A Phase 2 Trial of an Extended-release siRNA Implant Targeting KRAS G12D/V in Locally Advanced Pancreatic Cancer

Abstract

Purpose

Locally advanced pancreatic cancer (LAPC) accounts for 30% of pancreatic cancers. We assessed the efficacy and safety of a novel extended-release siRNA targeting KRAS G12D/V mutations (siG12D-LODER) combined with chemotherapy in LAPC.

Patients and Methods

This two-cohort, phase 2 multicenter, open-label study (NCT01676259) evaluated siG12D-LODER with chemotherapy in patients with LAPC, irrespective of KRAS status. In cohort 1, patients were randomized to siG12D-LODER plus gemcitabine/nab-paclitaxel (Arm 1) or gemcitabine/nab-paclitaxel alone (Arm 2). In cohort 2, patients with LAPC or borderline resectable disease received siG12D-LODER plus standard chemotherapy (modified FOLFIRINOX or gemcitabine/nab-paclitaxel) in a single-arm, non-randomized design. Primary endpoints were overall survival (OS) for cohort 1 and objective response rate (ORR) for cohort 2. Secondary endpoints included progression-free survival (PFS), duration of response (DoR), OS (cohort 2), and ORR (cohort 1).

Results

Fifty-nine patients were enrolled across two cohorts. In cohort 1, median OS in the modified intent-to-treat (mITT) population unselected for KRAS status was 22.7 months for siG12D-LODER + gemcitabine/nab-paclitaxel versus 21.9 months for chemotherapy alone (p>0.05). Among patients with KRAS G12D/V mutations, OS was 22.7 vs 13.4 months (HR 0.59, 95% CI 0.18–1.96, p=0.36). In cohort 2, ORR was 31.6% (95% CI 0.13–0.57) in the mITT (unselected for KRAS); in the G12D/V subgroup, ORR was 57.1%, similar to 63.6% in cohort 1. TEAEs were mainly procedure-related, including grade 1/2 gastrointestinal events and higher infection rates in the intervention arm.

Conclusions

siG12D-LODER plus chemotherapy is safe, tolerable, and warrants further investigation in KRAS G12D/V-mutant LAPC.

Statement of Translational Relevance

Patients with locally advanced pancreatic cancer have high rates of morbidity and mortality and few available treatment options. This study presents the results from a two-cohort phase 2 trial evaluating the efficacy and safety of a novel KRAS G12D/V-targeted siRNA in combination with standard of care chemotherapy in patients with locally advanced pancreatic cancer. In an unselected population no difference in primary endpoints were observed for patients treated with siG12D-LODER plus chemotherapy versus chemotherapy alone. Exploratory analyses suggest higher response rates and improved overall survival for patients whose tumors harbored KRAS G12D/V-mutations who were treated with siG12D-LODER an provide the basis for further investigation in a pre-selected biomarker identified population.

Introduction

Pancreatic cancer is typically detected at an advanced stage with most patients presenting with locally advanced (~30%) or metastatic disease (~50%) ^1,2. In current paradigms, patients with LAPC receive induction chemotherapy with or without radiation therapy, however, only a small percentage respond sufficiently to render surgical resection feasible ³. In many cases, systemic treatment is associated with high rates of failure characterized by local progression incurring substantial morbidity and/ or development of metastatic disease. New therapies and combinations are needed to improve outcomes for patients with unresectable, LAPC.

Oncogenic missense mutations in Kirsten rat sarcoma viral oncogene homologue (KRAS) are present in >90% of pancreatic ductal adenocarcinoma (PDAC) with the majority (~70%) being G12D and G12V mutations ⁴. Constitutively active mutant KRAS promotes cell proliferation, tumor growth, and therapy resistance. Two KRAS G12C-selective inhibitors have received FDA approval for the treatment of KRAS G12C-mutant advanced non-small cell lung and colorectal cancer, although their use has been met with emerging resistance through varied mechanisms including acquired mutations in KRAS/MAPK pathway genes, other genomic alterations, and histologic transformation ^5–8. These drugs, adagrasib and sotorasib have National Comprehensive Cancer Network (NCCN) guideline endorsement in patients with PDAC and KRAS G12C mutations (approximately 1% of PDAC). Recent studies have suggested that transcriptional subtypes of pancreatic cancer cells may respond differently to KRAS inhibition and that a combination therapy (e.g. KRAS inhibition plus chemotherapy) may be required to improve durability of response^9,10. Consequently, there is a clinical need for more universal anti-KRAS therapeutics with manageable toxicity profiles that could be combined with chemotherapy, immunotherapy, or other targeted therapeutics.

