Altered Gemcitabine and Nab-paclitaxel Scheduling Improves Therapeutic Efficacy Compared with Standard Concurrent Treatment in Preclinical Models of Pancreatic Cancer

In the United States, there will be an estimated 56,770 new cases of pancreatic carcinoma and 45,750 estimated deaths in 2019 (1). The 5-year overall survival (OS) rate remains dismal at only 10%, and consequently pancreatic ductal adenocarcinoma (PDAC) is projected to become the second leading cause of cancer-related deaths by 2030 (2). With curative surgery, 5-year survival rates approach 27%, but distant failure rates remain high at 44% (3). Historically, gemcitabine was the chemotherapy of choice for both metastatic and resected PDAC on the basis of clinical trial results which showed an improved OS benefit and greater response rates compared with observation of 5-fluoruracil (5-FU) chemotherapy in both the metastatic and adjuvant setting (4, 5). Recently, the combination regimens of either 5-FU/leucovorin with irinotecan and oxaliplatin (FOLFIRINOX) or gemcitabine and nanoparticle albumin-bound paclitaxel (NP, nab-paclitaxel or Abraxane) are now standard of care in multiple settings based on phase III clinical trials showing superior outcomes with either of the combination chemotherapy regimens over gemcitabine alone (6–8). However, minimal attention is paid to the effects of sequencing different combinations of chemotherapy agents before clinical testing.

Caveolin-1 (Cav-1) is the principal structural component of caveolae and is a key player in cellular endocytosis (9). PDAC has been shown to have high expression of Cav-1, and higher Cav-1 expression is associated with poor clinical outcomes, enhanced tumor progression, and resistance to gemcitabine in PDAC preclinical studies (10–12). Furthermore, knockdown of Cav-1 reduces PDAC uptake of albumin and NP, leading to reduced response to NP (13). For patients receiving gemcitabine and NP, the current standard of care is concurrent treatment of the drug combination on the same day. Interestingly, we observed that gemcitabine treatment of PDAC cells led to an increase in Cav-1 expression after 24–48 hours (10). We, therefore, hypothesized that pretreatment with gemcitabine could be a strategy to increase NP uptake into PDAC cells through caveolae-mediated endocytosis, and ultimately result in greater efficacy of the drug combination. Herein, we provide evidence that gemcitabine treatment up to 48 hours prior to NP treatment increases the therapeutic efficacy of the combination, and represents a novel, simple, and highly translatable approach to potentially significantly improve outcomes for patients with PDAC.

Gemcitabine (Hospira) was dissolved in water. Nab-paclitaxel (Abraxane, provided by Celgene) was dissolved in 0.9% saline. Trametinib (Selleckchem) and RO-3306 (MilliporeSigma) were dissolved in DMSO. RPMI1640 media, minimum essential media (MEM), McCoy 5a media, penicillin (100 U/mL)–streptomycin (100 μg/mL), and 0.25% w/v trypsin/1 mmol/L EDTA were purchased from Gibco Life Technologies. DMEM and PBS were purchased from GE Healthcare Bio-Sciences. FBS and lyophilized powder Human Serum Albumin (HSA) were purchased from MilliporeSigma. Anti-cleaved caspase-8, cleaved caspase-7, cleaved caspase 3, cleaved PARP, human albumin, and GAPDH primary antibodies were purchased from Cell Signaling Technology. Caveolin-1 primary antibody (N-20) was purchased from Santa Cruz Biotechnology. Anti-rabbit immunofluorescence secondary antibodies were purchased from LI-COR Biosciences.

AsPC-1, HPAF-II, MIA-PaCa-2 (MP2), BxPC3, Capan-2, and CFPAC-1 cells were obtained from ATCC and authenticated (via short tandem repeat profiling). FHs-74Int cells were also obtained directly from ATCC. Mouse PDAC cells derived from a pancreatic tumor of a LSL-KRas^G12D, LSL-Trp53^−/−, PDX-1-CRE, (KPC) genetically engineered mouse model of PDAC and transfected with enhanced firefly luciferase (KPC-Luc), as described previously, were provided by Z. Cruz-Monserrate (14, 15). Stable short hairpin RNA (shRNA) Cav-1–knockdown MP2 cells were generated as described previously (13). Cells were maintained at 37°C in 5% CO₂ in RPMI1640 media (AsPC-1), MEM (HPAF-II), DMEM (MP2, G37, and KPC-Luc), or ATCC Hybri-Care Medium 46-X (DMEM-like) + 30 ng/mL EGF (FHs 74Int) supplemented with 10% FBS and 1% penicillin–streptomycin. Cells were cultured for no more than 3 months continuously. Gemcitabine, nab-paclitaxel, trametinib, and RO-3306 were added to media with a vehicle final concentration of no more than 0.1%.

