All procedures were performed under approved Institutional Animal Care and Use Committee (IACUC) protocols (#08-01-001).

The human MPM cell line, MSTO-211H (American Type Culture Collection), was maintained RPMI-supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 ug/mL streptomycin in a 5% CO₂ humidified incubator at 37°C. Green fluorescent protein (GFP)-Firefly luciferase fusion and human mesothelin (MSLN) genes were cloned into SFG retroviral vectors, which were then transfected into H29 packaging cell line using calcium phosphate mediated transfection protocol. MSTO-211H cells were plated in 24-well plates 24 hours prior to retroviral transduction. Media containing filtered virus was added to cells permeablized using 8 ug/mL polybrene (Sigma-Aldrich, MO). Cells were reinfected with freshly collected virus 24 hours later. The human MSLN variant 1 was isolated from the human MSLN-expressing ovarian cancer cell line OVCAR using TRIzol Reagent. RT-PCR synthesis of full length cDNA of human MSLN was performed using SuperScript™ III One-Step RT-PCR System with Platinum® Taq High Fidelity Kit. The 5′ primer is GAT CTA CAC AGA CCA TGG CCT TGC CAA CGG, including a Nco I site, and the 3′ primer is GCG CAG ATC TTA CGT ATC AGG CCA GGG TGG AGG CTA G, including a Bgl II site and a Sna BI site. Plasmid DNA was isolated, subcloned into the SFG retroviral vector, confirmed by sequencing, and used to stably transduce MSLN.

Flow activated cell sorting (FACS) was performed using FACSAria (BD Biosciences). Human mesothelin cell-surface expression was detected using a phycoerythrin-conjugated or allophycocyanin-conjugated anti-human mesothelin rat IgG_2A (R&D systems, MN). Flow cytometry for GFP and mesothelin expression was performed on LSRII cytometer (BD Biosciences) and analyzed using FlowJo (Tree Star) software.

Female SCID/beige and male athymic nude mice (6–10 weeks of age, Taconic Farms, NY) were used. Mice anesthetized with inhaled isoflurane and oxygen underwent direct intrapleural injection of tumor cells in 200 uL serum-free media via a right thoracic incision. Following inoculation, tumor take and progression occurred in >95% of mice. Mice were sacrificed when moribund as per IACUC guidelines. The details of our intrapleural xenografting technique have been previously described [10], [11], [12].

Immunofluorescence staining for angio- and lymphangiogenesis was performed using CD34 rat monoclonal (eBioscience) and LYVE-1 goat polyclonal (R&D Systems) antibodies, respectively. Immunofluorescencent detection was performed with Streptavidin-HRP D (Ventana Medical Systems), followed by incubation with Tyramide-Alexa Fluor 488 (Invitrogen) or Tyramide Alexa Fluor 568 (Invitrogen) for CD34 and LYVE-1, respectively. The protocols were established and experiments performed at the Molecular Cytology Core Facility, MSKCC, using Discovery XT automatic processor from Ventana Medical Systems.

In vitro bioluminescence imaging (BLI) standardization was performed using GFP-Firefly luciferase expressing MSTO-211H cells with and without MSLN expression. Cells were plated in 96-well tissue culture plates in serial dilution from 1.6×10⁶ to 2.5×10⁴ cells/100 µL/well. Twenty minutes after the addition of 100 µL D-Luciferin to each well (15 mg/mL; Caliper Lifesciences, MA), plates were imaged using the Xenogen IVIS 100 Imaging System (Caliper Lifesciences). BLI data were analyzed using Living Image 2.60 software. Cell number versus total BLI flux (photon/s) was evaluated by Pearson's correlation.

In vivo BLI in tumor-bearing mice was performed using a single intraperitoneal dose of 150 mg/kg D-Luciferin. Mice were imaged with the Xenogen IVIS 100 Imaging System 20 minutes following D-Luciferin injection and images acquired for 5–30 seconds depending on signal strength. BLI data were analyzed using Living Image 2.60 software; BLI signal from regions of interest (ROI) were reported as total flux (photons/s).

Magnetic resonance imaging (MRI) was performed on mice anesthetized with 1% isoflurane in 20% Oxygen and imaged in a Bruker 4.7T USR scanner (Bruker Biospin Inc., Ettlingen, Germany) equipped with a 400 mT/m gradient coil and a 32 mm ID custom build birdcage resonator. Thoracic axial MRI images were acquired using a RARE fast spin-echo sequence with a repetition time (TR) of 1.7 s, echo time (TE) 40 ms and 12 averages. The acquisition was triggered by animal respiration to reduce respiration induced motion artifacts. The slice thickness was 0.7 mm and the in-plane image resolution was 117×156 mm. Tumor volumes (mm³) were measured by tracing tumor boundaries in each slice using Bruker ParaVision Xtip software (Bruker Biospin Inc., Ettlingen, Germany) and then calculated from the areas of tumor regions in each slice.

Blood was collected in K₂EDTA tubes, immediately centrifuged to separate serum, and stored at −20°C until assayed. Soluble mesothelin-related peptide (SMRP) was measured using an ELISA microplate sandwich assay (Mesomark® Fujirebio Diagnostics, Inc). The detection limit of this assay was 0.3 nM. Serum osteopontin (OPN) was measured using Quantikine® OPN immunoassay. The assay, performed per manufacturer's instructions, is a solid-phase ELISA employing a sandwich immunoassay to quantitate human OPN. Samples were measured in duplicate with standard controls.