RNA silencing therapy offers an alternative approach to targeting KRAS by suppressing gene expression and manufacturing of the mutant protein. Small interfering RNAs (siRNA) targeting KRAS can specifically and potently silence gene expression through degradation of its mRNA leading to decreased cellular proliferation and impaired tumor growth in preclinical models of PDAC ^11–13. Previously, a novel therapeutic platform comprising an anti-KRAS G12D siRNA enclosed in a biodegradable polymer matrix (siG12D-LODER) was developed for intratumoral delivery via endoscopic ultrasound (EUS) biopsy needle ^13,14. The LODER^™ delivery platform enables prolonged release and activity of siRNAs using reduced doses to minimize associated toxicities. The local delivery method can overcome challenges typically associated with systemic administration, such as peripheral degradation, off-target effects, and immune activation¹⁵. Preclinical analysis of siG12D-LODER suggested it possesses inhibitory activity against KRAS G12D and G12V alleles in vivo^13,16, and has demonstrated activity against wildtype KRAS in vitro (data not shown). A phase 1/2a clinical trial of siG12D-LODER in combination with chemotherapy (gemcitabine or modified (m) FOLFIRINOX) in patients with LAPC established that siG12D-LODER was safe and well tolerated at doses up to 3mg with no dose limiting toxicities observed¹⁶. Additionally, siG12D-LODER combined with standard chemotherapy demonstrated clinical promise with a median overall survival of 15.12 months and most patients achieving stable disease.

Based on these preclinical/early-stage clinical findings and the high prevalence of oncogenic KRAS mutations in PDAC, we hypothesized that the addition of siG12D-LODER combined with standard chemotherapy would lead to improved clinical outcomes in patients with LAPC. In this context, we conducted a phase 2 trial that evaluated the safety, tolerability, and efficacy of endoscopically delivered siG12D-LODER in patients with LAPC.

Patients and Methods

Study Design

The trial was designed as a multicenter two-cohort Phase 2 open-label trial conducted at sites in the United States and Israel between 2018 and 2023. Patients were enrolled in cohort 1 from 2018 to 2019 and, subsequently, in cohort 2 from 2020 to 2023. Cohort 1 was designed to assess the efficacy, safety, and tolerability of siG12D-LODER plus gemcitabine/nab-paclitaxel (“siG12D-LODER + GnP”) versus gemcitabine/nab-paclitaxel (“GnP”) alone in a randomized fashion. To facilitate enrollment, cohort 2 was subsequently designed as a single arm study designed to assess safety, tolerability, and efficacy of siG12D-LODER plus standard chemotherapy (either modified FOLFIRINOX or gemcitabine nab-paclitaxel, “siG12D-LODER+SOC”). Safety was assessed across both cohorts. This trial was registered with ClinicalTrials.gov, ID NCT01676259.

Patients

Male and female patients ≥ 18 years of age with Eastern Cooperative Oncology Group (ECOG) status 0 to 1 were enrolled. Cohort 1 was conducted in patients with unresectable LAPC, whereas cohort 2 included patients with LAPC or borderline resectable pancreatic cancer. Other key eligibility criteria for both cohorts included the presence of measurable disease with a target tumor accessible by EUS. Prior pancreatic cancer treatment, resectable disease, or metastatic disease were exclusionary. Other exclusion criteria included bulky celiac adenopathy, significant coagulopathy, secondary malignancy, or evidence of ascites (beyond trace). For both cohorts, patients were enrolled to the study irrespective of KRAS mutation status. In cohort 1, 37 patients were enrolled with 19 patients randomized to Arm 1 and 18 patients randomized to Arm 2. Following completion of cohort 1, 22 patients were enrolled to a single arm in cohort 2 (Supplementary Figure 1).