G37, a patient-derived xenograft tumor model (a low Cav-1–expressing model obtained from J.G. Trevino) established from a patient with surgically resected pancreatic adenocarcinoma, was passaged in the flanks of NOD-scid IL2Rgamma^null (NSG) mice using 2–3 mm³ tumor fragments. G37 tumors were harvested 30–60 days postimplantation, and a portion (6–8 mm in diameter) of tumor was cut into fine tissue particles. These were trypsinized at 37°C for 10 minutes, neutralized by DMEM containing 10% FBS, and then passed through 21-G needles followed by 25-G needle three times each, before transfer through 100-μm cell strainers. The pass-through was transferred onto a fresh 10-cm dish coated with 6–10 μg/cm² collagen (MilliporeSigma). This tumor cell mixture was cultured in DMEM containing 10% FBS and 10 nmol/L Dexamethasone (MilliporeSigma) in 37°C incubators. Following 7–14 days of growth, G37 cells were passaged and subcultured for future use.

alamarBlue proliferation assay was performed according to the manufacturer’s instructions (Bio-Rad Antibodies). Briefly, cells were seeded in 96-well plates in six replicates at a density of 1,000–5,000 cells per well in 100 μL medium (day 0). Between 48 and 72 hours after plating, alamarBlue reagent was added and incubated at 37°C for 4–8 hours, and absorbance was measured at 570 and 600 nm.

Cells were seeded at 1,000–2,000 cells per well in 96-well plates. The next day, cells were treated according to schedule. Cell confluence, as a measure of cell growth over time, was monitored every 4 hours for up to 4 days using the IncuCyte ZOOM Live-Cell Imaging System (Essen Biosciences).

Cells were trypsinized to generate single-cell suspensions. AsPC-1 (1,500 cells/plate), HPAF-II (1,500 cells/plate), G37 (3,000 cells/plate), and KPC-Luc (1,000 cells/plate) cells were seeded overnight onto 60-mm dishes. Plates were treated and allowed to grow until sufficiently large colonies were observed in the control plate (9–14 days after seeding). Colonies were fixed with methanol/acetic acid, and stained with 1% Crystal Violet (MilliporeSigma). The numbers of colonies or colony forming units containing at least 50 cells were counted using a Dissecting Microscope (Leica Microsystems, Inc.).

Cells were plated on coverslips and treated with or without gemcitabine. Two hours prior to fixation, cells were pulsed with 100 nmol/L nab-paclitaxel or molar equivalent concentrations of HSA. To remove membrane-bound albumin, two acid/salt washes were performed with 0.1 mol/L glycine and 0.1 mol/L NaCl, pH 3.02, on ice for 2 minutes each, followed by two washes with PBS to remove membrane-bound albumin. Cells were then fixed with 2% PFA for 15 minutes at room temperature, and washed with PBS two times for 5 minutes each. Cells were incubated in 1% Triton-x-100 for 10 minutes on ice to permeabilize, and then washed twice with PBS prior to blocking with 3% BSA in PBS overnight at 4°C. In a humidified chamber, primary antibodies, diluted 1:50 in blocking buffer, were added and incubated for 1 hour at 4°C, followed by three 10-minute rinses with blocking buffer. Secondary antibody (conjugated to Alexa Fluor 488) was added along with DAPI for 1 hour at room temperature. Cells were then rinsed, and coverslips were mounted onto slides and then sealed. Cells were then imaged with a confocal microscope.

