer lower risk. The knowledge of mutational status lends insight into guidance for anticoagulant prophylaxis management for this subgroup and integration with KS.

628 Background: Biliary tract cancers (BTC) are aggressive malignancies with poor survival outcomes and limited treatment options. This study compared the epidemiological and molecular differences of advanced BTC patients through a global collaboration involving the Mayo Clinic and National Cancer Center Hospital East. Methods: We included patients with advanced BTC who underwent comprehensive tumor molecular profiling who received at least 30 days of systemic therapy. Patients were grouped based on specific alterations. Overall survival (OS) was calculated from the start of first-line therapy until death and compared among subgroups using Cox regression models. Results: 332 (36%) out of 932 patients had at least one actionable mutation : somatic BRCA1/2 mutations N=106, FGFR2 fusions N=61, HER2 amplifications N=65, KRAS G12C/D mutations N=46, IDH1 mutations N=38, and TMB >10 mut/Mb N=16. BRCA1/2 -mutated BTC were more commonly found in older Asian males with intrahepatic tumors and often presented with peritoneal and liver metastases upon initial diagnosis. FGFR2 fusions were more prevalent in younger White females with intrahepatic tumors, while HER2 amplifications were mainly seen in Asian females with gallbladder tumors. KRAS mutations were primarily observed in Asian males, and IDH1 mutations were common in White males with intrahepatic tumors, frequently presenting with liver metastasis. Survival outcomes varied significantly based on the alteration type: patients with FGFR2 fusions had the longest median OS (mOS) at 23.3 months (mo) (95%CI 18.4-34.9), followed by those with IDH1 mutations (19.3 mo; 95% CI 17.4-NE) and TMB >10 mut/Mb (17.8 mo; 95% CI 13.6-NE). BRCA mutations had a mOS of 16 mo (95% CI 14.6-19.8), while HER2 amplifications and KRAS mutations had mOS of 15.5 (95% CI 11.9-20.2) and 11.6 mo (95% CI 10.0-20.9), respectively. Conclusions: These findings underscore the heterogeneity in clinical and genomic characteristics among BTC patients, highlighting the importance of tailored therapeutic approaches. Baseline patient characteristics stratified by mutation status. KRAS G12C/D (N=46) BRCA 1/2 (N=106) IDH1 (N=38) HER2 amplification(N=65) FGFR2 fusion(N=61) TMB>10 mut/Mb(N=16) P-value Median Age, year 67 68 64 66 52 73 <0.001 Female, n (%) 19 (41) 34 (32) 18(47) 34 (52) 45(74) 6 (38) <0.001 Race, n (%) <0.001 Asian 31 (67) 97 (92) 13 (34) 51 (79) 10 (16) 6 (38) White 14 (30) 8 (8) 24 (63) 13 (20) 48 (79) 10 (63) Primary tumor location, n (%) <0.001 Intrahepatic 28 (68) 45 (45) 33 (100) 19 (31) 54(95) 10 (67) Extrahepatic 9 (22) 27(27) 0 (0) 10 (16) 2 (4) 4 (27) Gallbladder 4 (10) 28 (28) 0 (0) 32 (53) 1 (2) 1 (7) Sites of Metastasis at diagnosis, n (%) Abdominal Wall / Peritoneal 9 (24) 15 (19) 4(13) 7 (12) 6 (10) 4 (27) 0.309 Bone 33 (92) 70 (90) 26 (87) 2 (3) 9(15) 0 (0) 0.212 Liver 20 (49) 52 (61) 28 (82) 45 (73) 45 (76) 6 (38) 0.001 Lung 26 (70) 65 (82) 22 (69) 36 (58) 27(45) 11 (69) 0.006