Study Treatment

Cohort 1: Patients randomized to arm 1 received up to 8 units (2.8mg) of siG12D-LODER administered by insertion into the primary tumor via endoscopic ultrasound (EUS 19G needle) once every 12 weeks +/− 7 days. In arm 1, patients commenced systemic therapy within 7 days of the initial siG12D-LODER insertion. For patients in both arms of cohort 1, gemcitabine (1000 mg/ m²)/nab-paclitaxel (125 mg/ m²) was administered intravenously (IV) on days 1, 8, 15 of a 4-week schedule. For both arms, treatment was repeated until disease progression, toxic effects, or study withdrawal for other reasons occurred.

Cohort 2: Patients in cohort 2 received siG12D-LODER (as described above) along with standard chemotherapy beginning up to 7 days following initial LODER insertion. Patients were assigned to either GnP or FOLFIRINOX/mFOLFIRINOX at the discretion of the investigator. GnP was administered as described above. (m)FOLFIRINOX was given on day 1 of a 2-week cycle as follows with either standard dosing: oxaliplatin (85 mg/ m² IV over 2 hours), leucovorin (400 mg/ m² IV), irinotecan (180 mg/ m² IV), 5-fluorouracil (400 mg/m² IV push), 5-fluorouracil (2400 mg/m² continuous IV infusion over 46–48 hours) or as mFOLFIRINOX: same with exception for modified irinotecan (150 mg/ m² IV) and no bolus 5-fluorouracil.

siG12D-LODER administration: The siG12D-LODER was inserted into the pancreatic tumor with the use of a standard 19-gauge endoscope ultrasound needle. Before insertion, the product was pre-loaded into the needle using a Silenseed-developed LODER Loading Device (LLD). Each dose of siG12D-LODER consisted of up to eight units (2.8mg). In the first step, four siG12D-LODER units were administered into the tumor using a 19-gauge (19G) EUS needle. This step was repeated using an additional four siG12D-LODERs for a total of eight units. Needles used included EchoTip^® Ultra (Cook Medical), or Expect 19ga FlexTM, (Boston Scientific) or 19G Aspiration Needle EZ Shot 3 Plus (Olympus).

Treatment schema is depicted in Figure 1. For both cohorts, treatment was continued until patients experienced disease progression, intolerable toxicity, or withdrew from study for any other reason. Chemotherapy dose modifications were made according to standard institutional practice. Written informed consent was obtained from all patients. This study was conducted with approval from an Institutional Review Board/Independent Ethics Committee (IRB/IEC) and in accordance with the Declaration of Helsinki.

Figure 1. Treatment Schema.

^aTreatment was continued until disease progression, intolerable toxicity or withdrawal from study for any other reason.

^bGemcitabine/nab-paclitaxel or (m)FOLFIRINOX per investigator’s choice.

Created in BioRender. Alagesan, B. (2025) https://BioRender.com/gr39h3a

Study Objectives

Primary endpoints were overall survival (OS) for cohort 1 and objective response rate (ORR), defined as the proportion of patients with best overall confirmed response of either a CR or PR, for cohort 2. Additional endpoints included progression-free survival (PFS), duration of response (DoR), OS (for cohort 2), and ORR (for cohort 1). Safety was evaluated based on vital signs, physical examination, ECOG status, pain assessment, clinical laboratory assessments, and adverse events. Exploratory endpoints included analysis of primary endpoints by KRAS mutational status.

Assessments

OS was defined as the time from randomization until death from any cause. ORR was defined as the proportion of patient with best overall confirmed response of either a partial response (PR) or complete response (CR). Safety was evaluated for both cohorts by adverse events (AE), vital signs, physical examination, as well as laboratory values (hematological, biochemical, coagulation studies and amylase/lipase) for cohort 1 and ECOG status for cohort 2. Genotyping of KRAS mutation was performed locally per standard of care assays. Pharmacokinetic (PK) analysis was performed on plasma from the first eight patients enrolled in arm 1 of cohort 1 at the following timepoints: pre- and post- EUS procedure, 15 ±3 minutes pre-siG12D-LODER treatment and post-siG12D-LODER treatment (at15±3 min, 30 ±3 min, 1 hour ±10min, 2 hours ±10min, 4 hours ±10min, 8 hours ±1hour, and 24 hours ±1 hour). Serum CA 19–9 level was assessed by local laboratory throughout the study duration.