For assessment of Cav-1 expression, Octyl-β-D-glucopyranoside (MilliporeSigma) was added at 60 mmol/L final concentration to RIPA Buffer (Thermo Fisher Scientific) containing Protease and Phosphatase Inhibitor Cocktails (Roche). Protein concentration was determined with a Dc Protein Assay Kit (Bio-Rad). For albumin immunoblots, two acid/salt washes with 0.1 mol/L glycine and 0.1 mol/L NaCl, pH 3.02, were performed on ice for 2 minutes each, followed by a PBS wash. Proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were incubated in 5% BSA in TBS-Tween blocking buffer for 1 hour at room temperature. Primary antibodies were allowed to bind overnight at 4°C, and used at a dilution of 1:500–1,000. After washing in TBS-Tween for three times for 10 minutes each, the membranes were incubated with immunofluorescence secondary antibodies at a 1:5,000 dilution for 1 hour at room temperature. Membranes were washed with TBS-Tween prior to imaging via LI-COR Odyssey CLx Imaging System (LI-COR).

Mass spectrometry for determination of intracellular paclitaxel concentrations following NP treatment was performed as described previously (13). Briefly, NP- (100 nmol/L) or paclitaxel- (100 nmol/L) treated and pretreated gemcitabine (50 nmol/L) samples in both AsPC-1 and HPAF-II cells were trypsinized, washed with cold PBS, and cell pellets were resuspended with 300 μL of 2:1 acetonitrile to water mixture. Metabolites were extracted by freeze thawing the cell suspension in aqueous-acetonitrile followed by vortexing for 45 seconds. This procedure was repeated three times, and then centrifuging the contents at 13,000 rpm for 10 minutes at 4°C separated cell debris. The supernatants were separated from the debris and immediately the samples were analyzed in LC/MS Triple Quad (LC-MS QQQ, Agilent Technologies, 6430) instrument for quantitative assessment of paclitaxel in the intracellular compartment.

Following treatment, cells were trypsinized to generate single-cell suspensions. Cells were pelleted at 500 × G for 5 minutes and washed with PBS. Cells were pelleted again and resuspended in 400 μL ice-cold PBS. Ice-cold 100% ethanol (800 μL) was added drop wise while mixing via vortexing on low speed, then incubated at 4°C overnight. Fixed cells were allowed to equilibrate to room temperature and gently resuspended by pipetting. Cells were pelleted at 500 × G for 5 minutes and washed with PBS followed by pelleting again. Pellet was resuspended in 200 μL Propidium Iodide (Roche) staining solution containing RNase (MilliporeSigma) and allowed to incubate for 30 minutes at 37°C protected from light. Cells were put on ice, still protected by light, and immediately taken to Flow Cytometer (BD FACSCalibur, BD Biosciences) for analysis.

Animal studies were conducted in accordance with an approved protocol adhering to the Institutional Animal Care and Use Committee policies and procedures at The Ohio State University (Columbus, OH). Six- to 8-week-old male athymic nude mice (Taconic Farms Inc.) were caged in groups of 5 or less, and fed a diet of animal chow and water ad libitum. AsPC-1, HPAF-II, and G37 cells were injected subcutaneously into the flanks of each mouse at 2 × 10⁶ (AsPC-1 and HPAF-II) or 1 × 10⁶ (G37) cells per injection. Treatment regimens were started once tumors reached approximately 100–200 mm³ in size, 1–3 weeks postinjection. Nab-paclitaxel (20 mg/kg) in 0.9% saline was administered intravenously via retro-orbital injection, and 50 mg/kg gemcitabine in water was administered via intraperitoneal injection accordingly. To obtain a tumor growth curve, perpendicular diameter measurements of each tumor were measured every 1–5 days from the first day of injection with digital calipers, and volumes were calculated using the formula (L × W × W)/2. For ultrasound imaging of the pancreatic tumors following orthotopic injection of 1 × 10⁴ G37 cells, we utilized our institutional Vevo 2100 ultrasound imaging system. Following anesthetization using isoflurane, mice were placed supine on the imaging platform. A MS200 transducer was placed on the ultrasound gel over the abdomen. Following identification of the pancreas, a 3D image was obtained using a 3D transducer setup of the entire abdomen starting superior at the diaphragm and ending inferior at the bladder. Tumors were identified on this 3D image and two measurements were recorded, length and depth. Volume was calculated on the formula V = (4/3) × π × (L/2) × (L/2) × (D/2).

