Tumor-Targeted IL-12 (PDS01ADC) with Hepatic Artery Infusion Pump Therapy for Colorectal Liver Metastases: Interim Analysis of a Non-randomized Phase II Trial.

Introduction
Colorectal cancer (CRC) is the third most common cancer worldwide, with over 1.9 million new cases in 2022, and the second leading cause of cancer-related mortality 1. The incidence of CRC is on the rise in young patients (<50 years of age), carrying significant economic and lifestyle ramifications 2. Despite the identification of molecular targets and an expanding armamentarium of FDA-approved agents, 5-year survival for patients with metastatic colorectal cancer (mCRC) remains less than 20% 3. Efficacy of immune checkpoint inhibition (ICI), which has had an enormous impact on cancer care, has largely been ineffective for the ~95% of mCRC patients with microsatellite stable (MSS) or mismatch repair-proficient (pMMR) disease. Attempts to increase efficacy of ICI for patients with mCRC through coupling with vascular endothelial growth factor receptor-tyrosine kinase inhibitors have been met with marginal success, especially in patients with liver metastases (the most frequent site of metastatic disease and often life-limiting) 4.
Interleukin-12 is a potent immunostimulatory cytokine that enhances cell-mediated immunity by activating natural killer (NK) cells, natural killer T (NKT) cells, and CD8+ T cells, promoting Th1 differentiation, interferon-gamma (IFN-γ) production, and angiogenesis inhibition 5,6. Importantly, IL-12 expression within tumor nodules was demonstrated to facilitate lymphocyte infiltration in mice 7. While recombinant IL-12 (rIL-12) demonstrated anti-tumor activity in multiple preclinical models 8, clinical application has been constrained by poor tolerability and limited intratumoral drug accumulation 9,10. To overcome these barriers, PDS01ADC (previously designated NHS-IL12) was developed as a tumor-targeting immunocytokine 11. This construct is a fusion of two human IL-12 heterodimers to the NHS76 human immunoglobulin G1 (IgG1) antibody, which selectively binds histone-DNA complexes exposed in necrotic tissue enabling targeted IL-12 delivery 12. Early-phase clinical studies have explored the effects of PDS01ADC in patients with solid tumors, including MSS/pMMR mCRC. In a Phase I trial (NCT01417546), monotherapy with PDS01ADC demonstrated modest clinical activity, with stable disease (SD) as the best overall response in 15 of 30 evaluable patients, which included 4 of 8 patients with CRC 13. Preliminary data from an ongoing Phase II trial (NCT04491955) evaluating PDS01ADC in combination with additional immunotherapies in patients with advanced small bowel and CRC showed disease reduction in 2 (−26.7% and −70.6%) of 18 patients14.
Hepatic artery infusion pump (HAIP) therapy capitalizes on the hepatic artery’s role as the primary blood supply for liver tumors 15, enabling direct infusion of floxuridine into the tumor microenvironment with concomitant systemic chemotherapy administration 16. This approach achieves a 94–99% first-pass metabolism of floxuridine by the liver 17, allowing for significantly higher localized drug concentrations while minimizing systemic toxicity. As a result, the maximum tolerated dose (MTD) of floxuridine administered via HAIP is estimated to be 400 times greater than what is achievable through systemic delivery, and leads to massive tumor cell death within the liver 18. Preclinical studies support combining PDS01ADC with therapies that induce cell death to enhance accumulation of IL-12 within the tumor, and thereby improve efficacy 19–22. We hypothesized that by combining subcutaneous PDS01ADC with HAIP therapy, the robust tumor cell death following HAIP therapy will expose histone-DNA complexes within CRC liver metastases (CRLM) and facilitate IL-12 accumulation, triggering local IFN-γ production, recruitment of cytotoxic immune effectors, and initiation of a robust anti-tumor immune response. Herein, we report the interim safety and efficacy analyses of PDS01ADC in combination with HAIP therapy in patients with metastatic MSS/pMMR CRC 23. As a pre-specified exploratory endpoint, results on clinical activity and safety are compared to a recently conducted, unpublished trial of HAIP therapy (without PDS01ADC) performed by the same care team at the same institution. This comparison provides context to convey the highly cytotoxic effects of HAIP therapy alone in this patient population, which is critical to the central hypothesis that robust cytotoxicity drives the efficacy of PDS01ADC.