Statistical Methods

All efficacy analyses were performed on the modified intention to treat (mITT) population defined as cohort 1: randomized participants who received ≥ 1 dose of study intervention, or cohort 2: enrolled participants who received ≥ 1 dose of study intervention with measurable disease at baseline and at least one post-baseline tumor assessment per RECIST v1.1. Time-to-event endpoints (OS and PFS) were estimated using the Kaplan-Meier approach with log-rank tests used for comparisons. Continuous variables are represented using descriptive statistics (mean +/− SD or median and IQR). ORR and secondary categorical responses are represented as proportions with Clopper-Pearson 95% confidence intervals. Time-to-event endpoints and ORR by KRAS mutation were analyzed using methods analogous to those indicated for the primary and secondary endpoints above. Descriptive statistics were used to summarize safety and other endpoints.

Data Availability

Data from this study are not publicly available to maintain patient privacy but can be made available upon reasonable request to the corresponding author.

Results

Patient Characteristics

Overall, N= 59 patients were enrolled in cohorts 1 and 2 across 9 centers in the United States and Israel between March 2018 and July 2023 (Figure 1). Patient demographics and characteristics were balanced across treatment groups (Table 1). The representativeness of the study participants is shown in Supplementary Table S1. A total of N= 38 participants with LAPC or BRPC (cohort 2 only) were treated with siG12D-LODER (Figure 1). In cohort 1, N= 37 patients were enrolled with N= 19 patients randomized to Arm 1 (siG12D-LODER+GnP) and N= 18 patients randomized to Arm 2 (GnP). Of these, the modified intention-to-treat (mITT) group included N=18 patients treated with siG12D-LODER and N= 11 treated with GnP. Twenty-two patients were enrolled in cohort 2 with 20 patients who ultimately received treatment and had at least one post-baseline tumor assessment. In cohort 2, chemotherapy choice was determined by the treating physician; nineteen patients received FOLFIRINOX/mFOLFIRINOX and one patient received GnP.

Table 1.

Demographic and baseline patient characteristics

Characteristic	Cohort 1		Cohort 2
	Arm 1 siG12D-LODER+GnP (n= 19)	Arm 2 GnP (n=18)	siG12D-LODER + SOC (n=22)
Age (years)
Median (range)	71 (54,82)	70.5 (44, 85)	65.5 (50, 84)
Gender, n (%)
Male	8 (42.1)	10 (55.6)	12 (54.5)
Race, n (%)
American Indian or Alaska Native	1 (5.6)	-	-
Asian	1 (5.6)	1 (6.3)	1 (4.5)
Black or African American	-	1 (6.3)	1 (4.5)
White	15 (83.3)	14 (87.5)	20 (90.9)
Other	1 (5.6)	-	-
Missing	1 (5.6)	2 (12.6)	-
ECOG performance status, mean (± standard deviation)	0.7 (0.46)	0.7 (0.47)	0.5 (0.51)

Clinical Efficacy

The mITT population in both cohorts was unselected for KRAS mutational status. In the mITT population in cohort 1, the median OS for patients treated with siG12D-LODER + GnP was 22.7 months compared to 21.9 months in the GnP group (HR 1.47 95% C.I. 0.643–3.365, P= 0.361) (Table 2). The secondary endpoint of ORR, defined as at least PR, was 44.4% in the siG12D-LODER+GnP compared to 27.3% in the GnP group (OR 2.13, 95% CI, 0.422–0.778, P= 0.360). There were no patients in either arm who achieved a CR. Additionally, there was no significant difference between the PFS or DoR for patients in Arm1 vs Arm 2 of cohort 1. The mean (±SD) baseline serum CA 19–9 levels were higher in the siG12D-LODER +GnP group (1390.8 ± 2789.1) versus GnP (890.8 ± 1091.7) and remained higher throughout all comparable collected time points. In cohort 1, 13/18 patients (72%, 95% CI, 0.465–0.903) treated with siG12D-LODER + GnP and 9/11 patients (81%, 95% CI, 0.482–0.977) treated with GnP alone experienced a maximum reduction in serum CA 19–9 level of ≥ 50%. For cohort 2, the primary efficacy endpoint of ORR in the mITT population was 31.6% overall (95% CI, 0.126, 0.566). The secondary endpoint of OS was 22.1 months for cohort 2. The mean (±SD) baseline serum CA 19–9 levels in the siG12D-LODER group of cohort 2 (2774.9 ± 6135.3) were higher than those in cohort 1. In cohort 2, 11/17 patients (64.7%, 95% CI, 0.383–0.858) treated with siG12D-LODER experienced a maximum reduction in serum CA 19–9 level of ≥ 50%.

