537 Background: Biomarkers do not yet exist to triage patients with oligometastatic ccRCC for de-escalation strategies. Two promising biomarkers are kidney injury molecule-1 (KIM-1) and circulating tumor DNA (ctDNA). Methods: Patients with ccRCC and ≤5 metastases were enrolled on a single-arm phase 2 trial of metastasis-directed therapy (MDT) without systemic therapy (NCT03575611), powered for the primary endpoint of systemic therapy–free survival (STFS). Plasma samples for ctDNA and KIM-1 measurement were collected at baseline and 3-month follow-up. Personalized ctDNA detection panels (≤2000 somatic variants) were constructed from tumor whole-genome sequencing (Myriad Genetics). To create the Kidney Cancer OligoMetastasis Prognostic Assessment Systemic Score (K-COMPASS), we screened 24 candidate variables by elastic net-penalized Cox regression with 5-fold cross-validation repeated 100 times, with performance assessed by C-index. Predictors with nonzero coefficients were refit with Weibull regression for STFS. Results: Among 112 patients with KIM-1 measurements available, median baseline KIM-1 level was 127.0 pg/mL (IQR 71.5–228.0). KIM-1 and ctDNA were associated with STFS at baseline and 3 months in univariable and multivariable analyses (table). Progression-free survival (PFS) and overall survival (OS) were also associated with KIM-1 at baseline (PFS: HR=2.2, 95% CI 1.5–3.3; OS: HR=5.1, 95% CI 2.5–10.2, P <0.001) and at 3 months (PFS: HR=3.5, 95% CI 2.2–5.5; OS: HR=5.0, 95% CI 2.3–10.9) (all P <0.001). K-COMPASS construction selected baseline KIM-1, baseline ctDNA MRD, and 4 clinical variables (prior systemic therapy lines, ECOG performance status, number of metastatic lesions, time from diagnosis to metastasis). The complete model showed strong discrimination (C-index=0.76) and excellent calibration (slope=1). A user-facing K-COMPASS tool is available online (www.trialdesign.org). Conclusions: To our knowledge, this is the first evaluation of KIM-1 in oligometastatic ccRCC and the first to analyze KIM-1 and ctDNA in tandem for RCC. Both biomarkers were strongly and independently associated with outcomes in patients receiving MDT without systemic therapy. Integrating KIM-1 and ctDNA with clinical factors (K-COMPASS) yielded a well-calibrated model with high discrimination to support risk-adapted decision-making. External validation is planned. Univariable and multivariable association of baseline KIM-1 and ctDNA MRD status with systemic therapy free survival. HR (95% CI) Baseline 3-Month Follow-Up Univariable Multivariable Univariable Multivariable KIM-1(per log10[pg/mL]) 2.5 (1.5–4.1) P <0.001 1.9 (1.0–3.6) P =0.041 3.2 (1.9–5.4) P< 0.001 2.2 (1.1–4.5) P =0.028 ctDNA (MRD+ vs MRD–) 2.8 (1.3–5.9) P =0.0089 2.5 (1.1–5.6) P =0.032 4.4 (2.1–9.5) P< 0.001 2.7 (1.1–6.7) P =0.032

509 Background: Venous tumor thrombus (VTT) in patients with renal cell carcinoma (RCC) represents an aggressive disease feature associated with poor prognosis and management challenges. The efficacy of immunotherapy-based combinations in this subgroup remains insufficiently characterized. Methods: The PELET trial is an ambispective study enrolling adult patients with histologically confirmed advanced RCC and unresected VTT. Patients received lenvatinib 20 mg orally once daily in combination with pembrolizumab 200 mg intravenously every 3 weeks. The primary endpoint was a ≥30% reduction in thrombus size. Secondary endpoints included changes in thrombus level, progression-free survival (PFS), and safety. Results: Twenty patients were included, predominantly male (85%), with a median age of 58.5 years. Most had clear cell histology (90%) and no sarcomatoid features. The median baseline thrombus size was 6.55 cm, which decreased to 3.0 cm after at least 12 weeks of therapy (p < 0.001). A ≥30% reduction in thrombus size was achieved in 45% of patients, meeting the primary endpoint. Moreover, 35% experienced a reduction in thrombus length > 50%, and 45% demonstrated a decrease in thrombus level by at least one level. Thrombus response was not correlated with baseline thrombus size. Median PFS was not reached at the time of analysis. Grade ≥3 adverse events occurred in 15% of patients. Conclusions: Twenty patients were included, predominantly male (85%), with a median age of 58.5 years. Most had clear cell histology (90%) and no sarcomatoid features. The median baseline thrombus size was 6.55 cm, which decreased to 3.0 cm after at least 12 weeks of therapy (p < 0.001). A ≥30% reduction in thrombus size was achieved in 45% of patients, meeting the primary endpoint. Moreover, 35% experienced a reduction in thrombus length > 50%, and 45% demonstrated a decrease in thrombus level by at least one level. Thrombus response was not correlated with baseline thrombus size. Median PFS was not reached at the time of analysis. Grade ≥3 adverse events occurred in 15% of patients. Clinical trial information: KCRB-092025-01.