556 Background: Combined hepatocellular carcinoma-cholangiocarcinoma (cHCC-CCA) is a rare and complex liver cancer with features of both hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA), making its management challenging. This study aims to analyze the clinical characteristics, genomic profiles, and treatment outcomes of cHCC-CCA patients in a real-world setting. Methods: We conducted a retrospective analysis of cHCC-CCA patients across three first-line treatment groups: chemotherapy alone(Gemcitabine/Cisplatin, FOLFOX), HCC-directed therapies (atezolizumab/bevacizumab, lenvatinib, nivolumab, durvalumab, sorafenib), and chemoimmunotherapy (Gemcitabine/Cisplatin combined with immunotherapy). This international study, involving Mayo Clinic and the National Cancer Center East (Japan), assessed overall survival (OS) using Kaplan-Meier and Cox regression models. Results: A total of 68 patients were included, with a mean age of 66 years; 74% were male and 66% were White Non-Hispanics. Cirrhosis was present in 74% of patients at diagnosis, and Hepatitis B and C were common etiologies in 57%. Prior primary resection and regional therapies (Transarterial chemoembolization (TACE), Transarterial radioembolization (TARE)) were recorded in 39% and 50% of patients, respectively. Advanced/metastatic disease was present in 52% of the cohort. The distribution of first-line treatments was as follows: 47% received chemotherapy alone, 39% received HCC-directed therapy, and 14% received chemoimmunotherapy. Genomic alterations were detected in 44% patients, with TP53 (28%), ARID1A (25%), and TERT (22%) being the most frequent. The HCC-directed therapy group demonstrated the best overall response rate (52%) and had a median OS of 15.8 months (mo), compared to 11.8 mo for chemotherapy and 4.7 mo for chemoimmunotherapy (p=0.8). Patients who received regional treatments had a longer median OS of 15.8 mo compared to those who did not (10.2 mo) (p=0.4). Conclusions: HCC-directed therapy showed a trend towards better survival outcomes compared to other treatments, though not statistically significant. A considerable proportion of patients had genomic alterations, which may have influenced survival outcomes. Ongoing research with larger cohorts is underway to further elucidate the impact of genomic variations on treatment efficacy and refine therapeutic strategies for this challenging malignancy. Baseline characteristics of patients with combined hepatocellular carcinoma-cholangiocarcinoma (cHCC-CCA). Total(N=68) Mean age at diagnosis, years (SD) 66(6.4) Gender, n (%) Male 50 (74%) Ethnicity, n (%) White, Non-Hispanic 45 (66%) Cirrhosis at diagnosis, n (%) 50 (74%) Etiology of Liver disease, n (%) Viral (Hepatitis B and C) 39 (57%) History of Primary resection, n (%) 27 (39%) Regional treatment, n (%) 34 (50%) Afp, n (%) < 200 23 (64%) ca19_9, n (%) < 50 27 (82%)

277 Background: While HER2-targeted therapies benefit patients (pts) with advanced HER2-positive mCRC, the development of resistance remains inevitable. This study aims to explore the landscape and frequency of genetic alterations putatively associated with acquired resistance to anti-HER2 therapy in HER2-positive mCRC. Methods: In this retrospective-multicenter international study involving Mayo Clinic, Duke Cancer Center and National Cancer Center Hospital East (Japan), we descriptively compared comprehensive genomic analyses from tissue or blood samples of pts with HER2-positive CRC pre-treatment and post-treatment with HER2-targeting agents. Results: A total of 36 pts were included in this study. 29 (80%) had HER2-positive mCRC as confirmed by immunohistochemistry (IHC) and fluorescent in situ hybridization (ISH), the remaining pts had HER2 amplification detected on blood or tissue . 34/36 (94%) pts were administered their first anti-HER2 agent after progressing on ≥2 prior lines of treatment. The most common anti-HER2 regimen administered was trastuzumab/pertuzumab (50%, 18/36), followed by trastuzumab-tyrosine kinase inhibitors (TKIs) (tucatinib or neratinib) (30%, 11/36) and trastuzumab deruxtecan (14%, 5/36). Putative genomic mechanisms of acquired resistance were detected in 72% (26/36), while the remaining pts lacked new genomic alterations that might explain resistance to anti-HER2 therapy, suggesting alternative escape mechanisms. Acquired alterations in downstream or alternate bypass pathways were identified in 72% (26/36) of pts; most commonly in the RAS/RAF pathway and in EGFR (34.6%, 9/26 each), followed by 27% (7/26) of pts with MET point mutations (mts) and amplification; activating muts in PI3K/AKT and in 27% (7/26), most commonly CDK12 (60%, 4/7) . HER2 receptor alterations were detected in 3 pts (8.3%), all of whom received trastuzumab-tucatinib and included HER2 L755S, HER2 V777L, HER2 D769Y, HER2 G727A and HER2 S310F, which are known to cause resistance to , however, information about tucatinib resistance is not yet available. Other altered pathways included AR, RB, NOTCH and HRD. Two pts lost HER2 expression on blood and tissue NGS post anti-HER2 therapy. Conclusions: The emergence of resistance might be in part related to acquired alterations that constitutively activate signaling pathways parallel or downstream of HER2, bypassing HER2 inhibition. Additional histologic studies, deep mutational scanning, larger populations and correlative analysis are needed to validate these findings.