Table 2.

Treatment response

	OS, months Median	Hazard Ratio (95% CI)	PFS, months Median	Hazard ratio (95% CI)	ORR, % (95% CI)	Odds ratio (95% CI)	DoR, months Median	Hazard ratio (95% CI)
Cohort 1 Arm 1:siG12D-LODER+GnP(N= 18)	22.7	1.47 (0.643, 3.365)	15.5	1.81 (0.744, 4.415)	44.4 (21.5,69.2)	2.13 (0.422, 10.778)	9.3	6.28 (0.762, 51.79)
Cohort 1 Arm 2: GnP (N=11)	21.9		20.1		27.3 (6.0,61.0)		34.0
Cohort 2a: siG12D-LODER + SOC (N=19)	22.1	--	14.5	--	31.6	--	15.8	--

^aCohort was non-randomized.

Abbreviation: ECOG, Eastern Cooperative Oncology Group.

An exploratory analysis was performed to determine KRAS mutational status at codon 12 (G12x) for N=31 patients for whom that data was available (Figure 2A). Amongst all patients in cohort 1 with known KRAS G12x status, there was no difference in OS between Arms 1 and 2. In patients with KRAS G12D/V mutations however, the median OS was 22.7 months in those treated with siG12D-LODER+GnP (N=11) compared to 13.4 months in the GnP arm (N=5, HR 0.59, 95% CI, 0.18–1.96, p=0.39) (Figure 2B). In the KRAS G12D/V subgroup of cohort 1, the secondary endpoint of ORR was 63.6% compared to 20.0 % in GnP (p>0.05). This subgroup represented a small group of patients, and the study was not powered to detect statistically significant differences. Among patients in the KRAS G12D/V subgroup of cohort 2, the ORR was 57.1%, which was similar to the proportion seen in the G12D/V subgroup of siG12D-LODER-treated patients in cohort 1 (Table 3).

Figure 2.

A) KRAS mutation frequency by treatment group. B) Exploratory survival analysis of patients in Cohort 1 with KRAS G12D/V mutations. GnP: Gemcitabine/nab-paclitaxel

Table 3.

Exploratory analysis of treatment response in patients with KRAS D/V mutations

KRAS D/V mutation	Cohort 1		Cohort 2
	Arm 1:siG12D-LODER + GnP (N= 11)	Arm 2: GnP (n =5)	siG12D-LODER + SOC (N= 7)
OS, median	22.7 m	13.5 m	13.6 m
PFS, median	12.7 m	13.5 m	12.7 m
ORR, % (95% CI)	63.6 (30.8, 89.1)	20.0 (0.5, 71.6)	57.1(18.4,90.1)

A fraction of patients with LAPC who undergo neoadjuvant therapy ultimately proceed to resection, while non-surgical candidates may receive radiation therapy to control local disease and relieve symptoms. In this study, post-hoc analysis identified that 3/18 patients (16.7%) in the siG12D-LODER+GnP of cohort 1 and 6/20 patients (30.0%) in cohort 2 treated with siG12D-LODER terminated the study once their tumors reached resectable endpoint, whereas only 1/11 patients (9.1%) in Arm 2 of cohort 1 receiving standard of care therapy terminated study for this reason. Of patients treated with siG12D-LODER, 2/18 patients (11.1%) and 1/20 patients (5%) in cohorts 1 and 2 discontinued study to pursue radiation therapy +/− chemotherapy as did 5/11 patients (45.5%) treated with GnP alone in cohort 1.

