Inhaled recombinant human IL-15 in dogs with naturally occurring pulmonary metastases from osteosarcoma or melanoma: a phase 1 study of clinical activity and correlates of response

Abstract

Purpose: Although recombinant human interleukin-15 (rhIL-15) has generated much excitement as an immunotherapeutic agent for cancer, activity in human clinical trials has been modest to date, in part due to the risks of toxicity with significant dose escalation. Since pulmonary metastases are a major site of distant failure in human and dog cancers, we sought to investigate inhaled rhIL-15 in dogs with naturally occurring lung metastases from osteosarcoma (OSA) or melanoma. We hypothesized a favorable benefit/risk profile given the concentrated delivery to the lungs with decreased systemic exposure.

Experimental design: We performed a phase I trial of inhaled rhIL-15 in dogs with gross pulmonary metastases using a traditional 3+3 cohort design. A starting dose of 10 µg twice daily × 14 days was used based on human, non-human primate, and murine studies. Safety, dose-limiting toxicities (DLT), and maximum tolerated dose (MTD) were the primary objectives, while response rates, progression-free and overall survival (OS), and pharmacokinetic and immune correlative analyses were secondary.

Results: From October 2018 to December 2020, we enrolled 21 dogs with 18 dogs reaching the 28-day response assessment to be evaluable. At dose level 5 (70 μg), we observed two DLTs, thereby establishing 50 µg twice daily × 14 days as the MTD and recommended phase 2 dose. Among 18 evaluable dogs, we observed one complete response >1 year, one partial response with resolution of multiple target lesions, and five stable disease for an overall clinical benefit rate of 39%. Plasma rhIL-15 quantitation revealed detectable and sustained rhIL-15 concentrations between 1-hour and 6 hour postnebulization. Decreased pretreatment lymphocyte counts were significantly associated with clinical benefit. Cytotoxicity assays of banked peripheral blood mononuclear cells revealed significant increases in peak cytotoxicity against canine melanoma and OSA targets that correlated with OS.

Conclusions: In this first-in-dog clinical trial of inhaled rhIL-15 in dogs with advanced metastatic disease, we observed promising clinical activity when administered as a monotherapy for only 14 days. These data have significant clinical and biological implications for both dogs and humans with refractory lung metastases and support exploration of combinatorial therapies using inhaled rhIL-15.

Keywords: clinical trials as topic; cytokines; killer cells, natural; melanoma; sarcoma.

Figure 1

Study design and assessment of clinical responses. (A) Schema of first-in-dog clinical trial of inhaled rhIL-15 delivered by nebulizer twice daily for 14 total days. Blood was drawn prior to treatment as well as on days 7, 14 and 28, and response was evaluated by chest radiograph (CXR) on days 28, 42, 70, and 98. (B) Waterfall plot demonstrating percent change of target tumor sizes in responders and non-responders. Progressive disease (PD) ≥20% increase from baseline, and partial response (PR) >30% decrease from baseline. (C) Spider plot summarizing changes in sum of each patient’s longest tumor diameters from baseline over time. Evaluable patients were considered responders (blue) or non-responders (red) based on RECIST criteria. Stable disease was defined as within the thresholds of <20% increase and <30% decrease in tumor diameter as illustrated by dashed lines. (D) Representative radiographs illustrating changes in pulmonary lesions prior to and following treatment with inhaled rhIL-15. Patient 4 (top panel) experienced a PR with complete resolution of three out of five lesions at dose level 1, while patient 14 (bottom panel) showed a complete response with resolution of diffuse pulmonary metastatic lesions leading to remission that lasted >1 year after completion of treatment. (E) Survival from first inhaled rhIL-15 treatment for responders/patients having clinical benefit versus non-responders. (F) Kaplan-Meier curve from initiation of treatment showing a trend for improved survival among responders/patients with clinical benefit. rhIL-15, recombinant human interleukin-15.