Data are presented as the mean ± SEM for proliferation assays, mass spectrometry data, and tumor growth experiments. The group comparisons of the percent change in tumor volume were performed at individual timepoints. Statistical comparisons were made between the control and experimental conditions using the unpaired two-tailed Student t test with significance assessed at P < 0.05. For the orthotopic in vivo study, a one-tailed Student t test was used because our xenograft modeling showed better outcomes with altered sequencing versus normal sequencing. GraphPad Prism (GraphPad Software Inc.) was used to perform the statistical analyses.

AsPC-1 (KRAS^G12D; p53^C135fs) and HPAF-II (KRAS^G12D; p53^P151S) PDAC cell lines have relatively low Cav-1 expression compared with other PDAC cell lines (10). We first tested whether gemcitabine could increase Cav-1 levels in AsPC-1 and HPAF-II cells. These two PDAC cells were treated with gemcitabine (50 nmol/L) for 8, 24, and 48 hours followed by isolation of protein and RNA for analysis of Cav-1 expression. Similar to the findings we observed in other PDAC cell lines (10), gemcitabine treatment for 48 hours resulted in an approximately 2-fold increase in both RNA and protein Cav-1 expression in AsPC-1 cells (P < 0.05; Fig. 1A and C) and a 6- and 2-fold increase in RNA and protein expression, respectively, for HPAF-II cells (P < 0.0001; Fig. 1B and D). Similar results were also observed in the G37 cell line and the murine KPC-Luc pancreatic tumor–derived cell line (Supplementary Fig. S1, top). Next, we tested whether gemcitabine treatment would increase albumin uptake in AsPC-1 and HPAF-II cells in vitro. Gemcitabine treatment of PDAC cells for 24 or 48 hours followed by a 1-hour pulse of HSA increased the amount of intracellular albumin detected via immunoblotting compared with the control-treated sample (HSA pulse without gemcitabine pretreatment; Fig. 1E and F). Extended 7-day treatment of AsPC-1 and HPAF-II cells with gemcitabine demonstrated that Cav-1 expression was maintained over extended time periods (Fig. 1G). Next, we measured the albumin bound to paclitaxel in NP by immunofluorescence in both AsPC-1 and HPAF-II cells following a 1-hour pulse of NP (Fig. 2A and B). Pretreatment with gemcitabine for 24 hours followed by a 1-hour NP pulse on day 2 or day 3 after gemcitabine treatment resulted in a significant increase in intracellular albumin signal compared with the untreated NP-treated control cells (Fig. 2A–D; P < 0.0001). To directly quantify intracellular NP, we performed mass spectrometry analysis and measured uptake of paclitaxel in both AsPC-1 and HPAF-II cells. We found that gemcitabine pretreatment of 24 or 48 hours increased intracellular paclitaxel uptake in both PDAC cell lines treated with NP for 1 hour compared with NP alone and gemcitabine delivered concurrently with NP (Fig. 2E and F; Supplementary Fig. S2). Taken together, these results suggest that gemcitabine pretreatment for up to 48 hours results in increased Cav-1 expression, as well as albumin and NP uptake, leading to increased intracellular paclitaxel in vitro.

