Multicenter Validation Study of Quantitative Imaging Mass

Apr 2, 2019 - approaches to IMS quantification, the mimetic tissue model .... bioanalytical group at GSK to quantify CLZ and NCLZ from a ... software...
0 downloads 0 Views 2MB Size
Subscriber access provided by OCCIDENTAL COLL

Article

Multicenter Validation Study of Quantitative Imaging Mass Spectrometry Jeremy A Barry, Rima Ait-Belkacem, William M Hardesty, Lydia Benakli, Clara Andonian, Hermes Licea-Perez, Jonathan Stauber, and Stephen Castellino Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.9b01016 • Publication Date (Web): 02 Apr 2019 Downloaded from http://pubs.acs.org on April 3, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Multicenter Validation Study of Quantitative Imaging Mass Spectrometry. Jeremy A. Barry;*1 Rima Ait-Belkacem;2 William M Hardesty;1 Lydia Benakli;2 Clara Andonian;3 Hermes Licea-Perez;3 Jonathan Stauber;2,4 Stephen Castellino1 1 Bioimaging, GlaxoSmithKline, 1250 S. Collegeville Rd, Collegeville, PA 19426, USA 2 Imabiotech SAS, Parc Eurasanté, 152 rue du Docteur Yersin, 59120 Loos, France 3 Bioanalysis, GlaxoSmithKline, 1250 S. Collegeville Road, Collegeville, PA 19426, USA 4 Imabiotech Corp, 44 Manning Rd, Billerica, MA 01821, USA JAB and RAB contributed equally to study design *Corresponding Author: Jeremy A. Barry – [email protected] Abstract: The aim of this study was to assess potential sources of variability in quantitative imaging mass spectrometry (IMS) across multiple sites, analysts, and instruments. A sample from rat liver perfused with clozapine was distributed to three sites for analysis by three analysts using a pre-defined protocol to standardize the sample preparation, acquisition, and analysis parameters. In addition, two commonly used approaches to IMS quantification, the mimetic tissue model and dilution series, were used to quantify clozapine and its major metabolite norclozapine in isolated perfused rat liver. The quantification was evaluated in terms of precision and accuracy with comparison to liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). The results of this study showed that across three analysts with six replicates each, both quantitative IMS methods achieved relative standard deviations in the low teens and accuracies of around 80% compared to LC-MS/MS quantification of adjacent tissue sections. The utility of a homogenously coated stable-isotopically labeled standard (SIL) for normalization was appraised in terms of its potential to improve precision and accuracy of quantification as well as qualitatively reduce variability in the sample tissue. SIL normalization had a larger influence on the dilution series where the use of the internal standard was necessary to achieve accuracy and precision comparable to the non-normalized mimetic tissue model data. Normalization to the internal standard appeared most effective when the intensity ratio of the analyte to internal standard was approximately one and thus precludes this method as a universal normalization approach for all ions in the acquisition.

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 18

Introduction: The combination of spatial localization with the inherent benefits of mass spectrometry detection such as sensitivity, selectivity, multiplex detection, and molecular characterization have allowed imaging mass spectrometry (IMS) to emerge as a unique and powerful tool in the biomedical sciences for the determination of molecular distributions. Even within a single IMS dataset there is a vast wealth of biological and chemical knowledge to be gained and while the ion distributions alone can be useful, the contextualization of this data is arguably far more valuable. Advances in bioinformatics and multi-modal imaging have helped to provide this connection between the ion distribution and underlying biology through correlation with histology or metabolic pathways.1-3 The development and validation of quantitative IMS has had a significant impact on a variety of disciplines, particularly those surrounding the application to drug discovery and development.4-9 The ability to quantify molecular distributions with high sensitivity and selectivity has broad implications for the assessment of the efficacy and safety of dosed compounds.10-13 There are many factors to consider when conducting a quantitative IMS experiment and several of these key topics have been discussed in detail recently.14 Importantly, the measured intensity of any ion from a complex matrix such as tissue is effected by the conditions of mutual extraction, ionization, and detection with molecular species in the sample, requiring that standard curves be prepared in comparatively matched matrices for accurate quantitation. Two common experimental designs for quantitative IMS have emerged including the mimetic tissue model15 and the dilution series.16 In the dilution series experiment, several concentrations of a standard solution are spotted onto control tissue sections where the underlying tissue section acts as a surrogate background matrix. Due to its relative simplicity, this approach is more prevalent in the literature. One drawback, however, is that it is unclear how well topically applied standards mimic extraction of the analyte from within the tissue. The mimetic tissue model was designed to account for this by spiking the standard series into tissue homogenates which are then frozen and sectioned alongside the tissue to be quantified. This method does require additional biological tissue and greater effort in terms of sample preparation but provides benefits to quantitative accuracy and reproducibility. A recently revised protocol17 improves upon the efficiency of the mimetic model preparation and application.18 A comparator for quantitative IMS has traditionally been achieved through correlation with liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) of serial tissue sections.15-16, 19-20 However, the tissue distribution determines the legitimacy of comparison to a homogenate value.18 In cases where the distribution is localized, extraction from whole tissue homogenates or even serial sections may not be reasonable. This issue is largely addressed by incorporating laser capture microdissection (LCM) to add a degree of spatial information to the quantification. LCM coupled to LCMS/MS would therefore be a more appropriate comparator for IMS quantification in most cases.13, 21-23 A crucial aspect of quantitative IMS is the consideration of how tissue matrix effects influence the observed ion intensity. Fluctuations in ion intensity from an evenly coated standard across a heterogeneous tissue are often attributed to differences in ion suppression.24-26 It is likely that other matrix effects such as differences in tissue binding and extraction efficiency also contribute to this observation. Several articles have been published on the use of normalization to a stable isotopicallylabeled (SIL) standard; however, the findings on the benefit of such an approach are confounding. Prior work investigating quantification using liquid extraction surface analysis (LESA)-IMS has shown that the internal standard actually increased the variability at each of the calibration points.27 Hansen and Janfelt also called into question the benefit of a topically applied internal standard because it was not able to account for all matrix effects across tissue types.28 Other work has indicated that the ion suppression effect may be both tissue and drug specific. Given that ion intensity in MALDI is dependent on matrix application, molecule specific ionization efficiency and tissue matrix effects, Hamm and Bonnel proposed the use of a normalization factor referred to as the tissue extinction coefficient (TEC).29 ACS Paragon Plus Environment