Materials and Methods
Detailed Materials and Methods are located in the Online Supplement. Treatment schematic is shown in Supplementary Figure S1.

Eligibility and Ethics
Eligible patients had histologically or cytologically confirmed unresectable colorectal cancer (CRC) metastatic to the liver, had failed at least first-line chemotherapy, were ≥18 years old, and had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1. Of note, unresectable was defined as the inability to extirpate all disease in a single operation and was made based upon pre-operative imaging in conjunction with laboratory/clinical factors related to liver injury from prior therapies and not based upon pre-defined criteria. Both study protocols were approved by the National Institutes of Health Intramural Institutional Review Board (Federalwide Assurance # 00005897), which complies with all applicable United States Regulations, including 45 CFR 46. Written informed consent was obtained from all participants, and the studies were conducted in accordance with institutional and federal guidelines.

Trial Registry
Trials were registered at clinicaltrials.gov:
PDS01ADC + HAIP Therapy Trial Registration ID: NCT05286814

HAIP Therapy Trial Registration ID: NCT03366155

Data Sharing Statement
De-identified individual participant data will be made available to qualified researchers beginning at the time of publication and up to 36 months after publication. Interested investigators should contact [Jonathan.Hernandez@nih.gov]. A data access agreement will be required prior to data release.

Results
Between October 2022 and May 2024, 16 patients with mCRC were screened for PDS01ADC + HAIP therapy; 3 (18.8%) were excluded due to ineligibility. Of 13 enrolled, 2 were excluded due to peritoneal disease at surgery and 2 for clinical deterioration (heparin-induced thrombocytopenia and persistent transaminitis secondary to progressive disease) prior to administration of PDS01ADC. The remaining 9 patients completed at least 1 treatment cycle and comprised the interim analysis cohort. Between June 2019 and June 2022, 27 patients with mCRC were screened for HAIP therapy alone; 4 (14%) were excluded due to ineligibility (Figure 1). Of the remaining 23 enrolled, 2 were excluded due to peritoneal disease at surgery, and 1 was lost to follow-up. Thus, 20 patients received at least 1 full cycle of treatment and were included in this analysis. Patients were non-randomized but balanced between the 2 trials (Table 1 and Supplementary Tables S2–S4).
Termination criteria for NCT05286814 (PDS01ADC + HAIP therapy) under the Simon two-stage design specified that if >4 patients required a dose hold of floxuridine during the first 3 cycles, accrual must stop due to elevated risk of biliary toxicity. Among patients who received PDS01ADC + HAIP therapy, 33.3% (3/9) of patients had floxuridine holds within the first 3 cycles vs. 30% (6/20) of patients who received HAIP therapy alone, meeting the safety criteria for protocol continuation (Supplementary Table S5). Over the first 6 cycles, the quantity of floxuridine administered at each cycle was similar between studies (Figure 2), as was the number of cycles in which floxuridine was administered (Table 2). Median dose holds were 1 (range 0–3) for PDS01ADC + HAIP therapy and 0.5 (range 0–3) for HAIP therapy. Mean floxuridine dose administered over 6 cycles (as % of max cycle 1 dose) was 52.0% (range 33.3–70.8%) for PDS01ADC + HAIP therapy vs. 55.5% (range 16.7–91.7%) for HAIP therapy alone (p=0.70). No patients receiving PDS01ADC had bilirubin levels exceeding 3.0 mg/dL nor developed biliary sclerosis or strictures requiring intervention within 6 months of initiating treatment vs. 1 patient who received HAIP therapy alone (Supplementary Table S6).
