Neoadjuvant antiangiogenic therapy reveals contrasts in primary and metastatic tumor efficacy

Abstract

Thousands of cancer patients are currently in clinical trials evaluating antiangiogenic therapy in the neoadjuvant setting, which is the treatment of localized primary tumors prior to surgical intervention. The rationale is that shrinking a tumor will improve surgical outcomes and minimize growth of occult micrometastatic disease-thus delaying post-surgical recurrence and improving survival. But approved VEGF pathway inhibitors have not been tested in clinically relevant neoadjuvant models that compare pre- and post-surgical treatment effects. Using mouse models of breast, kidney, and melanoma metastasis, we demonstrate that primary tumor responses to neoadjuvant VEGFR TKI treatment do not consistently correlate with improved post-surgical survival, with survival worsened in certain settings. Similar negative effects did not extend to protein-based VEGF pathway inhibitors and could be reversed with altered dose, surgical timing, and treatment duration, or when VEGFR TKIs are combined with metronomic 'anti-metastatic' chemotherapy regimens. These studies represent the first attempt to recapitulate the complex clinical parameters of neoadjuvant therapy in mice and identify a novel tool to compare systemic antiangiogenic treatment effects on localized and disseminated disease.

Keywords: VEGF; antibodies; neoadjuvant; surgery; tyrosine kinase inhibitors.

Figure 1. Primary tumor response to neoadjuvant sunitinib treatment is not predictive of post-surgical survival in multiple models of metastasis

1. BLI of SCID mice bearing orthotopic human SN12-PM6^LUC+ renal tumors receiving neoadjuvant sunitinib for 14 days.

2. BLI of mice before and after nephrectomy (SN12-PM6^LUC+ model).

3. Corresponding quantification of resected kidney BLI (left panel) and kidney weight (right panel) following neoadjuvant sunitinib treatment cessation (SN12-PM6^LUC+ model).

4. Post-surgical survival (SN12-PM6^LUC+ model).

5. BLI of BALB/c mice bearing orthotopic mouse RENCA^LUC+ renal tumors receiving neoadjuvant sunitinib for 14 days.

6. BLI of mice before and after nephrectomy (RENCA^LUC+ model).

7. Corresponding quantification of resected kidney BLI (left panel) and kidney weight (right panel) following neoadjuvant sunitinib treatment cessation (RENCA^LUC+ model).

8. Post-surgical survival (RENCA^LUC+ model).

9. Heatmap summary of metastatic distribution by visual scoring at individual mouse endpoint following neoadjuvant sunitinib treatment and tumor resection (SN12-PM6^LUC+, RENCA^LUC+, and human WM113/6-4L melanoma tumor model).

Data information: Symbols and bars for box and whiskers plot: median (line), upper/lower quartile (box), min/max (error bars). Survival analysis: hazard ratio (HR), confidence interval (CI), overall survival (OS) based on Kaplan–Meier or Cox regression analysis. N = 8–12 mice per group. BLI, bioluminescence imaging; Neoadj. Tx, neoadjuvant treatment; *P < 0.05, **P < 0.01 compared to control.

Figure 2. Modulating neoadjuvant sunitinib dose and surgical timing can improve post-surgical survival

1. A–C

   SCID mice implanted with LM2-4^LUC+ human breast cancer cells in the mammary fat pad and treated with three neoadjuvant regimens: vehicle, sunitinib (60 mg/kg/day), or CTX MTD (100 mg, 3 times weekly) for 14 days. (A) Comparison of tumor volume by caliper measurement, and (B) comparison of tumor weight following surgery (36 days post-implantation), with images of excised tumors shown (side panel). (C) Post-surgical survival following neoadjuvant CTX MTD or sunitinib treatment.

2. D–G

   SCID mice implanted with LM2-4^LUC+ human breast cancer cells in the mammary fat pad and treated with vehicle or sunitinib (120 mg/kg/day) for 7 days. (D) Comparison of tumor volume by caliper measurement, and (E) comparison of tumor weight following surgery (30 days post-implantation), with images of excised tumors shown (side panel). (F) Post-surgical survival following short-term (high-dose) sunitinib treatment compared to control. (G) Post-surgical survival comparison of short-term sunitinib treatment at either high (120 mg/kg/day) or lower (60 mg/kg/day) doses.

Data information: Symbols and bars for box and whiskers plot: median (line), upper/lower quartile (box), min/max (error bars). Survival analysis: hazard ratio (HR), confidence interval (CI), overall survival (OS) based on Kaplan–Meier or Cox regression analysis. N = 8–15 mice per group. Neoadj. Tx, neoadjuvant treatment; **P < 0.01, ***P < 0.001 compared to control.

Figure 3. Pre-surgical effects of neoadjuvant protein-based VEGF pathway inhibition predict for improved post-surgical survival in multiple metastasis models

1. Comparison of excised orthotopic LM2-4^LUC+ breast tumor weights following 14-day neoadjuvant therapy with VEGF RTKIs (sunitinib or axitinib), protein-based neutralizing antibodies to VEGF (G6.31 or B20), or a VEGFR-2 blocking adnectin (CT322).