96 Background: Preventing metastatic disease is the key goal of prostate cancer screening and local treatment. However, most evaluations of metastasis burden focus only on de novo metastatic disease at diagnosis and do not capture metastatic recurrences after local therapy, such as those based on cancer registries in the United States (U.S.). Methods: We inferred the full burden of metastatic prostate cancer in the Health Professionals Follow-up Study, a prospective cohort of U.S. male health professionals aged 40–75 years at enrollment in 1986 with follow-up in this analysis through January 2018. We captured three ways how metastatic prostate cancer manifested: (1) de novo metastases at initial diagnosis, (2) metastatic recurrences reported by participants during active follow-up, and (3) metastases were inferred from deaths attributed to prostate cancer, under the assumption that such deaths are almost invariably preceded by distant metastases. Cause-of-death adjudication assumed 90% specificity of coding, with a conservative 70% applied in sensitivity analyses. Absolute risk (cumulative incidence) of metastatic prostate cancer was estimated using the Aalen–Johansen method, accounting for death from other causes as a competing risk. Results: Our study population of 50,420 men had a median age of 54 years at enrollment (interquartile range, 46 to 63 years) when they were free from prostate cancer. The study population was predominantly of European descent (95%), with 2% Asian and 1% African American participants. Over up to 30 years of follow-up, we documented 8,305 prostate cancer diagnoses. Of these, 1,504 cases were adjudicated to be metastatic at diagnosis or to have metastasized after initial diagnosis. De novo metastatic disease accounted for only 249 (17%) of all metastatic cases. Conservative causes-of-death adjudication estimated 1,314 metastatic cases, with 19% being de novo metastases. The risk of metastatic prostate cancer among men was 0.8% (95% confidence interval [CI] 0.8, 0.9) by age 70, 2.4% by age 80 (95% CI 2.3, 2.6), and 4.0% (95% CI 3.8, 4.2) by age 90 years. Four in ten metastatic prostate cancers occurred after age 80 years. De novo metastatic disease risk represented about one-quarter of metastases by age 70, with lower relative contributions with advancing age. The risk of metastatic disease was particularly high in African-American men (3.6% by age 80, 95% CI 2.2, 5.8), while the risk among Asian men (2.3% by age 80, 95% CI 1.4, 3.7) was comparable to that among European men (2.5% by age 80, 95% CI 2.3, 2.6). Conclusions: One in 25 men in a widely PSA-screened population developed metastatic prostate cancer over the lifetime. Much of this substantial disease burden would remain hidden if using de novo disease as the only measure. Clinical and registry-based evaluations of early detection and local treatment need to capture the full burden of metastatic disease, which disproportionally impacts the oldest age groups.