Grade ≥3 adverse events (AEs) occurred in 35.0% (7/20) of patients receiving HAIP therapy alone and 77.8% (7/9) of patients receiving PDS01ADC + HAIP therapy within 6 months (p=0.05, Table 2). Among patients receiving HAIP therapy alone, treatment-related AEs occurred in 25.0% (5/20); surgery-related in 15.0% (3/20), HAIP-related in 10.0% (2/20), and chemotherapy-related in 10.0% (2/20). Thromboembolism was most common. In patients receiving PDS01ADC + HAIP therapy, treatment-related AEs occurred in 55.6% (5/9): surgery-related in 22.2% (2/9), PDS01ADC-related in 11.1% (1/9), and chemotherapy-related in 22.2% (2/9). Diarrhea was the most frequent AE, attributed variously to systemic chemotherapy, foodborne illness, or PDS01ADC, and occurred more frequently in patients receiving PDS01ADC compared to patients receiving HAIP therapy alone (p<0.05). No significant difference was observed in treatment-related grade ≥3 AEs (p=0.20).
RECIST responses at 3 months were comparable between the 2 trials, with partial or complete responses observed in 67% (6/9) of patients receiving PDS01ADC + HAIP therapy and 55% (11/20) of patients receiving HAIP therapy alone (p=0.69, Supplementary Figure S2A–B). Though not statistically increased with addition of PDS01ADC, the pre-specified criteria for continuation of NCT05286814 (PDS01ADC + HAIP therapy) in favor of an improved response rate (>5 patients with partial response) was met at 3 months. By 6 months, response rates increasingly diverged: 78% (7/9) of patients receiving PDS01ADC + HAIP therapy demonstrated partial or complete responses compared to 35% (7/20) of patients receiving HAIP therapy alone (p=0.05, Supplementary Figure S2C–D). Median hepatic progression-free survival (PFS) did not differ between patients receiving PDS01ADC + HAIP therapy and those receiving HAIP therapy alone: 12.7 months with a minimum follow up of 13.1 months at the time of data cutoff vs. 10.8 months, respectively (p=0.9, Supplementary Figure S2E). Interestingly, median extra-hepatic PFS was not reached for patients receiving PDS01ADC + HAIP therapy prior to the minimum follow-up of 13.1 months. Notably, extra-hepatic PFS was 8.1 months for patients receiving HAIP therapy alone (p=0.06, Supplementary Figure S2F).
Detailed clinical information and course for each patient are shown in Figure 3. Subsequent liver-directed interventions, such as resection of previously unresectable disease or ablation to achieve no evidence of disease (NED), were more frequently performed in patients receiving PDS01ADC + HAIP therapy (7 of 9 patients) compared to patients receiving HAIP therapy alone (6 of 20 patients; p=0.04). Neither survival nor subsequent interventions were obviously associated with individual patient factors across both studies including prior therapy, presence of lung micronodules at baseline, positive regional lymph nodes, or KRAS mutation status (Figure 3). Median OS in patients treated with HAIP therapy alone was 17.9 months, whereas median OS in patients treated with PDS01ADC + HAIP therapy remained undetermined at a median follow-up of 19.6 months (p=0.02), with survival analysis ongoing (Supplementary Figure S2G).
Limited changes in serum analytes were seen following HAIP therapy alone [at cycle 1 day 15 (C1D15), Figure 4A]; however, multiple changes reflective of enhanced immune activation were observed after patients received PDS01ADC (Figure 4A, Supplementary Figure S3A). For example, following the first dose of PDS01ADC, increases (at C1D17 and C2D1) in pro-inflammatory cytokines [IFN-γ (Figure 4B), TNF-α, IL-10], T cell recruiting chemokines [CXCL9 (Supplementary Figure S3B), CXCL10, CXCL11], and analytes involved in cytotoxicity [GZMB (Figure 4C), GZMA, GZMH] and reflective of enhanced T cell and NK cell activity [CRTAM (Supplementary Figure S3C), KLRD1, NCR1, TRAIL] were noted (Figure 4A). Reductions in immunosuppressive analytes, such as VEGF (Supplementary Figure S3D) and sCD40L (Supplementary Figure S3E), were also observed after PDS01ADC was given (Figure 4A). Ingenuity pathway analysis of these changes showed increases in multiple immune related pathways after PDS01ADC administration, including increases in pathogen-induced cytokine storm signaling and NK cell signaling pathways (Supplementary Figure S3F–G), which were not present with HAIP treatment alone (C1D15).