Page 3 of 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Reproducibility and accuracy are key metrics which govern the true value of any analytical methodology. These metrics can be assessed through multicenter studies which also serve to validate and build confidence in relatively novel technologies such as IMS. A few such studies have been performed over the last five years assessing various IMS capabilities.30-35 Common to each of these studies is the large number of analytical variables that need to be standardized in order to deconvolute biological effects. Here we report on the repeatability, reproducibility, and accuracy of quantitative IMS employing both the mimetic tissue model and dilution series approaches. Three analysts at three different sites analyzed tissue sections of rat liver which had been isolated and perfused with clozapine (CLZ). IMS quantification was performed using a standardized protocol to determine the concentration of CLZ and its primary metabolite norclozapine (NCLZ). A homogenously coated internal standard, clozapine-d8 (CLZ-d8) was also included to test the impact of SIL normalization on the precision and accuracy of both IMS quantification approaches. Experimental Section Materials CLZ, NCLZ, ketoconazole (KCZ), 2,5-dihydroxybenzoic acid (DHB), trifluoroacetic acid (TFA) and methanol were obtained from Sigma Aldrich (St. Louis, MO, USA) and CLZ-d8 was supplied by Tocris through Fisher Scientific (Pittsburgh, PA, USA). Structures for CLZ, NCLZ, CLZ-d8, and KCZ are shown below. NH

N N

N N

N Cl

Cl N H

Clozapine (CLZ) [M+H]+: 327.13710

Cl N H

D D N D D D D D N D N

Clozapine-d8 (CLZ-d8) [M+H]+: 335.18731

O

O H N

N H

Norclozapine (NCLZ) [M+H]+: 313.12145

N

O

N

O Cl

N

Cl

Ketoconazole (KCZ) [M+H]+: 531.15604

Isolated Perfused Rat Liver (IPRL) sample generation Animal studies were conducted in accordance with the GSK Policy on the Care, Welfare and Treatment of Laboratory Animals and were reviewed by the Institutional Animal Care and Use Committee at GSK. The CLZ perfusate consisted of bovine red blood cells, bovine serum albumin, glucose, and an appropriate volume of CLZ to achieve the target dose which was dissolved in Krebs-Ringer saline. Liver from a male Han Wistar rat was surgically prepared with bile duct, hepatic portal and vena cava cannulae. The liver was perfused for 1.5 hours at a rate of 30 mL/min giving a target dose of 20 mg/kg. At the end of the experiment, the liver was flushed with 5 mL of Krebs-Ringer bicarbonate and excised. The left and right lateral and median lobes of the liver were separated, frozen in liquid nitrogen, and stored at -80 °C. Study Outline Greater experimental detail as well as the study protocol are provided in the supporting information. Briefly, the IPRL right median lobe was portioned into two tissue blocks and distributed to GSK and Imabiotech for IMS analysis. In total, three analysts (two from GSK and one from Imabiotech) performed the protocol at different sites where each analyst was independently responsible for mimetic model preparation, dilution series preparation, tissue sectioning, matrix application, IMS acquisition, and data analysis. IMS quantification was conducted using both the mimetic tissue model and the dilution series calibration approaches. To assess IMS accuracy, a continuous series of tissue sections ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 18