Next, we tested whether treating PDAC cells in vitro with gemcitabine prior to NP could improve the therapeutic efficacy of the drug combination. We first determined the IC₅₀ dose for gemcitabine in various PDAC cell lines (Supplementary Fig. S1, bottom). We next compared the therapeutic efficacy for the current clinical standard of gemcitabine and NP delivered on the same day [Gem(D1)NP(D1)] against two altered schedules: (i) gemcitabine treatment on day 1 for 24 hours, followed by replacement with media lacking gemcitabine, but containing NP on day 2 for 24 hours [Gem(D1)NP(D2)] or (ii) gemcitabine treatment on day 1 for 24 hours, followed by replacement with media lacking gemcitabine for 24 hours, followed by NP treatment on day 3 for 24 hours [Gem(D1)NP(D3)]. The IncuCyte and alamarBlue assays were utilized to assess tumor cell proliferation with different treatment regimens in vitro. There was a significant reduction in cell proliferation in both AsPC-1 and HPAF-II cells treated with any of the three combination gemcitabine and NP schedules as compared with the vehicle control (Fig. 3A–D). Both the Gem(D1)NP(D2) and Gem(D1)NP(D3) schedules resulted in significant reductions in growth rates in both the alamarBlue and IncuCyte assays compared with the Gem(D1)NP(D1) schedule (Fig. 3A–D; P < 0.01). In addition, Gem(D1)NP(D3) had a further significant reduction in cell proliferation compared with the Gem(D1)NP(D2) schedule after 72 hours in the alamarBlue assay (Fig. 3A and B; P < 0.01). We next tested AsPC-1 and HPAF-II cells in the colony forming assay under different drug schedules. Survival was significantly reduced for both the Gem(D1)NP(D2) and Gem(D1)NP(D3) schedule compared with Gem(D1)NP(D1) schedule with one cycle of therapy (Fig. 3E–H). In addition, tumor cell survival was significantly reduced in the Gem(D1)NP(D3) schedule compared with Gem(D1)NP(D2) in both AsPC-1 and HPAF-II cell lines (Fig. 3E–H). These results were reproduced in the G37 and KPC cell lines (Supplementary Fig. S3A–S3D). Furthermore, we performed extended time course colony formation assays with multiple “cycles” of the treatment regimens and similarly found that altered scheduling produced the most profound reductions in colony formation (Fig. 3E–H). Interestingly, we tested higher Cav-1–expressing PDAC cell lines (BxPC3, Panc-1, and CFPAC-1) with the altered scheduling regimen and found heterogeneity of response between the Gem(D1)NP(D1) and the Gem(D1)NP(D2) scheduling (Supplementary Fig. S3E–S3G). Consistent with decreased cell survival noted with the Gem(D1)NP(D3) schedule, apoptotic markers, such as cleaved PARP and cleaved caspase 8, 7, and 3, were all increased in the Gem(D1)NP(D3) schedule (Fig. 4A). We next tested whether altering the schedule of the drug combination would impact normal cell growth. Normal small intestine FHs-74 Int cells were treated as described for the PDAC cells, and cell growth was measured by the IncuCyte assay over 136 hours (Supplementary Fig. S4). There were no observed significant differences in cell growth between any of the different chemotherapy treatment schedules.

Next, to determine whether Cav-1 expression is critical for the observed gemcitabine-induced albumin uptake, we performed stable knockdown of Cav-1 by shRNA (vs. control scrambled shRNA) in a relatively high-Cav-1–expressing cell line, MP2 (Fig. 4A). We confirmed that gemcitabine treatment for 24 hours increased both Cav-1 expression and albumin uptake in the control shRNA MP2 cells. However, in the Cav-1–depleted MP2 cells (shCav-1), there was less intracellular albumin uptake following gemcitabine treatment, compared with the shRNA control cells (Fig. 4A). This result supports the important role of Cav-1 in gemcitabine-induced albumin uptake. Next, to confirm that Cav-1 is partly responsible for the increased therapeutic efficacy of the altered scheduling in vitro, we performed colony formation again for the different schedules of gemcitabine and NP in both the MP2 sh-control and MP2 sh-Cav-1 cells. In the sh-control cells, Gem(D1)NP(D3) resulted in a significant reduction in survival compared with either the Gem(D1)NP(D1) or Gem(D1)NP(D2), similar to the results seen in the AsPC-1 and HPAF-II cells. However, in the sh-Cav-1 MP2 cells, the benefit of the altered scheduling of gemcitabine and NP was abrogated (Fig. 4B).

The mechanism of action of paclitaxel is to stabilize microtubules during mitosis and prevent proper cell separation into two daughter cells (16). Therefore, one strategy to optimize NP response might be to increase the proportion of cells in the G₂–M-phase of the cell cycle when paclitaxel is administered to tumor cells. We speculated that gemcitabine treatment could be altering distribution of PDAC cells in the cell cycle allowing for enhanced NP-mediated cell killing at 48 hours after gemcitabine. To assess this, AsPC-1 and HPAF-II cells were treated with gemcitabine for 24 hours followed by exchange with media lacking gemcitabine for an additional 48 hours. Cells were fixed at 12-hour timepoints and stained with propidium iodide for cell-cycle analysis. We noted a reduction in the percentage of cells in the G₂–M-phase within the first 24–36 hours after gemcitabine treatment was initiated in both AsPC-1 and HPAF-II cells, likely due to intra-G₁- and S-phase arrest mechanisms (Fig. 5A). Interestingly, at 24–36 hours following the removal of gemcitabine, there was an observed “rush” of cells into the G₂–M-phase at 48–60 hours from treatment initiation (Fig. 5B and C), likely due to release of cells from G₁- and S-phase arrest. Taken together, it is possible that this progression of cells into G₂–M-phase at about the time when NP is administered is another mechanism that underlies the benefit of altered scheduling of gemcitabine and NP.