LBA639 Background: PPARG is a master regulator of luminal lineage in UC with two-thirds of adv tumors classified as luminal. FX-909 is a first-in-class oral small molecule that potently and selectively inhibits both ligand-mediated and basal PPARG activity, offering a novel approach to targeting key cancer cell-intrinsic biology. Preliminary phase 1A data have demonstrated objective responses with FX-909 monotherapy (Gao, AACR Targets 2025). Here we report updated safety, tolerability, antitumor activity and predictive biomarker discovery to identify pts most likely to benefit from FX-909. Methods: 56 pts enrolled in a phase 1A open-label dose escalation study (30-100 mg PO QD, 28-day cycles), including additional 10 adv UC patients in 30 mg PO and 50 mg PO QD backfill cohorts. Baseline archival tissue (<30 months old) or a fresh biopsy collected for PPARG IHC (SP500 clone) and NGS correlative biomarker analysis. A provisional PPARG TPS cutoff was determined using linear regression modeling leveraging molecular real-world data (RWD) (N=2609 adv UC pts, Tempus xT). Serial blood samples collected for ctDNA analysis. Results: As of Nov 10, 2025, 46 pts with adv UC have been treated across four dose levels; 30 mg (17); 50 mg (16); 70 mg (11); and 100 mg (2). Median age was 71 (range 44-86), 71.7% ECOG PS 1, and 100% had mUC. Median lines of prior therapy was 3 (range, 1-8), including prior EV and anti-PD(L)1 treatment in 69.6%. Concordance between PPARG mRNA and TPS (r = 0.88, p<0.001) supported linear regression modeling, which inferred a provisional TPS cutoff of ≥60% (PPARG high ). Among 40 efficacy-evaluable adv UC pts, 35 had PPARG IHC results. 18/26 PPARG high pts showed tumor regressions, including 5 confirmed and 1 unconfirmed partial response. Decreases in ctDNA VAF occurred in 7/8 PPARG high pts with available results at cycle 2 (3 PR, 3 SD). Additional data demonstrating the relationship between PPARG high status, luminal subtype, and specific genomic alterations (PPARG(CN≥3), RXRA S427F and FGFR3) will be presented. At 30mg and 50mg doses, the most common ≥Gr3 TRAEs were anemia (18.9%), thrombocytopenia (16.2%) and fatigue (10.8%). Other common TEAEs were diarrhea (32.4%), hypertriglyceridemia (27%) and hyperglycemia (24.3%). The 30 mg dose was associated with fewer Gr3 TRAEs (35.3%), longer time to TRAE onset (52 days, range 44, 85), and fewer dose interruptions (29.4%) and dose reductions (11.8%). Conclusions: FX-909, a first-in-mechanism orally bioavailable PPARG inhibitor, demonstrates clinical proof-of-concept for targeting the master regulator of luminal lineage in heavily pretreated adv UC pts and may offer a novel therapeutic strategy for PPARG high adv UC. A randomized phase 1B expansion study is ongoing (NCT05929235). A combination study of FX-909/anti-PD1 is planned for early 2026. Clinical trial information: NCT05929235 .

TPS270 Background: ~15,000 men with metastatic prostate cancer (MPCa) receiving care in the Veterans Administration (VA) could benefit from germline testing to inform treatment options. However, there are insufficient genetics providers to arrange for germline testing. In 2021, the VA implemented multiple strategies to shift the task of germline testing to oncologists (i.e., mainstreaming) to increase guideline-concordant germline testing. These efforts led to improved adoption of germline testing by oncologists, and mainstreaming was more efficient and timely than genetics referral. However, germline testing remains low for many oncologists due to numerous barriers, e.g., lack of preparedness and confidence in germline testing, from identifying eligible patients to results-informed management. Methods: We will conduct a cluster-randomized trial comparing the effectiveness of implementation strategies to increase mainstreaming through randomization of oncologists not meeting a pre-specified test-order benchmark of 25%. Eligible participants include VA oncologists who: 1) have authored progress notes in oncology clinics for ≥10 patients with MPCa in the past two years, 2) are practicing at VA facilities with germline test orders in the electronic health record (EHR), and 3) <25% of their MPCa patients have had mainstream germline test orders. In the first 6 months, we will disseminate a germline testing handbook with provider education materials, information resources and EHR tools. At month 7, for oncologists who do not exceed the 25% benchmark, we will administer audit-feedback reports describing their germline testing rates compared to peers and the 25% benchmark. At month 12, oncologists whose order rates are <25% will be randomized to either continue audit-feedback or receive facilitation (interactive problem-solving and support) and audit-feedback. At month 24, we will stop the audit-feedback and facilitation strategies. Throughout the project, we will conduct semi-structured interviews to understand local germline testing workflows, challenges and enabling factors. We will assess fidelity to the implementation strategies and any adaptations made using the Longitudinal Implementation Strategy Tracking System. To evaluate the effectiveness of the implementation strategies on germline testing, we will use a mixed-effects model to account for the clustered nature of the data and the effect sizes of the implementation strategies. The project is informed by the RE-AIM (Reach, Effectiveness, Adoption, Implementation, Maintenance) framework. The study has not yet enrolled participants. There is no clinical trial registry number for this quality improvement project. This Department of Defense-funded project is a part of a greater quality improvement effort to increase guideline-concordant germline testing in the VA.