were collected from the GSK block, homogenized, and analyzed by LC-MS/MS by an independent bioanalytical group at GSK to quantify CLZ and NCLZ in the IPRL tissue. Mimetic Tissue Model Preparation Each analyst prepared a mimetic tissue model following the recently published revised protocol.17 The tissue model was prepared to include an inverse gradient of tissue concentrations for CLZ and NCLZ as well as a blank region spiked with KCZ as shown in Figure S1. The calibration levels for CLZ included 30, 24, 18, 12, 6, and 0.5 µg/g. Within those respective layers, NCLZ was spiked at calibration levels of 10, 60, 120, 180, 240, and 300 µg/g. These spiked homogenates were serially frozen into a cylindrical mold resulting in a plug of tissue with a stepped gradient of CLZ and NCLZ concentrations. Dilution Series Preparation Solutions covering a range of concentrations for CLZ (0, 0.05, 0.25, 0.5, 1, 1.5, 2, 2.5, and 3 µM) and NCLZ (0, 5, 15, 25, 35, 45, 55, 65, and 75 µM) were prepared in 50:50 methanol:water. Prior to matrix application, 1 µL of each of these solutions were deposited onto sections of control rat liver tissue and allowed to air dry which resulted in tissue concentrations which were comparable to the mimetic tissue model range. Concentrations were converted to µg/g by accounting for the tissue volume of each calibration spot (droplet area x tissue thickness) and assuming a tissue density of 1 g/cm3. Sectioning and Matrix Application Each IMS analyst prepared six replicate slides which contained: a section from the IPRL tissue, mimetic tissue model and control liver tissue sections for the dilution series. Tissues were sectioned on a cryomicrotome to a thickness of 12 µm and were thaw-mounted onto indium tin oxide (ITO) coated microscope slides (Bruker, Germany) which had been coated with poly-lysine to improve tissue adherence. Slides were stored at -80 °C until the day of analysis. On the day of analysis, a slide was removed from the freezer and equilibrated to room temperature under reduced humidity and the dilution series was spotted onto the control liver sections. Matrix solutions were prepared and sprayed for each individual slide on the day of analysis to capture potential variability associated with sample preparation. The matrix solution consisted of 40 mg/mL DHB dissolved in methanol:water (50:50) with 0.1% TFA and CLZ-d8 was spiked at a concentration of 0.5 µM. This solution was applied using an HTX TM Sprayer (Carrboro, NC) in 8 passes with 0.1 mL/min flow rate, 10 psi nitrogen pressure, 70 °C nozzle temperature, 1350 mm/min nozzle velocity, and 3 mm track spacing with a 1.5 mm offset. IMS Acquisition MALDI IMS was performed on a Bruker 7 T Solarix FT-ICR mass spectrometer (Billerica, MA) in positive ion mode at a spatial resolution of 100 µm. Key acquisition parameters included mass range (2001000 m/z), spectral resolution (~66,000 at m/z 400), laser focus (small), data reduction (95%), the number of laser shots (300) and the laser frequency (2 kHz). Other method parameters were optimized for the individual instrument. A summary of the acquisition regions is provided in Figure S1. A region of interest (ROI) was collected from the QC region (KCZ-spiked layer) at the start and end of each replicate analysis which included ROIs for the IPRL, mimetic tissue model, and dilution series. Data Analysis The two GSK datasets were loaded into fleximaging (Bruker, Billerica, MA). Pixel intensities within a 10 ppm window for each analyte of interest (CLZ, NCLZ, CLZ-d8, and KCZ) were exported from the calibration regions. Weighted linear regression was used to build calibration curves of CLZ and NCLZ for both the dilution series and mimetic tissue model according to a recently submitted manuscript.18 SIL normalization was achieved by obtaining the ratio of CLZ or NCLZ to the internal standard (CLZd8) at each pixel. Calibration using the normalized data involved averaging these pixel ratios for every pixel in the ROI of each calibration level. ACS Paragon Plus Environment

Page 5 of 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

The Imabiotech dataset was loaded into Multimaging™ software. ROIs were drawn for each dilution series concentration surrounding the dried droplet based on the scanned optical image. Normalization to CLZ-d8 was applied to each pixel in the dataset within the software. Calibration curves were generated for each method and the dilution series was automatically converted from µM to µg/g after inputting the tissue section thickness and assuming a density of 1 g/cm3. Statistica (StatSoft, Tulsa, OK, USA) was used for mean comparisons and box plot generation. Use of the term “significant” in the discussion below refers to a statistically significant difference (