Changes in circulating immune cell subsets reflective of enhanced immune activation were also seen primarily after patients received PDS01ADC (Figure 4D, Supplementary Figure S4A). Following cycle 1 of PDS01ADC (at C2D1), increases in NKT cells (Supplementary Figure S4B), activated NK cells (NKp30+, PD-1+, or TIM3+), and CD19+ B cells (Supplementary Figure S4C) were noted, whereas decreases in T regulatory cells (Tregs) with a suppressive phenotype (HLADR+, Supplementary Figure S4D–E) were observed (Figure 4D). After 2 cycles of treatment (at C3D1), total CD8+ T cells (Supplementary Figure S4F) and activated CD8+ T cells expressing PD-1 (Supplementary Figure S4G) were increased, whereas myeloid derived suppressor cells (MDSCs, Supplementary Figure S4H) were reduced compared to baseline (Figure 4D), resulting in increased CD8:Treg (Figure 4E) and CD8:MDSC (Figure 4F) ratios. Moreover, an increase in T cells expressing TCF1, a transcription factor crucial in T cell self-renewal 24, was observed after 2 cycles of therapy (Figure 4D, Supplementary Figure S5A–B).
Serum analytes and PBMC subsets were also evaluated for associations with clinical response to PDS01ADC + HAIP therapy. While not statistically significant due to the small number of patients with progressive disease (PD, n=2), patients who developed a complete or partial response at 6 months (n=7) had clear evidence of less immune suppression prior to therapy (e.g., lower circulating TGFβ levels and frequencies of PD-1+ Tregs and plasmacytoid dendritic cells) and evidence of greater immune activation early after PDS01ADC (e.g., higher activated and memory CD4+ and CD8+ T cell frequencies) than patients with PD (Figure 4G). Most notably, clinical responders had higher frequencies of circulating T cell subsets with stem-like features (expressing TCF1) by the end of cycle 1 as well as elevated levels of IFN-γ and multiple other serum analytes involved in T cell costimulation (CD28, CD70), cytotoxicity (GZMA, GZMH) and NK cell activity (NCR1, TRAIL, KLRD1) by the end of cycle 2 (Figure 4G). While several of the peripheral immune measures at post treatment timepoints that were associated with clinical response also significantly changed compared to baseline in analyses of all patients combined, several others did not significantly change with therapy.
Tumor biopsies were available at baseline and at the end of cycle 1 for 8 patients who received PDS01ADC + HAIP therapy. Viable tumor cells were not detected in three samples post treatment, which instead revealed hepatic parenchyma with fibrosis and/or necrosis. All three of these patients are alive without evidence of liver disease at 19–21 months of follow-up. Of note, CD8+ T cells were mostly abundant at baseline and in all biopsies following 1 cycle of PDS01ADC + HAIP therapy. Most of these T cells did not express granzyme B, indicating lack of cytotoxicity, but did express TCF1, suggesting a migration of stem-like T cells into the tumor microenvironment (Supplementary Figure S6). Although the number of samples is too small to draw any conclusions, it is noteworthy that among total CD8+ T cells, the percentage expressing TCF1 increased to the greatest extent in the patient with a complete response, rising from 16.1% at C1D1 to 41.3% at the end of cycle 1 (C2D1). In contrast, a patient with progressive disease who ultimately succumbed at 10 months demonstrated an increase in T regs after receiving PDS01ADC (Figure 4H), leading to a steep decrease in CD8+ T cell: Treg ratio from 200 to 17 at C1D1 and C2D1, respectively. (Supplementary Figure S6).