Biomarker performance in evaluating therapy response was assessed in Nude Athymic male mice, in which orthotopic pleural tumors, expressing both GFP-Luciferase and human mesothelin, were established as described above. Tumors progressed for 22 days following inoculation at which time equivalent tumor burden was confirmed by BLI signal and mice were divided into experimental groups. Mice received either: (a) cisplatin (4 mg/kg IP weekly), (b) isolated thoracic radiation (20 Gy total in 5 fractions), (c) combination chemoradiation, or (d) no treatment. Isolated thoracic radiation was provided using the XRAD-320 (Precision X-ray, CT) 250 kVp, 12 mA, 0.25 mm Cu filtration in conjunction with lead barrier protection of the mouse peritoneal cavity. Treatment was provided for two weeks and the mice were monitored serially with BLI and serum SMRP assessment. Differences in BLI signal and SMRP levels between treatment groups were assessed by Kruskal-Wallis test using area under the curve analysis and Student's t-test, respectively. Predictive value of SMRP for survival following treatment was evaluated using analysis of maximum likelihood estimates and reported as hazard ratios.

To evaluate biomarkers for MPM, we utilized an intrapleural mouse model. This model recapitulates the human tumor microenvironment grossly and histopathologically (Fig. 1A). In addition, the extensive vascularization of the MPM tumors in our model (Fig. 1B) facilitates detection of serum biomarker. This clinically relevant model provided an appropriate platform to investigate biomarkers for clinical translation. To non-invasively monitor tumor progression, we validated bioluminescent imaging (BLI) as an accurate, quantitative modality to assess tumor burden. We confirmed a linear correlation between number of mesothelioma cells (transduced to express firefly-luciferase) and BLI photon counts in vitro (Pearson r = 0.999, p<0.0001) and demonstrated that BLI can detect as few as 1×10³ tumor cells in the pleural space of mice (data not shown). We next demonstrated a strong correlation between bioluminescent flux and pleural tumor volume as determined by magnetic resonance imaging (MRI), the gold standard for tumor volume assessment [13] (r = 0.86, p<0.0001, adjusted for within mouse clustering; Fig. 1C and D). Accurate, quantitative BLI is possible in this model because tumor grows along the chest wall as a thickening of the pleural rind minimizing tumor depth and BLI signal attenuation (Fig. 1A). These experiments established that BLI in the orthotopic pleural model provides an accurate, non-invasive, serial, and quantitative evaluation of tumor burden.

We next sought to evaluate the candidate mesothelioma tumor biomarker, SMRP. SMRP is a cleavage product of the cell-surface protein mesothelin, overexpressed by epithelioid MPM (the most common MPM subtype), and has shown promise as a tumor biomarker [3], [14], [15], [16]. We first confirmed sustained mesothelin expression in the orthotopic MPM model by immunohistochemistry and flow cytometry even at advanced stages of disease (data not shown). We observed that SMRP in the pleural MPM model correlated with tumor burden and progression as confirmed by MRI tumor volume and BLI signal (Fig. 2A–C). We simultaneously evaluated serum OPN, another candidate serum MPM biomarker. Using the aforementioned quantitative imaging parameters, serum OPN levels neither correlated with tumor burden (Fig. 2D) nor SMRP level. Based on these observations, we only evaluated SMRP in subsequent experiments.

To examine the importance of a serum biomarker in predicting therapy response in MPM, we evaluated serum SMRP levels following multimodality treatment of mice with orthotopic pleural tumors. Immunodeficient nude athymic mice were engrafted with human MPM tumor cells intrapleurally and then treated for two weeks with multimodality therapy (Fig. 3A). Mice were monitored using serial serum SMRP, BLI, and followed for survival. We noted a significant decrease in serum SMRP level in the treatment groups compared to controls. Although SMRP increased during chemoradiation treatment cycles (days 0–11), the rate of increase was significantly decreased (slope of 0.04 versus 0.21 for controls, Wilcoxon test, p = 0.01; Fig. 3B). Decreased serum SMRP correlated with slowed tumor progression as reflected in BLI signal following treatment compared to control mice (data not shown). In addition, elevated serum SMRP was associated with an increased risk of death (hazard ratio = 4.5, 95% CI 1.53–13.1). Using this mouse model we demonstrated that the percentage change in SMRP levels correlates with survival and response to therapy (defined as <2-fold increase in BLI signal over the treatment period). Following treatment, the percent increase in SMRP level was significantly lower in mice that responded to therapy compared to non-responders (10-fold vs. 17-fold increase, p = 0.009, Fig. 3C). Mice with a lower percent increase in SMRP (less than the group median) demonstrated improved survival compared to mice with a higher percent SMRP increase (p<0.001, Fig. 3D).

Having evaluated the use of SMRP in the mouse model to detect therapeutic response following standard therapies, we next assessed the utility of the model in evaluating the efficacy of an emerging targeted therapy. Our laboratory is developing a targeted T-cell immunotherapy for MPM based upon our experience with genetically modified targeted human T cells [17]. Treatment response following MPM-targeted T-cell therapy versus control (unpublished observations), as defined as <2-fold increase in BLI signal over the treatment period, was associated with percentage change in SMRP (0.62-fold decrease vs.11-fold increase, p = 0.002, Fig. 4A). This noninvasive biomarker assessment has guided the dosing schedule in our ongoing immunotherapy experiments. Serum SMRP was a sensitive indicator of tumor progression following a response to an initial treatment with targeted T cells (Fig. 4B). These experiments demonstrate that serum SMRP is a sensitive, noninvasive marker of tumor response to therapy and relapse.