Figure 2

Plasma rhIL-15 quantitation and cytokine response. (A) Samples were collected prior to first inhaled rhIL-15 treatment and then at indicated time points for 6 hours after treatment initiation and measured by ELISA. Changes in plasma rhIL-15 concentrations were plotted over time for two patients in each of the 33–70 µg cohorts. (B) Concentration of rhIL-15 in plasma of patients in the 33 µg, 50 µg, and 70 µg inhaled rhIL-15 cohorts as well as those from a separate canine clinical trial where dogs with locally advanced melanoma were treated with palliative RT, allogeneic NK transfer, and 3 µg /kg rhIL-15 SQ to promote NK engraftment. Bars represent plasma rhIL-15 pretreatment and at the presumed Tmax of 4 hours post-treatment for each patient. Plasma rhIL-15 was detectable in all patients but below the limit of quantitation in several patients. Individual groups were compared using one-way analysis of variance with Tukey’s multiple-comparison test. ***P<0.001; **p<0.01. (C) Cytokines induced by rhIL-15, IL-8 and KC-like were quantified in the plasma of all patients (left) as well as separately within the SQ rhIL-15 trial cohort (center) and inhaled rhIL-15 cohorts (right) using a canine multiplex assay. (D) Four-hour KC-like levels (left) were significantly higher in responders compared with non-responders (p=0.046), and a similar trend was noted with 4-hour IL-8 levels (middle) (p=0.092). Responders and non-responders did not show differences in 4-hour rhIL-15 levels (right). rhIL-15, recombinant human interleukin-15; SQ, subcutaneous.

Figure 3

Baseline lymphopenia is associated with response to therapy. (A) Baseline absolute lymphocyte counts (ALCs) were determined. Day 0 ALC was lower among responders compared with non-responders (p=0.03 by Kruskal-Wallis Test). (B) Fold change from baseline ALC was on average 0.32 higher in responders than in non-responders (SE=0.12, p=0.020). (C) There is no difference in baseline neutrophil counts between responders and non-responders (p=0.3). (D) Neutrophil counts were higher in non-responders at all time points, but these differences were not statistically significant (p=0.16). (E) Representative flow cytometry staining is shown for PBMCs, including parent gating and CD3+, CD8+, and Nkp46 +lymphocyte subsets. NKp46 staining is shown for one patient for all time points along with primary cultured NK cells and Fluorescence Minus One (FMO) of positive and negative controls. (F) Frequencies of CD3+, CD8 + and NKp46 +lymphocyte subsets in the peripheral blood in patients over time. (G) Fold change of CD3+, CD8 + and NKp46 + lymphocyte subsets were calculated compared with day 0 values as the reference. (H) Multidimensional scaling (MDS) plots based on mathematical distances of differential gene expression for CD5 depleted versus CD5^bright subsets using RNA sequencing on PBMC samples at indicated time points. Plot demonstrates significant differences in clustering. (I–K) MDS plot for differential gene expression of CD5 depleted PMBCs enriched for NK cells showing mathematical distances of gene profile based on inhaled IL-15 dose, cancer diagnosis, and best clinical response, respectively. (L) Heatmap of 21 genes of interest demonstrates subset of genes induced in CD5 depleted PBMCs at time points postinitiation of inhaled IL-15 compared with day 0 gene expression. NK, natural killer; PBMC, peripheral blood mononuclear cell.

Figure 4

Cytotoxic function of patient PBMCs pretherapy and post-therapy. (A) Representative flow cytometry gating distinguishes PBMC effectors from CFSE-labeled OSA and melanoma target cells with dead cells staining positive for Viability Dye 780. (B) Cytotoxicity, calculated by flow cytometry, of PBMCs against osteosarcoma (OSCA-78) and melanoma (M5) cell lines was significantly increased postinhaled rhIL-15 treatment (p<0.001). PBMCs targeting OSCA had significant correlation (Spearman correlation coefficient, ρ) of (C) maximal cytotoxicity and (D) change in cytotoxicity with survival. (E) Maximal cytotoxicity of PBMCs targeting M5 did not show significant correlation to survival. (F) Minimal correlation existed between change in cytotoxicity and survival in PBMCs targeting M5. Representative examples of changes in cytotoxicity, at 1:1 effector to M5 (red) or OSCA (blue) target ratios, in a patient with (G) stable disease (SD), (H) complete response (CR), and (J) progressive disease (PD). PBMCs, peripheral blood mononuclear cells.

Figure 5

Plasma cytokine levels. Plasma cytokine values measured by canine Luminex assay are depicted comparing responders to non-responders over time: (A) cGM-CSF, (B) cINFγ, (C) cIL-2, (D) cIL-6, (E) cIL-7, (F) cIL-8, (G) cIL-10, (H) cIL-15, (I) cIL-18, (J) cCXCL10, (K) cKC-like, (L) cMCP-1, and (M) cTNFα. Fold change of (N) cINFγ, (O) cIL-2, (P) cIL-6, and (Q) cIL-18 levels fluctuated over time but were not significantly different between responders and non-responders. Concentrations of (R) cGM-CSF, (S) cIL-6, (T) cIL-7, (U) cIL-10, (V) cIL-15, and (W) cTNFα at baseline were higher in patients with OSA compared with those with melanoma with a significant difference observed in cIL-10 (p=0.004). OSA, osteosarcoma.