Discussion
In this interim analysis, we demonstrated that HAIP therapy can be leveraged for tumor-targeted IL-12 delivery, with associated increases in systemic immunity and alterations within the intratumoral immune microenvironment. This is the first study to report on feasibility, clinical activity, and the identification of immune correlates with clinical response in patients with mCRC treated with PDS01ADC and cytotoxic chemotherapy. Importantly, the addition of PDS01ADC did not increase biliary toxicity, which has been observed with prior attempts to improve conventional HAIP therapy 25,26. In a broader sense, the use of PDS01ADC combined with highly active cytotoxic chemotherapy is particularly significant for patients with mCRC given the RECIST responses observed with doublet (50%) or triplet chemotherapy therapy (62%) in the first-line setting 27.
Patients treated with PDS01ADC + HAIP therapy more frequently had partial or complete responses at 6 months. There is no indication of reduced FUDR tolerability or survival detriment due to the addition of PDS01ADC to HAIP therapy at the interim analysis. Importantly, the results described herein are consistent with or improved relative to similar studies of HAIP therapy for mCRC 28–30. Although we observed no difference in hepatic progression-free survival between patients receiving HAIP therapy with or without PDS01ADC, these early results indicate that liver progression may be more amenable to adjunct management strategies, and that systemic progression may be delayed or reduced with the use of PDS01ADC. However, these results are preliminary, and it should be noted that decisions regarding subsequent resection/ablation were made on a per patient basis and not using standardized, uniform criteria.
The use of recombinant IL-12 as a clinical agent was terminated due to toxicity10. Prior preclinical studies have shown that PDS01ADC outperformed rIL-12, both as a monotherapy and in combination with other agents in terms of anti-tumor response due to its ability to target necrotic areas in tumor19,20. More recent studies have also shown the potential clinical utility of PDS01ADC in terms of preliminary evidence of clinical benefit and limited toxicity13,31,32. While we were unable in the current study to evaluate the presence of PDS01ADC in the tumor, prior preclinical work in murine models has shown that PDS01ADC has the ability to target tumors, with enhanced accumulation in tumor tissue when given in combination with necrosis inducing therapies20,22,33. In addition, clinical studies using radiolabeled NHS-antibody have shown its ability to target tumors in patients with lung cancer34. Prior preclinical studies have also shown the ability of PDS01ADC + immune checkpoint blockade to have greater anti-tumor activity than either agent alone, including in a model of colorectal cancer35–37. Several clinical studies are also ongoing employing the combination of PDS01ADC with immune checkpoint blockade (summarized in38,39).
Multiple changes in serum analytes and peripheral immune cell subsets were observed post PDS01ADC, reflective of enhanced broad immune activation. PDS01ADC + HAIP therapy increased frequencies of peripheral CD8+ and CD4+ T cells expressing the transcription factor TCF1, a marker of T cell stemness important for anti-tumor immunity 24,40–43. Clinical responders to the PDS01ADC + HAIP regimen also demonstrated greater frequencies of TCF1+ T cells and evidence of increased immune activation following PDS01ADC administration. These observations are consistent with those of systemic immunomodulation following PDS01ADC monotherapy 38. Moreover, TCF1+ T cells were also identified within tumor core biopsies at both baseline and C2D1 following the first cycle of treatment by PDS01ADC + HAIP.
We recently reported on the importance NK cell evasion in the earliest steps of metastatic outgrowth in the liver and lung through coupled immuno-metabolic reprogramming that results in the suppression of NK cell ligands on the tumor cell surface 44. Intriguingly, we found that PDS01ADC increased NK cell activation and signaling pathways, which could have helped eliminate micrometastatic disease thereby limiting the amount of disease progression. These results suggest that PDS01ADC may augment control of micrometastases, both in the liver and at other sites, through systemic immune modulation. Taken together, our findings of immunomodulation suggest that PDS01ADC functions as an active participant in the treatment regimen and is not simply an inert component or delivered at an irrelevant dose, supporting further clinical and mechanistic evaluation of this combination. We did not evaluate changes in chemokines within tumor biopsies of patients receiving PDS01ADC + HAIP therapy; however, as shown in Figure 4, there was an increase in serum chemokines and cytokines in patients receiving PDS01ADC that was not seen at a timepoint when patients had only received HAIP therapy alone. Due to unavailability of biopsy and PBMCs, we were unable to evaluate immune changes in patients enrolled in the trial evaluating HAIP therapy alone.
This analysis has several limitations, including small number of patients, limited duration of follow-up at a minimum of 13.1 months, and non-randomized study design. Furthermore, comparison to historical controls could potentially introduce temporal variability in post-HAIP therapy management as this was done on a per-patient basis consistent with standard practice rather than by pre-defined criteria. In particular, survival data (OS and extra-hepatic PFS) are immature and while hepatic PFS did reach a median within the minimum duration of follow-up for these 9 patients, each of these findings may likely shift with increased numbers of patients. We urge caution in interpreting comparative results as only a randomized trial can conclusively determine differences in efficacy.
In conclusion, this interim analysis demonstrates that PDS01ADC in combination with HAIP therapy is tolerable for patients without limiting delivery of floxuridine. This combination is feasible for evaluation with promising immunologic and clinical activity, justifying continuation of the study to meet its planned accrual ceiling of 22 patients and further evaluation of clinical and scientific endpoints.

Supplementary Material
Data SupplementAppendix