Elien De Thaye (1,2), Elly De Vlieghere (2,3), An Vermeulen (1), Jan Van Bocxlaer (1) , Pieter Colin (1,2)
(1) Laboratory of Medical Biochemistry and Clinical Analysis, Ghent University, Ghent, Belgium; (2) Cancer Research Institute Ghent (CRIG), Ghent, Belgium; (3) Laboratory of Experimental Cancer Research, Ghent University, Ghent, Belgium
Objectives: Cytokeratin 18 (CK18) is specifically cleaved by caspases during apoptosis. We evaluated the potential use of circulating full-length and caspase-cleaved (cc) cytokeratin 18 (CK18) as a plasma marker for monitoring chemotherapy-induced cell death in ovarian cancer.
Methods: In vitro study: High-grade serous SK-OV-3lucIP1 and mixed-germ cell PA-1 ovarian cancer cells were exposed to increasing concentrations of paclitaxel (PTX, 0 to 1000 nM) during 24 hours. Following drug exposure of both cell lines, levels of cleaved and intact CK18 were measured in the culture medium, up to 5 days after washing away treatment, using specific ELISA assays (M30 Apoptosense® and M65 EpiDeath® ELISA, VLVbio, Sweden). In addition, cell survival was investigated using an MTS assay (CellTiter 96® AQueous One Solution Cell Proliferation Assay, Promega, The Netherlands). Cell survival and ccCK18 release data were analysed using a full inhibitory and stimulatory sigmoidal Emax model, respectively, as implemented in NONMEM®. In vivo study: Both non-tumor bearing and SK-OV-3lucIP1 and PA-1 xenografts were established in SCID/Beige mice. Single doses of PTX (Abraxane®, Celgene, US) were administered intravenously (iv; 5 or 50 mg/kg) or intraperitoneally (ip; 5, 15 or 50 mg/kg). Blood was collected within 24 hours after treatment in order to characterize PTX pharmacokinetics. Plasma was collected up to 3 weeks after treatment for measuring levels of (cc)CK18. Based on the PK data, a non-linear mixed-effect model was developed using NONMEM®. The discriminatory power of the pharmacodynamic markers (cc)CK18 were assessed by receiver operating characteristic (ROC) analysis.
Results: Exposure-response modeling of the in vitro PTX effects regarding cell survival resulted in an Emax of 90.3% (0.3% RSE and 95% confidence interval (CI): 89.8 – 90.8) and 46.7% (4.1% RSE and 95% CI: 43.0 – 50.4) for PA-1 and SK-OV-3lucIP1, respectively. C50 values were estimated at 9.06 nM (0.1% RSE and 95% CI: 9.04 – 9.08) and 26.4 nM (23% RSE and 95% CI: 14.5 – 38.3) while the estimated Hill coefficient (γ) was equal to 6.56 (0.9% RSE and 95% CI: 6.44 – 6.68) and 0.744 (13.8% RSE and 95% CI: 0.542 – 0.946) for PA-1 and SK-OV-3lucIP1, respectively. Estimated Emax parameters resulting from the exposure-response modeling of the drug effects regarding the released amount of ccCK18 were equal to 565 (5.5% RSE and 95% CI: 504 – 626) and 30.5 (9.6% RSE and 95% CI: 24.8 – 36.2) for PA-1 and SK-OV-3lucIP1, respectively. C50 values were equal to 1.15 nM (0.5% RSE, 95% CI: 1.14 – 1.16) and 8.05 nM (5.4% RSE and 95% CI: 7.20 – 8.90), estimates for the Hill coefficient (γ) were 19.6 (6.3% RSE and 95% CI: 17.2 – 22.0) and 9.38 (13.9% RSE and 95% CI: 6.83 – 11.93) for PA-1 and SK-OV-3lucIP1, respectively. A two-compartmental model was built based on all available in vivo PK data, and the following population PK parameters were obtained: clearance (0.0476 L/h), central and peripheral volume of distribution (0.0401 and 0.0429 L), intercompartmental clearance (0.0061 L/h), an absorption rate constant (0.594/h) and bioavailability (77.7%). Inter-individual variability was taken into account for clearance, the central and the peripheral volume of distribution. Both tumor presence and xenograft type (no tumor versus SK-OV-3lucIP1 or PA-1 xenograft) were not withheld as covariates. Based on the observed difference in area under the ROC curve (AUC), ccCK18 (pmol/g tumor) was considered to be a better marker to differentiate between responder and non-responders to PTX therapy in PA-1 (AUC of 0.831 and 95% CI: 0.668 to 0.995) compared to SK-OV-3lucIP1 (AUC of 0.556 and 95% CI: 0.380 to 0.731) xenografts.
Conclusions: In this study, the difference in cell sensitivity to the chemotherapeutic agent PTX and drug-induced apoptosis was reflected in the released ccCK18 biomarker levels. However, given the fact that the investigated PA-1 and SK-OV-3lucIP1 cell lines represent very different ovarian cancer subtypes and taking into account the rather aspecific character of CK18, the application and implementation of this marker in a clinical setting aiming to optimize or personalize cancer treatment is debatable.
References:
[1] NONMEM® version 7.4 (ICON Development Solutions, Hanover, MD, USA)
Reference: PAGE () Abstr 9403 [www.page-meeting.org/?abstract=9403]
Poster: Drug/Disease Modelling - Oncology