536 Background: The prognosis of advanced BTC remains poor. Although targeted agents have expanded late-line options, many patients (pts) become ineligible for these therapies after progressing on first-line treatment. Herein, we examined the real-world impact of targeted therapy on survival in advanced BTC. Methods: In this international collaborative study between the Mayo Clinic (US) and the National Cancer Center East (Japan), we included pts with advanced BTC with comprehensive tumor molecular profiling who had received at least 30 days of systemic therapy. Pts were categorized into 3 groups: 1) non-actionable, 2) pts with alterations who received matched targeted therapy at any point during their treatment (matched group), and 3) those with mutations who did not receive matched therapy (unmatched group). OS was calculated from the start date of first-line therapy until death and compared among three subgroups using Cox regression models. Results: Of 932 pts, 366 (39.3%) had an actionable alteration, and 135 of these received matched therapy. 566 (61.7%) did not have an actionable alteration (Table). The most common alterations were somatic BRCA1/2 (N=106, 11.4%), HER2 amplification (N=65, 7%), FGFR2 fusion (N=61, 6.5%), KRAS G12C/D (N=46, 4.9%), and IDH1 (N=38, 4.1%). 78.7% of FGFR2 fusions were found in Caucasian pts, whereas BRCA1/2 (97.5%), HER2 (78.5%), and KRAS G12C/D (67.4%) alterations were more prevalent in Asian pts. The median OS was significantly longer in the matched group (21.4 months; 95% CI 18.4-27.9) compared to the unmatched group (14.6 months; 95% CI 12.8-17.6) and the non-actionable group (17.2 months; 95% CI 15.5-19.0). The 36-month OS rate was 33% in the matched group vs. 7% in the unmatched group (p < 0.0001). After adjusting for age, gender, prior surgical resection, primary tumor location, and country of origin, matched targeted therapy remained a strong independent predictor for improved OS (HR 0.61; 95% CI 0.48-0.79). Conclusions: These real-world findings suggest that matched targeted therapies significantly improve survival in advanced BTC, underscoring the need to incorporate them earlier in the treatment course and address barriers to treatment access. The regional prevalence of actionable alterations should also be considered when developing new targeted therapies. Patient baseline characteristics stratified by actionable alterations and targeted therapy. Non-actionable(N=566) Unmatched group(N=231) Matched group(N=135) Site, n (%) Japan 409 (72%) 166 (72%) 50 (37%) US 157 (28%) 65 (28%) 85 (63%) Age at Dx > 65, n (%) 324 (57%) 122 (53%) 54 (40%) Female, n (%) 235 (42%) 92 (40%) 77 (57%) Primary tumor location (category), n (%) Intrahepatic 217 (43%) 115 (55%) 99 (76%) Extrahepatic 150 (30%) 39 (19%) 17 (13%) Gallbladder 142 (28%) 54 (26%) 14 (11%) Prior surgical resection, n (%) 220 (42%) 74 (34%) 48 (37%)

532 Background: HCC accounts for 80% of primary liver cancers, and worldwide, is the third common cause of cancer-related death. Two FDA approved 1st line systemic treatment options include atezo/bev and durva ± treme. Herein we report outcomes of these regimens in the 1st line setting via a real-world, multi-institutional retrospective cohort study. Methods: We reviewed clinical outcomes of patients (pts) who initiated 1st line atezo/bev or durva ± treme for unresectable HCC between March 2017 and February 2024, across 5 institutions (Mayo Clinic, University of Alabama, Cedars-Sinai Medical Center, University Hospitals Seidman Cancer Center and University of Michigan). Time to treatment failure (TTF) and overall survival (OS) were analyzed using the Kaplan-Meier method. Best (objective) response was tabulated. Multivariable analyses were performed using Cox models and logistic regression for time-to-event and binary outcomes, respectively, while adjusting for age, sex, race, HCC etiology, Child-Pugh class, and ALBI grade. Results: A total of 433 HCC pts were included; atezo/bev (336), durva ± treme (97). Median age at the start of therapy was 68 years; 77% were male, 82% were Caucasian. Etiologies of HCC were viral hepatitis (38%), MASLD (20%), alcohol (10%), non-viral multifactorial (10%), or unknown (22%). Pts receiving durva ± treme were less likely to have cirrhosis (11% vs. 28%, p<0.01) but similar in terms of macrovascular invasion (63% vs. 58%, p=0.35) compared to pts receiving atezo/bev. Reasons for treatment discontinuation were similar between the two groups, with the most common being disease progression - atezo/bev (56%) and durva ± treme (47%). The differences in clinical outcomes were not statistically significantly and remained consistent after multivariable adjustment. (Table, atezo/bev as the reference group). Outcomes were significantly different between Child-Pugh classes (CP: A,B,C) respectively, OS: 19.0 m, 6.4 m, 1.9 m (p<0.001); TTF: 6.1 m, 2.7 m, 1.3 m (p<0.001). Conclusions: In this real-world retrospective cohort study of unresectable HCC, the differences in overall survival, objective response, and time to treatment failure were not statistically significant between atezo/bev and durva ± treme in the 1st line setting. Outcome Atezo/bev Durva±treme Univariate p-value Multivariable adjusted hazard ratio (95% CI) Multivariable adjusted odds ratio (95% CI) Overall Survival, Median (month) 14.0 14.6 0.77 0.78 (0.54-1.14) --- Time to Treatment Failure, Median (month) 4.9 4.1 0.71 0.95 (0.72-1.26) --- Objective response, % 26.3% 21.1% 0.31 --- 0.78 (0.43-1.39)