To further explore this hypothesis, we used drugs known to synchronize cells into either the G₀–G₁- or G₂–M-phase. First, we used the MEK inhibitor, trametinib, to arrest cells in the G₀–G₁-phase of the cell cycle. After 24 hours of trametinib treatment of AsPC-1 cells, the percentage of cells in the G₀–G₁-phase of cell cycle increased from 57.3% to 92.9% (Fig. 5D). In colony forming assays, trametinib treatment for 24 hours resulted in no significant changes in clonogenic survival (Fig. 5E). Similarly, the addition of trametinib concurrently with NP on day 1 [Tram(D1)NP(D1)] did not alter the effectiveness of NP. However, when PDAC cells were pretreated with trametinib for 24 hours prior to NP treatment [Tram(D1)NP(D2)], there was a marked reversal of the inhibition of clonogenic survival observed with either NP treatment alone or the combination of trametinib and NP delivered on the same day (Fig. 5E). Similar results were obtained with HPAF-II cells (not shown). Bliss independence testing revealed trametinib to be antagonistic to NP, with a combination index (CI) score of 14.06 for the AsPC-1 cells and a CI of 2.14 for HPAF-II cells (Supplementary Fig. S5A). We next tested the contrasting scenario by forcing cells into the G₂–M-phase of the cell cycle prior to NP treatment. AsPC-1 cells were treated with the drug RO-3306, a selective ATP competitive inhibitor of the G₂–M-phase checkpoint regulator, CDK-1. This drug has been previously shown to induce G₂–M-phase cell-cycle arrest (17). Following 24 hours of treatment with R0-3306 (10 μmol/L), the percentage of cells in the G₂–M-phase increased from 19.1% to 66.8%, as expected (Fig. 5F). In colony forming assays, RO-3306 treatment alone had no impact on survival, but concurrently treating [RO(D1)NP(D1)] or pretreating with RO-3306 for 24 hours followed by NP on day 2 [RO(D1)NP(D2)] significantly reduced clonogenic survival compared with NP alone (Fig. 5G). In addition, RO(D1)NP(D2) demonstrated reduced tumor cell survival compared with concurrent treatment [RO(D1)NP(D1)]. Sequential treatment with RO(D1)NP(D2) was also determined to have a synergistic effect based on Bliss independence testing with a CI of 0.67 (Supplementary Fig. S5B). These cell-cycle results provide an additional explanation for the increased efficacy of the altered gemcitabine and NP schedules.

We next tested whether the altered scheduling of gemcitabine and NP could increase the therapeutic efficacy of the drug combination in subcutaneous (heterotopic) and orthotopic models of PDAC. Following tumor formation, mice were randomized to one of four treatment schedules delivered in three cycles over 9–11 days: (i) vehicle control, (ii) gemcitabine alone, (iii) Gem(D1)NP(D1), or (iv) Gem(D1)NP(D3) (Supplementary Fig. S6). There were no significant differences in tumor volumes at baseline; however, by day 10 (AsPC-1 and HPAF-II) or day 5 (G37) posttreatment initiation, there was a significant reduction in tumor volume in mice receiving the Gem(D1)NP(D3) regimen compared with Gem(D1)NP(D1) (Fig. 6A, C, and E; P < 0.001). The tumor doubling times in the Gem(D1)NP(D3) group compared with the Gem(D1)NP(D1) group were markedly longer (log-rank test, P < 0.0001; Fig. 6B, D, and F; P < 0.0001). In terms of toxicities, there were no significant differences in weight loss between the groups or in clinical tolerance of the mice to the different scheduling regimens (Supplementary Fig. S7). We next tested the efficacy of the altered schedule of gemcitabine plus NP in an orthotopic murine model by injecting 1 × 10⁴ G37 cells into the pancreata of athymic mice. One week postinjection, the mice were randomized to (i) vehicle control, (ii) Gem(D1)NP(D1), or (iii) Gem(D1)NP(D3) treatment regimens as described in the xenograft model. Abdominal ultrasound imaging was utilized starting 1 week posttreatment course to measure tumor dimensions and volume in vivo. There was a significant reduction in tumor volumes at 1 and 2 weeks posttreatment in both the Gem(D1)NP(D1) and Gem(D1)NP(D3) regimens compared with vehicle control group (Fig. 6G; P < 0.05). Importantly, there was also a significant reduction in tumor volumes at 1 and 3 weeks posttreatment in the Gem(D1)NP(D3) group compared with Gem(D1)NP(D1) group (Fig. 6G; P < 0.05). We next injected 1 × 10⁴ KPC-Luc (luciferase bearing) cells into the pancreata of athymic mice, and randomized the mice into the three treatment groups 1 week later, as described in the G37 orthotopic experiment. After 3 weeks, mice were sacrificed, and the ratio of pancreas to body weight was significantly reduced in both the Gem(D1)NP(D1) group (P = 0.05) and the Gem(D1)NP(D3) group compared with the control mice (P = 0.01; Fig. 6H). The pancreas to body weight ratios in the Gem(D1)NP(D3) group compared with Gem(D1)NP(D1) group were also significantly reduced, indicating increased efficacy with altered gemcitabine and NP scheduling (P = 0.05).