Safety and Pharmacokinetic Results

The safety population included patients in cohorts 1 and 2 who received ≥ 1 dose of treatment (N=49 total, N=38 for sG12D-LODER across cohorts, N=11 for GnP cohort 1). The average number of treatment cycles with siG12D-LODER ranged from 1 (n=18, 100%) to 5 (n=5, 27.8%) in cohort 1 and 1 (n=20, 100%) to 3 (n=8, 40%) in cohort 2. An average of 20.08 units of siG12D-LODER (range 2.0–40.0) was administered per patient in cohort 1 and 15.2 units (range: 8.0–24.0) per patient in cohort 2. The combination of siG12D-LODER and standard chemotherapy was generally well-tolerated across both cohorts with similar toxicities to those previously described ¹⁶. All patients experienced at least one treatment-emergent adverse event (TEAE, incidence of 100%) (Table 4). The most common TEAEs of any grade were gastrointestinal disorders (97%), blood disorders (93%), and generalized disorders including fatigue, fevers, chills (90%). A total of N= 27 patients (N= 24 /63.2% in siG12D-LODER+GnP, N=3/27.3% in GnP) reported serious TEAEs. The most frequently reported serious TEAE was sepsis, which was reported in N=5 patients treated with siG12D-LODER (all in cohort 1) and not seen in the GnP arm of cohort 1. Of all serious TEAEs across both cohorts, 9 events (pyrexia x2, abdominal pain, hyperbilirubinemia, pancreas infection x2, sepsis, procedural hemorrhage/presyncope, gastric hemorrhage) were possibly related to either siG12D-LODER or the EUS procedure and each of these resolved. Treatment-related adverse events (TRAEs) and procedure-related TEAEs were reported in 42.1% and 44.7%, respectively, of patients treated with siG12D-LODER (Table 4). Two Grade 5 events were reported in the siG12D-LODER population (one event each in cohort 1 and 2). One participant, while hospitalized for management of diarrhea, abdominal pain, and nausea, developed acute intestinal ischemia complicated by bowel infarction and pneumatosis leading to death on day 12 post-treatment initiation. CT imaging revealed complete occlusion of the superior mesenteric and portal veins. A second grade 5 event occurred in a patient who was hospitalized with recurrent gram negative (Klebsiella pneumoniae) sepsis 11 weeks following the last dose of treatment. This patient was hospitalized for grade 3 acute cholangitis requiring ERCP and biliary stent exchange 5 weeks prior to the grade 5 event. Both grade 5 events were assessed for causality and determined to be unrelated to treatment and related to underlying disease

Table 4.

Summary of treatment emergent adverse events

Adverse events, N (%)	Cohort 1 Arm 1 + Cohort 2: siG12D-LODER + GnP or SOC (N= 38)	Cohort 1: Arm 2 GnP (N= 11)
With ≥ 1 TEAE	38 (100.0)	11 (100.0)
With ≥ 1 serious TEAE	24 (63.1)	3 (27.3)
Discontinued from study due to TEAE	4 (10.5)	-
TEAE with outcome of deatha	2 (5.3)	-
With ≥ 1 treatment-related TEAE	16 (42.1)	-
With ≥ 1 procedure-related TEAE	17 (44.7)	-
Related TEAEs occurring in at least 5% of patients
Gastrointestinal disorders
Overall	6 (15.8)	-
Vomiting	3 (7.9)	-
Abdominal pain	2 (5.3)	-
Blood and lymphatic systemic disorders
Overall	2(5.2)	-
Thrombocytopenia	2(5.2)	-
General disorders
Overall	5 (13.2)	-
Pyrexia	3 (7.9)	-
Fatigue	2(5.2)	-
Hepatobiliary disorders
Overall	4 (10.5)	-
Elevated AST	2 (5.3)	-
Nervous system disorders
Overall	2 (5.3)	-

^aTwo grade 5 events (intestinal ischemia complicated by bowel infarction on day 12 post-treatment, gram negative sepsis) occurred in the siG12D-LODER groups (1 event each in cohorts 1 and 2). Both grade 5 events were assessed to be unrelated to treatment.

^bIncludes possibly related, probably related, and definitely related adverse events. Adverse events with missing relatedness are considered related. Events in the SOC arm were not assessed for relatedness.

The first eight patients enrolled in arm 1 of cohort 1 were evaluated for pharmacokinetic analysis. The plasma concentration of siG12D-LODER was below the lower level of quantitation (<0.25ng/mL) in all patients at all timepoints tested (data not shown).

