A Platform for Rapid, Quantitative Assessment of Multiple Drug Combinations Simultaneously in Solid Tumors In Vivo

Abstract

While advances in high-throughput screening have resulted in increased ability to identify synergistic anti-cancer drug combinations, validation of drug synergy in the in vivo setting and prioritization of combinations for clinical development remain low-throughput and resource intensive. Furthermore, there is currently no viable method for prospectively assessing drug synergy directly in human patients in order to potentially tailor therapies. To address these issues we have employed the previously described CIVO platform and developed a quantitative approach for investigating multiple combination hypotheses simultaneously in single living tumors. This platform provides a rapid, quantitative and cost effective approach to compare and prioritize drug combinations based on evidence of synergistic tumor cell killing in the live tumor context. Using a gemcitabine resistant model of pancreatic cancer, we efficiently investigated nine rationally selected Abraxane-based combinations employing only 19 xenografted mice. Among the drugs tested, the BCL2/BCLxL inhibitor ABT-263 was identified as the one agent that synergized with Abraxane® to enhance acute induction of localized apoptosis in this model of human pancreatic cancer. Importantly, results obtained with CIVO accurately predicted the outcome of systemic dosing studies in the same model where superior tumor regression induced by the Abraxane/ABT-263 combination was observed compared to that induced by either single agent. This supports expanded use of CIVO as an in vivo platform for expedited in vivo drug combination validation and sets the stage for performing toxicity-sparing drug combination studies directly in cancer patients with solid malignancies.

Conflict of interest statement

Competing Interests: RK, JD, WK, MG, JC, IT, SD, DT, JF are full time paid employees and shareowners of Presage Biosciences. DP is a full time paid employee and shareowner of Celgene Corporation. MC and IT are former employees of Presage Biosciences.

Figures

Fig 1. CIVO injection of Abraxane® results in localized mitotic arrest but minimal tumor cell death.

MiaPaCa2 tumors (n = 5) were injected using the CIVO device with Abraxane® (ABX) and resected 24 hours post injection. Tissue sections were stained for phospho Histone H3 (pHH3) as a marker of mitotic arrest and cleaved caspase 3 (CC3) as a marker of apoptosis with DAPI as the counterstain. Representative image shows a single site of ABX microinjection with the fluorescent tracking marker (FTM) denoting the site of injection. Representative radial effect curve shows the fraction of CC3+ cells as a function of radial distance from the site of injection.

Fig 2. CIVO multi drug combination analysis in a xenograft model of pancreatic cancer.

Using the CIVO device, MiaPaCa2 tumors were injected with single agents and combinations thereof with Abraxane® (ABX) into the same tumor (n = 4 tumors per single agent/ABX combination pair) and resected 24 hours post injection. Tissue sections were stained with cleaved caspase 3 (CC3) as a marker of apoptosis with DAPI as the counterstain. Representative radial effect curves show fraction CC3+ cells as a function of radial distance from the site of injection as demarcated by FTM. Data are averaged across four tumors. Error bars denote standard error of the mean (SEM). Representative images show cytotoxic responses at injection sites of single agents and combinations thereof with ABX.

Fig 3. CIVO analysis identifies ABT-263 as the agent that leads to the greatest increase in cytotoxic response when combined with Abraxane.

(A) Plots show the difference in fraction CC3+ cells averaged across n = 4 MiaPaCa2 tumors for each agent combined with Abraxane® (ABX) versus the corresponding single agent, as a function of radial distance from the injection site. Among the curves, ABT-263 shows a greater than two-fold higher increase compared to other agents. The increases for other agents were similar to the single-agent effect of ABX alone (dashed curve) suggesting simple additivity of responses for these agents. (B) For statistical comparisons, the areas under each curve were integrated and normalized, dividing by the average area under the ABX curve. The bar chart shows estimates of these integrated differences with standard errors derived from a linear mixed effects model. Of the combinations tested, only ABT-263 plus ABX elicited a statistically significant greater apoptotic response than ABX alone (p<0.01 Wald’s test).

Fig 4. CIVO combination analysis leads to identification of a synergistic combination between ABT-263 and Abraxane® but not with other selective BCL2 inhibitor, ABT-199.

(A Representative images show immunohistochemical staining of target proteins BCL2 (ABT-263 and ABT-199) and BCLxL (ABT-263 only) in MiaPaCa2 xenograft tumors. (B) Using the CIVO device, MiaPaCa2 tumors were injected with vehicle, ABT-199, ABX, ABT-199+ABX or vehicle, ABT-263, ABX, ABT-263+ABX into each of 5 tumors per combination set and resected 24 hours post injection. Tissue sections were stained with cleaved caspase 3 (CC3) as a marker of apoptosis, phospho Histone H3 (pHH3) as a marker of mitotic arrest with DAPI as the counterstain. FTM demarcates the site of injection. Representative images show localized cytotoxic responses induced by ABT-263, ABT-199 and combinations thereof with ABX. (C) Plots show fraction CC3+ cells as a function of radial distance from the site of injection. Data are averaged across five tumors. Error bars denote SEM. (D) A linear mixed effects model was used to estimate the difference (βs(r)) between the response due to each combination versus the sum of the individual responses, which were plotted with 95% confidence intervals for each radial distance. Only ABT-263 showed statistically significant synergy (P<0.05; confidence intervals do not intersect 0) at several radial distances.

Fig 5. Abraxane® combined with ABT-263 leads to 50% complete response and durable tumor remission in a pancreatic cancer model.

(A) MiaPaCa2 xenografted mice (n = 8 per treatment cohort) were treated systemically with vehicle (control), ABT 263 100 mpk PO Days 1–3, Abraxane® (ABX) 30 mpk IV Days 1–3 or combination of ABT 263 and ABX using the same dosing regimen as the single agents. Drug efficacy with respect to vehicle was assessed in all arms via tumor volume measurements and (B) via ex vivo tumor weight measurements. p < 0.05 (two sided Student’s t-test) for ABX+ABT-263 vs ABT263 or ABX. Data are averaged across all tumors in the respective cohorts. Error bars represent SEM. (C) Representative ex vivo images of tumors from each treatment arm on Day 24 and Day 42 (D) Tumor Growth Inhibition (TGI) index was calculated at Day 24 when vehicle and ABT-263 arms met censoring criteria (see Methods) and on Day 42 for the remaining ABX and ABX+ABT-263 arms. p values for TGI comparisons were calculated using the Wilcoxon Rank Sum test.