Nab-paclitaxel and gemcitabine combination therapy is now one of the preferred treatment options for advanced PDAC. NP (originally called ABI-007), an albumin-bound chemotherapy, was generated to overcome the poor solubility issues of the Cremophor-EL solvent required for free paclitaxel (18). Subsequent clinical trials showed that NP improved efficacy and toxicity profiles over standard paclitaxel in breast and non–small cell lung cancer (NSCLC) trials (19, 20). Interestingly, minimal attention is often paid to optimizing treatment schedules of combinations of chemotherapy drugs prior to clinical testing. With this is mind, we tested alternative delivery schedules based on the current knowledge of albumin physiology and biology in tumor cells using preclinical models of PDAC and found marked improvements in efficacy with delaying administration of NP for 24–48 hours after gemcitabine. Our findings appear to be related to increased expression of Cav-1 and cell-cycle–dependent effects of gemcitabine pretreatment. If clinically validated, our results could lead to a new standard of care for patients receiving gemcitabine and nab-paclitaxel

Immunoprecipitation studies have shown membrane-associated albumin-binding glycoprotein (gp60) associates with Cav-1 after gp60 binding to albumin (21). Caveolae and, thus, Cav-1 are critical for the process of gp60-mediated albumin transcytosis (22). Caveolae are 50–100 nm flask-shaped invaginations of the plasma membrane that are important for endocytosis and cholesterol homeostasis, and Cav-1, a scaffolding protein, is necessary for caveolae formation (9). Cav-1 has been found to be overexpressed in a variety of cancer types, including prostate, esophageal, non–small cell lung, breast, and pancreatic cancers (23–28). Interestingly, Cav-1 has been shown to be either a tumor suppresser or tumor promoter depending on the cancer type (29). In the setting of PDAC, the majority of evidence supports Cav-1 as a tumor promoter, and Cav-1 has been previously shown to promote resistance to chemotherapy and radiation (10, 11, 30). In the large-molecular subtyping classification studies by Collisson and colleagues, as well as The Cancer Genome Atlas publication, Cav-1 was associated with the more aggressive subtypes of PDAC, that is, quasi-mesenchymal and basaloid subtypes (31, 32). Previous work from our laboratory showed Cav-1 is a critical mediator for NP uptake and the efficiency of NP was reduced following genetic knockdown of Cav-1 in preclinical PDAC models (13). There is recent evidence in other disease sites that Cav-1 expression correlates with NP response. In a single-arm phase II clinical trial evaluating nab-paclitaxel and gemcitabine in 85 patients with metastatic breast cancer, both high tumor and stromal Cav-1 expression as determined by IHC was associated with longer progression-free survival (log-rank P = 0.03; 33). In addition, for advanced NSCLC, a single-arm phase II trial with carboplatin and nab-paclitaxel also showed high Cav-1 stromal expression by IHC was associated with both improved response rates and OS (34). Herein, we show that loss of Cav-1 through RNA inhibition resulted in both reduced albumin uptake and lower sensitivity to NP and gemcitabine altered scheduling combinations (Fig. 4). Our data support that gemcitabine-mediated increases in Cav-1 over 24–48 hours facilitate increased uptake of NP, resulting in enhanced tumor cell killing.