13 Background: Penile squamous cell carcinoma (PSCC) is a rare tumor with limited therapeutic options in advanced disease. Given the role of PD-L1 as a prognostic biomarker in many cancer subtypes, assessing its value in PSCC may inform prognosis and guide the use of immune checkpoint inhibitors in metastatic disease. In this systematic review and meta-analysis, we evaluate the prevalence of PD-L1 tumor expression in PSCC and assess its prognostic association with overall survival (OS), cancer-specific survival (CSS), and lymph node metastasis (LNM). Methods: We searched PubMed, Embase and Cochrane for relevant studies. The primary outcomes were OS, CSS, and LNM, analyzed using HRs or ORs, each with 95% CIs. Outcomes were prespecified in the PROSPERO protocol (CRD42019129410). Random-effects meta-analyses (restricted maximum-likelihood estimator) were performed using inverse-variance weighting. Heterogeneity was assessed by Q, τ², and I² statistics. Publication bias was examined by Egger’s test and funnel plots. Results: Fifteen retrospective studies were included (N = 1,445). Overall, 57% of tumors were PD-L1 positive (IQR, 44%–66%). Among 796 patients, PD-L1 expression was significantly higher in HPV-negative than in HPV-positive tumors (42% vs 13.5%). Eight studies (N = 930) evaluated OS, nine studies (N = 1,091) assessed CSS, and five studies (N = 622) investigated LNM. PD-L1 positivity was associated with reduced OS (HR 1.52; 1.12–2.05; I² = 6.8%) and CSS (HR 1.61; 1.16–2.23; I² = 38.3%), and with higher odds of LNM (OR 2.94; 1.28–6.78; I² = 57.3%). In meta-regression, HPV adjustment significantly strengthened CSS associations (QM = 7.13, p = 0.008; HR = 2.06, 1.21–3.51), accounting for all observed heterogeneity. Other moderators (sample size, adjustment status, tumor stage, study location, tumor-cell vs TIL scoring) were non-significant. Conclusions: PD-L1 expression is more frequent in HPV-negative PSCC and is consistently associated with adverse clinical outcomes, including shorter OS, shorter CSS, and higher odds of LNM. These findings support PD-L1 as a potential prognostic biomarker in PSCC and underscore the importance of considering HPV status in prognostic and therapeutic stratification.

140 Background: Chemotherapy-induced peripheral neuropathy (CIPN) is a painful and frequently persistent and debilitating toxicity of cancer treatment that greatly affects the quality of life of cancer survivors. There is an unmet need for effective treatments against established CIPN and for preventative approaches. The enzyme soluble epoxide hydrolase (sEH) converts epoxy fatty acids (EpFAs), which are polyunsaturated fatty acid (PUFA) derivatives with inflammation-resolving, analgesic and anti-cancer properties, into corresponding pro-inflammatory and hyperalgesic dihydroxy metabolites (diols). Thus, sEH is a master regulator of inflammation. The sEH inhibitor (sEHI) EC5026 has completed phase 1 studies in healthy volunteers, with no significant adverse effects noted. The EpFAs of the ω-3 PUFAs are particularly effective at promoting analgesia and resolution of inflammation, suggesting that a diet rich in ω-3 PUFAs would enhance the inflammation-resolving effects of a sEHI. Methods: We examined the prophylactic efficacy of the sEHI EC5026 in a mouse model of docetaxel-induced painful neuropathy, by measuring mechanical hind paw withdrawal thresholds (PWTs) using an electronic von Frey aesthesiometer (IITC, Woodland Hills, CA). EC5026 was formulated in the drinking water at a concentration calculated to provide the mice with a dose of ~5 mg/kg/day through their daily unrestricted water intake. In addition, we studied docetaxel-induced CIPN in adult sEH ( Ephx2 ) male homozygous knock-out (KO) mice and wild-type (WT) mice on a diet high in ω-3 PUFAs vs a diet high in ω-6 PUFAs (both from Research Diets Inc., New Brunswick, NJ), starting 3 weeks prior to beginning docetaxel treatment and continued throughout the duration of the experiment. We also measured the levels of non-esterified PUFAs and their corresponding EpFAs and diols by UPLC-MS/MS in mouse plasma. Results: Docetaxel decreased the PWT in male C57BL/6J WT mice, while prophylactic administration of EC5026 mitigated this neuropathic allodynia. After 3 weeks on the ω-3 vs ω-6 diet, the respective PUFAs and their corresponding dihydroxy metabolites were significantly enriched. As expected, genetic ablation of Ephx2 increased the plasma levels of the corresponding epoxides. Importantly, sEH KO mice exhibited less severe docetaxel-induced CIPN and recovered from it faster than their WT counterparts when fed a diet high in ω-3 PUFAs, but not when fed a diet high in ω-6 PUFAs. Conclusions: We propose that sEH is a druggable therapeutic target in CIPN pathophysiology. The benefits of targeting sEH can be augmented by a diet rich in omega-3 PUFAs.