Discussion

Patients with LAPC incur significant morbidity, commonly with pain and biliary obstruction, and associated mortality due to locally destructive tumor growth and/or eventual development of metastatic disease ^17,18. Surgery is the only potentially curative option for LAPC but is available to a small subset of patients whose tumors can be successfully downstaged with neoadjuvant systemic therapy¹⁹. Data from a large retrospective analysis by the Trans-Atlantic Pancreatic Surgery (TAPS) Consortium demonstrated that 17.6% of patients with LAPC proceeded to undergo resection following induction chemotherapy with mFOLFIRINOX ²⁰. Adjunctive radiation therapy can be used in LAPC to restrain local progression, relieve symptoms, provide time away from therapy, and potentially achieve resectability, however, it has not been shown to provide a clinically significant improvement in OS in randomized controlled trials ²¹.

KRAS-directed therapies hold significant promise for patients with pancreatic cancer with advanced disease and limited treatment options. Among patients with advanced stage pancreatic cancer, KRAS G12D/V mutations are associated with worse outcomes to chemotherapy and worse OS ^22–24. Secondary activating mutations in KRAS, including additional oncogenic G12x mutations, have been identified as a resistance mechanism to allele-specific inhibition ^5,6,25. Therefore, universal KRAS-directed therapies targeting multiple mutations, particularly G12D/V mutations, may enhance therapeutic efficacy. Several small molecule K(RAS) inhibitors targeting both the active GTP-bound (KRAS-ON) and inactive GDP-bound (KRAS-OFF) states, as well as single or multiple isoforms, are currently under clinical investigation (GDC-6036: NCT04449874; RMC-9805:NCT06040541; LY4066434:NCT06607185; INCB161734: NCT06179160) 4. In a Phase 1 study, the pan-RAS inhibitor daraxonrasib (RMC-6236) has shown early signs of efficacy with an initial disease control rate of 86% in heavily pretreated metastatic PDAC patients and an updated PFS of 8.1 months (95% CI, 5.9-not evaluable [NE]) in the second-line for patients with KRAS G12x mutations (NCT05379985)^26,27 ; a phase 3 RASolute trial investigating the use of daraxonrasib in the second line versus SOC chemotherapy is currently underway (NCT06625320). The first-in-class KRAS^G12D protein degrader, ASP-3082, preliminarily showed acceptable toxicity and demonstrated antitumor activity in a cohort of patients with extensively pretreated metastatic PDAC ²⁸. Nucleic acid-based therapies, including antisense oligonucleotides (ASO) and siRNA, act post-transcriptionally to target degradation of KRAS mRNA, preventing protein synthesis. AZD4785, an antisense oligonucleotide targeting both mutant and wildtype KRAS mRNA, was evaluated in phase I but failed to progress due to concerns over efficacy ^29,30. siG12D-LODER, an siRNA designed to inhibit KRAS G12V/G12D mRNA expression, demonstrated tolerability and clinical promise in a phase 1 study with most patients (76%) achieving stable disease and 23.1% with PR ¹⁶. In the locally advanced disease setting, , significant morbidity can occur making local control a key therapeutic goal. Unlike most other KRAS-directed therapies, siG12D-LODER is administered intratumorally, which may be advantageous to overcoming the desmoplastic barrier and relative hypovascularity of pancreatic tumors and avoiding off-target effects³¹. This local delivery approach facilitates direct intratumoral target suppression while minimizing systemic toxicity, including hepatotoxicity, mucositis, and rash that have been observed with oral RAS inhibitors and other systemic KRAS-directed therapies.

Herein, we present the results of a phase 2 clinical trial evaluating the efficacy and safety of siG12D-LODER for the treatment of nonresectable LAPC. Our final analysis demonstrated that there was no difference between the overall survival for patients treated with siG12D-LODER+GnP compared to GnP in the general study population. However, a numerical trend toward improved OS and ORR for patients with KRAS G12V/D was observed with siG12D-LODER+GnP (N=11) compared to GnP in cohort 1 (N=5). In the KRAS G12D/V subgroup of cohort 2 single arm N=7), a similar proportion of patients had at least a partial response to that seen in the corresponding cohort 1 analysis. Among patients treated with siG12D-LODER in this study, a slightly higher rate (23.7%) achieved resectability than previously analyzed retrospective cohorts ²⁰; however, the study’s small sample size and inclusion of borderline resectable patients in cohort 2 limit confidence around these conclusions.