Interestingly, our findings seemed to suggest that tumor cells with lower Cav-1 expression might be more effectively killed by altered scheduling (Fig. 3 vs. Supplementary Fig. S3). AsPC-1, HPAFII, and G37 human pancreatic cancer cells that have low Cav-1 expression all significantly benefited from altered sequencing (Fig. 3; Supplementary Fig. S3), while higher Cav-1–expressing BxPC3 and CFPAC-1 cells demonstrated less tumor cell suppression (Supplementary Fig. S3). However, other moderate to high Cav-1–expressing cells lines, Panc-1 (Supplementary Fig. S3) and MP2 (black bars on Fig. 4C), also seemed to benefit from altered scheduling. Taken together, our results suggest that in cells with higher baseline Cav-1, the effects of altered sequencing may be blunted, and that more study is warranted given the heterogeneity of response in higher Cav-1–expressing cell lines.

Another important finding in our study was that gemcitabine treatment resulted in synchronization of cells in the G₀–G₁-phase of the cell cycle at 24 hours, followed by a relatively synchronized enrichment of cells passing into G₂–M-phase 24–36 hours following gemcitabine removal (Fig. 4A and B). The ideal cell-cycle phase for NP sensitization is the G₂–M-phase, due to the timing of microtubule separation. Thus, by treating cells with gemcitabine and NP concurrently, there might be less efficacy due to a greater proportion of cells arrested in the G₀–G₁-phase by gemcitabine, which could counteract the efficacy of NP, as supported by our finding of pretreating cells with a G₀–G₁-arresting agent followed by NP (Fig. 5D). Similar evidence from Wang and colleagues demonstrated that cell synchronization into G₀–G₁-phase via thymidine block followed by fresh media exchange resulted in cells entering into the G₂–M-phase, which likewise resulted in increased effectiveness of paclitaxel in ovarian cancer cell lines (35).

In summary, our data indicate that gemcitabine treatment increases Cav-1 expression and albumin uptake. Treating cells first with gemcitabine followed by NP at 48 hours was found to have the greatest inhibition of tumor cell growth both in vitro and in vivo. This strategy may be most useful in low-Cav-1–expressing tumors and, thus, Cav-1 might serve as a predictive biomarker for patients who would benefit most from sequential gemcitabine and nab-paclitaxel treatment. As such, patients with low Cav-1 would stand to benefit most from sequential therapy, while patients with high Cav-1 would benefit less. Overall, our studies support further preclinical and clinical testing of altered scheduling of gemcitabine and NP for patients with PDAC, and represents a relatively simple and highly translatable novel treatment strategy that could result in dramatic improvements in treatment efficacy and a new standard of care with successful clinical validation.

Z. Cruz-Monserrate reports grants from NCI outside the submitted work. A. Chakravarti reports grants from Varian Medical Systems outside the submitted work. No disclosures were reported by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Advancing Translational Sciences or the NIH.

A.R. Wolfe: Conceptualization, formal analysis, validation, writing-original draft, writing-review and editing. R. Robb: Formal analysis, investigation, methodology, writing-original draft, writing-review and editing. A. Hegazi: Resources, data curation, formal analysis. L. Abushahin: Resources, writing-original draft. L. Yang: Data curation, software, formal analysis, investigation, writing-original draft. D.-L. Shyu: Data curation, investigation, methodology. J.G. Trevino: Resources, investigation. Z. Cruz-Monserrate: Resources, investigation, writing-original draft. J. Jacob: Data curation, investigation. K. Palanichamy: Data curation, investigation. A. Chakravarti: Resources, data curation. T.M. Williams: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

The authors thank M. Summers and M. Venere for use of IncuCyte system. The authors also thank the Analytical Cytometry and Small Animal Imaging Core (SAIC) at OSU for assistance with flow cytometry as well as in vivo ultrasound experiments. This work was supported by the following grants: NIH R01 CA198128-01 (to T.M. Williams). Research reported in this article was also supported by The Ohio State University Comprehensive Cancer Center (OSU-CCC) and NIH (P30 CA016058).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.