577 Background: Hepatocellular carcinoma (HCC) is the most common primary liver cancer. Immune checkpoint inhibitors (ICPI), whether alone or in combination with other agents including tyrosine-kinase inhibitors (TKIs), has shown strong anti-tumor activity among patients with advanced HCC. We aim to perform a comparative analysis of ICPI-based combination treatment versus TKI monotherapy among advanced HCC patients. Methods: Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, we conducted a meta-analysis to evaluate the demographics, overall survival (OS), progression-free survival (PFS), and the frequency of all-grade adverse events (AEs) in patients with advanced HCC. Individual patient data (IPD) were reconstructed from Kaplan-Meier curves published across multiple clinical trials from 2018 to 2025. Results: We evaluated data from 4,429 patients, including 389 who received immune checkpoint inhibitors (ICPI) monotherapy, 878 treated with ICPI combined with bevacizumab, 728 treated with double ICPI therapy, 705 treated with ICPI plus TKI, and 1,729 who received tyrosine-kinase inhibitor (TKI) alone. The analysis revealed a median OS of 16.84 months (95% CI: 13.85-NA) in the ICPI monotherapy arm,19.09 months (95% CI: 17.10-21.27) in the ICPI+ bevacizumab arm, 19.58 months (95% CI: 17.56-23.32) in the duplet ICPI arm, 19.23 months (95% CI: 16.82-21.88) in the ICPI+ TKI group, and 15.30 months (95% CI: 13.96-16.27) in the TKI monotherapy arm (p<0.001). The median PFS values were 3.73 months (95% CI: 3.34-3.89), 5.75 months (95% CI: 5.55-6.74), 5.51 months (95% CI: 4.75-5.76), 6.17 months (95% CI: 5.58-7.07), and 5.42 months (95% CI: 4.76-5.51), respectively. Safety analysis outcomes revealed significantly higher rates of hypertension and platelet abnormalities in the ICPI+ TKI group, and increased diarrhea incidence in the TKI monotherapy and ICPI duplet groups. Conclusions: We found that duplet ICPI demonstrated improved OS and PFS compared to TKI and ICPI monotherapies and varying but tolerable safety profiles across regimens.

658 Background: Pancreatic ductal adenocarcinoma (PDAC) is often accompanied by radiographically occult metastatic disease. Diagnostic laparoscopy (DL) prevents non-therapeutic laparotomies by detecting occult metastases and guiding treatment. The aim of this study was to assess the clinical utility of DL in patients with PDAC stratified by perceived resectability (resectable, borderline resectable (BR), and locally advanced (LA)) and to identify risk factors for a positive DL. Methods: We retrospectively reviewed the electronic medical records of all PDAC patients without confirmed metastases who underwent DL at a tertiary medical center from January 2018 to June 2024. Data on demographics, imaging, tumor markers, DL findings, and outcomes were collected. Results: 435 patients met the study criteria. Before DL, 115 (26.4%) had perceived resectable, 165 (37.9%) had BR, and 155 (35.6%) had LA disease. DL revealed metastases in 20.0%, 23.0%, and 54.8%, and resection rates were 68.7%, 30.1%, and 5.2% in resectable, BR, and LA cases, respectively. In multivariable analysis, patients with resectable disease had smaller tumors, lower CA 19-9 levels, and were older (all p<0.05). Overall, 146 (33.6%) had metastases on DL, 151 (34.5%) had localized disease on DL but did not proceed to resection, and 138 (31.7%) underwent resection. Metastatic disease on DL was associated with larger tumors, indeterminate extra-pancreatic lesions on imaging, higher CA 19-9 levels, and distal pancreas involvement (all p<0.05). Of metastatic DL cases, 35.6% lacked concerning imaging, and 15.5% had a normal CA 19-9. Conclusions: DL found metastatic disease in one-third of patients and altered management across all resectability groups. Despite imaging advancements and normal