In this study, siG12D-LODER was generally well tolerated, however, with a higher incidence of sepsis reported compared to prior investigation¹⁶. Although 5 patients treated with siG12D-LODER patients (all in cohort 1 Arm 1) experienced sepsis, only one episode was determined to be possibly related to siG12D-LODER treatment by the safety review committee. All other events occurred beyond the last administration of study treatment (≥1 month) and/or were attributable to alternative causes (e.g. pneumonia, febrile neutropenia/typhlitis). Two grade 5 events, which occurred in the siG12D-LODER population, were determined by the safety review committee to be unrelated to siG12D-LODER. Both deaths were attributable to well-recognized complications of unresectable pancreatic ductal adenocarcinoma—one from splanchnic venous thrombosis leading to intestinal ischemia and infarction, and the other from recurrent biliary infection culminating in gram-negative sepsis— which are associated with increased risk of morbidity and mortality^32–35. The most common adverse events were gastrointestinal disorders and most serious related adverse events were reversible and primarily related to the endoscopic procedure. The intra-tumoral administration of siG12D-LODER, and its and resultant low systemic levels, may serve to minimize systemic toxicity and off-target effects while allowing for increased local concentrations. Indeed, the lack of systemic side effects typically observed with oral small molecule inhibitors of KRAS (e.g. rashes), maybe attributed to siG12D-LODER’s intra-tumoral administration as well as its mechanism of action targeting KRAS. Silencing the oncogene, and thereby preventing the intracellular production of the mutated KRAS protein, as compared to inhibiting it, could have important implications for limiting resistance observed with the KRAS-off small molecule inhibitors. Additionally, concomitant silencing of wild-type and mutant KRAS may prevent the adaptive resistance observed through feedback activation of RTK signaling and downstream MAPK and PI3K pathway reactivation in the setting of mutant KRAS inhibition^36,37.

There are a series of limitations to our study. First, KRAS mutational status was not determined for patients prior to enrollment and could not be ascertained for all patients (24.1% and 52.6% of patients KRAS mutational status was unknown in cohorts 1 and 2, respectively) as this was not mandated prior to study entry and was obtained from locally collected data and retrospectively correlated and was not standard in this multi-site study at the time of enrollment. The resultant small sample size for certain mutational groups limited the power for statistical analyses. Second, post-treatment tumor specimens were lacking, which limited molecular comparisons between responders and non-responders to determine target engagement and identify biomarkers of response and resistance. Additionally, cohort 2 was designed as a single arm study, due to the high rate of attrition in cohort 1 arm 2 and included a spectrum of both borderline resectable and locally advanced PDAC. The practicality facilitated timely enrollment and ensured adequate follow-up; nonetheless, the lack of concurrent control group for cohort 2 and the heterogeneity of disease stage limits the interpretability of study results. Accordingly, the results from cohort 2 should be considered in the context of historical benchmarks and results from cohort 1 and interpreted with these limitations in context.

The vast majority of PDAC is characterized by activating KRAS mutations with ~70–75% of those being G12D or V mutations. In LAPC, pathologic diagnosis typically requires endoscopic ultrasound and can yield low volumes of tissue and limited genomic DNA for biomarker identification. Due to the prevalence of KRAS mutations and the relative difficulty of determining mutational status at the time of study initiation, we adopted a pragmatic approach with a non-biomarker enriched design. Recently, the advent of more cost-effective and sensitive sequencing techniques, including cell free and PCR DNA-based approaches, has positioned universal biomarker testing for KRAS mutations within reach ³⁸.

Although this study did not meet its primary endpoint in the mutation-agnostic overall population, the exploratory subgroup analysis suggests that siG12D-LODER may provide benefit in combination with chemotherapy for LAPC patients with KRAS G12D/V mutations. These findings should be interpreted as exploratory and hypothesis-generating, intended primarily to inform the design and powering assumptions of a future Phase 3 trial. To this end, a next-generation optimized siRNA product, SIL-204, with increased stability and broadened silencing activity across a spectrum of activating KRAS mutations (G12x, Q61H, and G13D) is currently in late-stage preclinical development with promising activity in reducing tumor growth and inducing tumor necrosis ³⁹. Future clinical investigations with mandatory prospective KRAS testing will assess the use of SIL-204 in molecularly defined non-resectable pancreatic cancer.