2. Forest plot summary of post-surgical Cox regression survival analysis following neoadjuvant treatment cessation for groups described in (A).

3. Combined analysis of pre- and post-surgical effects to assess effects on primary tumor and metastatic growth following 14 days of neoadjuvant treatment with sunitinib, axitinib, or B20. Models include LM2-4^LUC+ (red circle), WM113/6-4L (green square), SN12-PM6^LUC+ (blue triangle), and RENCA^LUC+ (purple diamond).

4. Corresponding values for primary tumor burden, survival hazard ratio, and Spearman coefficient analysis (see Materials and Methods for details).

Data information: Symbols and bars for box and whiskers plot: median (line), upper/lower quartile (box), min/max (error bars). Survival analysis: hazard ratio (HR), confidence interval (CI), overall survival (OS) based on Kaplan–Meier or Cox regression analysis. N = 6–15 mice per group. Treatment (Tx), vehicle control (Veh). Open symbols were used to indicate data points from animals that were still alive when the experiments were terminated. Crossed lines represent the standard deviation of the vehicle-treated (gray cross) and drug-treated (black cross) from primary tumor burden data (vertical line) and median survival data (horizontal line). *P < 0.05, **P < 0.01, ***P < 0.001 compared to control.

Figure 4. Short-term treatment ‘preconditioning’ in an experimental metastasis model shows range of host-mediated effects on survival

1. A, B

   Numerous anticancer treatments, including multiple VEGF pathway inhibitors, were administered during a 7-day period prior to i.v. tumor inoculation to evaluate effects on overall survival. Cox regression survival analysis summary for experiments in SCID mice inoculated with human breast LM2-4^LUC+ cells (1.5 × 10⁶ cells) (A) and human melanoma MeWo cells (1 × 10⁶ cells) (B). Treatments and doses include XRT (5 Gy/1×), MTD CTX (100 mg/kg/3×), LDM CTX (20 mg/kg/DW), LDM UFT (15 mg/kg/D), LDM CTX/UFT, OXI4503 (50 mg/kg/1×), sunitinib-120 (120 mg/kg/D), DC101 (800 μg/3×), CT322 (100 mg/kg/3×), B20 (5 mg/kg/3×), G6.31 (5 mg/kg/3×), crizotinib (50 mg/kg/D). Terms used: three times (3×), one time (1×), drinking water (DW), daily (D), hazard ratio (HR), treatment (Tx), vehicle control (Veh), radiation (XRT), maximum tolerated dose (MTD), low-dose metronomic (LDM), and cyclophosphamide (CTX). For treated groups in (A), N = 7–10 mice, for vehicle groups, N = 29 mice. For all groups in (B), N = 7–8 mice. Survival analysis: hazard ratio (HR), confidence interval (CI), overall survival (OS) based on Kaplan–Meier or Cox regression analysis. See Supplementary Figs S4 and S5 for individual survival curves. *P < 0.05; ***P < 0.001 compared to control.

Figure 5. Rational combination of sunitinib and low-dose chemotherapy can improve neoadjuvant benefits before and after surgery

1. Comparison of orthotopic LM2-4^LUC+ breast tumors weights surgically removed following neoadjuvant therapy with LDM CTX/UFT combination for 14 days.

2. Following treatment cessation and surgical resection of primary tumor from (A), comparison of post-surgical survival following LDM CTX/UFT treatment compared to vehicle controls.

3. In a similar comparison, mice bearing WM113/6-4L orthotopic melanoma tumors were given neoadjuvant treatment with neoadjuvant LDM CTX/VBL, sunitinib (60 mg/kg/day), or a combination for 14 days and tumor weights compared following surgical resection.

4. Comparison of post-surgical survival in mice from (C).

5. Combined analysis of pre- and post-surgical effects shown in (C) and (D) to assess effects on primary tumor and metastatic growth following 14 days of neoadjuvant treatment with LDM CTX/VBL, sunitinib (60 mg/kg/day), or a combination of both together.

Data information: Symbols and bars for box and whiskers plot: median (line), upper/lower quartile (box), min/max (error bars). Survival analysis: hazard ratio (HR), confidence interval (CI), overall survival (OS) based on Kaplan–Meier or Cox regression analysis. N = 7–9 mice per group. Open symbols were used to indicate data points from animals that were still alive when the experiments were terminated *P < 0.05 compared to control.

Figure 6. Comparison of primary tumor and overall survival benefits to define neoadjuvant efficacy score

1. Neoadjuvant efficacy score (NES) was defined as the difference between measured benefit of primary tumor response to therapy, and overall survival (OS) benefit after therapy is stopped and tumor resected. All values are compared to control with negative (−) and positive (+) indicating improvement or detriment for primary tumor benefit, respectively, and vice versa for overall survival benefit.

2. NES values for all therapies tested in breast, melanoma, and kidney models shown in Figs3 and 5. See Materials and Methods for NES value determination details.