tumor markers, DL often detects missed metastases. Given its diagnostic yield, prognostic value, and safety profile, DL should be considered more routinely in staging PDAC. Patients stratified by perceived resectability prior to diagnostic laparoscopy (N = 435). Variable Resectable (n=115) Borderline Resectable (n=165) Locally Advanced (n=155) P-value Age at DL, years, Median [IQR] 70.0 [62.0-75.5] 67.0 [59.0-72.0] 64.0 [59.0-69.0] <0.001 1 Gender, female (%) 44.3 47.3 47.7 0.85 2 Tumor Location (%) <0.001 2 Head of Pancreas 64.3 83.0 52.9 Distal Pancreas 35.7 17.0 47.1 Tumor Size, cm, Median [IQR] 2.7 [1.9-3.5] 3.1 [2.5-3.8] 4.0 [3.4-5.3] <0.001 1 CA 19-9 Prior to DL, U/mL, Median [IQR] 86.0 [27.0-248.0] 210.0 [56.6-608.5] 378.0 [86.3-1576.0] <0.001 3 CEA Prior to DL, ng/mL, Median [IQR] 3.0 [1.9-4.2] 3.2 [2.1-5.7] 5.0 [2.2-10.2] <0.001 3 Underwent Resection of Primary Tumor (%) 68.7 30.1 5.2 <0.001 2 Metastatic DL (%) 20.0 23.0 54.8 <0.001 2 N is the number of non-missing values. DL (Diagnostic Laparoscopy); IQR (Interquartile Range

53 Background: Liver metastases are the main cause of death in colorectal cancer (CRC), yet only ~20% of patients qualify for surgery. For unresectable colorectal liver metastases (CRCLM), hepatic arterial infusion chemotherapy (HAIC) and drug-eluting bead transarterial chemoembolization (DEB-TACE) are common treatments. This study assesses the efficacy and safety of combining DEB-TACE with HAIC using FOLFOX or FOLFIRI. Methods: This retrospective study included 221 CRCLM patients treated at Peking University Cancer Hospital from 2018 to 2023. Of these, 182 received DEB-TACE with HAIC-FOLFOX and 39 with HAIC-FOLFIRI. Propensity score matching (PSM) minimized selection bias. Primary endpoint was overall survival (OS); secondary endpoints included progression-free survival (PFS), hepatic PFS (hPFS), objective response rate (ORR), disease control rate (DCR), and treatment-related safety. Results: Propensity matching resulted in 38 pairs, achieving comparable baseline between DEB-TACE-HAIC-FOLFOX and DEB-TACE-HAIC-FOLFIRI cohorts. Of these pairs, 45 were male and 31 were female. Around 65.8% patients were younger than 65 years. The original locations of CRC in 67.1% of patients were on the left side. The middle to high grades of tumor were dominant (88.2%). Notably, 56.6% of patients harbored mutations in KRAS, NRAS or BRAF and extrahepatic metastasis occurred in 78.9% patients. Additionally, 75% patients were refractory to second line systemic therapy. Median hepatic progression-free survival (hPFS), progression-free survival (PFS), and overall survival (OS) were all slightly better in the matched DEB-TACE-HAIC-FOLFOX cohort, but not statistically significant (median hPFS: 8.6 vs. 5.8 months, P=0.206; median PFS: 6.4 vs. 4 months, P=0.062; median OS: 11.8 vs. 10.8 months, P=0.249). The objective response rate (ORR) was 36.8% in the DEB-TACE-HAIC-FOLFOX group and 39.5% in the DEB-TACE-HAIC-FOLFIRI group, with disease control rates (DCR) of 71.1% and 76.3%, respectively. These patients received 224 DEB-TACE plus HAIC treatments. There were no treatment-related deaths. Serious adverse events were comparable between the two groups, except for elevated transaminase. However, transaminase elevation was actually due to drug-eluting beads difference during DEB-TACE. With the further analysis of drug-eluting beads’ effect on adverse events, transaminase elevation, nausea, vomiting and pain occurred much frequently with DCB drug-eluting bead than HepaSphere drug-eluting bead; these differences were statistically significant. Conclusions: The combination of DEB-TACE with FOLFOX or FOLFIRI HAIC shows promising efficacy and tolerable safety for unresectable CRCLM. The FOLFIRI-based regimen is a viable alternative, especially for oxaliplatin-intolerant patients.